Gary Gilleskie, Charles Rutter, Becky McCuen

Biopharmaceutical Manufacturing
Also of Interest
Solubility in Pharmaceutical Chemistry
Christoph Saal and Anita Nair (Eds.), 
ISBN ----, e-ISBN (PDF) ----,
e-ISBN (EPUB) ----

Downstream Processing in Biotechnology

Venko N. Beschkov and Dragomir Yankov (Eds.), 
ISBN ----, e-ISBN (PDF) ----,
e-ISBN (EPUB) ----

Bioorganometallic Chemistry
Wolfgang Weigand und Ulf-Peter Apfel (Eds.), 
ISBN ----, e-ISBN (PDF) ----,
e-ISBN (EPUB) ----

Industrial Biotechnology
Mark Anthony Benvenuto, 
ISBN ----, e-ISBN (PDF) ----,
e-ISBN (EPUB) ----

Integrated Bioprocess Engineering

Clemens Posten, 
ISBN ----, e-ISBN (PDF) ----,
e-ISBN (EPUB) ----
Gary Gilleskie, Charles Rutter, Becky McCuen

Biopharmaceutical
Manufacturing

Principles, Processes, and Practices

Authors
Dr. Gary Gilleskie
Golden LEAF Biomanufacturing
Training and Education Center
NC State University
850 Oval Drive, Campus Box 7928
RALEIGH NC 27695-7928
USA
[email protected]

Dr. Charles Rutter

Golden LEAF Biomanufacturing
Training and Education Center
NC State University
850 Oval Drive, Campus Box 7928
RALEIGH NC 27695-7928
USA
[email protected]

Becky McCuen
KBI BioPharma, Inc.
1101 Hamlin Rd
DURHAM NC 27704
USA
[email protected]

ISBN 978-3-11-061687-3
e-ISBN (PDF) 978-3-11-061688-0
e-ISBN (EPUB) 978-3-11-061701-6

Library of Congress Control Number: 2021935540

Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available on the Internet at http://dnb.dnb.de.

© 2021 Walter de Gruyter GmbH, Berlin/Boston

Cover image: NC State University
Typesetting: Integra Software Services Pvt. Ltd.
Printing and binding: CPI books GmbH, Leck

www.degruyter.com
Preface
Upon completing a Ph.D. in chemical engineering, I chose an industrial career. Look-
ing back, I remember my start in the biopharmaceutical industry as being mired in
mistakes and characterized by a general lack of understanding of the methods and
technologies used for production. I recall my boss at the time mentioning the need to
perform an “OQ” (operational qualification) on a piece of equipment. I had no idea
what OQ meant, much less what the equipment being “OQ’d” was used for. And I
wasn’t at all sure what I would do to figure out how to accomplish the task at hand.
Somehow, I managed to survive and even thrive in the biopharmaceutical in-
dustry. There is no doubt that my training in chemical engineering provided a solid
foundation in topics such as fluid dynamics, heat transfer, mass transfer, and reac-
tor design, all of which are fundamental to the unit operations used in a biopharma-
ceutical manufacturing environment. But I still had much to learn. I had only a
modicum of familiarity in the practical and theoretical aspects of the unit operations
common to the biopharmaceutical industry, such as cell culture, chromatography,
or ultrafiltration. I had no knowledge of the stringent documentation rules, the impor-
tance of clear and detailed procedures, qualification and validation requirements,
methods for addressing deviation root causes, and the numerous other expectations
required under Good Manufacturing Practice (GMP) standards that govern production
of biopharmaceutical products. And I had no understanding of the biological product
landscape to help put my work into context – the variety of biological products on
the market, how those products work, and the diseases that were being treated or pre-
vented. When I speak with colleagues of a similar age about their early days in the
industry, most have similar stories.
As I progressed in my career, it became clear to me that a university student
who was offered an educational experience with even just a few courses tailored to
biopharmaceutical production could really hit the ground running when he or she
joined a biopharmaceutical company. My desire to work with others to create these
learning opportunities led me to join the Golden LEAF Biomanufacturing Training
and Education Center (BTEC) at North Carolina State University, which opened in
July 2007 to provide education and training opportunities to develop skilled profes-
sionals in the biomanufacturing industry. This mission has led to the development
of numerous courses – for both university students and professionals – in the area
of GMP manufacturing of biopharmaceuticals. The curriculum that has been devel-
oped is unique and has produced a cadre of graduates who, by all accounts, have
hit the ground running when they enter the biopharmaceutical industry. And our
professional programs have helped to fill knowledge gaps for literally thousands of
professionals affiliated with the industry in a variety of ways. BTEC – its mission,
the many courses that have been developed, the many colleagues who have sup-
ported these efforts during the past decade, and the many students who have (hope-
fully) benefitted – is the inspiration for this book.

https://doi.org/10.1515/9783110616880-202
VI Preface

As you read the book, I think it will become clear that the design and execution
of biopharmaceutical processes is a multidisciplinary effort. For example, expertise in
molecular biology is required to create the cell lines used to produce biopharmaceuti-
cal products. Cell culture and fermentation require training in microbiology, and
chromatography is best undertaken with a background in biochemistry and/or chem-
ical engineering. As a result, covering all the topics addressed in the book required
three co-authors – myself, Charlie Rutter, and Becky McCuen. I took on the down-
stream processing material, Charlie shared his expertise in fermentation and cell cul-
ture, while Becky, who works in the industry, shared GMP/regulatory expertise.
Our vision for the book has been to provide concise, comprehensive coverage of
the principles, processes, and practices underlying biopharmaceutical production.
Our choice of the material included has been influenced by both our industry expe-
rience and teaching experience. We are especially grateful for the many questions
posed over the years by university students looking at this topic with fresh eyes and
by current professionals who bring more experience to the classroom. The questions
from both groups have informed the topics covered here and the way they are cov-
ered. We hope that the book strikes the right balance between practical and theoret-
ical information and the story it tells benefits a variety of audiences: students using
it as a textbook for a course on biopharmaceutical manufacturing or professionals
working in the biopharmaceutical (or a related) industry who simply need informa-
tion on how biopharmaceutical production is done.
Chapter 1 provides an overview of biopharmaceutical products, explaining what
they are, the components of a biopharmaceutical drug product, and what we mean
by quality – a critical topic given that biopharmaceuticals are delivered to patients by
some form of injection and products must be effective for the sake of the patient. This
first chapter is intended to set the foundation for all subsequent discussion by clearly
defining the product that is being produced. Chapter 2 provides an overview of the
process steps used in production of biopharmaceutical products, how these process
steps are designed, and the design of the facilities in which the products are made. It
is intentionally written at a high level, as subsequent chapters provide greater detail
on process steps (i.e., individual unit operations) and practices. An important theme
in the chapter is the criticality of the manufacturing process and production facility to
overall product quality. The discussion around quality continues in Chapter 3, which
covers details on GMP requirements – the minimum requirements that ensure a prod-
uct is safe for use and effective – that apply to biopharmaceutical manufacturing.
Chapters 4–9 take a deeper look at production processes by considering each of
the unit operations commonly used to produce drug substance of the required qual-
ity. Those chapters are presented in the order that unit operations are likely to occur
in a manufacturing process. Chapter 4 covers fermentation and cell culture, the steps
in which most biopharmaceutical products are produced. Chapter 5 covers methods
for cell lysis, a necessary step when a biopharmaceutical product remains within the
host cell once produced. Chapters 6 (centrifugation) and 7 (filtration) look at solid-
 Acknowledgements VII

liquid separations – required because, among a number of applications, the product

must be separated from the cells used to produce it. Chapter 8 discusses chromatogra-
phy, the unit operation commonly used for removing soluble impurities from the
product, which at this point in a process is typically the active ingredient dissolved in
an aqueous solution that also contains a number of undesirable soluble impurities
that would cause harm to a patient. Chapter 9 focuses on ultrafiltration, a technology
that can be used in a number of locations within a process but is almost always
needed to formulate (e.g., exchange buffer systems) the final product. And, finally,
we wrap up in Chapter 10 by summarizing some of the key concepts in the book and
briefly considering some topics that were not able to be covered.
We have striven to make the concepts presented in each chapter clear and accu-
rate. Despite our efforts, we suspect that there may be flaws – hopefully very few –
that need to be fixed. We would be grateful if you would inform us of any such issues
so that they can be corrected.
We would be remiss not to make special mention of vaccines, given the COVID-19
pandemic that has wreaked havoc on much of the world’s population in a matter
of months. The pandemic started two-thirds of the way into the writing of this
book, and it has been fascinating to see the speed at which two vaccines (at the
time of this writing) were developed, authorized for use, and manufactured for
use in the United States. The fact that those vaccines rely on mRNA technology,
when no mRNA vaccines had previously been licensed, makes the COVID-19 vac-
cine story even more interesting. The need to rapidly develop and manufacture
billions of doses of COVID-19 vaccines highlights the importance and societal im-
pact of biopharmaceutical manufacturing. Indeed, many aspects of biopharma-
ceutical manufacturing presented in this book apply to production of those mRNA
vaccines used against COVID-19 and to production of other types of COVID-19 vac-
cines likely to be approved in the near future.

Acknowledgements
A number of colleagues and friends helped in many different ways to bring this
book together. Simply listing them here to acknowledge their contribution in no
way matches their efforts, for which we are most grateful. Lauren Lancaster, Ben
Lyons, Michele Ray-Davis, Arjun Shastry, and John van Zanten provided data and
process-related information that helped in providing real-world examples and prob-
lems. Mark Burdick, Sahr James, Mark Pergerson, Jennifer Sasser, and John Taylor
supported the creation of a number of figures. Donna Gilleskie, Matt Gilleskie, Erik
Henry, Russ McCuen, Baley Reeves, and Caroline Smith-Moore all read through early
versions of various chapters and provided helpful feedback that led to continuous im-
provement in content and clarity during the writing process. We also thank several
VIII Preface

collaborators at MilliporeSigma who read through chapters on filtration and provided

images that greatly enhanced the quality of those chapters: John Amara, Suzanne
Bellemore, Akshat Gupta, Thomas Parker, and Jonathan Steen. We offer our heartfelt
thanks to Brian Herring, who worked tirelessly to create most of the figures that are
included in the book, and to Patty Brown, who undertook many editing duties. Her
knack for picking the right word and ensuring that sentences and paragraphs are
properly structured has resulted in text that is clearer and more readable than it
would have been otherwise. And her organizational skills helped keep the writing
process on track. Lastly, we thank all of the other colleagues at BTEC and KBI Bio-
pharma with whom we’ve worked over the years who have helped enhance our un-
derstanding of all topics related to biopharmaceutical manufacturing and our ability
to teach these topics.

On behalf of the authors,

Gary Gilleskie
Contents
Preface V

Chapter 1
An introduction to biopharmaceutical products 1
1.1 Definitions and background 1
1.1.1 Biopharmaceuticals 1
1.1.2 Recombinant medicines explained 3
1.1.3 Proteins: key ingredients in most biopharmaceutical
products 3
1.1.4 A few final comments regarding terminology 7
1.2 Biopharmaceuticals vs. conventional small-molecule
pharmaceuticals 7
1.3 Biopharmaceutical product classes and examples 10
1.3.1 Protein therapeutics 10
1.3.2 Vaccines 11
1.3.3 Gene therapies 15
1.3.4 Cell therapies 16
1.3.5 Biopharmaceutical impact 17
1.4 Biopharmaceutical drug product: formulation and containers 18
1.5 Biopharmaceutical drug product: quality 20
1.6 Summary 23

Chapter 2
An overview of processes and facilities for biopharmaceutical production 25
2.1 Biopharmaceutical manufacturing processes – an overview 26
2.1.1 The main process 26
2.1.2 Supporting activities 34
2.1.3 Equipment for the main process and support activities 36
2.1.4 Amount of product to be produced 40
2.2 Process design 42
2.3 Manufacturing facilities 50
2.4 Summary 58
2.5 Review questions 60

Chapter 3
Good Manufacturing Practice (GMP) for biopharmaceutical production 64
3.1 GMP regulations – an overview 64
3.2 cGMP regulations/guidelines and guidance documents 65
3.3 Notable cGMP requirements for biopharmaceutical
manufacture 67
X Contents

3.3.1 Personnel 67
3.3.2 Personnel gowning 69
3.3.3 Cleanroom behaviors 70
3.3.4 Equipment 71
3.3.5 Documentation system: procedures, records/reports
and specifications 74
3.3.6 Testing: raw materials, in-process, drug substance, and drug
product 79
3.3.7 Qualification and validation 80
3.3.8 Deviations and corrective and preventive actions (CAPA) 84
3.3.9 Batch disposition 87
3.3.10 Post-marketing surveillance 88
3.4 Summary 89

Chapter 4
Upstream operations 92
4.1 Describing cell growth 92
4.2 Upstream processing and production modes: batch vs. fed batch
vs. continuous 98
4.3 Bioreactor equipment design 103
4.4 cGMP considerations for upstream processing 108
4.5 Upstream process parameters and input material attributes 113
4.6 Upstream performance parameters 115
4.7 Upstream process development 116
4.8 Scaling up fermentation and cell culture 118
4.9 Summary 124
4.10 Review questions 125

Chapter 5
Harvest operations, Part 1: cell lysis 128
5.1 Structure of cells commonly used in biopharmaceutical
production 129
5.2 The need for cell lysis: intracellular vs. extracellular products 133
5.3 Cell lysis methods 137
5.4 High-pressure homogenizers 140
5.5 Homogenizer process and performance parameters 143
5.6 Procedure for a high-pressure homogenization step 147
5.7 Summary 147
5.8 Review questions 149
 Contents XI

Chapter 6
Harvest operations, Part 2: solid-liquid separations by centrifugation 151
6.1 Solid-liquid separations in biopharmaceutical processes 152
6.2 Centrifugation principles 153
6.3 Solid-liquid separation by gravity 154
6.4 Production centrifuges 158
6.4.1 Disc-stack centrifugation: an introduction 159
6.4.2 Disc-stack centrifuges: equipment details 160
6.4.3 Other types of production centrifuges 164
6.5 Disc-stack centrifuge cGMP operating procedure 164
6.6 Disc-stack centrifuge process and performance parameters 166
6.7 Sigma factor, centrifugation step design and scale-up 170
6.8 Summary 173
6.9 Review questions 174

Chapter 7
Harvest operations, Part 3: solid-liquid separations by filtration 177
7.1 Filtration definition 177
7.2 Filtration applications in biopharmaceutical processing 178
7.3 Normal-flow vs. tangential-flow filtration 179
7.4 Solids retention by filters 180
7.5 Depth filtration 181
7.5.1 Depth filters and filter modules 182
7.5.2 Depth filtration equipment and procedures for production 185
7.5.3 Depth filtration parameters, performance and design 187
7.6 Tangential-flow MF overview 195
7.7 Unit operations for solid-liquid separation methods:
a comparison 198
7.8 Other operations for solid-liquid separation 199
7.8.1 Acoustic wave separation 199
7.8.2 Flocculation 199
7.8.3 Magnetic cake filtration 200
7.9 Summary 200
7.10 Review questions 202

Chapter 8
Purification operations: chromatography 206
8.1 Soluble impurities 207
8.2 Chromatographic principles 210
8.3 Creating separation: stationary phases, mobile phases,
and properties of the solutes to be separated 211
8.3.1 Protein properties 212
XII Contents

8.3.2 The most common stationary phase: resins used in packed

beds 213
8.3.3 Other stationary phase formats: membrane adsorbers
and monoliths 218
8.3.4 Mobile phase properties and their impact on separation 220
8.4 Chromatography performance 221
8.5 Operational modes 229
8.6 Typical chromatography procedures for cGMP manufacturing 230
8.7 Chromatography equipment for biopharmaceutical
production 233
8.8 Process and performance parameters for chromatography 240
8.9 Chromatography process design 240
8.10 Validation considerations for chromatography steps 246
8.11 Summary 248
8.12 Review questions 250

Chapter 9
Formulation operations: ultrafiltration 253
9.1 Basis of separation in UF and applications 254
9.2 UF equipment and process configurations for production 256
9.3 TFF membranes and modules 260
9.4 Typical TFF procedures for cGMP manufacturing 265
9.5 Process and performance parameters for tangential-flow UF
and MF 271
9.6 TFF development and scale-up 279
9.7 Validation considerations 284
9.8 Single-pass TFF 286
9.9 Summary 288
9.10 Review questions 289

Chapter 10
Summary and trends in biopharmaceutical processing 292

Glossary 297

References 313

Index 329
Chapter 1
An introduction to biopharmaceutical products

Because this book is focused on the processes and methods for manufacturing bio-
pharmaceuticals, a logical starting point is an in-depth discussion of the output of
these processes – biopharmaceutical drug products – to establish a clear under-
standing of what exactly is being produced. As you will see, biopharmaceuticals are
a class of medicines that is among the most effective and biggest selling in the
world. Their active ingredients are complex biological molecules and components
(e.g., proteins, viruses, and cells) that are used to treat, prevent, and, in some cases,
cure illnesses that medicines produced by chemical synthesis alone often cannot.
Their development has had significant societal impact and marks one of the great
achievements of modern science.
This chapter provides an overview of biopharmaceuticals by addressing the fol-
lowing questions:
– What are biopharmaceuticals, how are they different from more conventional
medicines such as aspirin, and why is it worth making the distinction?
– What are some examples of biopharmaceutical products and the diseases they
treat?
– What exactly are the components of biopharmaceutical drug products, and how
are these products administered to patients?
– What quality attributes are critical for biopharmaceutical drug products?

1.1 Definitions and background

Let’s start by answering the question, “What is a biopharmaceutical?”

1.1.1 Biopharmaceuticals

The term biopharmaceutical has a number of different definitions that depend on

whom you ask. In this book, we take a broad view of biopharmaceuticals and define
them as pharmaceuticals (that is, medicines) inherently biological in nature and man-
ufactured by or from living organisms (including cells from living organisms) [1].
Examples include humans and human cells, animals and animal cells, plants and
plant cells, and microorganisms.
An example of a biopharmaceutical that most are familiar with is insulin, which
is used to treat diabetes. Insulin is a protein hormone produced by the pancreas that
helps to regulate the metabolism of carbohydrates by keeping one’s blood sugar from
getting too high or too low. Diabetes is a disease that occurs when the body does not

https://doi.org/10.1515/9783110616880-001
2 Chapter 1 An introduction to biopharmaceutical products

make enough insulin (Type 1) or does not use it properly (Type 2), resulting in high
blood glucose levels. If glucose levels stay elevated for an extended period, complica-
tions such as heart disease, stroke, high blood pressure, neuropathy (nerve damage),
kidney disease, vision loss, and others may arise [2]. Treatment for type 1 diabetes
requires that a patient take insulin to make up for the insulin the body does not pro-
duce. Treatment for type 2 diabetes may or may not involve taking insulin.
So how is insulin for therapeutic use made? From its first use in the 1920s until
1982, all insulin was animal sourced. Specifically, insulin was extracted from pig
and cattle pancreases and injected into human patients. Insulin produced in this
way saved millions of lives but produced adverse reactions in many human patients.
In addition, only three days’ worth of insulin for a diabetic patient could be prepared
from a single pig pancreas [3]. Then in October 1982, the U.S. Food and Drug Adminis-
tration (FDA) approved an insulin product made in a very different way: Humulin®,
developed by Genentech and marketed by Eli Lilly. Humulin® became the first insulin
analog produced through recombinant deoxyribonucleic acid (DNA) technology –
that is, through genetic engineering – and, notably, the first recombinant medicine
approved for use in the United States. Recombinant technology is discussed in more
detail in the next section. In the case of Humulin®, E. coli was manipulated to pro-
duce human insulin through the addition of a gene (a segment of DNA that codes for
a certain protein) for human insulin [4]. Most insulin (or its analogs) in use today is
produced through recombinant DNA technology and produced in either E. coli or
yeast, such as Saccharomyces cerevisiae. Producing insulin through genetic engineer-
ing allows for greater control over production – and results in fewer adverse reactions
– compared to using pig and cattle pancreases as the insulin source.
Many insulins are now on the market. These products represent one of many
different types of commercially available protein therapeutics that are produced by
genetic engineering of cells. Many more examples are discussed throughout this
chapter.
It is worth noting that some writers limit their definition of biopharmaceuticals to
only those medicines that have been produced through genetic engineering techni-
ques. While genetic engineering has opened up a world of possibilities for production
of biomolecules, the definition of biopharmaceutical used in this book encompasses
all medicines made by or from living organisms; this includes medicines from non-
engineered organisms or cells. Numerous examples exist, but let’s consider one with
which you are likely familiar: flu vaccine. Many types of flu vaccine are on the mar-
ket, and most are produced using natural (i.e., non-genetically engineered) sources.
Vaccines aim to prevent diseases, in contrast to therapeutics such as insulin, which
treat existing diseases. Vaccines often consist of live attenuated or inactivated virus
that resembles the disease-causing virus and stimulates the body’s immune system,
without causing illness, to build defenses against the virus. Consider the specific ex-
ample of Flucelvax®, a seasonal flu vaccine made from four different influenza virus
strains. Each virus strain is grown in Madin-Darby Canine Kidney (MDCK) cells [5].
 1.1 Definitions and background 3

The cells are not genetically engineered; however, because the different virus strains
are propagated in MDCK cells, Flucelvax® fits our definition of biopharmaceutical.
After propagation, each virus strain is inactivated, disrupted, and purified. The four
virus strains are then pooled to produce the vaccine.
We take a broad definition of biopharmaceuticals in this book because, whether
or not genetically engineered cells are used, many similarities exist in manufactur-
ing operations and regulatory expectations for products produced by or from living
organisms. It is worth noting an exception to our definition of biopharmaceuticals.
Medicines based on small molecules that can be produced from cells, such as anti-
biotics, are generally not classified as biopharmaceuticals, even though they seem
to fit our definition, because (1) the active molecule is much smaller than a protein
therapeutic or the antigen that makes a vaccine, for example, and (2) the process for
production, particularly after the culture step, is different from processes used to
produce large molecules such as proteins and larger viruses.

1.1.2 Recombinant medicines explained

Let’s start this discussion by considering DNA. DNA is a macromolecule that consists
of four chemical bases and exists within cells. DNA provides the code – via transcrip-
tion to ribonucleic acid (RNA) and translation of RNA to protein – for protein produc-
tion in living organisms. Recombinant medicines are those produced by recombinant
DNA technology, also referred to as genetic engineering. Recombinant DNA is DNA
that has been combined from multiple sources. Recombinant DNA inserted into a cell
allows the cell to produce a protein that it normally would not. By manipulating DNA
in cells such as E. coli, the genetically modified cells can be grown and used as a kind
of living factory to produce a product, such as a protein like insulin, that it would not
naturally produce. (E. coli is a bacterium and most strains are harmless, despite the
bad press the pathogenic strains give it.) This concept is particularly useful for medi-
cines based on large and complex molecules because chemical synthesis of these mol-
ecules can be cost prohibitive or just not possible with current technology.

1.1.3 Proteins: key ingredients in most biopharmaceutical products

The majority of biopharmaceuticals are either proteins – such as insulin, or protein

based – such as flu vaccines made from influenza virus, which has significant pro-
tein content. Given their prevalence, let’s consider proteins in more depth.
Proteins are macromolecules composed of amino acids. In humans, 20 amino
acids are used in protein synthesis. As shown in Figure 1.1, an amino acid consists of
an α carbon in the center, to which is attached an amine group, a carboxyl group, a
hydrogen atom, and side chain designated as “R,” that is different for each amino acid.
4 Chapter 1 An introduction to biopharmaceutical products

Amine group Carboxyl group

α carbon Organic
side chain

Figure 1.1: Chemical structure of an amino acid, the building block for proteins. Note that at neutral
pH, the carboxyl group is deprotonated and carries a negative charge, while the amine group is
protonated and carries a positive charge. Image © NC State University; reprinted with permission.

Figure 1.2 shows R groups for the 20 different amino acids. R groups have different
properties – for example, some are positively charged, some negatively charged, some
are polar (soluble in water) but not charged, and some are nonpolar (limited solubility
in water). Differences in R groups among amino acids mean different proteins have dif-
ferent properties, such as charge and hydrophobicity. These properties can be exploited
in the design of a manufacturing process, particularly for the separation of product
from impurities. This topic is discussed in more detail in Chapter 8.
To form proteins, amino acids combine through the reaction of the amine group
of one amino acid with the carboxyl group of another. The reaction is referred to as
a condensation reaction because water is formed as a product. Within cells, conden-
sation reactions are catalyzed by ribosomes. The resulting bond is referred to as a
peptide bond. As more amino acids react, the polymer becomes larger and a protein
is formed. Proteins have a characteristic three dimensional structure. To describe
protein structure, biochemists typically refer to four different levels – primary, sec-
ondary, tertiary, and quaternary – each defined below and illustrated in Figure 1.3,
using insulin as an example:
– Primary structure refers simply to the sequence of amino acids in the protein.
– Secondary structure refers to regularly repeated local structures created by hy-
drogen bonds on nearby amino acids. The result of this hydrogen bonding is the
formation of alpha helices and beta strands and sheets.
– Tertiary structure refers to the overall three-dimensional structure of a single
protein molecule.
– Quaternary structure is the structure that results from the interaction of more
than one protein molecule.
 1.1 Definitions and background 5

Charged amino acids

Aspartic Acid Glutamic Acid Arginine Histidine

Asp/D Glu/E Arg/R His/H

Polar amino acids

Lysine Asparagine Cysteine Glutamine

Lys/K Asn/N Cys/C Gln/Q

Nonpolar amino acids

Serine Threonine Tyrosine Alanine

Ser/S Thr/T Tyr/Y Ala/A

Glycine Isoleucine Leucine Methionine

Gly/G Ile/I Leu/L Met/M

Phenylalanine Proline Tryptophan Valine

Phe/F Pro/P Trp/W Val/V

Figure 1.2: Side chains of the 20 amino acids that make up proteins in humans. Image © NC State
University; reprinted with permission.
6 Chapter 1 An introduction to biopharmaceutical products

(a)

A-chain

B-chain

(b) (c)
N C
C

Figure 1.3: The primary, secondary, tertiary, and quaternary structure of human insulin. (a) is the
primary structure and shows the 51 amino acids that make up insulin along with its three disulfide
bridges. (b) is a ribbon diagram illustrating the secondary and tertiary structure, with the A chain
(blue) forming two short helices, and the B chain (green) a helix and a strand structure. (c) is the
quaternary structure, a hexameric form of made up of three insulin dimers. Insulin is stored within
the cells as hexamer. (Note that each of the six insulin monomers is shown in a different color.)
Image (a) © NC State University; reprinted with permission. Image (b) by PDB-101 (PDB101.rcsb.org)
at https://pdb101.rcsb.org/global-health/diabetes-mellitus/drugs/insulin/insulin, used under
CC-BY-4.0 license (https://creativecommons.org/licenses/by/4.0/legalcode). Image (c) Benjah-bmm27,
Public domain, via Wikimedia Commons.

Three-dimensional structure is relevant to biopharmaceutical manufacturing because

protein structure is critical to its function. If that structure is disrupted during process-
ing, it is likely that the protein will not function as intended. Loss of secondary, tertiary,
and/or quaternary structure is referred to as denaturation.
To avoid denaturing a biopharmaceutical product, it is important to know the
causes for denaturation. A short list is given below:
– Heat. For many proteins, thermal denaturation begins at approximately 40 °C [6],
due primarily to the breakdown of hydrogen bonds.
– High or low pH. Acidic and basic conditions impact the charge of amino acids that
make up the protein. A change in charge of the amino acid side chains can disrupt
hydrogen bonding and salt bridges, both of which impact the three-dimensional
structure of the protein molecule.
 1.2 Biopharmaceuticals vs. conventional small-molecule pharmaceuticals 7

– Shear. Shear can be a cause of denaturation, particular when a gas/liquid inter-

face is present [7].
– Certain chemical agents. Chaotropic agents destabilize proteins by disrupting
hydrogen bonding between water molecules that surround a protein in solution
or by directly interacting with the protein [8]. Examples include ethanol, guani-
dine hydrochloride, and urea.

It is worth noting that the terms protein and peptide are sometimes used interchange-
ably; however, there is a difference: peptides, also made up of a chain of amino
acids, are short proteins. The U.S. FDA defines a peptide as having 40 or fewer amino
acids [9].

1.1.4 A few final comments regarding terminology

In these first few sections, a number of definitions and clarification of terms have
been presented. Before moving on, we want to clarify a couple of other terms used
throughout the book.
The term biological product is one that is commonly used in a regulatory con-
text. In this text, the terms biopharmaceutical and biological product are used syn-
onymously. According to the U.S. FDA, biological products “can be composed of
sugars, proteins, or nucleic acids or complex combinations of these substances, or
may be living entities such as cells and tissues. Biologics are isolated from a variety
of natural sources – human, animal, or microorganism – and may be produced by
biotechnology methods and other cutting-edge technologies.” Further, “biological
products include a range of products such as vaccines, blood and blood components,
allergenics, somatic cells, gene therapy, tissues, and recombinant therapeutic pro-
teins” [10]. This definition and the examples provided align with the definition of bio-
pharmaceutical used in this book.
In addition, the term drug is often used to describe medicines that are processed
using chemical and not biological methods. In this text, we use the term drug in a more
general sense, to include both small-molecule pharmaceuticals produced by chemical
synthesis as well as biopharmaceuticals/biological products.

1.2 Biopharmaceuticals vs. conventional small-molecule

pharmaceuticals
Many of the most commonly used medicines in the world are made via chemical syn-
thesis (think aspirin) rather than through biological methods. However, if we relied
solely on chemical synthesis for pharmaceutical production, the medicines with large,
relatively complex active ingredients such as insulin or flu vaccine, would likely not
8 Chapter 1 An introduction to biopharmaceutical products

exist. As the size of these active components (e.g., proteins) become larger and more
complex, chemical synthesis becomes more challenging, particularly at the scales re-
quired for commercial production, or even impossible as mentioned previously.
In this section, we take a deeper look at biopharmaceuticals by comparing them
to the more familiar small-molecule pharmaceuticals that are produced via chemi-
cal synthesis. A number of important differences exist, as described below.
1. Biopharmaceuticals are large, complex molecules, viruses, or even cells, while
more traditional pharmaceuticals are small molecules with well-defined structures.
Consider an example of each. Aspirin, shown in Figure 1.4, which is produced
by chemical synthesis, is one of the most widely used drugs in the world. It pro-
vides relief from pain, fever, and inflammation caused by a variety of conditions.
Humira®, a biopharmaceutical, is the largest selling (based on revenue) drug in
the world [11]. It is a human immunoglobulin G of subclass 1 (IgG1) monoclonal
antibody, with a structure similar to that shown in Figure 1.4, and is used to treat
rheumatoid arthritis, among a number of autoimmune diseases. It has a molecu-
lar weight of approximately 148,000 Da [12] and contains over 20,000 atoms; in
comparison, the molecular weight of aspirin is only 180 Da and consists of 21
atoms [13]. Table 1.1 shows a comparison of sizes of some well-known small-mol-
ecule drugs with biopharmaceuticals.
2. Biopharmaceuticals are produced using relatively complex processes that are bio-
logical in nature. Small-molecule pharmaceuticals like aspirin are most often
produced by chemical synthesis. We have already discussed that the biophar-
maceuticals Humulin® and Flucelvax® are produced using E. coli and MDCK)
cells, respectively. Humira® is produced by large-scale culture of genetically en-
gineered Chinese hamster ovary (CHO) cells [14]. Again, the complexity of the
molecules that are the active ingredient in biopharmaceuticals necessitate that
they be produced by or from living organisms or cells of living organisms – be-
cause chemical synthesis of such complex structures is neither economically vi-
able nor, in many cases, currently possible.
3. Biopharmaceuticals are largely administered to patients parenterally, [15] which
literally means other than through the gastrointestinal tract. They are not taken
orally. Common parenteral routes of administration for biopharmaceuticals are
intravenous, subcutaneous, and intramuscular injection. Why parenteral deliv-
ery? Because most biopharmaceuticals are proteins or contain protein compo-
nents as their active ingredients. If administered orally, these proteins would be
digested in the stomach due to low pH and/or digestive proteases, thereby ren-
dering the biopharmaceutical ineffective. In addition, their large size would
make transport of the molecule across the intestinal wall difficult [16]. Small-
molecule drugs, on the other hand, are primarily administered orally – as a tablet,
capsule, or liquid – but also topically and parenterally. All parenteral drugs, in-
cluding biopharmaceuticals, must be sterile because the parenteral route of drug
delivery bypasses the body’s natural defenses [17].
 1.2 Biopharmaceuticals vs. conventional small-molecule pharmaceuticals 9

Other differences between biopharmaceutical and traditional small-molecule drugs

are that biopharmaceuticals are typically more immunogenic – that is, likely to cre-
ate an unwanted immune response – and less stable than small-molecule drugs.

Figure 1.4: A comparison of an IgG1 monoclonal antibody, like Humira®, to aspirin, a small-
molecule, chemically synthesized pharmaceutical. Note that the IgG1 model shows the antibody
heavy chains in blue and light chains in red. In the aspirin representation, white represents
hydrogen, black carbon, and red oxygen. IgG1 image by molekuul © 123RF.com. Aspirin image,
© NC State University, includes ball-and-stick illustration by molekuul © 123RF.com.

Table 1.1: A size comparison of some well-known small-molecule drugs to biopharmaceuticals.

Product Active ingredient Biopharmaceutical Molecular weight Method of

example or small molecule (Da)a,b productionb

Aspirin Acetylsalicylic acid Small molecule  Chemical synthesis

Tylenol ®
Acetaminophen Small molecule  Chemical synthesis

Lipitor® Atorvastatin Small molecule , Chemical synthesis

calcium trihydrate

Humulin® Insulin Biopharmaceutical , Produced by

growing genetically
modified E. coli

Humira® Anti-TNF monoclonal Biopharmaceutical , Produced by

antibody growing genetically
modified CHO cells

Novoeight® Factor VIII Biopharmaceutical , (does Produced by

not include post growing genetically
translational engineered CHO
modifications) cells

Flucelvax® Subunits of Biopharmaceutical Subunits of Produced by

Quadrivalent influenza virus  nm virus propagating
particles influenza virus in
MDCK cells
10 Chapter 1 An introduction to biopharmaceutical products

Table 1.1 (continued)

Product Active ingredient Biopharmaceutical Molecular weight Method of

example or small molecule (Da)a,b productionb

Luxturna® Adeno-associated Biopharmaceutical ~,, [] Produced by

virus serotype  transfection of
(AAV) genetically HEK cells with
modified to express plasmid DNA
the human RPE
gene

Kymriah® T cells Biopharmaceutical  µm T cells [] Produced by

genetically
modifying T cells
a
For the small-molecule drugs, molecular weight values were obtained using the PubChem
database [20].
b
For the biopharmaceuticals, method of production taken from their package inserts. For Humira® and
Novoeight®, molecular weight was also taken from their package inserts [5, 12, 21–23].

1.3 Biopharmaceutical product classes and examples

There are a variety of different types of biopharmaceuticals, beyond the examples

presented up to now. They are powerful medicines that treat, prevent, or cure numer-
ous diseases. The majority of biopharmaceuticals can be classified broadly as protein
therapeutics, vaccines, gene therapies, and cell therapies. Each is considered in greater
detail below.

1.3.1 Protein therapeutics

As the name implies, protein therapeutics are proteins that treat an existing disease.
Insulin and Humira®, both discussed previously, are two examples. Why do pro-
teins make good medicines? The thousands of proteins in your body [24] serve a va-
riety of functions. The many biological activities carried out by proteins lead to a
number of ways they can be used for therapeutic benefit [25], including:
– Replacing a protein that is lacking or abnormal. Insulin for treatment of diabetes
is an example; a variety of insulin-based biopharmaceuticals, such as Humulin®
and Lantus®, are commercially available. Factor VIII for treatment of hemophilia
A – a disease in which blood does not clot normally, caused by a deficiency of the
protein clotting factor Factor VIII – is another example; a number of different Fac-
tor VIII-based biopharmaceuticals are on the market, such as Eloctate®.
 1.3 Biopharmaceutical product classes and examples 11

– Augmenting an existing biological pathway. Neulasta® is an example. It is a granu-

locyte colony-stimulating factor (G-CSF) manufactured by recombinant DNA tech-
nology. In the human body, G-CSF causes neutrophils, a type of white blood cell
that fights infection, to grow in the bone marrow. In patients being treated with
chemotherapy or radiation, neutrophils may stop being produced, which increases
a patient’s risk of infection. Neulasta® stimulates production of neutrophils thereby
helping patients undergoing chemotherapy or radiation to fend off infection [26].
– Interfering with a molecule or organism. Avastin® is an example. It is a recombi-
nant humanized IgG1 antibody produced in CHO cells that is used to treat various
cancers. The antibody binds vascular endothelial growth factor (VEGF). In the
human body, VEGF is involved in the formation of new blood vessels in cancer
cells, allowing them to grow and metastasize. To work, it must bind to receptors
on cells; however, because the Avastin® antibody binds VEGF, VEGF cannot bind
to the cell receptors, which inhibits formation of new blood vessels, thereby inhib-
iting cancer development and growth [27].
– Providing a function that does not normally exist. Botox® is an example. It is a botu-
linum toxin type A, produced from fermentation of Hall strain Clostridium botulinum
type A. It is used to treat a number of conditions, including chronic migraines and
upper limb spasticity. It is also used for cosmetic purposes. It “blocks neuromuscular
transmission by binding to acceptor sites on motor or autonomic nerve terminals,
entering the nerve terminals, and inhibiting the release of acetylcholine” [28].
– Delivering other compounds. Antibody-drug conjugates – a monoclonal antibody
covalently linked to a small, cytotoxic compound – are an example. The monoclonal
antibody binds to specific receptors found on certain types of cells, such as cancer
cells. The linked drug enters these cells and kills them without harming other cells.

Protein therapeutics can be classified according to the type/function of the protein

as shown in Table 1.2.

1.3.2 Vaccines

A vaccine is a biopharmaceutical used to prevent a disease by improving immunity to

that disease. It typically contains an agent that either: resembles a disease-causing mi-
croorganism or virus, resembles a portion of the microorganism or virus, or has the
ability to produce a part of the disease-causing microorganism or virus. The agent (or
its action) stimulates the body’s immune system to recognize and destroy the disease-
causing microorganism or virus if encountered again. Vaccines take many forms, as
shown in Table 1.3, including viral (based on attenuated – i.e., weakened – virus, inac-
tivated virus, or fractions of a virus), microbial (usually an attenuated microorganism
or inactivated toxin from a microorganism), virus-like particles, recombinant proteins,
polysaccharides, DNA, and RNA.
Table 1.2: Different classes of protein therapeutic biopharmaceuticals.
12

Class Description Examples of Commercial Products/

Company/Primary Indication

Blood clotting factors Proteins in the blood, such as Factor VIII and Factor IX, that Eloctate®/Bioverative/Hemophilia A
control bleeding. Most are for treatment of hemophilia A or B.

Growth factors Molecules that simulate cell growth. Common examples Epogen®/Amgen/anemia
include erythropoietin (EPO), a protein that plays a role in Neulasta®/Amgen/Reduces risk of infection during
production of red blood cells, and colony-stimulating factor chemotherapy
(CSF), which promotes the growth and differentiation of stem
cells. Numerous EPO-based biopharmaceuticals are available
for the treatment of anemia. There are also numerous CSF-
based biopharmaceuticals for the treatment of low white
blood cell count.

Hormones Molecules that serve as messengers, controlling and – Humulin®/Eli Lilly/Diabetes

coordinating activities throughout the body. Insulin, human – Lantus®/Sanofi/Diabetes
growth hormone, and follicle-stimulating hormones are – Norditropin®/Novo Nordisk/ Short stature, growth
protein hormones produced as biopharmaceuticals. hormone deficiency

Interferons, interleukins, Relatively small proteins that impact immune response. Plegridy™/Biogen/Multiple sclerosis
Chapter 1 An introduction to biopharmaceutical products

tumor necrosis factors Among this group, interferon-based biopharmaceuticals are

the most common. They are used to treat hepatitis C and
multiple sclerosis among other diseases.
Monoclonal antibodies Proteins that have a high affinity for the same epitope on an – Humira®/AbbVie/Rheumatoid arthritis, numerous othersa
and monoclonal antibody- antigen. They work through multiple mechanisms in providing – Keytruda®/Merck/Melanoma, lung cancer, various other
based products treatment for diseases. This category also includes cancersa
monoclonal antibody-based products such as antibody-drug – Opdivo®/Bristol-Myers Squibb/Metastatic melanoma,
conjugates and bispecific antibodies. Many of the world’s other cancersa
largest selling drugs are monoclonal antibodies. – Avastin®/Genentech (Roche)/Metastatic colorectal
Antibody-drug conjugates (ADCs) are monoclonal cancers, other cancersa
antibodies covalently linked to a small, cytotoxic compounds. – Rituxan®/Genentech (Roche),Biogen/Non-Hodgkin’s
The monoclonal antibody binds to specific receptors found on lymphoma, numerous othersa
certain types of cells, such as cancer cells. The linked drug – Stelara®/Johnson & Johnson/Plaque psoriasis, psoriatic
enters these cells and kills them without harming other cells. arthritis, othersa
All currently approved ADCs are cancer treatments. – Herceptin®/Genentech (Roche)/Breast cancera
Bispecific antibodies are antibodies that can bind to two
specific antigens at the same time.

Therapeutic enzymes This category represents a variety of recombinant protein Kyrstexxa®/Savient Pharmaceuticals/Gout
enzymes – molecules that catalyze reactions.

Thrombolytic agents Drugs that dissolve clots; many are based on the protein Activase®/Genentech (Roche)/Acute ischemic stroke,
tissue plasminogen activator. Most are for treatment of heart myocardial infarction
attacks, strokes, and other conditions caused by blood
clotting.

Fusion proteins Combined proteins created by combining genes originally Eylea®/Regeneron/Macular degenerationa
coding for separate proteins. Frequently, one partner has a Enbrel®/Amgen/Rheumatoid arthritis, numerous othersa
molecular recognition function while the other may impart
added stability or novel targeting.
a
Drugs among the top 15 biggest selling (based on revenues) in the world in 2019 [11].
1.3 Biopharmaceutical product classes and examples
13
14 Chapter 1 An introduction to biopharmaceutical products

Table 1.3: Examples of vaccines by type. Indication and host information taken from package
inserts for each vaccine.

Type Commercial For the Prevention of Host (Production

Example System)

Viral

Live attenuated FluMist® [] Flu Chicken eggs

®
Inactivated Ipol [] Polio Vero cells
®
Fractional Flucelvax [] Flu MCDK cells

Microbial

Live attenuated Vaxchora™ Cholera V. cholerae

[]

Toxin/Toxoid Daptacel® Diphtheria, tetanus, C. diphtheriae, C. tetani,

[] pertussis B. pertussis

Virus-like particle Gardasil® Diseases by human S. cerevisiae

[] papillomavirus

Recombinant protein Recombivax Hepatitis B S. cerevisiae

HB® []

Polysaccharide/ Conjugated Prevnar ®a Diseases caused by S. S. pneumoniae,

Polysaccharide [] pneumoniae C. diphtheriae

DNA, RNA, and Recombinant In Various Various

Vector Vaccine development
a
Among the top 15 biggest selling drugs (based on revenue) in the world in 2019 [11].

Numerous vaccines are available in addition to the flu vaccine discussed previously. Ex-
amples are given in Table 1.3. You are likely familiar with at least some of these from
vaccinations that you or loved ones have received. Vaccines have gotten a lot of atten-
tion recently in light of the coronavirus disease 2019 (COVID-19) pandemic. As of the
writing of this book, two vaccines have been authorized for use against COVID-19 in the
United States. Both are messenger RNA vaccines, a relatively new type of vaccine that
does not rely on a weakened form of the virus [36].
Messenger RNA provides a blueprint to produce proteins in cells through the
translation process mentioned in the previous discussion on recombinant medicines.
Viruses contain a protein coat that surrounds their genetic material. If the virus’s
genome – that is, the hereditary information in the form of DNA or RNA that is respon-
sible for producing the protein coat – is known, then genes can be determined that en-
code for the specific virus proteins. Once these genes are known, messenger RNA
sections can be designed that, along with our own cells, produce a viral protein. The
immune system recognizes the protein and mounts an immune response that builds
 1.3 Biopharmaceutical product classes and examples 15

the body’s defenses against the pathogen, like the SARS-CoV-2 virus that causes
COVID-19. One advantage to RNA vaccines is the relatively short time for development.
Note that while most vaccines are preventive, some vaccines are therapeutic. In
therapeutic vaccines, antigens associated with an illness are introduced to a patient
to stimulate the body’s immune system to fight an illness that is present, hence use
of the term therapeutic. Because they work to stimulate a patient’s immune system,
these therapies are referred to as vaccines.

1.3.3 Gene therapies

Gene therapy is a technique that treats or cures diseases by inserting a gene into a pa-
tient’s cells instead of using drugs or surgery. The inserted gene replaces a disease-
causing gene or produces a product, such as a protein, that treats the disease. Genes
are typically delivered to patients using viruses that act as vectors, such as adenovirus,
lentivirus, or adeno-associated virus. Viruses make good vectors – a “vehicle” used to
transport DNA into a host cell – because they have the ability to deposit DNA or RNA
into cells by infection. Gene therapy offers the potential to treat a range of disorders
that result from defects in a single gene (i.e., monogenic disorders); these disorders
include sickle cell anemia, cystic fibrosis, hemophilia, Parkinson’s, hypercholesterol-
emia, and Gaucher disease. It also offer the potential to treat polygenic disorders
such as heart disease, cancer, diabetes, and infectious diseases such as HIV.
Broadly, gene therapies may be classified as in vivo, in which a gene is trans-
ferred to cells inside the body, and ex vivo, in which patient cells are harvested and
cultivated in the laboratory. A gene is then transferred to the cultivated cells, and the
cells with the new genetic information are harvested and transplanted back into the
patient from whom they were derived. The diagram in Figure 1.5 illustrates the differ-
ences in the two therapies.

In Vivo Ex Vivo
Cells are taken
from the patient

Genes are
transferred into Genes are transferred
cells while still
in the patient cells expanded

back into the patient

Figure 1.5: In vivo versus ex vivo gene therapy. Image © NC State University; reprinted with
permission.
16 Chapter 1 An introduction to biopharmaceutical products

A list of all U.S. FDA-approved gene therapy products as of August 2020 is provided
in Table 1.4.

Table 1.4: In vivo and ex vivo gene therapies currently approved by the U.S. FDA [37]. Information
for each taken from package inserts..

Product Company Indication Type Vector Approval

Year

Tecartus™ [] Kite Pharma Relapsed or Ex vivo Retrovirus used to 

refractory mantle cell genetically modify
lymphoma a patient’s T cells

Zolgensma® [] Novartis Spinal muscular In vivo AAV 

atrophy (SMA) with
bi-allelic mutations
in the survival motor
neuron  (SMN)
gene, patients less
than  years old

Luxturna® [] Spark Retinal dystrophy In vivo AAV 

Therapeutics

Kymriah® [] Novartis B-cell precursor Ex vivo Lentivirus used to 

acute lymphoblastic genetically modify
leukemia and large a patient’s T cells
B-cell Lymphoma

Yescarta® [] Kite Pharma Relapsed or Ex vivo Retrovirus used to 

refractory large B- genetically modify
cell lymphoma a patient’s T cells

1.3.4 Cell therapies

Cell therapy refers to the use of living, whole cells for treatment of a disease. A variety
of cells are used including T cells and various stem cells. Cell therapies have the po-
tential to treat numerous disorders, including various cancers, autoimmune diseases,
urinary problems, and infectious diseases. Cell therapies can be categorized accord-
ing to whether cells are taken from and administered to the same individual – re-
ferred to as autologous cell therapy (this includes ex vivo gene therapies described
previously) – or derived from a donor or multiple donors other than the patient who
receives the cells – referred to as allogeneic cell therapy.
Kymriah® and Yescarta®, described in Table 1.4, are classified as autologous cell
therapies as well as ex vivo gene therapies. They are both prepared from a patient’s
own T cells, which are key to immune response in humans. A gene is introduced ex
 1.3 Biopharmaceutical product classes and examples 17

vivo into the T cells using a viral vector, such as lentivirus, that expresses a chimeric
antigen receptor (CAR). The CAR on the surface of the T cells helps them find and kill
cancer cells more effectively than if the CAR is not present. This type of therapy is
referred to specifically as CAR-T cell therapy.
If you are interested in reading more about the products listed or other biophar-
maceuticals, perform an Internet search for “full prescribing information” for the
product of interest. The resulting document contains a plethora of information on
the product, including indications, contraindications, dosage and dosage forms, a
general product description, basic information on the process used for production,
the drug’s mechanism of action, and results from clinical studies.

1.3.5 Biopharmaceutical impact

As you can see from Tables 1.2–1.4, the health benefit of biopharmaceuticals is sig-
nificant, with biopharmaceuticals treating a variety of diseases, including a number
of cancers, stroke, heart attack, multiple sclerosis, flu, hepatitis, rheumatoid arthri-
tis and a host of other autoimmune diseases, and various eye diseases.
The biopharmaceutical examples listed in the previous tables represent only a
fraction of all biopharmaceuticals approved for marketing throughout the world. In
the United States and European Union alone, 316 genetically engineered biophar-
maceuticals have active licenses as of July 2018. (Note that the reference from which
this information is taken defines biopharmaceuticals as recombinant proteins, in-
cluding recombinant antibody-based products, nucleic acid-based and genetically
engineered cell–based products. It does not include vaccines not produced through
genetic engineering or proteins purified from human plasma.) Among all new drugs
approved in the United States between January 2014 and July 2018, 47% were bio-
pharmaceuticals, with monoclonal antibodies the dominant class. A look at bio-
pharmaceutical approvals by the U.S. FDA and European Medicines Agency (EMA)
from January 2015 to July 2018 shows that 53% of first-time approvals in the United
States and European Union were for monoclonal antibody products. Hormones and
clotting factors also had a significant number of approvals during this period, and in
vivo and ex vivo gene therapies increased in number. Based on products in late-stage
clinical trials, it appears that monoclonal antibody-based products will continue to
dominate new biopharmaceutical approvals. It is also worth noting that approval of
biosimilars – a biological product highly similar to and interchangeable with an ap-
proved biological product – is rapidly increasing, with 52 approved in the United
States and European Union since 2006 [41].
In addition to the health benefits, economic impact of biopharmaceuticals is signif-
icant. In 2019, 10 of the top 15 selling medicines worldwide were biopharmaceuticals
[11], each with annual sales in excess of US$5 billion and combined sales of more than
US$85 billion. Among the 10, seven are monoclonal antibodies, two are fusion products
18 Chapter 1 An introduction to biopharmaceutical products

each of which includes a portion of an antibody, and one is a vaccine. A recent study
estimates the direct economic impact (revenues from biopharmaceutical businesses) of
the biopharmaceutical industry in the United States in 2017 to be US$560 billion [42].

1.4 Biopharmaceutical drug product: formulation and containers

To understand the processes involved in manufacturing biopharmaceuticals, it is

worth starting at the end of the process by considering the finished biopharmaceutical
product. The term drug product is used to describe this finished dosage form that con-
tains the active ingredient (i.e., the drug that is administered to the patient) [43]. Gener-
ally speaking, drug products can take a variety of forms, including tablets, capsules,
solutions, and lyophilized (also referred to as freeze-dried) powder. For the majority of
biopharmaceuticals, drug product is a liquid solution or lyophilized powder. The liquid
solution is made up of the active ingredient dissolved in water and formulated with a
number of other ingredients. Lyophilization is a process by which the biopharmaceuti-
cal is dried – that is, the water is removed – at very low temperature and under vac-
uum. If heat were used to evaporate the water, the product might be denatured or
degraded. Lyophilized drug products are powders. To be administered parenterally,
as most biopharmaceuticals are, the drug product must be in a liquid state, which
means lyophilized products must be reconstituted with water prior to injection.
So why are some biopharmaceuticals lyophilized while others are liquid solu-
tions? The answer relates to product stability. Biopharmaceuticals are prone to un-
dergo chemical and/or physical changes over time; for example, a biopharmaceutical
may denature, aggregate, or undergo chemical degradation while the drug product is
in storage. Generally speaking, biopharmaceuticals are more stable as lyophilized
products than as liquid solution [44]. Stability is critical at the drug product stage, as
most biopharmaceutical drug products are stored prior to use and commonly have a
shelf life for at least two years at 4 °C [45]. When liquid solution cannot provide the
drug product stability needed for shelf life, then lyophilization is necessary. Images
of liquid and lyophilized products in vials are shown in Figure 1.6.

(a) (b)

Figure 1.6: (a) Lyophilized (freeze-dried) and

(b) liquid solution forms of biopharmaceutical
drug product. Photo © NC State University;
reprinted with permission.
 1.4 Biopharmaceutical drug product: formulation and containers 19

While details of biopharmaceutical stability are beyond the scope of this book, it
should be noted that a main objective of drug product formulation is maintaining
chemical and physical stability of the product. When a drug product liquid formula-
tion can be developed that provides the necessary product stability, then liquid
form is desirable because it can be administered directly to a patient, without recon-
stitution, and enables a wider range of delivery devices to be used. Drug delivery
devices are designed to create a patient-friendly delivery method that supports pa-
tient compliance with the prescribed regimen and to accommodate the broad dos-
age range typical of biopharmaceutical products. Historically, biopharmaceutical
drug products have been delivered intravenously, with the product contained in
vials, as shown in Figure 1.6, and administered with a sterile needle/syringe or by
intravenous infusion. While intravenous administration offers the advantage of 100%
bioavailability (the percentage of drug that enters systemic circulation and is there-
fore able to have the intended effect), it is inconvenient as it typically requires admin-
istration by medical personnel and is painful.
A growing number of liquid formulations are becoming available, particularly
for use in prefilled syringes for subcutaneous administration and autoinjectors,
which are also prefilled devices but are equipped with a spring-loaded syringe that
delivers a dose of medication when forced against the body [46]. These prefilled syrin-
ges and autoinjectors are convenient and effective for self-administration of the drug,
which better enables patients to comply with their drug prescription. For example,
Enbrel® is available in a prefilled syringe or autoinjector and continues to be avail-
able as a lyophilized powder in a vial for reconstitution [47]. Much work is happening
in the field of drug delivery to engineer more patient-centric delivery methods,
thereby improving patient compliance with the prescription regimen. For a good
review, see Anselmo et al. [48].
A variety of components is used to formulate biopharmaceutical drug product.
Generally speaking, beyond water and the active ingredient, components are cho-
sen to provide product stability and clinical efficacy. Components that are not the
active ingredient are referred to as excipients. Typical components of biopharma-
ceutical drug product in their liquid form include the following [49, 50]:
– WFI (water for injection). A type of purified water used to formulate parenteral
products. It is discussed in greater detail in the next chapter.
– The active biomolecule.
– Buffers. These are agents used for pH control such as phosphate and acetate.
– Product-stabilizing agents. Nonionic surfactants such as polysorbate 20, polysor-
bate 80, and poloxamer 188 are commonly used. They serve to stabilize proteins
by reducing interfacial stresses that may result in product aggregation [50]. Sugars
such as sucrose, trehalose, maltose, and lactose are also commonly added to en-
hance product stability.
20 Chapter 1 An introduction to biopharmaceutical products

– Tonicity agents. They are added to increase the concentration of solutes to en-
sure that the formulation has the same osmotic pressure as cells. Sugars such as
sucrose and polyols such as sorbitol are used.
– Antioxidants. These are added to inhibit product oxidation. The amino acid me-
thionine is an example.
– Preservatives. Benzyl alcohol is an example.
– Product- and process-related impurities (at low levels). These are soluble compo-
nents not intended to be part of drug product. Impurities are discussed in greater
detail in Chapter 8.

As an example, a list of the components in Avastin® drug product is shown in

Figure 1.7. Avastin® is a humanized IgG1 antibody, as described previously, with
a liquid formulation. Avastin® is manufactured by Genentech, a member of the
Roche group.

Components of Avastin drug product

• Water for injection (WFI)
• mAb at a concentration = 25 mg/mL
100 mg/4 mL single vial or
400 mg/16 mL single vial
For the 100 mg vial:
• Sodium phosphate (monobasic, monohydrate) (23.2 mg)
• Sodium phosphate (dibasic, anhydrous) (4.8 mg)
• α, α-trehalose dehydrate (240 mg)
• Polysorbate 20 (1.6 mg)

Figure 1.7: Components of Avastin® drug product [27]. Image © NC State University; reprinted with
permission.

Note that while most biopharmaceutical drug products are liquid solution (like Avastin)
or freeze dried – Eloctate® is an example – some are not. For example, Kymriah®, the
ex vivo gene therapy described previously, looks very different from the final product
images in Figures 1.6 or 1.7. It comes as a frozen cell suspension in an infusion bag and
is intravenously infused into patients.

1.5 Biopharmaceutical drug product: quality

An essential characteristic of drug products is quality . But what is meant by quality?

The guidance document from the International Conference on Harmonization (ICH),
ICH Q6A, Specifications: Test Procedures and Acceptance Criteria for New Drug
 1.5 Biopharmaceutical drug product: quality 21

Substances and New Drug Products: Chemical Substances (October 1999), defines it
as follows: “The suitability of either a drug substance or drug product for its intended
use. This term includes attributes such as identity, strength, and purity” [51].
Drug substance and drug product are terms frequently used throughout this book.
So let’s differentiate between them. We’ve already defined drug product as the finished
dosage form – tablet, capsule, or in the case of biopharmaceuticals, a liquid solution or
lyophilized product – that contains the active ingredient. Drug substance is the active
ingredient that is intended to furnish pharmacological activity or other direct effect in
the diagnosis, cure, mitigation, treatment, or prevention of disease [43]. It may have
additional excipients added to it to create drug product. We elaborate on drug sub-
stance versus drug product from a processing perspective in the next chapter.
As part of the process of developing drug products, a quality target product pro-
file (QTPP) is needed. From the document ICH Q8(R2): Pharmaceutical Development
(November 2009), the QTPP is: “[a] prospective summary of the quality characteris-
tics of a drug product that ideally will be achieved to ensure the desired quality,
taking into account safety and efficacy of the drug product” [52]. Note that the use
of the term prospective in that definition sets the expectation that the QTPP is to be
defined at the outset of the development process. Components of the QTPP include
[52]:
– Intended use in clinical setting;
– Route of administration, dosage form, delivery systems;
– Dosage strength(s);
– Container closure system;
– Therapeutic moiety release or delivery and attributes affecting pharmacokinetic
characteristics appropriate to the drug product dosage form being developed;
and
– Drug product quality criteria (e.g., sterility, purity, stability and drug release)
appropriate for the intended marketed product.

In addition to the QTPP, a list of critical quality attributes (CQAs) is developed.

Again from ICH Q8(R2), the definition of a CQA is as follows: “a physical, chemical,
biological or microbiological property or characteristic that should be within an ap-
propriate limit, range, or distribution to ensure the desired product quality.” CQAs
are generally associated with the drug substance, excipients, intermediates (in-pro-
cess materials), and drug product [52].
When discussing CQAs in this book, we are most often referring to attributes of
the drug substance and drug product. CQAs are developed from a variety of sources –
the QTPP, prior product knowledge, scientific literature, etc. – and vary from product
to product. We’ve already discussed at least one CQA common to biopharmaceuti-
cal drug products, sterility, which is critical to ensuring safety of a parenteral drug.
There are numerous others to consider. Generally speaking, they can be categorized
as follows:
22 Chapter 1 An introduction to biopharmaceutical products

– General quality
– Identity
– Quantity
– Potency
– Purity
– Safety

Biopharmaceutical processes must be designed and executed to ensure the CQAs

are consistently met. And how the manufacturing steps are conducted is as impor-
tant as what the steps are. Biopharmaceutical products are manufactured using
Good Manufacturing Practice (GMP) [53, 54]. GMP are regulations enforced by regu-
latory authorities throughout the world that assure proper design, monitoring, and
control of manufacturing processes and facilities, thereby ensuring that products
are consistently produced with quality [55]. These regulations are important for any
type of drug, whether small molecule or biopharmaceutical. There are many impor-
tant aspects to GMP, and these are explored more fully in Chapter 3.
GMP regulations require that drug products be tested to ensure batch-to-batch
consistency and product quality. Specifications, including both testing methods and
acceptance criteria, are established to routinely confirm the quality of product; con-
formance to specifications is required for the product to be deemed acceptable for re-
lease and for use. The quality control (QC) lab is typically responsible for this testing.
These specifications are tied to the CQAs that have been determined for a product,
although several CQAs may be measured by a single test method. An example of typi-
cal release testing for a monoclonal antibody product is shown in Table 1.5.

Table 1.5: An example of specifications for a monoclonal antibody drug product. Adapted from
Anurag Rathore, Setting Specifications for a Biotech Therapeutic Product in the Quality by Design
Paradigm, BioPharm International, 23(1) [56].

Test Category Examples of Specific Tests for Each Category Range

General quality Appearance Colorless. No visible

particles

pH .–.

Identity Capillary zone electrophoresis Conforms to standard

Peptide mapping Conforms to standard

Quantity Protein concentration by UV absorbance at – mg/mL

 nm

Potency Antigen binding assay –% compared to

reference standard
 1.6 Summary 23

Table 1.5 (continued)

Test Category Examples of Specific Tests for Each Category Range

Purity (and impurities) SE-HPLC for aggregates Aggregates < .%

Percent deamidation by IEC ≤.%

Immunoassay for host cell proteins ≤ ng/mg

PicoGreen for residual DNA ≤ pg/mg

Safety LAL assay for endotoxin ≤. EU/mg

Sterility Sterile (USP)

Note that details of QC/analytical testing are beyond the scope of this book and are
not covered.
Now that we have covered biopharmaceutical products, including their quality
attributes, we shift focus in the subsequent chapters to the methods, equipment, fa-
cilities, design principles, and regulations used in production of these products
from biological sources.

1.6 Summary

In this book, we define biopharmaceuticals as medicines inherently biological in na-

ture and manufactured by or from living organisms. Biopharmaceuticals are among
the biggest selling medicines in the world. In fact, Humira®, a monoclonal antibody
produced in CHO cells, has been the biggest selling drug worldwide for at least seven
years, with sales of nearly US$20 billion annually [11]. Biopharmaceuticals are differ-
ent from more traditional chemically synthesized drugs like aspirin. In particular,
they are:
1. large, complex molecules, viruses, or cells, while more traditional pharmaceut-
icals are small molecules with well-defined structures,
2. produced using relatively complex processes that are biological in nature, and
3. largely administered to patients parenterally.

Major classes of biopharmaceutical include protein therapeutics, vaccines, gene thera-

pies, and cell therapies. Many products within these classes are produced via genetic
engineering of cells, but many, including a number of vaccines, are not. Table 1.6 pro-
vides examples of biopharmaceuticals in each major category.
Most biopharmaceutical drug products, with perhaps the exception of cell thera-
pies, are either a lyophilized powder or liquid solution. The latter is preferred due to
ease and flexibility in administration; however, the former is sometimes required to
24 Chapter 1 An introduction to biopharmaceutical products

Table 1.6: Summary of biopharmaceutical categories, with example products.

Biopharmaceutical Example Description Use

Category

Protein therapeutic Humira® Human immunoglobulin (IgG) Rheumatoid arthritis and a

monoclonal antibody number of other autoimmune
diseases

Vaccine Flucelvax® Fragments of inactivated Prevention of influenza

influenza virus disease (i.e., the flu) caused
by influenza virus type A and
type B

Gene therapy Luxturna® Live, non-replicating adeno- Retinal dystrophy

associated virus serotype  that
has been genetically modified to
express the human RPE gene

Cell therapy Kymriah® T cells genetically modified to B-cell precursor acute

produce anti-CD chimeric lymphoblastic leukemia and
antigen receptor large B-cell lymphoma

provide necessary shelf life, typically on the order of two years. Drug products must
be designed and manufactured for their intended use, or, alternatively, we would say
they must have suitable quality. Quality attributes for a biopharmaceutical product
must be developed and translated to specifications, which include both test methods
and acceptance criteria. Drug product is tested to these specifications to ensure batch-
to-batch consistency and product quality. An example of specifications for a monoclo-
nal antibody product is given in Table 1.5.
Subsequent chapters cover the regulations that must be followed to ensure bio-
pharmaceutical product quality and details of the fundamentals, design, and execu-
tion of the unit operations used to produce these products.
Chapter 2
An overview of processes and facilities
for biopharmaceutical production

Having described biopharmaceutical products, our attention turns to their manufac-

ture, with an emphasis on product quality and producing the required quantity.
Producing quality product in the required quantity depends significantly on the pro-
cess – including the equipment and methods – used in the manufacture of the prod-
uct, the facility in which manufacturing takes place, and (in the case of quality) the
practices implemented to comply with Good Manufacturing Practice (GMP) regula-
tions. This chapter provides an overview of the processes for manufacture of biophar-
maceuticals, including their design and the facilities that house them. More in-depth
coverage of individual unit operations used to carry out these processes is provided
in Chapter 4 and beyond. The topic of GMP regulations is covered in detail in the next
chapter.
Specific questions that this chapter addresses are:
– Many biopharmaceutical processes can be viewed as consisting of four stages.
What are these stages and the objectives of each?
– What unit operations are involved at each stage?
– What is the difference between the upstream and downstream biopharmaceuti-
cal manufacturing process?
– How is drug product produced from drug substance?
– What advantages does continuous processing offer over the more common batch
processing in biopharmaceutical production? Disadvantages?
– What advantages does single-use equipment offer over reusable equipment? What
are the disadvantages?
– How much product is made? And what do we mean by the terms titer, step yield,
and process yield?
– What are the basic steps involved in designing a biopharmaceutical process?
– What important concepts are used to guide the design of a biopharmaceutical
facility?

Before going further, let’s take a minute to explain a couple of terms that are
used throughout this chapter and the remainder of the book: unit operation and
product. Unit operation refers to any individual step in the process that serves a
specific function. For example, centrifugation is a unit operation commonly used
in biopharmaceutical manufacturing. Centrifuges are used to separate solids from
liquids in order to isolate the product. In some cases, product may be in the liquid
phase; in other cases, product may be in the solid. Either (or both) can be col-
lected from a centrifuge. Individual unit operations, such as centrifugation, are

https://doi.org/10.1515/9783110616880-002
26 Chapter 2 An overview of processes and facilities for biopharmaceutical production

connected to create the overall process. The second term, product, seems clear
enough but deserves some explanation. In the previous chapter, drug product
was defined as the final dosage form that contains the active ingredient plus a
number of other ingredients, referred to as excipients, which have no pharmacologi-
cal effect. In this and subsequent chapters, we frequently use the term product
(without “drug” as a descriptor) by itself. When used in the context of the biophar-
maceutical process, product has several meanings. It may refer to the specific active
ingredient (e.g., protein or virus) at any step in the process. It may also refer to the
intermediate form of the product that includes the active ingredient dissolved in
the process liquid at any process step. Throughout the book, we typically refer to
this form of the product as the intermediate, product intermediate, or process
intermediate.

2.1 Biopharmaceutical manufacturing processes – an overview

This section of the chapter provides an overview of topics related to biopharmaceu-

tical processes in order to put subsequent chapters in the proper context. We begin
with a description of the main process steps required to make drug product from
cells or other starting materials. Activities that support the main process – such as
preparing solutions for use in different unit operations – are covered as well. We
then discuss features of the equipment used to carry out all process activities and
define terminology related to the amount of product produced.

2.1.1 The main process

Two main goals are to be achieved by the processes used to manufacture biophar-
maceutical products: to consistently produce drug substance and drug product with
the intended quality [52, 57] and to produce the necessary quantity of drug product
for clinical or commercial use. Rather than proceeding immediately into the se-
quence of unit operations used to accomplish these goals, which vary from product
to product, let’s first consider a generalized process divided into four stages. Each
stage consists of multiple steps (or unit operations) with a common overall purpose.
These four process stages are depicted in Figure 2.1 and are common to the manu-
facture of many products.
Stage 1 is the product synthesis or production stage. The objective of this stage is
to produce the product in quantities suitable for clinical trials and/or commercial dis-
tribution. For the many biopharmaceuticals produced using genetically engineered
cells, this stage involves cell growth (i.e., growth of individual cells and an increase
 2.1 Biopharmaceutical manufacturing processes – an overview 27

Cells

Upstream Stage 1:
Product Synthesis
(Production)

Downstream
Stage 2: Cells or spent
Harvest (Recovery) media/cell debris

Soluble impurities both

Stage 3:
different from and
similar to the product

Stage 4:
Unwanted buffer
Formulation
components
and Fill

Drug product

Figure 2.1: The four stages of biopharmaceutical production. Note that Avastin® is used as an
example of a biopharmaceutical in this figure. As described in Chapter 1, Avastin® is a humanized
immunoglobulin G subclass 1 (IgG1) mAb produced by genetically engineered Chinese hamster
ovary (CHO) cells and used for treatment of metastatic colorectal cancer. Image © NC State
University; reprinted with permission.

in the number of cells) in bioreactors. It is important to note that two different terms
are commonly used to describe cell growth, based on the type of cells used:
– Cell culture: the growth of animal cells, including mammalian cells like CHO
and insect cells like Sf9 (cells from ovaries of a Fall armyworm).
– Microbial fermentation (or just fermentation): the growth of microorganisms,
like E. coli or the yeast Saccharomyces cerevisiae.

Both processes begin with a working cell bank (WCB) that has been prepared from a
master cell bank (MCB). A WCB is typically made up of 1 mL cryogenically preserved
vials of cells. This small number of cells must be expanded to a much greater num-
ber to produce the required amount of product. To produce an adequate number of
cells for inoculation into a production bioreactor, the WCB is used to initiate a seed –
also referred to as inoculum – train. That train may include multiple cell growth steps
28 Chapter 2 An overview of processes and facilities for biopharmaceutical production

of increasing volume to generate the required number of cells. In both fermentation

and cell culture steps, cells are typically grown in an aqueous medium that contains
nutrients and other chemicals for growth.
It is worth noting that cell growth may also be required for biopharmaceuticals
that are not the result of genetic engineering. An example is Flucelvax® flu vaccine,
described in Chapter 1. The influenza virus used to make the vaccine is propagated
in non-engineered Madin-Darby canine kidney (MDCK) cells, so virus stock must be
added to the bioreactor containing MDCK cells for production [5]. Another example
is Prevnar® 13, a vaccine produced by non-engineered Streptococcus pneumonia
[58] that protects against various pneumococcal bacteria that cause disease in
children and adults. There are even biopharmaceuticals whose production does
not involve the use of cells at all. Gamunex®-C, an immune globulin injection
used to treat primary humoral immunodeficiency, is an example [59]. It is pro-
duced by purifying IgG from pools of human plasma; thus, this product is synthe-
sized by humans rather than in a bioreactor.
Stage 2 is the harvest stage, also referred to as recovery. The objective of this stage
is to separate product from the production (i.e., host) system. Stage 2 prepares the
product intermediate for stage 3 by producing a clarified (i.e., particle-free) intermedi-
ate that is suitable for loading onto a chromatography column in stage 3, to be dis-
cussed shortly. The process intermediate from the harvest stage is commonly product
dissolved in liquid – typically media from stage 1 – that has been cleared of cells and/
or cell debris if cells are involved in the process. However, the clarified liquid from
the harvest stage contains numerous soluble impurities that need to be removed.
There is a key distinction to make at this point regarding where product resides
once produced: some remain within the cell and are referred to as intracellular
products, while others are found dissolved in the aqueous bioreactor medium and
referred to as extracellular products. This difference impacts the design of the har-
vest process and is discussed in greater detail in Chapter 5.
Stage 3 is the purification stage. The objective is to purify the biopharmaceutical
product; that is, separate the product from soluble impurities that are present. This
stage starts with a clarified aqueous solution (from the previous stage) in which the
product is dissolved. As mentioned previously, intermediate from the harvest stage
contains numerous soluble impurities that may not be separated from the product
as part of the harvest operations. Impurities common in many biopharmaceutical
processes include host cell proteins, endotoxin, host cell DNA, and product aggre-
gates. These and other impurities are discussed in much greater detail in Chapter 8.
This stage is essential to ensuring that critical quality attributes (CQAs) related to
purity, potency, and safety are met.
In biopharmaceutical manufacturing, the most common unit operation for puri-
fication is column chromatography. The need to begin stage 3 with a clarified inter-
mediate results from the fact that chromatography typically involves a bed packed
with resin beads that is easily clogged if particles are present in the feed. Usually,
 2.1 Biopharmaceutical manufacturing processes – an overview 29

three or more chromatography steps, each with a different interaction mode, are re-
quired to meet purity requirements. Other types of purification operations include
precipitation, used to a much lesser extent than chromatography, and viral clearance
steps, which inactivate or remove adventitious virus (i.e., viruses that are not sup-
posed to be part of the process). Viral safety is an important consideration for any
process that uses mammalian-derived components. Processes that use mammalian
cell lines, such as CHO cells, typically require viral clearance steps. Examples include
low pH incubation of the product to inactivate enveloped virus, virus filtration, which
holds back adventitious virus while allowing product to flow through the filter, and
certain types of chromatography. Processes that use cell lines that are not of animal
origin, such as E. coli, are typically not expected to be designed with viral clearance
steps because those cells cannot harbor adventitious mammalian viruses that might
be infectious to human beings [60].
Stage 4 is the formulation/filling stage. The objective of this stage is to produce
the final dosage form, that is drug product, and it involves a number of steps. These
steps are central to ensuring acceptable quality attributes related to appearance,
quantity, potency, and safety.
The term formulation refers to the steps required to ensure that purified product
in liquid solution from stage 3 is at the correct concentration, is in the correct buffer
system, and includes the proper excipients (in addition to the buffering agents).
Formulation is necessary because it is likely that the product concentration and
buffer in the effluent from stage 3 are not what is required for final drug product but
rather are dictated by the final chromatography step in the process. Figure 2.2 shows
a common sequence of operations to meet these objectives: concentration/buffer ex-
change by ultrafiltration followed by addition of remaining excipients. After these
steps, bulk filling takes place to produce bulk drug substance.
Bulk filling prepares drug substance for storage prior to executing final fill-
finish steps required to produce final drug product. It is carried out by dispensing
product intermediate from the UF step through a 0.2 µm filter (that is, a filter with
0.2 µm pores) into an appropriate storage container, which can have a range of
sizes depending on the product. Examples of commonly used bulk filling containers
include disposable polymeric bags, rigid polymeric containers, or stainless steel
portable vessels. The bulk drug substance contains active ingredient in aqueous
buffered solution largely free of soluble impurities. It is typically stored frozen or
refrigerated until it is ready to be filled into the drug product container (e.g.,
vial, prefilled syringe, autoinjector, etc.).
At this point, it is worth noting that drug substance is usually produced as a
low-bioburden intermediate, while biopharmaceutical drug product is sterile, as
discussed in Chapter 1, given that the common route of delivery is parenteral. Since
we are using terms related to microbial contamination, let’s take a minute to define
these and a couple of other relevant ones.
30 Chapter 2 An overview of processes and facilities for biopharmaceutical production

(a) Production of bulk drug substance (b) Production of drug product from drug substance

Bulk drug
(from Stage 3) substance thaw

Concentration/buffer Pooling and additional

compounding

Excipient addition
(if applicable)

lyophilization if needed)

BDS storage
Labeling and packaging
(typically frozen)

Shipping

Figure 2.2: Process steps associated with stage 4. (a) represents the steps starting from intermediate
from the purification stage leading to bulk drug substance (BDS). (b) represents steps used to
produce drug product from drug substance. Note that these diagrams follow the path of the product
and do not include other necessary steps, such as washing and sterilization of vials to be filled or
preparation of vial stoppers and caps. Image © NC State University; reprinted with permission.

– Bioburden: the level and type of microorganisms that can be present in raw materi-
als, process intermediates or drug substance. Bioburden should not be considered
contamination unless the levels have been exceeded or defined objectionable or-
ganisms have been detected [61].
– Sterile: free from living organisms (i.e., free from bioburden). A sterile product,
such as parenterally-administered biopharmaceuticals, contains no bioburden.
Further, bioreactors used to grow cells during the product synthesis stage are
sterilized prior to inoculation with the cells.
– Axenic: free from living organisms other than the cells required for the process.
When growth of cells is carried out in a bioreactor, the fermentation or cell cul-
ture is axenic, not sterile.
 2.1 Biopharmaceutical manufacturing processes – an overview 31

– Sanitized: a reduction in viable microorganisms to a defined acceptable level.

Downstream process equipment is typically cleaned and sanitized, not sterilized.
– Aseptic: a process and/or facility designed to keep a product sterile.

The steps used to produce drug product from drug substance are also shown in
Figure 2.2. Drug substance containers are thawed (if frozen) then pooled. Exci-
pients are added to the drug substance as needed after pooling (referred to as com-
pounding in Figure 2.2), followed by sterile filtration with a 0.2 µm filter and filling of
product into its final container. Filling of biopharmaceutical drug product is usually
done under aseptic conditions to ensure drug product sterility. To provide these con-
ditions requires a space designed to maintain sterility of the product, delivery device,
and equipment.
Note that the alternative method for ensuring sterility of a filled product is ter-
minal sterilization. This term refers to filling a low-bioburden product into the final
container followed by sterilization after filling. Sterilization can be accomplished in
a number of ways, such as applying heat or irradiation, which would denature most
biopharmaceuticals. Thus, aseptic filling rather than terminal sterilization is used
for producing biopharmaceutical drug product.
It is important to note that in a manufacturing environment, unit operations are
commonly categorized as being part of the upstream or downstream process, as
shown in Figure 2.1. The upstream process includes product synthesis (stage 1) opera-
tions and may include harvest (stage 2), although some facilities may include harvest
operations with the downstream process. The downstream process includes purifica-
tion and formulation/filling operations, up to and including the bulk filling step.
Figure 2.3 brings together the individual unit operations in each stage in a pro-
cess flow diagram for production of a monoclonal antibody (mAb). The process be-
gins by thawing a vial of cells from the working cell bank. Those cells are expanded
in number until there are enough to be inoculated into the production bioreactor.
Once in the production bioreactor, cells are grown until they are no longer producing.
In the case of CHO cells grown in fed-batch mode for production of a mAb, the pro-
duction reactor step can last as long as two weeks. The product in the bioreactor is
harvested by sending the broth, often referred to as harvest cell culture fluid (HCCF),
to a series of clarification steps. Figure 2.3 shows clarification as being accomplished
by a combination of centrifugation, for removal of most cells, followed by depth filtra-
tion for removal of residual cells and cell debris. The resulting process intermediate is
often referred to as cell culture supernatant. That intermediate is fed to a series of
chromatography and viral clearance steps as part of the purification stage. These
steps remove the soluble impurities and contaminants (or inactivate virus in the case
of low pH incubation), including adventitious virus that may be present. Note that
effluent from the chromatography steps is often referred to as column eluate. Filtrate
from the viral filtration goes into the formulation/fill stage, where product concentra-
tion and buffer exchange take place by ultrafiltration. Once the proper concentration
32 Chapter 2 An overview of processes and facilities for biopharmaceutical production

and buffer for drug product are achieved, the resulting drug substance is filled. The
harvest, purification, and formulation/fill steps, stopping at filled drug substance,
would typically require 5–7 days to complete.

Stage 1: Stage 2: Stage 3: Stage 4:

Product synthesis Harvest

WCB thaw Centrifugation Concentration/buffer

(Chapter 4) (Chapter 6) chromatography exchange by
(Chapter 8)
(Chapter 9)
Seed culture
expansion (Chapter 7) Low pH incubation
(Chapter 4) (Chapter 8)
(Chapter 2)

Production in Cation exchange

bioreactor chromatography mAb drug substance
(Chapter 4) (Chapter 8)

To harvest Anion exchange

chromatography
(Chapter 8)

(Chapter 8)

Figure 2.3: A process flow diagram showing individual unit operations in a typical process for drug
substance production of a mAb from genetically engineered CHO [62]. The chapter in which each of
these operations is covered is also shown. Note that cell lysis and tangential-flow microfiltration,
which are not part of this mAb process, are covered in Chapters 5 (cell lysis) and Chapters 7 and 9
(microfiltration). Image © NC State University; reprinted with permission.

This discussion makes clear that multiple different, somewhat complex unit opera-
tions are used in the manufacture of biopharmaceutical drug substance. In fact, bio-
pharmaceutical drug substance manufacture involves significantly more steps than
production of drug product from drug substance. This book covers in greater detail
the operations required for the product synthesis, harvest, purification, and for-
mulation/fill stages of drug substance manufacture. The process for drug product
manufacture from drug substance, however, is not covered in more detail than
has already been presented.
It is important to note that sampling and a variety of testing is performed through-
out the process as part of the overall control strategy to ensure quality product is
 2.1 Biopharmaceutical manufacturing processes – an overview 33

consistently manufactured. We discussed release testing of drug product against an

item specification in Chapter 1. Likewise, drug substance is tested against an item
specification as part of its release process. In addition, testing of raw materials and
in-process material (i.e., product intermediate) is performed. Testing requirements
vary from product to product. Most of this testing is performed by the quality control
(QC) lab; so samples must be pulled, sent to the QC lab, then tested. In the future, it
is likely that replacement of slower offline testing by rapid inline sensor measure-
ment, particularly for in-process samples, will occur.
The unit operations used in drug substance production are most often run in
batch mode, or in the case of a bioreactor step, fed-batch mode, rather than contin-
uous mode. In batch mode, a finite amount of feed is “charged” to a unit operation,
processed, and all product intermediate removed at some time later. The product
intermediate is then charged to the next operation, where it is processed and re-
moved at some time later. This continues at each processing step until drug sub-
stance is produced. In fed-batch operation of a bioreactor, cells are inoculated and
grown in batch mode for a certain amount of time, while nutrients are fed (hence
the term fed-batch) throughout the duration of the step to feed them. In a continu-
ous process, input and output streams for each unit operation flow continuously. So
if you were to draw a box around the entire drug substance process, there would be
a continuous flow into and out of that box.
Despite batch being the common mode of operation, some continuous processing
is used in biopharmaceutical manufacturing, most notably perfusion systems for con-
tinuous cell growth. Perfusion systems are covered in greater detail in Chapter 4.
Implementation of continuous processing in downstream operations has lagged
in development due to the complexity of converting steps like chromatography
and diafiltation to continuous operation. Further, to the best of our knowledge,
no fully continuous process for biopharmaceutical drug substance manufactur-
ing exists for an FDA-approved drug product. Regardless, there are a number of
potential advantages that continuous processing offers over batch processing
[63, 64]:
– Higher equipment utilization. Equipment is utilized constantly rather than only
for a single step of every batch as in batch operation;
– Smaller equipment. For a fixed rate of production, relatively low flow rates are
used in continuous operations, resulting in smaller equipment;
– Higher volumetric productivity;
– Lower capital investment in equipment. Smaller equipment means less up-front
investment;
– Smaller, less costly facilities. Smaller equipment translates to a smaller, less ex-
pensive facility;
– Reduced risk of product quality issues as product residence time (within a biore-
actor) and hold times are reduced.
34 Chapter 2 An overview of processes and facilities for biopharmaceutical production

The number of sessions being devoted to continuous processing at bioprocessing

conferences suggests that much effort by industry, academia, and the vendor com-
munity is being placed on development of continuous processes for biopharmaceu-
tical drug substance manufacture. It will become clearer in the coming years as to
whether the promise held by continuous processing comes to fruition. And there are
a few potential disadvantages that will have to be overcome, including an increase
in the time required to design processes.

2.1.2 Supporting activities

A number of activities are required for production that are not explicitly shown in
Figure 2.3 and yet are part of most biopharmaceutical processes. We provide an over-
view here so that the scope of these activities is clear.
The preparation of various solutions and media, commonly referred to as solu-
tion prep and media prep, respectively, are among the activities needed to support
a biopharmaceutical process. Biopharmaceutical manufacturing processes require
numerous solutions. For example, as will become clear in Chapter 8, chromatography
steps are likely to require solutions of NaCl, NaOH, and various buffers, to name just
a few. A single chromatography step may require eight different solutions. Looking at
Figure 2.3, there are three separate chromatography steps in the process shown,
which could require a total of as many as 20 different solutions. Other process steps
require solutions as well. For a multi-product facility running at capacity, it is easy to
see how the number of different solutions required may exceed 100. Consequently,
solution prep tends to be a high-throughput operation in many facilities; further, fail-
ure to correctly prepare solutions can have negative consequences to process perfor-
mance and product quality, as will become clear in later chapters. Solution prep is
commonly performed by adding dry components (such as NaOH pellets) or solution
concentrates (e.g., 5 M NaOH) to purified water at the required quantity and mixing in
a vessel to produce a solution at the correct concentration and with the necessary vol-
ume. From the mixing vessel, solutions are transferred to storage vessels, often via a
filter to reduce the likelihood of microbial contamination. Modifications to this conven-
tional process to prepare solutions have been made. An example is the use of inline
buffer dilution, using equipment like Asahi Kasei’s IBD™ systems. This skid-based sys-
tem uses stock concentrates as starting material and automatically dilutes them with
water and adjusts pH (if required) to produce solutions of a given volume, concentra-
tion, and pH [65]. Similar to solution prep, media prep is commonly carried out in sup-
port of upstream processing steps.
Cleaning, sanitization, and sterilization operations are also critical to processes
like the one shown in Figure 2.3. Reusable equipment must be cleaned after each batch
to ensure that residues of product, buffers, cleaning agents, etc. from the previous use
 2.1 Biopharmaceutical manufacturing processes – an overview 35

are removed. This cleaning is typically accomplished using clean-in-place (CIP)

techniques. CIP is a cleaning mode in which removal of soil from product contact sur-
faces is carried out by spraying and/or flowing cleaning agents and water rinses over
the surfaces to be cleaned without disassembling equipment. This method is in con-
trast to the less common method of disassembling equipment and manually cleaning
in a location away from the processing area, referred to as clean out of place (COP).
For vessels, such as bioreactors or portable process vessels, CIP is usually accom-
plished with a separate CIP system that delivers purified water and cleaning solutions
to the vessel through a spray ball. The spray ball creates high velocity jets of water or
cleaning solution that soak process fluid-contact surfaces in the interior of the vessel.
Some process equipment, however, does not require a separate CIP system to carry
out clean-in-place operations because pumps capable of delivering cleaning solution
are built into the equipment. Chromatography systems and ultrafiltration systems are
two such examples.
In addition to cleaning, some equipment is sanitized as well, particularly down-
stream equipment that does not undergo a separate sterilization step. Sanitization is fre-
quently done by chemical methods and often in combination with cleaning. As an
example, downstream equipment such as ultrafiltration systems are often simulta-
neously cleaned and sanitized with a 0.5 M NaOH solution. Sterilization of equipment,
particularly reusable bioreactors that must be free from microbial contamination, is typ-
ically performed by heating up the equipment surfaces using clean steam, in a process
referred to as steam in place (SIP). Cleaning, sanitization, and sterilization as they apply
to the different unit operations are covered in more detail in subsequent chapters.
We should also add that for cleaning small parts used in a biopharmaceutical
process, parts washers are used that work like large kitchen dishwashers. To steril-
ize small parts, an autoclave is used. The autoclave is basically a chamber into which
small equipment is loaded and exposed to high temperature steam for sterilization.
Both parts washers and autoclaves are designed to run automated cycles. There are
many good references that provide details on the topics of cleaning, sanitization, and
sterilization, including the book Clean-In-Place for Biopharmaceutical Processes [66]
and Sterility, Sterilisation, and Sterility Assurance for Pharmaceuticals [67].
To support operations related both to the main process presented in 2.1.1 and
the supporting activities described so far, there are a number of clean utilities, also
referred to as critical utilities, that must be generated. They are clean and critical
because they may make direct contact with the product. Clean utilities include the
following:
– Various purified waters. Biopharmaceutical processes require large volumes of
water for cleaning and preparing media and solutions. The two types of purified
water most commonly used begin as drinking water that has been purified to pro-
duce one type commonly referred to as purified water, which we will refer to as
high purity water (HPW), and the second type known as water for injection (WFI),
often referred to as “wiffey.” In the United States, requirements for these purified
36 Chapter 2 An overview of processes and facilities for biopharmaceutical production

waters are given in the U.S. Pharmacopeia (USP), a compendium that contains
specifications for various ingredients used in preparation of drugs. (Other regions
of the world have similar compendia; for example, there is a European Pharmaco-
peia and Japanese Pharmacopeia). For water, the relevant chapter is USP 1231,
which describes processes and testing performed for various water types [68].
HPW and WFI must meet the same requirements for conductivity and total or-
ganic carbon. An important difference between the two is that WFI has an endo-
toxin specification that must be met, while HPW does not. In biopharmaceutical
processes, HPW is often used for equipment cleaning. WFI is used as an excipient
in production. It is also used for equipment cleaning and to prepare most solutions.
The processes for producing either HPW or WFI are somewhat complex. For exam-
ple, an HPW-generation process may consist of the following steps, starting with
potable water feed: filtration for particulate removal; water softening for cation re-
moval; carbon bed filtration for removal of chlorine and organic compounds;
filtration for removal of carbon fines; treatment with ultraviolet (UV) light to
kill microorganisms; reverse osmosis for removal of small molecules, including
dissolved salts; a deionization step to further remove ions; and a 0.2 µm filtra-
tion step.
– Clean steam, also referred to as pure steam. It is used for sterilization of equip-
ment and/or process fluids. It is often generated from treated water, such as HPW.
– Clean compressed gases. These include air, O2, CO2, and N2. Air, O2, and CO2 are
typically used for fermentation and cell culture. N2 is used to blanket product
when needed. These gases often come from a supplier. Many facilities generate
their own compressed air with a system consisting of an air compressor, filters
for removing particulate, and solid media for reducing humidity.

A number of other support utilities are required for operation of a biomanufacturing

facility, including building heating, ventilation, and air conditioning (HVAC) systems,
waste inactivation systems, chilled water generation systems, electricity and water. De-
tails about these utilities are beyond the scope of this book and are not be covered fur-
ther. There are also a number of other supporting activities required for the main
manufacturing process that have not been discussed. These include filter integrity test-
ing, cell bank storage, and testing of in-process samples.

2.1.3 Equipment for the main process and support activities

Specialized equipment is required to execute upstream processing, downstream proc-

essing, and the support activities discussed so far. Equipment details appear in each
of the subsequent chapters on the various unit operations required for biopharmaceu-
tical manufacturing, while this section provides general information. Process equip-
ment comprises more than just the main equipment component in which product is
 2.1 Biopharmaceutical manufacturing processes – an overview 37

being processed. For example, cell culture requires more than just the bioreactor in
which cells are grown. Likewise, performing a chromatography step requires more
than just the packed column that performs the necessary separation. In addition to
these main components, bioprocessing equipment includes lots of “plumbing,” such
as inlets that allow process fluids to be connected to the functional component and
outlets that allow process fluids to be collected or discarded. Equipment also includes
a variety of valves that direct flow and pumps that move fluids through the system. It
also includes many sensors used to monitor parameters important to the operations
and send input to the control software. For example, bioreactors are likely to be
equipped with sensors to measure temperature, pH, dissolved oxygen, and dissolved
carbon dioxide. Chromatography systems are equipped with sensors to measure con-
ductivity, temperature, the presence of air bubbles in the liquid feeds, pressure, flow
rate, UV absorbance, and pH.
Other important characteristics of process equipment include scalability; that is,
equipment that allows for operations to be conducted at all scales, from bench-scale
development to manufacturing. Reusable process equipment must be cleanable, san-
itizable, and sterilizable (if sterilization is necessary). For example, cleanable equip-
ment is designed so that all fluid-contact surfaces are accessible during cleaning and
that system pumps are able to deliver the necessary flow rate for cleaning solutions
when an automated CIP system is not used. For more information on design of
equipment for biopharmaceutical processing see the American Society of Mechanical
Engineering Standard (ASME) on Bioprocessing Equipment [69].
Another common feature of equipment in biopharmaceutical processes, partic-
ularly downstream equipment, is that it is skid mounted. The term skid refers to a
collection of components mounted on a frame to support a given activity. For down-
stream equipment, skids are often on wheels so that they are easily moved. Exam-
ples of a centrifugation skid and chromatography skid are shown in Figure 2.4.
Each skid contains all of the inlets, outlets, valves, pumps, sensors and (in some
cases) control hardware and software required to execute a centrifugation or chro-
matography step for a biopharmaceutical process.
Both single-use and reusable equipment are used in biopharmaceutical pro-
cesses. With single-use systems, components that come in contact with process
fluids are generally made from polymeric materials intended for one-time use and
discarded afterwards. In contrast, reusable equipment is typically made from stain-
less steel. One key difference between single-use equipment and reusable equipment
is that the latter needs to be cleaned, sanitized and/or sterilized between uses, while
single-use equipment does not because it is disposed. Further, the cleaning, sani-
tization, and sterilization processes for reusable equipment must be validated for
commercial manufacturing, a time-consuming and resource-intensive effort that
is discussed in greater detail in the next chapter. A summary of the advantages of
single-use equipment is given below:
38 Chapter 2 An overview of processes and facilities for biopharmaceutical production

(a) (b)

Figure 2.4: An example of a centrifugation skid (a) and chromatography skid (b). The centrifuge is
an Alfa Laval LAPX 404 disc-stack centrifuge, and the chromatography system photo shows the
side and back view of a Cytiva (formerly GE Healthcare) AKTAprocess system. Photos © NC State
University; reprinted with permission.

– Speed to market;
– Lower capital costs, particularly for a new facility;
– Reduction or elimination of cleaning and sanitization, which reduces the cost of
water, cleaning/sanitization agents, and cleaning validation;
– Elimination of sterilization of bioreactors, which reduces the cost of water, power
for steam generation, and sterilization validation;
– More rapid, less expensive changeover between campaigns of different prod-
ucts, which can improve facility productivity;
– Reduced probability of cross contamination and microbial contamination.

With all of those advantages, why isn’t single-use equipment universally adopted?
A few reasons are listed below:
– Scale limitations. For example, the largest single-use bioreactor currently avail-
able is 6,000 L [70];
– Lack of standardization of single-use components;
– Increased reliance on outside suppliers to provide the components required for
each run;
– Additional consumables cost. A key question that must be addressed in deciding
between single-use and reusable equipment is does the savings in capital costs,
water, and labor associated with single-use equipment offset the cost associated
with purchasing the single-use components required for each batch produced.
For example, high costs of chromatography resins and tangential flow filtration
 2.1 Biopharmaceutical manufacturing processes – an overview 39

membrane modules often limit use of these unit operations in a truly single-use
way;
– Leachables/extractables. Extractables are defined as compounds that migrate
from any product-contact material when exposed to an appropriate solvent under
exaggerated conditions of time and temperature. Leachables are compounds that
are typically a subset of extractables that migrate from any product-contact mate-
rial under normal process conditions [71]. Their presence potentially impacts the
safety and quality of drug product;
– Increase in the amount of solid waste to be disposed.

Use of single-use equipment and components began in the early 1980s with the in-
troduction of membrane filters within plastic capsules. The mid-1990s saw the intro-
duction of “buffer” bags for storage of various solutions (and other process liquids),
along with totes to contain and move the bags and solutions they held [72]. An
example of buffer bags and totes is shown in Figure 2.5. Since the mid 1990s, sin-
gle-use technology options have continued to evolve to include mixing systems,
bioreactors, depth filters, filtration (including UF) units, and chromatography sys-
tems. Single-use versions of the latter two operations, in particular, have been en-
abled by the development of disposable flow paths. Single-use options for specific
process steps are discussed in subsequent chapters that cover individual unit op-
erations in more detail.

(a) (b)

Figure 2.5: Options for storage of solutions in biopharmaceutical processes. (a) shows buffer bags
and totes used for storage of solutions for chromatography processing. A single-use bag holds the
solution and is supported by a stainless-steel tote. (b) is a stainless steel portable vessel. Photos
© NC State University; reprinted with permission.
40 Chapter 2 An overview of processes and facilities for biopharmaceutical production

While the economic benefit of single-use equipment must be established on a case-

by-case basis, implementation of single-use bags to supply solutions is generally
considered to result in a significant savings compared to stainless steel storage
tanks. A net present value analysis of a process requiring 100 or more fermentation
batches annually demonstrated that bags for storage of media, solution, and inter-
mediate product clearly win from a process economic perspective over stainless
steel storage vessels [73].
Reusable equipment may be shared between different products or dedicated to
one product. Either is acceptable. However, for multi-product operations in which
reusable equipment is shared, steps must be taken to ensure that product-contami-
nation risks, such as contaminating product A with product B, are minimized. These
steps include development of effective cleaning procedures that can be validated to
ensure that one product is completely removed from a piece of equipment prior to
introducing a new product into that equipment. It is important to note that there are
a number of process components that are not usually shared between products, such
as chromatography resin, ultrafiltration membrane devices, filling equipment, and
gaskets. While these components may be reusable, they are typically dedicated to a
product due to concerns that if used across multiple products, the polymers from
which they are made may desorb product A into a batch of product B.

2.1.4 Amount of product to be produced

Now that we have addressed the process to make biopharmaceutical drug substance
and drug product with the desired quality, let’s shift focus to the quantity of product
produced. For any given product, the amount needed depends on a number of fac-
tors. If producing for preclinical studies or clinical trials, the amount of product
needed will likely be less than what is required for commercial distribution. And
different types of products will require different amounts just to meet market de-
mand. For example, market demand for all recombinant Factor VIII-based products,
classified as blood clotting factors in Chapter 1 and used in the treatment of hemo-
philia A, is approximately 300 g of protein annually. By comparison, the single
drug Humira®, the world’s largest selling drug, has a demand of about 700 kg of
protein annually, significantly higher than that for all Factor VIII products com-
bined [74].
There are a number of different terms related to quantifying the amount of prod-
uct that are commonly used. These are given below:
– Titer. The term generally refers to the concentration of a component in a solu-
tion. In biopharmaceutical manufacturing, it often specifically refers to the
amount of product produced per unit volume of bioreactor broth, with units of
g/L. As an example, the average titer for mAb products is estimated to be ap-
proximately 3.25 g/L in 2020 [75]. Titers for mAb products have grown through
 2.1 Biopharmaceutical manufacturing processes – an overview 41

the years due to improvements in engineering of cell lines, media optimiza-

tion, and improved process control.
– Percent Yield. This term is defined by the following equation:

amount of product at the end of one step or series of steps

% yield ¼ × 100 (2:1)
amount of product at the begining of the step or series of steps

The “amount” is often in grams, but can be of any unit of measure relevant to a
particular product. For example, in vivo gene therapies are often quantified in
terms of vector genomes, which refers to the concentration of viral capsids car-
rying the gene of interest. Note that the term percentage step yield is used when
equation (2.1) is applied to one step in the process and percentage process yield
when applied to the entire process. Percent recovery is a term used synonymously
with either percent step or process yield.
Even though GMP regulations are not discussed until the next chapter, it is
worth noting that several yield terms are defined in those regulations in the
United States: actual yield, theoretical yield, and percentage of theoretical yield
[53]. Actual yield refers to the quantity (e.g., grams) that is actually produced at
any phase of manufacture, processing, or packing of a particular drug product.
Theoretical yield is the quantity that would be produced at any phase of manu-
facture, processing, or packing of a particular drug product in the absence of
any loss or error in actual production. And percentage of theoretical yield is the
ratio of the actual yield to the theoretical yield applied to the same phase of
manufacturing, stated as a percentage. Therefore, percentage of theoretical yield
defined in 21 CFR Part 210 is analogous to the more commonly used terms percent-
age step yield and process yield as defined here.
– Productivity. The term refers to the amount of product produced per relevant
volume per time (e.g., units of g product/L/h). For example, the productivity of a
bioreactor could be given as the grams of product produced per liter of culture
broth per hour.

It is also worth noting that when referring to the scale of production, we typically
use the volume of a bioreactor. For example, a process that utilizes a 10,000 L biore-
actor is referred to as a 10,000 L process.

Example: Estimating the number of bioreactor runs required to meet demand for a mAb product
A company wants to produce 1,000 kg annually of a mAb drug product using a process that de-
livers a titer of 3.25 g mAb/L bioreactor broth. The process uses bioreactors with a 10,000 L
working volume, and the % process yield from the end of the bioreactor step to final fill-finish is
70%. How many bioreactor runs are required to meet the annual demand?
42 Chapter 2 An overview of processes and facilities for biopharmaceutical production

Solution
First, using the definition of titer and process yield, calculate the amount of mAb (in grams) pro-
duced as drug product from each bioreactor run given a titer of 3.25 g/L and process yield of 70%:

Amount of mAb drug product ¼ bioreactor broth volume × titer x % process yield=100
g mAb g mAb drug product
= 10, 000 L broth × 3.25 × 0.7
L broth g mAb in broth
= 22, 750 g mAb
Next, divide the total amount of mAb to be produced annually as drug product – 1,000 kg – by
the amount of mAb produced (as drug product) from each bioreactor run:

1, 000, 000 g mAb=year

=
22, 750 g mAb=run
= 44 bioreactor runs=year
Note that if a production run in the bioreactor takes 10 days, then one bioreactor is not enough
to perform the required number of runs in a year. At least two bioreactors running simulta-
neously are necessary.

2.2 Process design

This section briefly discusses the methodology used to design processes for com-
mercial production of biopharmaceuticals. We use the term process design to en-
compass both process development – those activities required to define the design
space (a term discussed in more detail shortly) – and scale-up, among other activi-
ties described shortly.
The major objectives of process design efforts are to ensure that a sufficient
amount of drug product is produced for commercial distribution; to establish a com-
mercial manufacturing process capable of consistently supplying product of the in-
tended quality [57]; and to reflect the design in planned master production and
control records [76]. That latter goal makes clear that a key deliverable from design
activities is a set of master batch records that provides detailed instructions for exe-
cution of a process. Batch records are considered in greater detail in the next chap-
ter. There are other important deliverables from process design as well, including
the actual physical process, a finalized list of CQAs, a complete process description
showing each unit operation and associated material attributes and operating
ranges, a list of raw materials and testing specifications, analytical methods and
specifications for product release testing, in-process test methods and acceptance
criteria, a complete process control strategy, and a host of reports to be written
that document the results from and rationale for the various design activities that
we will describe shortly.
 2.2 Process design 43

Before discussing how process design is done, let’s put it into the context of the
life cycle for a biopharmaceutical product. Figure 2.6 shows some of the activities
leading to commercialization of a biopharmaceutical product, including process de-
sign. Upon completion of drug discovery – the process by which new drugs are
identified – the biopharmaceutical candidate goes through preclinical studies,
which determine initial dosing and define pharmacological and toxicological ef-
fects. These studies usually involve animals. Following preclinical activities, the
drug can move into clinical trials – studies that involve humans – that are con-
ducted in phases. Success in one phase is required before progressing to the next.
The number of subjects increases with each phase, and the final phase – phase 3 –
looks at how the new treatment compares to the current treatment(s). Success in
clinical trials may lead to commercial production and distribution. There are nu-
merous sources available that describe clinical trials in greater detail [77].
As you can see in Figure 2.6, production of material for human use must be
done under GMP. But even as GMP manufacturing for clinical trial material takes
place, process design activities are ongoing. Thus, the process used to produce
phase 1 clinical trial material should be better defined and more robust by the time
commercial manufacture begins. After the process design stage, process qualifica-
tion takes place. Its main purpose is to bring the facility, utilities, equipment, and
trained personnel together along with all applicable procedures to produce com-
mercial batches to demonstrate that the process consistently performs as expected.
Multiple production-scale runs (3–6 are common) are performed as part of this exer-
cise. Once process qualification is successfully completed, and assuming success in
phase 3 clinical trials and FDA approval, commercial production ensues, and the
process is monitored throughout its lifetime. This monitoring is referred to as con-
tinued process verification, and its purpose is to ensure that the process remains in
a state of control. The process design, process qualification, and process verification
stages of the product life cycle are all part of a larger process validation program
that is discussed in greater detail in the next chapter [76].
There is no template for designing a manufacturing process that works for every
product. However, a general set of steps involved is shown in Figure 2.7. Discussion
that follows makes reference to this figure.
To begin process design, it is necessary to know the quality target product profile
(QTPP) and CQAs for the drug product because it is important to know the details of
the product for which the process is being designed. For example, CQAs related to
purity guide design of a chromatography step (e.g., choosing the type of resin to use
or the composition of an elution buffer) so that the impurities to be removed and the
required extent of removal are known in advance. Design of final fill-finish opera-
tions requires knowledge of the dosage form and delivery device, information that is
typically part of the QTPP. In addition, demand for the commercial product should
be understood so that production rates can be estimated. This information may influ-
ence choice of the expression system used as well as the scale of operations required.
 44

Phase 1 Phase 2 Phase 3 FDA

Drug Preclinical Commercial
Clinical Clinical Clinical Review and
Discovery Studies Manufacture
Trials Trials Trials Approval

cGMP Manufacturing

Process Continued
Process Design/Development

3–6 years 6–7 years for clinical trials + 1–2 years for FDA Review

Figure 2.6: Activities leading to the commercialization of a biopharmaceutical [78, 79]. Image © NC State University; reprinted with permission.
Chapter 2 An overview of processes and facilities for biopharmaceutical production
 2.2 Process design 45

Propose a manufacturing process

Establish design space

Identify control strategy

Figure 2.7: Steps involved in the design of a
biopharmaceutical process. Even though each step is shown
sequentially, there may be some iteration involved. For
Scale up example, design space studies may result in reconsideration
of the unit operations being used in the process. Image © NC
State University; reprinted with permission.

It is also worth noting that defining the QTPP and CQAs likely started prior to process
design, in the early stages of product development.
Typically, producing biopharmaceutical drug product from drug substance does
not involve a significant amount of manipulation. As a result, specifications for drug
substance and drug product are similar – with a few notable exceptions, such as a
sterility test for drug product but not (typically) drug substance.
Armed with an understanding of the product through the QTPP and CQAs, a
manufacturing process is proposed. A host cell system (when cells are used for
production) is selected along with other raw materials and consumable items, and
the unit operations to be used are identified. A preliminary list of process parame-
ters, a term we define shortly, is put together for each unit operation along with
an estimate of their allowable ranges. Proposing a process requires knowledge of
the process stages from Figure 2.1, the unit operations available to meet these ob-
jectives, processes for similar products, and any existing process and product in-
formation, including physical and chemical properties of the active product molecule
(or virus or cell). For example, selection of harvest operations is impacted by whether
the product is intracellular or extracellular. The selection of purification steps is
impacted by whether or not animal cells are used that require viral clearance
steps. The unit operations selected may change as process design proceeds and
process understanding deepens. Further, scalable operations, discussed previously,
are chosen when large quantities of product are being produced for commercial
distribution.
Once unit operations are proposed, including a list of process parameters for
each and their ranges, effort focuses on defining the design space, as shown in
46 Chapter 2 An overview of processes and facilities for biopharmaceutical production

Figure 2.7. Let’s start the discussion by covering a few definitions that help in under-
standing the concept:
– Process parameter, also referred to as an operational parameter: an input parame-
ter (e.g., the flow rate to a chromatography column) or process state parameter
(e.g., the temperature in a bioreactor) that is directly controlled. Process pa-
rameters can be further categorized as critical, key, and non-key. A critical pro-
cess parameter is one whose variability has an impact on a CQA. A key process
parameter is one whose variability is important to process performance but
does not affect a CQA. Non-key process parameters are easily controlled or have
wide limits and therefore pose little risk to impacting a CQA or performance pa-
rameter (defined below) [80].
– Performance parameter: an output parameter that cannot be directly controlled,
but is indicative of process performance [80]. Note that CQAs are not usually cat-
egorized as performance parameters. An example of a performance parameter is
the percentage step yield for a specific process step.

Every unit operation that makes up a biopharmaceutical process has inputs and
outputs. The most obvious are the product into and out of the manufacturing step.
In addition, as illustrated in Figure 2.8, process parameters and materials used in
an operation (and their attributes) are inputs to a processing step, and CQAs and
performance parameters are outputs. With this understanding we are in a position
to define design space. It is the multidimensional combination and interaction of
the inputs that have been demonstrated to provide assurance of quality [52]. Defin-
ing the design space for a process involves identifying process parameters and ma-
terial attributes that have an effect on CQAs and performance parameters as well as
determining the relationships that link these inputs to the outputs.

Process parameters
CQAs
(PP)
Unit operation
Material attributes Performance parameters
(MA) (that are not CQAs)

CQAs, Performance parameters = f(PP1, PP2, PP3,..., MA1, MA2, MA3,...)

Figure 2.8: A diagram illustrating the relationship between inputs – material attributes and process
parameters – and outputs – CQAs and performance parameters – for a single unit operation. Image
© NC State University; reprinted with permission.

Some form of risk analysis is often used to determine which of the process parame-
ters and material attributes identified as part of the proposed process have a high
probability of affecting CQAs. Details of risk analysis methodology are not covered
here, but the outcome would be an initial classification of process parameters and
 2.2 Process design 47

material attributes as critical, key, or non-key. Note that classification may change
as process knowledge is gained through process development activities. The rela-
tionships between the inputs (material attributes, process parameters) and outputs
(CQAs and performance parameters) are then established using a variety of methods,
including studies conducted using a one-factor-at-a-time approach, studies utilizing a
design of experiment (DoE) approach, and/or through the use of more fundamental
mechanistic models. DoE methods typically adjust multiple parameters at once, in
contrast with the one-factor-at-a time-method, resulting in fewer experiments and an
understanding of parameters that may interact. As you can ascertain, the development
activities to define the design space rely heavily on empirical studies. In addition, de-
velopment activities usually focus on one unit operation at a time. To keep costs and
time down, these studies are typically executed at small scale or even miniaturized
scale operated in high-throughput mode. The resulting design space is used to define
appropriate ranges for material attributes and process parameters to be used in pro-
cess batch records.
One of the challenges in defining the design space is the large number of pro-
cess parameters and material attributes that exist for some unit operations, in par-
ticular cell culture, fermentation, and chromatography steps. Let’s consider the
specific example of a cell culture step in the production bioreactor for a mAb in
CHO cells, as might be used in the process shown in Figure 2.3. The A-Mab case
study [62] provides a comprehensive example of how to design a process to pro-
duce A-Mab, a fictitious humanized IgG1 mAb intended as treatment for non-Hodg-
kin’s lymphoma. Figure 2.9 shows a list of process parameters and material attributes
considered in developing the design space to ensure that applicable CQAs and perfor-
mance parameters are met for the production bioreactor step. This list was deter-
mined from knowledge of the production of other related mAbs, production of A-Mab
for preclinical studies and phase 1 and 2 clinical trials, and previous process optimiza-
tion studies for A-Mab. Risk analysis was then used to assess the potential of each
process parameter to affect a CQA or performance parameter. The results of this exer-
cise are also shown in Figure 2.9. Reducing the number of parameters in this way
makes design space studies more manageable. Following risk analysis, DoE studies
were performed using an appropriate scale-down bioreactor model to determine pro-
cess parameters that deserved further consideration and to define the relationship
between these process parameters and CQAs. Plots showing the resulting design
space are also shown in Figure 2.9. Specifically, those plots show values for each
process parameter ultimately determined to potentially impact CQAs – osmolality,
dissolved CO2, temperature, pH, culture duration – and the probability that all the
quality attributes included in the model will be within the acceptable limits. This
example shows that defining the design space – particularly for unit operations
like a production bioreactor step that have a large number of potential process pa-
rameters – can be a complex undertaking.
48 Chapter 2 An overview of processes and facilities for biopharmaceutical production

(a) Process parameters, material attributes

• Inoculum viable cell concentration
• Inoculum viability
• Inoculum in vitro cell age
• Osmolality Critical quality attributes
• Antifoam concentration • Product aggregate concentration
• Nutrient concentration in medium • aFucoylsation
• Medium storage temperature • Galactosylation
• Deamidation
• Host cell protein concentration
• Medium age • Host cell DNA concentration
• Timing of feed addition
• Volume of feed addition
Performance parameters
• Component concentration in feed
• Timing of glucose feed addition • Product yield

• Amount of glucose fed • Cell viability at harvest

• Dissolved oxygen • Broth turbidity at harvest

• Dissolved carbon dioxide

• Temperature
• pH
• Culture duration (days)
• Remnant glucose concentration

(b) Perform risk analysis to determine

those process parameters likely to
have an impact on CQAs

Process parameters considered in

the design space study

• Inoculum viable cell concentration

• Osmolality
• Nutrient concentration in medium
• Volume of feed addition Perform DoE
studies to
• Component concentration in feed determine
• Dissolved oxygen design space
• Dissolved carbon dioxide
• Temperature
• pH
• Culture duration (days)
• Remnant glucose concentration

Figure 2.9: Defining the design space for the production of A-Mab in CHO. (a) is a list of process
parameters, material attributes, CQAs, and performance parameters initially considered. (b) traces
 2.2 Process design 49

With an understanding of the design space, defining the control strategy is the next
design step as shown in Figure 2.7. The goal of the control strategy is to minimize
variability, which can be significant, and ensure that CQAs for drug substance and
drug product are met. Elements of a control strategy typically include the following:
monitoring process parameters and controlling them within acceptable ranges, moni-
toring performance parameters and ensuring acceptable ranges are met, testing and
accepting raw materials, sampling and testing process intermediates to ensure ac-
ceptable ranges are met, maintaining and calibrating equipment used in manufactur-
ing, and release testing on drug substance and drug product. A key aspect of the
design of the control strategy is setting appropriate ranges on the critical process pa-
rameters and controlling within the set ranges. For the A-Mab study, the authors [62]
ultimately developed one inequality for each CQA limit that describes the combina-
tion of process parameter values that ensure that each CQA limit is met. Here we take
a less rigorous approach and propose a possible set of ranges for each of the critical
process parameters, based on evaluation of the curves in Figure 2.9, that ensure with
greater than 99% probability that product from the bioreactor step would meet CQA
limits:
– pH: 6.85–7.10
– Temperature: 34.5–35.5 °C
– Dissolved CO2: 100–160 mm Hg
– Osmolality: 360–400 mOsm
– Culture duration: 15–17 days

These parameters could be controlled by a combination of process instructions,

such as a batch production record, and equipment control features. It is also impor-
tant to note that ranges are typically specified for key and even some non-key pro-
cess parameters to ensure process consistency from batch to batch of product.
The last step in Figure 2.7 is scale-up, required because the studies used to de-
fine the design space are typically conducted at a scale significantly smaller than
production. There are different approaches to scale-up for the different unit opera-
tions; these approaches are discussed in subsequent chapters. Generally speaking,
scale-up is the process of transferring (1) material attributes and process parameters
and (2) equipment dimensions from a small scale to larger scale to increase through-
put for commercial production while preserving the quality of performance achieved
at small scale.

Figure 2.9 (Continued)

the process of reducing the number of process parameters considered to arrive at the design space
through risk analysis and characterization studies [62]. Note that regions in dark red have > 99%
probability of meeting all CQA limits. Image (a) and (b) © NC State University; reprinted with
permission. Graphs in image (b) reprinted from “A-Mab: a Case Study in Bioprocess Development,”
by CMC Biotech Working Group, p. 84. 2009.
50 Chapter 2 An overview of processes and facilities for biopharmaceutical production

Also worth noting is that the development of platform processes can greatly re-
duce the cost and increase the speed of process design and potentially improve
quality of the biopharmaceutical drug products manufactured. The term platform
refers to the use of a common or standard method, equipment, procedure or work
practice applied to the development or manufacture of different products [81]. Prod-
ucts with active ingredients that are similar lend themselves to platform processes.
Examples include production of mAbs or RNA vaccines. In both cases, development
of a platform is enabled by the physical and biochemical similarities in the product
being produced; for example, mAb products based on IgG1 antibodies all have simi-
lar chemical and physical properties. Therefore, upstream and downstream processes
can be similar from one mAb-based product to another, which enables more rapid
and less costly process design once the platform is in place.

2.3 Manufacturing facilities

In the beginning of this chapter, we emphasized the importance of facility design in

meeting product quality and quantity requirements. In this section, we review some
of the basic facility design considerations required to ensure that these quality and
quantity requirements are met.
At the most basic level, facilities must be designed to house all of the equip-
ment required for both the main process and for supporting activities. Therefore,
the processes (including supporting activities) housed within the facility and the
amount of product to be produced, which dictates the scale of operations, will impact
the final design. In addition to the supporting activities previously mentioned,
space will be needed for other support activities, including a warehouse, weigh
and dispense areas, quality control labs, and office space for groups such as pro-
cess engineering, quality, and environmental health and safety. Because product
types, production capacity, and processes differ from one facility to another, facil-
ities come in a variety of shapes and sizes. Further, the facility for manufacturing
drug product (from drug substance) is often separate from drug substance manufactur-
ing facilities.
A clear understanding of the design of the process and demand for the product
is required before embarking on designing the facility. Some specific examples of
how the design of the process impacts the design of the facility include the follow-
ing. A process designed with completely single-use systems will not require space
for CIP skids for cleaning because equipment cleaning between batches or product
campaigns is not required. Space for clean steam generators for SIP steps would
also not be required because those single-use components are generally received
sealed and sterile. Likewise, a process designed for continuous operation would
lead to a relatively small facility due to the reduced equipment footprint relative to
a process for the same product designed for batch operation. Facility design also
 2.3 Manufacturing facilities 51

requires a clear understanding of the demand for the product to allow for proper
equipment sizing, which will determine the space required to house the equipment.
There are a number of considerations in the design of the facility that focus on
“protecting the product,” a paradigm that is often referred to in facility design liter-
ature. Protecting the product is most often used to refer to steps taken to keep mi-
crobial contamination out of the process and to avoid cross contamination. Let’s
consider each separately.
Microbial contamination refers to contamination of product by microorganisms
such as bacteria and fungi. We discussed in Chapter 1 the need for biopharmaceu-
tical drug products to be sterile, given that they are typically delivered parenter-
ally. Therefore, the need to control bioburden during production of sterile drug
product from drug substance is clear. But even in the manufacturing steps for
drug substance upstream of drug product, the need for bioburden control exists.
Drug substance manufacture involves aqueous systems at temperatures and pH
ranges conducive to growth of microorganisms. Further, there are many potential
sources of microbial contamination within facilities. Generally, these sources in-
clude people, the process (e.g., equipment and raw materials), the facility itself
(e.g., viable and nonviable particulate in room air), and clean utilities. Therefore,
biopharmaceutical process streams are susceptible to microbial contamination.
There is a concern that this contamination may degrade product by release of mi-
crobial enzymes or produce exotoxin, endotoxin, and other by-products that may
not be removed by subsequent processing steps and consequently create safety
issues for patients [82, 83].
The potential for cross contamination also exists within a process and can
take many forms. For example, product A contaminating product B in multi-product
facilities is a form of cross contamination, as is batch 001 of product A contaminat-
ing batch 002 of the same product. Product intermediate from the bioreactor
step making its way into intermediate from a chromatography step would like-
wise be considered cross contamination. There are other types of contamination
as well, such residues like lubricants and oil making their way into product
from equipment.
Before getting into details of facility design features that protect the product,
it’s worth considering closed versus open processing, as the implementation of one
versus the other can have a significant impact on facility design considerations re-
lated to contamination. Definitions for each are given below:
– Closed system. A system designed and operated so that product is not exposed
to the surrounding environment [84]. There are numerous examples in the in-
dustry of steps that have been closed, including bioreactor steps, fluid transfer
systems, and bulk filling steps. Expectations for facility design features that
minimize contamination are lessened in areas that house closed systems.
– Open system. A process that is not closed and therefore must be performed in
an environment where the probability of contamination is low. For example,
52 Chapter 2 An overview of processes and facilities for biopharmaceutical production

chromatography and ultrafiltration steps in downstream processes are often

open. When connections are made to the chromatography system in Figure 2.4,
the lines connected are typically filled with room air; thus, a product-contact
surface has been exposed to the environment. If after making all connections,
the chromatography skid were sterilized, or all flow paths were initially sterile
and connected to feeds (i.e., solutions and product load required for the step)
in a way to maintain sterility, then the system could be considered closed.

There are a number of ways to close a system. Those details are beyond the scope of
this book, but it is important to recognize that a claim of system closure must be
adequately validated. For more details on closure methods and demonstration of
closure, the ISPE Baseline Guide Biopharmaceutical Manufacturing Facilities is a
good resource [84].
Numerous design practices are used in biopharmaceutical manufacturing facilities
to mitigate the risk of both microbial and cross contamination of product. Table 2.1
describes some of these. For processes that use reusable equipment, space for CIP sys-
tems and clean steam generation must be included. Because personnel are a main
source of microbial contamination, gowning areas are needed within the facility to im-
plement suitable gowning procedures. Gowning refers to the apparel that staff enter-
ing a facility must wear to protect the product. Note that the more closed a process is,
the less stringent the gowning requirements are. There are also a number of design
features that fall under the category of segregation, which can be spatial, temporal,
procedural, or carried out by environmental control. In the context of this discussion
on facility design, spatial and environmental are most common. Open processes are
executed in cleanrooms designed to keep particulate – both viable and nonviable –
concentration in the surrounding environment low. This type of segregation would be
categorized as environmental control. There are also a variety of spatial segregation
concepts listed in Table 2.1 that can be implemented to reduce the likelihood of con-
tamination. Air locks are designed as a buffer that keeps air from flowing from less
clean to cleaner areas. Separating each unit operation into its own dedicated space is a
common approach for reducing risk of cross contamination. In a multi-product facility,
clear physical separation of the process for product A from product B may be used.
Because cleanrooms are a particularly important method of environmental con-
trol, we provide a few more details here. A cleanroom is a room in which the con-
centration of airborne particles is controlled and that is constructed and used in a
manner to minimize the introduction, generation, and retention of particles inside
the room [85]. Cleanrooms are achieved by:
– Filtering air entering from outside with high-efficiency particulate air (HEPA)
filters.
– Maintaining a positive air pressure relative to less clean surrounding space so
that air flows from cleaner to less clean rooms. A pressure differential of 0.02–
0.08 inches of water is recommended [86].
Table 2.1: Facility design features used to protect product.

Design Feature Microbial or Required for Open Segregation Comments

Cross Systems, Closed Systems, Category
Contamination or Both

Space for CIP skid and parts Cross Both Enables Needed to house equipment necessary for cleaning
washing systems procedural process equipment and small parts. Unnecessary for
single-use equipment.

Space for clean steam Microbial Both Enables Needed to house equipment required to generate steam
generation procedural for sterilization. Unnecessary for single-use equipment
that comes sterilized.

Gowning areas Both Both Enables Allows for gowning procedures to be implemented,
procedural which are critical to microbial contamination control.
Less gowning is likely required when using closed
systems.

Cleanrooms Microbial Open Environmental Rooms in which the concentration of airborne

control, particulate is controlled, primarily through filtration of
procedural outside air.

Appropriate facility finishes Microbial Open Environmental Enables effective cleaning and disinfection of
control, manufacturing spaces and reduces risk of microbial
procedural growth. Examples include floors, walls, and ceilings
with smooth, hard surfaces.

Air locks Microbial Open Environmental, Separates areas of different cleanliness levels by
spatial creating a buffer between clean and less clean areas.
2.3 Manufacturing facilities

(continued)
53
Table 2.1 (continued)
54

Design Feature Microbial or Required for Open Segregation Comments

Cross Systems, Closed Systems, Category
Contamination or Both

Separate space for individual Cross Open Spatial Reduces risk of contaminating one step of a process
unit operations for one with product intermediate from another step. The more
product of the process that is closed, the less separation into
different rooms required. For example, if all upstream
and downstream processing steps were closed, they
could all take place in the same room.

Separate space for processes Cross Both Spatial Reduces risk of contaminating product A with product
for different products running B. Multi-product facilities can also be operated using
concurrently temporal segregation that relies on equipment cleaning
and changeover procedures to minimize risk of cross
contamination between products.

Separating pre-viral Microbial Open Spatial Reduces risk of contaminating product intermediate
clearance and post-viral that has undergone viral clearance with product
clearance operations for one intermediate that has not yet undergone viral
product clearance.

Separating supply and return Cross Open Spatial Combined with appropriate flow of personnel,
corridors equipment, and materials, reduces risk of cross
contamination.
Chapter 2 An overview of processes and facilities for biopharmaceutical production
 2.3 Manufacturing facilities 55

– Including air locks within the facility. An air lock is a room that typically has
two doors in series to separate a cleanroom from a less clean environment, such
as a corridor. The two doors are typically interlocked to avoid being opened at
the same time [87].
– Using appropriate facility finishes. For example, floors, walls, and ceilings should
have smooth hard surfaces that are easily cleanable and do not harbor microbial
contamination.

There are several cleanroom classification systems in use. The two most common are
systems from the International Organization for Standardization (ISO), often used in
the United States, and the European Union (EU) system. The specific classification
under each system is based on the number of particles of a particular size permit-
ted per volume of air. For example, a cleanroom meeting ISO 5 classification would
have no more than 3,520 particles/m3 of size greater than or equal to 0.5 µm. The
corresponding classification in the EU system is Class A. In the ISO system, higher
numerical designations mean a higher particle concentration in the environment;
for example, ISO 8 allows a higher maximum concentration for a given size particle
than ISO 5, 6, or 7 classifications. In the EU system, higher letters correspond to a
higher particle concentration; for example, Grade D allows for a higher particulate
concentration than Grades A, B, and C. Specific limits for particulate concentration
in the EU classification system can be found in Annex 1 of the EudraLex Volume 4
[88]. For the ISO system, the limits are contained within the ISO 14,644 standard
[85]. Generally speaking, rooms get cleaner farther downstream in the process, as
the product gets closer to drug product.
To illustrate how the facility design concepts in this section are put into prac-
tice, consider Figure 2.10, which shows a facility layout for production of protein
therapeutic drug substance. Take note of the following features of the design, which
follows the list provided in Table 2.1:
– The layout features space for both the main process and supporting activities. There
are rooms for media prep/storage and buffer prep/storage. There is also space in-
cluded for a parts washer, autoclave, and/or CIP skid to clean equipment. And there
is space for generating clean utilities such as HPW, WFI, and clean steam.
– Spaces for support activities are located close to the process areas they support.
This is referred to as defining adjacency. For example, buffer prep and storage
are located close to the DS processing areas in which the buffers are used.
– Gowning rooms for both the upstream and downstream processing areas are in-
cluded so that gowning procedures can be implemented.
– Support activities and the main process are conducted in cleanrooms. Cleanrooms
from ISO 8 down to ISO 5 are included. Note that generally, as the process moves
toward drug substance, cleanroom classification becomes more stringent.
 56

Return corridor (ISO 8)

AL

US staging
out (CNC) AL
AL

Gowning US Equip clean Media Media prep Inoc prep Cell culture Harvest
(CNC) US storage (ISO 8) (ISO 7) (ISO 8) (ISO 8)
(ISO 8) (ISO 8) (open steps
ISO 5)

Central hall (CNC)

Return corridor (ISO 8)

AL AL AL Clean utility
US staging
in (CNC) generation
AL
Supply corridor US (ISO 8) AL

Supply corridor DS (ISO 8)

AL AL

DS staging Harvest
in (CNC) AL AL AL AL staging
(CNC)

Gowning DS Equip clean Buffer Buffer prep BDS

(CNC) DS storage (ISO 8) 2 1
(ISO 8) (ISO 8) (ISO 7) (ISO 7) (ISO 7)

AL AL AL AL
DS staging
out (CNC)
AL
Return corridor (ISO 8)
Chapter 2 An overview of processes and facilities for biopharmaceutical production

Figure 2.10: A facility layout for production of protein therapeutic drug substance, such as a mAb,
with open processes and reusable equipment. Red arrows represent the product path through the
 2.3 Manufacturing facilities 57

– Rooms with a cleaner classification are separated from areas with a less clean
classification by an air lock.
– Separate rooms have been included to separate processing steps. The two purifica-
tion suites can be used to house operations that are pre- and post-viral clearance,
respectively. Note that two purification suites may not be enough to separate every
viral clearance and chromatography step, so each purification suite may contain
multiple unit operations. For example, Purification 2 may house the final two chro-
matography steps (post viral clearance) in the process. In this case, two batches of
the same product – one being processed with the first (of the final two) chromatog-
raphy steps in the space and one process with the second – would not be allowed
in the room at the same time. This type of segregation is temporal.
– Separate supply and return corridors exist.

Note also in Figure 2.10 that product flow is unidirectional, always moving forward,
which minimizes the likelihood of a “purer” form of the product intermediate being in
the same space as a less “pure” form. Like product, flow of people, materials, and equip-
ment is typically unidirectional. “Clean” people, materials, and equipment enter the fa-
cility and move through it via the supply corridor, also referred to as the clean corridor.
Personnel would enter into the room in which they work, execute their tasks, and then
leave the room and enter the return corridor, sometimes referred to as the dirty corridor.
Note that if multiple products are to be produced within a facility that uses open
processing, either spatial or temporal segregation may be used to keep products sepa-
rated. To segregate spatially, many of the process areas in Figure 2.10 would have to be
replicated within the facility. Temporal segregation would involve running processes
for products A and B, for example, in campaigns (i.e., a series of runs to produce a
specified amount of one product) separated by time to avoid cross contamination. Be-
tween campaigns, equipment would go through changeover. Changeover refers to the
replacement of wetted soft parts on equipment – such as gaskets, O-rings, and valve
diaphragms – to avoid the possibility of release of one product that may have been
absorbed into the soft parts into the next product.
Much of what has been discussed previously in this section has focused on protect-
ing the product. One other consideration with the potential to impact facility design is
biohazard containment. When working with some biological agents, steps must be
taken to protect the worker and the environment. These steps can involve process
design, facility design, and procedures. Biosafety refers to the degree of precautions
required to isolate dangerous biological agents. The Centers for Disease Control and

Figure 2.10 (Continued)

facility. CNC stands for controlled not classified. AL stands for air lock. BDS is bulk drug substance,
and US and DS stand for upstream and downstream. Image © NC State University; reprinted with
permission, and adapted from Biopharmaceutical Processing Development, Design, and
Implementation of Manufacturing Processes, 2018 [74].
58 Chapter 2 An overview of processes and facilities for biopharmaceutical production

National Institutes of Health have developed four biosafety levels (BSL): BSL 1, 2, 3,
and 4 [89, 90] and the levels are designated before work with the infectious agent
actually begins. BSL 1 applies to agents posing the least risk to human health and
have the least stringent safety requirements. BSL 4 applies to biological agents that
pose the highest risk to human health and requires the most stringent containment
measures. Similar concepts are in place in most countries. The final assignment of
biosafety level for a particular agent should be done through risk analysis as de-
scribed in the U.S. Department of Health and Human Services’ publication Biosafety
in Microbiological and Biomedical Laboratories [89]. From that document, the primary
risk criteria used to define which of the four ascending levels of containment to assign
are infectivity, severity of disease, transmissibility, and the nature of the work being
conducted. Generally speaking, facility design and procedures put in place in a BSL-1
facility will not differ significantly for what is designed for standard GMP operations.
Examples of agents in which BSL-1 is likely to apply include CHO cells and E. coli
strains typically used for biopharmaceutical manufacturing. Agents typically requiring
BSL-2 facilities include influenza virus, commonly used in production of flu vaccines,
and lentivirus, commonly used in production of gene and cell therapies. A BSL-2
facility requires limited access to the space and physical containment devices,
such as a biosafety cabinet, for all manipulations of agents that cause splashes or
aerosols of infectious materials.
There are numerous resources that provide more depth on the topic of facility
design than we have been able to cover in this chapter [84, 91]. Please refer to them
for additional details.
Finally, rather than building a facility, a biopharmaceutical company may choose
to outsource its manufacturing to a contract manufacturing organization. Outsourc-
ing is common and growing [92]. Companies may choose outsourcing because they
lack necessary existing facilities or lack funding to build a new facility, which can
range from $200 MM to $500 MM [93]. Other reasons to outsource include insufficient
qualified personnel in-house, pressure to reduce time to market, and alignment with
a company’s business model (e.g., a virtual company).

2.4 Summary

Well-designed biopharmaceutical processes and facilities are essential to ensuring

that drug product is manufactured with the intended quality and in the required
quantity. This chapter provided an overview of both. Most biopharmaceutical pro-
cesses require four stages: production, harvest, purification, and formulation/fill. Each
stage has a common objective carried out by multiple unit operations.
– In the production stage, the product is synthesized. For most biopharmaceutical
products, this involves cell growth in bioreactors. At the end of this stage, the
cells are part of a broth that includes spent aqueous media.
 2.4 Summary 59

– In the harvest stage, the product is separated from the production system. For intra-
cellular products, a cell lysis step such as homogenization must be used to release
the product into the surrounding liquid. Regardless of whether product is intracellu-
lar or extracellular, removal of solids from the aqueous process stream – either cells
or cell debris – is required. This can be accomplished by a variety of unit operations,
including centrifugation, depth filtration, and tangential-flow microfiltration. By the
end of this stage, the clarified process stream with the product in aqueous solution
may be essentially free of solid particles, but it still contains numerous soluble im-
purities that must be removed for the sake of patient safety.
– The purification stage removes these impurities most often by chromatography. Pro-
cesses that rely heavily on mammalian-derived components, such as CHO-based
processes for mAb production, also require viral clearance steps aimed at inactivat-
ing or removing viruses from the process stream. Upon completion of the purifica-
tion stage, the product is still in a clarified liquid solution, but that solution is now
free from most of the soluble impurities that were present at the harvest stage. How-
ever, the buffer system in which the product is dissolved is dictated by the last chro-
matography step and is likely not what is needed for drug product.
– During the formulation/fill stage, the product matrix is adjusted through unit op-
erations such as ultrafiltration, other excipients may be added, and the product
aseptically filled into its final container to produce the drug product described
in detail in Chapter 1.

Generally speaking, steps involved in production of drug substance and drug product
are operated in batch, rather than continuous, mode. And drug substance is a low-
bioburden intermediate, while drug product is sterile. It is important to note that not
all drug substance is produced in exactly the same way. In this chapter, we provided
details in Figure 2.3 on a process to produce a mAb product. Processes for other prod-
ucts may look different. For example, Gamunex®-C is an immune globulin injection
produced by purifying IgG from large pools of human plasma [59]; thus, product is
synthesized in humans rather than in cells grown in a bioreactor. Kymriah®, is an ex
vivo gene therapy described in Chapter 1. T cells from a patient are transduced with a
lentiviral vector, expanded in culture, washed to remove impurities, formulated for
cryopreservation, and then reinfused into the patient [23]. Thus, the final product is a
suspension of modified cells rather than an active ingredient dissolved in an aqueous
solution. Additionally, the original cells required for production are taken from the
patient. The processes required for drugs like Gamunex®-C and Kymriah® would obvi-
ously be different than those for a mAb.
Equipment for biopharmaceutical processes is often skid mounted; that is all parts
needed to support a given activity (e.g., chromatography) are mounted to a frame that
is often on wheels. Further, there are processes that use reusable, typically stainless
steel equipment, others that rely on single-use equipment (scale permitting), and
others that use both. Additional equipment details are presented in subsequent
60 Chapter 2 An overview of processes and facilities for biopharmaceutical production

chapters. The numerous activities required to support the main process were also dis-
cussed in this chapter. These include solution and media preparation, clean-in-place
and steam-in-place operations, and operations to generate clean utilities like high pu-
rity water, water for injection, and clean steam.
There are many steps involved in the design of a process for biopharmaceutical
production. Understanding what the product is – through the QTPP and CQAs – is
an important first step in the design process. From there, significant effort is de-
voted to bench-scale studies to develop the design space. The design space provides
the relationship between inputs, such as material attributes and process parame-
ters, and outputs, such as CQAs and performance parameters, for a given unit oper-
ation or series of unit operations. Design space definition informs a control strategy.
Central to that control strategy are batch production records that provide instruc-
tions and tracking of data for the process.
The facility in which the central process and supporting activities are conducted
is also key to ensuring product quality and quantity needs are met. A paradigm cen-
tral to facility design is to protect the product. To meet this objective, a good design:
– Allocates space for CIP skids and parts washing systems
– Allocates space for clean steam generation
– Includes air locks and gowning areas
– Includes cleanrooms for processing
– Uses appropriate facility finishes
– Provides separate spaces for individual unit operations for one product
– Provides separate space for processes for different products running concurrently
– Separates pre-viral clearance and post-viral clearance operations for one product
– Separates supply and return corridors

The extent to which these design features are implemented depends on whether a
process or processing step is open or closed. A validated closed process – one that
is designed and operated so that product is not exposed to the surrounding environ-
ment – allows for much more flexibility in design. For example, a validated closed
system does not require operation in a clean room or segregation of unit operations for
a given product.

2.5 Review questions

1. The stage 3 and 4 steps used to produce Xembify®, a 20% IgG solution for treatment
of Primary Humoral Immunodeficiency, is shown below as reported by Alonso et al
[94].The IgG for Xembify® is from human plasma, the starting material for the
manufacturing process. List all steps used in the production of Xembify® by stage,
keeping in mind that the steps in the diagram below apply only to stages 3 and 4.
 2.5 Review questions 61

Plasma collected from healthy donors

Thaw and pool plasma

Fractionation (II + III)

Caprylate precipitation,

Anion exchange chromatography

Concentrate to 10% protein

Additional glycine added

Concentrate to 20% protein

pH adjusted (to 4.3) and

Polysorbate 80 added

Low pH incubation

Figure 2.11: Stage 3 and 4 steps used to produce Xembify®. Image © NC State University; reprinted
with permission.
62 Chapter 2 An overview of processes and facilities for biopharmaceutical production

2. Human embryonic kidney 293 (HEK293) cells are used for production of adeno-
associated virus serotype 2 (AAV2), a viral vector commonly used for gene ther-
apy. Because a significant amount of the AAV2 produced is intracellular, cells
must be lysed within the bioreactor using Triton X-100, a nonionic surfactant.
The concentration of capsids containing the gene of interest is 1.11 × 1011 vector
genomes (vg)/mL, while the concentration of vector genomes in the filtrate
from the depth filter is 1.00 × 1011 vg/mL. The vector genomes are measured
using quantitative polymerase chain reaction (qPCR).
a) If the volume fed to the depth filter is 50 L and the volume of filtrate col-
lected is 48 L, calculate the step yield.
b) List three issues that could contribute to the step yield being <100%.

3. The table below shows the step yields for the mAb process shown in Figure 2.3.

Table 2.2: Step yields for the mAb process shown in Figure 2.3.

Step % Step Yield of Target Protein

Centrifugation/depth filtration 

Protein A chromatography 

Low pH viral inactivation 

Cation exchange chromatography 

Anion exchange flowthrough chromatography 

Virus filtration 

Concentration/buffer exchange by ultrafiltration 

Bulk Fill 

Protein production is extracellular. The 5,000 L bioreactor step used to produce

the protein has a titer of 4 g mAb/L bioreactor volume.
a) What is the % process yield, through bulk drug substance?
b) How many grams of product are bulk filled?
c) The final concentration of product is 110 mg mAb/mL. You fill 1 L bags to a
volume of 0.9 L during the fill. How many bags are required to perform the fill?

4. You are working on the design of a process and facility to produce 2 metric tons
of a mAb product (as drug substance) annually. Information about the process
is given below:
– Titer in the bioreactor is 5 g/L
– Production bioreactors are on a 12-day cycle (10 days for production, 2 days
for turnaround)
 2.5 Review questions 63

– DS process is on a 3-day cycle

– % yield for the DS process is 70%
a) How many kg of mAb will have to be produced in the bioreactor annually
for a 2 metric ton production rate (drug substance)?
b) What total volume of bioreactor fluid is required?
c) How many bioreactors are required to keep the DS process running nonstop
(i.e., to feed cell culture broth to DS every 3 days)? Assume you will operate
365 days/yr, but then explain why this is likely an overestimate.
d) How large will each bioreactor be? Will it be stainless steel or single use?

5. The image below shows different rooms planned for a multi-product facility
(Products A and B) for production of drug substance by open processing.
a) Arrange the rooms in an order that aligns with the design concepts pre-
sented in this chapter.
b) Draw arrows representing the product flow through the facility.
c) Draw arrows representing non-product flows, such as equipment, media,
and solutions.
d) To reduce size and cost of a new facility, you choose to have only one pro-
cess train that can be used for both Products A and B. Discuss differences
and similarities between the two facilities (i.e., one with equipment dedi-
cated to each product, and another with equipment shared between prod-
ucts) related to design features, operation, and cost.

Product A Product B

Recovery Fermentation Media prep/storage

Fermentation Recovery Buffer prep/storage

Inoculum preparation Equipment cleaning

Inoculum preparation

Figure 2.12: Rooms in a multi-product facility. Image © NC State University; reprinted with
permission.

6. Describe five changes that could be made and their advantages to the facility
design in Figure 2.10 if the process for manufacturing drug substance was closed
rather than open and all equipment was single use.
Chapter 3
Good Manufacturing Practice (GMP)
for biopharmaceutical production

As discussed in Chapter 1, biopharmaceutical drug products must have quality; that

is, they must be suited to their intended use. In Chapter 2, we learned about meeting
quality requirements through the biopharmaceutical manufacturing process and
design of the manufacturing facility. This chapter focuses on another element for
ensuring product quality, Good Manufacturing Practice (GMP).
Questions explored in this chapter include:
– What is GMP and how is it defined?
– What role do people play in meeting GMP requirements?
– What are the GMP expectations for people and for equipment used for biopharma-
ceutical processing?
– Why are documents so important to biopharmaceutical manufacturing, what
types of documents are used, and what are the expectations regarding good doc-
umentation practice?
– What role does analytical testing by the Quality Control lab play in GMP produc-
tion of biopharmaceuticals?
– What does it mean to qualify equipment, systems, and utilities? And what is the
purpose cleaning validation and process validation, and how are these activities
carried out?
– What is involved with release of a biopharmaceutical batch after manufacture?
– How does GMP apply even after biopharmaceutical products are on the market?

3.1 GMP regulations – an overview

To understand the importance of quality as it relates biopharmaceuticals, it is first im-

portant to acknowledge the human element that is involved. We described in Chapter 1
that many of the biggest selling drugs worldwide are biopharmaceuticals; therefore,
their use directly helps many patients. However, to be of benefit to patients, biophar-
maceuticals products must be safe, efficacious, or, more generally, have quality. While
it is indisputable that biopharmaceuticals made with quality benefit people, it is also
important to recognize that it is people who design and execute the manufacturing pro-
cesses and design and build facilities for production. The way in which these activities
are conducted impacts product quality, and this chapter focuses on the standards that
ensure appropriate systems are in place to manufacture products that truly benefit
patients.

https://doi.org/10.1515/9783110616880-003
 3.2 cGMP regulations/guidelines and guidance documents 65

Good Manufacturing Practice (GMP) standards define the minimum requirements

that must be met to ensure that products are consistently produced with the required
safety, identity, strength, purity, and quality (SISPQ). They achieve this by laying out
expectations for proper design, monitoring, and control of manufacturing processes
and facilities. GMP is designed to minimize the risks involved in biopharmaceutical pro-
duction, risks that cannot be eliminated through testing of a small sample of the final
product alone. Further, their focus is on what to do, but not specifically how to do it.
While GMP addresses general practices to be followed across the biopharma-
ceutical industry, new innovations constantly arise that call for additions and/or
modifications to these practices. Instead of perpetually rewriting GMP standards,
biopharmaceutical manufacturers may need to adapt their processes and practices
to align with the current methods related to manufacturing, testing, and quality sys-
tems implementation that offer improved assurance of product quality. Because of
this, GMP is often referred to as cGMP – current Good Manufacturing Practice. These
terms are used interchangeably in the remainder of the book.
Just as cGMP requirements change over time, they may also vary from country to
country, although at a high level, basic cGMP concepts are similar throughout the
world. Examples of regulatory agencies outside of the United States include the Euro-
pean Medicines Agency (EMA), the Pharmaceuticals and Medical Devices Agency in
Japan, or Health Canada.
With so many different biopharmaceutical products being produced, variation in
cGMP expectations from country to country, and a variety of agencies enforcing cGMP,
a company has many considerations in how best to implement cGMP. It is the respon-
sibility of the drug manufacturer to not only define the methods for meeting cGMP re-
quirements but to justify their approach to the appropriate regulatory agencies. Failure
to meet cGMP requirements can result in regulatory observations that a company must
remediate prior to receiving approval for marketing in a particular country. Other en-
forcement actions that may be taken by regulatory agencies when cGMP is not fol-
lowed include seizure of product, injunctions or court orders preventing individuals or
companies from specific actions that violate cGMP, fines, and for the most serious of-
fences, criminal prosecution. Understanding the various cGMP requirements through-
out the world is therefore especially important so that companies can comply.

3.2 cGMP regulations/guidelines and guidance documents

Examples of cGMP regulations and guidelines from various regions throughout the
world are presented in Table 3.1.
In addition to the various cGMP regulations/guidelines in the biopharmaceutical
industry, numerous guidance documents are available to help companies determine
the best practices to use for achieving cGMP requirements. While there may be legal
ramifications for not following the regulations and guidelines, guidance documents are
66 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

Table 3.1: Examples of cGMP guidelines and guidance documents.

Selected cGMP Regulations/Guidelines Relevant Guidance Documents

throughout the World

 CFR Part  – Current Good Manufacturing ICH Q – Good Manufacturing Practice for Active
Practice in Manufacturing, Processing, Packing Pharmaceutical Ingredients
or Holding of Drugs (United States)

 CFR Part  – Current Good Manufacturing ICH Q(R) – Pharmaceutical Development
Practice for Finished Pharmaceuticals
(United States)

EudraLex Volume  – Pharmaceutical legislation ICH Q – Development and Manufacture of

for medicinal products for human use (European Drug Substances
Union)

EudraLex Volume  – Guidelines for good United States Pharmacopeia – National

manufacturing practices for medicinal products Formulary (USP-NF)
for human and veterinary use (European Union)

Good Manufacturing Practices (GMP) Guidelines European Pharmacopoeia (Ph. Eur.)

for Active Pharmaceutical Ingredients
(GUI-) (Canada)

Ministerial Ordinance on Standards for Quality FDA Guidance for Industry – Process Validation:
Assurance for Drugs, Quasi-Drugs, Cosmetics General Principles and Practices
and Medical Devices (Japan)

Ministerial Ordinance on Standards for FDA Guidance for Industry – Sterile Drug
Manufacturing Control and Quality Control for Products Produced by Aseptic Processing –
Drugs and Quasi-Drugs (Japan) Current Good Manufacturing Practice

not usually enforceable by law. They do, however, provide a level of detail not typical
in regulations and guidelines that is helpful for determining how to put cGMP into
actual practice. Regulatory agencies such as those mentioned above publish guidance
documents in addition to regulations. The International Council for Harmonization of
Technical Requirements for Pharmaceuticals for Human Use (ICH) is an organization
that provides guidance by bringing together requirements from multiple regulatory
agencies and establishing harmonized guidance that can be applied worldwide. A few
relevant ICH documents are listed in Table 3.1 along with other cGMP guidance docu-
ments well known in the biopharmaceutical industry. Additional information about
best practices for implementing cGMP can be found through publications from trade
organizations and industry associations such as the International Society for Phar-
maceutical Engineering (ISPE) and the Parenteral Drug Association (PDA). Both
organizations are good sources for guidance documents related to biopharmaceutical
manufacturing in general, including cGMP.
cGMP regulations and guidelines encompass all aspects of biopharmaceuti-
cal manufacturing, including personnel, facilities and equipment, documentation,
 3.3 Notable cGMP requirements for biopharmaceutical manufacture 67

production, quality control, and quality systems. Chapter 2 provided an overview of fa-
cility design considerations to ensure product quality requirements and conformance
to cGMP. We examine some other cGMP expectations in the following sections.

3.3 Notable cGMP requirements for biopharmaceutical

manufacture

In this section, we look at some of the notable cGMP requirements and how they
apply to biopharmaceutical manufacturing. The basis for this discussion is the cGMP
standards from the U.S. Code of Federal Regulations and from the European Union’s
EudraLex, although as mentioned previously, cGMP requirements throughout the
world are similar at a high level.

3.3.1 Personnel

People play an important role and are necessary in some way for essentially every
step in biopharmaceutical manufacturing processing. People are needed for all types
of tasks, such as execution of manufacturing processes, testing of product samples,
review of manufacturing documentation, cleaning of manufacturing facilities, main-
tenance and calibration of manufacturing equipment, and management of other per-
sonnel, just to name a few. Because personnel are so essential to successful execution
of a biopharmaceutical manufacturing process, it is not surprising that the cGMP reg-
ulations and guidance documents include sections specific to personnel.
For example, it is a requirement that biopharmaceutical companies employ an ap-
propriate number of qualified personnel to manufacture a quality product[95, 96].You
might think that the focus of this requirement would be the personnel who are directly
responsible for manufacturing and testing of product. However, the regulations make it
clear that it is equally important to ensure sufficient ancillary personnel are in place to
provide oversight and support of manufacturing operations. The EMA requirements go
so far as to require organizational charts showing the relationships between different
departments and specifying general responsibilities for production and quality[97]. It is
also a requirement that personnel are qualified for their particular roles, having
the appropriate education, training, and experience to perform their assigned func-
tions[95]. From the previous chapter, it is clear that biopharmaceutical manufacturing
is complex and therefore requires input from many different people with many different
backgrounds. Having an adequate number of qualified personnel in place is key to en-
suring SISPQ.
Every biopharmaceutical company must also establish a training program that
provides initial and ongoing training for their cGMP personnel. Even with appropri-
ate education and potentially prior experience in the biopharmaceutical industry,
68 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

employees still need training on company-specific procedures and practices. In ad-

dition, it is an expectation that employees are provided training on topics related to
cGMP and quality systems at some frequency to reinforce basic principles and to
provide awareness of new regulatory requirements, company procedures, or indus-
try best practices.
Expectations regarding personnel hygiene and cleanliness are also part of cGMP
requirements. From Chapter 2, one of the key inputs to product quality is design of
the manufacturing facility with a focus on protecting the product from contamination.
Protecting the product from contamination from personnel is important as well, given
the risk to contamination that personnel pose. A large number of personnel enter a
facility every day. They have close proximity to the product being manufactured, and
people can shed 1,000,000,000 skin cells per day, of which approximately 10% may
contain microorganisms [98]. No matter how well a facility is designed to prevent mi-
crobial and cross contamination, if the cleanliness of the people inside the facility,
especially those with close proximity to manufacturing and support operations, is not
controlled, the product is not adequately protected. Specific ways in which product is
protected from employees include:
– Restricting access to cGMP areas for any employees who are sick or have open
wounds/lesions
– Ensuring personnel wash their hands and follow good sanitation practices
– Ensuring that employees who access cGMP areas are gowned appropriately, which
is discussed in greater detail in the next section
– Controlling cleanroom behaviors

The recent coronavirus pandemic has brought an increased awareness of the impor-
tance of practicing good hygiene to prevent spread of infectious diseases. These practi-
ces, although important in day-to-day life when faced with a global crisis such as a
pandemic, have always been important in the manufacture of biopharmaceuticals,
where it is critical to protect the product from contamination. To this end, the regula-
tions specify that companies are to prohibit any person with an apparent illness or
open lesions from direct contact with product or product materials and components
and that personnel are to report any health conditions that may adversely affect drug
products to supervision [99]. A biopharmaceutical manufacturing employee is also re-
quired to wash his or her hands frequently, especially prior to entering the manufactur-
ing areas and before donning an initial set of gloves. Rules around good hygiene also
typically prohibit wearing makeup, jewelry, and perfumes and storing or consuming
food and drink in the cGMP working areas.
 3.3 Notable cGMP requirements for biopharmaceutical manufacture 69

3.3.2 Personnel gowning

Personnel gowning, which refers to the special protective apparel that staff entering
the manufacturing facility must wear primarily to protect the product, is a basic cGMP
requirement for biopharmaceutical manufacturing; however, the specific gowning re-
quirements will differ based on the operations within a manufacturing area. In general,
as a process moves closer to drug substance or drug product, more protective gowning
is required. Likewise, open operations are conducted with more gowning than closed.
Some examples of protective gowning typical in this industry are hairnets, gloves, shoe
coverings, frocks/lab coats, coveralls/jumpsuits, masks, sterile sleeves, and safety
glasses/goggles. In addition, some companies restrict employees from wearing street
clothes – that is, their personal clothing – in the manufacturing areas and instead re-
quire employees to put on scrubs or cleanroom clothing and even dedicated facility
shoes. Whatever the requirements, it is important to gown according to approved facil-
ity procedures, which are discussed shortly, using the correct gowning supplies and
donning gowning in the correct order.
As a first gowning step when entering manufacturing areas, personnel will likely
don laundered scrubs or a plant uniform followed by facility-dedicated shoes or shoe
coverings. Donning shoes/shoe coverings is generally done such that the gowning
only touches the “clean” side of the gowning area. Initial gowning also likely includes
donning a hairnet and pair of gloves. As personnel make their way into specific
manufacturing suites where more critical operations taken place, they will be required
to add additional layers of protective gowning. For example, when entering an ISO 8
area, it is likely that personnel would add a coverall over their scrubs, a pair of shoe
covers over existing shoe covers, and a pair of gloves over existing gloves. If transition-
ing to an ISO 5 cleanroom from an ISO 8 area, more stringent gowning would be re-
quired as ISO 5 areas are required to meet a tighter criterion for particulates than an
ISO 8 cleanroom (see Chapter 2). Personnel would likely add a frock over their cover-
alls, a hood over their hairnet, boot covers over their shoe covers, etc. Typical gowning
items required for ISO 8 and ISO 5 areas are described below.
Example ISO 8 Gowning Procedure
1. Apparel from initial gowning
2. Safety glasses
3. Hairnet (if not already present) – ensure all hair is contained within the hairnet
4. Beard cover (if applicable)
5. Coverall – do not let coverall touch the floor or any unclean surfaces, ensure
any personal clothing is covered
6. Shoe covers – don one at a time while crossing from dirty side of room to clean
side
7. Gloves – spray hands with 70% isopropyl alcohol or use hand sanitizer after
donning
70 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

Example ISO 5 Gowning Procedure

1. Sterile mask – ensure nose and mouth are covered
2. Hood – tuck bottom of hood into top of coverall
3. Frock – added over existing gowning (i.e., coverall)
4. Boot covers – don one at a time while crossing from dirty side of room to clean side
5. Sterile sleeves – avoid touching outside of sleeves when donning
6. Sterile gloves – ensure sterile sleeves are captured under each glove

A comparison of typical gowning for ISO 8 operations versus ISO 5 operations is

shown in Figure 3.1.

Hairnet

Hood

Safety glasses

Mask

Beard cover

Coverall

Sterile sleeves

Gloves

Sterile gloves

Frock

Shoe covers

Boots

Figure 3.1: Gowning for ISO 8 (on left) operations versus ISO 5 (on right) operations. Photos © NC
State University; reprinted with permission. Note that ISO 5 gowning is typically donned on top of
existing gowning (e.g., gowning from ISO 8).

3.3.3 Cleanroom behaviors

Once gowned and in the cleanroom areas, personnel must take additional precau-
tions to remain as “clean” as possible and to reduce the risk of contaminating the
environment and the product. Often referred to as aseptic technique, these precau-
tions include the following:
 3.3 Notable cGMP requirements for biopharmaceutical manufacture 71

– Refrain from touching clean items and surfaces unless necessary. Minimizing
contact between the hands and other items/surfaces leads to less opportunity
for contamination.
– Frequently sanitize gloved hands, typically with an alcohol-based solution.
The frequency may vary based on the area in which you are working, but a
good rule of thumb is to sanitize gloves every 10–15 min or after touching any
other surface.
– Replace gloves if heavily soiled or damaged. Once gloves are damaged, they are
no longer effective for containing any contamination or shedding skin cells on
the hands.
– Avoid leaning or sitting on clean surfaces. Avoiding contact with these areas
with any part of your gowning helps to minimize contamination of critical
areas.
– Use slow, deliberate movements. Rapid movements can disrupt the laminar air-
flow in a cleanroom, causing the air, and any particulates, to swirl and move in
a more disorganized manner throughout the room.
– Avoid movements directly above open containers or near open processing. Since
the airflow in the room is likely flowing from top to bottom, placing your hand
directly above an open processing container could result in any particulates or
organisms on your hand falling into the vessel.

Learning appropriate aseptic techniques is an important part of good cleanroom be-

havior. It is likely your company will require you to complete training in these tech-
niques specifically for their particular process and the tasks you are responsible for
executing.

3.3.4 Equipment

GMP applies not only to the personnel involved with biopharmaceutical manufactur-
ing but to the processing equipment as well. The focus of cGMP related to equipment
is to protect the product and to ensure that equipment functions properly for its in-
tended use, both of which ensure product quality. Protecting the product means mini-
mizing potential for contamination, which can be accomplished in a number of ways.
From 21 CFR 211 subpart D [100], key equipment considerations related to cGMP are
as follows:
– Equipment should be designed and located appropriately for its intended use
and allow for cleaning and maintenance.
– It is important to ensure the equipment on its own does not introduce any con-
taminants to the product stream. Equipment surfaces that come in direct contact
with the product should be non-reactive, and not be a source of foreign materi-
als, residues, microbial contamination, or particulates.
72 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

– The product contacting surfaces should also be easy to clean and compatible with
various cleaning and sanitization solutions. Common materials of construction for
biopharmaceutical equipment include stainless steel, glass, and thermoplastics.

Unless disposable and intended for one-time use, equipment must be cleaned between
uses and therefore methods for cleaning must be developed. Cleaning of equipment
can be performed manually or by an automated means. Automated cleaning may be
performed either while the equipment stays in place – referred to as clean in place or
CIP, as described in Chapter 2 – or by transferring the equipment to a separate location
specifically for cleaning, such as a wash room containing a parts washer. In addition,
depending on the purpose of the equipment in the process, it may be necessary to es-
tablish a sanitization or sterilization procedure as well. We won’t go into details here
given that these topics were covered in the previous chapter. Regardless of the exact
process for cleaning and sanitization, GMP requires that written procedures be estab-
lished so that the same process can be consistently executed. Procedures should also
include details related to required frequency of cleaning and a requirement for inspec-
tion of equipment for cleanliness prior to use. It is expected that a cleaning history
or record is maintained for process equipment. We’ll discuss cGMP requirements
regarding documentation later in this chapter, but a general rule of thumb is that
if it isn’t documented, it didn’t happen. For this reason, it is important to record
each time cleaning and sanitization/sterilization is performed. In general, the fre-
quency of cleaning and sanitization/sterilization should be established to ensure
there is no build-up of residual product/process components and no carryover of
potential contaminants between equipment uses. Cleaning procedures are often
validated – meaning that testing is performed under a protocol to demonstrate
that the cleaning process consistently meets expected results. More specific infor-
mation related to execution of cleaning validation is covered in Section 3.3.7.
Maintaining equipment in an appropriate state of repair is also part of cGMP.
Doing so ensures that the equipment operates consistently during multiple production
batches and is accomplished through routine inspection, calibration, and preventive
maintenance. Written procedures as well as records of execution are also required for
these activities. The intended use of the equipment influences the schedule for perform-
ing calibration and maintenance activities, with more critical equipment needing the
most frequent oversight. In addition, it is expected that calibrations are performed
against established acceptance criteria and that equipment is prevented from use when
these criteria are not met. To that end, most companies employ a means of designating
the current calibration status, either by physically labeling the equipment or through
an electronic asset management system where personnel can easily find calibration sta-
tus and calibration due dates.
 3.3 Notable cGMP requirements for biopharmaceutical manufacture 73

Example: Calibrating a flow meter on a chromatography system

A calibration check on a chromatography system flow meter is scheduled at six-month intervals.
The procedure involves pumping water through the chromatography system at flow rates that
cover the range of operation for the skid. The flow rate reading on the control screen (transmit-
ted by the inline flow meter) is recorded during the procedure. The actual flow rate (i.e., the
standard against which the control screen reading is compared) is measured by weighing the
amount of water collected over a one-minute period and converting to a flow rate in units of L/h,
assuming 1 kg = 1 L. The percent deviation between the reading from the control screen and the
standard is determined. This percent deviation must fall within +/- 2.5% to be deemed accept-
able. Data from the most recent calibration check is shown in Table 3.2.
Does the flow meter require any adjustment to meet the allowable range for percent deviation?

Table 3.2: Data collected for a calibration check on a chromatography flow meter.

Flow Set Control Screen Volume Water Time for Calculated % Deviation
Point, L/h Flow Rate Collected, L Collection, s Flow Rate, L/h
Reading, L/h

 . . 
 . . 
 . . 
 . . 
 . . 

Solution
Calculation of the actual flow rates based on the volume of water collected over a 60 s period
result in the following values, listed from the low to high set point: 36.6, 71.4, 109.2, 144.6, and
180.6 L/h. The calculation for the 36 L/h set point is shown below as an example:

Volumetric flow rate at 36 L/h set point = 0.61 L/60 s × 60 s/min × 60 min/h = 36.6 L/h

The corresponding percent deviation values are 2.19, 0.14, 0.55, −0.28, and 0.61%, respectively.
Therefore, the flow meter is accurate based on the percent deviation criteria required, and no
adjustment to flow rate readings is required. As an example, the calculation for the percent devi-
ation at 20% set point is shown below:

% deviation = (36.6–35.8)/36.6 × 100 = 2.19%

There are additional cGMP considerations for equipment that involve automation or
computerized systems. Regulations require that the computer systems associated
with this equipment are validated to demonstrate that the computer system hard-
ware and software performs tasks consistently and as expected [101]. In addition,
access to a computerized system should be controlled to prevent unauthorized ac-
cess, and to restrict the ability to change data or records generated by the system
[102]. In many cases, the data generated by the computerized system becomes part
74 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

of the official manufacturing batch documentation, so it is important that data are

verified for accuracy, complete, and protected from modification or deletion. Regu-
lations also require that companies have processes in place to routinely back up
such data as a safeguard against data loss. Written procedures describing the spe-
cific requirements for each computerized system are expected.

3.3.5 Documentation system: procedures, records/reports and specifications

As mentioned in the previous sections, a familiar phrase in biopharmaceutical

manufacturing is “If it wasn’t documented, it didn’t happen.” Further, use of proce-
dures, records, and specifications is critical to controlling the process and product. Eu-
draLex, Volume 4 indicates that the main objective of a documentation system is to
“establish, control, monitor and record all activities which directly or indirectly impact
on all aspects of the quality of medicinal products” [103]. As a result, multiple regula-
tions and guidance documents focus on cGMP requirements related to documentation.
Documents are generated to execute the manufacture of a biopharmaceutical
product and may be captured in paper or electronic format. In general, documentation
falls within one of three categories – procedures, records and reports, and specifications.
– Procedures are documents that provide guidance or instructions to execute a spe-
cific task or process. In a sense, procedures are an input to a task. Procedures de-
fine what is supposed to happen, in what order, and what the expected outcomes
are. Written procedures provide control by ensuring that an activity is performed
the same way every single time. They must be written in clear and unambiguous
language and be accurate. The use of pictures/diagrams/flow charts where appro-
priate can be helpful in making a procedure easier to follow. Examples of proce-
dures are included in Table 3.3.
– Records and reports document the output of the executed tasks. Specific exam-
ples are included in Table 3.3.
– Specifications are similar to a procedure in that they can be viewed as an input to a
task. However, specifications are generally limited to stating requirements or accep-
tance criteria and serve as a basis for quality evaluation [103]. Specifications do not
generally include individual instructions. For example, an analytical testing specifi-
cation might include the acceptance criteria for several different tests but would not
include instructions for how to complete each assay. The analyst would instead
have to reference the specific test procedures to understand how to complete each
test. Note that analytical specifications are written for a variety of samples, includ-
ing raw materials, process intermediates, drug substance, and drug product.

Because procedures, records/reports, and specifications are an important means of

controlling a biopharmaceutical manufacturing process, and thus important for en-
suring product quality, cGMP requires these documents be approved by appropriate
 3.3 Notable cGMP requirements for biopharmaceutical manufacture 75

Table 3.3: Example document types.

Document Type Document Category

Standard Operating Procedure (SOP) Procedure

Master Production Record (the version of a Procedure

production record from which the batch
production record is created)

Batch Production Record Procedure and Record

Validation/Qualification Protocol Procedure and Record

Validation/Qualification Report Report

Forms Record

Logbook Record

Analytical Testing Report Record/Report

Status Tag Record

Stability Protocol Procedure and Record

Stability Report Report

Material Release Specification Specification

Product and Sample Labels Record

Complaint Files Record

Certificate of Analysis Record/Report

personnel and assigned an effective date. Any revisions to procedures are also con-
trolled and must be justified and preapproved prior to use by personnel. Figure 3.2
illustrates a typical controlled document process flow. It is never acceptable to know-
ingly deviate from approved instructions and procedures. In fact, a common observa-
tion by regulatory agencies during inspections is companies failing to follow their
written procedures.
Records and reports provide evidence that tasks were performed correctly and per
approved procedures and specifications. This document category includes not only the
manufacturing documentation, such as batch records and test results, but also docu-
mentation related to supporting activities, such as validation, training, investigations,
storage, maintenance, and cleaning. How things are documented is an important con-
cept in cGMP. In fact, this is such a crucial topic in the industry that there are standards
related to how to document results, referred to as good documentation practice or GDP.
GDP applies to both written and electronic records/reports supporting the manu-
facture of biopharmaceuticals. When dealing with written documentation, some spe-
cific best practices are universal [103]. For example, no areas where data is expected
76 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

Author writes or revises document

(e.g., batch records, SOP, etc.)

Document routed for review to relevant

areas (manufacturing, process develop-
ment, QC, validation, QA (always), etc.)

Author addresses comments

from review

Author routes revised document for

approval from relevant areas

Once approved, QA Document Control

creates a master of the document

QA Document Control distributes

document for use Figure 3.2: Controlled document process flow. Image ©
NC State University; reprinted with permission.

should be left blank. Handwritten entries should be made at the time of performance;
the individual making the entry and when it was made should be able to be traced.
This is generally accomplished by requiring the person making the entry to provide
their initials and the date, where the date includes the day, month, and year and does
not use slash marks or dashes. All entries should be clear and legible and should be
captured in indelible ink so that records are permanent and cannot be changed. Nu-
merical entries should be recorded to the same number of significant figures as associ-
ated acceptance criteria or specifications. Military time is generally preferred for time
entries as it does not require designation of am versus pm. Writing over previous entries
to make corrections is not allowed. If corrections or changes are made to original docu-
mentation, GDP requires that these are made in a certain way as explained below:
– What changed?
– Draw a single line through the error so that the original entry is not obscured
and can still be easily read.
– Do not erase, use white-out or otherwise obliterate the existing information.
– Who made the change?
– Document with your initials.
– Ensure that initials can be traced to a single individual.
– Why was the change made?
– Document an explanation for the change.
 3.3 Notable cGMP requirements for biopharmaceutical manufacture 77

– Legibly add the correct information as close to the original entry as space
allows.
– If there is not enough space near the original entry to make the correction,
include a footnote at the location requiring correction and enter a corre-
sponding footnote and explanation in the margin or at the bottom of the
page where space allows.
– When the change occurred
– Document the date the correction is made.
– Use the date format required by your SOPs.

Example: Finding documentation errors in an executed batch record

Figure 3.3 provides an excerpt from an executed batch record used in the production of tris solution,
a buffer commonly used in chromatography steps for biopharmaceutical processes. Focusing on the
written entries made by operators, identify specific examples of poor documentation practice.

Figure 3.3: Excerpt of an executed batch record for preparing tris solution. Image © NC State
University; reprinted with permission.
78 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

Solution
Documentation errors in the above record include:

Step 2 – The original data entry is obscured, and the corrected value does not include a reason
for why the data was changed, initials of who made the correction, and a date to indicate when
the correction was made.

Step 3 – The data field was completed incorrectly. An “X” was used in data field rather than “✓”
as indicated in the “Item” field.

Step 4 – The incorrect number of significant figures was used for agitator speed, and, when
rounded, the result does not meet the acceptance criteria for this parameter. Instead of correct-
ing the “Start Mix Time” data entry, the operator attempted to write over the original data (1026
changed to 1028). The “Performed By” entry appears to be a day later than when all other steps
were performed. The same person initially signed for “Performed By” and “Verified By.” Correc-
tion of the “Verified By” entry was not done correctly; the original entry is obscured and the
correction does not include why, who, and when.

Step 5 – The date entry in the “Performed By” field is not in the correct format. The slashes (/)
may be mistaken for the numeral 1. A more common format would be 02Nov2020.

Step 6 – Data for the pH meter ID is not legible (could be 1637 or 1037). Entry for pH is not re-
corded to the correct number of significant figures, which prevents determination of whether
the value is within the acceptable range.

Step 7 – The correction to the data entry was made correctly; however, the footnote explaining
the correction at the bottom of the page should also be initialed/dated. In addition, because the
original value recorded for “Final HPW Volume” was outside the acceptable range, a more de-
scriptive note would be necessary to explain why the value was changed.

The acronym ALCOA or ALCOA+ is commonly used to summarize principles associ-

ated with GDP, where ALCOA stands for: Attributable, Legible, Contemporaneous,
Original, and Accurate. The + has been added more recently to indicate that docu-
mentation must also be Complete, Consistent, Enduring, and Available. Table 3.4
provides additional information about each of these document characteristics.

Table 3.4: ALCOA+ principles for good documentation practices.

Letter Principles Description

A Attributable Entry is traceable to the person who performed the action or generated
the data.

L Legible Entry is easy to read and understand.

C Contemporaneous Entry was made or data generated at the time of the activity.

O Original Data is from original source.

A Accurate Entry is valid, correct, and contains no errors.

 3.3 Notable cGMP requirements for biopharmaceutical manufacture 79

Table 3.4 (continued)

Letter Principles Description

+ Complete All data is included, including repeat analysis.

Consistent Date/times associated with data follow expected sequence.

Enduring Entry/data is permanent and sustainable.

Available Data can be accessed for review or inspection over the lifetime of the
record.

A notable trend in the biopharmaceutical industry is the increasing use of electronic

records and reports, including electronic batch records. Electronic systems for docu-
mentation may help to minimize the occurrence of documentation errors and has
the added benefit of allowing data to be readily accessible for review and analysis.
Electronic documentation, however, is subject to the same requirements as hand-
written documentation, and ALCOA+ principles still apply. Electronic documentation
has the added component of electronic signatures in lieu of a handwritten signature/
date or initials/date. The requirements for the use of electronic signatures are defined
in several regulations and guidance documents, including 21 CFR Part 11. Electronic
signatures must capture the identity of the signer, the date and time when the signa-
ture was made, and the meaning or purpose of the signature. In addition, electronic
signatures must be made using at least two distinct identification components (e.g.,
username and password) [104].

3.3.6 Testing: raw materials, in-process, drug substance, and drug product

There are many points throughout a biopharmaceutical manufacturing process where

testing is performed. On the front end of the process, cGMP requires that each lot of
raw materials and components used in the manufacture of the product are sampled,
tested, or examined and released by the Quality unit prior to use [105]. Minimally,
this testing usually includes a confirmation of identity, which may only require in-
spection of the certification of analysis for the raw material/component. However, de-
pending on the criticality of the material and its use in the manufacturing process,
more comprehensive testing may be required. Approved material specifications indi-
cating the testing requirements and associated acceptance criteria are generated for
each material.
Once the manufacturing process is initiated, various tests are performed along
the way on process intermediates to ensure the process is performing as expected or
to prepare product streams for further processing [96, 106]. A variety of analytical
methods may be employed for measuring characteristics of intermediates such as
80 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

protein concentration, product purity, impurities, pH, conductivity, endotoxin or bi-

oburden. The specific testing performed depends on the type of biopharmaceutical
being produced. The requirements and acceptance criteria for such in-process testing
are generally specified in the manufacturing batch records or associated specifica-
tions. Other types of testing that may occur during processing of a biopharmaceutical
manufacturing batch include environmental monitoring (EM), which is performed to
verify the manufacturing environment in which the production process takes place is
under a state of microbial control.
At the end of the manufacturing process, testing of the bulk drug substance and
drug product is performed [107]. As we learned in Chapter 1, cGMP regulations re-
quire that drug products are tested to ensure batch-to-batch consistency and to ver-
ify product quality. These tests are commonly referred to as release testing because
the drug substance or drug product is required to conform to the approved accep-
tance criteria for the material to be released by the Quality unit. Approved product
specifications are typically generated to identify the testing requirements and ac-
ceptance criteria for this type of testing. Table 1.5 provides an example specification
for a monoclonal antibody drug product.

3.3.7 Qualification and validation

Another important cGMP concept is qualification/validation. It is a cGMP require-

ment that manufacturers control the critical aspects of their particular operations
through qualification and validation activities over the lifecycle of the product
and process [88]. Qualification and validation are performed to provide evidence
that a specific system or process (including analytical tests) is capable of meeting
predetermined requirements or acceptance criteria that demonstrate that the sys-
tem or process works as expected. Qualification/validation can apply to many as-
pects of biopharmaceutical manufacturing including facilities, utilities, equipment,
computerized systems, manufacturing processes, cleaning processes, and analytical
test methods. These two terms are often used interchangeably, however, there are
typically differences between qualification and validation activities. Although both
qualification and validation are executed to ensure a system or process is fit for in-
tended use, validation usually has an added factor of demonstrating consistency –
that is, the particular requirements for a specific intended use can be consistently ful-
filled [108]. In many cases, qualification activities are a pre-requisite to validation
activities. Another distinction between the two terms is that qualification is often as-
sociated with equipment, facilities, and utilities whereas validation is associated with
processes (cleaning, testing, manufacturing).
Qualification and validation activities are executed under protocols that are pre-
approved by the appropriate functional areas, including Quality Assurance. After
testing is executed, it is typical to summarize the results in a report that includes a
 3.3 Notable cGMP requirements for biopharmaceutical manufacture 81

discussion of any testing failures or departures from the approved protocol. Specific
activities related to qualification of equipment, facilities, and utilities include the
following [88]:
– Writing a Validation Master Plan (VMP) – a document that summarizes how
qualification/validation activities will be conducted for the entire process for a par-
ticular product, or for a particular project.
– Writing a User Requirements Specification (URS) – a document to summarize
the minimum requirements that the system must meet and which becomes the
basis for subsequent qualification testing.
– Design Qualification (DQ) – demonstration that the design of equipment, facili-
ties, and utilities complies with cGMP and meets user requirements.
– Factory Acceptance Test (FAT) – testing performed at the system manufac-
turer’s site to verify user expectations are met prior to delivery.
– Site Acceptance Test (SAT) – testing performed at the user’s site to verify no
negative impact during transportation and to confirm user acceptance.
– Installation Qualification (IQ) – verification that the system has been installed
correctly, consists of all expected components and instrumentation, consists of
the expected materials of construction, and that all associated documents, draw-
ings, specifications are available.
– Operational Qualification (OQ) – verification that the system operates as de-
signed, including verification of acceptable operation at the upper and lower op-
erating limits. The OQ is generally performed after IQ completion but may also
be performed as part of a combined Installation/Operational Qualification (IOQ).
– Performance Qualification (PQ) – verification that the system, in conjunction
with ancillary systems and associated personnel and procedures, performs appro-
priately during normal operation with normal operating limits.

Cleaning validation is performed for multi-use equipment (dedicated to one product or

used for multiple products) to demonstrate that a cleaning process is consistently able
to remove process soils and cleaning agents to acceptable levels between uses. Carry-
over of processing and product materials from one batch to another has impact to prod-
uct quality and therefore must be minimized. Cleaning validation is also performed
under a preapproved protocol with established acceptance criteria. It is typically initi-
ated during manufacturing of product for Phase 3 clinical trials and may continue into
commercial manufacturing. Until a cleaning procedure is validated, cleaning verifica-
tion must be conducted that demonstrates equipment is clean prior to moving forward
with processing. Once cleaning is validated, cleaning verification is not required.
Execution of cleaning validation involves taking samples during the final water
rinses of the cleaning cycle (referred to as rinse water samples) and from the prod-
uct contact surfaces of the equipment after the cleaning cycle is complete. Rinse
water samples may be analyzed for pH, conductivity, total organic carbon (TOC),
endotoxin, and bioburden. TOC, as the name suggests, measures the amount of organic
82 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

carbon present, which serves as a nonspecific indicator of product left behind. The sur-
face samples are typically collected using swabs to physically remove any soil that may
be remaining on the surface. Swab samples are usually analyzed for TOC or for the spe-
cific product that is being manufactured. Regardless of the analytical method used to
test cleaning validation samples, it is important to set acceptance levels that provide
assurance that the equipment has been adequately cleaned prior to its next use. To
demonstrate consistency of a cleaning procedure, cleaning validation requires that a
specific piece of equipment be successfully cleaned (i.e., meets established acceptance
criteria) multiple consecutive times using the applicable procedure. Three consecutive
cleanings is the minimum number commonly used.
Process validation is performed to ensure that a biopharmaceutical manufactur-
ing process is capable of consistently producing a product of appropriate quality.
Process validation is not a single activity or protocol but instead consists of multiple
activities that are performed as three stages [76]:
– Stage 1 – Process Design: Based on knowledge gained throughout development
and scale-up activities, the commercial process is defined during this stage. Crit-
ical aspects of the process are defined and operating ranges established to en-
sure manufacture of a quality product. The steps illustrated in Figure 2.9 for
designing an antibody process serve as an example.
– Stage 2 – Process Qualification: During this stage, the process is confirmed as
being capable of reproducible commercial manufacturing. This stage includes
execution of multiple batches, usually a minimum of three, at full manufactur-
ing scale, which are referred to as consistency batches or process performance
qualification (PPQ) batches. PPQ runs combine the actual facility, utilities, qual-
ified equipment, and trained personnel to produce commercial batches. There is
more sampling and testing conducted than during routine commercial production
in order to sufficiently establish consistency and product quality throughout the
process. Qualification of equipment, utilities, and systems must be performed
prior to PPQ runs. In addition, PPQ runs must be completed successfully be-
fore product can be commercially distributed.
– Stage 3 – Continued Process Verification: Ongoing assurance is gained during
routine production that the process remains in control. Continued process verifi-
cation (CPV) is often accomplished through routine trending of process and test
data to detect shifts in process performance.

In Figure 2.6, we showed that the three process validation stages span the entire bio-
pharmaceutical product lifecycle, from drug discovery to commercial manufacture
and beyond. Many different specific studies may be required to support validation of
a biopharmaceutical manufacturing process. Table 3.5 includes some of the most
common that are performed for biopharmaceutical processes along with a basic de-
scription of the study purpose. These studies are typically performed once during the
product lifecycle. Timing for completion of these studies may occur during Stage 1
 3.3 Notable cGMP requirements for biopharmaceutical manufacture 83

(Process Design) as a pre-requisite for Stage 2 (Process Qualification) or may occur

concurrently with Stage 2. Generally, a process validation master plan (PVMP) is writ-
ten to summarize all activities to be completed as part of the process validation effort
and to identify the timing for study execution.

Table 3.5: Process validation studies.

Study Type Description

End of Production Cell Demonstrates that the cells used to produce the product are stable
Bank Testing and free of contamination.

Media Hold Shows that media maintains sterility for a specific duration and
promotes cell growth in the bioreactors used for production.

Resin/Membrane Reuse Determines number of times chromatography resin and microfiltration

(Lifetime) or ultrafiltration membranes may be reused for processing of the same
product while maintaining expected performance with minimal
degradation. Requires studies at both bench- and production-scale.

Resin/Membrane Cleaning Validates that the cleaning process defined for resins/membranes
Validation results in minimum product carryover. Requires studies at both bench-
and production-scale.

Viral Clearance Studies Validates that process steps inactivate or remove viruses to
acceptable levels. Conducted by spiking model viruses into a bench-
scale version of manufacturing steps to measure clearance.

Residual/Impurity Demonstrates consistent removal of process residuals or product

Removal impurities to acceptable levels. Typically executed at bench scale.

Process Intermediate Hold Determines the amount of time a product intermediate pool can be
Times held without impacting product quality.

Sterile Filter Validation Demonstrates a . um filter is able to reduce microbial load in a given
process stream to an acceptable level.

Shipping Validation Demonstrates the procedure for shipping product to other locations
has no impact on product quality.

Leachable Validation Characterizes the amount/types of impurities that may be leached

from plastic/thermoplastic components used during processing.

Solution Hold Time Determines the amount of time a solution used in the manufacturing
process can be held without impacting quality. Typically requires
studies at both bench- and production-scale.

Once qualification and validation are completed, cGMP requires that any planned
changes to the facilities, equipment, utilities, and processes that may affect the quality
of the product should be formally documented and the impact on the validated status
or control strategy assessed [88]. The system put in place for this documentation and
84 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

assessment is referred to as change control or change management quality system.

So-called “change controls” are created for each proposed change according to a
company’s individual procedure and assessed by subject matter experts to deter-
mine impacts of the change. Evaluation may result in a determination that requa-
lification/revalidation must be completed to return the system or process to an
acceptable validated state. For example, consider a case where a new high-density
shelving system is to be installed in a qualified walk-in cold room. When the cold
room was originally qualified, temperature mapping was performed to verify that
the cold room can maintain temperature between 2–8 °C throughout all locations for a
24-hour period. This mapping was performed both in an empty chamber and while
holding materials representing a typical load. How might this new shelving system af-
fect the original qualification performed? Depending on the configuration, the racks
may change the airflow dynamics within the chamber, which may impact the tempera-
ture distribution within the cold room. In addition, high-density shelving will likely
allow for an increased amount of materials to be stored in the walk-in cold room; there-
fore, the temperature mapping performed previously for a loaded chamber may no lon-
ger be representative. In this case, the unit will likely need to undergo requalification to
maintain a qualified state. The change control system would not only identify actions
needed to ensure the unit remains appropriately qualified, but also any other actions
necessary to implement the change, such as updates to SOPs, changes to cleaning pro-
cedures, and changes to preventive maintenance and/or calibration requirements. Im-
plementation of planned changes through a formal change control system helps to
ensure that changes are made in a controlled manner, that all potential impacts are
appropriately planned for, and that a qualified/validated state is maintained.

3.3.8 Deviations and corrective and preventive actions (CAPA)

GMP requires that a biopharmaceutical manufacturer has a system in place to investi-

gate when events deviate from what is planned or expected. A variety of terms are used
to refer to such events, including deviations, exceptions, nonconformances, failures, or
atypical events. ICH Q7, GMP Guidance for Active Pharmaceutical Ingredients, defines
a deviation as a “departure from an approved instruction or established standard” [61].
FDA regulations require any “unexplained discrepancy or the failure of a batch or any
of its components to meet any of its specifications shall be thoroughly investigated”
and a “written record of the investigation shall be made and shall include the conclu-
sions and follow-up” [109]. Some examples of deviations that may occur during bio-
pharmaceutical processes are summarized in Table 3.6.
 3.3 Notable cGMP requirements for biopharmaceutical manufacture 85

Table 3.6: Deviation examples. Note that the categories indicate that type of deviation and not
necessarily root cause.

Deviation Category Deviation Examples

Production/Process Deviation from batch record instructions, such as executing  homogenizer

Controls passes when batch record requires .
Failure to meet an in-process limit, such as pH, conductivity, or step yield.

Documentation Failure to complete documentation at the time of execution.

Use of a non-effective or superseded version of controlled document.

Materials Failure to release raw materials prior to use in batch.

Management Use of expired material in a product batch.

Facilities/Equipment Power outage.

Use of equipment that is past the calibration due date.

Quality Control Failure of bulk product sample to meet release criteria.

Analytical test method system suitability failure.

When documenting a deviation, the following information is typically included:

– A full description of the atypical event that occurred, including details around
what happened, who was involved, when the event occurred, where the event
occurred, and how it was different than what was expected. Information regard-
ing root cause is typically left out of the description since an investigation has
yet to be performed.
– Any immediate actions taken to correct the event. Examples include contain-
ment of materials/batches impacted, restricting access to impacted facilities/
equipment, or performing specialized cleaning in response to the event.
– Results of the investigation to determine the root cause of the event, which is
important for determining how to prevent similar deviations from occurring in
the future. The root cause investigation should be systematic, where all possible
“whys” are evaluated and either confirmed or ruled out as a root cause. Fish-
bone analysis, 5-Why Analysis, and Cause and Effect analysis are some common
investigation techniques for determining root cause. Details on how these meth-
ods are used for root cause determination can be found elsewhere [110, 111].
– An assessment of impact to product quality.

Once a root cause is identified for a deviation, the logical next step is to determine if
there is a way to eliminate that cause to prevent recurrence of the deviation, referred to
as a corrective action. In addition, it is important to consider if additional actions can
be taken preemptively to prevent a similar deviation from occurring in the future, re-
ferred to as a preventive action. These actions are generally captured through CAPAs –
corrective and preventive actions. To fix/prevent a deviation, CAPAs must be directly
related to the root cause(s) identified during the deviation investigation. CAPAs are
86 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

typically managed by a company’s quality organization to ensure the proposed actions

are completed according to a defined timeline and that the actions are effective. If a
corrective action is ineffective, meaning that it does not prevent the deviation from re-
curring, then it may be that the “true” root cause of the deviation determined during
the original investigation is not correct. For this reason, if a corrective action is found to
be ineffective, further root cause investigation is typically required.

Example: Identifying the root cause and CAPAs for a step-yield deviation
Consider a deviation in which the step yield for a chromatography run is 40%, while the required
range in the batch record is 70%–90%. The root cause investigation has determined that the
volume reading from the level indicator on the process vessel used to collect eluate (i.e., prod-
uct) from the step is erroneously low, which has led to the out-of-range step yield. (Recall that
step yield is defined by equation (2.1). To calculate the amount of product at the end of the chro-
matography step – the numerator in equation (2.1) – the eluate volume is multiplied by the prod-
uct concentration in the eluate.) The erroneously low volume reading was confirmed by putting
the process vessel on a floor scale, adding a known amount of water by weight (and assuming
1 kg = 1 L), and observing that the level indicator on the vessel read less than the amount of
water added.
What are possible root causes and associated CAPAs for this deviation?

Solution
There are a number of possible root causes, and more investigation is required to determine the
true cause. Some possibilities include:
– Possible root cause #1: The investigation determines that the level sensor that provides the
volume reading from the process vessel was just calibrated. Interviews with the calibration
technician uncovered that he failed to follow the applicable SOP during the procedure in a
way that led to the low volume reading.

Corresponding CAPAs: The level sensor should be recalibrated to read accurately and, given
that this is a personnel error, the technician should be retrained on the procedure.

– Possible root cause #2: The investigation reveals that the level sensor was calibrated just
prior to the deviation, and interviews with the calibration technician reveal that the applicable
SOP is missing a key instruction necessary for proper sensor calibration.

Corresponding CAPAs: The level sensor should be recalibrated to read accurately and, given
that the root cause is related to a method, the applicable SOP should be revised to include
the missing instruction. Further, all calibration technicians would be retrained on the re-
vised SOP.

– Possible root cause #3: The investigation reveals that the level sensor malfunctioned during
the chromatography run leading to an erroneously low eluate volume reading.

Corresponding CAPAs: Given that the root cause is related to equipment, the level sensor
should be repaired or replaced with a new sensor. Further, the level sensor should be placed
on a preventive maintenance program.
 3.3 Notable cGMP requirements for biopharmaceutical manufacture 87

3.3.9 Batch disposition

Every batch of a biopharmaceutical manufactured is ultimately dispositioned by the

Quality unit, meaning it is assigned a specific status or product usage. This applies to
batches of both drug substance and drug product but may also apply to product inter-
mediates, solutions that are used in the drug substance or drug product process, raw
materials, and consumable items. The Quality unit is responsible for determining the
final disposition status, which is commonly described as one of the following:
– Approved or Released – Material is approved for further processing or distribution.
– Conditionally Approved or Conditionally Released – Material is approved for fur-
ther processing, but additional action is required for final disposition.
– Rejected – Material does not meet specification requirements and is not intended
for cGMP use in further processing or distribution.

We’ll focus on disposition of drug substance and drug product in the remainder of
this section. For each batch of drug substance or drug product, cGMP requires that
all of the records associated with that batch are reviewed to ensure compliance with
established specifications and procedures. The records included in this review in-
clude any completed batch production records, solution preparation records, equip-
ment preparation records, laboratory testing records, and packaging and labeling
records. The batch review likely also includes a genealogy review to verify that all of
the inputs into the process – for example, raw materials, consumables, product inter-
mediates, and buffer solutions – have been properly released and are within expira-
tion at time of use, and to verify that every input can be traced back to its original lot
information. Most records undergo an initial review by the department responsible
for the execution (for example, Manufacturing reviews batch production records and
Quality Control reviews lab testing records) followed by a final review by Quality
Assurance.
Additionally, the batch review process includes a review of all deviations associ-
ated with the batch, including those related to any materials, product intermediates,
and buffer solutions used in that batch. It is important to ensure that all deviation in-
vestigations have been completed prior to release of a batch, with an assessment that
the deviations are deemed to have no product quality impact. A review of any batch-
related change controls is also performed to ensure no impact to batch release. Other
items that may be reviewed as part of the batch release process are environmental mon-
itoring results, equipment work orders, and facility alarm reports. A manufacturing fa-
cility will have an approved procedure to define what must be considered in the record
review. It will also have a procedure describing the steps involved in batch disposition.
An example of checklists used for record review and batch disposition in cGMP facili-
ties is provided in Table 3.7.
88 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

Table 3.7: Example of a document review checklist and batch disposition checklist used to
determine the final disposition of drug substance and drug product.

Document review checklist

(complete checklist for all records associated with the batch)

Are all pages present and header information correct? □ Yes □ No

Is all equipment used within the record within calibration? □ Yes □ No

Are all rooms used within the record released per appropriate procedures? □ Yes □ No

Have all steps been initialed/dated by both the performer and verifier? □ Yes □ No

Has the record been signed by Manufacturing for review? □ Yes □ No

Has the record been reviewed for GDP standards? □ Yes □ No

Have all calculations been verified to be correct and any values carried to subsequent □ Yes □ No
step(s) correctly?

Has the record been signed by QA for review? □ Yes □ No

Batch Disposition Checklist

Have all records associated with the batch been reviewed by Manufacturing and QA? □ Yes □ No

Are all deviations related to the batch closed? □ Yes □ No

Have all environmental monitoring results been reviewed by QA and any related □ Yes □ No
deviation investigations closed?

Has the batch genealogy been created and approved by QA? □ Yes □ No

Have all change controls associated with the batch been reviewed by QA? □ Yes □ No

Has a Certificate of Analysis (COA) been generated for the batch and approved by QA? □ Yes □ No

Final Batch Disposition

□ Approved □ Conditionally Approved □ Rejected

QA Signature/Date:

3.3.10 Post-marketing surveillance

You might think that cGMP would no longer be applicable following completion of
biopharmaceutical manufacturing and final batch disposition. However, even after
biopharmaceutical drug product has been released and distributed for patient use,
regulations require that “a system and appropriate procedures should be in place to
record, assess, investigate and review complaints including potential quality de-
fects and if necessary, to effectively and promptly recall … medicinal products from
the distribution network” [112] Similar language can be found in the U.S. Code of
Federal Regulations [113]. Investigations related to post-marketing complaints are
 3.4 Summary 89

handled much the same way as those described in Section 3.3.8 for deviations and
include assessment of impact and determination of root cause(s) and corrective/
preventive actions. However, because these investigations relate to product already
on the market and thus may have impact to patient safety, there is an increased level
of urgency to resolve these issues quickly and thoroughly. In addition, complaints
may require notification to regulatory authorities, and when there is high risk to pa-
tient safety, may require recalling product lots from the market. A product recall is
a profoundly serious situation that may have significant impact to patients as well
as to the biopharmaceutical company. Use of cGMP throughout the biopharmaceu-
tical manufacturing process ensures that products are consistently produced with
SISPQ, minimizing the risk for product recalls.
An additional post-marketing responsibility of the biopharmaceutical manufac-
turer is to provide ongoing reporting of new information regarding the approved
product and manufacturing process. The FDA requires these reports to be provided
annually and include a summary of significant new information that might affect
product safety, effectiveness, or labeling of the drug product [114]. In addition, the
annual report may also include summaries of any stability data, investigations, and
change controls generated over the past year as well as manufacturing process data
trended to demonstrate the manufacturing process continues to perform consistently.

3.4 Summary

In this chapter, we learned how Good Manufacturing Practice (GMP) is essential to en-
suring production of biopharmaceuticals of appropriate quality. GMP standards define
the minimum requirements necessary to ensure production of products with the appro-
priate safety, identity, strength, purity, and quality (SISPQ). These standards encom-
pass many aspects of biopharmaceutical manufacturing and are intended to reduce
risks that can be encountered during biopharmaceutical production. Compliance with
GMP is enforced by various agencies throughout the world, such as the U.S. Food and
Drug Administration (FDA), the European Medicines Agency (EMA), and Health Can-
ada. Biopharmaceutical companies must therefore be aware of the GMP requirements
in the area where their products are marketed. As the industry evolves and new tech-
nologies are introduced and processing improvements are made, companies may have
to adapt the ways they meet GMP. For this reason, GMP is often referred to as cGMP or
current Good Manufacturing Practice.
This chapter touched on some of the main aspects of cGMP and specific require-
ments biopharmaceutical companies must adhere to in areas such as personnel, equip-
ment, documentation, testing, qualification and validation, deviations and CAPA, and
batch disposition. Best practices for meeting cGMP requirements in these areas were
also discussed.
90 Chapter 3 Good Manufacturing Practice (GMP) for biopharmaceutical production

– Personnel are a critical factor throughout all steps of a biomanufacturing process.

GMP requires that companies employ an appropriate number of people with the
right qualifications to adequately support both the manufacturing process and sup-
porting functions to produce quality product. Training programs must be established
to ensure personnel follow company-specific policies and procedures and to rein-
force basic GMP principles. Because personnel can be a major source of product con-
tamination, GMP also includes requirements around personnel hygiene, gowning,
and appropriate behavior while in the manufacturing cleanrooms.
– Equipment used for manufacture of biopharmaceuticals must function properly for
its intended use and not introduce contamination to the product, such as foreign
materials, residues, microbial contamination, or particulates. Reusable equipment
must be cleaned between runs, which is most often accomplished by automated
means. Equipment must also be maintained appropriately through routine mainte-
nance, calibration and inspection. GMP requires that the history of equipment use,
cleaning, and maintenance is tracked and documented over the equipment lifetime.
– A document system must be implemented that allows for control, monitoring
and recording all activities which directly or indirectly impact the quality of bio-
pharmaceutical products [103]. Various documents are generated and used as
part of GMP manufacturing. Procedures specify instructions for execution of spe-
cific tasks. Records and reports capture information or data as tasks are exe-
cuted. A common phrase in biopharmaceutical manufacturing is “if it wasn’t
documented, it didn’t happen” – records and reports provide proof of activities.
Specifications provide acceptance criteria or serve as a basis for evaluating qual-
ity at certain steps within the manufacturing process [103]. Control of docu-
mentation requires that changes are not made without appropriate justification,
review, and approval.
Good documentation practice (GDP) is an important concept within GMP that
applies to both written and electronic records and reports. There are many best
practices within the industry related to GDP, including specific ways to correctly
record information and make corrections to original documentation. Documenta-
tion must be attributable, legible, contemporaneous, original, accurate, complete,
consistent, enduring, and available (ALCOA+). When electronic records are used,
there are additional GMP requirements regarding electronic signatures and main-
tenance of data integrity.

– Testing is required throughout the biopharmaceutical process to verify compliance

with established acceptance criteria and specifications. Such testing is required
for process inputs such as raw materials, process consumables, and process solu-
tions. Testing is also performed throughout the manufacturing process on product
intermediates to monitor process performance. Testing of the product – both drug
substance and drug product – at the end of the process is also necessary to ensure
conformance to approved specifications for product SISPQ.
 3.4 Summary 91

– Qualification and validation are performed to provide evidence that a specific pro-
cess or system is capable of meeting predetermined requirements to demonstrate
that it functions as expected. These requirements apply to facilities, utilities, equip-
ment, computerized systems, manufacturing processes, cleaning processes, and an-
alytical test methods. Various documents are generated to plan for and to execute
qualification and validation activities, including master plans, specification and de-
sign documents, and protocols for testing execution and documentation. Validation
of cleaning processes for multi-use equipment is necessary to demonstrate process
and product residues are removed to acceptable levels, to ensure carryover between
batches is minimized. Manufacturing processes are validated to demonstrate con-
sistent production of quality product. In general, process validation is separated
into three stages: process design, process qualification and continued process ver-
ification. In addition to demonstrating consistency of the manufacturing process
through execution of multiple successful at-scale runs, multiple specific studies,
many executed at bench scale, are required to demonstrate an understanding and
control of the process and to establish a complete process validation package.
– Batch disposition is the responsibility of the Quality unit, and refers to drug sub-
stance or drug product being assigned a specific status such as released, condi-
tionally released, or rejected. The decision as to batch disposition depends on
the outcome of review of all records associated with the batch manufacture for
compliance with documentation procedures, a genealogy review to verify that
all inputs to the process were within expiry and are traceable to their original
source, and a review of any testing results and deviations associated with the
batch. If the results of the batch review are acceptable, the batch is disposi-
tioned as released and may be used for further processing or for distribution to
patients. If there are issues identified during batch review, the batch might be
conditionally released pending additional actions to disposition as approved or
may be rejected barring any further GMP use.
– Post-marketing activities take GMP beyond just the manufacture and release of
biopharmaceutical products for use by patients. Biopharmaceutical manufac-
turers must have systems and procedures in place to address complaints or po-
tential quality issues after their product is marketed. In addition, companies
have a responsibility to continue to provide updated information regarding their
product to regulatory agencies on a routine basis. The updates may take the
form of an annual report that summarizes additional process and stability data,
changes, and deviations generated during the previous year.
Chapter 4
Upstream operations

The first step in a biopharmaceutical production process is the generation of product.

Often, this is achieved by cultivation of an engineered cell line at large volumes. Dur-
ing this cultivation, growth substrates (i.e., nutrients that feed the cells) are con-
sumed by the cells and converted into desired product via a combination of native
and recombinant metabolic pathways. Once the upstream process is complete, broth
is harvested that contains the cells used as a catalyst for production, spent medium,
and the desired product. This chapter covers upstream production of biopharmaceut-
icals by answering the following questions:
– How do we manipulate engineered cells into producing our product?
– How do we operate the process to maximize productivity and product quality?
– How are bioreactor vessels designed to support a cell growth process?
– How does cGMP influence bioreactor design, process development, and execu-
tion of an upstream process?
– What are the process parameters that influence cell behavior and process perfor-
mance?
– What process attributes are used to assess the upstream process?
– What are the major considerations during process development, and how are
bioreactor operations scaled up?

Growing cells are the catalyst for the upstream process. As such, their behavior de-
termines the behavior of the overall process. Accordingly, our discussion begins
with information on kinetics of cell growth, which refers to the growth of an individ-
ual cell and and increase in the number of cells, and the factors that influence it.

4.1 Describing cell growth

A variety of different cells are used as factories for production of biopharmaceuticals.

The host is selected based on a variety of criteria. Characteristics of the various cell
types used in biopharmaceutical manufacturing are shown in Table 4.1. The organism
must be generally recognized as safe (GRAS), a designation granted by regulators that
a substance is recognized as safe by qualified experts. Cost of cultivation is another
main consideration that can vary widely from cell type to cell type. Growth rate is simi-
larly important because a faster growing host cell results in a shorter process time,
which leads to higher productivity. A cell line that is tolerant of a range of process con-
ditions may be desirable as it simplifies processing. Many therapeutic proteins require
modifications post-translation in order to be active. Commonly, glycosylation is an es-
sential modification for protein function. If a glycosylated final product is required, a

https://doi.org/10.1515/9783110616880-004
 4.1 Describing cell growth 93

cell type capable of performing that modification must be chosen. Finally, secretion of
proteins to the extracellular medium may simplify and reduce the cost of downstream
processing. Commonly used cell lines are selected for their ability to meet the above
criteria: E. coli (bacteria), S. cerevisiaie (yeast), Spodoptera frugiperda (insect), Chinese
hamster ovary (CHO, mammalian), and human embryonic kidney (HEK, human). Table
4.1 below highlights some of the attributes of each type.

Table 4.1: Characteristics of various cell types used in biopharmaceutical processing.

Bacteria Yeast Insect Mammalian Human

(Nonhuman)

Medium cost Low Low Medium High High

Batch time Hours–Days Days–Weeks Days–Weeks Weeks–Months Weeks–Months

Expression level High High Low–High Low–High Low–Medium

Process ranges Broad Broad Medium Narrow Narrow

Secretion No Yes Yes Yes Yes

Glycosylation None Simple Simple Complex Human

Using living cells as a catalyst offers a variety of benefits that make them suitable for
biopharmaceutical manufacturing. Metabolic processes are specific (catalyzing only a
single reaction) and selective (producing only a single stereoisomer product in most
cases). Furthermore, complex polymerization processes such as DNA replication and
RNA translation are carried out with a fidelity and efficiency that is impossible with
traditional chemical synthesis. These useful processes are all performed during the
course of the cell’s main objective: proliferation. In an upstream process we aim to
facilitate cell growth while simultaneously appropriating these processes to our own
ends: synthesis of a biopharmaceutical. In many cases, the rate of a process, such as
protein or DNA production, is directly proportional to the rate of cell growth [115–117].
As such, understanding the process of cell growth provides insight into how we
might optimize a production process.
The growth of industrially relevant cell lines can be described by four distinct
phases of behavior: the lag, the exponential, the stationary, and the death phase. A
typical growth curve representing each of these phases is shown in Figure 4.1. It
should be noted that the distinction between fermentation and cell culture is de-
fined in earlier chapters. In this chapter, we also use the term culture to refer to ei-
ther the process of cultivating growing organisms or the actual process fluid that
results from culturing cells and that contains a population of growing cells of any
type.
94 Chapter 4 Upstream operations

The lag phase can be described as a period during which the cells are acclimatiz-
ing to their new environment. In most cases, production is initiated by diluting a
high-density culture into fresh medium. This represents a major environmental
change, and the cells require time to adjust. Indeed, physiological changes occur
during the lag phase as activities such as gene expression and regulation adjust.
During this time, there is no appreciable change in the cell density (i.e., the num-
ber of cells per unit volume of broth).
Next is the exponential phase, during which cells begin to grow. Industrial cell
lines proliferate by division of a single cell into two cells. Often called daughter
cells, these two products of cell division are genetically identical (barring any ran-
dom mutation events). This process goes by different names, depending on the type
of cell: binary fission for bacteria like E. coli, budding for yeast like Saccharomyces
cerevisiae, or mitosis for mammalian cells like CHO cells. Regardless of the nomen-
clature, the cells undergo the process of replicating all genetic material as well as
cell components such as organelles and membrane components before segregating
the copies and dividing into two cells.
Log cell number

Stationary phase

Death phase
Lag phase

Log phase

Time

Figure 4.1: Typical growth curve of an industrial cell line. Image © NC State University; reprinted
with permission.

Regardless of the specific type of cell line used, the ultimate outcome of cell growth
and division is the doubling of cells. One cell will become two cells; those two cells
will divide to form four cells; those four will become eight, and so on. This exponen-
tial growth is perhaps most easily described by the Monod model of cell growth,
which assumes the growth rate, dX/dt, is proportional to the concentration of cells:

dX
¼ μX (4:1)
dt
 4.1 Describing cell growth 95

where: X = cell concentration, t = time, and µ = specific cell growth rate [118]. This
simple model plainly states that the rate of increase in cell density is proportional to
some growth constant multiplied by the current cell density. Upon integration of
the differential equation, we see that cell density follows an exponential trend with
respect to time.

X ¼ X0 eμt (4:2)

where: Xo = initial cell concentration. Upon further examination, it is clear that

the rate of cell growth is entirely determined by the numerical value of the growth
rate, µ. Further, growth rate varies as a function of process parameters such as
temperature, pH, and media composition, to name a few [119–121]. This exponential
growth rate continues throughout the exponential phase.
If a nutrient is depleted or byproducts accumulate to toxic levels, growth enters
the stationary phase. During this phase there is little if any cell growth. Cell mass
can be maintained for a short time as the cells scavenge the waste products they
produced during growth on their preferred substrate.
Eventually the medium will be truly depleted of nutrients to the point it can no
longer even sustain the culture that has grown. As cells experience prolonged stress
created by this depletion, they begin to die. This death phase is characterized by a
drop in the viable cell count in the bioreactor.
The overall dynamics of an upstream process are driven by the behavior of the
cells. Cell proliferation is directly coupled to both consumption of nutrients (often
called growth substrate(s)) and formation of desired product. Furthermore, because
of its impact on the rate of cell growth, specific growth rate determines much of the
overall performance of an upstream process.
First, we consider substrate consumption. In order for cells to divide, they must
duplicate everything from DNA to proteins. The synthesis of these cellular compo-
nents is supported by the growth medium. The medium must be able to provide the
atoms required to generate all cellular components – carbon, nitrogen, phosphorus,
sulfur, oxygen, and hydrogen. Generally, the first four are supplied by soluble media
ingredients, while oxygen and hydrogen are provided by aeration and the water in
which the ingredients are dissolved. Carbon is often provided by a carbohydrate such
as glucose or glycerol. Nitrogen can be supplied by a variety of chemicals ranging
from ammonium sulfate to less chemically defined components such as yeast extract
or enzymatic digests of plant or animal proteins. Though media comprises many dif-
ferent components, its function is to provide essential nutrition to the growing cells.
As such, the faster the cells grow, the faster the media components will be consumed.
For example, the rate of consumption of a carbon source can be given by the follow-
ing expression:
96 Chapter 4 Upstream operations

dS μX
¼ (4:3)
dt YXjS

where:
S = unconsumed substrate concentration
t = time
µ = specific cell growth rate
X = cell concentration
Yx|s = yield of cells on substrate.

Here we define a yield of cells on substrate, which is simply the number of grams of
cells produced by consuming a single gram of carbon source. The value of the yield is
influenced by a variety of process factors, such as temperature, pH, carbon source,
and oxygen availability to name a few. Theoretically, under constant conditions the
value of this yield should be constant. This equation shows that not only is the rate of
substrate consumption directly proportional to growth rate but also that substrate
consumption increases as cell density increases. Cell growth continues until the
growth substrate has been exhausted from the medium.
Next, we consider product formation. Biopharmaceutical products are formed as
a result of the proliferation and metabolism of the host cell. During these processes, a
portion of the substrate consumed is directed toward the synthesis pathway for the
desired molecule, while the remaining substrate is consumed to produce components
required for proliferation or waste products. Additionally, the product will constitute
some fraction of the total cell mass generated. As such, there are two ways to describe
product yield: on a substrate-consumption basis or on a biomass-generation basis.
Yield of product on substrate, Yp|s, is simply the number of grams of product formed
when one gram of substrate is consumed. Yield of product on cells, Yp|x, is the num-
ber of grams of product formed when one gram of biomass is generated. We can write
expressions to describe product yield similar to those for substrate consumption in
equation (4.3) using either one of these yields. As with substrate dynamics, we would
see that product formation is proportional to growth rate and cell density. Because
cell growth ceases upon exhaustion of substrate, so too will product formation.

Example: Calculation of the specific growth rate and Yx|s for E. coli BL21
Consider the following growth curve and corresponding data. Use this information to calculate the
following: (a) the growth rate between hours 3 and 2.5 of the fermentation, (b) the doubling time
between hours 2.5 and 3 of the fermentation and (c) the yield of cells on glucose, Yx|s, assuming
an initial glucose concentration of 1.5 g per liter.
 4.1 Describing cell growth 97

Growth of E. coli BL21

0.8 Time Cell Mass
(hours) Concentration (g/L)
0.7
 .
0.6
 .
Cell Density (g/L)

0.5 . .

 .
0.4 . .
0.3  .
. .
0.2  .
. .
0.1

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (hours)
Image © NC State University; reprinted with permission.

Solution
The growth rate between two time points within the exponential phase can be calculated by re-
arranging equation (4.2).


X2
ln ¼ μðt2  t1 Þ
X1
(a) Plugging in the values from above, an expression for the growth rate between hours 3 and
2.5, or µ2.5,3 can be written:
 
ln 0:63 0:231
μ2:5;3 ¼ 0:50
¼ ¼ 0:462 h1
3  2:5 0:5
(b) By definition, a doubling time is the time it takes for cell density to double. In that case the
natural log term would be equal to the natural log of 2, or 0.693. Solving for the time difference
yields the doubling time:

lnð2Þ 0:693
doubling time ¼ ¼ ¼ 1:5 h
μ 0:462 h1
This doubling time indicates that every 1.5 h the total amount of cells in the culture will double.
(c) Finally, the yield of cells on substrate over the course of the fermentation can be calculated
as:

Cell Mass Generated ΔCell Mass

YxjS ¼ ¼
Substrate Consumed ΔSubstrate
Note that in the above equation, Δ refers to the quantity at a later time minus the quantity at an
earlier time. In order to solve for yield with the data at hand, it can be assumed that all the glu-
cose initially present was consumed coincidentally with the start of the stationary phase. As
such:

0:63  0:03
YxjS ¼ ¼ 0:41
ð0  1:5Þ
Every one gram of glucose consumed will lead to generation of 0.41 g of cell mass.
98 Chapter 4 Upstream operations

Finally, it is important to understand that cells will consume substrate to produce

things other than cells and product. This waste production is essential for balanced
and efficient growth. The specific identity of waste products is often cell line and
process specific, but there are a handful that are universally produced. First, carbon
dioxide is a product of several of the steps of the tricarboxylic acid pathway that is
used to produce energy as well as the carbon-containing precursors of amino and
nucleic acids among other molecules. Additionally, organic acids such as lactic acid
and acetic acid are common byproducts of sugar metabolism. These waste products
are important to understand. Their accumulation often has inhibitory effects on cell
growth. Additionally, in some cases cells can utilize waste products as a carbon
source when the initial carbon source has been exhausted.
Ultimately, the dynamics of cell growth and the interaction of the living catalyst
with its medium determine the operational details of a production-scale bioreactor
for biopharmaceutical production.

4.2 Upstream processing and production modes: batch vs.

fed batch vs. continuous
Production of a biopharmaceutical product begins with a vial from a working cell
bank (WCB). For the purpose of this discussion, we assume the process relies on an
engineered cell line to produce a protein therapeutic. Of course, a number of bio-
pharmaceuticals are produced using cells that are not genetically engineered, as
discussed in Chapter 2 (e.g., some flu vaccines); however, much of this discussion
applies to that production scenario as well. The WCB comprises many identical
vials of the cell line that has been grown to a desired point in the growth cycle and
then frozen. This process involves the addition of a cryoprotectant, such as glycerol
or dimethyl siloxane, before lowering the temperature below −70 °C. This combina-
tion of cryoprotectant and low temperatures captures the cells in a state of sus-
pended animation so that when they are thawed, they are resuscitated to the state
they were in prior to freezing.
The WCB is generated carefully from a master cell bank (MCB). The MCB con-
tains cryogenically frozen vials of the host cell line for the process. The characteriza-
tion requirements for the MCB are discussed later in this chapter. Generation of the
WCB is initiated by thawing of a vial of the MCB and subsequently inoculating those
cells into a flask containing sterile growth media. The cells are allowed to grow for a
specific amount of time, corresponding to a specific part of the cell cycle, and pre-
pared for freezing. As discussed above, this involves addition of a cryoprotectant
and controlled freezing to temperatures at or below −70 °C. The size of a WCB is
determined by the expected number of production runs.
 4.2 Upstream processing and production modes: batch vs. fed batch vs. continuous 99

As illustrated in Figure 4.2, a production run is initiated with a single vial from
the WCB. Generally, a cell bank vial contains 1 mL of frozen cells. A production-
scale bioreactor is typically operated at a volume ranging between 2,000 to 20,000 L.
Inoculating such a large volume with a single mL of cells is problematic as seen from
the growth equations (equations (4.2) and (4.3)): cell growth and product formation is
proportional to cell density, and starting with a low cell density in the vessel means
more time is required to reach a desired cell density and production rate. To speed up
growth to the desired number of cells, a typical inoculum volume is 10 percent of the
final working volume of the production bioreactor. Using this 1:10 ratio, a 20,000 L
production bioreactor requires a 2,000 L inoculum. As such, culture expansion from
1 mL of WCB to 2,000 L of inoculum is required. As shown in Figure 4.2, this expansion
is achieved by operation of a series of bioreactors of increasing size leading up to the
production bioreactor, often referred to as a seed train.

Base solution/N2, Acid/CO2

Antifoam
Air Air
Media

CO2 CO2

WCB Shake Shake

thawed
seed vial Seed bioreactor
(Disposable/ Production bioreactor
stainless) (Stainless steel
stirred tank)

Figure 4.2: Expansion of a cell line from cell bank to production scale. Image © NC State University;
reprinted with permission.

The first step in the seed train expansion typically involves inoculation of the working
cell bank into between 500 mL and 10 L of fresh medium. This medium is specifically
formulated to support desired growth rates and final cell densities as well as to estab-
lish a population of host cells that is ready for optimal performance in subsequent
steps. This culture is incubated until the cells are at the desired cell density and stage
of their life cycle. This culture is then used as the inoculum for the next bioreactor in
the series. Each culture step increases the volume of culture and the total cell popula-
tion by a factor of 10. This passaging is continued until the culture has been ex-
panded to the volume necessary to inoculate the production bioreactor.
100 Chapter 4 Upstream operations

Example: Calculation of inoculum volume for a production bioreactor using CHO cells
The target starting cell density for a 4,000 L production bioreactor is 0.4 million cells/mL. After
10 days of cultivation, you measure the cell density in the seed bioreactor to be 5 million cells/
mL. What volume of seed culture should be transferred to the production bioreactor as inocu-
lum? What working volume, that is what total volume of culture, would you recommend for your
seed vessel?

Solution
The process of inoculation is essentially a dilution of one culture with fresh medium. As such a
dilution equation that equates the number of cells before and after dilution can be written:

C1 V1 ¼ C2 V2 (4:4)
Here, C and V are the concentration of cells and volume of culture, respectively. Subscripts 1 and
2 refer to before and after dilution, respectively. Applying the information we have, our initial con-
centration of cells is 5 x 106 cells/mL, and our final target concentration is 0.4 x 106 cells/mL at a
volume of 4,000.

0:4 × 106 cells

mL × 4000 L
V1 ¼ ¼ 320 L
5 × 106 cells=mL

It is good practice to cultivate a volume slightly in excess of what you need for inocu-
lation, so in the example above, a culture of 350 to 400 L would be suitable.
Once the production bioreactor has been inoculated, the process of generating
product begins. In the case of genetically engineered cells, providing the host cell
population with the optimal environment for cell growth and product formation is re-
quired. This is achieved by careful formulation of the growth medium to provide all nec-
essary nutrients as well as control of process parameters that are known to affect the
behavior of the cell population. Once the production run is complete, the bioreactor
contents are harvested and downstream processing begins.
Though the general process of seed expansion and production is similar across
different types of cell lines and products, there are many process parameters that are
manipulated differently for each particular process. In addition to controlling process pa-
rameters to a specific setpoint, which will be discussed in greater detail in subsequent
sections, a ubiquitous strategy for maximizing the productivity – that is, the amount of
product produced per unit volume of bioreactor per unit of time – is through the use of a
particular mode of operation of the production bioreactor. Production modes fall into
three different categories: batch, fed-batch, and continuous. Each mode has its own ben-
efits and drawbacks to be considered before deciding on which to implement.
Batch production is the simplest mode. With this approach, the media is formu-
lated and sterilized with the bioreactor itself. The bioreactor is then inoculated, and
the process is allowed to proceed until the stationary phase is reached. Completion
of the batch typically coincides with depletion of one or more nutrients from the
medium, though it may also follow accumulation of an inhibitory substance to toxic
levels. Alternatively, the batch may be ended at a specific time determined to be the
 4.2 Upstream processing and production modes: batch vs. fed batch vs. continuous 101

most efficient or cost effective. These events correspond with the beginning of the sta-
tionary phase as previously discussed. Batch cultures are easy to perform. One major
consideration in designing a batch culture process is that the cells see a constantly
changing nutritional environment as substrates are consumed and products, including
waste, are formed. This changing environment combined with the exponential dynam-
ics of cell growth can lead cells to experience many different stresses that may have
deleterious effects on growth and product formation. On the other hand, a contamina-
tion event is nearly impossible in this operation mode as the production system is
completely closed upon inoculation. Additionally, executing an effective batch pro-
duction requires very little process knowledge and generally requires relatively simple
controls to maintain process setpoints. Batch mode is ideal for production of products
that have solubility limits in the medium or are toxic to the host cell at lower concen-
trations. Batch processing is also very common in seed-train expansion steps.
Fed-batch production is simply a batch process that has one or more feeds intro-
duced at some point before completion. This additional feeding has many advantages. It
can be used to maintain a constant substrate concentration, leading to more consistent
cell behavior. In some cases it can be used to switch growth substrates in the middle of
the run, which may improve growth or product formation at a key time in the process.
For example, a methanol feed can be introduced to initiate protein overexpression in the
methylotrophic yeast P. pastoris. One of the biggest advantages of fed-batch operation is
the potential to extend the exponential growth phase by feeding additional nutrients.
The exponential nature of cell growth means that the final hour of a production run may
be more productive than the previous 5 or 10 h. As such, extending the exponential
phase by only a few hours may as much as double the final product titer. In some cases,
extension of the exponential phase can be achieved by simply altering the initial media
formulation to contain more substrate. Practically, however, these high initial substrate
concentrations are often inhibitory to cell growth. Alternatively, by accounting for the
rate at which cells are consuming substrate, a complementary feed rate can be set to en-
sure the substrate concentration never becomes prohibitive to cell growth, either too
high or low.
Though fed-batch operation is an effective way to increase productivity, it does have
limitations. First, despite substrate concentrations remaining stable, waste products, in-
cluding toxic ones, accumulate. Furthermore, the more substrate that is fed, the more
waste that is generated and the more severe the inhibitory effects of the waste products
during the later stages of the process. In many cases, this inhibitory effect is what limits
productivity. There are other engineering challenges as well. As cell density increases
exponentially so does the demand for nutrients. At sufficiently high cell densities, it may
be impossible to supply these nutrients in needed amounts due to limitations of sub-
strate solubility, pump speed, air flow, or even total available volume in the bioreactor (i.
e., the volume of nutrient solution required may cause the bioreactor to exceeds its al-
lowable working volume). Additionally, extremes of metabolic activity caused by in-
creased cell density create challenges with respect to controlling parameters such as pH
102 Chapter 4 Upstream operations

and temperature. Finally, design of an optimal feeding strategy requires a more detailed
understanding of the system than batch operation. This understanding must be gained
by analysis of data generated by process runs. Improper design or unexpected departure
from expected behavior often leads to unexpected departure from control of process
parameters, which is catastrophic in a fed-batch process. Despite these limitations,
fed-batch operation is the most commonly used mode for biopharmaceutical produc-
tion owing to increased productivity relative to batch operation with fairly straightfor-
ward feeding strategies.
Continuous production is a mode in which cells are grown in a bioreactor into
which fresh media is being constantly fed at the same rate that spent medium is
removed. This can be done with or without retention of the cells from the waste ef-
fluent stream. If cells are not retained within the process, it operates as a continu-
ous stirred tank bioreactor. If cells are retained within the vessel, the process is
referred to as perfusion. A perfusion process typically employs a microfiltration
membrane as a cell retention device. Perfusion has been used for a number of years
and is in use for approximately 20 biopharmaceutical products [122]. This mode of
operation has the benefit of a substrate feed increasing the productivity of the sys-
tem with the added benefit of removing toxic and inhibitory compounds in the efflu-
ent. Furthermore, in the case that the cell line secretes the product of interest, such
as the production of monoclonal antibodies in CHO cells, this effluent will contain
product. Typically, perfusion culture is only used for production of extracellular
products. This operation mode benefits from a truly constant nutritional environ-
ment for the cell population, leading to much more consistent behavior throughout
the run. This constant productivity means that utilization of the process equipment is
high. This production mode, however, can be costly due to the need for large volumes
of media. Constant feeding of medium also carries an increased risk of microbial con-
tamination. In theory, continuous production can be carried out indefinitely because
cells continue growing exponentially as long as continuous feed and harvest continue.
Practically, however, there is a risk of a genetic mutation occurring during each cell
division cycle. A mutation can potentially lead to reduced cell growth or product forma-
tion or, in the worst case, production of an altered product. Consequently, a limit on
the number of generations in each run is typically defined. The duration of a generation
is defined as the time between cell division events, or the doubling time. Doubling
times range from 30 min for bacteria, to several hours for yeast, to a day or more for
insect and mammalian cell lines.
Because each mode has its benefits and drawbacks, considerations related to
cell physiology, engineering, regulatory, cost, and market considerations are care-
fully weighed when deciding on a production mode.
 4.3 Bioreactor equipment design 103

4.3 Bioreactor equipment design

Proper design of the bioreactor vessel is critical for proper execution of a fermentation or
cell culture process from seed expansion to production. There are two main design crite-
ria for any bioreactor: (1) it must enable aseptic operation so that the risk of contamina-
tion is minimized, and (2) it must be capable of providing an optimal environment for
growth and product formation for each cell grown in the bioreactor. It should be noted
that these two criteria are not of equal importance: the ability to maintain aseptic opera-
tion is paramount to cGMP manufacturing. Any design changes that improve feeding of
media components or oxygen delivery, for example, must be done in a way that does not
compromise the ability to prevent microbial or cross contamination in the process. With
these guidelines in mind, we must also consider the material of construction and its in-
teraction with the growing culture as well as the geometry and configuration of the
vessel and its various attachments. These criteria, combined with the nature of the
process fluid and regulations for cGMP manufacturing, lead to designs that fall
into two categories: reusable or traditional bioreactors and single-use or dispos-
able bioreactors.
Traditional bioreactors are also commonly referred to as stainless-steel bioreactors
in reference to their material of construction. An example is shown in Figure 4.3. Stain-
less steel is used because it is tolerant to the wide range of conditions that a reusable
bioreactor may be subject to. Because the bioreactor is reusable, it must tolerate the
caustic and acidic cleaning solutions used during clean in place (CIP) without degrada-
tion or decomposition. Additionally, it should be inert to all the components of growth
medium and not introduce any toxic components during the production process.
Finally, it must be able to withstand the high temperatures of the sterilization pro-
cess, which can exceed 121 °C. Stainless steel is the most practical material (from
an availability and cost standpoint) that satisfies the above needs. In addition to
the metal used for the vessel construction and the piping used to distribute gas and
liquid components, soft parts are needed to form seals, piping junctions, and sites
of flow regulation, such as harvest or sampling ports. These soft parts must be
made of a material that satisfies the same criteria above. In this case, choices in-
clude Teflon, silicone, ethylene propylene diene monomer (EPDM) rubber, and flu-
ororubber compounds. The American Society of Mechanical Engineers publishes
an excellent standard describing requirements for equipment design that apply to
bioreactors [69].
Single-use, or disposable, bioreactors have different material considerations.
An example is shown in Figure 4.4. First, disposable bioreactors do not need to be
cleaned because they are disposed after use. Further, they don’t need to be sterilized
by the user prior to use because they are typically sterilized, via gamma radiation, by
the vendor. This eliminates the need for a temperature- and acid/base-resistant
material. Typically, these bioreactors, sometimes called “bags,” are made of poly-
meric materials, which are well suited to sterilization by gamma irradiation. Though
104 Chapter 4 Upstream operations

housing

housing

housing
Vessel jacket

DO/pH probes

Agitator motor
Harvest valve

Figure 4.3: Example of a 30 L stainless steel bioreactor. Image © NC State University; reprinted
with permission.

sterilization and cleaning have been eliminated, the materials chosen must be inert
to the conditions present during cell cultivation while providing the structural integ-
rity to contain the volume of liquid and achieve control over process conditions. In
general, no single polymeric material has the appropriate combination of physical
and chemical properties required. Because of this, the walls of the disposable vessel
are made of laminated materials referred to as multi-layer films. The innermost layer
that is exposed to the contents of the vessel is chosen for its chemical stability and
reduced amount of compounds that can be leached by the culture medium. The out-
ermost layer is chosen for its rigidity and mechanical strength to give the bag the abil-
ity to hold a shape. These layers are joined by an adhesive. Between the inert and the
structural layer there may also be a layer of material that confers impermeability to
moisture and gas to prevent losses through the film. Examples of commercially avail-
able single-use bioreactors are given in Table 4.2.
In addition to material considerations, the configuration of the bioreactor is criti-
cal to the ability to provide an optimal environment for the growing cells. The biore-
actor must be configured such that mixing can be achieved, air can be introduced,
heat can be added or removed to maintain temperature, and solutions can be added
or removed to allow for pH control or a variety of feeding strategies. Probes and sensors
 4.3 Bioreactor equipment design 105

(a)

Disposable
bioreactor
bag

Water Jacket/
Support

Agitator motor
Aseptic probe
assembly

(b)
Strength Gas barrier Inert—non leachable
Adhesive Adhesive
“tie” layer “tie” layer

Exterior Bag
of bag interior

Figure 4.4: (a) An example of a 50 L disposable bioreactor and (b) an illustration of the typical
construction of a single-use vessel. Image © NC State University; reprinted with permission.

for monitoring important process parameters must also be incorporated into the system
in a way that useful measurements can be taken.
In stirred-tank bioreactors, whether reusable or single use, mixing is achieved by
attaching impellers to a central agitator shaft, which is turned by a motor. Mixing
serves to distribute nutrients to each and every growing cell within the reactor volume
and to maintain uniformity of important parameters such as pH and temperature.
106 Chapter 4 Upstream operations

Table 4.2: Examples of commercially available single-use bioreactors.

Manufacturer Trade Name Mixing Type Volume

Cytiva Xcellerex Stirred Tank  L to , L

Cytiva WAVE Rocker Bag  mL to  L

Pall Allegro™ STR Stirred Tank  L to , L

Pall Allegro™ XRS Rocker Bag  L to  L

Sartorius Ambr® Stirred Tank  mL to  mL

Sartorius Biostat STR® Stirred Tank  L to , L

Sartorius Flexsafe® RM Rocker Bag  L to  L

Impeller size and shape affect the physics of mixing and must be selected based on the
mixing needs of the production system. Disposable bioreactors can also be configured
so that mixing is achieved by a rocking motion. These aptly named “rocker bags” use a
platform that tilts back and forth at a specified frequency to drive mixing of the liquid
inside. This design does not use a shaft or any impellers and, consequently, delivers
reduced mixing intensity and lower shear.
The two biggest considerations when selecting a mixing strategy and/or an impel-
ler type are the fluid shear generated and the mixing paradigm. Flat and vertical impel-
lers, often called Rushton impellers and shown in Figures 4.5(a) and 4.5(b), push fluid
outward, or radially. This mixing is very efficient, but high shear rates are generated at
the edges of the impeller. These high shear rates can lead to considerable cell lysis with
certain cell lines. Lowering the shear generated can be achieved by curving the impeller
blades as shown in Figures 4.5(d) and 4.5(e). This reduction in shear comes at the cost
of reduced mixing efficiency. Additionally, flow can be translated from a radial to an
axial direction by tilting or pitching the impellers. This axial flow causes pumping mo-
tion in the upward or downward direction, which can help increase mixing efficiency.
Other mixing design considerations are the number of impellers as well as their spac-
ing. Multiple impellers on the agitator shaft should be distributed to ensure consistent
mixing throughout the entire volume of the bioreactor vessel without creating dead
zones. Generally, microbial cell lines can withstand much higher shear rates than
those used in cell culture processes (insect and animal cells).
The location of the drive motor must also be carefully chosen. A bottom-mounted
motor has good power efficiency, but the point of insertion of the agitator shaft will
be below the surface of the liquid. This necessitates the design of a sanitary seal, to
ensure the bioreactor contents do not make contact with the dirty motor, to maintain
sterility. A failure of this seal can lead to either introduction of contaminants into the
culture broth or loss of culture volume through the seal. While top-mounted drive
motor configurations do not carry this risk, the power transfer is less efficient, meaning
 4.3 Bioreactor equipment design 107

(a) (d)

(b) (e)

(c) (f)

Figure 4.5: Examples of flat (a, d), curved (b, e), vertical (a, b) and pitched (d, e) impellers for use in
bioreactor mixing. Illustrations of radial (c), and axial (f) flow patterns. Image © NC State
University; reprinted with permission.

the same mixing performance will require a larger motor. Furthermore, larger bioreac-
tors can be upwards of 40 feet tall. Mounting a large motor over 40 feet off the ground
comes with maintenance, safety, and infrastructure challenges.
Recall from previous discussion that gases, such as air and oxygen, must be fed
to the bioreactor for cell growth. Gas feeds are achieved in one of two ways: below
the liquid surface or into the vessel headspace. Subsurface gas addition is achieved
by sparging a gas through the liquid. A sparger is a device that creates fine bubbles of
a gas to be introduced into the bioreactor’s liquid phase. Gases are pumped through a
108 Chapter 4 Upstream operations

sparger, which is generally located below the bottom-most impeller to allow for opti-
mal incorporation of gas bubbles into the liquid, thereby facilitating maximum mass
transfer. Spargers can vary in shape, diameter, size and number of the openings
through which gases flow, to name a few characteristics. These sparger design param-
eters are manipulated to provide the necessary flow rate and gas transfer rate as de-
termined by process development studies. The second method for introducing
gases into a bioreactor – directly into the headspace, or overlay – is also used. Com-
pared to sparging, it can provide gas delivery without increasing foaming and bubble
formation, both of which can negatively impact cell growth and product formation.
Additionally, feeding gas into the headspace can aid in removing unwanted gases,
namely carbon dioxide, from the culture medium. In any case, these gases must
be sterile upon entry into the vessel. Sterilization is typically achieved by sterile a
0.22 µm filter.
A variety of liquids may need to be introduced during a run, including feeds
and various solutions for pH adjustment to name a few. Liquid feeds are introduced
through feed ports. These ports consist of a length of piping or tubing downstream
of a control valve. These valves open to allow pumping of a solution into the vessel.
These liquid feeds include solutions of carbon or nitrogen sources, acid or caustic
for pH adjustment, or antifoaming agents and are used in all modes of operation.
Inoculation is also carried out using these ports as well. Ports can be configured to
introduce the feed either above or below the surface of the culture broth. This choice
is determined by the nature of the feed, with substrates and inoculum typically
being introduced in a subsurface fashion. These same ports can also be used for re-
moval of culture broth during continuous operation. In this case, the piping must
be below the surface of the broth.

4.4 cGMP considerations for upstream processing

In addition to the considerations made for facility design and environmental control
discussed in Chapter 2, a major consideration for upstream processing in a cGMP en-
vironment is hygienic design of the equipment. This includes everything from the
configuration of piping junctions, to valve architecture, to bioreactor vessel geometry
to allow for adequate cleaning. The goal of hygienic design is to reduce the likelihood
of contamination by carryover from previous processes, adventitious microorganisms,
or other sources.
The need for adequate cleaning between runs drives much of the design of
stainless steel bioreactors. Cleaning is achieved using a CIP system as described in
Chapter 2. Cleaning the interior of a bioreactor vessel is facilitated by a sprayball
device, which uniformly sprays the interior surfaces of the vessel with cleaning and
rinse solutions. Additionally, flow paths (e.g., piping) are cleaned by turbulent flow
of cleaning and rinse solutions across all surfaces that may potentially come into
 4.4 cGMP considerations for upstream processing 109

contact with process fluid. To facilitate this process, the bioreactor must be config-
ured to ensure that no surfaces are obstructed from flow of the cleaning solutions or
shadowed by other parts of the bioreactor, which may reduce or deter proper contact
with cleaning solutions. Flat surfaces, including horizontal runs of piping, should be
eliminated, and the vessel should be capable of fully draining by gravity to ensure
that wet spots where microbes can settle and proliferate are eliminated. Good clean-
ing by flow also requires very smooth surfaces. Rougher surfaces have peaks and val-
leys that may be of the appropriate depth for microbes to attach and avoid removal
by the fluid flow. It is also important that material junctions such as welds and
clamped connections do not create deep pits or grooves where microbes can establish
themselves. This effectively prohibits threaded connections from any area which sees
process fluid. As an alternative, flanged connections are clamped together with a san-
itary gasket forming a tight seal. The worst-case scenario with surface imperfections
is the development of biofilms, which are notoriously difficult to remediate.
The orientation and design of flow paths to allow adequate cleaning is also very
important. Because flowing liquids are used to clean surfaces, it is paramount that
all surfaces are exposed to that flow. One of the biggest challenges in this area is
the elimination of flow paths that cannot be configured to accept this flow during
cleaning. These paths that cannot accommodate flow during cleaning are often
called “dead legs.” Without flow across these surfaces, they cannot be adequately
cleaned of any residual cells, product, or media components between batches. It is
desirable to eliminate any dead legs through system design. In the case of a tee that
is perpendicular to a main run of pipe, as illustrated in Figure 4.6(a), a depth-to-
width ratio for the tee (i.e., L/D in Figure 4.6(a)) of less than 2 is required to ensure
that the tee is adequately exposed to cleaning solution. A typical double block-and-
bleed valve configuration, shown in Figure 4.6(c), is an excellent way to avoid dead
legs while allowing repeated sterilization of piping paths. This configuration arranges
four valves (two block and two bleed valves) such that steam can be directed in one
of two flow paths. These valves can then be manipulated to isolate the region that
has been sterilized to maintain sterility.
Valves and pumps must also be designed so that the surfaces that come into con-
tact with process fluid can be completely cleaned. Diaphragm valves are commonly
used to satisfy this requirement. Diaphragm valves work by actuating a soft dia-
phragm up and down against a weir to permit or stop flow as shown in Figure 4.6(b).
This diaphragm is the only moving part that is exposed to process fluid and is
completely cleanable. A variety of hygienic pump designs exist and should be se-
lected based on flow needs such as viscosity, pressure, and flow rate. Perhaps the
most commonly used pump is a peristaltic type where a soft tubing can be placed in
the pump head, and the flow is controlled by the rate of rotation of rollers that com-
press the soft tubing and force liquid to be pumped through the tube. In this case, the
pump never needs to be cleaned because none of the pump parts ever contact process
fluid (and the soft tubing placed in the pump head is disposed after each use).
110 Chapter 4 Upstream operations

(a) (b)

Diaphragm

Flow

(c)
Hose
connection

Flow

To condensate
drain
From
steam line

Figure 4.6: (a) an example of a dead-leg pathway. (b) an illustration of a typical diaphragm valve
configuration. (c) an illustration and a photograph of a typical double block-and-bleed valve
assembly. Image and photo © NC State University; reprinted with permission.

As discussed in Chapter 3, validation of cleaning procedures is required for reusable

equipment (either dedicated to one product or used for multiple products) to dem-
onstrate that a cleaning process is consistently able to remove process soils and
cleaning agents to acceptable levels between uses. This includes equipment used in
cGMP fermentation and cell culture steps. As previously mentioned, bioreactor clean-
ing is accomplished by pumping the cleaning solutions to a spray device located in-
side the bioreactor. The spray device sprays the interior components of the bioreactor
vessel with water and other cleaning solutions. Consistent performance of CIP proce-
dures is achieved through careful control of pressure of the cleaning solution (at
the spray ball), temperature of cleaning solution, contact time, and concentrations
of cleaning agents. This process must be validated for the removal of cells and
other components of the process between runs as well as the removal of any resid-
ual cleaning agents by final rinsing steps. Cleaning performance is measured by
testing rinse water and swab samples, as described in Chapter 3, and comparing
 4.4 cGMP considerations for upstream processing 111

results to allowable limits. Swabbing locations are often chosen as areas within a
bioreactor that are judged particularly difficult to clean. These challenging loca-
tions may include the underside of impellers or sparge rings, probe insertion sites,
or baffle/wall junctions to name a few. Typically, a minimum of three consecutive
successful cleanings – that is, cleanings that meet all acceptance criteria – are re-
quired to call a cleaning procedure validated. Once validated, the extensive sam-
pling and testing performed during the validation runs are no longer required.
In similar fashion, sterilization protocols must also be validated to ensure that
contaminating organisms are not present at the start of a run. Sterilization is achieved
by exposing all surfaces to steam to raise the temperature to a desired value for a de-
fined time period. The temperature is selected to provide a necessary degree of cell
and spore destruction, typically 121 °C. Validation of sterilization-in-place (SIP) serves
to assure that no residual microbes or their spores have survived the process. If they
do survive, they can potentially begin to proliferate in the growth medium, ultimately
leading to process failure. Detailed guidelines for equipment and sterilization process
design can be found in Technical Report 61 from the Parenteral Drug Association
[123]. Validation of sterilization protocols involves executing the procedure and then
aseptically introducing sterile growth medium into the sterilized vessel. The sterile
growth medium is held for a period of time to allow any surviving organisms to grow.
The process is monitored for signs of growth, such as pH and dissolved oxygen (DO)
changes. Samples are also taken and tested for bioburden. The sterilization process is
modified until all indications of cell growth during a sterile hold are eliminated. Once
validated, evaluation of both cleaning and sterilization efficacy is performed periodi-
cally to guarantee their ongoing acceptable performance.
Use of disposable bioreactors eliminates many – though not all – of the hygienic
design considerations given that cleaning is not required. Many probes and sensors are
still without single-use options. In the case where a reusable probe must be used in a
single-use bioreactor, a method for sterilization and insertion of the probe without
compromising the sterility of the bioreactor is necessary. A variety of proprietary sys-
tems exist to meet this need. Additionally, the connection needed for feeding or harvest
cannot be sterilized by steam as with a stainless steel system. In these cases a tubing
welder is used to make an aseptic connection. As such, the destination vessel as well
as the feed container must have the appropriate weldable tubing present to make the
connection. One final consideration for single-use systems is chemicals from the poly-
meric material leaching into the process fluid. Though complete elimination of leach-
ables may not be possible, identification and characterization of potential chemicals
is necessary as part of process validation (as described in Chapter 3). This characteri-
zation should be performed under conditions comparable to those present through-
out the bioprocess. These conditions may include media composition, temperature,
pH, and duration of exposure.
As discussed previously, the upstream process begins with expansion of cells from
a WCB. There are a number of regulatory considerations related to the generation and
112 Chapter 4 Upstream operations

maintenance of both MCBs and WCBs. The process for generating both is shown in
Figure 4.7. The initiation of a process from a WCB was described previously. That work-
ing bank is itself generated from an MCB. This MCB is generated from a single clone of
the cell line in use for a process. Use of a working cell bank generated from a master
cell bank ensures that each new production batch starts with cells from a population
that is genetically identical. Once a clone is selected as part of cell line development
activities, it is cultivated in flasks until it reaches a desired density and volume. This
culture is then harvested, aliquoted into vials, and carefully frozen. Controlling freez-
ing rate and incorporation of a cryoprotectant help prevent damage to cells during the
freezing and subsequent thawing process. A typical storage temperature is −70 °C for
microbial and −130 °C for animal cell lines. Each and every batch for production of a
particular drug must be initiated from WCBs generated from a single MCB. As such, it
is important to identify an appropriate size for the bank based on expected usage.

Production
process
Single
colony Culture Culture

Master Cell Bank

(MCB)
Working Cell Bank
(WCB)

Figure 4.7: Typical process for generation of MCB and WCB. Image © NC State University; reprinted
with permission.

Once generated, extensive testing is performed on the MCB. Specific guidelines can be
found in FDA and EMA guidance documents [124, 125]. In short, three criteria are as-
sessed: purity, identity, and stability. Purity is confirmed by testing for the presence of
potential contaminating species such as microbes or viruses. DNA sequencing can be
performed to confirm the species and the clonality of the cell stocks (i.e., the identity).
Finally, the bank is grown and cells compared to a similar sample taken from the pro-
duction bioreactor, in which cells have been passaged multiple times, upon the com-
pletion of a batch. This testing is described in Table 3.5. This comparison confirms that
the cell line is suitable for stable cultivation over the course of a production run.
Working cell banks are generated from the master bank as described previously.
Much of the same testing is performed on working cell banks to confirm purity, identity,
 4.5 Upstream process parameters and input material attributes 113

and stability. Generally WCB characterization is less stringent than MCB generation
with the assumption that, barring a contamination or mutation event, the WCB will be
of suitable quality when generated with fidelity from the master bank. Additionally, pu-
rity and identity of cells are monitored throughout the seed train to ensure the same
cells are being cultivated all the way through production [124]. In addition to their gen-
eration, storage of banks is important. The temperature of freezers must be carefully
controlled and monitored to ensure their proper preservation. Generally, redundant
and distinct storage locations are maintained, and access by personnel is strictly
regulated.
Overall, a typical upstream production campaign follows a fairly consistent series
of steps. The initial shake flask is sterilized and filled with fresh sterile medium. The
WCB vial is carefully thawed and used to inoculate the first seed vial. During this cul-
tivation, the next vessel (assuming it is reusable) in the seed train is being cleaned,
sterilized, and filled with sterile culture medium. A sterile hold for at least 24 h is per-
formed to test for latent microbial contamination. The preparation of the next vessel
should be scheduled so that completion coincides with the completion of the previ-
ous cultivation step. A sterile “before-inoculation” sample is taken and sent to QC for
sterility testing, and the vessel is inoculated from the previous seed-train step. End-
point samples for each step can also be collected for analysis, to include potential
contaminants and comparison of cell behavior to a validated process. This process is
repeated at increasing scale until the vessel being inoculated is the production vessel.
This vessel is then operated under the appropriate conditions until completion. At
this time a final sample is collected for analysis, and the contents of the vessel are
harvested for subsequent downstream processing.

4.5 Upstream process parameters and input material attributes

The performance of the cell drives the performance of an upstream process. Because
of this, process parameters for an upstream process are those that affect the physiol-
ogy, metabolism, and ultimately the growth rate of the host cell line. The founda-
tion for cell growth is the nutrients provided by the growth medium. As mentioned
earlier in the chapter, certain elements must be available for the cells to assimilate
and convert into energy and compounds necessary for proliferation and product for-
mation. Of particular importance are the carbon and nitrogen sources. These ele-
ments comprise the bulk of the mass of both cells and products. The specific carbon
source significantly affects energy production as well as the waste products pro-
duced. Glucose, for example, is an ideal carbon source because it is highly oxidized
and able to produce large amounts of adenosine triphosphate (ATP) and nicotin-
amide adenine dinucleotide (NADH) needed to produce the energy required for cell
metabolism. Glycerol, on the other hand, can be catabolized to produce a similar
amount of energy as glucose while producing more reducing equivalents [126]. This
114 Chapter 4 Upstream operations

increased reducing power may be advantageous in production of products that re-

quire excessive reducing power in their metabolic production pathways. In some
cases, this benefit comes at the cost of reduced carbon consumption and lower cell
growth rates.
The specific nitrogen source can significantly affect product formation. Defined
sources such as ammonium salts provide pure, soluble nitrogen for cell metabolism.
This nitrogen must combine with carbon-containing molecules to form amino and nu-
cleic acids before being incorporated into proteins or DNA, however. In some cases,
pure amino acids can be used instead. Cell strains with a deficiency in production of
a certain essential metabolite, or auxotroph strains that require a nutrient that the
normal strain does not, often require supplementation of one or more specific amino
acids. Often, use of pure amino acids can be cost prohibitive, and other nitrogen sour-
ces are used. Complex nitrogen sources such as proteolytic extracts of bulk protein
from varied sources such as soy, animal tissue, milk, or yeast contain a mixture of the
proteinogenic amino acids and nucleic acids along with other vitamins and minerals.
While these complex sources may be more readily metabolized, they come with an
increased lol-to-lot variability. In many cases, a medium will be formulated from mul-
tiple nitrogen sources.
While the identity of media components is known to impact cell growth and
product formation, so too can the relative amounts of different components. The
carbon-to-nitrogen ratio, for example, affects regulation of a variety of metabolic
pathways, which will affect the proliferation of cells and product synthesis. The
ratio of ammonium to free amino acids can affect the uptake of amino acids and
formation of proteins. Other important components of media include buffering agents,
divalent cations, and necessary vitamins and cofactors to name a few. Initial media for-
mulation and feeding strategies (i.e., fed-batch or continuous operation) are carefully
balanced and designed to optimize growth rate and product formation with respect to
total process time, product quality, formation of impurities, and other downstream
processing considerations.
Oxygen is a critically important nutrient that cannot be supplied in excess as a
media component due to its low liquid solubility. As such, it must be continuously sup-
plied from a gas phase in contact with the liquid. Oxygen in the bioreactor is controlled
by achieving a specified DO concentration, an extremely important process parameter.
In the presence of oxygen, cells generate large amounts of energy in the form of adeno-
sine triphosphate (ATP) as well as important cofactors necessary for other metabolic pro-
cesses [127]. When oxygen concentrations drop below a certain threshold, cell
metabolism changes, and growth and production suffer. On the other hand, exces-
sively high oxygen concentrations can lead to oxidative stress from formation of
reactive oxygen species. Agitation, air flow rate to a sparger, oxygen enrichment
of the inlet gas, and vessel pressure are process parameters that are controlled to
 4.6 Upstream performance parameters 115

maintain the optimal DO for the specific process by ensuring the rate of oxygen
supply to the liquid matches the rate at which it is being consumed by the cells.
Another environmental process parameter that affects cell physiology and me-
tabolism is extracellular pH. Each cell regulates the pH within its cytosol and various
organelles. This is achieved by the pumping of protons, or hydrogen ions, across the
various membranes [128–130]. These proton pumps are evolved to function within a
narrow range of surrounding pH values. Any excursions from this pH range mean reg-
ulation is reduced, and a host of cell stresses occur due to protonation or deprotona-
tion of metabolites and enzymes. Extracellular pH is an important process parameter
and control strategies are designed to maintain its value. In some cases, this can be
achieved by including simple buffering components in the medium, but in many
cases acid and/or base addition may be necessary throughout a run. The specific opti-
mal pH value can vary from as low as 4 to as high as 8 depending on the cell line and
the product of interest.
Temperature affects many of the processes occurring at a cellular level in an up-
stream process. The activity of metabolic enzymes is precariously dependent on tem-
perature as are many physiological factors, such as membrane and cytosol fluidity
and regulation of gene expression [131, 132]. Changes in temperature, therefore, lead
to changes in metabolism that in turn affect growth rate, product formation, and
waste generation. Temperature also affects the properties of the medium. Decreasing
temperature lowers the solubility of media components and increases the solubility
of gases, notably carbon dioxide and oxygen. Temperature also affects pH.

4.6 Upstream performance parameters

Though a process is generally designed with a target product titer in mind, other pa-
rameters, such as those that describe cell growth and metabolism dynamics, also give
insight into process performance. Parameters that describe the rates of cell growth or
product formation, for example, capture the kinetics of the cell-based processes. Other
parameters describe the stoichiometry of the relevant reaction, i.e., substrate to prod-
uct, and inform on the metabolic and physiological state of the cell catalyst. Consis-
tency in all of these parameters from run to run gives assurance that the cells are
behaving in the desired way and, by extension, consistently producing the expected
product at the expected quantity and quality.
Lag time, growth rate (µ as defined in equation (4.1)), and total process time are
all performance parameters that may be monitored and characterize the kinetics of
the cell catalyst. The lag time, or duration of the lag phase defined previously, is af-
fected by the state of the cells in the inoculum as well as the conditions in the biore-
actor at the time of inoculation. A departure of the lag time from expected values may
indicate that an error occurred during the seed-train process, including variability in
116 Chapter 4 Upstream operations

the composition of the growth medium. As discussed previously in this chapter, cell
growth varies with a variety of process parameters. Excursions in pH, temperature,
and substrate feed composition, for example, often manifest as a change in the
observed growth rate. Related to growth rate is the total process time. As men-
tioned above, a process may be terminated based on the time of substrate exhaus-
tion or accumulation of inhibitory compounds. Though this time will certainly
vary with growth rate, it may also be affected by variation in substrate availability
or marked changes in the metabolism of the host cell independent of growth rate.
Consistency from run to run in lag time, growth rate, and total process time indi-
cate that fermentation or cell culture step is producing the expected product at the
expected quantity and quality.
In-process measurements such as final total cell mass concentration and final
product titer provide abundant information about process performance as well.
Consistency in these values indicates that the cell catalyst is converting substrate into
more cells and desired product in the expected proportions. This consistency in stoi-
chiometry strongly suggests cellular processes are generating product in the expected
way, implying fidelity of critical quality attributes. Another parameter that suggests
the process is performing in a way to produce product of the intended quality is the
concentrations of select waste products. Though direct measurements of waste prod-
ucts give useful information, many other parameters indirectly inform on the behav-
ior of the system. Because agitation is used to control DO, time trends for agitation
act as a history of the oxygen demand, which is directly related to metabolic activity.
Similar trends can be collected for pH control pumps or exhaust gas carbon dioxide
concentrations, for example.
It should be noted that specific processes may include additional performance
parameters or even omit any of the above. Indeed, the parameters that may influ-
ence performance are diverse and complex. Simply put, however, changes in these
performance parameters can typically be directly attributed to the cell-based cata-
lyst’s significant departure from expected behavior. This departure strongly suggests
that the physiology and metabolism have diverged from that which was validated.
This divergence can have impacts on process productivity and product quality and
can often lead to batch failure.

4.7 Upstream process development

Process development for fermentation or cell culture requires the methodical ex-
ploration of the effects of the numerous process parameters that can be expected
to impact various process outputs. Though mathematical models can be used to
predict process performance with changes in process conditions, the overall com-
plexity and variation in a living biological system makes it difficult, if not impossible,
to completely capture the performance of bioreactor operations in a mechanistic
 4.7 Upstream process development 117

mathematical model (i.e., the use of mathematical expressions that govern the un-
derlying biological processes). As such, development of bioreactor steps is largely
an experimental exercise necessary to generate the understanding required for pro-
cess optimization. The necessary studies are typically conducted using bench scale
(or miniaturized) systems to keep material requirements low. Once the process is
defined, it is scaled up to production.
A major endeavor during upstream process development is determination of an
optimal media formulation. As discussed at length in this chapter, the components
of the media and their concentrations have a tremendous effect on process perfor-
mance. Generally, industrial cell lines have a set of fairly standardized media com-
positions such as a Luria-Bertani broth for E. coli or Dulbecco’s modified Eagle medium
for CHO cells. These general recipes provide a good starting point for media formulation
development, but it is often necessary to modify these formulae to get optimal perfor-
mance from the system. A common approach is to thoroughly analyze the spent me-
dium after a typical run to determine which nutrients may be limiting or are poorly
utilized for that particular cell line. Knowledge of the underlying biochemistry of the
organism can also be combined with this information to determine other supplements
that may need to be added.
Design of experiment (DOE) approaches should be employed to ensure that effects
of process variables – as well as interactions between different process variables –
on the desired outputs of the process are captured with appropriate resolution. It
should be noted that the impact of different process variables may depend on media
formulation. Therefore, relevant variations on media recipes should be included at this
stage. The overall design process is outlined in Chapter 2 as an example for a CHO cell
culture process. Identification of an optimization goal, as well as process limitations, is
an important part of this analysis. An obvious goal is maximization of final product
titer while producing product of the intended quality, as defined by CQAs. This is the
approach described in the example in Chapter 2, but many other criteria are typically
worth considering. When comparing the performance of different batch or fed-batch
production strategies, it can be helpful to optimize another performance parameter:
productivity. As mentioned previously, it is defined as the ratio of product titer to pro-
cess time. This parameter enables direct comparison on both a titer and a time basis.
Other pertinent criteria to consider as part of the design process for a fermentation or
cell culture step may include cost and availability of raw materials, feasibility of control
strategies, or impacts on cost and efficacy of downstream steps. Finally, the selected
conditions must meet the necessary targets for critical quality attributes of the product.
Once complete, process development studies should provide the design space for the
upstream process.
118 Chapter 4 Upstream operations

4.8 Scaling up fermentation and cell culture

As mentioned in Chapter 2, process development studies are typically performed at

small scale with volumes not exceeding two liters. Execution at small scale facilitates
the completion of the high number of necessary runs in a cost- and time-effective
way. Though the information gleaned at development scale is useful, it is unwise to
assume that cell growth and the production process will behave identically at produc-
tion scale, which typically ranges from 2,000 to 20,000 L or more. As such, care must
be taken when scaling up to ensure that the necessary consistency and performance
defined during process development can be achieved at production scale. Scale-up
methods for bioreactors and challenges are presented here.
This chapter has discussed the importance of environmental conditions (e.g., me-
dium pH and temperature) at length. A cell’s environment, in this case, can be de-
scribed by the conditions only nearby, within microns of the cell surface. As such,
heterogeneity of relevant process conditions within the bulk volume of the bioreactor
leads to different portions of the cell population experiencing different environments.
This discrepancy can lead to varied behavior of different cells within the same biore-
actor. This inconsistency, in turn, leads to variability in process performance.
The bioreactor system relies on efficient and thorough mixing to create a system
that is spatially homogenous in process parameters such as dissolved oxygen and
medium components, among others. The effect of mixing on oxygen delivery is dis-
cussed later in this chapter. Many of the problems that arise when scaling up can be
attributed to differences in mixing dynamics at different scales. A bioreactor system
is a complex one. All three phases of matter are present, and their behavior is mean-
ingful to the process. Furthermore, process parameters change across a wide range
of time and length scales: micrometer-sized cells reside in a vessel potentially sev-
eral meters wide growing over hours while oxygen transfer takes place on the order
of seconds. Achieving consistency throughout the bioreactor becomes more chal-
lenging as liquid volumes increase. More power is required to generate the same
sort of turbulent flow because bubbles, solutes, and solids must travel farther to be
evenly distributed, and gravity and pressure effects become more significant.
Because mixing is so critical, scale-up approaches generally aim to keep certain
parameters related to mixing constant as culture volumes increase. Many parameters
can be used to describe mixing: power/volume, Reynolds number, or mixing time, for
example, capture the momentum transfer in the system. Unfortunately, these different
parameters scale differently with volume, and holding one value fixed (e.g., power/vol-
ume) as scale increases means others (e.g., Reynolds number and mixing time) will
vary. If those parameters that vary upon scale-up have a significant impact on mixing
dynamics, performance may begin to decrease with increasing volume. Reynolds num-
ber, for example, increases with the square of the impeller diameter, while the power/
volume ratio increases with the fifth power of impeller diameter. So if bioreactor scale-
up is performed by holding the Reynolds number constant, the calculated impeller
 4.8 Scaling up fermentation and cell culture 119

diameter will result in a power/volume ratio that is lower for the production biore-
actor than the smaller scale unit. From this example, it is clear that the decision
on which parameter to design around is, therefore, critically important and some-
what perilous.
Let’s consider another example to illustrate how holding different mixing param-
eters constant on scale up affects the Reynolds number around the impeller. The Rey-
nolds number is a dimensionless number that captures the ratio of inertial forces of
fluid flow (i.e., the input forces) to the resistive forces from fluid viscosity. At high
Reynolds numbers, above 10,000, the flow resulting from the impeller is turbulent,
which is desirable for effective mixing. For an impeller, it can be written as follows:

D2i nρ
Re ¼ (4:5)
μ

where:
Re = Reynolds number
Di = impeller diameter
n = rotational speed (in revolutions per minute)
ρ = fluid density
μ = fluid viscosity

Consider the case of scaling up a bioreactor by a factor of 1,000 (e.g., scaling up

from 2 L to 2,000 L), so that V2 = 1,000 × V1 (where subscripts 1 and 2 refer to small
and large scales, respectively). To scale up, the following geometric parameters will
be held constant:
– the aspect ratio, which is the ratio of the vessel height to diameter, at a value of 3:1;
– the ratio of impeller diameter to tank diameter.

The volume of each vessel is approximated as the volume of a cylinder and can be
written as follows for each:

2
2
DT;1 DT;2
V1 ¼ πh1 ; V2 ¼ πh2 (4:6)
2 2

where:
DT,1 = diameter of tank 1
DT,2 = diameter of tank 2
h1 = height of tank 1
h2 = height of tank 2
120 Chapter 4 Upstream operations

Using a constant aspect ratio of 3:1, equation (4.6) becomes:

2
2
DT;1 DT;2
V1 ¼ 3πDT;1 ; V2 ¼ 3πDT;2 (4:7)
2 2

Because the ratio of V2 to V1 is equal to 1,000, the ratio of tank diameters can be writ-
ten as:

4 πDT;2
3 3
V2
¼ 1000 ¼ (4:8)
4 πDT;1
V1 3 3

Simplifying the expression by cancelling and taking the cube root of both sides
leads to:

DT;2 ¼ 10 × DT;1 (4:9)

Because we assume the ratio of tank to impeller diameter is constant upon scale-up,
then

Di;2 ¼ 10 × Di;1 (4:10)

where:
Di,1 = diameter of impeller on tank 1
Di,2 = diameter of impeller on tank 2

With the relationship between impeller diameters at each scale established, we con-
sider the value for Reynolds number, Re, at each scale. Assuming that the density and
viscosity of the fluid in the bioreactor are the same at each scale leads to the following:

Re2 ðD2 Þ2 n2
¼ (4:11)
Re1 ðD1 Þ2 n1

Next, we consider holding different mixing parameters constant at scale-up and

then assessing the impact that each scenario has on the Reynolds number ratio in
equation (4.11). In the first scenario, rotational speed (rpm) is held constant at each
scale. In the second scenario, the linear velocity of the impeller tip is held constant
at each scale. If the same rotational speed is used at each scale (i.e., n1 = n2), the
ratio of Reynolds numbers at each scale becomes:

Re2 ð10  Di,1 Þ2

¼ ¼ 100 (4:12)
Re1 ðDi,1 Þ2

If the linear velocity of the impeller tip is held constant between scales (note that
the linear velocity of the impeller is proportional to the impeller diameter, so that
n2 = 1/10 × n1), the following relationship results:
 4.8 Scaling up fermentation and cell culture 121

2
Re2 ð10  Di,1 Þ 101 n1
¼ ¼ 10 (4:13)
Re1 ðDi,1 Þ2 n1

This analysis demonstrates how keeping certain mixing parameters constant – in this
example, either rotational speed or linear velocity of the impeller – can result in sig-
nificantly different values for the impeller Reynolds number, an important measure
of mixing performance. Finding the parameter that best captures the behavior of your
system is therefore critical for successful scale-up.
Aeration is another important consideration during scale-up. Delivery of oxygen
is often the limit on cell growth and is affected by process conditions and bioreactor
architecture. Because of this, it is important that aeration efficiency is maintained
as an optimized process is scaled up. Simple parameters describing aeration, such
as vessel volumes per minute (vvm), superficial gas velocity (flow rate/cross sec-
tional area of bioreactor), or gas residence time, are all straightforward and are eas-
ily scaled. Furthermore, if a consistent bioreactor aspect ratio is maintained, these
will all scale linearly with volume. It should be noted that airflow to the overlay
does not contribute as efficiently to oxygen transfer compared to sparging.

Example: Scaling air flow rates and gas velocities

Consider a pilot scale bioreactor with total liquid volume of 2,000 L and a height-to-diameter
ratio of 3:1. The setpoint for airflow in this process is 1,250 L of air per minute. If the final pro-
duction reactor is to be 15,000 L with the same aspect ratio, what gas flow rate is necessary to
match the vessel volumes per minute used at pilot scale?

Solution
First, the flow rate in vvm is calculated by simply taking the ratio of gas flow rate to liquid volume

gas flow rate ðL=minÞ GFR

vvm ¼ ¼
liquid volume ðLÞ V
GFRPilot 1250 L=min
vvmPilot ¼ ¼ ¼ 0:625 vvm
VPilot 2000 L
Keeping vvm constant allows the following expression:

GFRProduction GFRProduction
vvmProduction ¼ 0:625 ¼ ¼
VProduction 15000 L
0:625 vvm  15000 L ¼ GFRProduction ¼ 9375 L=min

Another parameter used to characterize aeration is the overall oxygen mass transfer
coefficient, often denoted as kLa. The transfer of oxygen from the gas into the liquid is
driven by the concentration difference between the surface of the bubble and the bulk
of fluid as illustrated in Figure 4.8. This gradient exists across a finite distance known
as the boundary layer. kLa is a lumped parameter – the product of kL, the specific
122 Chapter 4 Upstream operations

mass transfer coefficient, and a, the total interfacial surface area between the gas and
liquid phases. kL is a function of the thickness of the boundary layer (i.e., the stagnant
liquid layer represented by the outer layer in tan in Figure 4.8) that surrounds a gas
bubble and through which oxygen must diffuse to be distributed throughout the biore-
actor. Boundary layer thickness is affected by the fluid properties of the liquid to a
large degree, and the intensity of mixing to a lesser degree. The interfacial surface area
is affected by the total volume of air flow as well as the size of the air bubbles. Smaller
bubbles have a higher surface-area-to-volume ratio compared to larger bubbles, and
increasing mixing efficiency helps to create smaller bubbles. It should be noted that
various physical limits, namely surface tension, put a practical lower limit on bubble
diameter. When lumped, kLa is fairly easily determined for a system based on vessel
geometry, total gas flow, and agitation.

Medium

Air Figure 4.8: Diffusion of oxygen from a gas bubble into the bulk
liquid for consumption by growing cells. C* represents the
concentration of oxygen in the liquid medium in equilibrium with
C* the gas-phase oxygen. CL represents the concentration of oxygen
CL in the bulk liquid medium. Image © NC State University;
reprinted with permission.

It should be noted that mixing and aeration within the system are likely to lead to
the formation of entrained foam at the top of the liquid, and increases in agitation
and gas flow will increase foam formation. Foam is problematic for a few reasons.
First, it will entrain cells within the bubble film and create an environment that is
not able to be replenished with nutrients as readily as the bulk liquid. Second, foam
occupies significant bioreactor volume if not controlled and can actually cause the
total volume of the contents to exceed the available volume in the bioreactor and
result in what is referred to as “foaming out.” As such, delivering similar process
performance across scales without causing excessive foaming is an ongoing chal-
lenge with upstream process development and scale-up.
One final consideration for bioreactor scale-up is temperature control. As men-
tioned previously, temperature affects cell growth as well as a variety of other process
parameters, which themselves affect cell growth. Metabolically active cells produce
heat as a waste product, and this heat must be removed in order to control the tem-
perature. This is accomplished by circulating a cooling fluid through the jacket of the
bioreactor vessel to transfer heat away from the vessel. Bioreactor vessels are typically
cylindrical and, as such, as they get larger the surface-area-to-volume ratio decreases.
This decrease in surface area to volume can lead to a scenario in which there is not
enough surface area for heat to be removed as quickly as it can be generated. It
should be noted that this limitation is more commonly encountered during microbial
 4.8 Scaling up fermentation and cell culture 123

fermentation than animal cell culture. This can be overcome by either using cooling
fluid whose temperature can be reduced beyond that of water (water can be cooled to
no lower than its freezing point of 0 °C) or by increasing cooling surface area by add-
ing coils to the interior of the vessel.
There are many different approaches to discern parameters that are important
for describing scale-up effects. First, development of a computational fluid dynamic
model for a bioreactor can help predict how mixing changes as volume increases.
While a computational approach only provides an approximation, it provides in-
sight into which parameters are useful for capturing changes with scale. Ultimately,
it can point to which parameters should be explored experimentally.
Often, performance studies are conducted at pilot scale – that is, a scale between
development and production. The choice of the size of this pilot bioreactor depends
on the ultimate final production scale. At pilot scale, performance is compared to
bench-scale and an assessment made of which mixing and aeration parameters ex-
plain departures in performance. This assessment then informs on which parameters
to hold constant and which to change on scale-up to the final desired volume. This
process is more practical at pilot scale than final production scale for cost reasons (e.g.,
material costs for the studies will be less than at production scale). In addition, the
magnitude of change is smaller and the source of variation is easier to discern.
Another common approach is to use a scale-down model. In this method, you
characterize your production-scale bioreactor and then try to impose similar behav-
ior at the bench-scale that matches the expected behavior of your final production
bioreactor. Limits on parameters such as agitation speed or airflow that exist in the
production bioreactor can be designed into the development bioreactor to restrict
values for characteristic parameters like kLa or Reynolds number. This approach re-
quires knowledge of the characteristics of the final production bioreactor, which
may not have been built or even designed in the earliest stages of process develop-
ment. However, heuristics can be applied that make this approach more certain.
Often, scale-down is applied during troubleshooting of an established process. In
this case, the production-scale bioreactor is scaled down to bench scale to explore
causes of problems at production scale in a fast and inexpensive way.
Scaling up of upstream operations is challenging. Each particular process will
have its own criteria, limitations, constraints, and challenges that can be hard to
accurately predict or explain even with ample data. In some cases where scale-up is
particularly problematic and the desired performance cannot be reached, scale-out
is another option for increasing process capacity. Scale-out increases capacity by
simply operating a larger number of smaller bioreactors that are able to meet neces-
sary desired performance parameters. While this strategy eliminates the need for
scale-up studies, it inherently increases variability in performance owing to each
vessel being a separate production run. This inconsistency comes with other regula-
tory and processing challenges. Furthermore, multiple small vessels have a higher
capital cost than one larger vessel.
124 Chapter 4 Upstream operations

4.9 Summary

The chemical complexity of biopharmaceuticals necessitates growing cells for their

production. A variety of cell lines are used – from simple bacteria like E. coli to mam-
malian (non-human) cell lines such as CHO to human cell lines such as HEK293 cells –
to generate products such as recombinant proteins, viral vectors for gene therapy, or
DNA. Though the specifics of the host cell and desired product may vary from process
to process, the principle is the same: cells consume nutrients and perform a complex
series of metabolic reactions to produce an incredibly complex and highly valuable
product. This chapter describes a general process for taking a cell from a frozen stock
(i.e., working cell bank) to product-containing broth ready for harvest. Key considera-
tions in producing a biopharmaceutical product in the required quantity and with the
intended quality include the following.
– Bioreactor design, which ensures our ability to fastidiously cultivate cells at large
scale. Bioreactor vessels must be designed to deliver an optimal environment for
cell growth and product formation while reducing the risk of contamination. Ma-
terial of construction is an important consideration. Additionally the architecture
of the vessels must be carefully designed to allow hygienic and aseptic operation
in order to eliminate both microbial and cross-contamination. Single-use poly-
meric bioreactors and stainless steel options are both suitable options for up-
stream processing, though each have their benefits and disadvantages.
– Adherence to regulatory requirements, which also drives equipment design consid-
erations. Effective procedures for cleaning and sterilization must be developed
and validated to applicable standards. Additionally, cell banks used to initiate
production runs must be thoroughly characterized to ensure that the desired cell
line and only the desired cell line is being grown for product formation.
– Control of cell growth, which is the main tool for influencing process behavior.
Cells rely on growth media for the nutrients necessary for growth. The specific
medium recipe for each process must provide all the components that cells re-
quire to proliferate and produce the desired product. Simple components such
as sugars and amino acids provide a majority of these needs. More complex com-
ponents such as growth factors, vitamins, or proteins like insulin may be neces-
sary for more challenging host cells or products. In addition to nutrients, cells
require other conditions such as pH, DO, and temperature to be carefully con-
trolled. All of these factors affect the metabolism and physiology of the growing
cells, which, in turn, affects the biological processes they use to produce the de-
sired product. Maintaining consistent cell behavior is therefore crucial for main-
taining consistent product quality. Finally, different feeding strategies can be
employed to improve the productivity and reproducibility of a process.
 4.10 Review questions 125

Finally, approaches and considerations for process development and scale-up were
presented. The goal of process development is twofold: generate process understanding
and define the optimal operating conditions to maximize profitability. This process in-
volves identifying those process parameters that impact productivity and product qual-
ity and testing different combinations until a set of conditions that lead to desired
process outputs is determined. Most commonly, details around medium formulation
and process parameters such as temperature, pH, and DO are identified. Process out-
puts for optimization may include product titer, volumetric productivity, and product
purity. Scale-up is complementary to process development with the goal of applying
the optimal conditions identified during development to a production process at a
larger scale in order to meet market demands. Challenges with scale-up center
around changing mixing dynamics with increasing volumes. Achieving homoge-
neity of process conditions throughout the entire volume becomes increasingly
difficult as volume increases. The heterogeneity that results creates heterogeneity
in the behavior of the host cells and, ultimately, product formation. Different ap-
proaches to scale-up were presented.

4.10 Review questions

1. Consider the growth curve in example 4.1. Identify the lag, exponential, and sta-
tionary phases.

2. You are trying to optimize a medium for production of a protein by fermenta-

tion. After performing several experiments, you determine the following:
– Complete exhaustion of 20 g/L of glucose yields 4 g/L biomass in an other-
wise nutrient-rich medium.
– Repeating the same fermentation with induction of protein expression yields
3.6 g/L of biomass and 0.65 g/L of protein product.
– The biomass is 60% carbon and 20% Nitrogen by mass.
– The protein product contains 15% Nitrogen by mass.

Based on the above, answer the following questions:

(a) What are the values for YX/S,YP/S, and YP/X?
(b) How many grams of nitrogen must be available to the cells to achieve the
biomass and protein yields obtained in experiment 2 above?

You decide to use ammonium chloride as a nitrogen source in your medium.

(c) What biomass and protein titers can you expect if you grow this strain and
induce protein expression in a medium that contains 20 g/L glucose and
2 g/L ammonium chloride?
(d) What will the residual glucose and ammonium chloride concentrations be?
126 Chapter 4 Upstream operations

3. Explain how the following would impact each of the performance parameters:
final total cell mass, growth rate, and final product titer.
(a) Nitrogen content of the medium was less than specified in the formulation
(b) The solution record for a glucose feed had an extra zero, leading to a 10-
fold excess of glucose in the feed
(c) Sterilization was not completely effective and an unknown microbe sur-
vived the process
(d) Double the correct amount of inoculum was added to initiate the run.

4. You are comparing data from three candidate runs. The first is a batch process
that produces 2 g/L product in 10 h. The second is a fed-batch process that pro-
duces 4 g/L in 24 h. The third is a continuous process that produces a harvest
stream with 1 g/L of product at a flow rate of 15 L per hour. Each has a working
volume of 100 L.
(a) Calculate the productivity for each process.
(b) Why might you recommend each over the other options?

5. This number is used to describe how efficiently a stirrer transfers its energy into
movement of the fluid. It can be described as the power consumed by the fluid
as it resists the motion of the impeller. This power is subsequently used to mix
the liquid contents. Consider the example given in 4.8 regarding Reynolds num-
ber at different scale. Perform a similar analysis to determine how power-to-vol-
ume ratio changes with scale.
P
Np ¼
ρn3 D5

Where:
Np: Power number
P: power applied to the fluid
ρ: fluid density
n: rotational speed
D: impeller diameter

6. Derive an expression for the surface-area-to-volume ratio for a sphere. Assum-

ing bubbles in a bioreactor to be spherical, describe how smaller bubbles leads
to increased surface area available for gas transfer.

7. The transfer of oxygen from the gas phase to the liquid phase can be described
by the following equation:
dCL
V ¼ kL aðC  CL ÞV
dt
 4.10 Review questions 127

Where:
V: Culture Volume
CL: Oxygen concentration in the bulk fluid
C*: oxygen concentration in the liquid at the bubble surface
kLa: overall mass transfer coefficient.

You decide to try and quantify the kLa at a specific agitation and airflow. You’ve
calibrated your probe to 100% saturation of the liquid with oxygen. You turn on the
agitation and airflow and record the reading every 10 min for 2 h. Use this data to
estimate kLa.

Time (minutes) Probe Reading (%)

 

 

 

 

 

 

 

 

 

 

 

 

 
Chapter 5
Harvest operations, Part 1: cell lysis

Once the product has been generated, it must be separated from the production sys-
tem (e.g., cells), and all solids must be removed to create a clarified liquid in prepa-
ration for subsequent chromatography steps, which may clog if solids are present in
the feed. The steps involved comprise the harvest stage and are the subject of the
next three chapters.
Most biopharmaceuticals are produced using cells, so feed to the harvest stage
is typically a broth – the cells used to produce the biopharmaceutical suspended in
the spent medium used to support cell growth. For those products that are extracel-
luar – that is, products that reside in the liquid medium surrounding the cell after
production – cells are removed from the broth so that the medium can be processed
further. We discuss methods for separating solids, like cells, from liquid in the next
two chapters. For those products that are intracellular, cell lysis, or the breaking
open of cells, is required to release the biopharmaceutical product into the sur-
rounding liquid phase. Separation of the resulting cell debris from the surrounding
liquid is required afterwards. This chapter covers cell lysis for biopharmaceutical
production by addressing the following questions:
– What is the structure of the cells typically used in biopharmaceutical produc-
tion, and if a lysis step is required, how does the cell structure impact the choice
of lysis technique?
– What are examples of intracellular and extracellular biopharmaceutical products?
– In addition to cell lysis, what other processing steps are used in the harvest
stage for intracellular products?
– What are the different methods used for cell lysis? Which are scalable and which
are not? And, in addition to cell type, what other factors impact the choice of
method?
– How does one of the most common cell lysis unit operations – high-pressure ho-
mogenization – work?
– What are the process parameters that impact performance of a high-pressure ho-
mogenizer, and how are ranges for these parameters determined?
– What is a typical operating procedure for a high-pressure homogenizer used for
current Good Manufacturing Practice (cGMP) production of biopharmaceuticals?

We start the discussion with information on basic cell structure, focusing on the outer
layers of cells that must be ruptured to release intracellular product. This background
information helps to put subsequent discussion on lysis methods in context.

https://doi.org/10.1515/9783110616880-005
 5.1 Structure of cells commonly used in biopharmaceutical production 129

5.1 Structure of cells commonly used in biopharmaceutical

production
Generally speaking, cells can be classified as prokaryotic or eukaryotic. Eukaryotes are
organisms made up of one or more complex cells and include animals, plants, fungi,
and protists. The eukaryotic cell, as shown in Figure 5.1(a), contains a membrane-
bound nucleus; numerous membrane-bound organelles, such as the endoplasmic retic-
ulum, Golgi apparatus, and mitochondria, each with a specialized function;and rod-
shaped chromosomes that carry genetic information for each cell. Eukaryotic cells may
be bound by only a cell membrane or by both a cell membrane and rigid cell wall,
which provide a barrier between the cell cytoplasm (i.e., the liquid that fills the inside
of a cell) and fluid surrounding the cell. Examples of eukaryotic cells commonly used
in biopharmaceutical production and discussed previously include Chinese hamster

(a)
Cell or cytoplasmic Various
membrane organelles

Phospholipid Proteins
Cytoplasm bilayer

Cholesterol
Various Nucleus
organelles (includes DNA)

(b)

Mannoproteins

β-glucans
Cell wall components
β-glucans + chitin

Mannoproteins

Periplasmic space

Cell membrane

Figure 5.1: Basic structure of eukaryotic cells. (a) shows an illustration of an animal cell (such as a
CHO cell), which has no cell wall but does have a lipid bilayer membrane surrounding the cell,
which is shown to the right; (b) is an illustration of the thick cell wall of the eukaryote
Saccharomyces cerevisiae, a yeast. Image © NC State University; reprinted with permission.
130 Chapter 5 Harvest operations, Part 1: cell lysis

ovary (CHO) cells and Saccharomyces cerevisiae, a yeast. CHO cells grown in suspen-
sion are spherical and have a diameter of approximately 15 µm [133]. Saccharomyces
cerevisiae is egg-shaped and has a diameter of 5–10 µm [134]. One big difference be-
tween animal cells and yeast is that animal cells like CHO are enclosed only in a cell
membrane primarily comprised of a lipid bilayer with embedded proteins, while yeast
such as Saccharomyces cerevisiae have a thick cell wall surrounding its cell membrane.
The cell wall for Saccharomyces cerevisiae is made of β-glucans (polysaccharides), pro-
teins, and chitin, and has a thickness of 110–200 nm [135]. The thick cell wall of yeasts
make them more difficult to lyse compared to animal cells bound only by a membrane.
Prokaryotes are less structurally complex single-celled organisms. All bacteria, in-
cluding E. coli, are prokaryotes. Prokaryotic cells do not have a membrane-enclosed nu-
cleus or membrane-bound organelles like eukaryotic cells. In essence, most prokaryotes
are molecules surrounded by a membrane and a cell wall that provide a barrier between
the cell cytoplasm and surrounding fluid. An illustration of bacteria cells, including the
cell walls for both gram-positive and gram-negative bacteria, is shown in Figure 5.2.

(a)
Cell or cytoplasmic
Cell wall membrane
Ribosomes

Plasmid
(extrachromosomal
DNA) Cytoplasm
Nucleoid
(chromosomal DNA)
(b)

Outer membrane
Cell wall
Peptidoglycan components

Periplasmic
space

Cell
Membrane

Gram negative Gram positive

Figure 5.2: (a) Structure of bacteria, a prokaryote. (b) cell walls of gram-negative and gram-positive
bacteria. Image © NC State University; reprinted with permission.
 5.1 Structure of cells commonly used in biopharmaceutical production 131

Note that E. coli is rod shaped with dimensions of approximately 1 × 3 µm [136] and as
such is smaller than either CHO or Saccharomyces cerevisiae.
Because the layers surrounding the cell must be disrupted during lysis, more de-
tail on the constituents of those layers is presented in Table 5.1. Give this table a care-
ful read, as subsequent sections of this chapter rely on the information found here.

Table 5.1: Description of the outer layers surrounding animal cells, bacteria, and yeast.

Cell Type Outer Layers of Cell Composition

Looking from Inside Out

Animal Cell membranea A lipid bilayer with embedded proteins. See Figure .(a).
Cell membranes generally have a thickness of – nm
[].

Yeast Cell membranea A lipid bilayer with embedded proteins, . nm wide in
S. cerevisiae []. See Figure .(a).

Periplasmic space Thin region external to the cell membrane comprised

mainly of secreted proteins that are unable to permeate
the cell wall. See Figure .(b).

Cell wall A thick layer (– nm in S. cerevisiae) made of

polysaccharides (β,- and β,-linked glucans),
mannoproteins, and chitin. Chitin is a polymer of N-
acetylglucosamine []. See Figure .(b).

Bacteria Cell membranea A lipid bilayer with embedded proteins, similar to that
(gram shown for animal cells in Figure .(a). Approximately
negative the same thickness as the cell wall in E. coli [].

Periplasmic space Thin region external to the cell membrane that contains
the peptidoglycan layer and proteins. Part of the cell
wall as drawn in Figure ..

Peptidoglycan layer See Figure .(b). Peptidoglycan is a polymer consisting

of alternating residues of two carbohydrates: β-(., )
linked N-acetylglucosamine (NAG) and N-acetylmuramic
acid (NAM), along with attached peptides. It is a
relatively thin layer, only about – nm thick in
E. coli []. Part of the cell wall as drawn in Figure ..

Outer membrane A lipid bilayer. The outer surface of the bilayer is made
up of lipopolysaccharide and protein.
Lipopolysaccharides are large molecules consisting of a
lipid and polysaccharide. Lipopolysaccharides are
endotoxins, a common impurity in biopharmaceutical
processes, at particularly high levels in E. coli-based
processes. Part of the cell wall as drawn in Figure ..
132 Chapter 5 Harvest operations, Part 1: cell lysis

Table 5.1 (continued)

Cell Type Outer Layers of Cell Composition

Looking from Inside Out

Bacteria Cell membranea A lipid bilayer with embedded proteins, similar to that
(gram shown for animal cells in Figure .(a).
positive)
Cell wall See Figure .(b). The cell wall is made up of
peptidoglycan, described above. In gram-positive
bacteria, the cell wall is thick relative to that in gram-
negative bacteria [].
a
Also referred to as the cytoplasmic membrane and plasma membrane.

The gram-positive and gram-negative designation of bacteria is based on a staining

method known as the Gram stain test. A violet dye is applied to cells, followed by a
decolorizing agent, followed by a second dye that is red. Bacteria with a thick peptido-
glycan layer retain the violet dye used in the Gram stain procedure and are referred to
as gram positive; bacteria with a thinner peptidoglycan layer lose the initial violet dye
during decolorization and are stained pink by the red dye. These bacteria are referred
to as gram negative. Also worth noting is that the outer membrane of gram-negative
cells such as E. coli is a source of endotoxins in biopharmaceutical processes. Endo-
toxin specifically means toxin “from within.” Endotoxins can cause fever, trigger the
coagulation cascade, and cause endotoxin shock among other adverse effects; there-
fore, they must be removed in the downstream process to ensure safe drug product
[140]. The source of endotoxins is the lipopolysaccharides that make up the outer
membrane. Lipopolysaccharides consist of both lipids and sugars. One of the lipids in
the outer membrane of gram negative cells is lipid A, the toxic portion of the lipopoly-
saccharide. When gram-negative bacterial cells are lysed, lipopolysaccharides (i.e.,
the endotoxins) are released. Consequently, endotoxin levels in E. coli-based processes
are particularly high. While endotoxins are a particular concern in an E. coli-based bio-
pharmaceutical processes, steps for removal of endotoxins are designed into processes
that rely on other host systems as well, such as CHO, as contamination by endotoxin-
producing gram-negative bacteria that enter the process stream adventitiously is a
possibility. Endotoxins along with other impurities are discussed in greater detail in
Chapter 8.
Generally speaking, bacteria and yeast, with their tough cell walls, are more dif-
ficult to lyse by mechanical means than animal cells. Among bacteria, gram-positive
cells are more difficult to lyse by mechanical methods than gram-negative cells due
to the presence of a thicker peptidoglycan layer in gram-positive bacteria.
 5.2 The need for cell lysis: intracellular vs. extracellular products 133

5.2 The need for cell lysis: intracellular vs. extracellular products

Once produced, the biopharmaceutical product may remain inside the cell as an in-
tracellular product or be released to the medium surrounding the cells as an extra-
cellular product. Whether a cell produces intracellular or extracellular products
depends on the host system. For example, products produced using E. coli are intra-
cellular. Processes for intracellular products require a cell lysis step for release into
the liquid surrounding the cell. Depending on the process design, this liquid may
be the aqueous spent medium from the bioreactor or a buffer solution used to sus-
pend whole cells recovered from the bioreactor. Processes for extracellular products
do not require a lysis step because the product is secreted out of the cell. Further
explanation of intracellular and extracellular products is provided below:
– Intracellular products reside within the cell cytoplasm or the cell periplasm. Bio-
pharmaceutical products produced in E. coli, such as insulin and human growth
hormone, are examples. The antigen produced for Recombivax HB® hepatitis B
vaccine, which is produced in S. cerevisiae, is another [34].
– Recombinant intracellular products can be soluble, insoluble, or some combina-
tion of both. Insoluble products are produced as inclusion bodies – micron-sized
solid aggregates that consist of relatively pure (typically greater than 80%), but
inactive product [141, 142]. Insoluble production as inclusion bodies can lead to
high purity going into the purification stage of a process because of the high pu-
rity of the inclusion bodies; however, because the protein that makes up the in-
clusion body is misfolded, the inclusion body must be solubilized by unfolding
the protein with a denaturant and then refolding the protein to create a bioac-
tive, soluble biopharmaceutical product. The recovery of product during this refold-
ing step is typically very low due to a significant amount of product precipitation
that occurs [143]. Therefore, soluble production is generally desirable over insoluble
because it results in good overall process yields.
– For intracellular product that resides within the cell periplasm, yields tend to be
low and recovery of the periplasmic protein can be poor. However, an advantage
of periplasmic product is the possibility of designing a cell lysis step that selec-
tively releases the product contained within the periplasm without releasing the
contents of the cytoplasm. This selective release would result in a higher product
purity going into the purification stage of the process. However, methods for se-
lective release of periplasmic product can be challenging to implement [144].
– Extracellular products are released into the medium used to grow the cells. A
common example is monoclonal antibodies produced in CHO cells, which are
capable of secreting proteins [145]. Extracellular products have the advantage of
requiring no lysis step that, in addition to releasing product, releases undesirable
cell components into the process stream; thus, extracellular product typically
results in fewer soluble impurities entering the purification stage. As mentioned
134 Chapter 5 Harvest operations, Part 1: cell lysis

previously, E. coli does not typically secrete recombinant proteins, although

work is being done to engineer E. coli strains to do just that [146].
– Some processes produce significant proportions of both intracellular and extra-
cellular product. An example is the production of adeno-associated virus (AAV)
using human embryonic kidney 293 (HEK293) cells, as shown in Figure 5.3 [147].
AAV viral capsids are commonly used as vectors for gene therapy, as discussed
in Chapter 1. From Figure 5.3, certain AAV serotypes, such as AAV5 and AAV9,
are produced as both intracellular and extracellular capsids using HEK293 cells.
Other serotypes, such as AAV2, are primarily intracellular, while most other se-
rotypes (AAV1, AAV6, AAV7, and AAV8) are primarily extracellular. AAV2 is pri-
marily intracellular in a HEK293 production system due to its affinity to heparin,
which is present on the HEK293 cell surface [147].

Whether product is intracellular, extracellular, or some combination impacts the de-

sign of a production process. Examples of two different harvest process designs for
intracellular products are shown in Figure 5.4 along with a design for an extracellu-
lar product for comparison. Each process starts with broth from a bioreactor and
ends with a clarified process intermediate, prepared for loading to a chromatogra-
phy column in the purification stage.
In the first process shown, cells from the bioreactor are recovered by centrifuga-
tion, then suspended in a buffered solution that is appropriate for the product and
for the first chromatography step in the purification stage. The resulting cell suspen-
sion is homogenized to create the lysate, also referred to as the homogenate when
using a homogenizer, then filtered to produce clarified lysate. In the second design
in Figure 5.4, product is lysed directly in the bioreactor using a detergent. The re-
sulting lysate is passed through a depth filter, or multiple depth-filter stages, for
clarification followed by an ultrafiltration step for product concentration and buffer
exchange. This latter step ensures that product is in the correct buffer system for the
first chromatography column in purification stage. This second design is particu-
larly useful when a significant portion of product is both intracellular and extracel-
lular, as is the case with certain viral vectors used for gene therapy, because all
bioreactor contents move forward in the process. By contrast, in the first process,
because only the cells from the bioreactor move forward to the next step, any prod-
uct that resides in the spent medium would be lost. However, the first process de-
sign offers the advantage of not requiring a buffer exchange step (ultrafiltration)
prior to the purification stage. Suspending cells in the buffered solution required for
the first column chromatography step eliminates this need. The topic of liquid–
solid separation steps (i.e., centrifugation and filtration) that follow cell lysis is cov-
ered in greater detail in the next two chapters.
 5.2 The need for cell lysis: intracellular vs. extracellular products 135

(a)

(b)
AAV2

20 nm

Figure 5.3: Distribution of intracellular (recovered after lysis, in the lysate) and extracellular (in the cell
culture medium) AAV capsids produced in HEK293 cells. The y axis shows the number of genome
copies (i.e., the number of capsids that contain the gene of interest) produced per HEK293 cell (GC/
cell). The x axis shows the AAV serotype. The + and – in the figure indicate whether or not cells were
incubated in fetal bovine serum following the transfection step [147]. Below the graph is an illustration
of an AAV2 capsid. The graph is reprinted with permission from “Efficient Serotype-Dependent Release
of Functional Vector into the Culture Medium During Adeno-Associated Virus Manufacturing,” by Luk H.
Vandenberghe, Ru Xiao, Martin Lock, Jianping Lin, Michael Korn, and James M. Wilson, 2010, Human
Gene Therapy, 21(10), 1253. ©Mary Ann Liebert, Inc [147]. The AAV2 capsid is adapted from PDB ID:
1LP3, The Atomic Structure of Adeno-Associated Virus (AAV-2), a Vector for Human Gene Therapy,
DOI: 10.2210/pdb1LP3/pdb, RCSB Protein Data Bank, rcsb.org, 2002 [148].
136 Chapter 5 Harvest operations, Part 1: cell lysis

Harvest process #1 for Harvest process #2 for Harvest process for

intracellular product: intracellular product: extracellular product:

Broth from Broth from Broth from

bioreactor bioreactor bioreactor

Incubation with
Centrifugation detergent (directly Centrifugation
in bioreactor)

Remove particulate
Recover cells Lyse cells

Cell suspension
(in a buffer)

Remove residual particulate

Remove cell debris
through homogenizer; put (because centrifugation
created during lysis
product in correct buffer alone is unlikely to remove
all particulate)

Homogenization

Concentrate product and

exchange buffer to put
Lyse cells
product into correct matrix

Remove cell debris created

Figure 5.4: Examples of two different harvest processes for intracellular products. Also shown for
comparison is an example of a harvest process for an extracellular product. Note whether or not a
UF step is needed in process #2 depends on the requirements of the next step, usually
chromatography. It is possible that a UF step may be required after depth filtration in the process
for extracellular product as well. Image © NC State University; reprinted with permission.
 5.3 Cell lysis methods 137

5.3 Cell lysis methods

There are a number of lysis methods in use to recover intracellular biopharmaceuti-

cal products. Table 5.2 provides a summary and classifies them as either nonme-
chanical or mechanical.

Table 5.2: Common cell lysis methods.

Method How It Works Scalable? Common Cell Types Suitable

for the Method

Nonmechanical (includes both chemical and physical disruption

methods)

Treatment with Surfactant solubilizes the cell ✔ Animal cells, such as CHO and
nonionic surfactant membrane. HEK
(e.g., Triton™ X-)

Treatment with Organic solvent extracts lipids ✔ Typically not used for
organic solvent (e.g., from the cell membrane of biopharma processing due to
ethanol, isobutanol, animal cells or cell walls, risk of denaturing proteins in
toluene, etc.) thereby increasing presence of organic solvent
permeability of intracellular
components.

Enzymatic treatment Enzyme digests bacterial or ✔ Gram-negative bacteria, such

(e.g., lysozyme) yeast cell walls. For example, as E. coli, gram-positive
lysozyme is a well-known bacteria, yeast such as
enzyme that digests the Saccharomyces cerevisiae
peptidoglycan layer of
bacterial cell walls []. To
use this method, however, a
gram-negative cell would first
be treated with an agent such
as Triton™ X- detergent to
remove the outer membrane
[].

Osmotic shock Osmolality in the surrounding ✔ Primarily animal cells; the

fluid is lowered to drive water force is not great enough to
into cells, breaking the cell lyse tougher bacterial and
membrane. yeast cell walls

Freeze/thaw Suspension to be lysed is X Bench-scale method primarily

exposed to multiple freeze- animal cells, but may find use
thaw cycles; the resulting ice for bacteria
crystals will disrupt the
integrity of cell membranes.
138 Chapter 5 Harvest operations, Part 1: cell lysis

Table 5.2 (continued)

Method How It Works Scalable? Common Cell Types Suitable

for the Method

Mechanical

High-pressure In this book, we use the term ✔ Variety of cells, including both
homogenization high-pressure homogenization eukaryotic and prokaryotic
to refer specifically to forcing a
cell slurry at high pressure
through the small orifice
created by a homogenizing
valve resulting in lysis.
Additional detail is provided in
the text.

Lysis using a A different type of high- ✔ Variety of cells, including both

Microfluidizer® pressure homogenization in eukaryotic and prokaryotic
which a cell slurry is forced
through microchannels at a
high pressure; high fluid
pressure is converted to shear
force, which results in cell
lysis. In this book, we refer to
this type of high-pressure
homogenizer as a
Microfluidizer®.

Bead milling Cells are ruptured by shaking ✔ Commonly used to disrupt

with glass or stainless steel yeast
beads, which are necessary to
promote an abrasive action to
grind the cells.

Sonication Electronic generator sends X Not commonly used in

waves of high-frequency biopharma production; may
energy through a metallic tip. not be suitable for lysis of
Vibration in the cell slurry tough cells like yeast
results in cavitation, which
disrupts cells.

The choice of lysis method depends on a number of factors, notably the ease with
which cells are disrupted so that product can be released (to enable high product
recovery), the ability to maintain the integrity of the target molecule (i.e., avoid de-
naturation), and the scalability of the method. Other factors to consider include the
ease of removal of the cell debris produced (generally, the smaller the debris, the
more difficult it is to clarify the lysate), processing time, and, of course, cost.
 5.3 Cell lysis methods 139

Among the nonmechanical methods shown in Table 5.2 are chemical treatments
used to “permeabilize” the membrane of animal cells, which allows for release of
intracellular product to the fluid surrounding the cell. Nonionic surfactants such as
Triton™ X-100 are used for this purpose, as is treating cells with organic solvents
like toluene. However, treatment with organic solvent is not commonly used in pro-
duction of biopharmaceuticals given the risk of denaturing the product.
Table 5.2 also shows that enzymatic treatment can be used to digest cell walls in
bacteria, such as E. coli, and yeast, such as Saccharomyces cerevisiae, which leads to
cell rupture. For example, the enzyme lysozyme breaks the cell wall of gram-negative
bacteria by catalyzing the splitting of the polysaccharide chains in the peptidoglycan
layer [149]. However, in gram-negative bacteria, access to the peptidoglycan is limited
by the outer membrane, which may inhibit effectiveness of the enzymatic treatment
[149]. Therefore, it may be necessary to expose the cells to Triton™ X-100 or some
other surfactant to solubilize the outer cell membrane. Generally speaking, a disad-
vantage to the use of enzymatic treatment for lysis of bacteria or yeasts is the high
cost and possible lack of availability of the required enzymes [150].
Osmotic shock can be used to lyse animal cells by creating an osmotic pressure
gradient between the liquid surrounding the cell and the interior of the cell. Osmotic
pressure is the hydrostatic pressure produced by a difference in concentration of sol-
utes between a semipermeable membrane. When the extracellular solute concentra-
tion becomes small, the osmotic gradient causes water to flow into the cell, which
eventually leads to rupture. The last nonmechanical method listed in Table 5.2 is the
use of freeze-thaw cycles, particularly for animal cells. Ice crystals that form during
the process disrupt the integrity of cell membranes. While commonly used in a bench-
scale lab environment, this method is not typically used in a manufacturing environ-
ment due to lack of scalability.
Numerous mechanical methods are also available for cell lysis. The four listed in
Table 5.2 are among the most common. High-pressure homogenization is discussed
separately in the next section, as it is the most common of the cell lysis techniques
used for biopharmaceutical production [151]. We use this term to specifically refer to
forcing a cell slurry at a high pressure through a small orifice created by a valve.
A Microfluidizer®, a different type of high-pressure homogenizer, pumps the cell
slurry to an interaction chamber at high pressures – up to 40,000 psi. The proprietary
interaction chamber contains microchannels in which pressure is converted to kinetic
energy, and the resulting shear and impact forces produce cell lysis. Bead mills consist
of a cylinder (grinding chamber) typically filled with glass or stainless steel beads. The
beads and cells are agitated using agitator disks on a shaft inside a grinding chamber.
Cells are lysed due to grinding between and collisions with beads. They are often used
for tough-to-lyse cells like yeast. Ultrasonic disruption (i.e., sonication) relies on high-
frequency sound that is transported through a metallic tip to a cell suspension. The
sound creates cavitation – the formation, growth, and collapse of vapor-filled bubbles,
140 Chapter 5 Harvest operations, Part 1: cell lysis

which leads to shearing and ultimately cell lysis. While effective, this method is often
limited to lab-scale applications.
The mechanical and nonmechanical cell lysis techniques have relative advan-
tages and disadvantages, which are summarized below.
– Mechanical methods are generally effective at lysing tough microbial cells; how-
ever, they tend to break cells completely, which means that all intracellular ma-
terials are released. This results in relatively high amounts of impurities going to
the purification stage. The nonmechanical methods may produce gentler lysis
and therefore less release of unwanted intracellular impurities.
– Mechanical methods tend to create smaller particles than nonmechanical, which
can make subsequent clarification of the lysate by centrifugation and/or filtration
difficult.
– Mechanical methods generate more heat than nonmechanical methods, which can
adversely impact the activity of the product. At manufacturing scale, design of the
lysis step must include ways to minimize the temperature increase. Equipment for
most mechanical methods is equipped with a heat exchanger to cool down product
post lysis. In addition, it is good practice to cool the process feed to these units
(e.g., to 2–8 °C) so that even with heat generation, the lysate temperature remains
low enough to avoid denaturation of the biopharmaceutical product. For many pro-
teins, thermal denaturation begins at 40 °C as explained in Chapter 1. If the feed is
not chilled, temperatures approaching this level are possible in homogenizer lysate.
– Chemical (nonmechanical) methods require chemical additives that must be re-
moved in subsequent downstream processing steps. Mechanical methods re-
quire no additives; therefore, there is no concern over the removal of additional
process-related impurities.

5.4 High-pressure homogenizers

Because high-pressure homogenization is the most common method for cell disruption
in processing biopharmaceuticals, greater detail is provided in this section. Lysis in a
homogenizer relies on pumping a cell slurry through a restricted orifice valve. An illus-
tration of the flow of the slurry through a GEA Niro Soavi Type NS3006H unit is shown
in the flow diagram in Figure 5.5(a). A photo of the unit is shown in Figure 5.5(b).
Cells, suspended in the spent medium from the bioreactor step or in a buffer, are
fed through the product supply line to the piston pump. This flow path is highlighted
in green. A rotary lobe pump in the product supply line keeps the piston pump
flooded with cell slurry and is referred to as a stuffing pump. The piston pump moves
the cell slurry to the homogenizer valve at pressures up to 1,000 bar (≈ 14,500 psi) for
this particular unit. The homogenizer valve is discussed in greater detail shortly.
Once cells are lysed, the lysate stream flows to a heat exchanger, in place to cool
the lysate as a result of the heat generated during lysis. A temperature increase of
 5.4 High-pressure homogenizers 141

(a)

Piston pump Heat exchanger

Homogenizer valve

(b)

Heat exchanger

Control panel
Piston pump

Homogenizing valve

Product return

Product supply

Figure 5.5: (a) Flow diagram for a Niro Soavi Type NS3006H homogenizer. The path of flow for
product is highlighted in green. Major components are circled. The blue path shows chilled water
entering and exiting the shell-and-tube heat exchanger located at the outlet of the homogenizing
valve. Note that variables shown in the green boxes were measured when the system was not
running. Image adapted from the NCSU MMI Manual and used with permission, © GEA Mechanical
Equipment US, Inc.; (b) Photo of a Niro Soavi Type NS3006H high-pressure homogenizer with
various components identified. Photo © NC State University; reprinted with permission.
142 Chapter 5 Harvest operations, Part 1: cell lysis

approximately 2.4 °C per 100 bar of operating pressure can be expected [152]. Once the
lysate passes through the heat exchanger, it exits the homogenizer through the product
return line. The lysate can be sent to the next processing step or recycled to the homog-
enizer for another pass. The number of times the slurry is fed to the homogenizer is
referred to as the number of passes, and it is typically more than one for a homogeniza-
tion step. Operating with multiple discrete passes requires at least two vessels: one that
contains the initial feed slurry and a second that collects lysate from the first pass. To
execute a second pass, the lysate collection vessel for the first pass becomes the feed
vessel for the second and the original feed vessel is used to collect the second-pass ly-
sate. As mentioned previously, it is good practice to cool the feed to the homogenizer
(e.g., to 2–8 °C) to ensure that the temperature of the lysate just after the homogenizing
valve but before the heat exchanger does not result in denaturation of product. Thus, it
is desirable that both process vessels used be jacketed to allow flow of cooling water.
Figure 5.6 shows a photo of the homogenizer valve assembly and an illustration of
how the slurry flows through it. The three pieces shown on the left side of the figure
are the key components. From right to left they are the impact head (that is, the valve),
the impact ring, and the passage head, also referred to as the valve seat. The image on
the right shows how the three components come together to create the homogenizing
valve assembly and the flow through it. The slurry passes through the passage head
(seat) and flows radially through the opening, which is only a few millimeters. The
slurry “hits” the impact ring, then exits the valve assembly as lysate. On one side of the
valve, the pressure is thousands of pounds per square inch as mentioned previously,
and it is controlled by adjusting the distance between the valve and the seat. Once the
cells pass through the small orifice, the pressure drops to approximately atmospheric.

(a) (b)
impact ring
lysed
valve product
seat

product impact pressure

head

valve
seat lysed
product
impact ring

Figure 5.6: The homogenizing valve assembly. (a) shows the three components that make up the valve
assembly, from right to left: impact head (the valve), impact ring, and the passage head or seat. (b)
shows the path of fluid through the homogenizing valve assembly. Note that the valve design shown in
this figure is referred to as a sharp-edge or knife-edge valve and is common for cell disruption. Other
valve designs exist. Photo and image © NC State University; reprinted with permission.
 5.5 Homogenizer process and performance parameters 143

There are several mechanisms likely to create lysis in a high-pressure homoge-

nizer, including (1) shear in the liquid, (2) impingement against the impact ring, (3)
turbulence (and turbulent eddies), (4) cavitation forces, and (5) pressure drop across
the valve. While numerous studies have been performed over the years to bring clar-
ity to which of these mechanisms is the most important, the precise mechanism re-
mains unclear [150, 152].

5.5 Homogenizer process and performance parameters

Let’s start our discussion of parameters that impact homogenizer performance by

considering a model presented by Hetherington et al. [153] for the homogenization
of yeast. It assumes that the lysis process follows first-order kinetics – meaning that
the rate of product released from a cell is proportional to the amount of releasable
product. From this assumption, they went on to derive the following expression,
which fit their data well:

Rmp
log10   ¼ KNpasses Pα (5:1)
Rmp  Rp

where:
Rp = product released per unit cell mass
Rmp = maximum amount of product that can be released per unit cell mass
K = empirical constant that is a function of temperature
Npasses = the number of times suspension passes through the homogenizer
P = pressure
α = empirically determined constant

We will use equation (5.1) as a guide to understanding the parameters that impact
lysis in a high-pressure homogenizer. From equation (5.1), pressure and number of
passes impact the amount of product released (a performance parameter) during a
homogenizer run; specifically as the pressure and/or number of passes increase, the
amount of protein released, Rp, increases. These two process parameters are directly
controllable and would require a range (for pressure) or target (for passes) to be speci-
fied in a batch production record for a cGMP homogenization step.
Data showing the impact of pressure and passes on the homogenization of E.
coli genetically modified to produce green fluorescent protein (GFP), in this case
used as a surrogate for a protein therapeutic, are shown in Figure 5.7. The data were
generated from a bench-top high-pressure homogenizer, in which five passes were
run at 300 bar (=4,350 psi) and at 900 bar (=13,050 psi). Lysate from each pass was
centrifuged, and the concentration of GFP in the resulting supernatant, reported
144 Chapter 5 Harvest operations, Part 1: cell lysis

in units of mg GFP/mL of solution, was measured. As equation (5.1) would predict, a

higher number of passes and higher pressure lead to more released product. At
900 bar, the concentration of GFP in the liquid levels off at approximately 2.66 mg/mL,
which represents the maximum amount of GFP to be extracted from the E. coli cells
used in this study. Note that data for the E. coli/GFP system (not shown here) demon-
strate that pressures above 900 bar do not result in release of additional product.
In Figure 5.7(b), the data shown in Figure 5.7(a) are plotted in a slightly differ-
ent way, as the extent of lysis for each pass. The extent of lysis is a performance
parameter that measures the percent of the maximum concentration of product that
has been released at any pass and is calculated by the following equation:

C  Co
Extent of lysis ð%Þ ¼ × 100 (5:2)
Cmax  Co

where C is the concentration of the target component (product), Co is the starting

concentration of the target, and Cmax is the maximum concentration of target after
infinite time (assuming no denaturation of this compound).
The data in Figure 5.7 can be used to set a range on homogenizer pressure and
target value for the number of passes. Specifically, three passes at 900 bar release
most product from the cells; therefore, for cGMP manufacturing, batch record instruc-
tions would specify using three passes at, say 850–950 bar with a target or set point
of 900 bar. No range on passes (e.g., 2–4 passes) is needed as executing three discrete
passes is straightforward. Most production homogenizers are equipped with auto-
mated pressure control, so the range around the 900 bar set point really depends on
the ability of the homogenizer to hold the set point. Note that operating with more
than three passes not only offers no benefit to product recovery but would have unde-
sirable consequences such as increasing the process time and decreasing the size of
the cell debris, making subsequent clarification steps more challenging. As men-
tioned previously data not shown for this E. coli/GFP system demonstrate that pres-
sures above 900 bar do not result in greater release of product in fewer passes.
Using the data shown in Figure 5.7 alone may be insufficient to set a range on op-
erating pressure and a target on the number of passes. Additional data showing the
impact of passes and pressure on the activity of the biopharmaceutical – that is, mea-
suring activity as a performance parameter in addition to product concentration in the
liquid – would more completely characterize the homogenization step and allow for a
choice of process parameter values that ensures no negative impact to product quality.
In addition to pressure and the number of passes, other parameters affect ho-
mogenizer performance. These are summarized below:
– Feed temperature. The empirical parameter K in equation (5.1) increases with tem-
perature. Consequently, the extent of lysis can generally be improved by increasing
the temperature of the slurry fed to the homogenizer [152]; however, doing so poses
a risk to the product given that increasing the temperature of the homogenizer feed
 5.5 Homogenizer process and performance parameters 145

(a)

3.0
2.63 2.65 2.66
2.56
2.5 2.41 2.37 2.41
2.23
GFP Concentration, mg/mL

2.0 1.94

1.60
1.5

1.0

0.5

0.03 0.03
0.0
Feed 1 2 3 4 5
Passes
300 bar 900 bar
(b)

100

90

80

70
Extent of lysis, %

60

50

40

30

20

10

0
Feed 1 2 3 4 5
Passes
300 bar 900 bar

Figure 5.7: Impact of operating pressure and number of passes on release of GFP from E. coli using a
GEA Niro Soavi Type NS1001 homogenizer. The E. coli cell line used is BL21(DE3)pET-17B GFPuv. (a)
shows the concentration of GFP in the liquid phase of the lysates generated. (b) shows data from (a)
converted to extent of lysis calculated by equation (5.2). The calculation was performed assuming that
the average of the GFP concentration values from passes three through five at 900 bar represent the
maximum concentration of product that could be obtained (Cmax). Images © NC State University;
reprinted with permission.
146 Chapter 5 Harvest operations, Part 1: cell lysis

would result in an even higher temperature after the slurry has passed through the
homogenizing valve, which could lead to product denaturation.
– Cell type. The parameters α and K, and therefore the extent of lysis, depend on
the type of cell. As discussed previously, yeast, with their thick cell wall, are
more difficult to lyse than gram-negative bacteria such as E. coli, which has a
thinner cell wall. In a biopharmaceutical process, the cell line being used is set
as part of process design and is not directly controllable.
– Numerous others. The growth phase of the cell at the time of bioreactor harvest,
the growth medium composition, and the design of the homogenizer and the ho-
mogenizing valve all impact the extent of lysis in a single pass [152].

It is also worth noting that cell concentration in the homogenizer feed has been
found to have only a minor effect on the extent of cell lysis in homogenizers, based
on a study done using E. coli [154].
The performance of a homogenizer can be quantified by either indirect or direct
methods. We’ve already mentioned the amount of product released and the extent
of lysis as performance parameters. Measuring either would be an indirect method
for measuring homogenizer performance. In contrast, counting the number of cells
destroyed would be considered a direct method. Quantifying the intracellular prod-
uct released is the method used to generate the data shown in Figure 5.7. Although
classified as an indirect method, it offers the advantage of providing a direct answer
to the question, “How much biopharmaceutical product can be recovered from the
cells?” – a question that needs to be answered for the sake of process design.
The time required to complete a homogenization step, tprocess, can be calculated
by the ratio of the total volume fed to the unit, which must account for the possibil-
ity of more than a single pass, to the volumetric flow rate or

Vfeed Npasses
tprocess ¼ (5:3)
Q

where Vfeed is the volume (e.g., units of L) of slurry fed in a single homogenizer pass,
Npasses is the number of passes, and Q is the volumetric flow rate (e.g., units of L/h)
through the homogenizer. Vfeed is determined by process requirements. The number
of passes needed is determined through a study, like the one previously described, to
determine the number of passes necessary for maximum product recovery. Q depends
on the operating pressure and size (e.g., model) of high-pressure homogenizer that
you are working with. The supplier should be able to provide data showing the de-
pendence of flow rate on operating pressure for a given homogenizer model [155].

Example: Calculating the homogenizer flow rate required to achieve a given processing time
You are developing a high-pressure homogenization step to process 300 L of suspended E. coli
cells. Development studies show that two passes at 900 bar are required for acceptable lysis.
 5.7 Summary 147

The homogenizer supplier wants to know what flow rate is required, so they can recommend a
model. What is your response?
Solution
Start by solving equation (5.3) for Q:

Vfeed Npasses
Q¼
tprocess
Vfeed and Npasses are known, but a suitable process time must be chosen. Three hours seems rea-
sonable given that you want to complete the homogenization step in a single eight-hour shift and
allow enough time for setup and cleanup. Therefore

L × 2 passes
300 pass
Q¼
3h
¼ 200 L=h

5.6 Procedure for a high-pressure homogenization step

The procedure that we use for a (simulated) cGMP high-pressure homogenization

step using the equipment in Figure 5.5 is depicted in Figure 5.8. After the homoge-
nizer is set up, slurried cells are fed to the unit. Multiple passes may be required to
completely lyse cells as described previously. The high-pressure homogenizer is
then flushed with a buffer, and recovered buffer is added back to the final lysate to
capture any residual product remaining in the hold-up volume of the unit. The sys-
tem is then immediately cleaned and sanitized using a validated method. Vendors
can make recommendations on appropriate cleaning solutions to use for their spe-
cific equipment – 0.5 M NaOH is common. Recall that an open downstream process
such as that being described is expected to be under bioburden control (rather than
sterile), so the high-pressure homogenizer is typically sanitized rather than steril-
ized between runs. The homogenizer is then rinsed with high purity water to remove
residual cleaning/sanitizing agents and stored dry.

5.7 Summary

For a biopharmaceutical process in which product is intracellular, a cell lysis step is

required that releases the product into the liquid phase that surrounds the cell. Be-
cause a variety of cells are used to produce biopharmaceutical products, there is var-
iation in the outer layers that must be disrupted to extract the target molecule from
the cell. For example, animal cells are enveloped with a membrane composed of a
lipid bilayer but lack a cell wall. Prokaryotic cells such as bacteria, other eukaryotic
cells like fungi, including yeast such as Saccharomyces cerevisiae, and plant cells all
148 Chapter 5 Harvest operations, Part 1: cell lysis

Suspend cell paste in

appropriate buffer

Set up homogenizer

Feed cell slurry to homogenizer

Feed lysate from previous pass if

multiple passes needed

product recovery

Clean/sanitize system with

0.5M NaOH

Flush system with

high purity water (HPW)
Figure 5.8: A typical procedure for recovery of
intracellular biopharmaceutical products using a high-
pressure homogenizer. Note that these steps are based
Store homogenizer dry on the first process design option presented in Figure 5.4.
Image © NC State University; reprinted with permission.

have both a cell membrane and a thicker cell wall, which make them more diffi-
cult to lyse by mechanical means than animal cells. The composition of these
outer layers surrounding the cell impacts choice of lysis method, and there are a
variety to choose from. The choice of method depends on a number of factors, no-
tably the ease with which cells are disrupted so that product can be released and
the ability to maintain the integrity of the target molecule (i.e., avoid denatur-
ation). Other factors to consider include the ease of removal of the cell debris pro-
duced – generally, the smaller the debris, the more difficult it is to clarify the
lysate – level of soluble impurities released, processing time, and, of course, cost.
Examples of nonmechanical methods that are scalable and that can be used in
biopharmaceutical applications include:
– treatment with nonionic surfactant to solubilize cell membranes, which can be
used to lyse animal cells;
 5.8 Review questions 149

– enzymatic treatment to digest cell walls of bacteria and yeast;

– osmotic shock to drive water into cells thereby breaking the cell membrane.

Examples of scalable mechanical methods that can be used in biopharmaceutical ap-

plications include high-pressure homogenizers, which send a slurry at high pressure
through a small orifice created by a homogenizing valve, and bead mills that rupture
cells by grinding with glass or stainless steel beads. Generally, mechanical methods are
effective at lysing tough microbial cells; however, they tend to break cells completely,
which means that all intracellular materials are released. They also create smaller par-
ticles relative to the nonmechanical methods, which can make subsequent clarification
difficult. And they generate heat, which can adversely impact product activity. Nonme-
chanical chemical methods, on the other hand, require additives that must be removed
in subsequent downstream processing steps.
High-pressure homogenization is the most common lysis method used in bio-
pharmaceutical production. Development studies should focus on specifying ranges
for the operating pressure and number of passes that ensure maximum recovery
without denaturing the product or creating cell debris that cannot be removed in
subsequent clarification steps. Allowable ranges for each would be included in a
batch production record.

5.8 Review questions

1. D.V. Yates, an operator in the large-scale recovery suite, has frequently failed to
follow instructions in the batch record for the high-pressure homogenization of E.
coli used to produce his company’s blockbuster protein therapeutic. You are the
process engineer assigned to this product and are responsible for writing the
product quality impact assessment for these deviation reports. Explain whether
or not the following deviation scenarios impact product quality.
(a) The homogenization pressure is above the range written in the batch record.
(b) Fewer than the number of passes required by the batch record are used.
(c) The homogenization pressure is below the range written in the batch record.
(d) Cooling water was not fed to the jacketed vessels used for the initial homog-
enizer feed or the lysate collection as directed by the batch record.

2. Lysis of cells using a high-pressure homogenizer usually requires multiple passes

to achieve the desired extent of lysis. In addition to implementing discrete passes
as described in the chapter, describe two other means to achieve multiple passes.
What are the advantages/disadvantages of each?

3. You are developing a high-pressure homogenization step to process 1,000 L of

suspended E. coli cells. Your development studies show that four passes at
700 bar are required. (a) Among the units listed in the table below, which will
150 Chapter 5 Harvest operations, Part 1: cell lysis

Table 5.3: Flow rates in L/h as a function of pressure for various high-pressure homogenizer
models.

Pressure (bar)
Model
     

A      

B ,     

C , , , ,  

D , , , , , ,

E , , , , , ,

you pick for the large-scale process and why? (b) For the model chosen, what
will the actual processing time be?

4. Using a high-pressure homogenizer to lyse E. coli, you use three passes at 300 bar.
Estimate the extent of lysis, assuming the first-order kinetic expression in equation
(5.1) applies, with constants of K = 6.5 × 10–4 and α = 1.71, for a pressure in MPa.

5. 400 mL of E. coli cell paste, generated by removing the spent medium from a fer-
mentation broth by centrifugation, was suspended in 4,000 mL of tris buffer to
conduct range-finding studies for a homogenization step. The resulting cell sus-
pension was homogenized to completion, using five passes at 900 bar, and the
concentration of the protein therapeutic in the clarified lysate was 2.5 mg/mL after
the fifth pass. The cells came from a 25-L fermentation with a solids concentration
of 8% by volume. Assuming the homogenizer run extracted all of the intracellular
protein therapeutic, calculate the fermentation titer.

6. Your course instructor gives the following quiz question: Which cell lysis method
is likely to lead to a simpler purification “process” – high-pressure homogenization
or treatment with a non-ionic surfactant? You circle treatment with a non-ionic sur-
factant. Your instructor says you are wrong, that high-pressure homogenization re-
sults in simpler purification because there are no chemicals added that must be
subsequently removed. You still think your answer is correct. Why?

7. You are working on the design of an animal cell-based process for viral vector
manufacture for a gene therapy product, and your company has set a goal to use
only single-use equipment to minimize capital expenditures. How would you de-
sign your lysis step to meet this goal? In answering this question, address the fol-
lowing: (a) what method/equipment would you use for the lysis step, (b) where
in the process would you implement the lysis step, and (c) what process and per-
formance parameters are important to consider in process development studies?
Chapter 6
Harvest operations, Part 2: solid-liquid
separations by centrifugation

In this chapter we turn our attention to solid-liquid separation, which is required

throughout a biopharmaceutical process and, in particular, in the harvest stage. As
we have discussed previously, regardless of whether the biopharmaceutical product
is intracellular or extracellular, the product intermediate from the harvest stage
needs to be clarified – that is, made free of solids. More specifically, that intermedi-
ate typically comprises the active biopharmaceutical dissolved in a clarified liquid,
such as spent media or a buffer solution, and is ready to be loaded to a chromatog-
raphy column in the purification stage. The intermediate must be clarified because
the first chromatography step typically involves a packed bed that is susceptible to
clogging if solid particles are present in the feed. In addition to this application,
solid-liquid separation steps find other important uses in biopharmaceutical pro-
cesses, and these are discussed shortly.
In this book, we focus on those techniques that are well suited to separating a
relatively large amount of solid from a slurry as is required in the harvest stage of a
biopharmaceutical process. This chapter is the first of two that considers these unit
operations. In it, we focus on one of the most common methods – centrifugation –
while the next chapter covers two common filtration methods – depth filtration and
tangential-flow microfiltration. This chapter opens with a brief general discussion
on solid-liquid separation in biopharmaceutical processes before turning its atten-
tion to centrifugation. Questions addressed include the following:
– What are examples of solid-liquid separations performed in biopharmaceutical
processes? And, generally speaking, what unit operations are used to carry out
these steps?
– What are the principles underlying centrifugation for separation of solids from
liquids?
– How does a disc-stack centrifuge compare to a gravity settling tank for continuous
liquid-solid separation?
– In addition to disc-stack centrifuges, what other types of centrifuges are used in
biopharmaceutical processes?
– What are the process and performance parameters for disc-stack centrifugation
steps?
– What is sigma analysis and how is it used in centrifuge scale-up?

https://doi.org/10.1515/9783110616880-006
152 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

6.1 Solid-liquid separations in biopharmaceutical processes

Previously Figure 2.3 showed a process flow diagram depicting monoclonal antibody
(mAb) production using Chinese hamster ovary (CHO) cells that secrete product to
the cell culture medium; that is, the product is extracellular. From that diagram, it is
clear that steps for solid-liquid separation are required for:
1. removing cells and cell debris from the cell culture broth, leaving a product that
is dissolved in the spent clarified liquid medium, and that this separation is car-
ried out by centrifugation and depth filtration.
There are also a number of other uses for solid-liquid separation steps in biophar-
maceutical processes, including:
2. recovering cells from the bioreactor for further processing of intracellular prod-
uct as depicted in Figure 5.4;
3. clarifying product following a cell lysis step, also shown in Figure 5.4;
4. recovering inclusion bodies for processes in which the host system (e.g., E. coli)
produces the product (a protein) in this form. Inclusion bodies were discussed
in Chapter 5;
5. recovering precipitated product or removing precipitated impurities in pro-
cesses that use precipitation steps for purification. Note that the term precipita-
tion refers to a process step in which a dissolved target component – either the
product or impurities – is separated from other components also in solution by
adding a reagent that decreases the solubility of the target and causes it to fall
out of solution as a solid.
6. removing microorganisms to produce a bioburden-free or sterile process stream.

Table 6.1 summarizes these scenarios and lists techniques that can be used to per-
form the separation.

Table 6.1: Solid-liquid separation examples common to biopharmaceutical processes and the unit
operations used to perform the separation.

Solid–Liquid Separation Scenario Commonly Used Technique

(Refer to text above.)

 Centrifugation, depth filtration, tangential-flow

microfiltration, flocculation followed by centrifugation or
depth filtration, expanded bed chromatography

 Centrifugation, tangential-flow microfiltration

 Tangential-flow microfiltration, depth filtration, flocculation

followed by centrifugation or depth filtration, expanded bed
chromatography
 6.2 Centrifugation principles 153

Table 6.1 (continued)

Solid–Liquid Separation Scenario Commonly Used Technique

(Refer to text above.)

 Centrifugation, tangential-flow microfiltration

 Centrifugation, depth filtration, tangential-flow

microfiltration

 Surface (membrane) filtration

In the next chapter, we discuss the choice of a solid-liquid separation method in

greater detail. Generally speaking, however, the method chosen and designed should
(1) provide effective separation of the solid from the liquid, (2) produce a good prod-
uct step yield, (3) be scalable, (4) be rapid, and (5) be cost effective. Centrifugation
typically meets these criteria and is the topic that we now turn to.

6.2 Centrifugation principles

Centrifugation uses the density difference between solids and the surrounding liquid
as well as centrifugal force to separate denser suspended solids from their surround-
ing liquid. We begin the discussion by considering centrifugal force. It is an apparent
force that acts on an object moving in a circular (or curved) path and creates move-
ment outward from the center of rotation. There are numerous examples of centrifu-
gal force in our everyday lives. It is the reason that clothes in a washing machine get
pushed to the wall of the machine drum during the spin cycle. The greater the speed
of rotation and distance from the center of the motion, the greater the apparent force.
Centrifuges separate liquid from solids by rotating – that is, generating centrifugal
force – which causes particles dispersed in a liquid to move away from the axis of rota-
tion of the centrifuge as long as the density of the particles is greater than the density
of the liquid in which they are suspended. Centrifuges have been used for many years.
A relatively simple bench-top model is shown in Figure 6.1(a). Samples are put into
small tubes, which are placed in the rotor. The rotor in the centrifuge in Figure 6.1(a)
spins at a maximum rate of 14,000 revolutions per minute (rpm), and solids quickly
settle to the bottom of the tubes as illustrated in Figure 6.1(b). To understand the mag-
nitude of the centrifugal force that can be generated, it is useful to compare centrifugal
force to gravitational force by defining a dimensionless acceleration, referred to as the
relative centrifugal force (RCF), as follows:

ω2 R
RCF ≡ (6:1)
g
154 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

where ω = angular velocity (in units of rad/s, for example), R = distance from center of
rotation, the product ω2R is the centrifugal acceleration, and g = acceleration due to
gravity (= 9.8 m/s2). RCF provides a measure of the outward force generated by the rota-
tion of the centrifuge relative to the force of gravity. In the Eppendorf centrifuge shown
in Figure 6.1(a), the unit spinning at 14,000 rpm generates an RCF of 16,888 (calcula-
tion shown in Figure 6.1(c)), often written as 16,888 × g; thus, the force acting on the
particles is 16,888 times the force of gravity, which results in much more rapid settling
of particles in the fluid than what gravity could produce. This is an important point
because fast particle settling leads to fast separation, which leads to short processing
times as we will discuss shortly. (Note that the terms settling and sedimentation are
used interchangeably in this book.)

(a) (b)

Increasing time

(c) RCF for the Eppendorf Centrifuge 5418 shown above:

RCF ≡ ω2 × R/g
ω = 14,000 rev/min × 1 min/60 s × 2π radians/rev = 1466/s
R = 7.7 cm (radius from center of axis of rotation to bottom of centrifuge tube)
RCF = (1466/s)2 × 7.7 cm/(980 cm/s2) = 16,888

Figure 6.1: A bench-top centrifuge. (a) shows a photo of the Eppendorf 5418, an example of a bench-
top centrifuge. (b) is a diagram showing the movement of solids in a spinning centrifuge tube with
respect to time. (c) is the calculation of RCF for the Eppendorf 5418. Note that samples in the centrifuge
rotor are at an angle of 45° from vertical. Images © NC State University; reprinted with permission.

6.3 Solid-liquid separation by gravity

Despite the high RCF values generated by the centrifuge shown in Figure 6.1(a), it
would clearly be difficult to scale up this system for use in a large-scale manufactur-
ing process. To better understand how centrifugation is implemented at larger scale,
we first consider a settling tank that relies on gravity to separate solids from liquid.
Gravity separation of solids may seem like a digression from our main topic – centri-
fugation – but this discussion sets the foundation for production-scale centrifugation.
 6.3 Solid-liquid separation by gravity 155

An example of an equipment setup for solid-liquid separation that relies on gravity

for particle sedimentation is shown in Figure 6.2(a). Once particles settle to the bot-
tom of the vessel, liquid can be pumped out and solids then removed. The time that it
takes for the settling process to be completed, tsettling, is equal to the time required for
a particle that starts at the top of the vessel to settle to the bottom and is given by the
following equation:

h
tsettling ¼ (6:2)
vg

where h = height of the liquid in the vessel and vg = settling velocity, which is sim-
ply the velocity at which the particle falls through the fluid.

(a) (b)
Fd = 3πμdpvg Fluid drag force (Stokes’ Law)

t=0 Fb = mlg Buoyant force

t > tsettling Fg = mpg Force due to gravity

(c) A force balance on a particle settling where:

under the force of gravity gives: mp = particle mass
gravity = buoyancy + drag ml = mass of liquid displaced

Fg = Fb + Fd g = acceleration due to gravity

μ = liquid viscosity
mpg = mlg + 3πμdpvg
dp = particle diameter
4πdp3ρpg = 4πdp3ρlg + 3πμdpvg ρp = particle density
3 8 3 8
ρl = liquid density
Solving for the settling velocity, vg:
vg = dp2(ρp − ρl)g
18μ

Figure 6.2: Particles settling in a vessel under the force of gravity. (a) shows solid-liquid separation
by gravity in a batch settling vessel. (b) shows forces acting on a single particle settling in a fluid
under the force of gravity. (c) is a derivation of the settling velocity for a particle from a force
balance. Image © NC State University; reprinted with permission.
156 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

When a particle moves through a fluid under the force of gravity, it quickly comes to a
constant velocity, referred to as a settling velocity, sedimentation velocity, or terminal
velocity. At steady-state (i.e., constant settling velocity), gravity acting downward on a
particle is balanced by the buoyant force and the Stokes’ drag force acting upward on
the particle, shown in Figure 6.2(b). Equating the downward acting force of gravity to
the upward acting drag and buoyancy forces results in the following expression for vg:
 
d2p ρp  ρl g
vg ¼ (6:3)
18μ

where dp = particle diameter, ρp = particle density, ρl = liquid density, g = accelera-

tion due to gravity, and µ = viscosity of the liquid through which the particle is set-
tling. The derivation is shown in Figure 6.2(c). From equation (6.3), it is clear that a
difference in density between the particle and fluid must exist for particle settling
and that the larger the difference, the faster the settling velocity. Equation (6.3) also
shows that large particles settle faster than small, and that particle settling becomes
slower as fluid viscosity increases. The latter relationship, in particular, should be
clear as a highly viscous fluid like motor oil impedes the movement of a particle to a
much greater extent than a less viscous fluid like water. Note equation (6.3) also in-
dicates that if the particle density is less than that of the liquid, the density differ-
ence is negative and the particle moves opposite the direction that gravity acts. This
is what happens when an ice cube settles at the surface of water in a glass rather
than sinking to the bottom of the glass.
We expand this analysis by considering gravity sedimentation with the continuous
flow unit in Figure 6.3. This analysis, adapted from a publication [156] from Alfa Laval,
a leading centrifuge manufacturer, provides insight into the principles underlying pro-
duction centrifuges, and, in particular, the disc-stack centrifuge that is commonly used
in the biopharmaceutical industry [157]. The illustration shows a slurry, such as a fer-
mentation or cell culture broth, being fed to a vessel at a volumetric flow rate, Q. We
assume that all particles in the feed are exactly the same size and that the inlet and
outlet flow rates are the same so that the liquid volume in the tank does not change
with time. The goal is to produce an effluent that is free of particles, which requires
that all the solid particles that are introduced at the inlet have ample time to settle to
the bottom of the tank before being forced into the exiting effluent stream. From a proc-
essing perspective, it is worth considering the maximum volumetric flow rate, Qmax,
that will allow for a particle-free outlet stream, noting that an inlet flow rate that is too
high will result in a liquid residence time within the unit that is too short to allow for
complete sedimentation of particles (i.e., the particles will flow out of the tank with the
liquid). Incomplete sedimentation would result in an unclarified effluent.
To derive an expression for Qmax, both the particle settling time and the resi-
dence time of the fluid in the tank must be considered. The time required for a parti-
cle to settle to the bottom of the unit is given by equation (6.2), just as it was for the
 6.3 Solid-liquid separation by gravity 157

Inlet volumetric

Liquid volume in tank Vliquid

Outlet volumetric

Figure 6.3: Continuous separation of solids from liquid using a gravity settling tank. Image courtesy
of Alfa Laval and adapted with permission.

batch settling vessel shown in Figure 6.2(a). The residence time of the liquid in the
tank, tresidence,liquid, is the liquid volume in the tank, Vliquid, divided by the volumetric
flow rate, Q:

Vliquid lwh
tresidence;liquid ¼ ¼ (6:4)
Q Q

where l, w, and h are the tank length, width, and liquid height, respectively. To produce
a particle-free effluent, the fluid must reside in the tank at least as long as is required for
particles to settle (i.e., tresidence,liquid ≥ tsettling); otherwise, particles will be swept out of
the vessel with the fluid to the tank outlet. The maximum inlet volumetric flow rate,
Qmax, that allows for a particle-free outlet stream corresponds to the scenario in which
tresidence,liquid time, given by equation (6.4), is equal to tsettling, given by equation (6.2):

lwh h
¼ ; (6:5)
Qmax vg

or
Qmax ¼ vg lw ¼ vg Asettling (6:6)

where Asettling (=l × w) is the area for settling. Equation (6.6) shows that Qmax de-
pends only on settling velocity and settling area. Interestingly, the height of the liq-
uid does not impact Qmax. This results because even though an increase in height
would increase the time required for particle settling, the residence time of the liq-
uid for a fixed Qmax likewise increases and offsets the longer settling time.
We have established that the maximum volumetric flow rate for a gravity settling
tank that produces a particle-free effluent is proportional to the settling velocity of
the particle and the area available for settling. However, as mentioned previously,
158 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

settling time under the force of gravity is relatively long because vg is small. How
small? Table 6.2 shows settling velocity values for E. coli, Saccharomyces cerevisiae,
and CHO cells along with the parameters used to calculate the values according to
equation (6.3). As you can see from the vg values presented, cells settle slowly under
the force of gravity; for example, E. coli settles only slightly faster than 0.10 cm/h! As
a consequence, solid-liquid separation by gravity is, simply put, slow and not often
used in biopharmaceutical processes. In general, the time to complete a process step,
tprocess, is the ratio of the total volume fed to a process step, Vfeed, and the volumetric
flow rate, Q. Applying this to the settling tank in Figure 6.3 leads to:

Vfeed Vfeed
tprocess ¼ ¼ (6:7)
Q vg Asettling

Clearly, a small particle settling velocity results in a long (and undesirable) process
time. As an example, complete settling of E. coli cells from a 2,000 L fermentation
under gravity in a vessel with 100 ft2 (= 92,903 cm2) of settling area would require
135 h (= 2,000,000 cm3/(0.16 cm/h × 92,903 cm2). So, small settling velocities trans-
late to undesirably long process times.

Table 6.2: Gravitational settling velocities calculated for cell types commonly used in
biopharmaceutical processing.

Cell Type Diameter (µm) Density (g/cm) [] Settling Velocity (cm/h)a

E. coli  ×  [] . .

Saccharomyces cerevisiae – [] . .

CHO  [] . .

a
The liquid density and viscosity are assumed to be that of water – 1 g/mL and 1 cp, respectively.
We have also used the largest length dimension shown.

6.4 Production centrifuges

Equation (6.7) suggests that to reduce the processing time required by the unit in
Figure 6.3 for a given Vfeed, the volumetric flow rate Q should be increased. As equa-
tion (6.6) suggests, increasing settling area, settling velocity, or both will increase
the maximum value for Q that will still allow for a clarified effluent from the tank.
An increase in settling area can be accomplished by placing plates or discs into the
settling tank. To increase settling velocity, equation (6.3) suggests that increasing
the acceleration may be most effective because the other equation variables relate
to particle and fluid properties that may not be adjustable in a process (e.g., their
values are dependent on the cells used and bioreactor conditions). If we replace
 6.4 Production centrifuges 159

gravitational acceleration, g, in equation (6.3) with centrifugal acceleration, ω2R,

the particle settling velocity can be expressed as follows:
 
d2p ρp  ρl ω2 R
vc ¼ ¼ vg × RCF (6:8)
18μ

We showed previously that RCF values for a centrifuge are in the thousands; thus,
vc ≫ vg and much faster processing times result when separation is performed by
centrifugation. This discussion points to the rationale for the design of the com-
monly used disc-stack centrifuge, which is the focus of much of this section.

6.4.1 Disc-stack centrifugation: an introduction

Imagine modifying the equipment for gravity separation in Figure 6.3 by adding set-
tling plates, flipping the tank vertically, and rotating the resulting unit. The result
would resemble a disc-stack centrifuge, as shown in Figure 6.4.

Settling area: Small Settling area: Large Settling area: Large

Relative force: 1g acting downward Relative force: 1g acting downward Relative force:
> 5000g acting outward
from center of rotation

Figure 6.4: Improvements to a continuous gravity settler lead to a disc-stack centrifuge. The disc-
stack centrifuge improves upon the gravity tank by increasing both the settling area and the force
acting on the particles (i.e., settling velocity). Image courtesy of Alfa Laval and adapted with
permission.

Figure 6.5 shows the bowl and discs (inside the bowl) of a disc-stack centrifuge and
flow of solid and liquid through the discs. The spacing between discs is small to
create a short distance for particle settling. The feed – a slurry such as fermentation
or cell culture broth – enters continuously at the top of the centrifuge through a sta-
tionary pipe and moves through the centrifuge bowl as it spins. The feed flow is
generated by pressurizing the slurry feed vessel, in which case the flow rate is con-
trolled by a valve in the feed line, or by pumping the slurry into the centrifuge inlet.
Feed flows downward then upward in the space between the discs. As the centrifuge
160 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

rotates, particles move outward, settle on the underside of the discs, and continue to
move to the bowl wall where they ultimately settle. The clarified liquid stream, also
referred to as the centrate or supernatant, exits near the top of the centrifuge, close to
where the feed enters.

Feed slurry

liquid liquid Liquid 5000 g

Solids Solids ~1.0 mm

Solids
Motor drive to The disc-stack Flow between
rotate bowl the discs

Figure 6.5: Schematic of the bowl (and flow through the bowl) of a disc-stack centrifuge. Image
courtesy of Alfa Laval and adapted with permission.

As the run proceeds, solids from the bowl must be removed to avoid clogging the
bowl and to keep them out of the clarified liquid. Solids from disc-stack centrifuges
can be removed manually, by disassembling the centrifuge and scraping solids out of
the bowl, or automatically. Automatic removal may be intermittent or continuous. In-
termittent removal is common and involves ejecting the accumulated solids from the
centrifuge bowl at regular, pre-determined intervals. The ejected solids from either
intermittent or continuous removal are referred to as the solids, or perhaps more ap-
propriately as the concentrate given that some liquid is ejected along with the solids.
Note that either intermittent or continuous removal of solids provides for a continu-
ous solid-liquid separation method unlike the bench-top centrifuge shown in Figure
6.1(a), which operates in batch mode. Note also that either the clarified liquid stream
or the solids stream is readily collected; therefore, the biopharmaceutical product
can be recovered regardless of whether it resides in the clarified liquid or solids
stream. Figure 6.6 shows a disc-stack centrifuge at its different stages of operation.

6.4.2 Disc-stack centrifuges: equipment details

Disc-stack centrifuges are mechanically complex, and covering the details of their de-
sign is beyond the scope of this book. However, we do want to highlight some features
that will likely be of interest and directly impact processing considerations. The basis
of this discussion is the Alfa Laval LAPX 404 SGP centrifuge, shown in Figure 6.7.
 6.4 Production centrifuges 161

Figure 6.6: Operation of a disc-stack centrifuge with intermittent discharge. The bowl shown in these
images discharges radially. Intermittent axially discharging bowls also existing in which solids flow
downward (axially) before being discharged. Image © NC State University; reprinted with permission.

Control panel
Centrifuge bowl hood.
Under the hood is the
bowl, which contains
the discs.
Concentrate
tank

Centrifuge
motor

Process lines

liquid, concen-
trate, etc.)

Concentrate
pump

Figure 6.7: An Alfa Laval LAPX 404 SGP/TGP centrifuge. Image © NC State University; reprinted with
permission.
162 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

Figure 6.8: A piping and instrumentation diagram (P&ID) for the Alfa Laval LAPX 404 unit. Image
courtesy of Alfa Laval and adapted with permission.
 6.4 Production centrifuges 163

Figure 6.8 shows three highlighted system lines that are central to the execution of
a disc-stack centrifugation step: process liquid (i.e., feed) line 201, clarified liquid
line 220, and concentrate line 222. A description of these lines, as well as other lines
and important components of the centrifuge, follows:
– The centrifuge bowl. The volume of the bowl for the LAPX 404 centrifuge shown
in Figure 6.7 is 2.2 L. Bowl rotation is driven by an electric motor via a belt and
bowl spindle. The unit in Figure 6.7 rotates at a maximum speed of 9,600 rpm.
The centrifuge is also equipped with a brake that can be used to stop bowl rota-
tion once the power is off.
– Process liquid (feed) line 201. Feed slurry to be separated is fed through line 201.
The temperature (TT201-1) and flow rate (FIT201-4) are measured for the incom-
ing feed. The flow control valve FCV201-2 is a diaphragm valve that controls the
flow rate of the feed contained in a pressurized vessel.
– Clarified liquid line 220. The resulting clarified stream flows through line 220,
where turbidity (QIT220-4), pressure (PT220-1), and temperature (TT220-1) are
measured. Pressure control valve PCV220-3 is in place to control back pressure
to the centrifuge.
– Concentrate line 222. Solids are discharged through ports in the bowl wall. For the
centrifuge in Figure 6.7, these ports are covered between discharges by a sliding
bowl bottom. Opening and closing of the bowl is controlled hydraulically, with
line 370 feeding the necessary water. “Opening” water fed through the 372 branch
of line 370 lowers the sliding bowl bottom, thereby exposing ports at the bowl
wall. “Closing” water is supplied via the 376 branch of line 370 to the built-in op-
erating water tank and pumped to the operating system through the bowl spindle.
The closing water pushes the sliding bottom up, thereby closing the ports at the
bowl wall. Opening and closing of the bowl only take a fraction of second. The
concentrate line 222 includes a small vessel for collecting solids and an air-oper-
ated diaphragm pump (P222-1) to move the solids through the concentrate line ei-
ther for collection or disposal to waste.
– Flushing liquid line 301. This line allows cleaning agent to be sent to the bowl
exterior and to the solids collection vessel, as shown in Figure 6.8. More detail
on cleaning a disc-stack centrifuge is provided later in this chapter.
– Buffer liquid line 301a. This line can be used to feed a buffer (or other liquid) to
the bowl if needed. This function is particularly useful when product is in the
liquid phase, as is the case with monoclonal antibodies produced using CHO
cells that secrete the antibody. During bowl discharge, liquid, in addition to
solid, may be ejected resulting in a yield loss for extracellular products since the
discharge stream flowing through concentrate line 222 is typically directed to
waste. To avoid product loss, the bowl can be flushed with a buffer through line
301a just before solids discharged – referred to as a predischarge flush – to push
product-containing liquid within the bowl out of the system through line 220,
thus enabling product recovery. Note that the bowl continues to spin during this
164 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

step, so solids remain within the bowl while buffer flushes out held-up liquid.
This step should be designed to minimize any disruption to solids in the bowl.
– Cooling water line 401 is used to feed water into the hood surrounding the bowl
to reduce heating of the product as the centrifuge bowl rotates.

6.4.3 Other types of production centrifuges

While the disc-stack is the most commonly used type of centrifuge for biopharmaceuti-
cal applications such as cell recovery and cell removal, other types are available. For
example, in biopharmaceutical processing, a tubular-bowl centrifuge is also common.
Decanter and multi-chamber bowl centrifuges are also commonly used industrially, al-
though less commonly for biopharmaceutical applications. A number of resources pro-
vide detailed coverage of the different types of centrifuges [159–161]. Here, we present
only a summary of these and relevant characteristics of each in Table 6.3.
In addition to the centrifuges listed in Table 6.3, there are also single-use centri-
fuges currently in use. These units offer many of the advantages described previously
in Chapter 2 for single-use equipment, including elimination of cleaning and sanitiza-
tion steps, which reduces the costs of water, cleaning agents, and cleaning validation.
Examples of single-use units include the Unifuge® from PneumaticScaleAngelus®
and its scaled up version, U2k®, both single-use tubular devices. Feed flows through
the center of the device. Cells settle to the periphery of the bowl and clarified liquid
(supernatant) continuously flows. Once the bowl is filled with solids, feed is stopped,
cell slurry is pumped out, and the process can start again. The Unifuge® operates at
flow rates of 1–4 L/min and at RCF values of 300–3,600. Its recommended applica-
tions include both cell recovery and slurry clarification, particularly in mammalian
and insect cell applications [164]. The larger U2k® system operates at up to 20 L/min
and RCF values up to 3,000. Based on the information provided by the supplier,
both clarified liquid and collected cells continuously flow from the unit during op-
eration [165].

6.5 Disc-stack centrifuge cGMP operating procedure

A typical operating procedure for an automated disc-stack centrifuge with intermittent

discharge, such as the one shown in the photo in Figure 6.7, is outlined below:
– Set up the unit;
– Measure the % solids in the feed and calculate the feed interval;
– Input parameters (e.g., feed flow rate, feed interval) to the centrifuge controller;
– Begin feeding;
– Allow multiple feed/discharge cycles to execute until feed is completely processed;
Table 6.3: Commonly used centrifuges and important characteristics of each. Note that the sigma (Σ) values shown in the table are discussed shortly, in
Section 6.7.

Centrifuge Type Illustrationa,b Feed Rate (L/ Typical Solids Content in Σ (m) [] Other Characteristics [, ]
h) [] Feed (% by Volume) []

Disc-stack –, ≤ ,–, – RCF up to 15,000 × g

– Continuous operation (i.e., solids discharge
possible)
– Bowl cooling
– Solids are wet, paste contains liquid
– Difficult to clean
Tubular –, < ,–, – RCF up to 20,000 × g
– Firm paste (i.e., good removal of liquid)
– Easy to clean
– Limited solids capacity
– Recovery of solids typically manual and difficult
Decanter –, – –, – RCF up to 10,000 × g
– Continuous solids discharge
– Handles high feed solids concentration
– Low RCF
Multi-chamber –, < – – RCF up to 9,000 × g
Bowl – Firm paste (i.e., good removal of liquid)
– Large solids holding capacity
– Bowl cooling
– Recovery of solids typically manual and difficult
6.5 Disc-stack centrifuge cGMP operating procedure

– Difficult to clean

a
Arrows indicate path of liquid phase; dashed lines indicate solids accumulation.
b
Images adapted from Fermentation and Enzyme Technology, p. 263, by D. I. C. Wang et al., New York: John Wiley & Sons, Inc. Copyright (1979) by John
Wiley and Sons, Inc. [163].
165
166 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

– Upon completion of feeding, perform system flush with high purity water (HPW)
or a buffer. Note this flush may be built into the clean-in-place (CIP) cycle;
– Clean and sanitize the machine;
– Perform final HPW rinse to remove cleaning/sanitizing agents. Note that this
rinse may be built into the CIP cycle;
– Store system dry.

As we discussed in Chapter 3, current Good Manufacturing Practice (cGMP) production

of biopharmaceuticals requires that equipment cleaning be validated. Centrifuge clean-
ing and sanitization are often carried out using a CIP skid programmed to run an auto-
mated CIP cycle. Table 6.4 provides details of a CIP cycle that has been validated for
the disc-stack centrifuge shown in Figure 6.7. As shown, the cycle comprises multiple
phases, each with a distinct objective. To execute the CIP cycle, connections are made
from the CIP skid, via a CIP station, to three inlets on the centrifuge. In the P&ID in
Figure 6.8, they are the process liquid (i.e., feed) line 201, the buffer liquid line 301A,
and the flushing liquid line 301. These connections enable all product-contact surfaces
to be cleaned, and the cleaning fluids are cycled through each line throughout the CIP
cycle. Spent cleaning fluid is returned to the automated CIP skid through the clarified
liquid line 220 and the concentrate (i.e., solids) line 222. In addition, during CIP, the
bowl is discharged at two-minute intervals to ensure adequate cleaning of the interior
of the bowl. It is also worth noting that immediately after use, prior to execution of the
automated CIP cycle, a “manual” rinse of the centrifuge is performed with 100 L of
HPW. HPW is fed only through line 201 at flow rate of 500 L/h, and the bowl is dis-
charged at two-minute intervals. The manual rinse is in place to enable extended time
between processing and running the automated CIP cycle. This time is needed because
the centrifuge step is often not completed until the end of a shift, making it necessary
to delay execution of the automated CIP cycle until the next shift arrives more than 12 h
later. (It is desirable to have staff present as the automated cycle runs.) The option of
holding the centrifuge dirty for more than 12 h presented challenges to validating clean-
ing of the system. Consequently, the manual HPW rinse immediately after centrifuge
use was implemented to allow for validation of centrifuge cleaning with a 12-plus hour
hold time after rinsing.

6.6 Disc-stack centrifuge process and performance parameters

To begin the discussion of parameters that impact performance of a disc-stack centri-

fuge, let’s write an equation similar to 6.6 for a centrifuge; that is, an equation that
relates the maximum volumetric flow rate that allows adequate removal of solids,
Qmax, to properties of the particle, liquid, and centrifuge geometry. It is important to
be able to estimate Q, because process time depends on it. For centrifuges, these
equations were first presented by Hebb and Smith [166], then Ambler [167, 168]. We
 6.6 Disc-stack centrifuge process and performance parameters 167

Table 6.4: CIP cycle details for the Alfa Laval LAPX 404 unit shown in Figure 6.7.

Phase Cleaning Fluid Details Purpose

Initial Rinse HPW – Single pass a

Remove free-rinsing soils from
– 24 min equipment and piping surfaces. Cold
– 7.5–8.0 L/min or ambient water is desirable to avoid
– 30 °C denaturing any proteinaceous soils.

Alkaline Wash CIP ®, KOH- – ~3% CIP 100® Remove organic (e.g., proteinaceous)
based detergent – Recirculated soils that are not easily rinsed.
– 120 min
– 7.5–8.0 L/min
– 75 °C
Post-wash HPW – Single pass Remove alkaline wash chemicals.
Rinse – 25 min
– 7.5–8.0 L/min
– 30 °C
Acid Washb CIP ®, – ~3% CIP 200® Neutralize traces of alkali residual on
phosphoric acid- – Recirculated equipment surfaces; remove inorganic
based detergent – 120 min soils.
– 7.5–8.0 L/min
– 75 °C
Final Rinse HPW – Single Pass Remove all cleaning agents.
– 60 min, >0.9 µS/cm
– 7.5–8.0 L/min
– 30 °C
a
Single pass means cleaning fluid only flows through the centrifuge once and is then sent to drain.
Recirculated means that cleaning fluid flows from CIP skid to the centrifuge and back to the CIP
skid continuously throughout the phase.
b
CIP was validated without an acid wash step, so acid wash is optional.

do not derive these equations here, as the derivations are available in the previously
cited references and in textbooks [169]. For a disc-stack centrifuge, the applicable
equation is:
2  3
 
d2p ρp  ρl 2Ndiscs πω2 R3  R3 cot θ
Qmax ¼ 4 5 0 1
(6:9)
18μ 3

where the variables in the first bracket have been defined previously in equation
(6.3), and the remaining variables are defined as follows: Ndiscs is the number of
discs, ω is the angular velocity, R0 and R1 are the radial distances from the center of
rotation to the outer and inner radii of the discs, and ϴ is the angle the disc makes
with the bowl axis.
168 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

Equation (6.9) suggests the following about properties of the feed to the centri-
fuge and their impact on disc-stack centrifuge performance:
– Increasing the particle diameter to be separated increases the maximum volu-
metric flow rate that can be used.
– Increasing the density difference between the fluid and particle and/or decreasing
the fluid viscosity allows for larger volumetric flow rates through the centrifuge;
alternatively, for a fixed flow rate, a larger density difference allows separation of
smaller particles.

When recovering or removing cells from fermentation or cell culture broth, making
adjustments to the particle size, particle density, liquid density, and liquid viscosity
is unlikely. Those parameters are dictated by the fermentation or cell culture step.
We can also use equation (6.9) to understand the effect of input parameters that
are readily controllable (i.e., process parameters). Ranges or set points for these param-
eters should be determined in development studies and provided in a batch record.
– Feed flow rate/residence time in the bowl. Generally speaking, increasing resi-
dence time of the slurry in the bowl results in better separation. Consider equation
(6.9) from a different perspective. If that equation is solved for dp, it becomes evi-
dent that dp is proportional to Qmax1/2. This relationship means that as Qmax is de-
creased, thereby increasing residence time in the centrifuge bowl, even smaller
particles may be separated (i.e., clarity of the clarified liquid, a performance pa-
rameter, may be improved). It is worth noting that disc-stack centrifuges typically
cannot effectively remove particulates of <1 μm in diameter [157]. To achieve com-
plete removal of solids, clarified liquid from the centrifuge typically undergoes
further processing using filtration – the topic of the next chapter – to remove re-
sidual particulate before subsequent downstream processing.
– Rotational speed. From equation (6.9), the maximum volumetric flow rate that al-
lows for separation of particles of diameter dp is proportional to the angular veloc-
ity (ω) squared. Therefore, for a given feed, higher rotational speeds enable larger
values of Q and shorter processing times; alternatively, for a fixed value of Q, a
higher rotational speed enables smaller particles to be separated. Note that high
rotational speeds create shear. Shear forces created by the liquid act along the sur-
face of the cell and may cause cell damage. Shear is particularly a problem at the
centrifuge inlet. Cell damage is undesirable in intracellular products because it
means that product ends up in the clarified liquid waste. In the case of extracellu-
lar products, cell damage results in undesirable soluble impurities being released
into the clarified liquid product stream. Disc-stack centrifuges for use in biophar-
maceutical processes are designed to minimize the effects of shear [159].
– Particle concentration in the feed. The settling velocity equations (6.3) and (6.8) pre-
sented previously were derived under the assumption that particles settle freely
through a fluid. As the particle concentration increases, however, particles becomes
obstructed by other particles, which lowers the settling velocity and results in
 6.6 Disc-stack centrifuge process and performance parameters 169

difficult separation of solids from liquid (i.e., more solids end up in the clarified liq-
uid and step yield is reduced for an intracellular product). As a result lower-than-
expected feed flow rates would have to be used to achieve the desired separation.
To avoid the effects of so-called hindered settling, feed to the centrifuge can be di-
luted to reduce particle concentration. Of course, diluting the centrifuge feed
leads to greater feed volume, which leads to a longer processing time. Additional
information on hindered settling can be found elsewhere [170].
– Feed interval/discharge frequency. Time between discharges in a disc-stack cen-
trifuge with intermittent discharge is referred to as the feed interval or discharge
interval and designated tfeed here. This value may have to be entered into the
centrifugation system prior to a run and can be estimated based on the following
calculation:

bowl capacity for solids Vbowl Fbowl

tfeed ¼ ¼ : (6:10)
Qsolids Csolids;f =100 × Q

where Qsolids is the volumetric flow rate of solids to the centrifuge, Vbowl is the vol-
ume of the centrifuge bowl, Fbowl is the fraction of bowl volume filled with solids
before discharging, Csolids,f is the solids content of the feed in volume percent, and
Q is the total volumetric feed rate (solids + liquid) to the centrifuge. It is common
to allow the bowl to fill to 50%–70% solids before discharging; that is Fbowl is
0.50–0.70 [171]. As you might imagine, if the feed interval is too long, solids overly
accumulate in the bowl and are carried into the clarified liquid stream. Solids in
the clarified liquid translate to a reduced step yield for intracellular products and
reduced clarification efficiency for extracellular products. Further, if the feed inter-
val is too short (i.e., the frequency of discharge too high), then for extracellular
products, more liquid (i.e., product) will be lost with the discharged solids, which
results in a decreased step yield.

Example: Calculating the feed interval for recovery of cells from an E. coli fermentation using a
disc-stack centrifuge
Cells are recovered from a 500 L E. coli fermentation with a disc-stack centrifuge. The fermenta-
tion contains 6% cells by volume. The flow rate to the centrifuge is 120 L/h, and the centrifuge
bowl has a volume of 2.2 L. The centrifuge requires that a feed interval be entered in units of
seconds. What value do you enter?

Solution
Calculate the feed interval by substituting the given values into equation (6.10) as follows:

2:2 L × 0:5
tfeed ¼ × 3600 s=h
6=100 × 120 L=h
¼ 550 s ðor 9:2 minÞ
170 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

– Discharge Ratio. Bowl discharge in a disc-stack centrifuge can be full or partial. A

full discharge empties most or all bowl contents – both solids and liquid. A partial
discharge, in which the bowl is open for less time, allows for only a fraction of the
bowl contents to be discharged, thereby minimizing loss of liquid to the solids
stream. When product is dissolved in liquid, partial discharges are particularly
useful for minimizing the amount of product lost with the solids waste (i.e., maxi-
mizing step yield). The discharge ratio is the ratio of partial to full discharges. By
setting a discharge ratio, buildup of solids in the bowl is not as excessive as if par-
tial discharges alone are used, and loss of liquid is less than if only full discharges
were utilized. The discharge strategy should be optimized during process develop-
ment and the discharge ratio (if used) specified in the batch record.

6.7 Sigma factor, centrifugation step design and scale-up

Central to centrifuge scale-up is sigma analysis, which uses a Σ value to characterize

centrifuge performance. The Σ value is arrived at as follows. If we multiply the first
part of the equation (6.9) in brackets by g, the gravitational acceleration, and divide
the second part in brackets by g, the following equation results.
2   3
 
d2p ρp  ρl g 2Ndiscs πω2 R3  R3 cot θ
Qmax ¼ 4 5 0 1
(6:11)
18μ 3g

The first bracketed expression is simply the settling velocity due to gravity. The second
bracketed expression is referred to as sigma, Σ, and the equation can be written more
simply as:

Qmax ¼ vg Σ (6:12)

where Σ = 2Ndiscsπω2(R03 – R13)cotϴ/3 g, for a disc-stack centrifuge. As you can see, vg

depends on particle and fluid properties, while Σ depends mostly on geometric varia-
bles for the centrifuge. Also, despite the fact that equation (6.12) applies to a centri-
fuge, vg instead of vc shows up, due to the fact that we intentionally multiplied and
divided the right side of equation (6.9) by g. Importantly, note the similarity between
equations (6.6) and (6.12), which is written for a gravity settling tank. The similarity
between the two equations results in the following interpretation for Σ: the area in a
gravity settling device, such as that shown in Figure 6.3, that would be required to
give the same performance as a centrifuge at equal volumetric flow rates.
To put Σ in perspective, consider the disc-stack centrifuge in Figure 6.7, which has
a Σ value of 5,230 m2. To match the performance of this centrifuge, a gravity settling
tank would need 5,230 m2 of surface area. For comparison, the area of an American
football field is approximately 5,350 m2. Disc-stack centrifuges are available in sizes
 6.7 Sigma factor, centrifugation step design and scale-up 171

much larger than the one shown in Figure 6.7. Machines with Σ values in excess of
100,000 m2 that handle flow rates in excess of 100,000 L/h are in use [159]. Also, it is
important to point out that the mathematical expression for Σ depends on the centri-
fuge type (i.e., the expression for Σ for a disc-stack centrifuge is different than that for a
tubular centrifuge).
Equation (6.12) can be used for scale-up as follows. Because vg is only a function
of particle properties, fluid properties, and the constant g, which are all scale inde-
pendent, the following is true for centrifuges at any two scales, designated 1 and 2
in the equations below:

vg;1 ¼ vg;2 (6:13)

Therefore

Qmax;1 Qmax;2
¼ : (6:14)
Σ1 Σ2

Equation (6.14) can be used as the basis for centrifuge scale-up as illustrated in
Figure 6.9, which shows determination of Q1/ Σ1 as a first step. This can be done experi-
mentally by using a bench-scale centrifuge of the same type to be used at production
scale. Once that value is determined, there are two scenarios for scale-up depending on
whether an existing production centrifuge is available. If there is, the flow rate to the
production centrifuge is calculated by solving equation (6.14) for Q2 and plugging in
known values for Q1, Σ1, and Σ2. Note that the centrifuge manufacturer should be able
to provide a value for Σ2. Alternatively, if a new centrifuge is to be purchased, Σ2 is un-
known. A value for Q2 will have to be determined based on an acceptable processing
time, and from that value, Σ2 is estimated using equation (6.14). An example calculation
is shown below.

Example: Scaling up a disc-stack centrifugation step from bench-scale data

You are purchasing a disc-stack centrifuge for recovery of cells from an E.coli fermentation used
to produce an intracellular therapeutic protein. A study using a bench-top disc-stack centrifuge
(Σ = 1,900 m2) suggests a flow rate of 1,200 mL/min is suitable to meet your centrifugation ob-
jectives. This value was determined by varying the flow to the bench-top unit and selecting a
value that resulted in no more than 2% loss of cell mass. You don’t currently have a disc-stack
centrifuge to use at manufacturing scale. Answer the following:
a. What flow rate will you operate at to process a 5,000 L fermentation?
b. What size centrifuge (Σ value) will you purchase?

Solutions
a. To determine a feed flow rate, assume an acceptable processing time and then use equa-
tion (6.7) to solve for Q, the volumetric feed rate to the centrifuge. Let’s assume a process-
ing time of 3 h, which does not include centrifuge setup and cleaning; these steps will
lengthen the total time to complete the centrifugation step to approximately 8 h.
172 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

Vfeed 5000 L
Q¼ ¼
tprocess 3h
¼ 1667 L=h

b. Now that Q is known, equation (6.14), solved for Σ2, can be used to determine the size of
production centrifuge to carry out the step:

Q2 Σ1 1667 L=h
Σ2 ¼ ¼ × 1900 m2
Q1 72 L=h

¼ 43; 990 m2

Determine Q1/Σ1 for small

centrifuge (e.g. Whisperfuge).
Σ1 can be provided by vendor.

Use
Yes an existing No
centrifuge?

Calculate Q2 to meet process

Calculate Q2 for a
needs, then calculate Σ2 for given
given Σ2. Use Q2 as
Q2. Ask centrifuge vendor to
provide a centrifuge with Σ = Σ2.

Figure 6.9: Using equation (6.14) for centrifuge scale-up. Image © NC State University; reprinted with
permission.

It is worth noting that the ratio of settling velocity in a centrifugal field to that in a
gravitational field has been found to fit the following equation for a disc-stack cen-
trifuge [159]:
 2 0:75
vc ωR
¼ (6:15)
vg g

Based on equations (6.3) and (6.8), we would expect the exponent to be 1 rather
than 0.75. However, the relationship in equation (6.15) results from increased shear
forces in a disc-stack centrifuge that act to drag particles to the clarified liquid out-
let, ultimately resulting in a reduced particle settling velocity in the centrifuge [159].
As a result of the reduced settling velocity, which is not accounted for in the
 6.8 Summary 173

derivation of Σ for a disc-stack centrifuge, the KQ value was developed as an empiri-

cal alternative to characterize disc-stack centrifuge size [172] and is given by:
 rpm 1:5  
KQ ¼ 280 Ndiscs cot θ R2:75
0  R1
2:75
(6:16)
1000

6.8 Summary

There are numerous solid-liquid separation applications in biopharmaceutical processes.

These include:
– removal of cells and cell debris from the cell culture broth, leaving a product
that is dissolved in the spent clarified liquid media;
– recovery of cells from the bioreactor for further processing;
– clarification of lysate streams;
– recovery of inclusion bodies in processes in which the host system (e.g., E. coli)
produces the product as an inclusion body;
– recovery of precipitated product or removal of precipitated impurities in pro-
cesses that use precipitation for purification;
– removal of microorganisms to produce a bioburden-free or sterile process stream.

The three most common methods for solid-liquid separation in biopharmaceutical

processes are centrifugation, covered in this chapter, depth filtration, and tangential-
flow microfiltration, both of which are covered in the next chapter. Centrifugation re-
lies on a density difference between solids and the surrounding liquid combined with
centrifugal force to cause suspended solids to settle rapidly in their surrounding liquid.
Once settled, the clarified liquid and solids can be recovered separately. The cen-
trifugal force used in centrifuges for biopharmaceutical applications is typically
around 5,000–10,000 × g; thus, particles settle much faster than by gravity, which
results in relatively short process times for the separation.
There are numerous types of centrifuges in use industrially. For biopharmaceu-
tical applications, the disc-stack centrifuge is the most common. Discs stacked in a
bowl provide surface area for settling, while rotation of the bowl provides the neces-
sary centrifugal force. The feed slurry flows continuously to the centrifuge and into
the rotating bowl, while a clarified liquid flows out. Accumulated solids are dis-
charged from the bowl either intermittently at predetermined intervals or continu-
ously, depending on the centrifuge design.
There are a number of process parameters for which ranges must be determined
and which are typically written into a batch production record and/or built into the
process automation. These include feed flow rate, rotational speed of the bowl, parti-
cle concentration in the feed, and the feed interval in the case of a centrifuge with
intermittent discharge. The procedure for a disc-stack centrifugation run includes
174 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

setup steps that precede the actual separation followed by multiple feed-discharge
cycles (for centrifuges with intermittent discharge). Once the entire feed slurry has
been processed, the system can be flushed with a buffer for product recovery (if the
product is in the clarified liquid phase), then cleaned and sanitized. For cGMP pro-
duction of biopharmaceuticals, cleaning/sanitization is performed using a validated
method, often with an automated CIP system.
To properly scale up results from bench-scale development studies to larger
scale, sigma analysis is used. The relevant equation is Qmax,1/ Σ1 = Qmax,2/ Σ2, where
Q is the volumetric flow rate at two different scales designated 1 and 2. While Σ is
represented by a relatively complex equation (see the second bracketed expression
in equation (6.11)), its interpretation is straightforward: it is the area in a gravity set-
tling device that would be required to give the same performance as the centrifuge
at equal volumetric flow rates. Centrifuge manufacturers can provide Σ values for
their models.

6.9 Review questions

1. Derive an expression for estimating vg using a swing-out test tube centrifuge as

follows. Starting with equation (6.8) and recognizing vc = dR/dt for the particle
settling, solve the resulting differential equation to show that vg = g × ln (R/Ro)/
(ω2t), where R = radius (from center of rotation) at which a particle is located at
time t and Ro = radius (from center of rotation) at which a particle starts. Explain
how you would experimentally determine vg based on your derived equation.

2. You are trying to recover E. coli cells from fermentation broth using an Eppen-
dorf 5418 centrifuge, similar to the one shown in Figure 6.1(a). The cells are sus-
pended in a liquid with a density of 1.00 g/cm3 and a viscosity of 1.0 centipoise
(1 cp = 0.01 g/cm/s).
(a) How much time in hours is required for all the E. coli cells to settle by grav-
ity to the bottom of an Eppendorf tube, which has a height of 4 cm? Assume
that settling is not hindered by neighboring cells.
(b) How much time in seconds is required for all the E. coli cells to settle to the
bottom of an Eppendorf tube while rotating in the Eppendorf 5418 centri-
fuge at 14,000 rpm? Assume that the Eppendorf tube is positioned horizon-
tally in the centrifuge. Also, the top of the tube is 4.5 cm from the center of
rotation. Again, assume that settling is not hindered by neighboring cells.

3. A lab exercise is being designed to determine the effect of feed flow rate on the
concentration of solids in the clarified liquid from a bench-top disc-stack centri-
fuge. The unit does not have automated discharge. You will use flow rates of 200
mL/min, 700 mL/min, and 1,000 mL/min. You anticipate operating at 200 mL/min
 6.9 Review questions 175

for 10 min followed by 700 mL/min for 7 min followed by 1,000 mL/min. How
long could broth flow at 1,000 mL/min before the bowl is half full of solids?
The concentration of solids in the feed is 6% by volume, and the bowl volume
for the unit is 1 L.

4. You are working with a suspension of cells that are approximately 2 μm in diame-
ter. The cells are lysed, and the resulting cell debris has a diameter of approxi-
mately 1/6 of the intact cells. The density of the intact cells and cell debris is
about the same. In a swing-out test tube centrifuge in which the tubes are per-
fectly horizontal during rotation, how many times faster do the whole cells settle
than the cell debris? Hint: begin by recognizing that vc = dR/dt for a particle, as-
sume no hindered settling, and solve the resulting differential equation for time.

5. Saccharomyces cerevisiae cell broth is fed to a bench-top disc-stack centrifuge

with Σ = 1,900 m2. You conduct a study on the impact of flow rate on separation.
At what flow rate would you expect to see yeast cells in the clarified liquid, as-
suming that all cells are exactly the same size? The bowl is not allowed to fill
significantly with solids.

6. You are a process engineer tasked with designing a disc-stack centrifugation

step for recovery of cells from a 2,000 L E. coli fermentation. You will be using
an existing production-scale disc-stack unit with Σ = 5,230 m2. In development
studies using a bench-top disc-stack centrifuge (Σ = 1,900 m2), you determined
Q = 1,200 mL/min to be a suitable flow rate. Calculate the time required in hours
to process the 2,000 L E. coli fermentation using the production centrifuge.

7. You are working on a centrifugation step for the clarification of CHO culture in
a monoclonal antibody (mAb) production process. Recall that mAbs are typi-
cally secreted by CHO cells. The mAb process you are working on requires that
you centrifuge 5,000 L of CHO cell culture broth. Estimate the minimum size for
a centrifuge to clarify the CHO cell culture.

8. In problem 6, a Qmax value = 1,200 mL/min was observed for separation of E.

coli using a bench-top disc-stack centrifuge (Σ = 1,900 m2). Using equation (6.12)
and the settling velocity for E. coli presented in this chapter, calculate the theo-
retical Qmax. Give three possible reasons for the difference in the observed and
calculated values.

9. A 1,000 L Pichia pastoris fermentation at a cell concentration of 50% by volume

is clarified (product is secreted) using a disc-stack centrifuge. In order to achieve
the required liquid clarification, fermentation broth is diluted to five times its
original volume. The bowl volume is 3 L and you will allow the bowl to fill to 1/2
of its solids capacity before discharging.
176 Chapter 6 Harvest operations, Part 2: solid-liquid separations by centrifugation

(a) If the flow rate to the centrifuge is 850 L/h, what is the feed interval in
seconds?
(b) Estimate the time required for the run, in hours.

10. Show that the expression for Σ in a tubular centrifuge is given by [π × L × (Ro2 – R12)
× ω2]/[g × ln(Ro/R1)]. (Note that Ro is the distance from the center of rotation to
the liquid surface within the bowl, and R1 is the distance from the center of
rotation to the bowl wall.) To do this requires developing a differential equa-
tion for the rate that a particle moves in the radial direction (dR/dt) and the
rate at which it moves in the axial direction, pushed by the flow of fluid (dz/dt).
The equations can then be combined in the form dR/dz = … to give a particle tra-
jectory equation (that is, an equation describing the radial and axial position of a
particle) from which a relationship in the form of Q = vg × Σ can be written.
Chapter 7
Harvest operations, Part 3: solid-liquid
separations by filtration

In addition to centrifugation, filtration is a commonly used operation for solid-liquid

separations in biopharmaceutical processes. In this chapter, we cover two distinctly
different but common types of filtration steps in biopharmaceutical processing: depth
filtration and tangential-flow microfiltration (MF). We begin by discussing fundamen-
tal aspects of filtration and then turn our attention to details related to depth filtration
and tangential-flow MF. Specific questions addressed are:
– What is filtration?
– What is the difference between tangential-flow and normal-flow filtration? What
are the advantages and disadvantages of each?
– What is the difference in how solids are retained by a depth filter and an MF
membrane?
– What materials are depth filters and MF membranes made from and how are
they “packaged” for large-scale processing?
– What is the basic equipment setup and procedure for a depth filtration step? A
tangential-flow MF step?
– What are the important process and performance parameters for depth filtration?
– What bench-scale studies are used to select a depth filter and how are depth fil-
tration steps scaled up?
– In addition to centrifugation (covered in the last chapter), depth filtration, and
tangential-flow MF, what other operations are used for solid-liquid separations
in biopharmaceutical processes?

You may notice that our coverage of filtration topics in this chapter is somewhat
lopsided, favoring depth filtration. That is by design. We only cover basic concepts
related to MF in this chapter, holding off on discussing a number of details until
Chapter 9, which focuses on ultrafiltration (UF). UF is a close “cousin” to MF, and
certain concepts – such as process parameters, development studies, and scale up –
are similar for both and, therefore, are covered together.

7.1 Filtration definition

Let’s start our discussion with a definition of filtration. A good one is given by Harrison
et al. [162] as follows: any technique used to separate particulate in a fluid suspension or
components in solution according to their size by flowing under a pressure differential
through a porous medium. The first part of the definition refers to the separation of solid

https://doi.org/10.1515/9783110616880-007
178 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

particles from a fluid suspension and covers applications such as removing CHO cells
from a cell culture broth to produce antibody-containing clarified liquid ready for the
purification stage. This type of application is what most of us think about when we use
the term filtration: removal of solids from a liquid. But the definition also includes the
separation of components in solution. An example is the use of UF to concentrate (i.e.,
remove water from) a protein solution or to exchange one buffering solution for another.
UF is commonly used as a formulation step for a drug product and, as previously men-
tioned, is discussed in detail in Chapter 9. This chapter focuses on the first part of the
definition: removal of solids from a fluid suspension.

7.2 Filtration applications in biopharmaceutical processing

There are numerous examples in which filtration is used to perform solid-liquid sep-
aration in biopharmaceutical processes, as summarized in Table 6.1 in the previous
chapter. Let’s take a closer look at one example in Figure 2.3, which shows a typical
process for production of a monoclonal antibody in CHO cells. In the process repre-
sented by that figure, CHO cells secrete the monoclonal antibody into the culture
medium, and a combination of centrifugation and depth filtration is used to clarify
(i.e., remove solids from) the cell culture broth in preparation for loading onto a Pro-
tein A chromatography column. This is a common use of depth filtration – removal
of residual particles from the clarified liquid from a centrifuge. In addition, depth
filtration alone – that is, without centrifugation – can be used to produce clarified
cell culture broth. Without centrifugation, it is common for depth filtration to be
performed in two stages: a first stage with a larger pore size filter medium followed
by a second stage with a smaller pore size. Alternatively, tangential-flow MF could
be used in place of both the centrifugation and depth filtration steps.
As described in Table 6.1, depth filtration and tangential-flow MF can also be
used for other applications in biopharmaceutical processes. For example, they can
be used to clarify lysate streams generated from lysis steps used to release intracel-
lular product. Examples of processes requiring lysate clarification include E. coli-
based processes for protein therapeutic production and production of certain viral
vectors for gene therapy using human embryonic kidney 293 (HEK293) cells. Both
types of filtration can also be used for removal of precipitated impurities from pro-
cess streams. And normal-flow MF is commonly used for bioburden removal, but we
focus on tangential-flow MF in this chapter.
 7.3 Normal-flow vs. tangential-flow filtration 179

7.3 Normal-flow vs. tangential-flow filtration

There are two common modes of operation for filtration steps: normal flow, also re-
ferred to as dead end, and tangential flow, also referred to as cross flow. As the names
suggest, in normal-flow filtration, feed flows perpendicular to the filter medium, in the
same direction as filtrate flow. Filtrate refers to the fluid that passes through the filter
and is therefore relatively free of solids. In tangential-flow filtration, feed is pumped
tangentially along the surface of the filter, perpendicular to the filtrate flow, with only a
portion of the feed permeating the filter. The two scenarios are illustrated in Figure 7.1.
It is important to note that even though the images depict what appears to be solid par-
ticles being removed from the filtrate, the same illustrations apply to the case in which
dissolved solutes, such as proteins, are retained by a filter. In this chapter we discuss
solid particles and discuss dissolved solutes in Chapter 9. As you can see, in normal-
flow filtration solid particles accumulate at the surface of the filter, and the filtrate rate

Pf

Feed pressure at Pf Retentate pressure at Pr

Membrane/Filter Membrane/Filter
Filtrate at pressure P Filtrate at pressure P

Figure 7.1: Normal-flow vs. tangential-flow filtration. Note that the graphs apply to the operating
scenario in which Pf, the feed pressure to the filter, is constant. Image © NC State University;
reprinted with permission.
180 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

quickly decreases with time for the scenario in which feed pressure is held constant.
The flow of the filtrate carries particles to the filter surface, but their build up is limited
due to the sweeping action of the tangential feed flow. Thus, the filtrate rate levels off
with time, rather than decreasing to zero. Depth filtration is operated in normal-flow
mode, while MF is run in tangential-flow mode when the solids load is high.
As described in the definition previously presented, filtration is a pressure-driven
process. In the normal-flow filtration diagram in Figure 7.1, the pressure differential
that drives filtrate flow is the difference between the feed and filtrate pressures, given
by:
ΔP ¼ Pf  Pfiltrate (7:1)

where ΔP is the pressure difference, and Pf and Pfiltrate are the pressures of the feed
and filtrate streams, respectively. Note that the filtrate is typically at atmospheric
pressure. As long as ΔP > 0 and the filter is not completely plugged, flow through
the filter will occur. In tangential-flow mode, the relevant pressure differential is de-
fined differently. Because feed flows tangentially along the filter surface, there is a
pressure drop that occurs from the feed inlet to the feed outlet. The effluent from
the feed side of the membrane is referred to as the retentate – think slurry that is
retained by the filter. Assuming a linear decrease in pressure along the length of the
filter feed channel, an average pressure on the feed side of the filter is calculated as:
(Pf + Pr)/2, where Pr is the pressure of the retentate. The relevant pressure differen-
tial that drives permeation across the filter is known as the transmembrane pres-
sure, TMP, defined as:

Pf + Pr
TMP ¼  Pfiltrate (7:2)
2

7.4 Solids retention by filters

Two ways in which particulates are retained by a filter are illustrated in Figure 7.2.
With surface filtration, particles are sieved out at the surface of the filter medium. An
MF membrane is an example of a surface filter, with pore sizes as small as 0.1 μm or
as large as 10 μm [173]. Depth filters, on the other hand, trap particles not only at the
surface, but within the filter medium. Depth filters are typically operated in normal-
flow mode, while surface filters are operated in either normal-flow or tangential-flow
mode. Depth filters have a high solids capacity relative to surface filters because they
are able to retain a large number of particles before becoming plugged; consequently,
depth filters are often the normal-flow filtration method of choice to clarify feeds con-
taining a large solids load, such as cell culture broth. Using a surface filter in normal-
flow mode for this type of application is rare beyond the bench scale, as the surface
area required would be large and therefore prohibitively expensive. Using a surface
 7.5 Depth filtration 181

Feed Filtrate Feed Filtrate

Solids load at surface Solids load throughout depth

Figure 7.2: Depth vs. surface filtration in normal-flow mode. Image © NC State University; reprinted
with permission.

filter in tangential-flow mode, on the other hand, minimizes solids build-up at the
surface and allows filtration to proceed for a longer time, which means less filter area
is required relative to normal-flow operation.
Note that filters that rely on a surface mechanism for filtration are often referred
to as membranes, such as the MF membrane already referred to. Depth filters, on
the other hand, are not typically referred to as membranes but rather just filters. We
use this terminology convention in this book. It is also worth noting that the filtrate
from a surface filter (i.e., membrane) is often referred to as permeate. The term per-
meate is typically used with MF membrane- and UF membrane-based steps.

7.5 Depth filtration

We now turn our attention to details on depth filtration. As discussed previously,

depth filtration is used to remove particles from feeds such as cell culture broth;
thus, removing the solids and achieving the necessary clarity in the resulting filtrate
is a primary objective. In addition, the design and implementation of a depth filtra-
tion step focuses on maximizing step yield (i.e., ensuring that product easily passes
through the filter and is present in the filtrate) and filter capacity for solids. Let’s
begin by focusing on the depth filters themselves.
182 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

7.5.1 Depth filters and filter modules

Depth filters used in biopharmaceutical processes are most commonly made of cellu-
lose fibers bound by a polymeric resin that adds a positive charge to the filter medium
[174]. Depth filters may also contain a filter aid, such as perlite or diatomaceous
earth, to provide increased surface area for improved solids capacity. There are also
depth filters made from polypropylene fibers, polyacrylic fibers, and activated carbon
that find use in biopharmaceutical applications [175].
Figure 7.3(a) shows an image of cellulosic depth filtration media. Note the wide
distribution of pore sizes, a common characteristic of depth filters. Filters for solid-
liquid separations are typically rated by a pore size, which equates to the size of the
largest particle that can pass through the filter. For depth filters, the pore size rating
is described as “nominal” rather than “absolute.” A filter with a nominal rating can
retain some percentage – significantly less than 100% – of solid particles of greater
than the stated pore size. For example, a nominal rating of 0.8 µm may mean that
the filter retains only 60% of particles with a diameter larger than 0.8 µm. Filters
with a relatively wide distribution of pore sizes, such as a depth filter, are typically
given a nominal rating. An absolute rating specifies the pore size at which a more
significant percentage of particles of a particular size (e.g., 99.9%) will be retained
by the filter. It is usually used for a filter that has a fairly narrow pore size distribu-
tion. MF membranes are an example.
Figure 7.3(b) shows an example of how the depth filtration media is packaged
to create a process-scale single-use depth filter, in this case a Millistak+® HC Pro
D0SP Pod device from MilliporeSigma. Flow through this type of device is illus-
trated in Figure 7.3(c), which shows that feed through the filter is split equally
among multiple identical filtration media “packets” within the Pod; that is, the Pod
is configured so that each packet is in parallel. These packets include one or more
layers of filter media and filter aid. Filtrate from the packets combine within the Pod
to create a single filtrate stream. This type of filter construction, in which feed is
equally distributed to each packet in a module, is common among depth filters cur-
rently on the market. However, not all filters have a rectangular geometry like the
MilliporeSigma Pod devices shown.
To provide the filtration surface area required for a given application, modules are
typically placed in a parallel configuration into a holder, as shown in Figure 7.3(d) for
Pod devices. Filtration area may be scaled by increasing the number of Pod devices
with the Pod rack. For the filter module shown in Figure 7.3(b), the process-scale holder
can accommodate up to 16 m2 (= 0.77 m2/Pod × 7 Pods/rack × 3 racks) of filtration area.
Depth filters from other vendors are commonly used in biopharmaceutical applications
as well, including Stax™ capsules from Pall, Sartoclear® depth filters from Sartorius
Stedim Biotech, and Zeta Plus™ encapsulated filter capsules from 3M.
In modern biopharmaceutical applications, depth filtration is typically a single-
use operation. All the depth filters mentioned in this chapter are sold as single-use
 7.5 Depth filtration 183

(a) (b)

(c) (d)

Figure 7.3: Depth filter images. (a) shows an electron micrograph of cellulose fibers that make up
Millistak+® CE40 depth filtration media. (b) is MilliporeSigma’s Millistak+® HC Pro D0SP Pod
device with 0.77 m2 of filter area. The media is made from silica filter aid with polyacrylic fiber
[176]. (c) shows schematically how the feed stream is equally distributed to layers (only two are
shown) of depth filtration media within a Pod device. (d) is a photograph of a Millistak+® Pod filter
holder showing how multiple filtration modules are placed in a parallel configuration to provide the
required surface area [177]. Images reproduced with permission from Merck KGaA, Darmstadt,
Germany and/or its affiliates.

encapsulated modules. The equipment required to execute the step is also available
in a single-use format, with details provided in the next section.
Depth filters come in a range of pore sizes, and, as already mentioned, for a given
filter type, there is a wide distribution of pore sizes. This range of pore sizes is shown
in Figure 7.4 for depth filtration products offered by MilliporeSigma. The chart shows
depth filtration media with nominal pore size ratings from slightly less than 0.1 µm
up to greater than 10 µm. It is also worth pointing out that multi-layer depth filtration
 184

Figure 7.4: Various depth filtration media grades for Millistak+® family and Clarisolve® depth filters (from MilliporeSigma) and their nominal pore size
rating in microns [178]. Image reprinted from “Millistak+® family and Clarisolve® depth filters at a glance” 2017, p. 1 © by MilliporeSigma Corporation
and reproduced with permission from Merck KGaA, Darmstadt, Germany and/or its affiliates.
Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration
 7.5 Depth filtration 185

media is available. Typically, multi-layer media is made of a more open first layer fol-
lowed by tighter second layers. For example, from Figure 7.4, MilliporeSigma’s
Millistak+® HC D0HC media combines a layer of CE25 media, which has a nominal
pore size of approximately 6 μm, with a layer of DE40 media, which has a nominal
pore size between 0.5 and 1 μm.
While solids retention in a surface filter, such as an MF membrane, is due
mainly to particle sieving at the surface, a number of retention mechanisms in a
depth filter enable it to trap particles. Primary mechanisms for retention are de-
scribed below [175] and illustrated in Figure 7.5:
– Sieving. Particles cannot pass into the filtrate because they are larger than the
filter pores. Sieving may occur at the entrance to a pore, such as at the surface,
or in a restriction within the pore. Both types of sieving are shown in Figure 7.5.
– Adsorption. Electrostatic interactions between the particles and filter occur, which
results in adhesion of particles to the filter surface and enables removal of particles
smaller than the size of the filter pores. Recall that the resin binder used in depth
filters imparts a positive charge to the filter that may attract negative charge on the
surface of the particles. It is worth noting that if the depth filter media has a posi-
tive charge and the active ingredient dissolved in the feed has a negative charge,
product loss (i.e., reduction in step yield) may occur during filtration due to bind-
ing of product to the filter media.

Surface sieving Pore sieving Adsorption Flow

Figure 7.5: Retention mechanisms in a depth filter. Note that an actual depth filter is more porous
than the image shown. Image © NC State University; reprinted with permission.

7.5.2 Depth filtration equipment and procedures for production

A schematic of a depth filtration setup is shown in Figure 7.6. Feed from a bioreactor or
process vessel is pumped to the depth filter. Gauges on the feed and filtrate streams
monitor pressure, in particular the pressure difference across the depth filter. For many
applications, including clarification of cell culture broth to prevent plugging of a subse-
quent chromatography column, a 0.2 μm MF membrane filter (i.e., a filter with 0.2 μm
pores) is placed in series with the depth filter to ensure the filtrate is largely free of
186 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

Bioreactor or
process vessel

Pump

Filtrate to

Figure 7.6: Schematic of a depth filtration setup for biopharmaceutical processing. Image © NC
State University; reprinted with permission.

solids. Thus depth filtration can be viewed as a pre-filtration step for the 0.2 μm filtra-
tion step. It is worth noting that many process flow diagrams do not show the 0.2 μm
MF membrane filter after the depth filter, as it is assumed to be part of the overall
depth filtration step and not a separate operation.
As mentioned previously, in addition to the filters themselves, the equipment
required to execute a depth filtration step is available in single-use format. Specifi-
cally, single-use skids that contain all the components shown in Figure 7.6 are
available. Examples include the Allegro™ MVP Single-use System from Pall Corpo-
ration and the Mobius® FlexReady Solution for Large Scale Clarification from Milli-
poreSigma. These systems rely on disposable flow paths that are replaced after each
run. The flow paths can be equipped with disposable sensors as well. For example,
the Allegro™ MVP system includes both single-use pressure and flow sensors [179].
It is also worth noting that depth filtration steps can be carried out without a skid
by bringing together the individual components, including tubing, valves, pressure
gauges, and a pump, and making proper connections to a holder that contains the
encapsulated filters.
A typical depth filtration procedure is summarized by the diagram in Figure 7.7.
Initially, the filters are flushed with high purity water (HPW) to remove entrapped
air and leachable components, followed by a buffer to condition the filter prior to
use. Then filtration of the feed stream begins, with product (i.e., filtrate) collection
beginning after all HPW or buffer held up in the system from the previous step is
purged to drain. Note that it is not required to purge the system of the flushing liq-
uid first; however, it will have a dilution effect on the product if collected with the
filtrate. Upon completion of the product filtration, buffer is pumped through the
system to recover product that remains held up in the system. A blowdown step in
which pressurized air is used to displace the buffer held up in the system may be
performed afterwards to reduce the weight of the filter modules and to facilitate a
 7.5 Depth filtration 187

air that would otherwise reduce adsorptive capacity, and

Removes particulate (e.g., cells, cell debris, precipitates)

Filtration
from feed stream

Buffer chase Recovers product in the system hold up

blowdown clean dissasembly

Cleaning,
sanitization disposal, removes product from surfaces of multi-use
and/or sterilization equipment in preparation for next run

Figure 7.7: Steps involved in a depth filtration procedure after the filter(s) has been installed in the
system. Image © NC State University; reprinted with permission.

cleaner disassembly. Finally, the filters and single-use components are disassembled
and discarded as biohazardous waste. Reusable components, such as a multi-use fil-
ter housing, must be cleaned and sanitized with a validated method (assuming a
GMP process) because there is a potential that some of their surfaces have contacted
product.

7.5.3 Depth filtration parameters, performance and design

Before beginning the discussion of depth filtration performance and development,

it’s necessary to define a few terms:
– Feed flux: the volumetric feed rate across the filter per unit area. It is controlled by
the system pump and typically expressed in units of LMH (L of suspension fed per
m2 of filter area per hour). If, for example, the feed flux across the Pod filter in
Figure 7.3(b) is 150 LMH, the total flow rate being fed during filtration is 150 L/m2/
h × 0.77 m2 or 116 L/h. The feed flux, Jf, across the depth filter can be generally
represented by the following expression [180]:

ΔP
Jf ¼ (7:3)
μRtotal
188 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

where ΔP – the driving force for feed flux – is defined by equation (7.1), μ is the
viscosity of the liquid making up the slurry, and Rtotal is the resistance to flow
through the filter. Contributions to resistance include the filter medium, particles
blocking the entrance to pores, and particles restricting pores internally. The resis-
tance increases as the depth filtration run proceeds because particles accumulate
on or in the filter and increasingly impede flow. Depth filtration is typically oper-
ated with a constant feed flux; therefore, based on equation (7.3), the driving
force for flow, ΔP, will increase to keep the flux constant because the resistance,
R, increases as the depth filtration run proceeds.
– Capacity (also referred to as throughput or loading): the maximum volume of suspen-
sion that can be filtered (because a maximum pressure difference has been reached,
for example) per unit area of filter. It is typically expressed in units of L/m2. Note that
we use the term throughput to more generally mean the volume of material filtered
per unit area of filter – but not necessarily the maximum volume that can be filtered.
Thus, capacity represents the maximum throughput for the filter system.

Example: Calculating depth filtration area requirements and process time

Consider the removal of cells from a CHO culture broth. 2,000 L of broth is to be clarified
using a depth filter with a capacity of 250 L/m2 at a feed flux of 150 LMH. Answer the following
questions:
(a) What is the depth filtration area, Adepth,min, required?
(b) The depth filtration medium in use is packaged in modules with a filtration area of 0.77 m2.
How many modules are required?
(c) How much time would be required to process this broth?

Solutions
(a) The area required is the ratio of the volume of broth fed to the depth filters to the capacity
of the filter:
Vfeed
Adepth;min ¼ (7:4)
capacity
2000 L
¼
250 L=m2

¼ 8 m2

(b) The number of modules is calculated by dividing the depth filtration area calculated previ-
ously by the area per module:

8m2
# modules ¼
0:77 m2 =module
= 11 modules ðrounding up from 10:4Þ
 7.5 Depth filtration 189

(c) 11 depth filtration modules at 0.77 m2 per module leads to 8.47 m2 of filtration area. The
feed rate, Q, is the product of the feed flux (150 LMH) and the depth filtration area:

Q ¼ Jf Adepth (7:5)
L
¼ 150 × 8:47m2
m2
¼ 1271L=h
The process time, tprocess, is calculated from equation (6.7) as:

Vfeed 2000 L
tprocess ¼ ¼
Q 1271 L=h
¼ 1:57 h ðor 94 minÞ
Note that often a safety factor is used in the design of filtration steps. For example, a safety
factor of 1.5 means 50% more filter area is used, Ainstalled, than is calculated using data from
bench-scale studies, Adepth,min; that is

Ainstalled ¼ safety factor × Adepth;min (7:6)

Safety factors are intended to account for the possibility of feed stream variability and perfor-
mance differences upon scale up [181]. So, if a 1.5 safety design factor were used here, Ainstalled
would be 12.7 m2 (=1.5 × 8.47 m2), and the number of modules and processing times would be
calculated based on this value.

– Turbidity: cloudiness of solution due to the presence of particles. It is typically

measured using a nephelometer, which measures the scattered light in an illu-
minated sample by placing a detector at a specific angle (often at a right angle)
to the incident light. The more scattered the light, the higher the turbidity. It is
expressed in nephelometric turbidity units (NTU). See Figure 7.8 for a photo of
suspensions with different turbidity values.
– Turbidity breakthrough: the point where particles have made it all the way through
the filter and are observed in the filtrate.
– Fouling: plugging of the filter due to accumulation of solids on and in the depth
filtration media.

Figure 7.8: Standards with turbidity values in vials A-F of 0, 50, 100, 250, 500 and 1,000 NTUs,
respectively. Reprinted from “Evaluation of Microflow Digital Imaging Particle Analysis for Sub-Visible
Particles Formulated with an Opaque Vaccine Adjuvant” by G. Frahm, A. Pochopsky, T. Clarke, M.
Johnston, 2016, PLoS ONE, under license by Creative Commons Attribution 4.0 International [182].
190 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

Let’s begin by considering input parameters and output (performance) parameters

in a depth filtration step. Input parameters include the following: (1) type of depth
filter, (2) feed flux, (3) solids concentration of the feed, and (4) particle size distri-
bution of feed. Note that the first three are controllable and therefore fit our defini-
tion of process parameter presented in Chapter 2. The last is a property of the
process intermediate fed to the depth filter and may not be controllable. Perfor-
mance parameters include (1) filter capacity, (2) filtrate turbidity or turbidity reduction
(= turbidityfeed – turbidityfiltrate), (3) pressure differential across the filter (ΔP defined by
equation (7.1)) at constant feed flux, and (4) percent recovery of product (i.e., step yield).
A number of studies have been conducted that shed light on the impact of pro-
cess parameters and other input parameters on depth filter performance. For exam-
ple, Figure 7.9 presents data on the impact of feed flux on filter capacity and filtrate
turbidity [183]. These data were obtained by varying the feed flux of a Chinese ham-
ster ovary (CHO) culture, at a density of 3.6 × 106 cells/mL and viability of 70%, to a
depth filter with a pore size rating of 8 µm. The data show that as feed flux in-
creases, filter capacity decreases and turbidity increases. The increase in turbidity is
attributed to adsorptive (electrostatic) capture of particles becoming less efficient as
the residence time within the filter medium decreases (due to increased flow rate).
This increase in filtrate turbidity with increasing feed flux would also result in a de-
crease in the capacity of a 0.2 µm surface filter placed after the depth filter, due to
increased solids loading to the 0.2 µm filter. Thus, more frequent changeout of the
filter and/or a larger filtration area would be required of a 0.2 µm filter as the feed
flux to the preceding depth filter increases.
Properties of the feed may have an impact on depth filtration performance as
well, as illustrated in Figure 7.10. These data were obtained from a process in which
a yeast (Pichia pastoris) fermentation broth was clarified using a centrifuge. The re-
sulting clarified liquid was processed for additional solids removal using Millipore-
Sigma’s Millistak+® A1HC depth filtration media at a constant feed flux of 250 LMH.
Two different batches of clarified liquid were used: one with a solids content of 0%
(that is, very low solids concentration) and another at 0.7%. Figure 7.10 shows pres-
sure differential, ΔP, and turbidity as a function of throughput for each feed. We
might expect that the 0.7% feed would show greater ΔP and filtrate turbidity values,
relative to the 0% feed, given the higher particle concentration. While ΔP for the
0.7% feed is always higher than that for the 0% feed, the rise in turbidity for the 0%
feed is more rapid than that of the 0.7% feed. So what’s going on? The feed with the
low solids concentration likely contains smaller particles than the feed with the
higher solids concentration. These small particles bind mainly by an adsorption
mechanism rather than a sieving mechanism. The adsorption binding sites quickly
become saturated, resulting in solids breakthrough for the 0% solids feed. On the
other hand, for the feed with the higher solids content, sieving is the main retention
mechanism, and as pores become blocked, ΔP increases much more rapidly than for
the low-solids feed. This interesting result clearly demonstrates that properties of
 7.5 Depth filtration 191

8μm Depth Filter Capacity as a Function of Different Flow Rates

3.6 x 106 CHO cells/ml at 70% viability level
Capacity (liter/m2)
350

300

250

200

150

100

50

0
Constant flow mode Constant flow mode Constant flow mode Constant flow mode
at 100l/m2/h flow at 200l/m2/h flow at 300l/m2/h flow at 500l/m2/h flow

Effluent Turbidity as a Function of Flow Rate

3.6 x 106 CHO cells/ml at 70% viability level into 8μm depth filter
Filter effluent turbidity
[NTU]
200

180

160

140

120

100

80

60

40

20

0
Constant flow Constant flow Constant flow Constant flow
mode at 100l/m2/h mode at 200l/m2/h mode at 300l/m2/h mode at 500l/m2/h
flow flow flow flow

Figure 7.9: Depth filter capacity and filtrate turbidity as a function of feed flux for filtration of a CHO
culture at a density of 3.6 × 106 cells/mL and viability of 70%. Note that a depth filter with a pore size
rating of 8 µm was used. Adapted from “Depth filtration: Cell clarification of bioreactor offloads,” by
M. Prashad and K. Tarrach, 2006, Filtration and Separation, 43(7), 29. © (2006) Elsevier B.V [183].
192 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

Figure 7.10: Pressure difference and filtrate turbidity versus throughput for depth filtration of
centrifuge clarified liquid from a Pichia pastoris fermentation broth. Millistak+® A1HC depth
filtration media was used at a constant feed flux of 250 LMH. (a) shows results for a clarified liquid
feed with very low solids concentration (0% solids). (b) shows results for a clarified liquid with
0.7% solids. Image copyright 2007 from Process Scale Bioseparations for the Biopharmaceutical
Industry by Shukla, A; Etzel, M; Gadam, S., eds. Reproduced by permission of Taylor and Francis
Group, LLC, a division of Informa plc [184].

the feed to the depth filter, in this case solids concentration and particle size, can
impact filter performance.
Table 7.1 presents a summary of the impact of process parameters and other
input parameters on process performance, evaluated by considering their impact on
capacity, filtrate turbidity, and pressure drop across the filter. Note that in addition
to the parameters shown, ionic strength of the feed may impact particle retention
 7.5 Depth filtration 193

Table 7.1: Depth filtration input parameters and their impact on performance.

Input Parameter Impact on Depth Filter Performance

Depth filter type For a fixed feed flux and a feed stream with fixed properties (solids
concentration, particle size distribution), as pore size decreases
– Filter capacity (based on pressure drop) decreases
– Filtrate turbidity decreases
– ΔP rises more rapidly as filtration proceeds
Feed flux For a fixed filter and feed stream, as feed flux increases
– Filter capacity decreases
– Filtrate turbidity increases
– ΔP rises more rapidly (either plotted against time or throughput) as
filtration proceeds
Solids concentration in For a fixed feed flux and filter, as solids concentration increases (at
feed constant particle size)
– Filter capacity (based on pressure drop) decreases
– Filtrate turbidity may increase
– ΔP rises more rapidly (either plotted against time or throughput) as
filtration proceeds
Note that solids concentration in the feed can be impacted by bioreactor
operating conditions

Particle size For a fixed feed flux and filter, as particle size decreases
distribution in feed – Filter capacity (based on filtrate turbidity) decreases
– Filtrate turbidity may increase
– ΔP rises less rapidly as filtration proceeds
Note that particle size distribution in feed can be impacted by bioreactor
operating conditions

for cases in which adsorption is a primary retention mechanism. At high concen-

trations, ions may occupy binding sites that would otherwise be used for adsorp-
tion of particles. Lot-to-lot variability of filters may also impact depth filtration
performance.
How do we select a filter for a given application and know how much filter area
will be required for a process? If information on the size distribution of particles to be
filtered is known, initial selection of depth filtration media may be done based on
pore size using charts like the one shown previously in Figure 7.4. However, ulti-
mately it is prudent to conduct a bench-scale study, using feed material that is repre-
sentative of production feed, as particle size distribution is but one of many variables
that impacts depth filtration performance, and the amount of filter area required is dif-
ficult to predict theoretically. Most filter manufacturers offer small-area depth filtration
devices for these bench-scale studies.
A common approach to gathering sizing data and screening filters is a constant
flow method referred to as a Pmax™ study, which stands for maximum pressure
194 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

[185, 186]. Recall that under constant flux conditions, the pressure drop across the
filter increases as the pores becomes blocked and plugged. The Pmax™ study meas-
ures the filter capacity by loading feed until a maximum pressure drop, ΔPmax, for the
filter is reached. The procedure is straightforward. A representative feed is pumped at a
constant flow rate through a small-scale version of the depth filter. Feed flux values of
100–200 LMH are common. The filtrate volume, feed pressure, filtrate pressure, and
filtrate turbidity are measured with time and recorded. The test is stopped when the
feed pressure reaches the chosen ΔPmax. Typically, the product concentration, turbidity
in the feed and the turbidity of final filtrate pool are measured to determine product
step yield and overall clarification efficiency. The filter capacity is simply the liters fed
per square meter of filter area at ΔPmax.
The maximum pressure drop can be based on the allowable pressure for the fil-
ter, accounting for the fact that at large-scale, the depth filter is likely to be followed
by a 0.2 µm surface filter that has its own pressure limitations. For example, Milli-
poreSigma’s Millistak+® process-scale Pod filters are rated for a maximum operating
pressure of 50 psig and a maximum pressure differential of 30 psi [187]. You might
choose to limit the pressure drop across the depth filter to 25 psi, estimating that the
0.2 um surface filter in the setup would contribute an additional 15 psi of pressure
drop. This additional pressure drop would require a feed pressure (to the depth fil-
ters) of 40 psig for operation, just below the allowable maximum pressure. (Note
that the depth filter filtrate pressure would be 15 psig with a 25 psi pressure drop,
and the 0.2 µm filtrate would be at 0 psig with a 15 psi pressure drop.)
Used for screening, the Pmax™ method provides data (ΔP vs throughput, filter
capacity, product recovery, and turbidity) that enables comparison of the perfor-
mance of different depth filtration media. Used for sizing, the data provide a capac-
ity value from which a production system can be sized if the same feed flux and
process time are used in production as in the bench-scale Pmax™ study. Alterna-
tively, if a different filtration process time is desired, data from the Pmax™ run are
plotted as the filter resistance versus throughput. Note that the filter resistance is
given by the expression R = ΔP/(μJf), obtained by solving equation (7.3) for R. If the
fluid viscosity μ is incorporated into the resistance, then resistance has units of
pressure over flux (e.g. psi/LMH). The resistance versus throughput data is fitted
with a (typically) third-order polynomial. The expected feed volume at production
scale and desired process time are then used along with the polynomial equation
that expresses resistance as a function of throughput to iteratively determine the
required filtration area for the chosen feed volume and time [186].
There is one last point to make regarding depth filtration before moving to tan-
gential-flow MF. Depth filtration is not only capable of clarification, but it can act as
a purification step as well. The charge on depth filters allows them to be effective at
removing soluble process-related impurities such as host cell proteins, host cell DNA,
and virus, presumably because many of these impurities have a negative charge at
process conditions, which promotes binding to the positively charged depth filtration
 7.6 Tangential-flow MF overview 195

media. A number of studies have demonstrated clearance of these impurities [188,

189]. Note that these impurities are discussed in greater detail in Chapter 8.
At this point, we turn our attention to another common filtration method used
for separation of solids from liquid: tangential-flow MF. As mentioned at the begin-
ning of the chapter, tangential-flow MF is not covered in as much detail as depth
filtration in this chapter. Instead, we wait to present some details, particularly those
related to membrane materials, parameters, process development, and scale up,
until we cover the related topic of tangential-flow UF in Chapter 9.

7.6 Tangential-flow MF overview

As the name suggests, tangential-flow MF involves the use of MF membranes oper-

ated in tangential-flow mode for solid-liquid separations. In contrast to depth filtra-
tion, in which only a clarified filtrate can be collected as product (i.e., solids cannot
be recovered from a depth filter), either the clarified permeate or the retained solids,
in the form of a concentrated cell slurry, can be recovered as product with a tangen-
tial-flow MF system. The primary mechanism of particle retention for MF mem-
branes is sieving at the membrane surface. MF membranes have pore sizes as small
as 0.1 μm or as large as 10 μm [173]. They are one of a number of different membrane
types used for separations and classified by their pore size. Others include UF mem-
branes, the topic of Chapter 9, which have a smaller pore size; they retain solutes
with a diameter of 0.005–0.15 µm [190]. Reverse osmosis membranes, another ex-
ample, have pores small enough to separate water from ions such as Na+ and Cl-.
Reverse osmosis is commonly used in the biopharmaceutical industry not as a pro-
duction step but as one of several steps used to purify water. A graphic comparing
different types of membrane separations is shown in Figure 7.11.
Tangential-flow MF steps are designed and implemented to deliver the desired
clarity in the resulting permeate if the product is dissolved in the liquid phase or to
achieve a target concentration of solids if solids recovery is the primary objective.
Maximizing step yield, short and reproducible processing times, acceptable mem-
brane area, and, of course, acceptable product quality are additional goals.
A typical equipment setup for tangential-flow MF is shown in Figure 7.12. To pro-
duce tangential flow through the MF membrane module, a different setup is used
than that shown for depth filtration in Figure 7.6. A pump moves feed slurry from the
feed tank to the MF membrane module. The retentate valve is used to increase pres-
sure on the feed side of the module, which creates the TMP, defined previously in
equation (7.2), that is required to force liquid to permeate the membrane. Two streams
are created as a result of the process: the permeate, which passes through the MF
membrane, and the retentate, the portion of the feed that does not permeate the
membrane and is recirculated to the feed vessel. (Note that in MF, a pump can be
used to lower the permeate flow rate thereby reducing the buildup of solids on the
196 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

Reverse Osmosis
Salts, buffer Cells (e.g., E. coli,
components, and Saccharomyces cerevisiae,
Feed other low molecular Proteins, CHO), cell debris, other
Water weight compounds viruses suspended solids

Reverse Osmosis
Membrane

Water

Salts, buffer Cells (e.g., E. coli,

components, and Saccharomyces cerevisiae,
Feed other low molecular Proteins, CHO), cell debris, other
Water weight compounds viruses suspended solids

Membrane

Water Salts, buffer

components, and
other low molecular
weight compounds

Salts, buffer Cells (e.g., E. coli,

components, and Saccharomyces cerevisiae,
Feed other low molecular Proteins, CHO), cell debris, other
Water weight compounds viruses suspended solids

Membrane

Water Salts, buffer Proteins,

components, and viruses
other low molecular
weight compounds

Figure 7.11: A comparison of different types of membrane separations. Image © NC State

University; reprinted with permission.

membrane). The retentate stream passes through the system multiple times and be-
comes concentrated in solids, while the permeate stream is void of solids but contains
the product in process scenarios like clarification of CHO culture broth. To maximize
recovery of product in the permeate, the retentate remaining in the vessel may be
 7.6 Tangential-flow MF overview 197

Feed tank Retentate

valve Retentate
Wash buffer
stream

Feed MF membrane
stream Permeate Permeate
Buffer feed Feed or pump stream
pump recirculation
pump

Figure 7.12: Equipment setup for a tangential-flow MF step. Image © NC State University; reprinted
with permission.

washed (also referred to as diafiltration) with a buffer. More specifically, as liquid

containing the product permeates the MF membrane, wash buffer is added to the
feed, often at the same rate that liquid permeates, thereby maintaining constant vol-
ume in the feed tank. This procedure washes product from the retentate and into the
permeate and results in enhanced recovery of the product (i.e., step yield) from the
initial feed. A variety of sensors are in place. For example, the pressure of the feed,
retentate, and permeate are typically monitored to measure the TMP. We continue dis-
cussion of TFF equipment in Chapter 9, including more information about MF mem-
branes and membrane modules.
In tangential-flow MF, the permeate (or filtrate) rate is not equal to the feed rate
(also referred to as the recirculation rate) as it is in depth filtration because only a
portion of the feed permeates the MF membrane. The permeate flow rate is usually
expressed as a permeate flux (i.e., the volumetric permeate rate across the mem-
brane per unit area, often in units of LMH). It can be represented by the same gen-
eral equation written for the feed flux across a depth filter (equation (7.3)) but in a
slightly modified form, with Jf replaced by Jp and ΔP replaced by TMP as the relevant
driving force for permeate through the membrane (refer to Figure 7.1)

TMP
Jp ¼ (7:7)
μRtotal

In the case of MF, the resistance can be broken into three contributions: resistance
from the clean membrane (i.e., the intrinsic membrane resistance, Rm), resistance
due to formation of a filter cake on the membrane surface, Rfc, and resistance from
internal membrane fouling, Rif [191]. Rfc and Rif increase with filtration time; there-
fore, permeate flux decreases with time.
198 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

As we have mentioned, there is much more to be covered on the topic of MF,

including procedures for current Good Manufacturing (cGMP) production, parame-
ters that impact and quantify performance, process development, and scale up.
These topics are addressed in Chapter 9.

7.7 Unit operations for solid-liquid separation methods:

a comparison
Now that we have covered in detail the three most common methods for solid-liquid
separations in biopharmaceutical processes, all of which are natural fits for the har-
vest stage, the logical question is, how do we choose one? To help in that decision, a
comparison among the three main operations considered in the last two chapters is
shown in Table 7.2. Note that normal-flow MF has not been included in this table be-
cause its solids capacity is too small to be useful for feeds with a relatively high con-
centration of solids. However as was mentioned previously, normal-flow MF is by far
the most common method for a very important solid-liquid separation application in
biopharmaceutical processing: removal of bioburden.
Table 7.2 shows that each of the operations discussed is effective at separating
solids from liquids and, if designed properly, can produce high product step yields.
If recovery of solids is desired, depth filtration is not an option; centrifugation and
tangential-flow MF are. Note that because cells can be recovered by microfiltration,
MF membranes are often used as cell-retaining devices in perfusion bioreactor set-
ups. Tangential-flow MF typically produces a filtrate with low solids concentration rela-
tive to centrifugation or depth filtration. Generally, the solids capacity for centrifuges is
higher than either depth or tangential-flow MF. Expect the capital cost for a centrifuge
to be higher than equipment for the other two operations. Consumables costs will be
higher for depth filtration, given the need to purchase new depth filters for every batch
of product. Because MF membranes can be reused for the same product, their costs

Table 7.2: Comparison among the solid-liquid separation operations covered in Chapters 6 and 7.

Unit Effectiveness Product Collection Particle- Solids Capital Consumables

Operation at Separating Yield of Solids Free Capacity Cost Cost
Solids from Possible Stream
Liquids

Centrifugation High High Yes No High High Low

Depth High High No No Medium Low High

Filtration

Tangential- High High Yes Yes Medium Medium Medium

Flow MF
 7.8 Other operations for solid-liquid separation 199

may be distributed across many product batches, which lowers the consumables cost
associated with membrane module purchases.

7.8 Other operations for solid-liquid separation

Centrifugation, depth filtration, and tangential-flow MF will continue to be used for

solid-liquid separation applications in biopharmaceutical processes into the foreseeable
future. Advances in these well-used technologies will continue, such as development of
enhanced filter/membrane materials that improve performance and less expensive
single-use options. However, there are also a number of other solid-liquid separation
technologies with strong potential for greater adoption over the coming years. These
include acoustic separation, flocculation followed by filtration or centrifugation, and
magnetic cake filtration. We’ll provide a brief synopsis of each in the following sections.

7.8.1 Acoustic wave separation

Acoustic wave separation is a technology that continuously removes solids, such as

CHO cells, from a feed. The feed slurry flows to a flow channel within an acoustic
chamber. A three-dimensional standing acoustic wave is created within the flow
channel, and the resulting forces cause solids flowing through the chamber to mi-
grate to the nodes of the standing wave. As the solid cell clusters grow, they quickly
settle by gravity, resulting in a clarified liquid exiting the acoustic chamber [192].
The cells can then be pumped out of the chamber. Pall currently offers commercially
the Cadence™ Acoustic Separator [193]. The system operates in continuous mode
and is single use.
Interest in acoustic wave separation is growing, as demonstrated by Millipore-
Sigma’s recent announcement that it will make acoustic technology available for
cell therapy manufacturing – specifically the Ekko™ platform for cell washing and
concentration [194].

7.8.2 Flocculation

Flocculation is a method that has been implemented for solid-liquid separation in

other industries for years and is gaining increasing interest for biopharmaceutical
manufacturing. It is a pre-treatment method in which a flocculating agent is added
to a slurry and causes the dispersed individual particles to adhere to one another,
thereby creating larger particles. This shift to larger particles can enhance separation
by methods such as centrifugation and filtration. A number of different flocculating
agents have been studied over the years. Because animal cell surfaces are known to
200 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

be negatively charged [195], cationic polymers have been found to serve as a good
flocculating agents by neutralizing the cell surface, thus allowing agglomeration of
cells and/or bridging between cell surfaces. Care must be taken in choosing a floccu-
lating agent as strongly cationic polymers can be toxic to cells [195]. One cationic
polymer that has been marketed for cGMP use and has been shown as effective for
flocculating CHO cells is poly (diallyldimethylammonium chloride) (pDADMAC) from
MilliporeSigma. A study comparing depth filtration of a CHO culture broth treated
with pDADMAC against depth filtration of untreated CHO culture broth showed that
filter capacity of the broth treated with pDADMAC was significantly higher than could
be achieved using untreated broth [196]. Of course, a process that incorporates a poly-
mer flocculation step must be designed to provide sufficient clearance of the floccu-
lating agent, particularly if the agent is cytotoxic at the levels used in the process.

7.8.3 Magnetic cake filtration

Magnetic separations use polymeric beads that are magnetic and to which a ligand
has been attached with a high affinity for a component (e.g., a product). The bead
selectively binds the product from a crude feed and is removed by some method
that takes advantage of its magnetic properties. The product is then recovered from
the beads. To date, magnetic separations have been limited to analytical and diag-
nostic applications in which only a small number of beads are required [197].
For biomanufacturing applications, a relatively large number of beads is needed.
To make this separation work at larger scale, magnetic cake filtration has been pro-
posed [197]. The magnetized beads are mixed with a feed and allowed ample contact
time to adsorb the target compound. The beads are then trapped in the filter with the
help of a magnet and washed for removal of cells, cell debris, and unwanted soluble
impurities. Finally the product is recovered by eluting it from the beads. While studies
have only been performed at bench scale [197], the method holds promise for larger
scale biopharmaceutical process applications as it streamlines operations by combin-
ing product harvest and initial purification step.

7.9 Summary

Filtration is any technique used to separate particulate in a fluid suspension or com-

ponents in solution according to their size by flowing under a pressure differential
through a porous medium. This chapter continued the discussion from Chapter 6 on
solid-liquid separation in biopharmaceutical processes by considering methods for
separation of solid particles from a fluid suspension (i.e., the first half of the defini-
tion) by filtration. There are numerous solid-liquid separation applications in bio-
pharmaceutical processes for which filters are well suited, including clarification of
 7.9 Summary 201

cell culture broths, clarification of lysate streams, removal of precipitated impurities

from process streams, and removal of bioburden.
Depth filters operated in normal-flow mode are commonly used and offer the
advantage of high capacity for solids, particularly when compared to surface filters
used in normal-flow mode. They come in a range of pore sizes, from approximately
0.1 µm to greater than 10 µm. Depth filter media is typically packaged into single-
use modules that are available from numerous vendors. The modules are placed in
a parallel array using holders, and the number of modules in the holder can be in-
creased or decreased to achieve the appropriate surface area for a given application.
In addition to the filter modules and holder, equipment required for depth filtration
includes a feed vessel, pump, and pressure gauges to measure pressure drop, as
shown in Figure 7.6. A typical depth filtration procedure for cGMP processing is
shown in Figure 7.7. Depth filtration steps are typically operated at a constant feed
flux (volumetric flow rate through the filter per unit area, measured in L/m2/h). To
select an appropriate filter for a given application, bench-scale screening studies
are conducted. These studies can also be designed to determine the capacity of the
filter – the maximum volume that can be filtered per unit area, typically measured
in L/m2 – for a given feed for scale-up purposes. Pmax™ methodology is commonly
used for capacity measurement and measures the volume of feed that can be loaded
until a maximum allowable pressure drop is reached. The data can then be used to esti-
mate the filter area (i.e., number of filter modules) required for a particular application.
In addition to normal-flow depth filtration, tangential-flow MF is also a common
filtration method for solid-liquid separation in biopharmaceutical processes. MF mem-
branes come in pore sizes similar to depth filters, although the pore size distribution is
narrower. They are surface filters that are run in tangential-flow mode when process-
ing feeds with relatively high solids concentration. A typical process setup is shown in
Figure 7.12. More information on tangential-flow MF, including a discussion of process
and performance parameters, procedures in a cGMP environment, development stud-
ies, and scale up, is provided in Chapter 9.
There are a number of advantages and disadvantages to the three solid-liquid
separation methods described in Chapters 6 and 7: centrifugation, depth filtration,
and tangential-flow MF. For example, if recovery of solids is required, depth filtra-
tion is not an option. For applications in which product is dissolved in the liquid
phase of a suspension, tangential-flow MF is capable of producing a product stream
with low solids concentration relative to centrifugation and depth filtration; how-
ever, the clarified liquids from centrifugation and depth filtrations can be further
processed using normal flow MF if needed. Capital costs are likely to be significantly
higher for centrifugation than the other methods, while consumables costs are typi-
cally higher for depth filtration and tangential-flow MF. In addition, a number of less
established methods for the harvest stage in biopharmaceutical manufacturing are
likely to become more widely used in the coming years. These include acoustic wave
separation, polymeric flocculation, and magnetic cake filtration.
202 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

7.10 Review questions

1. Qualitatively plot ΔP versus throughput and the feed flux, Jf, versus throughput
for the following two scenarios: a depth filtration step operated at constant feed
flux and a depth filtration step operated at constant inlet pressure.

2. Pichia pastoris is used to produce an extracellular therapeutic protein. Centrifu-

gation is used as a primary clarification step for the fermentation broth and
depth filtration as a secondary step (i.e., clarified liquid from the centrifuge is
fed to the depth filter). The depth filter that provides the best performance has a
capacity of 160 L/m2 at a feed flux of 250 LMH [198].
(a) How much depth filtration area is required to process a 2,000 L fermenta-
tion? Note that the centrifuge removes most of the yeast, which is at a con-
centration of 50% by volume in the fermenter. Use a 50% safety factor in
your calculation.
(b) How many depth filtration modules are required if the largest available has
a surface area of 1.1 m2?
(c) How much time is required to perform the filtration?

3. The capacity of a depth filter used to clarify a CHO culture is significantly less at
production scale than what was measured in bench-scale studies. High pressure
drop across the production setup has resulted in a premature end to filtration,
and the feed has to be directed to a backup filtration setup. Give three possible
reasons for the less-than-expected capacity.

4. Bench-scale studies are being planned to mimic a depth filtration step used to
clarify a production CHO culture, and an estimate of the volume of feed material
that will be required is needed. In the production process, 1,000 L of CHO cul-
ture is fed to Zeta Plus™ 60SP02A depth filter capsules from 3M. Seven capsu-
les with a surface area of 1.6 m2 per capsule are used. The small-scale setup will
use a capsule of the same filter media with an area of 170 cm2. What minimum
volume of culture is required at bench scale to equal the throughput of the
large-scale operation? (Data from Thomas P. O’Brien, “Large-Scale, Single-Use
Depth Filtration Systems for Mammalian Cell Culture Clarification,” Bioprocess
International, 10(5), May 2012 [174].)

5. Data from a study to determine the filter area required for a production-scale
depth filtration step for clarification of an insect cell (Sf9) lysate to produce a
gene therapy vector is shown below.
 7.10 Review questions 203

Table 7.3: Data required to determine depth filtration area for clarification of
an insect cell lysate.

Time (min) Volume Filtered Feed Pressure Filtrate Turbidity

(mL) (psi) (NTU)

.  .

 . .

 . .

 . . .

 . .

  . 

 . 

 . . 

 . .

 . . .

 . .

 . . .

 . .

  . 

  

 . . .

 , .

 . .

 . . .

® ®
This study used a Millistak+ HC depth filter in µPod format, D0HC media se-
ries. The experimental setup is shown in Figure 7.13.
®
(a) What is the nominal pore size rating Millistak+ HC D0HC filtration media?
(b) What is the surface area of the depth filter used? An Internet search is required.
(c) What average feed flux was used?
(d) Plot the pressure (psi) versus throughput in L lysate fed/m2 filter area.
(e) The maximum pressure difference for the filtration step is 15 psi. From the
plot, estimate the filter capacity at this pressure.
204 Chapter 7 Harvest operations, Part 3: solid-liquid separations by filtration

(f) How much filter area is required to process 2,000 L of Sf9 lysate assuming
the filters are loaded to the capacity estimated previously?
(g) You purchase filter modules with a 1.1 m2 area for production. How many
modules are required if 50% more area (safety factor) is purchased than cal-
culated above?
(h) What will the process flow rate be in L/h?
(i) What time will be required to process 2,000 L of Sf9 lysate?

Process Vessel

Depth Filter
Pump
Filtrate

Figure 7.13: Experimental setup used to generate the data shown in Table 7.3. Image © NC
State University; reprinted with permission.

6. To determine the intrinsic resistance of a 0.1 µm hollow fiber MF membrane,

Rm, a water permeability test is conducted using HPW. HPW permeates a new
membrane at a rate of 200 L/h, measured using a graduated cylinder and stop-
watch, at a pressure of 5 psi using a hollow fiber module with 0.16 m2 of area.
What is the value of Rm, in m−1? (Data from reference 184.)

7. To recover an extracellular protein therapeutic produced in Pichia pastoris, a

tangential-flow MF wash (diafiltration) step was performed, using the same
0.1 µm hollow fiber MF membrane from the previous problem. If a constant
permeate flux of 55 LMH results in an average TMP of 2.5 psi, what percentage
of the resistance to permeate flow is due to the cake on the surface and inter-
nal membrane fouling (i.e., Rfc + Rif)? (Data from reference 184.)

8. The tangential-flow diafiltration step referred to in the previous problem was

executed using 2 diafiltration volumes (DVs); DV is defined as the ratio of the
volume of buffer added to the volume of product processed. The diafiltration
was performed by maintaining constant volume in the feed vessel. What % step
yield would be achieved with 2 DVs, assuming that the product, a protein thera-
peutic, passes freely across the MF membrane? Note that the fermentation broth
 7.10 Review questions 205

being diafiltered contains 50% by volume P. pastoris; therefore 2 DVs based on

total volume correspond to 4 DVs based on liquid volume. Hint: to complete this
problem, a product balance in the form of input – output = accumulation is re-
quired on the MF system. First derive the equation for the simple case in which
no wash is done.
Chapter 8
Purification operations: chromatography

At the end of the harvest stage, the active product is typically dissolved in a clarified –
that is, solids-free – aqueous solution. A sample would likely look “clean” and clear,
yet numerous soluble impurities are present that must be removed to levels low enough
to ensure no risk to a patient. Liquid chromatography is the unit operation in biophar-
maceutical processing most commonly used to remove these soluble impurities from
the product and, therefore, is critical for meeting product quality attributes related to
purity.
The topic of chromatography in biopharmaceutical processes is a big one. The pro-
cesses occurring at the molecular level to create separation are complex. There are nu-
merous types of chromatography media to choose from, numerous process parameters
that can be varied in the design of chromatography steps to optimize performance, and
a number of validation considerations required for current Good Manufacturing Prac-
tice (cGMP) processing. This chapter attempts to succinctly explore all of these topics,
by specifically addressing the following questions:
– What are the impurities that must be removed in a process to produce a drug
product safe for patients?
– How does separation of the product from impurities occur in chromatography?
– Chromatography utilizes a stationary phase that interacts with solutes in the liq-
uid feed to enable separation. What different chemistries are utilized on these
stationary phases to create separation for biopharmaceutical processes?
– In addition to resin beads used to pack chromatography beds, what other sta-
tionary phase formats are in use and how does their performance compare to a
bed packed with resin?
– What factors describe the performance of a chromatography step?
– What is a typical sequence of phases (i.e., steps) used to carry out a chromatog-
raphy run in a biopharmaceutical production process?
– What are key components of a chromatography system for cGMP biopharmaceu-
tical manufacturing and how is the quality of a packed bed assessed prior to its
use?
– What are the important process parameters that potentially impact chromatographic
performance?
– How are chromatography steps developed at bench scale and scaled up to
manufacturing?
– What is required to validate chromatography steps for a cGMP biopharmaceuti-
cal process?

https://doi.org/10.1515/9783110616880-008
 8.1 Soluble impurities 207

We also note that in biopharmaceutical manufacturing, liquid chromatography is

commonly used not only for production but also for testing samples of raw materi-
als, product intermediates, and final product, in the form of high performance liq-
uid chromatography (HPLC) and ultra-high-performance liquid chromatography
(UHPLC). While the basic modes of interaction between chromatography media and
solutes used in analytical and process chromatography are the same, there are a num-
ber of differences between the two techniques. Analytical chromatography is a bench-
scale activity, unlike most process chromatography, and is used for quantitative and/or
qualitative analysis of a sample, which requires resolving individual components as
separate chromatographic peaks. To provide necessary resolution, analytical chroma-
tography uses relatively small resin beads of 5 µm diameter or smaller. In contrast, liq-
uid chromatography steps for biopharmaceutical production are designed to produce a
product intermediate with the necessary purity and potency (rather than trying to re-
solve each individual component in the intermediate), while also providing a high step
yield, high productivity, and a low cost.

8.1 Soluble impurities

Because the purpose of chromatography is purification, we begin the chapter with a

discussion of some of the impurities of concern in biopharmaceutical processes. As
mentioned previously, biopharmaceutical process streams contain a number of sol-
uble impurities, which are usually removed for the sake of patient safety and for
efficacy of drug product. Referring back to the release specification for a monoclo-
nal antibody (mAb) presented in Table 1.5, you see that a number of tests are per-
formed to quantify impurity levels in drug product. Generally speaking, an impurity is
“any component present in the drug substance or drug product which is not the de-
sired product, a product-related substance, or excipient including buffer components”
[199]. Further, impurities may be classified as either process- or product-related[199].
Process-related impurities are impurities originating from the manufacturing process
[199]. Common process-related impurities include cell-derived impurities such as host
cell proteins (HCPs) and host cell DNA, cell culture/fermentation-derived impurities
such as antibiotics and induction agents, and downstream impurities such as leached
protein A from protein A chromatography resin. Common process-related impurities
in biopharmaceutical processes are described more fully in Table 8.1.
Product-related impurities refer to molecular variants of the product that may
form during manufacture and storage. They are often physically and chemically
similar to the product but not as safe or effective [199]. Separation of these impuri-
ties from product can be challenging due to the similarity between each. There are
numerous examples of product-related impurities. Modification of amino acid resi-
dues during production, downstream processing, and storage may take place, leading
to oxidized, reduced, and deamidated forms of protein-based products. Product may
208 Chapter 8 Purification operations: chromatography

become fragmented or aggregated as a result of processing, both of which may result

in reduced product potency and undesirable immune response, particularly in the
case of protein therapeutics [200, 201]. Further, as new biopharmaceutical modalities
are approved, different product-related impurities must be considered. For example,
viral vectors used in gene therapies have as a quality attribute the concentration (or
percentage) of viral capsids that contain the gene of interest. Those capsids without
the gene of interest or capsids that contain the incorrect DNA are product-related im-
purities and are undesirable for a number of reasons, including that they are a source
of potentially antigenic material [202]. It is important to note that the U.S. FDA has no
specific guidance on acceptable levels of product-related impurities in biopharmaceu-
tical drug products. We will not go into detail on the formation mechanisms of these
product-related impurities, as that is beyond the scope of this book and discussion on
those mechanisms can be found elsewhere [203].
A term related to the discussion on impurities is contaminant, defined as an
external component introduced into the process that is not part of the process [199].
For example, bioburden that enters a process stream from the environment is termed
a contaminant rather than an impurity based on this definition.

Table 8.1: Common process-related impurities in biopharmaceutical processes.

Impurity Description Acceptable Level

Cellular Nucleic acids, including DNA, reside in < ng/parenteral dose [] or
deoxyribonucleic any host cell and can be released. Host decrease biological activity by
acid (DNA) cell DNA is a risk due to the possibility reducing size (< base pairs) []
that it might cause tumors or transmit
viral infections [].

Endotoxin Endotoxins (literally, toxins from within) <. endotoxin units/kg body weight/
are lipopolysaccharides that reside in hr []
the outer membrane of gram-negative
bacteria like E. coli, as discussed in
Chapter . Endotoxins produce a variety
of responses in humans, including fever
and a lowering of blood pressure. Even
in processes using hosts other than
gram-negative bacteria, endotoxins can
be introduced as a contaminant and
should therefore be cleared.
 8.1 Soluble impurities 209

Table 8.1 (continued)

Impurity Description Acceptable Level

Fermentation/ A variety of components, including None reported

cell culture serum, and other additives (selection
components agents, antibiotics, induction agents,
transfection agents) that may reside in
spent media. Some of these may be
removed from the process prior to the
purification stage, but some may not.

Host cell proteins Refers to the native proteins from the Levels are typically expected to be
(HCPs) host cell that make their way into < ng HCP/mg product [],
product intermediates. They are a although the USP suggests achieving
heterogeneous group made up of levels as low as reasonably possible
hundreds and possibly even thousands [].
of different proteins, with widely
different properties []. They may
cause an unwanted immune response,
and, for those with biological activity,
there may be a risk from the direct
activity of the impurity. They are also a
risk to the process due to the ability of
some to degrade product.

Leachables As defined in Chapter , leachables are Evaluation required as part of process

compounds that migrate (i.e., leach) validation
from any product-contact material
under normal process conditions. As
the use of single-use technologies
increases, leachables have become a
bigger concern due to the prevalence of
plastics used to make single-use
components.

Virus Production sources of mammalian Level of clearance depends on viral

origin, such as CHO cells, may be a load []
source of viruses that potentially infect
humans.

Note that for the purposes of this book, salts and buffer components introduced during
the process and not intended to be a component in the final product may also be classi-
fied as process-related impurities. However, these small molecular weight components
are all typically removed during the final diafiltration step, if not earlier in the process,
and are not discussed in this chapter as impurities.
210 Chapter 8 Purification operations: chromatography

8.2 Chromatographic principles

Chromatography’s dominant use as a purification method is due largely to its ability

to separate a variety of impurities from product while ensuring that product remains
potent and safe. Chromatography systems can readily and effectively be cleaned
and sanitized, ensuring that both batch-to-batch product contamination and micro-
biological contamination are minimized.
To begin the discussion on how chromatography works, consider Figure 8.1(a). A
solution containing three colored solutes (fictitious blue, green, and red solutes) is in-
troduced (i.e., loaded) to a cylindrical column that is packed with spherical particles. In
chromatography, these particles are referred to as resins or beads, and the resulting
column is a packed bed or packed column. The packed column contains a support on
each end that keeps the resin in place but allows liquid to flow through. (Note that in
addition to resin beads, the column always contains liquid in the space between the
beads.) Once packed into a column, the beads do not move and therefore are referred
to as the stationary phase. The solution containing the blue, green, and red solutes is
loaded under conditions in which each solute has an affinity for the resin and therefore
binds to it. After the feed solution is loaded to the column, a change to the liquid flow-
ing through the packed column is made. Any liquid that flows through the packed col-
umn is referred to as mobile phase, and the mobile phase used at this step of the
procedure is chosen to weaken the interaction between the solutes and the resin so
that the column flow can push the solutes downward through the column. The greater
the affinity of a solute for the stationary phase, the longer the time it spends bound to
the stationary phase and the slower its movement through the bed. Each solute has a
different affinity for the resin and therefore moves through the column at a different
speed, which leads to separation. In the example shown in Figure 8.1(a), the blue solute
has the greatest affinity for the resin and the red solute the least. The individual compo-
nents originally mixed together in solution have been separated because they each
move through the column with a different speed. If the green component is product
and the red and blue components impurities, it is easy to see how product that is free
from impurities can be collected in the column effluent.
It is important to note that not all types of chromatography involve binding of a
mobile phase solute to the stationary phase. In particular, size exclusion chromatogra-
phy is based on differences in the size of different components, as is discussed shortly.
Regardless, the idea that different solutes travel through the packed column at different
speeds still applies.
Figure 8.1(b) shows a plot of the concentration of each component in the column
effluent (i.e., flow from the column outlet) versus time or versus the volume. The con-
centration trace results in a chromatographic peak for each solute. This type of plot is
referred to as a chromatogram, which is a graphical presentation of any precolumn or
postcolumn measurement – for example, UV absorbance, pH, or flow rate – plotted
versus cumulative volume of mobile phase that has flowed through the column. Note
 8.3 Creating separation 211

(a)
load mobile phase continuously
applied applied to the top of column

t0 t1 t2 t3

resin

column

(b)

Time or volume

Figure 8.1: Illustration of a chromatographic separation. (a) shows the movement of three solutes
through a packed bed. (b) shows the solute concentration in the column effluent vs. cumulative
volume of mobile phase fed (or time), referred to as a chromatogram. The bell-shaped curves show
the concentration trace that results for each solute, while the rectangular curves are the traces that
would result in the absence of band-broadening effects. Images © NC State University; reprinted with
permission.

also that throughout this chapter, we will continue to use the term solute or component
to refer to molecules – products or impurities – in the mobile phase that may or may
not bind or interact with the stationary phase. We use the term target solute specifically
to refer to the molecule to be recovered – usually the product – in the mobile phase.
The target solute may or may not bind to the resin, as will become clear shortly.

8.3 Creating separation: stationary phases, mobile phases,

and properties of the solutes to be separated
In this section, we explore the nature of the interaction between a mobile phase sol-
ute and the stationary phase. The degree of interaction depends on properties of the
solutes to be separated, properties of the stationary phase, and properties of the mobile
212 Chapter 8 Purification operations: chromatography

phase. Let’s start by considering properties of proteins (i.e., the solutes) that im-
pact separation, given that the majority of biopharmaceuticals are protein therapeutics,
such as insulin or Humira®, or are protein based, such as inactivated influenza virus
used in the flu vaccine or adeno associated virus used for gene therapies.

8.3.1 Protein properties

Recall from Chapter 1 that proteins are macromolecules made up of amino acids and
have a three-dimensional structure that is tied to their function. Three protein proper-
ties particularly relevant to chromatography are charge, hydrophobicity, and size.
Proteins are charged mainly because some amino acids have side chains, desig-
nated as “R” in Figure 1.1, that carry a charge, as shown in Figure 1.2. The net charge on
a protein is the sum of charges of individual amino acids. Note that the amine and car-
boxyl groups on amino acids are also charged when in a solution at neutral pH; how-
ever, with the exception of the terminal amine and carboxyl groups of the protein, these
groups form the peptide bonds with other amino acids and therefore don’t contribute to
the protein charge. The pH at which the net charge on a protein is zero is referred to as
the isoelectric point, written pI. The pH of the solution in which a protein is dissolved
determines its charge. At pH values below the pI, the net protein charge is positive due
to the relative abundance of hydronium ions (H3O+) in solution. If the solution is at a
pH value above the pI, the net protein charge is negative. The dependence of charge on
pH is shown for three common proteins – bovine serum albumin (BSA), lysozyme, and
ovalbumin – in Figure 8.2. Note that the net charge vs. pH curve is different for each of
the proteins; therefore, at a given pH, different proteins have different net charges. For
example, if BSA and lysozyme are both dissolved in a buffer at pH 7.0, BSA carries a
negative charge, while lysozyme carries a positive charge. This difference in charge can
be exploited to separate proteins by chromatography as becomes clear shortly.

Figure 8.2: Net protein charge vs. pH for ovalbumin,

lysozyme, and BSA. The charge on the y-axis is in
e0. Reprinted with permission from “Specific Ion
Effects in Solutions of Globular Proteins:
Comparison between Analytical Models and
Simulation,” by M. Boström, F.W. Tavares, D.
Bratko, B.W. Ninham, 2005, The journal of physical
chemistry.B, 109(51), 24491. Copyright (2005)
American Chemical Society [209].
 8.3 Creating separation 213

Another property that differentiates proteins is their degree of hydrophobicity,

which literally means fear of water. The term is used to describe molecules that have
portions repelled by water. Figure 1.2 shows that amino acids such as phenylalanine
and isoleucine are hydrophobic, given the nonpolar nature of their side chains. The
presence of hydrophobic amino acids like these in proteins results in hydrophobic
patches on proteins. Again, because different proteins are made up of different amino
acids, hydrophobicity varies from protein to protein. Like charge, differences in hydro-
phobicity can be exploited to separate proteins in solution by chromatography.
Finally, in addition to charge and hydrophobicity, size is a property with signifi-
cant variation among proteins, as clearly shown in Table 1.1. The molecular weight
of insulin, a small protein with only 51 amino acids, is about 5,808 Da, while the
molecular weight of a mAb, like the active ingredient in Humira®, is approximately
150,00 Da. Further, viruses, such as adeno-associated virus, are made up of protein
shells and are significantly larger than even mAbs.

8.3.2 The most common stationary phase: resins used in packed beds

We begin this discussion by focusing on the interaction of solutes with the most com-
mon chromatographic stationary phase: resin beads. Resins are typically spherical
and porous, as shown in Figure 8.3, to provide ample surface area for binding. Most
resin beads, with the exception of those used in size exclusion chromatography, are
made up of a support that is chemically modified by attaching specific ligands. This
chemical modification, which is often referred to as the bead being functionalized,
enables the interaction between soluble components and stationary phase illustrated
in Figure 8.1. Resin supports are made from a variety of materials, including natural
polymers such as agarose, synthetic polymers such as acrylate polymers, and inor-
ganic materials such as silica. Diameters are typically 30–100 um [74]. In process
chromatography for biopharmaceutical applications, the area on the outer surface of
the particle does not contribute significantly to the total area available for binding;
that is, most of the area for binding is inside the resin beads, associated with the
pores [210]. Therefore, for a solute to bind in significant amounts to the resin, the
pore must be accessible to the solute, which means the pore diameter must be larger
than the size of the solute. Pore size depends on the type of resin and can range from
30–400 nm [210]. Desirable properties of a resin include resistance to the many chem-
icals to which the resin is exposed, mechanical rigidity to withstand pressure drop, low
non-specific binding to ensure adsorption occurs mainly by the mechanism for which
the resin is designed (a large amount of non-specific binding affects the degree of
purification achievable), high capacity for protein binding, and minimal leachable
components.
214 Chapter 8 Purification operations: chromatography

Flow

hbed

Dbed

Convection

Film diffusion (external mass transfer)

Intraparticle diffusion (intraparticle mass transfer)

Figure 8.3: Steps (convection, film diffusion, intraparticle diffusion) involved in the transport of solutes
from the mobile phase to the surface of the resin pores (in a packed column) where most adsorption
occurs. Note that the dotted line around the resin particle (closest to the particle) on the right
represents the stagnant film layer through which solutes diffuse. Image © NC State University;
reprinted with permission.

As mentioned previously, the type of ligand attached to the support defines the in-
teraction mode. The most common interaction modes are:
– Ion exchange (IEC). Adsorption of a component from the mobile phase is based on
attraction of opposite charges: a negatively charged solute binds to a positively
charged resin bead and vice versa. Separation between components in the mobile
phase is based on differences in charge between components. For example, a solute
with a strong positive charge binds more strongly to a negatively charged resin
than a solute with less positive charge or one with a negative charge.
– Affinity (AC). Adsorption relies on a highly specific interaction between a target
solute and the resin, due to specific chemical groups on the solute. Think of the
interaction like a key fitting in a lock. Solutes without the specific chemical
group do not bind to the resin.
– Hydrophobic interaction (HIC). Adsorption of a solute is based on hydropho-
bicity: a hydrophobic solute binds to a hydrophobic resin. The greater the hydro-
phobicity, the stronger the interaction between the solute and resin. Addition of
certain salts is required to adsorb hydrophobic solutes to the resin. Solutes are
eluted by decreasing the concentration of these salts.
 8.3 Creating separation 215

– Reversed phase (RPC). Binding of a component to a resin is based on hydro-

phobic interaction, similar to HIC, but the binding is stronger and typically re-
quires an organic solvent for elution.
– Size exclusion (SEC). Binding does not take place; instead, separation is based on
size differences between solute molecules. Larger molecules in the mobile phase
cannot diffuse into pores within the resin particles and elute after a volume of
buffer equal to the column void volume (i.e., the volume between resin particles)
has passed through the column. Smaller molecules can access the resin pores and
are therefore held up longer in the column. SEC is different from the other modes in
that interaction of the solute with the resin is not based on affinity of a component
to the resin bead.
– Multimodal or Mixed Mode (MMC). Adsorption of a solute is based on more
than one type of interaction. A common type of MMC resin includes a ligand that
is both charged and hydrophobic, which results in both ion exchange and hy-
drophobic interaction.

A summary of these different interaction modes, including examples of common li-

gands for each, appears in Table 8.2. There are numerous suppliers of chromatogra-
phy resins, including Cytiva (formerly GE Healthcare), Bio-Rad, Tosoh Bioscience,
MilliporeSigma, Thermo Scientific, Repligen, and Pall. Given the number of compa-
nies producing resins and the fact that some companies offer multiple different resins
for a given interaction mode, there are many resins to choose from when designing a
purification step.

Table 8.2: Major chromatography techniques by interaction mode and examples of each.

Technique Description Ligand Examples Mobile Phase Conditions

that Promote Solute
Adsorption and Desorption

Ion Exchange Adsorption of a component Anion exchange (AEC) Adsorption: pH that

(IEC) (solute) to a resin is based on [, ] produces a charge on the
opposite charge. Separation quaternary amine (Q): solute that is opposite that
is based on differences in -O-CHN+(CH) of resin; low ionic strength.
charge between components. also: Desorption: pH that
diethylaminoethyl produces a solute charge
(DEAE), that is the same as the
diethylaminopropyl resin; high ionic strength.
Cation exchange
(CEC) [, ]
Sulfopropyl (SP):
-O-CHCHSO−
also: carboxymethyl
(CM), sulfonate (S)
216 Chapter 8 Purification operations: chromatography

Table 8.2 (continued)

Technique Description Ligand Examples Mobile Phase Conditions

that Promote Solute
Adsorption and Desorption

Affinity (AC) Adsorption of a component to Protein A, Protein G, Mobile phase conditions

the resin based on a specific Protein L for human leading to adsorption or
interaction (“lock and key”) IgG containing Fc desorption of solute vary
due to the presence of a region [] according to specific affinity
specific chemical group on Single-domain [VHH] resin.
the component adsorbed. antibody fragment for
adeno-associated
virus []

Hydrophobic Adsorption of a component Butyl: -O-(CH) -CH Adsorption: high ionic

Interaction (solute) to a resin is based on [] strength resulting from
(HIC) hydrophobic interaction Octyl: -O-(CH)-CH including salts like
between component and [] (NH)SO or NaSO.
resin. Separation is based on Desorption: low ionic
differences in hydrophobicity strength resulting from
between components. lowering salt concentration.

Reversed Adsorption of a component C, C, C [] Adsorption: aqueous buffer
Phase (RPC) (solute) to a resin is based on that includes an ion pairing
hydrophobic interaction agent and organic modifier.
between component and resin Desorption: nonpolar,
(stronger than HIC). organic solvent required.
Separation based on
differences in hydrophobicity
between components

Size There is no adsorption in SEC. NA No binding involved.

Exclusion Solutes smaller than the resin Separation by SEC should
(SEC) pore diffuse in; solutes that be independent of mobile
are larger do not. Separation phase composition.
is based on size difference
between solutes.

Mixed Mode Adsorption based on more Often a ligand with See IEC and HIC.
(MMC) than one type of interaction, ion exchange and
typically charge and hydrophobic
hydrophobicity chemistry
 8.3 Creating separation 217

We have already mentioned that chromatography resins are packed into a cylindri-
cal column to create a packed bed. The volume of the bed is simply the volume of
the cylinder formed by the bed (refer to Figure 8.3) and is calculated as:

D2bed
Vbed = Abed hbed = πR2bed hbed = π hbed (8:1)
4
Abed is the cross sectional area of the bed, hbed is the packed bed height, Rbed is the
packed bed radius, and Dbed is the packed bed diameter or, equivalently, the inner di-
ameter of the column. The total bed volume, Vbed, is the sum of multiple components:
the volume of void space between resin beads (interparticle pores, Vinter) and the vol-
ume taken up by the resin particles themselves. The resin particle volume can be fur-
ther divided into the open volume within the pores (intraparticle pores, Vintra) and the
volume taken up by the polymer that forms the resin particle (Vp). Therefore, Vbed can
also be written as:

Vbed = Vinter + Vintra + Vp (8:2)

The different volume contributions are more commonly expressed as porosity val-
ues, as explained and illustrated in the example that follows.

Example: Calculating the total porosity in a packed bed

The total porosity in a packed bed, ɛt is defined as the ratio of the total void volume within the
bed, (= Vintra + Vinter), to the total bed volume (ɛt = (Vintra + Vinter)/Vbed). Calculate the total bed
porosity in a packed bed with an interparticle porosity, ɛinter, of 0.4 and an intraparticle poros-
ity, ɛintra, of 0.8.
Note that ɛintra is defined as the ratio of the pore volume within a resin bead to the total vol-
ume of the resin bead, and ɛinter is the ratio of the interparticle void volume to the total bed
volume (= Vinter/Vbed).

Solution
First, derive an expression for ɛt as a function of ɛinter and ɛintra, then substitute in the values to
calculate ɛt. Note that the volume of resin, Vresin, is the volume of resin particles – including
both porous and solid parts – in the bed:

Vintra + Vinter ðεintra Vresin + εinter Vbed Þ

εt = =
Vbed Vbed
εintra ð1 − εinter ÞVbed + εinter Vbed
=
Vbed
= εintra ð1 − εinter Þ + εinter (8:3)
= 0.8ð1 − 0.4Þ + 0.4 = 0.88

So, based on the porosity values given, 88% of the packed column is voids. Note that typical
values for ɛinter range from 0.3–0.4, while values for ɛintra are as high as 0.9 [210].
218 Chapter 8 Purification operations: chromatography

Because most of a resin’s surface area is associated with the resin pores, the majority
of ligands for binding are attached at the pore surface rather than the outer surface of
the resin bead. Therefore, for a solute to bind, it must be transported from the mobile
phase into the pore where it can be adsorbed by the resin. Likewise, the solute must
also be desorbed and make its way back to the mobile phase flow to be recovered.
The transport steps involved are illustrated in Figure 8.3. A solute is transported
through the porous packed bed by the mobile phase flow, usually created by a pump
and therefore referred to as convection. As the mobile phase moves through the
packed bed, a binding solute diffuses from the bulk mobile phase through a stagnant
film surrounding the particle surface in a process commonly referred to as external
mass transfer. (The term diffusion refers to the random movement of a component
from an area of high concentration to low concentration.) From the particle outer sur-
face, the solute then diffuses into the liquid-filled pores of the resin in a process re-
ferred to as intraparticle mass transfer (or intraparticle diffusion). It then adsorbs on
the pore surface as a result of the interaction with a ligand attached to the resin. In
chromatography steps for biopharmaceutical processing, intraparticle diffusion is typi-
cally the transport step that requires the most time [210], and we say that intraparticle
diffusion provides the greatest resistance to mass transfer.

8.3.3 Other stationary phase formats: membrane adsorbers and monoliths

In addition to resin beads packed in a column, membrane adsorbers and monoliths

are also common stationary phase options for biopharmaceutical manufacturing.
Membrane adsorbers are porous membranes, like the microfiltration membranes dis-
cussed in Chapter 7, whose surfaces have been functionalized by attaching ligands to
provide interaction chemistry, much like resins. In fact, many of the same chemistries
used for resin bead ligands are also used in the membrane adsorber and monolith
formats. Many membrane adsorbers are housed in a cylindrical (annular) module ge-
ometry, in which multiple layers of membranes are wrapped around a central core.
Monoliths are similar, but instead of multiple layers of membrane, a monolith con-
sists of a continuous porous stationary phase made up of a network of pores through
which fluid flows [210]. Both formats are most often made from polymers such as re-
inforced cellulose or modified hydrophilic polyethersulfone (for membrane adsorb-
ers) or polymethacrylate (for monoliths).
Membranes and monoliths are used in much the same way as packed chromatog-
raphy columns. The process intermediate to be purified is fed to the membrane or
monolith module, and product and/or impurities bind. In the case of a cylindrical ge-
ometry, it is common that feed enters the module, flows through an outer channel,
then moves radially inward through the pores of the membrane layers or monolith to
an inner channel and out of the module at the end opposite the feed. Recently, mem-
brane adsorbers in a cassette geometry have also been introduced [215].
 8.3 Creating separation 219

There is an important difference between solute transport through a bed packed

with resin beads and a membrane adsorber or monolith, as illustrated in Figure 8.4:
ligands are exposed to the mobile phase flowing through the membrane or monolith
pores. Pore access is by convection, unlike resin beads in which solute exposure to the
ligand typically requires intraparticle diffusion. Convective flow through pores of a
membrane or monolith results from their relatively large size and their interconnected-
ness relative to pores in resin beads. Pore sizes from 0.2 µm to greater than 3 µm are
common [215–217]. These differences in pore size and geometry lead to the following
important differences in performance between the resin and membrane adsorber/
monolith formats [218].
– Binding capacity (the amount of a component that binds per unit volume of sta-
tionary phase, as will be discussed shortly) and resolution of the chromatographic
peaks are independent of flow rate, allowing higher flow rates to be used. In
packed beds, both binding capacity and resolution typically decrease with in-
creasing flow rate. We will come back to this point shortly.
– Membrane and monolith adsorbers have lower pressure drops than packed beds
for a given volumetric flow rate.
– Membrane and monolith adsorbers are particularly good for binding large biomole-
cules, including larger proteins and viruses, because diffusion into small pores is
not required; however, binding capacity for “normal sized” (and smaller) biomole-
cules on membrane adsorbers or monolithic media is typically higher in packed col-
umns due to the higher surface area for binding per unit volume of column.

Convection

Film diffusion

Figure 8.4: Transport of solutes from the mobile phase to the surface of a membrane adsorber or
monolith pore where adsorption occurs. Note that intraparticle diffusion, which limits the rate at
which solutes bind to resin beads in packed beds, is not present in membrane adsorbers and
monolith devices. Image © NC State University; reprinted with permission.
220 Chapter 8 Purification operations: chromatography

Membrane adsorbers are offered by a number of vendors, including Pall, Sartorius,

MilliporeSigma, and Cytiva. BIA Separations (now part of Sartorius) is a major sup-
plier of monoliths.

8.3.4 Mobile phase properties and their impact on separation

To complete our discussion of how and why mobile phase components interact with
chromatographic stationary phases, let’s turn our attention to the impact of the mobile
phase. In Table 8.2, we have summarized key points related to mobile phase properties
required for solute binding and desorption with each interaction mode and briefly elab-
orate here.
A common feature of chromatography steps in biopharmaceutical processes is
that products are typically applied to chromatography stationary phases in a buff-
ered solution, regardless of interaction mode. pH fluctuations can occur for a num-
ber of reasons, and surrounding the product with a buffer during chromatography
(and other processing steps) minimizes these and helps ensure product stability.
In ion exchange, because binding is based on opposite charge, the solute to be
bound must be at a pH that creates this opposite charge. So, for AEC, the solute to be
bound must be in a buffered solution with a pH > pI of the binding solute during load-
ing. For CEC, the buffered solution must be at a pH < pI of the binding solute during
loading. Therefore, design of an ion exchange step requires that product be in a solu-
tion with buffering capacity at the specific pH necessary for binding. Further, the
concentration of the buffering agent is usually kept low (10–50 mM) to minimize con-
ductivity, so buffer ions are not competing for binding with the target solute. Because
most buffering agents have a limited useful pH range, careful selection of a buffering
agent to match the needed load pH is required. To help make this selection, tables are
available that provide information on the useful pH range for a variety of buffering
agents [211].
For AC, there are a variety of different ligands in use, as shown in Table 8.2, and
the mobile phase requirements depend on the nature of binding between the affinity
ligand and mobile phase solute. Commonly used protein A resins (i.e., resins that have
protein A attached as a ligand) have a high affinity for most immunoglobulin G anti-
bodies, which include mAbs used as biopharmaceuticals. Protein A binds with the anti-
body at its Fc region at neutral pH. The antibody is eluted by lowering the pH [219].
From Table 8.2, the mobile phase for binding a solute to a HIC resin should
have a high conductivity, and in particular should include kosmotropic salts such
as ammonium sulfate ((NH4)2SO4) and sodium sulfate (Na2SO4). Removal of those
salts – and therefore a lowering of conductivity – favors desorbing conditions. Why
are these salts required? When hydrophobic solutes and ligands are in aqueous so-
lution, the water molecules cannot wet the surface of the hydrophobic substance.
 8.4 Chromatography performance 221

Instead, water molecules form a highly ordered layer around the hydrophobic mobile
phase solute and ligand. For a hydrophobic solute to bind to a HIC resin, this struc-
tured water layer must be broken. Kosmotropic salts such as (NH4)2SO4 are able to do
this, which allows for binding to occur. More in-depth discussion on the nature of
protein binding on HIC resins is offered elsewhere [220].

Example: Separation by ion exchange chromatography

Consider the separation of the following proteins from a mixture: ribonuclease (pI = 9.5), cyto-
chrome C (pI = 10.3), and lysozyme (pI = 11). The proteins are dissolved in a solution of 50 mM
MES buffer at pH 6.0. (Note that MES is the common name for 2-(N-morpholino) ethanesulfonic
acid, a good buffering agent at pH 6.0). Answer the following:
(a) What type of ion exchange chromatography should be used for the separation – anion or
cation?
(b) The concentration of NaCl in the MES buffer is slowly increased to elute the proteins from
the cation exchange resin. In what order will the proteins elute from the resin?

Solutions
(a) At pH 6.0, all three proteins have a net positive charge. None will bind to an anion ex-
change resin, therefore separation by AEC is not possible. All will bind to a cation exchange
resin for separation, and because each protein has a different pI, each binds with a differ-
ent strength, which can be exploited for separation. So use CEC for the separation.
(b) Proteins with a lesser positive charge elute before proteins with a greater positive charge.
Further, the farther away the pI for a given protein is from the surrounding pH, the more
positively charged the protein. Therefore, the order of elution is ribonuclease, cytochrome
C, then lysozyme. Referring back to Figure 8.1., ribonuclease would be the red component,
cytochrome C the green, and lysozyme the blue (even though these proteins are not really
colored).

8.4 Chromatography performance

In this section, we turn our attention to chromatography performance. Note that

many concepts covered in the section are described in terms of packed beds but can
be applied to membrane adsorbers and monoliths as well. Because column perfor-
mance is often related to mobile phase flow rate, we begin by discussing velocity
and residence time – both related to flow rate – in a column. In chromatography at
process scale, flow rate typically refers to volumetric flow rate, Q, through a packed
bed in units of L/h. However, we also refer to the superficial velocity, vs, of the mo-
bile phase in units of cm/h. The superficial velocity represents the velocity of a mo-
bile phase fluid element if no resin is present in the column and is calculated as the
ratio of the volumetric flow rate to cross sectional area of the bed, Abed:

Q
vs = (8:4)
Abed
222 Chapter 8 Purification operations: chromatography

For a cylindrical column, Abed is calculated as π × Rbed2 or equivalently π × Dbed2/4,

as shown in equation (8.1). The actual velocity of a fluid element in the mobile
phase is higher than that calculated by equation (8.4) because a significant amount
of the cross sectional area in a column is occupied by resin particles and therefore
not available for liquid flow. If the interparticle porosity is known, then the actual
velocity, vactual, can be calculated as:

Q
vactual = (8:5)
εinter Abed

Related to fluid velocity is the residence time of a mobile phase fluid element in the
packed column. The residence time is calculated by taking the distance that a fluid
element must flow – the height of the packed bed – and dividing by its velocity
through the column:

hbed
tresidence = (8:6)
vactual

In practice, the superficial velocity is often used when calculating tresidence because
the value of ɛinter is not always known:

hbed Vbed
tresidence = = (8:7)
vs Q

Calculating tresidence by equation (8.7) is common, but it should be recognized that

the residence time calculated is that required for a fluid element to travel through
an empty column. It is also worth noting the residence time, defined by either of the
two previous equations, is different from the time required for a binding solute to
travel through a column. That time is more commonly referred to as the solute’s re-
tention time, tr. For example, the blue, green, and red solutes in Figure 8.1 are mov-
ing more slowly than the mobile phase because they interact with the stationary
phase.

Example: Calculating the mobile phase residence time in a packed column

A chromatography step is optimized at bench scale using a 1.5 cm diameter column packed to a
bed height of 25 cm and operated at a flow rate of 15 mL/min. What is the residence time of the
liquid phase in minutes?

Solution
To use equation (8.7), first calculate the superficial velocity, vs, using equation (8.4) and recog-
nize that 15 mL/min is the same as 15 cm3/min:

Q 15 cm3 =min
vs = =
Abed π × ð1.5 cm=2Þ2

= 8.4883 cm=min
 8.4 Chromatography performance 223

Now calculate the residence time using equation (8.7):

hbed 25 cm
tresidence = =
vs 8.4883 cm=min
= 2.9 min
Thus, a fluid element in the mobile phase moving at a volumetric flow rate of 15 mL/min through the
packed column would spend 2.9 min in the column if the column is not packed. The true residence
time is less, given that the volume available for mobile phase flow is less than then total bed volume.

You may have asked yourself previously why the bands in Figure 8.1 that represent
each solute in the chromatography column broaden instead of maintaining their origi-
nal rectangular shape (the shape that represents the load). Solutes moving through a
packed column get dispersed. It is desirable to minimize this dispersion as it can lead
to loss of resolution between solutes; that is, the concentration traces shown in Figure
8.1(b) begin to overlap as the dispersion occurs, which eventually makes it difficult to
collect the product in a pure form, thus compromising purity. The height equivalent
of a theoretical plate (HETP) is a theoretical concept used to describe this broadening
of solute bands in a column. It is based on the idea that a packed column contains
fictitious plates that allow for equilibration between the mobile and stationary phases.
Statistically, the number of plates in the column, Nplates, can be defined from the con-
centration trace for a single band eluting from the column as [221]:

Vr2
Nplates = (8:8)
σ2v
where Vr is the retention volume of the peak (refer to Figure 8.8) and σv2 is the
peak variance in volume units. The greater the peak variance, the broader the chro-
matographic peak, and the more likely it is that the solute band represented by the
peak will overlap with other bands and compromise separation. Thus, Nplates, serves
as a measure of the efficiency of the packed bed.
HETP is simply the height of a single plate and is given by the equation

hbed σ2
HETP = = hbed v2 (8:9)
Nplates Vr

HETP is often expressed in units of cm. Given that peak variance is a measure of the
width of a chromatographic peak, HETP is a direct measure of band broadening; that
is, the wider the solute band in a column, the larger the HETP. However, equation (8.9)
provides no insight as to the processes occurring within a column that lead to the band
broadening. To gain insight into those processes, consider the well-known van Deemter
equation,[222] given by the following equation:
B
HETP = A + + Cvactual (8:10)
vactual
224 Chapter 8 Purification operations: chromatography

A, B, and C are constants related to hydrodynamic dispersion, axial diffusion of sol-

ute, and transport of solute from the mobile phase to the stationary phase, respec-
tively. Each of these processes is described in greater detail below.
– Hydrodynamic (or eddy) dispersion (A). As illustrated in Figure 8.3,multiple
paths are available for mobile phase to travel through a chromatography col-
umn. Some liquid elements take a longer, more winding path, while other liquid
elements take a shorter path. Therefore, solute molecules of a specific type (e.g.,
product molecules) in the mobile phase may each take a different path through
the interparticle voids in the column, which results in broadening of the solute
band.
– Axial molecular diffusion (B/vactual). Solutes diffuse axially in the mobile phase
– that is, up and down the length of the packed bed, away from the center of sol-
ute concentration in the band. The result of this movement is broadening of the
solute band as the chromatography step proceeds. Axial diffusion is typically not
significant in liquid chromatography applications because diffusion of solutes in
liquids is relatively slow. Generally, the longer the solute bind resides in the col-
umn, the greater the impact of axial diffusion on band broadening. Thus, at low
fluid velocities, the B/vactual term becomes larger and the HETP increases (i.e.,
band broadening increases).
– Mass transfer of solute to the stationary phase (Cvactual). Among the three
contributions to HETP, the impact of solute transfer from the mobile phase to
the resin surface may be the most challenging to understand, but in process-
scale liquid chromatography, it is the most important. As a band moves through
the column, time is required for solute in the mobile phase to equilibrate with
the stationary phase due to the time required for transport of solute into the
resin pores for binding. At any axial position within the column, as the solute in
the mobile phase moves toward equilibrium with the stationary phase, some un-
bound solute is not be adsorbed to the stationary phase until it moves further
down the column. As the solute band passes the stationary phase, time is also
required for solute to be desorbed from the resin and transported back to the
mobile phase. These lags in solute transport into and out of the resin pore result
in broadening of the solute band. The impact of mass transfer resistance on
band broadening is proportional to velocity, because as the rate at which the
mobile phase moves through the column increases, the likelihood that solute in
the mobile phase is out of equilibrium with solute on the stationary phase in-
creases. Note that the time required for a solute to adsorb once it has diffused to
the resin surface is usually fast relative to the time required for mass transport to
the resin surface [210] and is not a significant source of band broadening.

From equation (8.10), it is clear that at low mobile phase velocities, the second term
in the equation becomes large, and the third term becomes negligible, resulting in a
high HETP. Likewise, at high velocities, the second term in the equation goes to
 8.4 Chromatography performance 225

zero, while the third term becomes large, resulting in a high HETP. Thus, there is an
optimal mobile phase velocity that results in a minimum (i.e., desirable) HETP value.
Minimizing HETP leads to sharper chromatographic peaks, which result in good reso-
lution between solutes in the feed and enhanced purity of product. We’ll come back
to the concept of HETP shortly when we discuss measurements to test that a column
has been packed correctly.
Related to HETP is resolution, a term that we have used several times already, but
have yet to specifically define. In chromatography, the term resolution is a measure of
the separation between two components and is specifically defined as the distance be-
tween the center of two neighboring peaks (in time or volume) divided by the average
of the peak widths (in time or volume). Poor resolution between a biopharmaceutical
product and impurities results in impure product eluting from a chromatography col-
umn. Both the number of plates in a column and the selectivity of the stationary phase
impact the resolution between two components to be separated. Selectivity is a mea-
sure of how much time one solute (e.g., product) spends on the stationary phase rela-
tive to another (e.g., impurities). The greater the selectivity, the better the resolution.
We mention it here because selectivity is often used to screen chromatography resins
for use in biopharmaceutical processes, as is discussed shortly.
Another important performance parameter for chromatography steps is binding
capacity. Generally, the term refers to the amount of a molecule that can be adsorbed
by a stationary phase under a specific set of conditions and is typically expressed as
the mg of solute (e.g., protein) bound per mL of resin. It is desirable to maximize the
binding capacity of the target component(s) to reduce the amount of stationary phase
required to perform the separation. There are two types of binding capacities that are
measured: the equilibrium binding capacity (EBC) (also referred to as total binding
capacity or static binding capacity) and the dynamic binding capacity (DBC). The EBC
is the total amount of a molecule that the resin is capable of binding under conditions
of no flow. Imagine that resin is added to a beaker filled with a solution that contains
a solute that binds to the resin. The resin is mixed with the solution long enough to
ensure that equilibrium between the target solute and resin is achieved. The mass of
solute adsorbed (e.g., mg solute) per unit volume of resin (e.g., mL resin) is the EBC
at the concentration of unbound solute in the liquid phase. A plot of EBC versus the
concentration of solute (i.e., unbound solute) with which the stationary phase is in
equilibrium at a given set of conditions (e.g., at fixed temperature, pH, or ionic
strength) is referred to as an adsorption isotherm.
The DBC for a specific solute is the amount of that solute that is adsorbed onto
the stationary phase while operating at a specific flowrate. To measure the DBC, a
solution that contains the solute of interest (e.g., product) at a known concentration
is fed to a chromatography column at the desired flow rate. When the concentration
of product in the column effluent reaches some specified percentage of the product
concentration in the column feed, we say that breakthrough has occurred. For that
226 Chapter 8 Purification operations: chromatography

reason, this type of test is often referred to as a breakthrough study. It is common to

use 10% of the target solute concentration in the feed as the acceptable concentra-
tion in the column effluent, which leads to the notation DBC10%. Note that the con-
cept of DBC can be applied to membrane adsorbers and monoliths as well, even
though the previous description referred to “columns.”
Figure 8.5 illustrates the DBC concept more fully. The data in that figure were
generated by feeding BSA in 50 mM tris, pH 8.0 to a 3 mL anion exchange mem-
brane capsule (Figure 8.5(a)) and 3 mL packed column packed with anion exchange
resin (Figure 8.5(b)) until the concentration of BSA in the column effluent became
constant – that is, until the stationary phase became saturated with BSA. To calculate
the DBC10% for the target solute, in this case BSA, the mass adsorbed, mads, is approx-
imated by the total mass of target solute fed to the point of 10% breakthrough and
divided by the bed volume to give:

mads,10% Cf Vloaded,10%
B10% = = (8:11)
Vbed Vbed

where Cf is the concentration of the target solute (BSA) in the column feed and V-
loaded,10% is the volume of solution fed to the column up to the point of 10% break-
through. The equation is only an approximation of the mass adsorbed because it
includes not only solute that binds to the stationary phase but also solute in the col-
umn effluent (i.e., solute that did not bind because binding sites in the stationary
phase are not available), up to the point of 10% breakthrough. Looking at Figure 8.5
(b), we estimate the DBC10% for BSA in the packed bed at a flow rate of 1.28 mL/min
to be 67 mg BSA/mL Q Sepharose FF resin (= 1 mg/mL × 200 mL)/3 mL). Likewise, at
3.83 mL/min, the DBC10% from Figure 8.5(b) is estimated as 37 mg/mL.
Comparison of Figures 8.5(a) and 8.5(b) also lead to the following observations
regarding performance of packed beds compared to membrane adsorbers:
– Breakthrough curves for each flow rate tested with the packed bed do not over-
lap; curves at low feed flow rates are much steeper than the curves at the higher
flow rates. Therefore, Vloaded,10% increases with decreasing flow rate, and, conse-
quently DBC10% for BSA in the packed bed increases with decreasing flow rate.
This relationship between DBC10% and feed flow rate is a direct consequence of
slow intraparticle diffusion: slower feed rates allow more time for intraparticle
diffusion to take place, and therefore greater binding of BSA to binding sites on
the resin occurs before BSA shows up in the column effluent.
– Breakthrough curves for each flow rate tested with the membrane capsule are
nearly identical and result in a DBC10% of approximately 21 mg BSA/mL membrane
capsule at each flow rate tested. The fact that DBC10% does not depend on flow rate
– in stark contrast to the DBC10% values measured for the packed bed – is a result
of the rapid transport of BSA to binding sites on the membrane (i.e., slow intrapar-
ticle diffusion does not exist within the membrane). The independence of the slope
of the breakthrough curve with flow rate is also predicted by the van Deemter
 8.4 Chromatography performance 227

(a) (b)

Figure 8.5: Breakthrough curves for adsorption of BSA at 1 mg/mL in 50 mM tris, pH 8.0 on (a) a
Sartobind® Q nano membrane capsule with a 3 mL volume and (b) a Q Sepharose® FF packed
column with 3 mL volume. Note that measurements for both the membrane capsule and packed
bed were performed at equal superficial velocity values. Note also that for these graphs, the
notation is slightly different that defined previously in this chapter. c is the concentration of BSA in
the column effluent, co is the concentration of BSA in the column feed, and V is the volume of BSA
solution fed in mL. Reprinted from “Direct comparison between membrane adsorber and packed
column chromatography performance,” by C. Boi, A. Malavasi, R. Carbonell, G. Gilleskie, 2020,
Journal of Chromatography A, 1612, 6. Copyright 2019 Elsevier B.V [223].

equation (equation (8.10)), as the Cv term for the membrane adsorber goes to zero
due to rapid mass transfer of solute to the stationary phase, and the B/v term is
negligible due to the fact that diffusion in liquids is slow. This leaves the HETP to
be a fixed value, independent of flow rate.
– Binding capacity values are higher for the packed bed than the membrane cap-
sule, likely a result of the greater surface area per unit volume in the packed bed.

DBC has practical implications for sizing of chromatography columns, as the capac-
ity of stationary phase for a solute dictates the amount of stationary phase required
to bind a solute (e.g., product) without product ending up in waste. The example
below illustrates the relationships between DBC and column sizing.

Example: Estimating column size (diameter) from the DBC

10 kg of a protein therapeutic is produced by a 2,000 L cell culture step. Assuming no loss of
product in the harvest steps, calculate the diameter of the column used for the first chromatog-
raphy step if:
– the column is loaded to 70% of its DBC10% (note that this amount of loading is significantly
greater than what is illustrated in Figure 8.1).
– the DBC10% at the flow rate used for this step has been measured as 50 g product/L resin.
– the bed is packed to a height of 25 cm.
228 Chapter 8 Purification operations: chromatography

Solution
The packed bed volume required to capture all product is calculated as the ratio of the mass of
product loaded, mload, to the product load, defined as the amount of product loaded per unit
volume of bed:

mload
Vbed = (8.12)
product load
10, 000 g product
=
0.7 × 50 g product=L resin

= 286 L of resin or 285, 714 cm3 of resin

From equation (8.1):

D2bed
Vbed = π h
4 bed

1=2
1=2
4Vbed 4 × 285, 714 cm3
Dbed = =
πhbed π × 25 cm
≈ 120 cm

Note that DBC and EBC are related, but DBC is impacted by the band broadening fac-
tors resulting from flow through a packed bed (or membrane adsorber or monolith). As
flow rate becomes smaller in a breakthrough study, the DBC should approach the EBC,
all other conditions (e.g., temperature, pH, conductivity) being the same. The amount
of solute that binds to a stationary phase measured either under static or flow condi-
tions depends on solute properties (e.g., solute charge in ion exchange), mobile phase
properties (concentration of binding solute, pH, ionic strength), and stationary phase
properties (pore size in resins, porosity of stationary phase, ligand, etc.). For example,
in IEC, if the mobile phase contains a high concentration of ions then the DBC for the
target solute on ion exchange media is reduced. A low concentration of certain salts,
such as NH4SO4, in the mobile phase decreases the binding capacity of target solute on
HIC media. We have only scratched the surface on the topic of solute binding to reins.
This topic, and in particular the relationship between the concentration of adsorbed
solute on the stationary phase in equilibrium with the solute in the mobile phase as it
applies to liquid chromatography is covered in greater detail in other references [210].
In addition to HETP, resolution, selectivity and binding capacity, other meas-
ures of chromatography performance include step yield and productivity. Building
on the definition of percent yield provided in equation (2.1), the percentage step
yield in a chromatography step can be calculated as the amount of product at the
end of the step divided by the amount of product loaded to the column converted to
a percentage:
Celuate Veluate
% step yield = × 100 (8:13)
Cload Vload
 8.5 Operational modes 229

where Celuate and Cload are the concentration of product in the eluate from the col-
umn and load to the column, respectively, and Veluate and Vload are the volumes of
the eluate from and load to the chromatography step. Productivity, P, of a chroma-
tography step is defined as the amount of product collected per bed volume per
time, and is given by the following equation:

Celuate Veluate
P= (8:14)
Vbed tprocess

where tprocess is the time for the chromatography step. Clearly, maximizing produc-
tivity is desirable.

8.5 Operational modes

The most common operational modes for chromatography in biopharmaceutical puri-

fication are bind-and-elute and flow-through. Bind-and-elute mode is designed and
operated so that product binds to the stationary phase during the load step. Impuri-
ties may or may not bind. The product is recovered during the elution step by making
a change in mobile-phase (i.e., elution buffer) composition that disrupts interaction
between the stationary and mobile phases. The mobile phase composition is either
changed in a step or in a stepwise fashion and referred to as a step elution, or changed
linearly with time and referred to as a linear gradient. These different elution methods
will be discussed in greater detail shortly. Note that the mobile phase requirements
that allow for release of components from the resin have been previously discussed
(refer to Table 8.2).
Flow-through chromatography is designed and operated so that product is not
bound to the stationary phase, but impurities are. Loading of the product solution
continues until impurities in the effluent from the column (or membrane adsorber or
monolith) exceed an acceptable limit. A common flow-through step is the use of AEC
as the final chromatography step in a process for mAb production, as shown in Figure
2.3, often performed with a membrane adsorber. mAbs have pI values ranging from
approximately 6–9 [224]. If the pH of the buffer in which the mAb product is dissolved
is less than the pI, then the mAb is positively charged and flows through the anion
exchange bed or membrane, while impurities that are negatively charged at even
these lower pH values, such as endotoxins, some HCPs, and DNA, bind to the chroma-
tography media [225]. Membrane adsorbers are attractive for flow-through applications
meant for polishing – that is, steps that remove low levels of remaining impurities –
because their relatively low binding capacities are suitable for binding the residual im-
purities that remain, and the membrane systems can be operated at higher flowrates
than packed beds.
230 Chapter 8 Purification operations: chromatography

8.6 Typical chromatography procedures for cGMP manufacturing

Once a chromatography column is properly packed, cleaned and sanitized, it is ready

for use. Every chromatography step, including steps using membrane adsorbers and
monoliths, in a downstream bioprocess is actually made up of a series of steps – that we
refer to as phases for clarity – necessary to ensure that purification objectives are met
and that the packed column can be reused. A typical sequence of phases for a batch
bind-and-elute chromatography step, regardless of interaction mode, is: equilibration,
product load, wash, elution, cleaning/regeneration, sanitization, buffer rinse, HETP/
asymmetry measurement, and storage. Each of these phases is described in Table 8.3.
Note that the sequence of phases in Table 8.3 can be adapted to SEC by removing the
wash step. The specific design of each phase varies, of course, according to the product
and specific resin used. And there are a number of variations on the typical sequence of
phases that are used. For example, cleaning and sanitization may be done with the
same solution; therefore the cleaning and sanitization phases become one. Further, mul-
tiple wash steps may be performed to maximize removal of all weakly bound impurities
from a column prior to elution.

Table 8.3: Phases in a typical bind-and-elute chromatography step.

Phase Description

() Equilibration – A buffer fed to the column removes storage solution from the previous
column run and prepares the column for binding product. For example, in
IEC, equilibration ensures that resin charge groups are in equilibrium
with counter ions of binding buffer and that column pH and ionic
strength are appropriate for product binding. The equilibration buffer is
often the same buffer that the protein loaded to the column (in the next
↓ step) is dissolved in.
– Buffer flows through the column until the pH and/or conductivity of the
effluent leaving the column equals that of the solution fed to the column.
Typically, 3–5 CVs of buffer is sufficient.
() Load – Product is applied to the column so that the product binds. Other
components (i.e., impurities) may or may not bind to the stationary
phase.
↓ – Overloading beyond the binding capacity of the column is avoided to
ensure no loss of product to drain.
() Wash – Buffer, often the same as equilibration buffer, is fed to the column to
wash away any unbound impurities, i.e., impurities in the voids between
resin particles or weakly bound impurities. The product remains bound.
↓ – Typically 3–5 CVs of wash buffer is sufficient.
 8.6 Typical chromatography procedures for cGMP manufacturing 231

Table 8.3 (continued)

Phase Description

() Elution – A change in mobile phase composition is made that weakens interaction
between the stationary phase and target solute (e.g., product). Elution is
designed and executed in such a way that the target component (product) is
↓ collected separately from bound impurities. Elution conditions for each
chromatographic interaction mode are shown in Table 8.2.
– Elution can be executed making a step change in mobile phase
composition (most common) or changing the mobile phase composition
linearly.
() Cleaning/ – Product is no longer bound to the column, but other strongly bound
regeneration impurities may be. A solution is fed to the column that is capable of
removing these impurities. For example, in ion exchange, a relatively
high concentration of NaCl solution is fed so that Na+ or Cl- ions displace
↓ bound components. The solution used varies according to type of
chromatography used.
– Typically 3–5 CVs of cleaning solution is adequate.
() Sanitization – A solution is fed to the column to remove any microbiological
contamination without degrading the chromatography resin. A common
sanitizing agent is 0.5 M NaOH solution; however, not all resins – most
↓ notably some protein A affinity resins – are compatible with these high
concentrations of NaOH. Sanitization and cleaning phases may be
combined into a single phase.
– Typically 3–5 CVs of sanitization solution is adequate. In addition, a
static hold of the column in sanitizing solution may be performed.
() Buffer rinse A buffer – often that used for equilibration – is used to quickly bring the pH
↓ to the desired value if an acid or caustic cleaning/sanitization agent is used.

() HETP/ Some biopharmaceutical manufacturers may choose to measure HETP/

asymmetry asymmetry after every use of a column to assess bed stability over the
↓ lifetime of the column. More information on this measurement is given in the
next section.

() Storage – A storage solution that inhibits growth of bioburden is fed to the column.
Solutions of ethanol in water are often used.
– Typically 3–5 CVs of storage solution is adequate.

The volume of solution fed to the column at each phase is typically described using the
scale-independent parameter “column volumes,” often referred to as “CVs.” As the name
implies, the number of CVs of a solution corresponds to the volume of solution fed rela-
tive to the volume of the column, calculated as:

Vmobile phase fed

CVs = (8:15)
Vbed
232 Chapter 8 Purification operations: chromatography

where Vmobile phase fed is the volume of a solution fed at any phase in the chromatog-
raphy step and Vbed is the volume of packed column, membrane adsorber, or mono-
lith. For example, 600 L of equilibration buffer fed to a column packed to a volume
of 200 L corresponds to 3 CVs of equilibration buffer.
Also, as shown in Table 8.3, elution can be carried out in a stepwise fashion or
using a linear gradient, each referring to a different method to make the change in
mobile phase composition required to elute bound solutes. For example, in HIC, the
concentration of (NH4)2SO4, which is required for binding, must be reduced to elute
product from the column. If the concentration of (NH4)2SO4 is decreased slowly, with
a constant slope over time, we refer to this type of elution as a linear gradient. This
gradient is typically described as going from a starting concentration of (NH4)2SO4 to
an ending concentration of (NH4)2SO4 over the applicable number of CVs, which is
the total volume of elution buffer fed during the gradient. In a step elution, a single
step change in the buffer composition is made (e.g., reducing the (NH4)2SO4 concen-
tration from 2 M to 0 M at the beginning of the elution step). Generally, linear gra-
dients provide better resolving capability between product and impurities but may be
more difficult to control in a production environment than a step elution. An example
of a linear NaCl gradient for an AEC step is shown in Figure 8.7 and can be identified
by the increase in conductivity during the elution phase.
Chromatography resins are typically reused and dedicated to a single product.
Membrane adsorbers and monoliths may also be reused. Consequently, after storage,
the sequence of phases can be repeated as many times for each new batch of a prod-
uct as the column lifetime allows. The column lifetime must be determined as part of
process validation. It is also worth mentioning that instead of the full amount of prod-
uct from the previous process step being loaded to the chromatography column, the
batch from the previous step may be split into sub-batches that are fed individually.
In so-called intrabatch cycling, the equilibration phase would begin the cycle for
each sub-batch. But the sequence of phases would likely proceed only to the sanitiza-
tion step and then return to equilibration for the next cycle. This pattern continues
until all sub-batches have been purified. Upon completion of the final sub-batch, the
buffer rinse, HETP/asymmetry, and storage phases are completed, and the column is
held until the next batch is ready for processing. Because the amount of product
loaded in a sub-batch is less than in a full batch, less resin is needed per batch and a
smaller, less expensive column is used. However, a smaller column that sees more
cycles for a given batch of product may use up its lifetime more quickly than a larger
column that processes the entire batch in a single cycle. In addition, multiple cycles
per batch increase process time relative to single-cycle operation. However, in some
scenarios, intrabatch cycling may lead to better utilization of resin lifetime. For exam-
ple, if producing product for clinical trials, the total number of runs required to pro-
duce the necessary amount of product is likely small and resin is likely to be disposed
after the production campaign given that it may be years before the clinical trial is
 8.7 Chromatography equipment for biopharmaceutical production 233

complete and more product is needed. In this case, cycling a column may lead to better
utilization of resin lifetime than would be achieved otherwise.

8.7 Chromatography equipment for biopharmaceutical

production
Let’s turn our focus to equipment required to conduct chromatographic separations.
Consider the schematic of a typical chromatography system for biopharmaceutical
manufacturing in Figure 8.6(a).
Moving from left to right in Figure 8.6(a), there are vessels, connected to the
chromatography skid, that contain the different solutions (including the product) re-
quired to execute the multiple phases of a chromatography run. Referring back to Fig-
ure 2.5, you see an example of one type of vessel – buffer totes – lined up for use in a
chromatography run. Stainless steel portable vessels may also be used. The skid is
equipped with at least one pump to move fluid to the chromatography column, and
often there are two. The second pump is necessary to create step and linear gradients.
The pumps on the AKTAprocess system in Figure 8.6(b) are diaphragm pumps, which
move fluid by the back-and-forth motion of a diaphragm. The skid is also equipped
with an air trap, sometimes referred to as a bubble trap, to remove air that inadver-
tently makes its way into the chromatography system with the various feed solutions.
Air trapped in a packed bed is undesirable because the mobile phase flows around
the air pockets and prevents uniform flow distribution throughout the entire column
cross section, thus compromising separation. A filter housing is also in place to house
a surface filter – typically 0.2 µm – to protect the column from particulate. There are
sensors both precolumn and postcolumn that are used to measure a number of differ-
ent parameters, which we will discuss shortly. And, of course, there is a vessel to col-
lect product during the load step (for flow-through chromatography) or elution step
(for bind-and-elute chromatography). The column is connected to the skid through a
number of valves that enable mobile phase to flow downward, upward, or around the
packed column. There are also numerous other valves on the system, including multi-
ple inlet valves to select feed from one or more of the feed vessels and outlet valves to
properly direct flow out of the system (e.g., to the product collection vessel or to
waste).
Sensors in place upstream of the column are used to measure flow rate, pres-
sure, conductivity, temperature (usually with conductivity), pH, and the presence of
air. Postcolumn, there are likely to be sensors for measurement of conductivity, tem-
perature (usually with conductivity), pH, and UV absorbance. The chromatography
system may also be equipped with a postcolumn pressure sensor. Two of the most
important to chromatography operations – conductivity and UV absorbance – are
discussed in more detail.
234 Chapter 8 Purification operations: chromatography

(a)
Feed vessels Chromatography skid Chromatography
column or
Air Filter membrane/monolith
trap housing module
Multiple
inlet
valves
Precolumn
sensors

Pump A

Postcolumn Product
sensors collection vessel

Multiple
outlet
Multiple valves
inlet
valves Pump B

Figure 8.6: A typical production chromatography system. (a) shows a basic flow diagram for a
production-scale chromatography system. (b) shows the front view and rear view of a Cytiva
 8.7 Chromatography equipment for biopharmaceutical production 235

Conductivity is a measure of a fluid’s ability to conduct electric current and can be

viewed as a way to monitor ionic strength. Conductivity is readily measured using a
bench-top conductivity meter or inline conductivity elements, as is common on
chromatography systems. The SI unit for conductivity is Siemens (S)/m; in biophar-
maceutical applications, however, mS/cm or µS/cm is more commonly used. Sol-
utes that readily ionize in water have relatively high conductivity values because of
their ability to conduct current; for example, 0.5 M NaOH has a conductivity of
98 mS/cm. Compounds that do not dissociate in aqueous solutions (e.g., ethanol)
have relatively low conductivity values; for example, a solution of 20% (by volume)
ethanol in water has a conductivity <1 µS/cm.
The measurement of conductivity in chromatography operations is extremely im-
portant as it provides a measure of a fluid’s ionic strength (a measure of the concentra-
tion of ions in solution), which plays a key role in protein binding for several of the
chromatographic interaction modes. Conductivity measurement is also used to monitor
the progress of a chromatography run. Because different solutions have different con-
ductivity values, a chromatogram showing either precolumn or postcolumn conductiv-
ity plotted versus time or cumulative volume of fluid fed can serve as confirmation that
the correct solutions were fed at each phase of the chromatography step and in the
proper amount.
Ultraviolet (UV) absorbance is the common method for detecting components in
the effluent from a chromatography device. Proteins absorb UV light – electromagnetic
radiation with a wavelength range of 10–400 nm – and do so strongly at a wavelength
of 280 nm due to the presence of amino acids with aromatic rings (tyrosine, phenylala-
nine, and especially tryptophan). Monitoring UV absorbance therefore serves as a
method for detecting proteins and protein-based components in the column effluent.
Most chromatography systems are equipped with inline UV sensors, many of which
have the capability of monitoring multiple wavelengths simultaneously. At low enough
concentrations of protein, the relationship between UV absorbance, A, and concentra-
tion of an absorbing solute such as a protein, Csolute, is given by the Beer-Lambert law:

A = ϵlCsolute (8:16)

where l is the optical path length and ɛ the extinction coefficient. This equation sug-
gests that a plot of A versus Csolute results in a straight line, making absorbance
measurements useful for quantifying protein concentration. However, the A versus
Csolute curve can deviate from this linear relationship for a number of reasons, includ-
ing that at higher concentrations, molecular interactions increase, which results in
proteins not contributing to absorbance independently. Proteins and protein-based

Figure 8.6 (continued)

AKTAprocess chromatography skid that operates at flow rates of up to 180 L/h. (b) also shows an
unpacked BPG 300/500 column, which has a diameter of 30 cm (300 mm) and allows a maximum
bed height of 50 cm (500 mm). Images © NC State University; reprinted with permission.
236 Chapter 8 Purification operations: chromatography

biopharmaceuticals may be at high enough concentrations in chromatography steps

to be outside the linear relationship defined by the Beer-Lambert law.
Figure 8.7 shows a chromatogram for the purification of green fluorescent protein
(GFP) in a clarified lysate generated from E. coli cells. Precolumn conductivity,
postcolumn conductivity, and postcolumn UV absorbance at 280 nm and 395 nm
are all plotted against cumulative volume (in L) of mobile phase fed to the column. The
equilibration and wash buffer is 50 mM tris, pH 8.0. As expected for an AEC step, con-
ductivity during the equilibration, load, and wash phases is kept low so that few ions
are present to disrupt protein binding. Tris was chosen due to its high buffering capac-
ity at pH 8, a pH that results in a net negative charge for GFP (pIGFP ≈ 6.0) so that it
binds to the resin. The method is programmed with a linear NaCl gradient – 0.1 M NaCl
in tris buffer, pH 8.0 to 0.2 M NaCl in tris buffer, pH 8.0 – which is reflected by the
linear increase in precolumn conductivity during elution. Conductivity increases signif-
icantly during the regeneration phase, which uses a 2 M NaCl solution. GFP absorbs UV
light strongly at a wavelength of 395 nm (A395) in addition to 280 nm (A280). The ele-
vated A280 during the load and wash phases, with no absorbance at 395 nm, indicates
that many impurities are not bound to the column and are present in the column efflu-
ent, which is directed to drain during these phases. The A280 is also high during the
initial portion of the elution phase, during which effluent is directed to drain. GFP
elutes near the end of the linear gradient as indicated by the A395 peak. Product collec-
tion is initiated when the A395 value rises to 0.1 absorbance units (AU) and ends as the
A395 decreases to 0.1 AU. Based on the A280 and A395 traces, it is clear that significant
separation of impurities from product has taken place. From the chromatogram, it is
also clear that about 7 L of eluate (product) was collected. (Note that a pilot-scale col-
umn with a 14 cm diameter and bed height of 25 cm was used in this run).
We’ve already discussed packed beds in some detail, but let’s briefly discuss col-
umn hardware. A chromatography column is basically a tube that holds packed chro-
matography resin in place. It is equipped with bed supports on each end that hold
resin stationary within the column and help distribute flow across the column cross
section and distributor plates that further ensure uniform flow distribution. Most col-
umns allow for adjustable bed heights. Some characteristics of columns used for bio-
pharmaceutical production, pulled together from a search of the websites from three
leading vendors for columns for the biopharmaceutical industry (Cytiva, Millipore-
Sigma, and Pall) are:
– Diameter: up to 2 m
– Bed height: up to 50 cm
– Column materials of construction: plastic, glass, or stainless steel
– Pressure ratings: up to 5 bar
 8.7 Chromatography equipment for biopharmaceutical production 237

Figure 8.7: A chromatogram for purification of GFP from an E. coli lysate using AEC. The y-axis
shows precolumn conductivity (brown), post-column conductivity (red), and postcolumn UV
absorbance at 280 nm (blue) and 395 nm (green) wavelengths. Cumulative volume, in L, is shown
on the x-axis. Note that the chromatogram goes only through the regeneration phase and does not
include the sanitization, buffer rinse, or storage phases. Image © NC State University; reprinted
with permission.
238 Chapter 8 Purification operations: chromatography

As you can see, the pressure rating for most columns used for biopharmaceutical ap-
plications is relatively low pressure. Columns for medium (20–40 bar) and high pres-
sure (>40 bar) are available but less commonly used.
Common methods for packing resin into chromatography columns can be found
in a number of references [226] and are not covered here. However, packing can be
challenging, and a number of defects can occur in a packed bed – such as channel-
ing, trapped air, clogged bed supports – that result in flow disturbances, which can
compromise separation in the column and therefore product quality. Thus, it is com-
mon to ensure packing is of high quality just after packing, prior to its first use, and
even between runs. Evaluation of column packing is most often accomplished by
measuring the HETP, which was discussed previously, and the asymmetry for the
packed bed. One procedure is used to determine both values and involves injecting a
pulse of tracer into the packed bed (or membrane adsorber or monolith) and then
comparing the calculated HETP and asymmetry values to acceptable ranges. The pro-
cedure is summarized below [227]:
1. Inject a small volume – usually less than one percent of the bed volume – of a
tracer to the column. The tracer should be small so that it can easily access
resin pores, nonbinding so that binding effects are eliminated from consider-
ation, and detectable at the column outlet. Commonly used tracers are acetone
(1–2% acetone in water with water as the mobile phase to push the acetone
pulse through) and concentrated NaCl solutions (e.g., 2 M NaCl).
2. Push the tracer pulse through the column using an appropriate mobile phase.
The broadened pulse results in a chromatographic peak as shown in Figure 8.8.
3. Calculate the asymmetry, As, of the peak that elutes using the following equation.

b
As = (8:17)
a

where a is the width of the first half of the peak at 10% peak height and b is the
width of the second half of the peak 10% peak height, as shown in Figure 8.8.
For As > 1, the peak is referred to as a tailing peak. As < 1 is a fronting peak. An
As value of one is ideal, although a range of 0.8 to 1.8 may be considered accept-
able [227].
4. Calculate the HETP from equation (8.9) as the height of the bed divided by the
number of plates. The number of plates is calculated as:

2
Vr
Nplates = 5.54 (8:18)
Wh

where Vr = volume of mobile phase required to move the tracer, once injected to a
column, to the column outlet, and Wh is the width of the peak at half height.
Both are illustrated in Figure 8.8. Note that this equation is derived from geometric
properties of a Gaussian peak that are related to the variance and mean of the peak
 8.7 Chromatography equipment for biopharmaceutical production 239

[210]. HETP is often expressed as a reduced plate height, H, defined as the ratio of
the HETP to diameter of particles (dp) packed into the column:
HETP
H= (8:19)
dp

H < 3 is considered optimal but assumes that optimal test conditions (such as
mobile phase velocity as discussed previously for HETP vs velocity curves) are
utilized, which may not always be the case in an industrial setting.

Vr
Absorbance or Conductivity

Wh

50%
a b
10%

Volume

Figure 8.8: Chromatogram showing the output (e.g. UV absorbance or conductivity measured in the
column effluent) from a pulse tracer test to determine the HETP and asymmetry of a packed column.
All variables shown are used in the equations used to calculate HETP and asymmetry. a and b are
the distances shown at 10% of peak height. Vr is the volume of mobile phase required to move the
tracer, once injected to a column, to the column outlet, and Wh is the width of the peak at half
height. Image © NC State University; reprinted with permission.

We’ll wrap up this section on chromatography systems with a few brief comments on
single-use systems. There are a number of them on the market, based on ready-to-
use, disposable flow paths fitted with the necessary sensors. These include systems
from Cytiva, Sartorius, and MilliporeSigma [228–230]. Single-use chromatography
“columns” come in two formats: membrane adsorbers, discussed previously, and
prepacked columns. Both are generally reusable, and, in fact, using a prepacked col-
umn only once may be too expensive to be considered truly single-use [231]. Membrane
adsorbers and prepacked columns can be purchased from a number of vendors. Both
types of devices offer the advantage of avoiding challenges with column packing. Cur-
rently, the largest prepacked column is Repligen’s Opus® 80 R column with a diameter
of 80 cm [232].
240 Chapter 8 Purification operations: chromatography

8.8 Process and performance parameters for chromatography

Numerous process parameters can impact chromatography performance. We have al-

ready discussed a number of these. Table 8.4 provides a summary of process parame-
ters and their potential impact on the following performance parameters and critical
quality attributes (CQAs): step yield, binding capacity, productivity, column lifetime,
and purity. A number of the parameters are written generally to apply to most interac-
tion modes, and specifics will vary according to mode. For example, the impact of
“load conditions,” such as the composition, pH, and conductivity of the buffer in
which the product is dissolved, differ according to whether IEC, AC, or HIC is being
used. In Table 8.2, we see that for IEC, low conductivity of the load buffer favors prod-
uct binding by ensuring few ions are present to compete for binding sites on the sta-
tionary phase. Thus conductivity impacts binding capacity and, potentially, step yield
for IEC. Further, the pH of the load buffer in IEC must be such to provide a charge on
the target solute opposite to that of the stationary phase. Therefore, pH of the load
buffer also has the potential to impact binding capacity and step yield. For HIC, the
presence of certain salts in the load buffer, such as (NH4)2SO4, encourages hydropho-
bic interaction between the target solute and stationary phase. As a result, the type
and concentration of these salts impact binding capacity and, potentially, step yield.
Note that the list presented in Table 8.4 makes no assessment as to whether the
process parameters are critical, key, or non-key (see Chapter 2). Their classification
is determined as part of process development studies, along with establishing the
specific relationship between these inputs and the outputs (performance parame-
ters and CQAs). These parameters must then be controlled to these ranges through
production control mechanisms such as batch records, automation, etc.

8.9 Chromatography process design

Let’s bring together the information we have covered thus far by considering an ex-
ample that illustrates an important design strategy for the purification stage of a
biopharmaceutical process: three or more chromatography steps, each with a differ-
ent interaction mode, are often necessary to achieve the purity required in the final
drug product. The first chromatography step is referred to generally as the capture
step, and the second and third as the intermediate and polishing steps, respectively.
Data from a purification stage designed with this approach for production of a mAb
in Chinese hamster ovary (CHO) cells is shown in Table 8.5 [233]. Specifically, step
yields and impurity levels for each of the three chromatography steps used – protein
A chromatography, CEC, and AEC – are shown. Review of these data lead to the fol-
lowing observations:
 8.9 Chromatography process design 241

Table 8.4: Common process parameters for a chromatography step and their potential impact on
performance.

Process Parameter Performance Parameters and CQAs

Potentially Impacted

Resina Binding capacity, purity, step yield,

productivity, lifetime

Packed bed (column) HETP/asymmetry Binding capacity, purity, step yield

Bed height Purity, step yield

Equilibration conditions (e.g., buffer composition, Binding capacity, step yield

pH, conductivity)

Equilibration flow rate and CVs Productivity

Load conditions (e.g., buffer composition, pH, Binding capacity, step yield
conductivity)

Product load (e.g., g product/Vbed) Purity, step yield

Wash conditions (e.g., buffer composition, pH, Purity, step yield

conductivity)

Wash flow rate and CVs Productivity

Elution conditions (e.g., buffer composition, pH, Purity, step yield

conductivity)

Elution gradient slope Purity, step yield, productivity

Elution flow rate Productivity, purity, step yield

UV gates (start and end collection points for product) Purity, step yield

Cleaning, sanitization solutions Column lifetime

Flow rate and CVs for all other phases not mentioned Productivity

Temperature Purity, step yield

a
Better identified as a raw material than a process parameter based on definitions presented in
Chapter 2, but it is an important input to a chromatography step and therefore included here.

– Protein A chromatography run in bind-and-elute mode – a commonly used AC

step as shown in Table 8.2 – is very selective for mAbs as shown by the data in
Table 8.5. Significant amounts of important process-related impurities such as
HCPs, host cell DNA, and endotoxin are removed from the harvested cell culture
fluid (HCCF) by the protein A step, while the step yield for the mAb product re-
mains high. However, the protein A step does add leached protein A (the ligand
from protein A resin) into the product.
– Cation exchange run in bind-and-elute mode provides significant clearance of HCPs,
protein A, and aggregate. The majority of HCPs, leached protein A, and aggregate
242 Chapter 8 Purification operations: chromatography

were eluted from the column after the elution step, during the regeneration phase,
which suggests they were more strongly bound to the CEC resin (i.e., more positively
charged) than the mAb product.
– Anion exchange run in flow-through mode provides removal of residual HCPs,
many of which have a negative charge at the pH of the load and therefore bind to
the AEC column. To create flow-through mode, the pH of the buffer in which the
mAb resides during loading to the AEC column was adjusted to 0.5–1 pH unit
below the mAb pI.

Turning our attention to the design of individual chromatography steps, we recall that
the goal of process development activities is to define the relationship between process
inputs (e.g., process parameters) and process outputs (e.g., performance parameters
and CQAs). Applied specifically to chromatography, process development seeks to pro-
vide the necessary purity to meet CQAs of drug substance and optimal performance of
each chromatography step by providing high step yield, high productivity, and low
cost, all while maintaining product quality. Thus, the relationships between the process
parameters listed in Table 8.4 to the various performance parameters and CQAs needs
to be established. As discussed in Chapter 2, process design requires an understanding
of the CQAs for a product, as well as a general understanding of processing, which in
the case of chromatography is required to make initial selection of the type of chroma-
tography media to use – e.g., ion exchange or affinity or hydrophobic interaction. Once
the initial choice of stationary phases has been made, development activities take
place. For chromatography, development relies heavily on experimental studies given
the complexity of the product intermediates involved (i.e., product with numerous,
often poorly characterized impurities). However, the introduction of mechanistic mod-
els – that is, mathematical models that describe the physical processes taking place,
rather than statistical models limited to describing data – into design activities is be-
coming more common [234].
Studies typically begin with screening of both resins and mobile phase conditions,
in which the capacity of resins for a solute and selectivity of resins for product relative
to key impurities are assessed to select resins that optimize binding capacity (minimiz-
ing the volume of resin required, thereby minimizing cost) and ensure acceptable selec-
tivity (maximizing purity). Even if the mode of interaction for a chromatography step
has been narrowed to one, there will still be numerous options among that one interac-
tion mode available from the different vendors. The specific design of these screening
studies varies according to product type. An excellent example describing screening of
CEC resins used to remove aggregates from an Fc fusion protein following a protein A
capture step and low pH viral inactivation step has been presented by Shukla [235].
Modern process development tools for chromatography steps include miniatur-
ized systems that make for rapid testing and low sample requirements. Options in-
clude 96-well filter plates filled with resin, such as PreDictor™ plates from Cytiva
Table 8.5: Typical step yield and purity values for a three-column purification process, similar to that shown in Figure 2.3, for a mAb produced in CHO as
reported by Fahrner et al [233].

Process intermediate Step yield HCP DNA Endotoxin Protein A Aggregate (%)
(%) (ng HCP/mg (pg DNA/mg (EU/mg (ng protein A/mg
product) product) product) product)

Harvested cell culture fluid (HCCF) NA ,–,, ,–,, – NA –

Protein A pool > –, –, <. – –

Cation exchange pool – – < <. < <.

Anion exchange pool > < < <. < <.

8.9 Chromatography process design
243
244 Chapter 8 Purification operations: chromatography

[236] and miniature columns, such as Opus® RoboColumns® from Repligen [237].
Both allow for multiple studies conducted in parallel with automation to conduct
screening studies. The Opus® RoboColumns® are small (50–600 µL volume), pre-
packed columns that come in a 96-well plate format and can be used in combination
with a liquid handler, such as the Tecan Freedom EVO®, for automated, parallel execu-
tion of studies screening studies [237]. The combined Tecan / Opus® RoboColumns®
systems are capable of running step elutions and collecting fractions for analysis.
High-throughput methods to analyze effluents from these high-throughput screening
studies are desirable as well. Combined with design of experiment (DoE), these meth-
ods are now referred to as high throughput process development. A nice example of
implementing 96-well filter plates, miniature columns, and lab columns packed to a
typical production height for development of a step to separate mAb product from
aggregates, testing CEC and MMC resins, is provided by Welsh et al [238].
Following screening studies, which if successful lead to selection of resin and
an initial set of load, wash, and elution conditions, method optimization/characteri-
zation typically takes place in which load, wash, and elution conditions are final-
ized and the remaining phases (see Table 8.3) specified as well. These bench-scale
studies may be conducted with a column packed to the same bed height as used in
production. Peak pooling criteria – that is, when to start product collection and end
production collection based on UV absorbance values from the product peak – are a
key outcome from these studies as well.
Once a chromatography step is designed at bench scale, it needs to be scaled
up. In addition, chromatography steps may need to be scaled down from production
for the purpose of conducting studies, such as those required to demonstrate viral
clearance. The objective of scale-up or scale-down is to keep the performance, as
measured by purity (or impurity profile), step yield, and other performance parame-
ters, consistent at each scale. We focus on scale-up in what follows, although the
concepts apply to scale-down as well. Scale-up could theoretically be carried out by
replicating enough lab-scale columns, each running in parallel (i.e., the feed is equally
distributed to each) at optimized process parameters, to increase total throughput to
meet production requirements. Because this approach is impractical, scale-up is in-
stead carried out by increasing the column diameter while holding the column height
and mobile phase superficial velocity constant to keep the fluid residence time constant
at each scale. Specifically, the following basic rules can be applied to column scale-up:
1. Optimize chromatography conditions at lab scale.
2. Use the same mobile and stationary phases at both scales; for example, use the
same resin, with same particle diameter and the same solutions for each of the
phases that make up the chromatography step.
3. Maintain the product load – the ratio of the amount of product loaded (typically
in grams) to bed volume – between scales.
 8.9 Chromatography process design 245

4. Keep mobile phase residence time constant between scales (designated as 1 and
2) constant. To do this, maintain bed height between scales, hbed1 = hbed2, and
maintain superficial velocity between scales, vs1 = vs2.
5. For gradients, keep the slopes constant between scales. The gradient slope is
(Chi – Clow)/(Vgradient/Vbed), where Chi and Clow are the high and low concentra-
tion of the component impacting binding of product to the stationary phase
(e.g., the high and low concentration of NaCl in a linear gradient for IEC).

The result of scaling up using the rules described above results in a column with a with
a larger diameter to process the larger amount of product to be loaded to the column
and a higher volumetric flow rate, which results in a processing time for the chroma-
tography step equal to that at bench scale despite the larger product volume to be proc-
essed. The example below demonstrates how this scale-up approach is applied.

Example: Scaling up a 1.5 cm diameter column to production scale

A chromatography step is optimized at bench scale using a 1.5 cm diameter column packed to a bed
height of 25 cm. The flow rate for the load phase is 15 mL/min, and the load volume is 260 mL.
(Note that the residence time of fluid in that column was calculated in a previous example in Section
8.4). You will scale up to an estimated load volume of 200 L in production. Answer the following.
(a) What is the bed height of the production-scale column?
(b) How large a column (expressed as the column diameter in cm) is required for production?

Solutions
(a) The bed height at each scale is kept the same (hbed1 = hbed2); therefore, the bed height at
production scale is 25 cm.
(b) Next, calculate the bed volume required for the separation by keeping product load equal
at both scales. Assuming that the concentration of product in the load solution is the same
at both scales (and designating bench and production scale by subscripts 1 and 2, respec-
tively), results in:
Vload1 Vload2
=
Vbed1 Vbed2
where the Vbed is given by equation (8.1). Solving for Vbed2 gives:

2
Vload2 200 L 1.5
Vbed2 = × Vbed1 = ×π× × 25 cm
Vload1 0.260 mL 2

= 33, 984 cm3 or ≈ 34 L

Solving equation (8.1) for Dbed2 gives:


1=2
4Vbed2 1=2 4 × 33, 984 cm3
Dbed2 = =
πhbed2 π × 25 cm
= 41.6 cm
But column diameters come only in certain values, so if you are purchasing a new column,
a diameter of 45 cm is a likely choice. Using a 45 cm column results in packing slightly
more than the necessary amount of resin if the bed height is kept at 25 cm; i.e., the column
is packed with more binding capacity than is actually needed.
246 Chapter 8 Purification operations: chromatography

From the example above, it is clear that keeping both the amount of product loaded
per volume of bed and bed height constant upon scale-up fixes the diameter of the col-
umn; that is, the column diameter cannot be independently selected. If you do not
have a column of the calculated diameter and do not want to buy one, it is possible to
scale up to a particular column diameter by removing the constraint of maintaining
bed height between scales. Like the direct scale-up method presented previously, the
method relies on keeping the mobile phase residence time constant between scales.
But rather than keeping residence time constant by keeping bed height and superficial
velocity constant, the volumetric flow rate, in units of CVs per time (e.g., CVs/h), is
held constant. In addition, the amount of product loaded per volume of bed is held
constant between scales [239]. This volumetric flow-based method allows the flexibil-
ity of using an existing column – that may not have the diameter estimated by the
direct scale-up approach previously described – by adjusting the bed height to provide
the same mobile phase residence time between scales. Of course, changing the bed
height leads to a different number of theoretical plates in the columns at each scale;
however, it has been shown that separation is always equal to or better when the bed
height is increased on scale-up [239]. Thus, this volumetric flow-based method of scale-
up should be satisfactory when a chromatography step has been optimized at bench-
scale using a short bed height relative to what is used in production.

8.10 Validation considerations for chromatography steps

Like all equipment, a chromatography system – both skid and column – must be
qualified prior to use, as described in Chapter 3, to provide assurance that the equip-
ment has been installed properly and operates through the intended ranges. Consis-
tency of the chromatography steps that are part of a process is demonstrated as part
of process performance qualification (PPQ) runs for process validation. Recall from
Chapter 3 that PPQ includes execution of multiple batches, usually a minimum of
three, at full manufacturing scale and with more sampling and testing than during
routine commercial production to sufficiently establish consistency and product qual-
ity throughout the process. For a chromatography step designed to remove impurities
X, Y, and Z, sampling and testing during PPQ would be performed to demonstrate
clearance of X, Y, and Z. A multitude of other data are also typically examined when
assessing consistency of chromatography steps during PPQ. For example, UV elution
profiles related to the product and impurities for each run may be reviewed for consis-
tency. HETP/asymmetry values may be evaluated after each PPQ run for consistency,
and so on. Following successful completion of the PPQ runs, this additional testing
and data review may be removed because the consistent performance of the chroma-
tography step has been demonstrated.
 8.10 Validation considerations for chromatography steps 247

In addition to equipment qualification, PPQ runs, and routine monitoring of

processing steps following the process qualification stage of process validation, a
variety of studies specific to chromatography systems must be conducted as part of
process validation. These may be viewed as an extension of process design activities
(stage 1 of process validation) and include resin (or membrane adsorber or mono-
lith) lifetime studies, viral clearance studies, impurity removal studies, and column
storage studies. Note that if chromatography columns or membrane adsorbers are
not reused, then the need for lifetime and storage studies is eliminated. The use of
the term column in what follows includes both membrane adsorbers and monoliths.
As mentioned previously, it is common to reuse chromatography media in
manufacturing for batches of the same product. Because of the importance of chroma-
tography steps to product purity and overall product quality, data must be obtained to
demonstrate that a column can maintain performance over a certain number of runs –
that is, the lifetime of the column expressed as a number of allowable runs (or cycles)
must be determined. A lifetime study must be performed for each chromatography step
used in a process. Studies performed to generate these data are typically conducted at
both small and production scales, with small-scale studies providing an initial estimate
of lifetime, in what are often referred to as cycling studies, and production-scale studies
providing confirmation of the lifetime. Deterioration of the chromatography stationary
phase can result in a decrease in product purity, increase in impurities, increase in pres-
sure drop across the column, variation in elution profiles, and decreasing step yields.
Consequently, these are all parameters that may be monitored as part of bench-scale
cycling studies and production-scale runs to confirm lifetime.
An important component of establishing a column lifetime is assurance of minimal
carryover of product/impurities from one batch to the next throughout the lifetime of
the chromatography media being used. Minimal carryover is typically demonstrated by
performing blank runs at production scale. These runs involve the usual sequence of
phases, except that the column load is replaced by a buffer – typically the buffer in
which the product is dissolved – so that any components that elute from the column
during the blank run, whether product or impurities, have come from the previous
product run and indicate undesirable carryover. Blank runs are executed on a single
column periodically throughout its lifetime, importantly including a blank run at the
end of its lifetime. Once column lifetime is established, blank runs are no longer exe-
cuted. Column lifetimes established in cGMP manufacturing can be significant. For ex-
ample, lifetimes for protein A resins for mAb purification used in cGMP manufacturing
are commonly 50–200 cycles [240].
As discussed in Chapter 2, processes that pose a risk of introduction of virus into
the product require viral clearance steps. Viral clearance refers to either physical re-
moval or inactivation of the virus. Viral clearance studies are performed on individual
steps identified to remove and/or inactivate virus in a process. The objective of these
studies is to assess the potential of a step to clear virus. They are performed using an
appropriately scaled-down version of the process step and involve spiking a panel of
248 Chapter 8 Purification operations: chromatography

relevant viruses into the process intermediate that feeds the unit operation being
evaluated. The clearance is typically quantified as a log reduction value, defined as
the base ten log of the ratio of total virus loaded to the step to the total virus in the
product (i.e., output) from the step. Because chromatography steps can provide viral
clearance, they are often part of the scaled-down viral clearance studies for a process.
For example, protein A chromatography [241] and AEC [242] used in mAb production
have been shown to provide significant viral clearance. For chromatography steps,
these studies should demonstrate the ability of a stationary phase to clear virus across
the process parameter ranges identified in the design of the process and should in-
clude resin that is at the end of its lifetime to ensure that viral clearance remains ro-
bust throughout the lifetime of the resin. Note that the overall goal of these clearance
studies is to demonstrate that a process has more than enough viral clearance capac-
ity to ensure drug product safety [243].
In addition to resin lifetime and viral clearance studies, impurity removal stud-
ies and storage studies are conducted. Impurity removal studies typically gather
bench-scale data that demonstrate consistent removal of impurities. These studies
can be conducted as part of the resin lifetime evaluation. Storage studies demon-
strate that the column storage solution effectively inhibits microbial growth and
does not degrade the stationary phase – that is, does not cause ligand leaching or
cause degradation of the support material – over the expected storage times. The
storage study should also determine conditions required for complete removal of
storage solution.

8.11 Summary

Numerous soluble process- and product-related impurities are present in biopharma-

ceutical product intermediates that must be removed to produce drug product that is
safe and effective. Host cell proteins, endotoxin, leachables from product-contact ma-
terials, virus, and product aggregates are some examples. Chromatography separates
by preferential interaction of certain solutes, such as product and impurities, in the
mobile phase (i.e., liquid phase that moves) with the stationary phase (e.g., resin
beads packed into a column). For example, product may bind more strongly to a chro-
matography resin than impurities, which results in the product moving through the
column (which contains the resin beads) more slowly than the impurities. The differ-
ence in speed in which product and impurities move through the column leads to the
separation. In addition to resin packed in columns, chromatography stationary phase
options also include membrane adsorbers and monoliths.
Interaction between mobile phase solutes and the stationary phase is a result of
solute properties, mobile phase conditions, and stationary phase properties. Stationary
phases come in a variety of chemistries classified according to their interaction mode:
ion exchange, affinity, hydrophobic interaction, reversed phase, and mixed mode. Size
 8.11 Summary 249

exclusion chromatography also exists, although separation is not the result of station-
ary phase chemistry but rather occurs because small molecules are able to diffuse into
resin pores that are not accessible to larger molecules. The physical process involved in
binding of solutes to resin beads includes transfer of the binding solute from the bulk
mobile phase to the resin bead, diffusion of the solute into the pores of the bead where
most binding sites exist, and finally adsorption to the resin. Intraparticle diffusion is
particularly slow and results in broadening of solute bands as they move through a
packed column.
Chromatography steps for biopharmaceutical processes are commonly run in
bind-and-elute mode, in which product binds to the stationary phase and impurities
may or may not bind. Flow through is another common mode, in which impurities
bind to stationary phase and product flows through. Each mode of operation com-
prises a number of phases. For example, bind-and-elute operation using reusable
columns requires equilibration, load, wash, elution, regeneration, sanitization, and
storage phases. Equipment used to carry out the chromatography procedure may be
either reusable or single use. Skids with disposable flow paths are available from a
number of vendors; however, single-use chromatography columns and membrane
adsorbers tend to be too expensive to be disposed of after each run.
Biopharmaceutical processes often consist of three or more different chromatogra-
phy steps (e.g., anion exchange, cation exchange, and hydrophobic interaction) to
achieve purity levels required for drug product. Development of individual steps is
fairly complex, as it involves relating numerous process parameters involved in chro-
matography to various performance parameters and CQAs. Empirical studies are typi-
cally required due to the complexity of the intermediates being purified. These studies
often use miniaturized automated equipment that can be used to perform multiple
studies in parallel. In addition to these development studies relating process parame-
ters to chromatography performance, a number of specific studies are conducted to
support validation of chromatography steps, including column lifetime, viral clearance,
column storage, and impurity profile studies.
Finally, we started off the chapter by saying that chromatography is a big topic,
and we conclude this chapter with the admission that there are many important
topics related to chromatography that could not be covered in this single chapter.
For example, continuous chromatography, which has been the focus of much devel-
opment activity over the past several years, was not covered here nor were details
around mechanistic modelling of chromatography steps, which will undoubtedly
play a larger role in chromatography development and process control in manufactur-
ing in the years to come. But we’ll have to end it here.
250 Chapter 8 Purification operations: chromatography

8.12 Review questions

1. Proteins X, Y, and Z are separated by AEC. Protein X does not bind under the
loading conditions (i.e., it elutes in the load flow through), but proteins Y and Z
do. Y and Z are eluted from the column using a linear NaCl gradient, with Z elut-
ing first. On a single graph, qualitatively draw charge vs. pH curves for each of
the proteins. Draw a vertical line indicating the pH at which the separation takes
place.

2. Derive an expression for the superficial velocity in a membrane adsorber with a

cylindrical geometry in which mobile phase flows from the exterior of the cylin-
der formed by the membranes, through the membrane layers, and out through
an internal flow channel. Note that at a constant volumetric feed rate, Q, the
superficial velocity increases as fluid moves from the exterior of the membrane
cylinder to the interior.

3. A solute is continuously fed to a packed column and binds to the resin. Do the
following:
(a) Draw (qualitatively) the resulting breakthrough curve.
(b) Using the breakthrough curve, graphically indicate the area that corresponds
to the total mass of solute adsorbed onto the column. To do this, perform a
solute balance over the column from the beginning to end of the run, and
use the result to determine the appropriate area.
(c) On the same curve, show the area that corresponds exactly to mads,10%.
(d) On the same curve, show the area corresponding to the mass mads,10% esti-
mated by equation (8.11).

4. A study is conducted to generate the adsorption isotherm for a mAb binding to

protein A resin. A number of conditions are tested to determine the EBC over a
range of mAb concentrations in a mobile phase. In the study, 1.00 mL of resin
slurry, which equates to 0.25 mL of resin, is added to a 2.00 mL vial and incu-
bated for 48 h with 1.00 mL of mAb solution. At the end of 48 h, the concentra-
tion of mAb in the liquid phase is measured. If the concentration of mAb in the
liquid phase added to the vial is 13.0 mg/mL and the concentration of mAb in
the liquid phase after incubation is 0.3 mg/mL, (a) what is the value of the EBC
at these conditions, and (b) given that the adsorption isotherm is a plot of EBC
against the solute concentration with which the stationary phase is in equilib-
rium, what are the x and y isotherm coordinates that this data represents?

5. The data in Table 8.6 apply to the chromatography step represented in the chro-
matogram shown Figure 8.7. Based on these data, calculate (a) the % purity for
the column load (clarified lysate) and the product (i.e., eluate), and (b) the %
step yield for the run.
 8.12 Review questions 251

Table 8.6: Data for chromatography run represented in Figure 8.7.

Process GFP Total protein Phase start Phase end

intermediate concentration concentration volume (L) volume (L)
(mg/mL) (mg/mL)

Load . . . .

(clarified lysate)

Eluate . . . .

(start collection) (end collection)

6. A column with an inner diameter of 60 cm is packed for a CEC step to a bed

height of 20 cm. The equilibration and wash phases each use 3 CVs of 50 mM
MES, pH 6.0 buffer. The elution phase is a linear gradient from 0% to 50%
pump B (one of two inlet pumps on the system) over 4 CVs of elution buffer,
with the 50 mM MES buffer connected to pump A (the other inlet pump on the
system). Estimate the following: (a) the total volume (in L) of MES buffer that
will have to be prepared for the step and (b) the amount of MES, in kg, that
would be required to prepare this total volume of solution.
7. Draw a flow diagram representing the phases involved in a flow-through chro-
matography step using a single-use chromatography skid and column.

8. Read the article “Defining Process Design Space for a Hydrophobic Interaction
Chromatography (HIC) Purification Step: Application of Quality by Design (QbD)
Principles” by Jiang et al [78]. Answer the following questions:
(a) What specific HIC resin is used in this study?
(b) What type of biopharmaceutical is being purified?
(c) What mode of operation is used to run the HIC step?
(d) Which process parameters are considered in the characterization study, and
how were these chosen from a much larger list?
(e) Among the process parameters considered in the characterization study,
which were found to have a significant impact on step yield, aggregates
(HMW) levels, and HCP levels?
(f) For the two process parameters that had the most significant impact on step
yield, explain the reason for their impact.

9. Development activities for a chromatography step used in the production of a

vaccine are conducted at bench scale, using a chromatography system whose in-
line UV detector has a path length of 2 mm. The chromatography system used at
production scale has a UV path length of 10 mm. If the UV gates (measured at
280 nm wavelength) based on development studies were set to start collecting
252 Chapter 8 Purification operations: chromatography

when the product peak hits 0.2 absorbance units (AU) and to stop collecting on
the downward slope of the product peak at 0.5 AU, what UV gate values should
be transferred to production to ensure that the start and stop of peak collection
occurs at the same product concentration at both scales?

10. As a follow up to the example presented on scaling up a 1.5 cm column used for
bench-scale studies to production scale in section 8.9, answer the following
questions. (a) At what load flow rate will the 45 cm diameter column be oper-
ated? (b) If the sequence of phases used for the step is the same as shown in
Table 8.3, estimate the amount of time required for the entire step if the flow
rate remains the same at each phase and 3 CVs of solution flow to the column at
each phase.

11. A chromatography step for a biopharmaceutical process has been developed

using a column with a height of 5 cm, an internal diameter of 1 cm, and a load
flow rate of 1.30 mL/min. During the development studies, 100.0 mg of target
protein was loaded to the column, which corresponds to 77.0 mL of load mate-
rial. At production scale, you estimate that you will load 400 L of the same ma-
terial. You have a 30 cm column available for production scale. (a) Estimate the
bed height of the production column based on the volumetric flow scale up
method described in the chapter. (b) Estimate the flow rate, in L/h, you will use
to load the column.

12. You have been tasked with designing a test for an OQ protocol to check that the
pumps on a production chromatography system can deliver the flow rates re-
quired for production of a viral vector for gene therapy. The chromatography
system is automated and controls flow rate based on a set point value in L/h.
Describe the OQ test design, including the following information: (i) the param-
eter(s) you would set to execute the test, (ii) any measurements you would need
to make as part of the test, and (iii) how you would demonstrate whether the
chromatography system passes or fails the test. Use the flow meter calibration
example in section 3.3.4 as a guide.
Chapter 9
Formulation operations: ultrafiltration

Ultrafiltration (UF) is an important operation common to the formulation/fill stage

of most biopharmaceutical processes. Following the final chromatography step in a
process, the product is in a clear – that is, essentially particle-free – liquid solution
and contains low levels of the process- and product-related impurities discussed in
the last chapter. However, it is also at a concentration and buffer composition dic-
tated by that last chromatography step and likely different from what is required for
the drug product. Consequently, an adjustment to the product concentration and
buffer composition is needed, which can be accomplished by UF. This adjustment
produces bulk drug substance, which is used to create the drug product as de-
scribed in Chapter 2. UF steps are also used for other purposes in biopharmaceutical
processes, as discussed shortly. Note that we use the term product here to mean the
active molecule at any step in the process and the term product solution or product
intermediate to refer to the solution in which product is dissolved.
In manufacturing processes, UF is run in tangential-flow mode and is closely
related to tangential-flow microfiltration (MF), which was introduced in Chapter 7.
The main difference between the two is membrane pore size, with pores in an MF
membrane larger than those in a UF membrane. The pore size in MF membranes
makes them well suited for solid-liquid separations, including clarification steps in
which solids are retained and product solution permeates the membrane. The smaller
pores in UF membranes allow them to be used for solute-solvent separation, includ-
ing concentration steps in which product (the solute) is retained by the membrane
while water (the solvent in biopharmaceutical applications) and other small molecu-
lar weight components permeate. Thus, in MF steps, product typically permeates the
membrane, while in UF steps, product is retained by the membrane.
This chapter begins by introducing UF and answering the following questions:
– What is the basis of separation in UF?
– In addition to formulation of drug substance, what are applications of UF in bio-
pharmaceutical processes?
– What equipment is used to execute a UF step?

As the chapter continues, we take advantage of the similarities in the two types of
tangential-flow filtration (TFF) by bringing tangential-flow MF back into the discus-
sion to complete the coverage started in Chapter 7. Questions to be addressed in-
clude the following:
– What are UF and MF membranes made from and how are they modularized for
use in production?
– What is a typical TFF procedure for a current Good Manufacturing Practice
(cGMP) manufacturing environment?

https://doi.org/10.1515/9783110616880-009
254 Chapter 9 Formulation operations: ultrafiltration

– What are the process and performance parameters for TFF steps?
– How are TFF steps developed and designed for cGMP use?
– How are TFF steps scaled up from bench-scale development studies?

9.1 Basis of separation in UF and applications

Recall from Chapter 7 that our definition of filtration included not only the separa-
tion of solid particles in a fluid suspension according to their size by flowing under
a pressure differential through a porous medium, but it also included separation of
components in solution by size. While the former applies to MF, the latter part of the
definition describes UF.
Like MF, the separation in UF is size based and pressure driven, as shown in
Figure 9.1. Small molecules, such as water, salts, and buffering agents, permeate
the UF membrane while larger molecules or components are retained. As discussed
in Chapter 7, UF membranes retain solutes with a diameter of 0.005–0.15 µm [190].
However, the pore size in a UF membrane is typically described by a molecular
weight cutoff (MWCO), which has a range from 1 to 1,000 kDa [244] and is discussed
in more detail shortly.

proteins, other
large biomolecules,
viruses

salts and other

small molecules Feed pressure Retentate
at Pf pressure at Pr
membrane

water

Permeate at pressure Pp

Figure 9.1: Illustration of UF. Small molecules like water, salts, and buffer components permeate
the membrane, while larger components like proteins and viruses are retained. Note that in
tangential-flow mode, Pf > Pr. Image © NC State University; reprinted with permission.

UF membranes are most commonly used to concentrate product (i.e., increase the
concentration of product in solution by removing solvent, in our case water) or for
diafiltration (i.e., exchange the matrix of the product). (Note that in this book, the
term UF includes any operations that utilize UF membranes, including concentration
 9.1 Basis of separation in UF and applications 255

and diafiltration). We use the terms diafiltration (DF) and buffer exchange inter-
changeably throughout this chapter. Buffer exchange, in this context, refers not
only to the actual exchange of buffering agents, such as tris, but also solutes that
are not true buffers, such as NaCl. Concentration and DF are required near the end
of the drug substance process for most biopharmaceuticals. During concentration,
product solution is fed to the membrane. Water and other small molecules perme-
ate, while product is retained. As water continues to permeate, the product be-
comes more concentrated and its volume is reduced. During DF, as the product
solution is fed to the membrane, a buffer solution is added, typically at the same
rate that water permeates the membrane. In a UF system operated in this configu-
ration, the new buffer system washes out the old, which results in an exchange of
buffers. We discuss each of these operations in more detail shortly, but first let’s
consider the questions below to illustrate these concepts.

Example: Calculating target volume during a concentration step

Green fluorescent protein in 50 mM tris is introduced to a UF device at a concentration of
1.50 mg/mL and a volume of 300 L. Assuming that no GFP permeates the membrane but that
water, tris, and all other small molecular weight components freely permeate, calculate the fol-
lowing: (a) the retentate volume for a final GFP concentration of 20.0 mg/mL, and (b) the amount
of water that has been removed from the product in the permeate stream.

Solution
(a) If no product is lost in the permeate, then the mass of GFP remains unchanged from begin-
ning to end of the concentration step, and the following relationship applies:

Cinitial Vinitial = Cfinal Vfinal (9:1)

where C refers to the concentration of product and V to volume. So the final volume is given
by
g
Cinitial Vinitial 1.50 L × 300 L
Vfinal = =
Cfinal 20.0 gL

= 22.5 L
Note that in a production environment, typically a product concentration target is met by
concentrating to a target volume measured by a level sensor on a process vessel or using a
scale, rather than measuring a concentration value inline.
(b) During concentration, water, tris, and any salts present permeate the membrane, with
most of the volume being water. Consequently, the volume of water removed is calculated
by Vinitial – Vfinal:

Vfinal − Vinitial = 300 L − 22.5 L = 277.5 L

(c) In a separate step, phosphate buffered saline (PBS) is fed to the concentrated GFP solution
at the same rate that permeate is removed from the system. If the volume of PBS fed is 10
times that of the concentrated product solution (i.e., GFP solution at 20 mg/mL), determine
the final volume of product and the composition.
256 Chapter 9 Formulation operations: ultrafiltration

Because PBS is fed to the membrane at the same rate that permeate is withdrawn, the product
volume remains unchanged at 22.5 L. Further, because all buffer components and salts freely
permeate the membrane, the tris is washed out of the product solution, leaving the product at
its starting concentration in the replacement buffer, PBS solution, with only trace amounts of
the original tris buffer remaining.

UF used for product concentration and buffer exchange is designed and imple-
mented with a number of objectives in mind. These include consistently achieving
the target product concentration, consistently reaching the desired extent of buffer
exchange, maintaining product integrity, delivering consistently high step yield,
and achieving reproducible and high permeate flux.
We mentioned that UF may be used for more than just concentration and buffer
exchange to formulate drug product. Other applications in biopharmaceutical pro-
cesses include:
– concentrating and/or diafiltering an intermediate product stream prior to a chro-
matography step. Concentration reduces the volume loaded to a chromatogra-
phy system, and DF adjusts the product matrix (e.g., removes salts or exchanges
buffer components) to optimize product binding to the resin and/or selectivity of
the resin for the product;
– purifying viruses as part of a vaccine or gene therapy viral vector process in
which the large virus particle is retained while smaller impurities such as host
cell proteins permeate the membrane;
– removing unreacted reagents from chemically modified biopharmaceuticals, such
as antibody-drug conjugates discussed in Chapter 1, and PEGylated proteins,
which are therapeutic proteins to which polyethylene glycol has been attached
to increase retention time in the blood or to stabilize the protein [245].

9.2 UF equipment and process configurations for production

At production scale, UF is typically operated in tangential-flow mode (refer to

Chapter 7) and the equipment used is similar to the setup shown in Figure 7.12 for
tangential-flow MF. In fact, systems for tangential-flow UF and MF can often be used
interchangeably, with only minor modifications. For example, MF setups are usually
equipped with a pump on the permeate to keep permeate flux at an optimal level. A
permeate pump may be used in UF applications in which the pore size is relatively
large (100 kDa or greater) [190] but seldom for UF applications with tighter pores.
Diagrams showing a basic UF system for product concentration and buffer ex-
change are shown in Figure 9.2.
 9.2 UF equipment and process configurations for production 257

(a)
Feed tank* Retentate
valve Retentate
stream

Feed UF membrane
stream module Permeate Permeate
Feed or pump stream
recirculation (optional (to waste)
pump for UF)

(b)
Feed tank* Retentate
DF buffer Retentate
valve
tank stream

Feed UF membrane
stream module Permeate Permeate
Buffer feed Feed or pump stream
pump recirculation (optional (to waste)
pump for UF)

*Also referred to as retentate, recycle, or recirculation tank

Figure 9.2: UF system examples. (a) is the setup used for batch concentration. (b) is the setup
typically used for buffer exchange. Note that a system used for fed-batch concentration would be
similar to that for DF, except the DF buffer tank is replaced by a feed tank, and the feed tank as
labelled in Figure 9.2(b) would be referred to as the recirculation (or recycle or retentate) tank.
Image © NC State University; reprinted with permission.

The main components of these basic UF systems are:

– a feed tank that holds the product solution being processed;
– a feed pump that moves the product tangentially across the membrane surface
and pressurizes the feed stream. It is often a rotary lobe pump [190]. As discussed
in Chapter 7, the portion of the feed that does not permeate the membrane is re-
ferred to as retentate. The portion that permeates is referred to as permeate or fil-
trate. Consequently,
258 Chapter 9 Formulation operations: ultrafiltration

Qf = Qr + Qp (9:2)
where Q represents volumetric flow rate and subscripts f, r, and p refer to feed,
retentate, and permeate streams, respectively;
– membrane module(s) that contain the UF membrane required for separation. We
discuss membranes and modules in greater detail in the next section of this chapter;
– a retentate valve that along with the feed pump establishes transmembrane pres-
sure (TMP), which drives permeate flow through the membrane. Recall from equa-
tion (7.2) that TMP = (Pf + Pr)/2 – Pp;
– pressure sensors, used to measure the feed, retentate, and permeate pressures.
These values are used to determine the TMP and enable control to a TMP set
point (for systems operated under TMP control).
– a DF tank and pump used to feed buffer into the product solution for performing
buffer exchange. This pump is often a peristaltic type [190].

In addition to pressure gauges on the feed, retentate, and permeate streams, a num-
ber of other sensors may be found on a UF system. These include flow meters to
measure flow rate through the feed or retentate line and the permeate line; conduc-
tivity and pH sensors on the permeate line to monitor the progress of the DF step (as
DF progresses, the conductivity of the permeate will shift to that of the DF buffer); a
UV detector on the permeate line to detect product loss to the permeate; a level sen-
sor on the feed vessel for volume measurement; and temperature sensors on the
feed or retentate line to detect temperature increase.
Concentration is commonly carried out in batch mode. Retentate is recycled to
the feed tank until the concentration target, usually measured by a volume reduc-
tion, is met; thus, retentate undergoes multiple passes through the system shown in
Figure 9.2(a). Permeate is directed to waste. As the concentration step proceeds and
the amount of permeate increases, the level in the feed tank is reduced. During DF,
which typically follows concentration, retentate recirculates through system and
permeate is sent to drain, as with concentration. However, buffer from the DF tank
is also added, often at same rate that permeate is withdrawn, to accomplish buffer
exchange. When the rate of buffer addition matches the permeate rate, the volume
of product solution in the feed vessel does not change. This method of operation is
referred to as constant-volume DF. DF can be viewed simply as washing out the old
buffer solution and replacing it with a new one.
Concentration by UF can also be carried out in a fed-batch configuration, which
is particularly useful when a large volume reduction is needed for concentration
(e.g., when the final volume is too small to be measured in a large feed vessel). Refer-
ring to Figure 9.2(a), in fed-batch operation, product intermediate from a separate
vessel would be pumped to the feed tank, better referred to as the retentate vessel in
this case, at the same rate that permeate is withdrawn. Because the feed rate to the
retentate vessel and the permeate rate are the same, the volume of solution in the
 9.2 UF equipment and process configurations for production 259

retentate vessel remains constant. Once all product is fed to the retentate vessel, con-
centration of product can continue as a typical batch concentration step.
An example of a pilot-scale TFF unit is shown in Figure 9.3. This system can
accommodate either UF or MF membrane devices. The system in Figure 9.3 sits on
wheels, which is common for TFF systems. However, there are a range of TFF sys-
tems in use for cGMP manufacturing – from relatively simple benchtop units oper-
ated manually that might be used for small-volume clinical lot manufacturing to
much more complex, highly automated permanent units mounted to the floor.

Figure 9.3: Photo of a pilot-scale UF system, with major components labeled. Photo © NC State
University; reprinted with permission.

The photo in Figure 9.3 shows a stainless steel TFF skid that is cleaned, sanitized,
and reused between batches. Single-use TFF skids are also available. Examples in-
clude the Pall Allegro™ Single-Use TFF system, which comes in a smaller CS1000
and larger CS4500 model. Both include a disposable flow path and single-use sen-
sors to measure pressure, UV absorbance, temperature, flow rate, and conductivity.
The CS4500 can accommodate between 3 and 10 m2 of membrane area and provide
flow rates up to 4,500 L/h [246]. Other single-use TFF systems on the market include
the Mobius® FlexReady Solution featuring Smart Flexware™ Assemblies for TFF
from MilliporeSigma [247] and the FlexAct® CF from Sartorius [248].
260 Chapter 9 Formulation operations: ultrafiltration

9.3 TFF membranes and modules

As mentioned previously, UF membranes are characterized by a molecular weight

cutoff (MWCO), also referred to as a nominal molecular weight limit (NMWL), with a
range from 1 kDa to 1,000 kDa. The MWCO represents the smallest molecular
weight component, in daltons, retained by the membrane. However, the passage
or retention of components across the membrane is not only a function of the mo-
lecular weight being smaller or larger than the MWCO, for a number of reasons,
including:
– UF membrane pores are not all the same size – there is a distribution;
– membrane surfaces and pores can become blocked by the retained components,
effectively lowering the MWCO of the membrane;
– the shape and flexibility of the molecule may impact its retention [190, 249].
For example, a solute that is long, thin, and/or flexible may align and even
elongate under the influence of flow through the membrane pore in such a way
that solute enters the pore despite having a molecular weight greater than the
MWCO.

In Chapter 7, we discussed that MF membranes have pores with sizes expressed as a µm

rating. For MF, the range of pore sizes is as small as 0.1 μm or as large as 10 μm [173].
The images in Figure 9.4 show electron micrographs of UF membranes made
from two of the most common materials: modified polyethersulfone on the left and
regenerated cellulose on the right. Their structure is asymmetric, which is typical of
UF and MF membranes: a thin top layer provides the separation on top of a substrate
with larger pores [250]. The porous substrate provides mechanical strength for the
thin ultrafiltration layer while not significantly impeding permeate flow through the
membrane.
In general, regenerated cellulose is hydrophilic, exhibits low protein adsorption
and low fouling, and is more compatible with organic solvents than polyethersulfone-
based membranes. However, regenerated cellulose is less tolerant of pH extremes and
oxidants, such as hypochlorite, a key component in bleach. Consequently, cleaning
and sanitization of regenerated cellulose membranes with solutions having a high
concentration of NaOH (high pH) or including bleach (an oxidant) may be problem-
atic, although these more stringent cleaning solutions may not be required due to the
low protein binding and fouling exhibited by regenerated cellulose membranes. Poly-
ethersulfone tends to adsorb proteins and other biological components but is stable
over a wide range of pH values and to oxidants, which may allow for more effective
cleaning and sanitization. For bioprocessing applications, it is often modified to
be more hydrophilic, which reduces protein binding. In addition to polyethersul-
fone and regenerated cellulose, other materials are used for UF and MF membranes
for biopharmaceutical processing. An Internet search of the websites of membrane
 9.3 TFF membranes and modules 261

Figure 9.4: Scanning electron micrograph images of two membranes: (a) a 10 kDa MWCO Biomax®
modified polyethersulfone membrane and (b) a 10 kDa MWCO Ultracel® regenerated cellulose
membrane. Images from MilliporeSigma Technical Brief “Protein Concentration and Diafiltration by
Tangential Flow Filtration,” page 5. © 2003. Reproduced with permission from Merck KGaA,
Darmstadt, Germany and/or its affiliates [244].

vendors such as MilliporeSigma, Pall, Repligen, Sartorius, and GE Healthcare shows

that polyvinylidene fluoride, mixed cellulose ester, and polysulfone are also in use.
An exhaustive review of membrane materials is beyond the scope of this book; how-
ever, vendors can provide information to help decide on the type of membrane suit-
able for a particular application.
For ease of use, UF and MF membranes are assembled into modules that pro-
vide (1) a high surface-area-to-volume ratio, (2) physical separation of the retentate
and permeate streams, (3) scalability, (4) cleanability/sanitizability, and (5) flow
characteristics required for suitable performance. The two most common module
geometries in biopharmaceutical manufacturing are flat-sheet cassettes and hollow-
fiber modules. These are shown in Figures 9.5 and 9.6. A hollow-fiber module con-
tains a bundle of hollow, parallel fibers, with inner diameters typically ranging
from 0.5 mm to 1 mm based on a review of offerings from membrane vendors [251,
252]. The fibers are housed in a cylindrical shell. The interior of the fiber is referred
to as the lumen, and the space outside the fiber is the shell. The thin retentive layer
of the membrane is usually on the lumen side of the fiber, in which case feed flows
through the lumen, while permeate flows radially through the fiber and to the shell
side. Hollow-fiber modules are typically equipped with a single feed inlet, a single
outlet for the retentate, and two ports for permeate.
Cassettes are the most commonly used membrane module type for TFF in bio-
processing [190]. Figure 9.6(a) provides a view of the interior of a cassette and the
262 Chapter 9 Formulation operations: ultrafiltration

Figure 9.5: Hollow-fiber modules for TFF applications. (a) is a schematic showing feed, retentate, and
permeate flow through the module. (b) is a photo of a Spectrum (now Repligen) KrosFlo® hollow-fiber
module. It is 50 cm in length, has modified polyethersulfone fibers of 0.5 mm inner diameter and a
pore size of 0.2 μm, and contains 2.60 m2 of membrane area. This is an MF membrane, but similar
devices are available with UF membranes. (c) shows the front of the module, to which feed flows and
enters into the fiber lumen. Images © NC State University; reprinted with permission.

flow through it. The cassette has multiple layers of membrane across and through
which feed flows. (Note that the number of membrane layers shown in that drawing
is not necessarily representative of an actual cassette.) The tight retentive side of
the membrane (refer to Figure 9.4) is on the feed side. Retentate flow from the cas-
sette is separated from permeate. Screens may be inserted into the feed channels to
promote turbulence on the feed side of the membrane, which serves to better sweep
dissolved components (in the case of UF) or particles (in the case of MF) away from
the surface. This sweeping enables high permeate rates to be achieved at lower tan-
gential flow rates, which results in a reduced pumping requirement and the use of
less membrane area. Details on screen designs appropriate for different applications
can be obtained from the supplier.
A more detailed view of the flow of feed, retentate, and permeate in a cassette is
shown in Figure 9.6(c), which illustrates flow in a membrane “packet” – defined as
two membrane sheets and a permeate spacer. The holes shown on each side of the
packet in Figure 9.6(c) are specific to the feed, retentate, and permeate streams, and
 9.3 TFF membranes and modules 263

Figure 9.6: Images of cassettes and holders for TFF applications. (a) shows flows through a cassette.
Image adapted and reproduced with permission from Merck KGaA, Darmstadt, Germany and/or its
affiliates. (b) shows actual TFF cassettes. These are MilliporeSigma Pellicon® 3 UF cassettes with an
area of 88 cm2, 0.11 m2, 0.57 m2, and 1.14 m2. Reprinted from “Pellicon® 3 Cassettes with Ultracel®
Membrane,” MilliporeSigma Data Sheet, 2018, Lit. No.: DS1209EN001. © Merck KGaA, Darmstadt,
Germany and/or its affiliates, and reproduced with permission [253]. (c) shows flow through a single
“packet” of membranes within a cassette. A cassette is made up of a stack of multiple packets in
parallel. Image from “Integrity Testing of Ultrafiltration Systems for Biopharmaceutical Applications,”
by S. Yee Lau, P. Pattnaik, and B. Raghunath, 2012, BioProcess International, 10(9), 54. Reprinted
with permission [254]. And (d) shows two 0.5 m2 Pellicon® 2 cassettes, 10 kDa MWCO, placed in a
stainless steel holder. Photo © NC State University; reprinted with permission.

their circumference is sealed in such a way as to prevent flow from feed/retentate

holes to permeate holes and vice versa. Feed is pumped to a holder, which contains
the membrane cassettes. The cassette holder directs the feed to the feed holes on
one side of the cassette. The feed stream flows through its designated holes (every
264 Chapter 9 Formulation operations: ultrafiltration

other hole, as illustrated in Figure 9.6(c)) throughout the depth of the cassette, which
is made up of many packets. That is, feed flows parallel to each membrane packet
within the cassette through the alternating feed holes. The resulting retentate flows to
the other side of the cassette, where it accesses only the retentate holes, combines
with retentate from other membrane packets, and is directed by the holder out of the
cassette to the retentate line. Permeate flows across the membrane to the permeate
spacer and then out the permeate holes, where it combines with permeate from other
packets and is directed to the system permeate line by the holder.
To increase the membrane surface area within a cassette for scale-up, the feed
flow channel is widened and/or more packets are stacked in parallel within a cas-
sette. Note that widening the flow channel does not mean that the space between
membrane layers is made greater; instead, the membranes themselves are widened
giving flow more area to “spread out” so that larger volumetric flow rates can be
accommodated. In addition, individual cassettes can be installed in parallel if even
more surface area is required. With hollow-fiber modules, area is increased by using
modules of the same length but with more hollow fibers included in the module.
And like cassettes, more modules can be installed in parallel if needed. We discuss
scale-up in more detail shortly.
Table 9.1 contains information on various commercially available UF and MF
membranes for commercial-scale biopharmaceutical applications.

Table 9.1: Various commercially available UF and MF membranes, and their properties, for use in
biopharmaceutical manufacturing.

Membrane Reference Manufacturer Module Membrane MWCO or Reusable or

Format Material Pore Size Single-Use
Module

Polysulfone [, ] Cytiva Hollow Polysulfone – kDa Reusable and
fiber .–. μm single use

Ultracel® [, ] MilliporeSigma Cassette Composite –, kDa Reusable

RC cassettes and
single-use
Pellicon®
capsules
available

Biomax® [] MilliporeSigma Cassette Modified –, kDa Reusable

PES

Durapore® [] MilliporeSigma Cassette PVDF  kDa Reusable

.–. μm

Supor® [] Pall Cassette Modified .–. μm Reusable

PES
 9.4 Typical TFF procedures for cGMP manufacturing 265

Table 9.1 (continued)

Membrane Reference Manufacturer Module Membrane MWCO or Reusable or

Format Material Pore Size Single-Use
Module

Omega™ [] Pall Cassette Modified – kDa Reusable

PES

Delta [] Pall Cassette RC – kDa Reusable

ProStream [] Repligen Cassette Modified .– kDa Both single use
PES and reusable
cassettes
available

HyStream [] Repligen Cassette Modified – kDa, Both single use
PES .–. μm and reusable
cassettes
available

Modified [] Repligen Hollow Modified – kDa, Single use

PES fiber PES . μm

Mixed [] Repligen Hollow Mixed ., . μm Single use

cellulose fiber cellulose
ester ester

Polysulfone [] Repligen Hollow Polysulfone – kDa, Single use

fiber . μm

PES [] Repligen Hollow PES . μm Single use

fiber

Hydrosart® [, ] Sartorius Cassette Stabilized – kDa, Reusable

cellulose ., . μm

PESU [, ] Sartorius Cassette PES – kDa, Reusable

. μm

Note that RC is regenerated cellulose, PES is polyethersulfone, and PVDF is polyvinylidene difluoride.

9.4 Typical TFF procedures for cGMP manufacturing

In describing a typical TFF procedure, we start with concentration and DF by UF,

similar to what might be used for product formulation just prior to bulk fill. This
procedure can be readily adapted to UF steps at other points in a biopharmaceutical
process and to MF steps as well. More on that at the end of this section. A diagram
of the individual steps that make up an overall UF step for a biopharmaceutical pro-
cess is shown in Figure 9.7.
266 Chapter 9 Formulation operations: ultrafiltration

Install membrane module(s)

Flush with water

Clean/sanitize new membranes

Flush with water

Perform integrity test

Measure Normalized Water Permeability (NWP)

for new membrane

Equilibrate system

Recover product from the UF system

Reuse system
between
Clean/sanitize system batches of the
same product

Perform integrity test*

Measure the NWP for the used membrane module(s)*

Store the system for reuse

Figure 9.7: A typical procedure for tangential-flow UF used for concentration and buffer exchange of
product. The steps shown apply to the case in which membranes are reused. Note that steps with
an asterisk (*) may be executed prior to the following run, just before the system equilibration.
Image © NC State University; reprinted with permission.
 9.4 Typical TFF procedures for cGMP manufacturing 267

The first step is installation of the membrane module(s). Specific instructions

for installation are provided by the supplier. Modules from the supplier are often
shipped in a storage solution, such as glycerin or a low concentration of NaOH that
keeps the membrane wet – so it doesn’t dry out and collapse – and inhibits microbio-
logical growth. Modules are flushed with water after installation to remove storage
solution. Referring to Figure 9.2, the term flush means a process configuration in
which both the permeate and retentate lines are directed to drain, with some back-
pressure on the system. The flush configuration enables removal of material held up
in the UF system. We also use the term recirculate, which refers to a process configu-
ration in which liquid in the retentate, permeate, or both is returned to the feed vessel
and back to the membranes. Following the water flush, the membrane modules are
cleaned/sanitized. We cover cleaning and sanitization in more detail shortly.
After the initial cleaning and sanitization step, modules are again flushed with
water to remove cleaning and sanitization agents. To ensure that the membrane has
not been damaged during shipping and installation, an integrity test is performed.
A membrane without integrity is one that does not perform as designed; in particu-
lar, for the case of a UF step designed to concentrate and diafilter a product, a lack
of integrity would be any defect that allows product that should be retained to be
lost to permeate. A hole in the membrane or improper sealing of the permeate and
feed streams within the module are two causes of loss of integrity. An air diffusion
test is often used for UF membranes to assess integrity. This test involves wetting
the membrane with water then flowing air to the feed side of the membrane at a
specified pressure. The only way for air to be transported to the permeate side in a
completely wetted membrane with integrity is by the very slow process of diffusion.
Thus, air flow rates in excess of what would be measured by diffusion alone suggest
loss of membrane integrity. Suppliers provide test conditions and a specification for
the largest air flow rate allowable for a passing test. For example, a Pellicon® 2 UF
cassette containing Biomax® membrane with a MWCO of 10 kDa and area of 0.5 m2,
like the cassettes in the holder shown in Figure 9.6(d), must have an air flow (from
feed side to permeate) of ≤18 cm3/min at an air pressure of 10 psig [269].
Following integrity testing, the normalized water permeability (NWP) is mea-
sured. NWP serves as a measure of membrane cleanliness. It is performed by feed-
ing high purity water (HPW) or water for injection (WFI) to the UF membrane at the
same flow rate and TMP each time, then measuring the rate at which water perme-
ates. Simply put, the permeate rate for water remains relatively constant if the mem-
brane is cleaned effectively (i.e., no fouling from run to run) and declines if the
membrane becomes fouled.
The equation used to calculate NWP is as follows:

Qp × TCF
NWP = (9:3)
TMP × Am
268 Chapter 9 Formulation operations: ultrafiltration

where Qp is the volumetric flow rate of the permeate water typically measured in L/
h, TCF is a dimensionless temperature correction factor (values are usually available
from the membrane supplier), TMP is the transmembrane pressure as defined previ-
ously by equation (7.2) and typically measured in psig, and Am is the membrane
area in m2. Note that the ratio Qp/Am is the permeate flux of the water, which we
designate Jw. Referring to equation (7.7), the flux of water in an NWP measurement
with an unused membrane can be written Jw = TMP/(μRm). Note that Rtotal in equa-
tion (7.7) has been replaced by Rm, the intrinsic membrane resistance, because re-
sistance from a filter cake and internal membrane fouling would be nonexistent for
a water flux measurement. The flux of water becomes dependent on temperature
through the viscosity, μ. Higher temperatures reduce the viscosity of water thereby
increasing permeate flux. For example, the viscosity of water at 25 °C is 0.8903 cp,
while the viscosity of water at 18 °C is 1.0532 cp [270]. Based on these viscosity val-
ues, we would expect Jw at 25 °C to be 1.1830 times higher than Jw at 18 °C (Jw,25C/
Jw,18C = μ18C/μ25C = 1.1830), all other conditions being the same. The TCF corrects for
differences in permeate flow rates due to temperature by converting water flux val-
ues measured at any temperature to a standard reference temperature, typically
25 °C; thus, the TCF at 18 °C is 1.1830.
So if NWP is a measure of membrane cleanliness, why measure it on a new, un-
used membrane? Because NWP is often reported as a percent of the original value,
with the value measured immediately after installation/initial cleaning serving as
the original value:

NWPafter run
% of original NWP = × 100 (9:4)
NWPoriginal

The percent of NWP recovered is a parameter typically trended for a campaign using
the same set of membrane modules. For cassettes, it is common to see a decline in
NWP after their first use. This initial decline has been attributed to the cassette screen
compressing into the membrane and obstructing the area for permeation [190]. Be-
yond this initial decline, a steady decrease in NWP from run to run suggests ineffec-
tive membrane cleaning. Relatively constant NWP suggests effective cleaning and
that process permeate flux values will not decline significantly from run to run. NWP
as a percentage of the original value is typically used to define the membrane module
lifetime. The exact criteria differ from company to company, but once the NWP falls
below a predetermined percent of original value (for example, NWP < 70% of the orig-
inal), existing membrane modules are replaced with new modules.
Prior to exposing the UF system to product during concentration and DF, the
system is equilibrated by flushing and recirculating buffer. Equilibration is neces-
sary to remove water from the system (for the case in which new membranes are in
use) or storage solution from the previous run. Exposure of product to either water
or storage solution can lead to precipitation and/or denaturation, which is to be
avoided. It is worth noting that the pre-concentration/DF steps just described may
 9.4 Typical TFF procedures for cGMP manufacturing 269

be greatly reduced by using a pre-sterilized single-use, self-contained TFF module,

which may only need to be installed and flushed with equilibration buffer before
the concentration/DF steps [271].
Following equilibration, concentration and/or DF take place. We discuss these
steps in more detail in the sections on parameters and process development. Following
the concentration and DF steps, product must be recovered from the UF system. Given
the value of biopharmaceutical products, it is critical to maximize their recovery while
ensuring product integrity is maintained and final concentration targets are met. Prod-
uct recovery begins by depolarizing the membrane. Depolarization involves recirculat-
ing the retentate, which resides in the feed tank, with the permeate valve closed. This
procedure allows product that is concentrated at the membrane surface to diffuse back
into the bulk retentate. Next, the system contents are collected in a separate vessel.
This can be accomplished by collecting the retentate from a drain located at a low
point on the system. We refer to product collected as the recovered retentate.
At this point in the procedure, most of the product has been removed from the
system, but residual product may remain in the system hold-up volume. A number of
different methods can be used to recover this residual, including blowing down the
system with air to push remaining product out of the system, flushing buffer through
the feed/retentate to displace and collect the residual, and recirculating buffer
through the system and collecting the resulting solution. Residual product recovered
by air blowdown is at the same concentration as the recovered retentate and can be
added back to that retentate directly. However, the residual product collected from
the buffer flush or recirculation steps contains product at a concentration less than in
the recovered retentate, which presents a problem: how can that product be returned
to the collected retentate, so that it is not wasted, while also achieving the target con-
centration for the step? This can be accomplished by concentrating to slightly above
the target – while ensuring no impact on product quality – then using the recovered
product-containing buffer to dilute recovered retentate to the target concentration.
For example, if the target concentration for final formulation is 10 mg of product/mL
solution, then the product might be concentrated to 12 mg/mL and subsequently di-
luted to 10 mg/mL with the buffer recovered from the flush or recirculation steps.
The following equation can be used to calculate the amount of recovered buffer,
Vb, at product concentration Cb, that can be added to an overconcentrated recov-
ered retentate to achieve the target final product concentration, Cfinal:

ðCr − Cfinal ÞVr

Vb = (9:5)
Cfinal − Cb
Cr is the product concentration in the recovered overconcentrated retentate and Vr
is the collected volume of the recovered retentate. Note that Cr and Cb must be mea-
sured (for example, in the quality control laboratory) to use equation (9.5). In prac-
tice, it is desirable to minimize overconcentration, while at the same time adding as
much of the recovered buffer flush as possible back to the recovered retentate to
270 Chapter 9 Formulation operations: ultrafiltration

maximize step yield. To write a procedure that consistently achieves the target con-
centration and maximizes step yield, a typical value for Cb is required. This value
depends on the amount of buffer used for the flush or recirculation steps. Knowl-
edge of the typical value for Cb along with the buffer volume used for flush or recir-
culation will determine the appropriate product concentration value to target in the
overconcentrated retentate. That value is necessary for the UF procedure in a cGMP
process. At-scale, engineering runs that precede cGMP runs provide a good opportu-
nity to define this overconcentration procedure. Note that in a well-design UF pro-
cess, step yields of >95% are readily achievable.
After product recovery, membranes are cleaned and sanitized to minimize car-
ryover of the product from one batch into the next, to keep process permeate flux
values high upon reuse, and to kill microbial contaminants. Cleaning and sanitiza-
tion of membranes can be conducted as a single step or two different steps, and the
solutions used depend on the membrane type. Cleaning typically involves flushing
and recirculating cleaning/sanitization solution; the sanitization step may also in-
volve a hold. Membrane manufacturers typically offer recommendations for clean-
ing and sanitization procedures and solutions. This information can be found in the
user’s guide for the cassette or hollow-fiber module. For example, for Pellicon® 2
cassettes with Biomax® membranes, MilliporeSigma recommends 0.1 M NaOH/250
ppm NaOCl – 0. 5 M NaOH/600 ppm NaOCl at 20–50 °C for cleaning, among a num-
ber of other possible cleaning agents. They also recommend sanitization be per-
formed as a separate step after cleaning, using NaOH between 0. 5 M and 1.0 M at
20–50 °C, among a number of other options [269]. Of course, for a commercial bio-
pharmaceutical product, the cleaning procedure must be validated as described in
Chapter 3. Following cleaning and sanitization, the system is flushed with water to
remove residual cleaning and sanitization agents, and a post-use integrity test and
NWP measurement are performed.
Reusable membranes are typically stored in a liquid solution to keep the mem-
brane wet and inhibit microbial growth until their next use. The storage step should
include a flushing procedure to remove the cleaning/sanitization solution or water
(if a water flush is performed) introduced in the previous step followed by recircula-
tion. As with cleaning and sanitization, manufacturers often recommend a storage
procedure and solutions. For the Pellicon® 2 cassettes with Biomax® membranes
mentioned previously, 0.1 M NaOH is recommended [269].
It is important to note that reusable membrane modules are typically used for
only a single product. If modules were used for multiple products, then there is the
risk that Product A adsorbs on the polymeric membrane and is released into a sub-
sequent batch of Product B.
Before concluding this section, let’s briefly summarize how a tangential-flow
MF procedure differs from the typical UF procedure depicted in Figure 9.7:
 9.5 Process and performance parameters for tangential-flow UF and MF 271

– For an MF step, the concentration/DF portion of the procedure is replaced by a

solid-liquid separation such as clarification of a cell culture broth containing ex-
tracellular product. For this example, product permeates; therefore the permeate
stream is collected rather than being sent to drain as in a UF step used for con-
centration and DF. Note that DF may be part of a tangential-flow MF step used
for clarification, but the objective is enhanced product recovery from the reten-
tate, once most of the product has permeated, rather than buffer exchange.
– The step yield for a tangential-flow MF step used for clarification is calculated as
follows because the product permeates:

Cp V p
% step yield = × 100 (9:6)
Cf V f

where Cp is the measured product concentration in the permeate, Vp is the vol-

ume of permeate collected, and Cf and Vf are the concentration of product in
and volume of the initial feed.
– Whereas the integrity of a UF membrane is typically measured by an air diffu-
sion test, for an MF membrane, a bubble point test may be used. Bubble point
refers to the pressure at which a wetting liquid, such as water, is pushed out of
the pores of a membrane. The supplier specifies the minimum bubble point
value for a particular wetting agent that demonstrates membrane integrity. The
bubble point pressure in a membrane pore is inversely proportional to the pore
diameter. The larger pores of the MF membrane allow for a bubble point pres-
sure that does not exceed the pressure rating for the membrane module, as it
would in a UF membrane with its much smaller pores.

9.5 Process and performance parameters for tangential-flow

UF and MF
The purpose of this section is to discuss process and performance parameters that
are of particular importance to tangential-flow UF and MF steps. We actually started
this discussion in the previous section by defining a couple of performance parame-
ters for both UF and MF: step yield and NWP. Let’s continue the discussion of pa-
rameters by considering the diagram in Figure 9.8, which represents flow of retained
product between two UF membranes, similar to the feed channel of a cassette. As dis-
cussed, flow on the feed side of the membrane is driven by a pump. The system is
operated so that the feed-side pressure is higher than the permeate-side pressure,
which forces liquid to permeate the membrane. Pf is greater than Pr, and the average
pressure on the feed side of the membrane, (Pf + Pr)/2 is greater than the perme-
ate-side pressure, Pp; that is, the TMP > 0. As liquid permeates the membrane, the
concentration of the product at the membrane surface, Cw, increases because the
permeating liquid carries dissolved product to the membrane surface, where it is
272 Chapter 9 Formulation operations: ultrafiltration

retained. Thus, Cw >> Cfeed channel. Elevated concentration at the membrane surface
is referred to as polarization, a term used previously in our discussion of the prod-
uct recovery procedure. Note that the same polarization occurs in a tangential-
flow MF system, except that it is solids that are accumulating at the membrane
surface rather than dissolved components (i.e., product).

Membrane Qf
Qr
Qf Qr Qp
Pf Cfeed channel Pr
CW Pf = pressure of the feed at the membrane inlet
Pr = pressure of the retentate (i.e., what’s left of the feed)
Membrane Qp at the membrane outlet
Cp Pp = permeate pressure
Pp
Cfeed channel = product concentration in the bulk feed side
of the membrane
CW = product concentration at membrane wall
Cp = product concentration in permeate

Figure 9.8: Schematic of flows, pressures, and product concentrations in flow between two UF
membranes. Image © NC State University; reprinted with permission.

UF is typically operated under TMP control; that is, the TMP is a process parameter.
Under TMP control, the permeate flux, designated Jp, is an important performance
parameter. It is desirable to maximize the permeate flux because high flux leads to
short processing times and/or lower membrane area requirements as the example
below illustrates.

Example: Estimating the time to concentrate a solution of viral vectors for a gene therapy product
A bench-scale UF system is used to concentrate a feed of adeno-associated virus serotype 2
(AAV2) vectors, used in gene therapies, from 8.5 × 1010 vg/mL to 8.5 × 1011 vg/mL. (Note: vg
stands for vector genomes, which is the number of AAV2 capsids that contain the gene of inter-
est). The membrane is 100 kDa MWCO regenerated cellulose with an area of 88 cm2. If the start-
ing feed is 2 L and the average permeate flux during concentration is 150 L/m2/h, how long will
it take to complete the concentration step? Assume no loss of AAV2 in the step.

Solution
First, determine how much feed must be removed as permeate to meet the concentration target
of 8.5 × 1011 vg/mL. Once the volume permeated is determined, the process time can be calcu-
lated using equation (6.7) (tprocess = Vfeed /Q), rewritten for the special case of a membrane sepa-
ration as:

Vp
tprocess = (9:7)
Jp Am
 9.5 Process and performance parameters for tangential-flow UF and MF 273

where Vp is the volume permeated, Jp is the permeate flux, and Am is the membrane area. The
retentate volume at the end of the concentration step is calculated from equation (9.1) as Vr =
(8.5 × 1010 vg/mL × 2,000 mL)/ 8.5 × 1011 vg/mL = 200 mL. So the volume permeated is 2,000 mL
– 200 mL, or 1,800 mL. Substituting these values into equation (9.7) results in:

1.8 L
=
150 L=ðm2 hÞ × 0.0088 m2
= 1.4 h
Obviously if the permeate flux is <150 L/m2/h, the time required for the UF step increases.

We’ve already discussed permeate flux in tangential-flow MF in Chapter 7 (refer to

equation (7.7)). For UF, the permeate flux during a concentration or DF step is
strongly dependent on the extent of polarization, among a number of other varia-
bles. A polarization layer adds resistance to permeate flow, similar to what was dis-
cussed for accumulation of particles at the surface of an MF membrane. In addition,
the accumulation of solute at the membrane wall creates an osmotic pressure gradi-
ent that opposes the TMP driving force. Osmotic pressure refers to the pressure that
would have to be applied to prevent a solvent, in our case water, from moving from
a solution of low solute concentration (permeate side) to high solute concentration
(feed side). In the case of UF, because solute concentration is significantly higher
on the feed side of the membrane than the permeate side, pressure must be applied
just to prevent water from flowing in the opposite of the desired direction – from
the permeate to the feed side of the membrane. With these polarization effects in
mind, an equation for the permeate flux, Jp, for UF can be written similar to equa-
tion (7.7) for MF [272]

ðTMP − ΔπÞ
Jp = (9:8)
μRtotal

where the variables are defined as previously in equation (7.7), with the exception
of Δπ, which is the osmotic pressure difference for the solute concentration at the
feed side and permeate side of the membrane surface. Its value can be quite high.
For example, monoclonal antibodies are often formulated at high concentration
(e.g., 200 g/L) so that only small volumes are required for subcutaneous administra-
tion. Binabaje et al. measured the osmotic pressure of a 200 g/L monoclonal antibody
solution to be as high as 7 psi [273]. In the presence of concentration polarization, the
osmotic pressure of a 200 g/L monoclonal antibody solution will be even higher, lead-
ing to a significant pressure difference that must be overcome to generate permeate
flux. Note also that in the case of UF, Rtotal is the sum of resistances from the mem-
brane and from any fouling that occurs, such as adsorption of solutes to the pore
walls and gel layer formation.
274 Chapter 9 Formulation operations: ultrafiltration

Three input parameters that impact the process permeate flux, and therefore
the process time, in tangential-flow UF are TMP, crossflow rate, and product con-
centration. Note that crossflow rate refers to the flow rate at which solution flows
through the feed channel of the membrane. To understand the impact of each of
these process parameters on permeate flux, consider the plots in Figure 9.9. As dis-
cussed previously, the flux of water across a membrane is given by the equation Jw
= TMP/(μRm), which shows that permeate flux varies linearly with TMP. This rela-
tionship is shown in Figure 9.9. When a solute is present that is retained by the
membrane, such as a biopharmaceutical product, the permeate flux initially in-
creases with TMP and then levels off, also shown in Figure 9.9. The levelling off oc-
curs because of concentration polarization, just discussed, which leads to a high
osmotic pressure gradient and greater resistance to permeation. Further increases
in TMP simply increase polarization, which counteracts the additional driving force
for permeate flow from the increased TMP.
Figure 9.9 also shows the impact of crossflow rate and product concentration in
the feed on permeate flux. Higher feed rates result in higher permeate flux values
than lower feed rates as a result of the enhanced sweeping action across the surface
of the membrane, which minimizes polarization. Higher product concentration, on
the other hand, leads to lower permeate flux because polarization effects are more
significant. For concentration/DF steps under TMP control, the TMP target value is
usually chosen at a point between the pressure-dependent and pressure-indepen-
dent portion of the curve, as shown in Figure 9.9.

Low solute or
Water particle concentration,
2 h)

Optimum
operating
Optimum point
operating
point
High solute or
particle concentration,

Transmembrane pressure (psi)

Figure 9.9: Typical permeate flux vs. TMP curves for TFF operations. Note that low or high solute
concentration applies to UF, and low or high particle concentration to MF. Image © NC State
University; reprinted with permission.
 9.5 Process and performance parameters for tangential-flow UF and MF 275

In addition to TMP and crossflow rate, a number of other process parameters

are controlled for the concentration and DF steps. Two important ones – concentra-
tion factor (CF) and DF volumes – are summarized in Table 9.2.

Table 9.2: Additional process parameters that are controlled during concentration and DF steps.

Parameter Definition Comments

Concentration CF: Cr =Cf (9:9) – For final formulation, the final

factor (CF) and product concentration in the UF
VCF: Vf =Vr (9:10)
Volume system, Cr, is dictated by what is
concentration required to formulate drug
factor (VCF) CF = VCF, assuming no loss of product to substance.
the permeate – A concentration target is usually
Note that in these equations, Cf and Cr met by targeting a final retentate
refer to the product concentration in the volume, which is determined by
initial feed and final retentate, assuming VCF = CF.
respectively. Like Vf and Vr are the
volumes of the initial feed and final
retentate.

DF volumes or DV: Vdf buffer =Vr (9:11) – In DF, the number of DVs dictates
diavolumes the extent of buffer exchange. For
(DVs) buffer components that freely
In this equation, Vdf buffer is the volume
permeate the membrane, almost
of DF buffer fed and Vr is the retentate
95% of buffer exchange occurs after
volume at the time of DF.
only 3 DVs. For “complete”
exchange, 7–12 DVs are common.
– For constant-volume DF, DV is often
monitored and controlled by
measuring Vp collected during DF
and assuming Vdf buffer = Vp.

For final formulation steps, the final production concentration, and therefore CF, is
dictated by the product concentration required for formulation of drug product.
Typically in a manufacturing process, the target concentration is achieved via the
volume concentration factor (VCF) in equation (9.10) (see Table 9.2). The VCF is as-
sumed to equal the required CF, and from the VCF, a final retentate volume is calcu-
lated and targeted. This strategy is used because volume (or weight) of retentate is
more readily measured inline (and in real time) than is product concentration in the
retentate.
During DF, the extent of buffer exchange depends on the number of DF volumes
(DVs) used, defined by equation (9.11) in Table 9.2. In a production process using
constant-volume DF, DVs, or alternatively the volume of buffer added based on the
number of DVs required, are often monitored by measuring the value of Vp during
DF and recognizing it is equal to the volume of buffer added in a constant-volume
276 Chapter 9 Formulation operations: ultrafiltration

DF setup. So how many DVs are required for complete buffer exchange? For con-
stant-volume DF, the theoretical extent to which a solute targeted for removal, such
as a buffer component or a salt, is remaining or removed is described by the follow-
ing equations:

Csolute,rf
% solute remaining = × 100 = exp½ðR − 1ÞDV × 100 (9:12)
Csolute,ro

% solute removed = ½1 − expðR − 1ÞDV × 100 (9:13)

where Csolute,rf and Csolute,ro are the solute concentrations in the final and initial re-
tentate during DF, respectively, and R is the rejection (also referred to as retention)
of the solute. Rejection is defined as

Csolute,p
R=1− (9:14)
Csolute,r

where Csolute,p and Csolute,r are the concentrations of the solute in the permeate
and the retentate being fed to the membrane, respectively. To avoid ambiguity,
we note the assumption that in the following analysis, the solute concentration in
the feed tank, feed line, and retentate is similar enough to be considered the
same and is designated Csolute,r. If a buffer component or salt passes through the
membrane unhindered, its concentration in the permeate and retentate will be
the same, and R = 0. If a solute is completely retained by the membrane, its con-
centration in the permeate is zero, and R = 1. Equation (9.12) (and by extension,
equation (9.13)) is derived by performing a mass balance on the solute that is re-
moved during DF, such as a salt or buffer component, over the DF system, as
shown in Figure 9.10(a). That is, the accumulation of the solute (=d(VrCsolute,r)/
dt) during DF equals its input (in the DF buffer) minus its output (in the perme-
ate). In the case of DF, the rate of input of a solute to be removed is zero, and the
output is Csolute,p × Qp, which can be rewritten as (1-R) × Csolute,r × Qp. For a con-
stant-volume DF, in which the rate of buffer addition is equal to the permeate
flow rate, the resulting differential equation can be solved to obtain equation
(9.12). Additional details on deriving equations (9.12) and (9.13) are included in
the questions at the end of this chapter.
Figure 9.10(b) plots the results of equation (9.12) for various R values for the sol-
ute. As you can see, for a solute that freely permeates the membrane (R = 0), only
5% of that solute remains in the feed after 3 DV; that is, 95% of the solute has been
removed. And after 12 DV, >99.999% of the solute has been removed. When per-
forming buffer exchange for final formulation of product, it is common to use 7–12
DVs to ensure as complete an exchange as possible.
 9.5 Process and performance parameters for tangential-flow UF and MF 277

(a)

Solute accumulation
DF buffer Feed tank* Retentate
Retentate
tank valve
stream

Permeate
stream
(to waste)

Solute
output
Solute Feed UF membrane
input = 0 stream Permeate
Buffer feed Feed or pump
pump recirculation
pump
*Also referred to as recycle tank and recirculation tank

(b)
100.000

10.000
Solute retained
(% of original)

1.000

0.100
Rejection = 0.4

0.010
Rejection = 0 Rejection = 0.2

0.001
0 2 4 6 8 10 12 14 16
Diavolumes

Figure 9.10: (a) The boundaries of the component balance used to derive equation (9.12). (b)
Graphical results of equation (9.12). Images © NC State University; reprinted with permission.

Example: Estimating the concentration of a buffering agent at the end of a DF step

50 L of a solution containing a protein therapeutic in 50 mM tris is diafiltered with a buffer that
contains no tris. If 200 L of DF buffer is used, estimate the concentration of tris at the end of the
DF step. Assume that tris freely permeates the UF membrane being used.
278 Chapter 9 Formulation operations: ultrafiltration

Solution
To begin, solve equation (9.12) for Csolute,rf, which represents the concentration of tris at the end
of DF:

Csolute, rf = exp½ðR − 1ÞDV × Csolute, ro

Recognize that because tris freely permeates the membrane, R = 0. Further, 200 L of DF buffer
added against 50 L of protein solution equates to 4 DVs of tris-free buffer added. Substituting
these values along with Csolute,ro = 50 mM leads to:

Csolute, rf = exp½ð0 − 1Þ × 4 × 50 mM tris

= 0.92 mM tris
This value represents a significant reduction from the 50 mM starting value.

Note that throughout this chapter, we have indicated that concentration of product
precedes DF. But the reverse sequence, in which DF precedes concentration, is also
an option. In fact, there is an optimal product concentration at which to perform
buffer exchange that will minimize the DF process time; specifically, it is desirable
to minimize the ratio of permeate volume (thus minimizing the amount of DF buffer
that must be fed during the DF step) to permeate flux (thus maximizing the rate of
permeation during DF). There are both empirical and theoretical methods for deter-
mining this optimal product concentration for minimizing DF time. Those details
can be found elsewhere [190, 244]. To take advantage of this optimal product con-
centration, a UF step may be designed to perform an initial concentration to achieve
the optimal product concentration for DF, followed by DF for buffer exchange, fol-
lowed by a final concentration step. As mentioned previously, this final concentra-
tion step may be designed to “overconcentrate” – that is, concentrate product to a
value slightly greater than the target for bulk drug substance – so that the recovered
product in the buffer from the flush or recirculation steps may be added back to the
recovered retentate. As discussed previously, addition of the recovered buffer di-
lutes the overconcentrated retentate to the target product concentration while at the
same time adding recovered residual product held up in the UF system back to the
product stream.
Much of the previous discussion on process and performance parameters for
tangential-flow UF applies to tangential-flow MF as well, with a few differences.
These differences are considered in the following points:
– The schematic in Figure 9.8 applies to tangential-flow MF, except that the profile
for product concentration (i.e., Cfeed channel and Cw) in the UF scenario applies to
particle (not product) concentration in the MF scenario.
– The permeate flux vs. TMP curves in Figure 9.9 apply to tangential-flow MF,
with the exception that product concentration would be replaced by particle
concentration.
 9.6 TFF development and scale-up 279

– For tangential-flow MF, the CF defined in Table 9.2 applies to particulate. For
MF steps used to clarify an extracellular product, the product is not retained but
should freely permeate the MF membrane. Therefore, CF and VCF are not used
as process parameters. Note, however, that dilution of the permeated product
from DF buffer during an MF step does occur, and there is a tradeoff between
step yield and product concentration; that is, the more DVs used for product re-
covery in the permeate, the more dilute the product becomes.
– Equations (9.12) and (9.13) apply to tangential-flow MF, but with a twist. For
clarification of extracellular product, the solute being removed is the product.
So equation (9.13) can be used to predict the percent of product remaining in the
feed vessel just prior to DF that will be recovered as a function of DVs (i.e., the
% solute removed in equation (9.13) for UF = % of product recovered for MF of
extracellular product).

It is also important to note that while feed flow rate and TMP are frequently con-
trolled parameters that are specified in a batch record and/or under automated con-
trol in a UF step for concentration/DF, tangential-flow MF may rely on permeate
flow control. Because of the relatively large pore size of MF membranes, permeabil-
ity is higher than with UF membranes, which results in a high permeate flux at the
outset of an MF run. In other words, most of the feed is converted to permeate. This
high initial permeate flux pushes particles onto the surface of the membrane and
into the pores, resulting in significant fouling. Consequently, rather than control
TMP in MF steps, permeate flow rate is often controlled by using a pump on the per-
meate line to maintain flux at a level that avoids significant membrane fouling.
Under this scenario, the TMP is a response to the permeate flow rate and is not a
controllable process parameter.

9.6 TFF development and scale-up

The goal of most tangential-flow UF and MF development efforts for biopharmaceu-

tical processing is to design a step that consistently meets the target concentration
of product and the correct extent of buffer exchange (for UF), or an acceptable prod-
uct clarity (for MF), and that results in acceptable product quality, high product re-
covery (i.e., step yield), and high permeate flux with reproducible process times. Of
course for UF, high product retention is needed for high step yields, while for clarifi-
cation using MF, low product retention is needed for high step yields. To produce
acceptable performance, an appropriate membrane and module must be selected.
In addition, process parameters and their ranges must be identified that lead to ac-
ceptable values for performance parameters such as step yield, process time, and
various quality attributes.
280 Chapter 9 Formulation operations: ultrafiltration

For TFF steps, development studies are likely to begin by screening different
membranes. In the particular case of concentration/DF of proteins by UF, a rule of
thumb is to select a membrane with a MWCO that is one-third to one-fifth of the
molecular weight of the target protein [244]. This guideline should lead to good
product retention while simultaneously allowing a high permeate flux. For clarifica-
tion by MF, select a membrane pore size that maximizes particle retention, mini-
mizes product retention, and maximizes process permeate flux. If using a membrane
cassette with a turbulence-promoting screen on the feed side, the type of screen to be
used is another choice that must be made. Also, the membrane chemistry must be
chosen as well. Advantages and disadvantages of two of the most common membrane
materials, regenerated cellulose and modified polyethersulfone, were presented pre-
viously. Clearly, there are a number of variables to consider when choosing the right
membrane (MWCO, pore size, membrane chemistry, module type, turbulence promot-
ing screens), and there is value in experimentally screening different membranes and
modules to select the one that is best suited for a particular tangential-flow UF or MF
application.
Following membrane screening, additional studies and information gathering
are required to identify and determine ranges for process parameters that lead to
acceptable values for performance parameters and quality attributes. Guidelines on
development activities for the various steps involved in a typical UF procedure for
concentration and DF of product are detailed in Tables 9.3 and 9.4. These tables
present an expanded list of process parameters, relative to what was presented in
the previous section, that should be considered. They also include performance pa-
rameters and quality attributes that are potentially impacted and strategies for iden-
tifying “ranges” for these parameters. These development studies, including the
membrane screening previously described, are typically conducted at bench scale
using systems that can reliably be scaled to manufacturing. Most of the membranes
and modules listed in Table 9.1 come in sizes suitable for these bench-scale studies,
and there are a number of TFF units commercially available that are likewise suit-
able for use with the smaller volumes typically used to conduct these bench-scale
studies (<1 L). It is also worth noting that suppliers can provide significant input on
membrane screening and parameter setting for each of the procedural steps via ven-
dor-performed studies for a specific process, user manuals, technical briefs, and
other sources of information.
It is worth noting that many of the process parameters listed in Tables 9.3 and
9.4 likely will have no effect on product quality. They are in place primarily to en-
sure processing consistency. Process parameters in Tables 9.3 and 9.4 that may be
critical – that is, must be controlled tightly to ensure drug substance critical quality
attributes meet their specifications – include the CF and DVs.
Because development studies are typically done at bench scale, scale-up is often
required for execution at production scale. Matching the optimized performance de-
veloped at bench scale is a primary objective of scale-up. To properly scale up, the
 9.6 TFF development and scale-up 281

Table 9.3: Process development activities by step in a UF procedure for concentration and DF, from
membrane installation to equilibration.

Step in UF Process Parameters Performance Source of Process

Procedure Parameter(s)/Quality Parameter Information
Attribute Potentially
Affected

Installation – Membrane/module Step yield, process time Membrane screening

type (for membrane/module studies
– Module area type and area) Scale-up calculations
– Torquing requirements Sealing, integrity Supplier instructions
for cassettes

Cleaning/ – Cleaning/sanitization For all process parameters Supplier

sanitization agents listed, corresponding recommendations; studies
– Cleaning/sanitization performance parameters at small scale to
agent concentration are process time, demonstrate
– Cleaning solution membrane lifetime, effectiveness, assessed by
temperature impurities from previous NWP measurement,
– Recirculation time batches, bioburden product carryover
– Flush volume measurement, sanitization
studies, etc.

Integrity test – Test pressure (for air Process parameters in Supplier instructions
diffusion test) place to ensure consistent
– Allowable air flow rate and accurate
(for air diffusion test) measurement of integrity

NWP – Feed rate Process parameters in Supplier instructions, with

– TMP place to ensure consistent the exception of pass/fail
– Pass/fail criterion (% and accurate criterion, which is
of original) measurement of NWP established by user

Equilibration – Buffer For all process parameters User chooses buffer

– Feed flow rate listed, corresponding solution-typically the
– TMP quality attribute is buffer in which the
– Recirculation time, product-related impurities. product is in or DF buffer;
flush volume Removal of previous supplier recommendations
system contents (e.g.,
water, storage solution)
during equilibration
minimizes formation of
these impurities.

Refer to Figure 9.7 for the procedure on which this table is based (adapted from Lutz, Herb, 2015,
Ultrafiltration for Bioprocessing) [190].
282 Chapter 9 Formulation operations: ultrafiltration

Table 9.4: Process development activities by step in a UF procedure for concentration and DF, from
the concentration step to system storage.

Step in UF Process Parameters/Performance Source of Process Parameter

procedure Parameters or Quality Attributes Information
Potentially Affected

Concentration/ – TMP (for TMP control)/Process time For concentration, perform TMP
DF – Permeate flow rate (for permeate flux scouting studies (Jp vs. TMP at
control)/Process time various flow rates as shown in
– Crossflow rate (Qf)/Process time Figure .) at initial and final
– Concentration factor (CF) or product concentration to choose
concentration end point (Cr)/Product TMP and Qf. These flux studies are
concentration, process time, conducted with permeate recycled
aggregates to the feed vessel.
– DF buffer/Drug substance formulation For concentration, concentrate
(pH, osmolality, conductivity) product to final target value.
– Diavolumes/Extent of buffer Measure Jp vs. time (or Cr), Cp to
exchange (pH, osmolality, ensure acceptable rejection of
conductivity), process time, product and to ensure final target
aggregates concentration can be consistently
met. Also ensure that step yield and
product quality are acceptable.
For DF, perform TMP scouting
studies (Jp vs. TMP at various flow
rates as shown in Figure .) at
product concentration during DF to
choose TMP and Qf.
Diafilter product with buffer to
determine DVs that produce
acceptable buffer exchange with no
loss of product or product quality.

Product recovery – Depolarization flow rate, TMP/Step Bench studies to determine

yield parameter values that maximize %
– Buffer flush volume or recirculation recovery of product while
volume and flow rate/Product minimizing product dilution
recovery

Cleaning/ See Table . See Table .

sanitization,
NWP, Integrity
test
 9.6 TFF development and scale-up 283

Table 9.4 (continued)

Step in UF Process Parameters/Performance Source of Process Parameter

procedure Parameters or Quality Attributes Information
Potentially Affected

Storage – Storage solution and concentration/ Supplier instructions

membrane lifetime and bioburden
– Flow rate/ membrane lifetime and
bioburden
– Recirculation time/ membrane
lifetime and bioburden
– Flush volume/ membrane lifetime and
bioburden
– Allowable hold time between runs/
membrane lifetime and bioburden

Refer to Figure 9.7 for the procedure on which this table is based (adapted from Lutz, Herb, 2015,
Ultrafiltration for Bioprocessing) [190].

pressure, flow, and product concentration profiles (or particulate concentration for
MF) along the length of the membrane module must be maintained across scales dur-
ing concentration and DF [190, 274]. To maintain these profiles across scales, a scale-
up method that maintains feed-side velocity and channel length on scale-up can be
used. The following steps apply in general:
1. Optimize the UF or MF step at lab scale by performing the studies described
previously.
2. Use the same membrane (chemistry, MWCO or pore size) and similar module
(geometry, feed channel or lumen dimensions, turbulence-promoting screen if
applicable) as used in bench-scale development/optimization studies. The TMP
is scale independent and should be transferred directly if operating under TMP
control.
3. Maintain Vf/Am between scales, where Vf is the initial volume of material fed to
the membrane unit and Am is the membrane area.
4. Maintain the length of the flow path between scales. Suppliers offer cassettes
and hollow-fiber modules with small surface areas suitable for bench-scale stud-
ies and much larger areas suited to production applications (as has been de-
scribed previously) while keeping the feed path length constant. Figure 9.6(b)
shows how this is done for Pellicon® 3 cassettes. It is clear in that photo that
membrane area is increased by adding feed channels (resulting in thicker cas-
settes) and widening feed channel width (resulting in wider cassettes) while
keeping the length of the flow path the same.
5. Maintain the volumetric feed rate per membrane area (e.g., L/m2/h) between
scales. As a result, the feed flow rate (L/h) (obviously) increases with increased
membrane area.
284 Chapter 9 Formulation operations: ultrafiltration

Membrane area requirements for cGMP production applications vary depending on

the product. There are UF systems in use for commercial manufacturing with <1 m2
of membrane area and much larger systems with up to 120 m2 of UF membrane area
in operation commercially [190].

9.7 Validation considerations

TFF systems must go through installation qualification (IQ) and operational qualifi-
cation (OQ) as a prerequisite to process performance qualification (PPQ) runs as de-
scribed in Chapter 3. There are numerous tests to be conducted during IQ and OQ,
and a detailed listing of these tests is beyond the scope of this book. However, it is
worth noting that during OQ, it is particularly important to check the following for
UF and MF systems: that the feed tank on the system provides good mixing (this is
particularly important to ensure adequate buffer exchange), and that product can
be readily drained from the TFF system. In addition, the minimum and maximum
operating volumes of the equipment must align with what is required for the pro-
cess. It is particularly important to confirm that a UF system is capable of accurately
measuring the volume that will be targeted for a concentration step. PDA Technical
Report No. 15 recommends these and a number of tests to consider as part of a TFF
system OQ [275].
Also as discussed in Chapter 3, PPQ runs combine the actual facility, utilities,
qualified equipment, and trained personnel to produce commercial batches. How-
ever, enhanced sampling, testing, and monitoring of process and performance pa-
rameters occur during a PPQ run compared to a normal commercial batch. For
example, you might also test for aggregates during PPQ to ensure that the scaled-up
version of the TFF step is not adversely impacting product quality. This test would
be in addition to monitoring performance parameters such as process permeate
flux, process time, and step yield, as well as drug substance quality attributes such
as product concentration, extent of buffer exchange (e.g., pH and conductivity),
and bioburden as would be monitored in a routine commercial batch. Upon demon-
strating that aggregate levels remain within expected limits during PPQ, you may
choose to stop testing for aggregates at the UF step during routine commercial
production.
Cleaning validation for TFF systems typically takes place concurrently with PPQ
runs and demonstrates that the cleaning/sanitization procedure for the system (in-
cluding both membrane modules(s) and skid) ensures that at process scale there is
minimal carryover of product from one batch to another using methods described
previously in Chapter 3; that bioburden is controlled within acceptable limits; that
the process permeate flux is recoverable after each run; and that cleaning/sanitizing
agents are removed. Note that an effective cleaning/sanitization procedure would
have been developed at bench-scale prior to the validation runs. That bench-scale
 9.7 Validation considerations 285

evaluation would include an assessment of the compatibility of cleaning/sanitization

agents with the membrane as well as sanitization studies that look at the effective-
ness of the sanitizing agent, often the same as the cleaning agent, for reducing micro-
bial growth.
For membranes that are used for multiple runs of the same product, validation
studies are also required to establish a membrane lifetime (i.e., the maximum num-
ber of runs). The evaluation may start with bench-scale studies aimed at assessing
how reuse affects membrane performance. Starting at bench scale offers the advan-
tage of allowing for adjustment of cleaning/sanitization and storage procedures if
they are leading to membrane degradation. The maximum time that membranes
can be exposed to cleaning/sanitization and storage solutions should also be deter-
mined as part of a membrane lifetime study. The supplier should be able to supply
data to support setting these limits. Ultimately, the membrane lifetime should be
validated under protocol at process scale by closely monitoring appropriate perfor-
mance parameters and ensuring they remain within applicable ranges throughout
the membrane lifetime. The example below considers the design of such a protocol.

Example: Designing a UF membrane lifetime protocol

A validation engineer is writing a lifetime protocol to validate the maximum number of uses for a
UF membrane in a process for a new vaccine. The protocol will be executed concurrently with
production of commercial batches. Suggest parameters to be monitored at production scale to
establish the lifetime.

Solution
The goal of the study is to demonstrate that performance of the membrane remains consistent
throughout its lifetime. To that end, the following performance parameters may be monitored
for each batch processed up to the maximum number of runs being validated. Note that accep-
tance criteria would have to be established for each.
– NWP
– Permeate flux during concentration and DF
– Membrane integrity
– Step yield

In addition, monitoring bioburden and endotoxin as part of the UF step is useful to assess the
ability of the sanitization procedure to control microbial contamination throughout the lifetime
of the membrane. Finally, it is also recommended to assess carryover of product from batch to
batch to ensure it remains within acceptable limits. One method to accomplish the carryover
assessment is by performing blank runs at some pre-determined frequency, including a blank
run at the end of membrane lifetime. Similar to the blank runs described in Chapter 8 for chro-
matography, these runs would use the cGMP normal procedure except that product replaced by
a buffer. The presence of product or impurities in the resulting “buffer product” suggests carry-
over from one batch to the next.
286 Chapter 9 Formulation operations: ultrafiltration

9.8 Single-pass TFF

In the UF process configurations we have discussed previously (batch concentra-

tion, constant-volume DF, and fed-batch concentration), retentate makes multiple
passes through the setup as shown in Figure 9.2. If instead of recycling retentate to
the system feed vessel, it is discharged, as shown in Figure 9.11(a), a single-pass
TFF (SPTFF) system configuration is created. It is a method that has gained popular-
ity over the last few years, with the first patent being awarded in 2008 [276].

(a)
Feed tank Retentate
valve
Retentate

Permeate

Feed UF membrane
stream module
Feed pump
(b)

(c)

Figure 9.11: Single-pass TFF. (a) shows a UF system configured for single-pass operation. Image ©
NC State University; reprinted with permission (b) shows conventional UF membranes placed in
series to create an SPTFF membrane stack. Image © NC State University; reprinted with permission
(c) shows the diagram for an SPTFF module. Adapted from Pall Life Sciences’ “Application Note:
Cadence™ Systems Employ New Single-Pass TFF Technology to Simplify Processes and Lower
Costs,” 2011, 2. Copyright 2011 Pall Corporation [277].
 9.8 Single-pass TFF 287

A parameter often used to characterize the performance of SPTFF is conversion,

defined as the ratio of the permeate flow rate, Qp, to the feed flow rate, Qf:

Qp
Conversion = (9:15)
Qf

To achieve the desired concentration target in SPTFF, high conversion values are re-
quired. High conversion can be achieved by operating at low flow rates through the
module and/or by increasing the path length to increase residence time. There are a
number of ways to implement a single-pass configuration with increased path length.
Conventional UF cassettes can be placed in series by the end user, similar to the config-
uration shown in Figure 9.11(b). Membrane providers like Pall and MilliporeSigma have
created special manifolds and gaskets to enable placement of conventional UF cas-
settes in series within a single existing holder [278, 279]. In addition, SPTFF cassettes,
comprising an internally manifolded series of UF cassette stages, can be used. Pall’s
Cadence SPTFF modules with Delta membrane are an example [277]. The flow path for
these cassettes is shown in Figure 9.11(c). The number of flow channels through each
stage decreases so that flow velocity remains high as the feed stream is converted to
permeate. Note that unlike conventional batch concentration by UF, in which product
concentration in the retentate changes with time, product concentration in the reten-
tate exiting the SPTFF device of a given length reaches a steady-state value.
One of the advantages of recirculating feed as shown in Figure 9.2 is the possi-
bility of minimizing the membrane area for a concentration or DF step. So why use
SPTFF? There are several potential advantages versus the traditional batch concen-
tration configuration shown previously:
– A process stream can be fed continuously, while retentate and permeate are con-
tinuously withdrawn, making SPTFF well suited for continuous processing.
– Unlike batch concentration by UF, product in single-pass UF sees only a single
pass through a pump or, alternatively, the process stream can be moved by pres-
surizing the feed tank. Reduced pump passes relative to batch concentration
makes SPTFF a gentler method of processing and useful for fragile molecules.
– The smaller hold-up volume in an SPTFF system can translate to better product
recovery.
– The equipment footprint for single-pass UF is smaller than for the batch concen-
tration configuration.
– Product concentration in the retentate from the single-pass UF setup reaches a
steady state, unlike batch concentration in which product concentration contin-
uously changes.

SPTFF is well suited to a number of applications in biopharmaceutical processes. As

we’ve mentioned, it is a preferable option for continuous processing. Even in a
batch process, it is a good option for inline concentration (i.e., volume reduction);
for example, it can be used to easily reduce the load volume for a chromatography
288 Chapter 9 Formulation operations: ultrafiltration

step, which can shorten the time for processing. It is also a good option for inline
desalting, which can be accomplished by concentrating via SPTFF then diluting
with a no- or low-salt buffer.
More recently, inline cassettes for DF have become available. These address a gap
in single-pass UF technology by enabling inline buffer exchange, to complement the
concentration capabilities of the SPTFF systems just described [280]. Few details of the
module configuration are provided; however, presumably it is configured to allow DF
buffer to be fed to stages within the SPTFF module while simultaneously allowing per-
meate to be withdrawn from each stage, thereby enabling buffer exchange.

9.9 Summary

Tangential-flow UF and MF are commonly used membrane separation techniques for

biopharmaceutical processes. Separation in both is sized based. UF is often used for con-
centrating product and performing buffer exchange in a step referred to as diafiltration,
and it is included as a formulation step for most biopharmaceuticals just before bulk
filling. MF is used for solid-liquid separation, such as clarifying a cell culture broth.
Most membranes are made from modified polyethersulfone or regenerated cel-
lulose, and numerous vendors supply UF and MF modules for biopharmaceutical
applications. For biopharmaceutical applications, UF and MF membranes are pack-
aged as modules in either a cassette or hollow-fiber format. To accommodate larger
scale applications that require higher flow rates, membrane area in a cassette is in-
creased by widening the membrane or adding more membrane layers in parallel
within the cassette while keeping the path length the same. With hollow-fiber mod-
ules, area is increased by adding more fibers of the same length within a module.
To achieve even greater surface area for large-scale applications, multiple cassettes
or hollow-fiber modules are installed in a parallel configuration.
Most UF and MF membrane modules are reusable, and a procedure similar to that
shown in Figure 9.7 is used for GMP processing. Design of a UF or MF step is typically
done at bench-scale, using small membrane modules that are readily scaled up. There
are numerous process parameters that apply to UF and MF steps, such as TMP, cross-
flow rate, concentration factor, and diafiltration volumes. Ranges must be set appropri-
ately through development studies, with the goal of designing a step that meets either
the target concentration of product and the correct extent of buffer exchange (for UF)
or an acceptable product clarity (for MF), while resulting in acceptable product quality,
high step yield, and high permeate flux with reproducible process times. Once the UF
or MF step has been optimized at bench scale, it can scaled up using a linear scale-up
methodology in which the length of the flow path, the volumetric flow rate per unit
area of membrane, and the feed volume per unit membrane area are held constant be-
tween scales. Among a number of expectations for validation of UF and MF steps is the
 9.10 Review questions 289

execution of a study, done under protocol, to establish a membrane lifetime for a spe-
cific process.
Single-pass tangential-flow filtration is a method that has gained in popularity
in recent years, particularly for concentration of products by UF. Rather than recir-
culating retentate through the system as is done in batch concentration or constant-
volume diafiltration, the retentate is discharged from the system after a single pass.
This process configuration may be particularly useful for concentration and buffer
exchange steps in continuous processes.

9.10 Review questions

1. As mentioned earlier in the chapter, an in-line UV sensor (set to a wavelength

of 280 nm) is often included on the permeate line of a tangential-flow UF system
to detect product loss to the permeate. Why not also include a UV sensor in the
feed tank of a UF system to provide a direct measure that a concentration target
has been met?

2. You are working on a procedure to measure the NWP for a UF step to be used as
part of a process to produce a protein therapeutic for clinical trials. That process
uses 2.5 m2 of membrane area, and you know from bench-scale studies that the
NWP for the membrane is 23.7 LMH/psi. The measurement of NWP in the scaled
up process is manual. Estimate the amount of HPW that will be collected as per-
meate over a 1-minute period at a TMP of 5 psi. The HPW temperature is 18 °C.

3. Below is a graph showing NWP data for 15 UF runs for a validated process.

5
4.5
4
NWP L/m2 h psig

3.5
3
2.5
2
1.5
1
0.5
0
0 2 4 6 8 10 12 14 16 18
Run #

Figure 9.12: NWP data for 15 UF runs for a validated process. Image © NC State University;
reprinted with permission.
290 Chapter 9 Formulation operations: ultrafiltration

Give three possible reasons why the NWP value suddenly increased at run #16.
4. A solution containing a 27 kDa protein was concentrated using a 10 kDa MWCO
UF membrane with an area of 50 cm2 at a TMP of 22 psig. The starting concentra-
tion was 1 mg/mL. Data showing the retentate volume and time for the run are
shown below. Answer the following: (a) what is the protein concentration of the
final retentate? (Assume no protein is lost in the permeate); (b) what is the aver-
age permeate flux for the run?; (c) plot the permeate flux in L/m2/h (i.e., LMH) vs.
the protein concentration in the retentate and explain the trend you see.

Table 9.5: Retentate volume versus time data for

concentration of a protein by UF.

time (s) Vr (mL) time (s) Vr (mL)

  , 

  , 

  , 

  , 

  , 

  , 

  , 

,  , 

,  , 

,  , 

,  , 

,  – –

5. A protein therapeutic is to be concentrated to a target of 7.00 ± 0.20 mg/mL. To

achieve this concentration, the product is initially overconcentrated and then
buffer flush from the product recovery step is added back to the retentate. The
overconcentrated retentate is drained from the pilot-scale UF skid and collected
into a vessel with a tare weight of 1.34 kg. The gross weight of the vessel following
collection is 6.43 kg, and the product concentration is 9.42 mg/mL. The system is
flushed with 3 L buffer, of which 2.89 L is recovered at a product concentration of
0.49 mg/mL.
a) How much of the recovered flush can be added to achieve the final target
concentration?
b) What is the step yield achieved?
 9.10 Review questions 291

6. It was shown that for batch concentration, VCF = CF assuming no loss of product to
the permeate. Derive the more general expression Vr = Vf × (Cr/Cf)−1/R that accounts
for the possibility of product permeating the membrane. To do this, (i) perform a
total mass balance (in the form: accumulation of mass = total mass input - total
mass output) on Figure 9.2(a) operated in batch concentration mode and (ii) per-
form a product mass balance in the form: accumulation of product = product mass
input – product mass output. Recall that R = 1 – Cp/Cr and solve the three equations
for Vr.

7. Referring to the above problem, if a protein therapeutic permeates a UF mem-

brane with a rejection value of 0.8, what final retentate volume would be tar-
geted for a concentration factor of 10 (in batch concentration configuration) if
the initial feed volume is 100 L? How does this value compare to the final vol-
ume required for a product that is completely retained by the membrane?

8. Derive equation (9.12) for the percent of a solute retained in the feed, with no
solute added in the DF buffer, during a constant-volume batch DF step. To do
this, (i) perform an overall mass balance (in the form: accumulation of mass =
total mass input - total mass output) on Figure 9.10(a) operated in constant-vol-
ume DF mode and (ii) perform a solute mass balance in the form accumulation
of solute = solute mass input – solute mass output.

9. For a solute that is completely permeable to a UF membrane, the solute con-

centration is reduced by a factor of 10 in a constant-volume DF after every
__________ DVs.

10. 500 L of product intermediate in a buffer solution of 50 mM tris and 0.1 M NaCl
is being diafiltered (by constant-volume DF) to reduce NaCl concentration to
below 0.005 M. For NaCl, R = 0, and for product R = 1 for the membrane in use.
If the average permeate flux is 40 LMH and 25 m2 of membrane area is used,
answer the following questions.
a) How many DVs of DF buffer are required to reduce the NaCl concentration
to the desired value?
b) How much permeate is generated during the DF step?
c) How much time is required for the DF step?

11. A concentration/DF step using UF for formulation of drug substance has an ex-
pected step yield of 93%–100%. After 15 runs of the same product using the
same membrane modules, the step yield measures 75%. A deviation is initiated
and root cause must be determined. Give eight possible reasons for the out-of-
range step yield value.

12. For an SPTFF system in which product is completely retained by the mem-
brane, what conversion value is required to produce a product concentration of
15 mg/mL from a feed at 4 mg/mL?
Chapter 10
Summary and trends in biopharmaceutical
processing

Rather than providing a traditional conclusion to the book, we close by sharing an

imagined conversation between a new employee, fresh out of school, at a biopharma-
ceutical company and a thoughtful manager at the same company who has signifi-
cant experience in the field. The new employee has little hands-on experience in
making biopharmaceuticals and only rudimentary book knowledge. The experienced
manager has significant knowledge of the biopharmaceutical field but only 15 minutes
before having to depart for an important meeting. This conversation is meant to sum-
marize salient points made in the book, to acknowledge some topics that we were
unable to cover (due to space limitations) but that deserve more in-depth discussion,
and to highlight trends in biopharmaceutical processing.
New employee: I would benefit from a better understanding of biopharmaceut-
icals. I know you need to get to a meeting, but can you give me a quick synopsis of
biopharmaceutical products?
Experienced (and thoughtful) manager: You will soon discover that the bio-
pharmaceutical field is an excellent way to apply the science and engineering courses
you took in college to solve real-world problems. You are probably aware that bio-
pharmaceuticals are medicines manufactured by or from living organisms such as
bacteria like E. coli, mammalian cells like Chinese hamster ovary (CHO) cells, or even
human beings. The active ingredients in these medicines are so large and complex
that they are difficult to produce by chemical synthesis, which is used to manufacture
many traditional pharmaceuticals. Cells have the ability to produce what the chemist
cannot. Some familiar examples include insulin, which is used to treat diabetes and
is produced through genetic engineering of E. coli or yeast cells. Another is Humira®,
a monoclonal antibody produced in CHO cells and used to treat rheumatoid arthritis
among a number of other diseases. Humira® is also the biggest selling drug (by reve-
nue) in the world. Vaccines, such as those currently being developed and approved
for COVID-19, are biopharmaceuticals as well. Many newer types of biopharmaceuti-
cals are coming on the market too, such as gene therapy products.
The final form of a biopharmaceutical product is not a pill or caplet. It is usually
a liquid solution or freeze-dried product that is reconstituted to a liquid solution be-
fore being administered. Biopharmaceuticals are delivered in liquid state because
they are injected or infused. If they were administered orally, they would break
down in the stomach and not be effective.

https://doi.org/10.1515/9783110616880-010
 Chapter 10 Summary and trends in biopharmaceutical processing 293

New employee: How are biopharmaceutical products made? Even a brief expla-
nation would help me as I begin working on the projects assigned to me.
Experienced manager: Processes for producing biopharmaceutical drug prod-
ucts typically start with steps necessary to grow the cells to high concentrations and
large volumes. Large-scale cell growth requires a bioreactor. Once the cells have pro-
duced the active ingredient, the product is separated from the culture broth that re-
sults. This separation requires a series of steps – referred to as the harvest process –
that may include cell lysis, centrifugation, depth filtration, and tangential-flow micro-
filtration. Those steps are capable of separating product from cells, but they typically
cannot separate the product from the many soluble impurities that are generated dur-
ing processing. The concentration of these impurities should be minimized to ensure
the safety and efficacy of the medicine. A clarified (i.e., particle-free) product interme-
diate, in which the active ingredient is dissolved, is produced by the harvest process
and is loaded to a chromatography column. In fact, product is likely to go through
several independent chromatography (or other purification) steps to separate the
product from impurities such as host cell proteins, endotoxin, host cell DNA, and
product aggregates. These chromatography steps are essential to ensuring that criti-
cal quality attributes (CQAs) related to purity, potency, and safety are met.
Once the product has been purified, the resulting product intermediate – the
product dissolved in a buffer – is formulated. Formulation involves steps like ultra-
filtration, which are performed to put the product into the proper buffer system
with all necessary excipients added. The formulated product can then be filled into
one of a variety of devices – such as vials, prefilled syringes, or autoinjectors – to
create the drug product that goes to a patient.
I would be remiss if I led you to believe that this is all there is to manufactur-
ing a biopharmaceutical product. There are also regulations, known as current
Good Manufacturing Practice (cGMP), that must be followed to ensure that prod-
ucts are consistently produced with the required quality. The manufacturing pro-
cess, facility, and cGMP regulations all come together to ensure product quality.
People are central to these activities as well. It is people who design and execute
the manufacturing processes, design and build facilities, test the product, and im-
plement systems in place to ensure quality. I’ve given you quite a bit of informa-
tion and need to get going, but I know of a good book that tells most of the “story”
of biopharmaceutical manufacturing. It’s called Biopharmaceutical Manufactur-
ing: Principles, Processes, and Practices.
New employee: This information is useful and has increased my interest. I know
your meeting is getting ready to start, but I have one more question. You said the
book you mentioned covers most of the “story” of biopharmaceutical manufacturing.
What doesn’t it cover?
Experienced manager: The book provides the details of most steps required to
produce final drug product. But there are a few operations that you may want to
know more about. For example, viral clearance steps, which are briefly discussed in
294 Chapter 10 Summary and trends in biopharmaceutical processing

Chapter 2 and in Chapter 8, could be the topic of a separate chapter. If you are not
familiar with viral clearance, it is a term used to describe steps required to remove or
inactivate unwanted viruses that can make their way into a process, particularly
those processes that use mammalian-derived components such as CHO cells. The
book also gives only brief mention of final fill-finish operations in Chapter 2. Fill-fin-
ish refers to the steps needed to produce final drug product from the purified active
ingredient. It, too, could have been the topic of a separate chapter. The book also
only touches on the different testing – of raw materials, in-process samples (i.e., sam-
ples of product intermediate), drug substance, and drug product – required as part of
a biopharmaceutical process. That topic is large enough to deserve its own book.
Remember, though, that biopharmaceutical manufacturing is a big and evolving
field. There are many advances being made in a number of areas within biopharma-
ceutical manufacturing.
New employee: I know you have to go to your meeting, like yesterday, but do
you have any information on these trends? I’m interested in everything you’ve said
so far and would like to do additional reading.
Experienced manager: In fact, I keep a running tab of trends based on my dis-
cussions with colleagues, seminars I attend, and articles that I read. The list includes:
– Growth in use of single-use technologies. Single-use technologies are already
widely adopted in processes to manufacture biopharmaceuticals for pre-clinical
and clinical trial studies. It is expected that adoption for commercial processing
will increase [281]. Single-use technologies offer a number of advantages, in-
cluding the flexibility offered by rapid changeover between products. Single-use
technologies have been and will continue to be particularly important to en-
abling personalized cell therapy processes.
– More processes designed to operate continuously. By creating more continuous
processing options, potential advantages such as smaller equipment, a smaller
manufacturing facility, and less cost for equipment and facilities may be realized.
– Development and implementation of intensified processes, in addition to devel-
opment of continuous process steps. There is no single accepted definition of
process intensification, but we use it to refer to “technologies to increase the
amount of intermediate or final product manufactured per volume, footprint,
unit of time, or expense by increasing efficiency in one or more unit operations”
[282]. There are numerous examples of advances that fit this definition. One
method of intensifying is to integrate multiple processing steps into a single unit
operation, resulting in less capital cost and high process yields. The Emphaze™
AEX Hybrid Purifier from 3 M is an example. It combines the ability to clarify a pro-
cess stream while simultaneously clearing soluble impurities, objectives that would
typically be met by the use of two separate unit operations rather than just one.
Intensified processes will lead to greater productivity, less equipment in a process,
smaller facilities, and less cost for equipment and facilities.
 Chapter 10 Summary and trends in biopharmaceutical processing 295

– Advances in process analytical technology (PAT) will lead to greater implemen-

tation. Process analytical technology refers to a system for designing, analyzing,
and controlling manufacturing through timely measurements (i.e., during proc-
essing) of critical quality and performance attributes of raw and in-process mate-
rials and processes, with the goal of ensuring final product quality [283]. It is
most often associated with real-time inline measurement (sensor placed directly
in a process vessel or process stream) and online measurement (sensor placed in
a sample stream diverted from the manufacturing process and then returned) of
critical quality attributes or performance parameters, with adjustments made to
process parameters to keep the CQA or performance parameter within range.
Central to PAT is sensor technology, and while inline and online sensors are com-
mon in biopharmaceutical processes, there remain many measurements that re-
quire a sample to be pulled and tested in a lab, away from the process. This takes
time and makes it difficult to control the process using the measured value. Conse-
quently, development of inline/online sensors is widely considered essential to ad-
vancing biopharmaceutical processing by enabling real-time testing (samples won’t
have to be sent to the QC lab for testing) and even real-time release [284]. Jiang et
al. [285] provide a nice summary of analytical technologies that may significantly
advance PAT implementation.
– More sophisticated automation and process control. These changes will reduce
manual interventions in processes and one day lead to “lights-out manufactur-
ing” in which no human presence is required. Interest in improving automation
and control in biopharmaceutical processes is high. BioPhorum, a consortium in
which member companies can collaborate to accelerate their progress, recently
published the results of a technology roadmapping exercise for biopharmaceuti-
cal processing [284]. Process automation was identified as one of six areas in
which industry would be investing innovation efforts over the next decade.

And now back to the readers of the book: thank you for your enthusiasm and inter-
est. We expect you will have a fulfilling and engaging experience in biopharma-
ceutical manufacturing for many years to come.
Glossary
Adsorption Attachment of a molecule, particle, etc. to a solid surface.

Affinity chromatography (AC) A type of chromatography in which a target solute, such as a protein
therapeutic, has a very specific affinity for a stationary phase, such as a resin.

Air diffusion test A type of filter integrity test that challenges the feed side of a wetted membrane
with pressurized air and measures the rate that air permeates. If the flow rate of air that permeates
is less than or equal to the maximum that would be expected by diffusion alone, the membrane has
integrity.

Air lock A room that typically has two doors in series to separate a cleanroom from a less clean
environment, such as a corridor. The two doors are typically interlocked to avoid being opened at
the same time.

Allogeneic cell therapy A type of cell therapy in which cells are derived from a donor (or donors)
other than the patient who will receive the cells.

Antigen Any substance (e.g., molecule) that induces the body to make an immune response.

Aseptic Describes a process and/or facility designed to keep a product sterile.

Aseptic filling Filling of drug product under aseptic conditions.

Aseptic technique Practices for working in a cleanroom that minimize the risk of contaminating the
environment and the product.

Asymmetric membrane Membrane consisting of a thin, dense layer that provides the desired
separation on top of a thicker, more porous bottom layer, which provides mechanical strength
without significantly hindering permeate flow.

Asymmetry A parameter that provides a measure of packing quality; it is typically calculated from
the peak resulting from a pulse injection of a tracer to a packed column. Specifically, asymmetry is
a measure of the symmetry of the elution peak that results from the pulse tracer test.

Autologous cell therapy A type of cell therapy in which cells are taken from and administered to the
same individual.

Auxotroph A strain that has been genetically modified to be deficient in production of an essential
metabolite.

Axenic Free from living organisms other than the cells required for the process. When growth of
cells is carried out in a bioreactor, the fermentation or cell culture should be axenic, not sterile.

Band broadening (or zone broadening) The widening of a solute band moving through a
chromatographic stationary phase. It results from hydrodynamic dispersion, axial diffusion of the
solute, and resistance to transport of the solute from the mobile phase to the surface of the
stationary phase. Band broadening dilutes the solute and reduces resolution between components
being separated.

https://doi.org/10.1515/9783110616880-011
298 Glossary

Batch A defined quantity of product processed in one process or series of processes so that it could
be expected to be homogeneous. Definition adapted from EudraLex – Volume 4 – Good
Manufacturing Practice (GMP) guidelines [286].

Batch disposition The process of assigning a specific status or product usage to a batch of product
or product intermediate. Disposition status may include released/approved, conditionally
released/approved, rejected.

Batch production record An approved document that provides instructions on and generally
requires data be recorded from an operation associated with the production of drug substance and
drug product.

Batch release Approval of a batch of product or product intermediate for further processing or
commercial distribution.

Bind-and-elute chromatography A mode of chromatography in which product binds to the resin

during the load step, and impurities may or may not bind. The product is recovered during an
elution step by making a change in mobile-phase (elution buffer) composition that disrupts
interaction between the stationary and mobile phase.

Binding capacity (equilibrium and dynamic) The amount of solute that can bind to a stationary
phase per unit volume of stationary phase. When measured under static conditions in which the
mobile phase and stationary phase are allowed to come to equilibrium, the term equilibrium
binding capacity is used. Measured under flow conditions, binding capacity is referred to as
dynamic binding capacity.

Bioavailability The fraction of drug that enters systemic circulation.

Bioburden The amount and type of microorganisms that can be present in raw materials, process
intermediates or drug substance.

Biological product In this text, synonymous with biopharmaceutical.

Biopharmaceutical A pharmaceutical (that is, medicine) inherently biological in nature and

manufactured by or from living organisms (or cells from living organisms).

Biosafety The discipline addressing the safe handling and containment of infectious
microorganisms and hazardous biological materials [89].

Biosimilar A biological product highly similar to and interchangeable with an approved biological
product.

Bubble point test A type of filter integrity test that measures the minimum air pressure required to
force liquid, such as water, from the pores of a filter or membrane. Measured bubble point
pressures equal to or greater than the minimum bubble point value specified by the vendor for a
particular wetting agent/gas (typically air) combination demonstrates membrane integrity.

Buffer A solution that resists changes in pH when acid or base is added to it. Buffers often include
a weak acid or weak base together with one of its salts. They are used to avoid shifts in pH in a
solution. In biopharmaceutical processing, the term is often used to refer to solutions – even those
that do not buffer against pH changes – in general.

Buffer exchange See Diafiltration.

 Glossary 299

Buoyancy The force experienced by a particle submerged in a fluid that results from the pressure at
the bottom of the particle being greater than that at the top. In the case of gravity settling, the force
is upward; in the case of centrifugation, the force acts toward the axis of rotation.

Calibration Comparison of the results from a particular instrument or device with results produced
by a reference or traceable standard over an appropriate range of measurements. The comparison
must fall within specified limits [96].

Cassette A tangential-flow filtration module, usually with a rectangular geometry, containing

multiple layers of flat sheet UF or MF membranes and separate channels for feed, retentate and
permeate.

Cavitation Phenomenon in which a sudden drop in the pressure of a liquid leads to the formation of
small vapor-filled cavities in a liquid. When these bubble collapse, shock waves are formed.

Cell culture The growth of animal cells, including mammalian cells like Chinese hamster ovary
(CHO) and insect cells like Sf9 (cells from ovaries of a fall armyworm).

Cell lysis (also referred to as cell disruption) Breaking open of cells, usually to recover intracellular
components. There are numerous methods used to lyse cells, some of which are more scalable
than others.

Cell membrane (also referred to as the plasma membrane or cytoplasmic membrane) A

semipermeable membrane that surrounds the cytoplasm of every cell. It is composed of lipids and
proteins. Generally speaking, the cell membrane serves as a barrier keeping the cell components in
and unwanted substances out. It also regulates transport of essential nutrients into the cell and
removal from the cell of waste products.

Cell therapy The use of living, whole cells for treatment of a disease.

Cell wall A flexible layer that surrounds the plasma membrane of the cells of bacteria, fungi such as
yeast, and plants. The composition of the cell wall varies by cell type. It provides cells with
structural support, which is particularly needed as water enters the cell due to the osmotic
pressure difference between the cytoplasm and surrounding fluid.

Centrifugal force An apparent force that acts on an object moving in a circular path that causes
movement outward from the center of rotation. When a slurry consisting of particles dispersed in a
less dense liquid is exposed to a centrifugal force, the particles move away from the axis of
rotation. This movement is the basis of centrifugation for solid-liquid separation.

Centrifugation A process step that uses density difference between two immiscible phases and
centrifugal force to separate the two phases. In biopharmaceutical manufacturing, the two phases
are typically solids, such as host cells or cell debris, dispersed in a surrounding liquid.

Centrifuge A machine used for centrifugation. A centrifuge sets immiscible phases in rotation
around a fixed axis, which applies a centrifugal force that enables separation of the two phases by
causing the denser phase to move outward in the radial direction. Common centrifuge types used
in biopharmaceutical production include the disc-stack centrifuge and tubular-bowl centrifuge.

Change control A formal system by which qualified representatives of appropriate disciplines

review proposed or actual changes that might affect the validated status of facilities, systems,
300 Glossary

equipment or processes. The intent is to determine the need for action to ensure and document
that the system is maintained in a validated state [88].

Changeover The replacement of wetted soft parts on equipment – such as gaskets, O-rings, and
valve diaphragms – to avoid the possibility of contamination of one product that may have been
absorbed into the soft parts into the next product.

Chromatogram A plot of a sensor response (e.g., UV absorbance, conductivity, pH, etc.) from a
chromatography system versus volume or time.

Chromatography column The tube (and other associated hardware) that holds chromatography
resin in place.

Chromatography support The nonfunctionalized base for the stationary phase medium.

Clarification The process of making a liquid clear by removing suspended solids.

Cleaning The removal of residues (e.g., proteins, buffer components, etc.), particulates, and
cleaning/sanitizing agents from equipment surfaces.

Clean-in-place (CIP) A cleaning mode in which removal of soil from product contact surfaces is
carried out with equipment in its process position by flowing and/or spraying cleaning agents and
water rinses over the surfaces to be cleaned.

Clean-out-of-place (COP) A cleaning mode in which equipment is disassembled and manually

cleaned in a location away from the processing area.

Cleanroom (also referred to as a classified area) A room in which the concentration of airborne
particles is controlled and classified (e.g., ISO 8) and which is constructed and used in a manner to
minimize the introduction, generation, and retention of particles inside the room [85]. Cleanrooms
are monitored to demonstrate that the classification is met [287].

Clonality The condition of being genetically identical.

Closed system A system (process or process steps) designed and operated so that product is not
exposed to the surrounding environment [84].

Column volumes (CVs) Term used to describe the amount of a mobile phase fed to a
chromatography column relative to the volume of the column. For example, 600 L of equilibration
buffer fed to a column packed to a volume of 200 L corresponds to 3 CVs of equilibration buffer.

Complex component A chemical component (e.g., cell culture medium) whose constituents are not
completely defined. All constituents may not be known and their relative abundance difficult or
impossible to discern.

Component From 21CFR210, any ingredient intended for use in the manufacture of a drug product,
including those that may not appear in such drug product [53]. Additionally, a component can refer
to constituent part of something (for example, a component of a piece of equipment).

Concentration Generally refers to the amount of a component per unit volume or mass. As a
biopharmaceutical processing step, the term refers to increasing the mass of product per unit
volume of solution by removing solvent, which is typically water in biopharmaceutical processes.
 Glossary 301

Concentration factor (CF) Ratio of the final concentration of the product to its starting concentration
during a UF concentration step.

Concentration polarization Elevated solute (usually product) concentration at the surface of a UF

membrane that results from permeating liquid carrying dissolved solutes, such as product, to the
membrane surface where they are retained.

Contaminants An external component introduced into the process that is not part of the process
[199].

Controlled not classified (CNC) Space in which the particulate levels in the room air are controlled,
but the rooms is not necessarily monitored and therefore cannot be classified (as ISO 8, for
example).

Corrective action An action taken to correct and/or eliminate the cause of a detected non-
conformity or other undesired situation to prevent recurrence.

Critical quality attribute (CQA) “A physical, chemical, biological or microbiological property or

characteristic that should be within an appropriate limit, range, or distribution to ensure the
desired product quality. CQAs are generally associated with the drug substance, excipients,
intermediates (in-process materials), and drug product” [288].

Crossflow rate Flow rate on the feed side of a membrane.

Current Good Manufacturing Practice (cGMP) a system for ensuring products are consistently
produced and controlled according to quality standards. The “c” stands for “current,” reminding
manufacturers that they must employ technologies and systems that are up to date in order to
comply with the GMP regulations [289].

Dead leg A portion of a piping or distribution systems through which fluid cannot flow when
connections are removed and capped or valves are closed.

Death phase The terminal phase of growth where cellular stresses due to starvation or toxic
compounds causes cell death.

Defined component A chemical component (e.g., cell culture medium) with clearly defined chemical
formulae. All chemical constituents and their abundance are known.

Denature To change the three-dimensional structure (i.e. secondary, tertiary, or quaternary) of

proteins or other biomolecules by application of an external stress such as heat, strong acid or
base, or an organic solvent.

Depth filtration A type of filtration that uses a porous filtration medium to capture particles
throughout the depth of the filter rather than just at the surface.

Design of Experiments (DoE) A statistical approach for planning and executing defined experiments
and analyzing the generated results to provide insight into factors that significantly affect the
outcome of a process.

Design qualification (DQ) The documented verification that the proposed design of the facilities,
systems and equipment is suitable for the intended purpose [88].
302 Glossary

Design space The relationship between the process inputs (material attributes and process
parameters) and the critical quality attributes [288].

Deviation An occurrence, problem or other undesirable event that represents a departure from an
approved process or procedure or from an established standard or requirement.

Diafiltration (DF) A processing step that uses UF membranes to remove solutes or exchange solutes
for others by adding a new solution and simultaneously permeating out the old. Used
synonymously with the term buffer exchange.

Diavolume (DV) The ratio of the volume of diafiltration buffer fed in diafiltration step to the volume
of retentate being diafiltered. For example, 500 L of a diafiltration buffer added to 100 L of product
solution (retentate) is five DVs.

Diffusion The movement of solute molecules from areas of high concentration to low concentration
driven by the random thermal motion of solute molecules.

Dissolved oxygen (DO) A measure of the amount of oxygen molecules dissolved in a liquid.

DNA replication The cellular process by which genetic material is copied and sorted in preparation
for cellular division.

Doubling time The time it takes for cell concentration to double in value.

Downstream (DS) process Later steps in a process for production of biopharmaceuticals. The
downstream portion of a process comprises the purification stage and formulation and fill stage. It
may also include steps in the harvest stage.

Drug In this text, both small-molecule pharmaceuticals produced by chemical synthesis as well as
biopharmaceuticals.

Drug product The finished dosage form – tablet, capsule, or in the case of biopharmaceuticals, a
liquid solution or lyophilized product – that contains the active ingredient. It contains the drug
substance generally, but not necessarily, in association with one or more other ingredients [43].

Drug substance The active ingredient that is intended to furnish pharmacological activity or other
direct effect in the diagnosis, cure, mitigation, treatment, or prevention of disease. It is intended to
be incorporated into the finished dosage form [43].

Dynamic binding capacity (DBC) See Binding capacity.

Effluent Flow exiting a unit operation, such as the flow from a packed bed during any phase of
chromatography.

Eluate Product intermediate from a chromatography step.

Elution Chromatography phase in which product bound to the resin is moved through the
stationary phase for recovery of the purified product. For chromatography that involves binding of a
mobile phase solute to the stationary phase, elution steps weaken the interaction between a solute
and the stationary phase by changing mobile phase composition, typically in a stepwise fashion or
linearly. Isocratic elution involves no change in mobile phase composition and is common in size
exclusion chromatography.
 Glossary 303

Endotoxin Literally, a toxin from within. It is most often used in reference to lipopolysaccharides
found in the outer membrane of Gram-negative bacteria like E. coli. It is a common impurity or
contaminant in bioprocess streams, and it can produce a variety of reactions in humans, including
fever and a lowering of blood pressure.

Equilibrium binding capacity (EBC) See Binding capacity.

Eukaryotic cell Cells from animals, plants, fungi, or protists. They contain a membrane-bound
nucleus that holds the cellular DNA and numerous membrane-bound organelles.

Excipient Anything in the drug product other than the active ingredient.

Exponential phase The period following the lag phase where cells are rapidly dividing. It is marked
by an exponential increase of cell mass with respect to time.

Extracellular Outside the cell.

Factory acceptance test (FAT) Testing performed at an equipment/system manufacturer’s facility to

demonstrate that a piece of equipment or system performs as expected prior to delivery to the end
user.

Feed flux Volumetric feed rate to a filter per unit area. It is often expressed in units of LMH (L of
feed per m2 of filter area per hour).

Feed interval The time between discharges in a disc-stack centrifuge.

Fill-finish Steps involved in putting drug product into its container (e.g., vial or syringe). Some
sources consider the labelling, packaging, and QA release of product part of fill-finish operations.

Filter A porous medium used for separation of components (e.g., solids from liquid, solute from
solvent) in a fluid stream.

Filter capacity (also referred to as loading or throughput) Maximum volume of suspension that can
be filtered – because a maximum pressure difference has been reached, for example – per unit
area of filter. It is typically expressed in units of L/m2.

Filtrate (also referred to as permeate) Portion of the feed that permeates a filter or membrane.

Filtration Any technique used to separate particulate in a fluid suspension or components in

solution according to their size by flowing under a pressure differential through a porous medium.

Flow-through chromatography A mode of chromatography in which impurities rather than product

bind to the stationary phase. Flow through also refers to material passing through column during
the load of bind-and-elute chromatography without being bound.

Formulation Refers to the buffer system and other excipients in which active ingredient is dissolved
in drug product. The steps required to place the active ingredient into the proper formulation are
part of the formulation and fill stage of a process.

Fouling Plugging of the filter due to accumulation of solids on or within the filter medium that leads
to poor filter performance (e.g., reduced filtrate or permeate flux).
304 Glossary

Gene therapy A technique that uses genes to treat or prevent disease by inserting a gene into a
patient’s cells instead of using drugs or surgery. The inserted gene replaces a disease-causing
gene or produces a product, such as a protein, that treats the disease.

Glycosylation attachment of sugar molecules to other molecules, such as proteins.

Gram staining A multi-step method of staining bacteria that uses crystal violet as the primary dye.
Results from the procedure are used to classify bacteria as gram positive or gram negative.
Bacteria with a thick peptidoglycan cell membrane are stained purple by crystal violet and referred
to as gram positive. Bacteria with thinner peptidoglycan layers do not retain the crystal violet dye,
and are counter-stained pink by safranin. These bacteria are referred to as gram negative.

Harvest The processing stage that separates product from the production system. Also referred to
as recovery.

Harvest cell culture fluid (HCCF) Broth resulting from the culturing of cells in a bioreactor.

Height equivalent to a theoretical plate (HETP) The height of one theoretical plate within a packed
column (or other stationary phase). HETP is based on the idea that a packed column (or membrane
adsorber or monolith) contains theoretical plates that allow for equilibration between the mobile
and stationary phases, and the larger the number of plates, or equivalently the smaller the HETP,
the better the separation.

High purity water (HPW) In this text, we use the term high purity water to refer to a type of purified
water that meets the requirements set for Purified Water in Chapter 1231 of the USP-NF. In
biopharmaceutical manufacturing, it is often used for equipment cleaning.

Hollow-fiber module A membrane module that contains a bundle of hollow fibers made from UF or
MF membranes that are housed in a cylindrical shell and used for tangential-flow applications. The
interior of the fiber is referred to as the lumen and has a typical diameter of 0.5–1.0 µm.

Homogenizer Machine that pumps a cell slurry at high pressure through a small orifice created by a
specially designed valve resulting in lysis.

Homogenizer passes The number of times a suspension of solids (or insoluble liquid) in liquid
(e.g., cells or cell debris in medium or buffer) is passed through a high-pressure homogenizer.

Host cell proteins (HCPs) Native proteins from the host cell that may become part of the product
stream. They are a heterogeneous group made up of hundreds and possibly even thousands of
different proteins, with widely different properties. They are problematic for a patient because they
may cause an unwanted immune response and must be removed from product.

Hydrophobic interaction chromatography (HIC) A type of chromatographic interaction based on

hydrophobicity, that is, a hydrophobic solute (such as the product) binds to a hydrophobic
stationary phase. The greater the hydrophobicity, the stronger the interaction between the solute
and stationary phase. Addition of certain salts is required to adsorb hydrophobic solutes to the
resin.

Impurities “Any component present in the drug substance or drug product which is not the desired
product, a product-related substance, or excipient including buffer components” [199].

Inclusion body A form of intracellular product (protein) in which the product takes the form of
micron-sized solid aggregates that are relatively pure. Because the protein that makes up the
 Glossary 305

inclusion body is misfolded, the inclusion body must be solubilized and then refolded to create a
bioactive, soluble biopharmaceutical product.

Installation qualification (IQ) The documented verification that the facilities, systems, and
equipment, as installed or modified, comply with the approved design and the manufacturer’s
recommendations [88].

Intermediate Material produced during processing of drug substance that undergoes further
processing before it becomes drug substance.

Intracellular Within the cell.

Ion exchange chromatography (IEC) A chromatographic interaction mode in which adsorption of a

component from the mobile phase is based on attraction of opposite charges. For example, a
negatively charged solute binds to a positively charged resin bead and vice versa. Separation
between components in the mobile phase is based on differences in charge between components.
Positively charged resins are used in anion exchange chromatography; negatively charged resins
are used in cation exchange chromatography.

Ionic strength A measure of the concentration of ions in a solution.

Isoelectric point (pI) The pH at which the net charge on a molecule (e.g., protein) is zero. The pH of
the solution in which a molecule is dissolved determines its charge. If the solution in which a
molecule is dissolved is at a pH < pI, the molecule has a net positive charge. If the solution is at a
pH value > pI, the net charge on the compound is negative.

Kosmotropic salt Salts such as (NH4)2SO4 that lead to protein aggregation and, by extension,
binding of hydrophobic solutes such as proteins to hydrophobic ligands in HIC.

Lag phase The period of time following inoculation where cells are becoming acclimatized to their
new environment. It is marked by a minimal amount of cell division.

Laminar flow An airflow moving in a single direction and in parallel layers at constant velocity from
the beginning to the end of a straight line vector [290].

Leachables Compounds that migrate from any product-contact material under normal process
conditions.

Ligand The molecule coupled to a chromatography support that creates interaction between a
mobile phase solute and the stationary phase.

Lysate The intermediate that results from a cell lysis step. In the case of lysis by homogenization,
the intermediate may be referred to as homogenate. Lysate that has had solids removed is often
referred to as clarified lysate.

Master cell bank (MCB) Cells that have been grown and dispensed into multiple vials and stored
under defined conditions, often cryogenically. The master cell bank is used to produce the working
cell bank. It prevents genetic variation by minimizing the number of times a cell line is passaged or
handled during the manufacturing process.

Membrane A thin porous medium used for separation of components (e.g., solids from liquid,
solute from solvent) in a fluid stream. A membrane is a type of filter, but not all filters would be
classified as membranes.
306 Glossary

Microbial fermentation The growth of microorganisms, like the bacteria E. coli or the yeast
Saccharomyces cerevisiae.

Microfiltration (MF) A type of filtration that relies on a microfiltration membrane to capture

particles at the membrane surface. Microfiltration membranes have pore sizes as small as 0.1 μm
or as large as 10 μm.

Mobile phase The liquid phase that flows through a packed chromatography bed or other
chromatography device (e.g., membrane adsorber or monolith).

Molecular weight cutoff (MWCO) Also referred to as the nominal molecular weight limit (NMWL),
represents the molecular weight of a solute, in daltons, retained 90% by the membrane. MWCO
values for UF membranes range from 1 kDa to 1,000 kDa.

Multimodal chromatography (MMC) A chromatography mode in which separation of a solute from

other components is based on more than one type of interaction. For example, a mixed mode resin
might include a ligand that is both charged and hydrophobic.

Normal-flow filtration (NFF) A mode of filtration in which feed flows perpendicular to the filter
medium, in the same direction as filtrate flow.

Normalized water permeability (NWP) A measure of membrane cleanliness in which the permeate
rate of water, at a given TMP, temperature, and crossflow rate, is measured and calculated per
equation (9.3). Typical units of NWP are L/(m2 h psig).

Open system A process that is not closed and therefore must be performed in an environment
where the probability of contamination is acceptably low.

Operational qualification (OQ) The documented verification that the facilities, systems, and
equipment, as installed or modified, perform as intended throughout the anticipated operating
ranges [88].

Osmolality/osmolarity Measures of solute concentration. Osmolarity is the number of osmoles of

solute per L of solution, while osmolality is the number of osmoles per kg of solvent (water).
Osmoles are the number of moles of solute that contribute to the osmotic pressure of a solution.
For example, one mole of NaCl would equate to two osmoles of solute particles, because NaCl
dissociates to Na+ and Cl− ions in water.

Osmotic pressure Pressure that would have to be applied to prevent a solvent, such as water, from
being transported across a semipermeable membrane from low solute concentration (such as UF
permeate) to high solute concentration (such as the feed side of a UF membrane).

Parenteral Literally, other than through gastrointestinal tract. Common parenteral routes of
administration for biopharmaceuticals are intravenous, subcutaneous, and intramuscular injection.

Passaging The act of transferring growing cells to a freshly prepared medium, typically to increase
the volume of growing culture.

Peptide Essentially a short-chain protein with 40 or fewer amino acids [9].

Percent yield The ratio of that amount of product at the end of one step or series of step to the
amount of product at the beginning of that step (or series of steps).The term percentage step yield
 Glossary 307

is used when applied to one step in the process and percentage process yield when applied to the
entire process.

Performance parameter An output parameter that cannot be directly controlled, but is indicative of
process performance [80].

Performance qualification (PQ) The documented verification that systems and equipment can
perform effectively and reproducibly based on the approved process method and product
specification [88].

Perfusion culture A production mode in which fresh culture medium is fed at the same rate that
spent medium is removed. Perfusion differs from a simple continuous operation in that cells are
retained within the bioreactor vessel as spent medium is removed.

Permeate (also referred to as filtrate) Portion of the feed that permeates a filter or membrane. In
this book, we use the term permeate in the context of membrane separations.

Permeate flux Permeate flow rate per unit area of membrane. Typically expressed in units of LMH (L
of permeate per m2 of membrane area per hour).

Platform The use of a common or standard method, equipment, and procedure applied to the
development or manufacture of different, but typically related, products.

Precipitation A process step in which a dissolved component – either the product or impurities – is
separated from other solutes by adding a reagent that alters the solubility of the component and
causes it to fall out of solution. The precipitated component is a solid phase.

Preventive action Actions taken to eliminate the cause of a potential deviation.

Procedure A document specifying instructions for carrying out an activity or process. Procedures
typically do not require recording of information.

Process parameter (also referred to as operational parameter) An input parameter or process state
parameter that is directly controlled. Process parameters can be further categorized as critical, key,
and non-key. A critical process parameter is one whose variability has impact on a critical quality
attribute. A key process parameter is one whose variability is important to process performance,
but does not affect a CQA. Non-key process parameters are easily controlled and therefore likely to
have no impact on a CQA or performance parameter [80].

Process-related impurity An impurity originating from the manufacturing process [199].

Process residuals Chemicals added to the biopharmaceutical manufacturing process that are not
fully consumed and therefore remain in the process stream in small amounts. Process residuals
must be removed to safe and acceptable levels within the final drug substance or drug product.

Production In Chapter 2, the term used to refer to the product synthesis stage of a process.

Product-related impurity Molecular variants of the product that may form during manufacture and
storage. They are often physically and chemically similar to the product, but not as safe or effective
[199].

Prokaryote A structurally simple single-celled organisms. Prokaryotes do not have a nucleus or

other organelles within the cytoplasm.
308 Glossary

Protein A macromolecule composed of amino acid monomers.

Protein therapeutic Protein used for therapeutic applications, i.e., to treat an existing disease.

Purification Generally refers to separating product from impurities present in a process stream. In
Chapter 2, the term refers to the process stage that purifies the product.

Quality assurance (QA) The unit within a biopharmaceutical facility with responsibility over quality
systems and ensuring GMP requirements are fulfilled. QA typically has responsibility for
disposition of raw materials, product intermediates, drug substance, and drug product.

Quality control (QC) The group within a biopharmaceutical facility whose primary responsibility is
conducting routine analytical testing on raw materials, product intermediates, drug substance, and
drug product to ensure acceptance criteria and specifications are met.

Quality target product profile (QTPP) A prospective summary of the quality characteristics of a drug
product that ideally will be achieved to ensure the desired quality, taking into account safety and
efficacy of the drug product [288].

Raw materials A general term used to denote starting materials, reagents, and solvents intended
for use in the production of intermediates or active ingredients [96].

Records and reports Documents stating results achieved or providing evidence of activities
performed.

Rejection factor (also referred to as retention factor) A measure of solute retention by a membrane,
defined specifically as 1 – Csolute,p/Csolute,r, where Csolute,p and Csolute,r are the measured solute
concentrations in the permeate and the retentate. Solutes that are completely retained by a
membrane have a rejection factor of 1.

Relative centrifugal force (RCF) A measure of the outward force generated by rotation of a
centrifuge relative to the force of gravity. It can be calculated using the following relationship: RCF
≡ ω2 × R/g, where ω = angular velocity (in units of rad/s, for example), R = distance from center of
rotation, the product ω2 × R is the centrifugal acceleration, and g = acceleration due to gravity
(=9.8 m/s2).

Resin Beads that serve as the stationary phase in a packed chromatography column.

Resolution A measure of the degree of separation between two components in a chromatography

step, specifically defined as distance between two neighboring peaks divided by the mean of the
peak widths.

Retentate The effluent from the feed side of UF or MF membrane module, after liquid has
permeated the membrane, in a tangential-flow filtration operation. UF retentate will be more
concentrated in retained components than the feed; likewise MF retentate will be more
concentrated in solids (for a clarification application) than the feed.

Reverse osmosis (RO) A filtration technique that uses reverse osmosis membranes, which are able
to reject ions, such Na+ and Cl-, from an aqueous solution.

Reversed phase chromatography (RPC) A chromatographic interaction mode in which binding to a

stationary phase is based on hydrophobic interaction, similar to HIC, but the binding is stronger
and typically requires an organic solvent for elution of bound solutes.
 Glossary 309

Sanitization A process used to reduce microbial contamination to acceptable levels. In

biomanufacturing, sanitization is often performed using chemical solutions such as isopropyl
alcohol, sodium hydroxide solutions, or disinfectants.

Selectivity A measure of how much time one solute (e.g., product) spends on the stationary phase
relative to another (e.g., impurities). The greater the selectivity, the better the resolution between
solutes.

Settling velocity The velocity of a particle as it moves through a fluid under a force, such as gravity
or centrifugal force. It increases as the size of the particle increases and the density difference
between the particle and surrounding fluid increases. It decreases with increasing fluid viscosity.

Shear stress A force, or more correctly a force per unit area, that acts along the surface of an
object, such as a cell, that may cause damage.

Sieving The removal of particles from a fluid stream by size (i.e., a particle is too large to flow
through the pores of the filter media).

Sigma (Σ) factor Centrifuge scale-up parameter that corresponds to the area in a gravity settling
device that would be required to give the same performance as the centrifuge at equal volumetric
flow rates.

Site acceptance test (SAT) Testing performed upon delivery to the end user’s facility to verify the
system or equipment performs as expected.

Size exclusion chromatography (SEC) A chromatography method in which separation is based on

size differences between solute molecules. Larger molecules in the mobile phase cannot diffuse
into pores within the resin particles and elute after a volume of buffer equal to the interparticle
column void volume (i.e., the volume between resin particles) has passed through the column.
Smaller molecules access the resin pores and are held up longer in the column, resulting in
separation from the larger molecules.

Skid A collection of components mounted on a frame to support a given activity. Skids are often on
wheels so that they are easily moved.

Slurry A mixture of particles suspended in liquid.

Standard Operating Procedure (SOP) An approved written document that provides instructions for
performing a task or operation not necessarily specific to a given product.

Stationary phase (related to chromatography) the fixed phase through which the mobile phase
flows in a chromatography device and with which solutes interact to create separation. Examples of
stationary phases used in biopharmaceutical processing include resin beads packed into a column,
the membrane used in a membrane adsorber, or a monolith.

Stationary phase (related to fermentation and cell culture) The phase of cell growth where division
is arrested due to the exhaustion of nutrients or accumulation of inhibitory substances.

Steam-in-place (SIP) The use of steam to sterilize or sanitize equipment in place.

Sterile Free from living organisms.

310 Glossary

Sterilization A process used to eliminate all microbial contamination, rendering the item sterile. In
biomanufacturing, sterilization is often performed using heat (i.e., autoclave), exposure to gamma
irradiation or through chemical means.

Superficial velocity The velocity of a mobile phase fluid element in a column or other
chromatography device if the stationary phase were not present. In a packed column, it is
calculated as the ratio of the volumetric flow rate to cross sectional area of the bed.

Tangential-flow filtration (TFF) (also referred to as crossflow filtration) A mode of filtration in which
feed is pumped tangentially along the surface of a filter or membrane with only a portion of the
feed permeating the filter.

Terminal sterilization Filling a low-bioburden product into the final container followed by
sterilization after filling.

Titer The term generally refers to the concentration of a component in solution. In

biopharmaceutical manufacturing, it is most often used to refer to the amount of product produced
per unit volume of bioreactor media. Units are often g/L.

Total organic carbon (TOC) An analytical method used to determine the amount of organic carbon
in a sample. This method is used for detection of product and process residues to verify equipment
cleaning procedures are effective.

Transcription Part of the process by which DNA is decoded to generate a functional product, such
as a protein. Transcription refers specifically to the process by which a segment of DNA is copied
into a new strand of messenger RNA.

Translation Along with transcription, part of the process by which DNA is decoded to generate
proteins. Translation specifically refers to the process by which messenger RNA, produced by
transcription, is decoded for protein synthesis.

Transmembrane pressure (TMP) Pressure difference across a membrane that drives permeate flow
in a TFF step. It is defined as the average pressure on the feed side of the membrane minus the
pressure on the permeate side: TMP = (Pf +Pr)/2 – Pfiltrate.

Turbidity Cloudiness of solution due to the presence of particles. It is typically measured using a
nephelometer, which measures the scattered light in an illuminated sample by placing a detector
at a specific angle (often at a right angle) to the incident light. The more scattered light, the higher
the turbidity. It is expressed in nephelometric turbidity units (NTU).

Ultrafiltration (UF) A type of filtration that relies on an ultrafiltration membrane, typically used to
separate large molecular weight components in solution from smaller components. UF membranes
have molecular weight cutoff values from 1 kDa to 1,000 kDa.

Unit operation Any individual step in a process that serves a specific function, often associated
with a single piece of equipment.

Upstream (US) process Initial steps in a process involved in the production of a biopharmaceutical
product. The upstream portion of a process includes steps in the production stage and may include
the harvest stage.
 Glossary 311

User requirements specification (URS) The set of owner, user and engineering requirements
necessary and sufficient to create a feasible design meeting the intended purpose of the system
[88].

Vaccine A biopharmaceutical used to prevent a disease by improving immunity to that disease.

Validation Establishing documented evidence that provides a high degree of assurance that a
specific process will consistently produce a product meeting its predetermined specifications and
quality attributes [290].

Viral clearance The inactivation or removal of unwanted virus in a biopharmaceutical process.

Water for injection (WFI) A type of purified water that meets the requirements for Water for Injection
in Chapter 1231 of the USP-NF or in other pharmacopoeia. WFI is used as an excipient in the
production of parenteral drugs, like biopharmaceuticals. HPW and WFI have similar testing
requirements, except WFI has a bacterial endotoxin specification that must be met that is not in
place for HPW.

Working cell bank (WCB) A (usually) cryogenically preserved vial of cells produced from a master
cell bank that is used to initiate a seed train that produces inoculum for a production bioreactor.

Yield of cells on substrate The ratio of cell mass produced to growth substrate consumed during
growth.

Yield of product on cells The ratio of product produced to cell mass accumulated during growth and
production.

Yield of product on substrate The ratio of product produced to growth substrate consumed during
production.
References
[1] Rader RA. (Re)defining biopharmaceutical. Nature Biotechnology 2008; 26: 743–51.
[2] Prevent Complications | Living with Diabetes | Diabetes | CDC 2019. (Accessed Oct 19, 2020,
at https://www.cdc.gov/diabetes/managing/problems.html).
[3] Schmid RD. Pocket Guide to Biotechnology and Genetic Engineering. Weinheim Germany:
Wiley-VCH, 2003, 1942.
[4] Lilly E and Company. Humulin [package insert]. 2019.
[5] Seqirus Inc. Flucelvax Quadrivalent [package insert]. 2020.
[6] Holme T. Denaturation. (Accessed Oct 19, 2020, at http://www.chemistryexplained.com/
Co-Di/Denaturation.html).
[7] Maa Y, Hsu CC. Protein denaturation by combined effect of shear and air‐liquid interface.
Biotechnology and Bioengineering 1997; 54: 503–12.
[8] Salvi G, De Los Rios P, Vendruscolo M. Effective interactions between chaotropic agents and
proteins. Proteins, Structure, Function, and Bioinformatics 2005; 61: 492–99.
[9] Impact Story: Developing the Tools to Evaluate Complex Drug Products: Peptides 2020. U.S.
Food and Drug Administration, Center for Drug Evaluation and Research. (Accessed Oct 19,
2020, at https://www.fda.gov/drugs/regulatory-science-action/impact-story-developing-
tools-evaluate-complex-drug-products-peptides).
[10] What Are: “Biologics” Questions and Answers 2019. Center for Biologics Evaluation and
Research. (Accessed Oct 19, 2020, at https://www.fda.gov/about-fda/center-biologics-
evaluation-and-research-cber/what-are-biologics-questions-and-answers).
[11] Philippidis A. Top 15 Best-Selling Drugs of 2019. GEN – Genetic Engineering and
Biotechnology News 2020.
[12] AbbVie Inc. Humira [package insert]. 2020.
[13] Aspirin. National Library of Medicine. (Accessed Oct 19, 2020, at https://pubchem.ncbi.nlm.
nih.gov/compound/2244).
[14] Jayapal KP, Wlaschin KF, Hu W, Yap MGS. Recombinant protein therapeutics from CHO cells –
20 years and counting. Chemical Engineering Progress 2007; 103: 40.
[15] Thomas F. Considering alternative dosage forms in biologics. BioPharm International 2019;
32: 16, 18, 50.
[16] Mantaj J, Vllasaliu D. Recent advances in the oral delivery of biologics. Pharmaceutical
Journal 2020; 304.
[17] Siew A. Ensuring sterility of parenteral products. Pharmaceutical Technology Europe 2013;
25: 37.
[18] rAAV Average Molecular Weight – SignaGen Blog. 2018.
[19] Tasnim H, Fricke GM, Byrum JR, Sotiris JO, Cannon JL, Moses ME. Quantitative measurement
of Naïve T Cell association with dendritic cells, FRCs, and blood vessels in lymph nodes.
Frontiers in Immunology 2018; 9: 1571.
[20] Kim S, Chen J, Cheng T, et al. PubChem 2019 update: improved access to chemical data.
Nucleic Acids Research 2018; 47: D1102–9.
[21] Novo Nordisk Inc. Novoeight [package insert]. 2020.
[22] Spark Therapeutics Inc. Luxturna [package insert]. 2017.
[23] Novartis Pharmaceuticals Corp. Kymriah [package insert]. 2018.
[24] Ponomarenko EA, Poverennaya EV, Ilgisonis EV, et al. The size of the human proteome: the
width and depth. International Journal of Analytical Chemistry 2016; 2016: 7436849–6.
[25] Dimitrov DS. Therapeutic Proteins. Totowa, NJ: Humana Press, 2012.
[26] Amgen Inc. Neulasta [package insert]. 2016.
[27] Genentech Inc. Avastin [package insert]. 2020.

https://doi.org/10.1515/9783110616880-012
314 References

[28] Allergan Pharmaceuticals. Botox [package insert]. 2010.

[29] MedImmune LLC. FluMist Quadrivalent [package insert]. 2020.
[30] Sanofi Pasteur SA. IPOL [package insert].
[31] PaxVax Bermuda Ltd. Vaxchora [package insert]. 2016.
[32] Daptacel [package insert]. 201x.
[33] Merck & Co. Inc. Gardasil [package insert]. 2015.
[34] Merck & Co. Inc. Recombivax HB. 2018.
[35] Pfizer (Wyeth Pharmaceuticals, Inc.). Prevnar 13 [package insert]. 2017.
[36] Coronavirus Vaccine Tracker 2020. Corum J, Wee S, Zimmer C (Accessed Nov 1, 2020, at
https://www.nytimes.com/interactive/2020/science/coronavirus-vaccine-tracker.html?
smid=nytcore-ios-share).
[37] Approved Cellular and Gene Therapy Products 2020. Center for Biologics Evaluation and
Research. (Accessed Nov 1, 2020, at https://www.fda.gov/vaccines-blood-biologics/cellular-
gene-therapy-products/approved-cellular-and-gene-therapy-products).
[38] Kite Pharma Inc. Tecartus [package insert]. 2020.
[39] AveXis Inc. Zolgensma [package insert]. 2019.
[40] Kite Pharma Inc. Yescarta [package insert]. 2020.
[41] Walsh G. Biopharmaceutical benchmarks 2018. Nature Biotechnology 2018; 36: 1136–45.
[42] TEConomy Partners LLC. THE ECONOMIC IMPACT OF THE U.S. BIOPHARMACEUTICAL INDUSTRY:
2017 National and State Estimates, 2019.
[43] 21CFR314.3 – Code of Federal Regulations – Title 21 --Food And Drugs. Chapter I--Food And
Drug Administration, Department Of Health And Human Services. Subchapter D – Drugs for
Human Use. Part 314 -- Applications For FDA Approval to Market a New Drug. Subpart A –
General Provisions. Sec. 314.3 Definitions. (Accessed Nov 9, 2020, at https://www.access
data.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?fr=314.3).
[44] Emami F, Vatanara A, Park E, Na D. Drying technologies for the stability and bioavailability of
biopharmaceuticals. Pharmaceutics 2018; 10: 131.
[45] Bye J, Platts L, Falconer R. Biopharmaceutical liquid formulation: a review of the science of
protein stability and solubility in aqueous environments. Biotechnology Letters 2014; 36:
869–75.
[46] Siew A. Drug delivery systems for biopharmaceuticals. BioPharm International 2015; 28: 14–19.
[47] Amgen (Mfg. by Immunex Corp.). Enbrel [package insert]. 2020.
[48] Anselmo AC, Gokarn Y, Mitragotri S. Non-invasive delivery strategies for biologics nature
reviews. Drug Discovery 2019; 18: 19–40.
[49] Gutka H. Rational selection of sugars for biotherapeutic stabilization: a practitioner’s
perspective. BioProcess International 2018; 16: 40–52.
[50] Khan TA, Mahler H, Kishore RSK. Key interactions of surfactants in therapeutic protein
formulations: a review. European Journal of Pharmaceutics and Biopharmaceutics 2015; 97:
60–67.
[51] European Medicines Agency. ICH Topic Q6A, Specifications: Test Procedures and Acceptance
Criteria for New Drug Substances and New Drug Products: Chemical Substances. 2000.
[52] US Dept. of Health and Human Services, Food and Drug Administration, Center for Drug
Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER).
Guidance for Industry – Q8(R2) Pharmaceutical Development. 2009.
[53] CFR – Code of Federal Regulations Title 21 – Title 21 -- Food and Drugs, Chapter 1 -- Food
and Drug Administration, Department of Health and Human Services. Subchapter C – Drugs:
General. Part 210 – Current Good Manufacturing Practice in Manufacturing, Processing,
Packing, or Holding of Drugs: General. (Accessed Nov 2, 2020 at https://www.accessdata.
fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?CFRPart=210).
 References 315

[54] CFR – Code of Federal Regulations Title 21 – Title 21 -- Food and Drugs, Chapter 1 -- Food and
Drug Administration, Department of Health and Human Services. Subchapter C – Drugs:
General. Part 211 – Current Good Manufacturing Practice for Finished Pharmaceuticals.
(Accessed Nov 2, 2020 at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/
cfrsearch.cfm?cfrpart=211).
[55] Facts About the Current Good Manufacturing Practices (CGMPs) 2019. Center for Drug
Evaluation and Research, U. S. FDA. (Accessed Nov 2, 2020, at https://www.fda.gov/drugs/
pharmaceutical-quality-resources/facts-about-current-good-manufacturing-practices-
cgmps).
[56] Rathore AS. Setting specifications for a biotech therapeutic product in the quality by design
paradigm. BioPharm International 2010; 23.
[57] U.S. Department of Health and Human Services, Food and Drug Administration, Center for
Drug Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER).
Guidance for industry – Q11 Development and Manufacture of Drug Substances. U.S. Food
and Drug Administration 2020.
[58] European Medicines Agency. EMA Assessment Report for Prevenar 13, EMA/798877/2009.
London, England: European Medicines Agency, 2009.
[59] Grifols Therapeutics Inc. Gamunex-C [package insert]. 2020.
[60] Remington KM. Fundamental strategies for viral clearance – Part 1: exploring the regulatory
implications. BioProcess International 2015; 13.
[61] U.S. Dept. of Health and Human Services, Food and Drug Administration, Center for Drug
Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER). Q7
Good Manufacturing Practice Guidance for Active Pharmaceutical Ingredients – Guidance for
Industry. U.S. Food and Drug Administration 2016.
[62] CMC Biotech Working Group. A–Mab: A Case Study in Bioprocess Development: ISPE |
International Society for Pharmaceutical Engineering.
[63] Warikoo V, Godawat R, Brower K, et al. Integrated continuous production of recombinant
therapeutic proteins. Biotechnology and Bioengineering 2012; 109: 3018–29.
[64] Hernandez R. Continuous manufacturing: a changing processing paradigm. Biopharm
International 2015; 28.
[65] Fluid Management Solutions for IBD™ Inline Buffer Dilution and Conditioning Systems.
UME91I01-1.1, Asahi Kasei Bioprocess America, Inc. 2019.
[66] Seiberling DA. Clean-In-Place for Biopharmaceutical Processes. Boca Raton: CRC Press, 2007.
[67] Sandle T. Sterility, Sterilisation and Sterility Assurance for Pharmaceuticals: Technology,
Validation and Current Regulations. Oxford, UK: Woodhead Publishing Ltd., 2013.
[68] United States Pharmacopeia and National Formulary (USP 42-NF37 2S). United States
Pharmacopeial Convention, 2018.
[69] American Society of Mechanical Engineers. Bioprocessing equipment ASME BPE-2019
(Revision of ASME BPE-2016); An American National Standard. New York, NY: ASME, American
Society of Mechanical Engineers, 2019.
[70] ABEC Advances Single-Use Bioreactor Volumes To 6,000 Liters 2019. (Accessed Nov 9, 2020,
at https://www.bioprocessonline.com/doc/abec-advances-single-use-bioreactor-volumes-
to-liters-0001).
[71] Extractables and leachables subcommittee of the bio-process systems alliance.
Recommendations for extractables and leachables testing, Part 1: introduction, regulatory
issues, and risk assessment. BioProcess International 2007; 5: 36–49.
[72] Martin J, Brief A. History of single-use manufacturing. BioPharm International 2011; 5–6.
[73] Barnoon B, Bader B. Lifecycle cost analysis for single-use systems. BioPharm International
2008; 2008 Supplement.
316 References

[74] Jagschies G, Lindskog E, Łącki K, Galliher P. Biopharmaceutical Processing: Development,

Design, and Implementation of Manufacturing Processes. Amsterdam, Netherlands: Elsevier,
2018.
[75] Rader RA, Langer ES. 30 years of upstream productivity improvements. BioProcess
International 2015; 2: 10–15.
[76] U.S. Department of Health and Human Services, Food and Drug Administration, Center for
Drug Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER),
Center for Veterinary Medicine (CVM). Guidance for Industry – Process Validation: General
Principles and Practices. 2011.
[77] Step 3: Clinical Research 2019. U.S. Dept. of Health and Human Services, Food and Drug
Administration, Office of the Commissioner. (Accessed Nov 10, 2020, at https://www.fda.
gov/patients/drug-development-process/step-3-clinical-research).
[78] Jiang C, Flansburg L, Ghose S, Jorjorian P, Shukla AA. Defining process design space for a
hydrophobic interaction chromatography (HIC) purification step: Application of quality by
design (QbD) principles. Biotechnology and Bioengineering 2010; 107: 985–97.
[79] DiMasi JA, Grabowski HG, Hansen RW. Innovation in the pharmaceutical industry: New
estimates of R&D costs. Journal of Health Economics 2016; 47: 20–33.
[80] PDA Task Force on Technical Report No. 60. Technical Report No. 60 – Process Validation:
A Lifecycle Approach: Parenteral Drug Association, Inc., 2013.
[81] European Biopharmaceutical Enterprises. Platform Manufacturing of Biopharmaceuticals:
Putting Accumulated Data and Experience to Work. EBE Concept Paper 2013.
[82] Bioburden control in biopharmaceuticals: managing the risks. 2016. (Accessed Nov 10, 2020,
at https://www.engineersireland.ie/Covid-19-information-base/bioburden-control-in-biophar
maceuticals-managing-the-risks).
[83] Von Wintzingerode F. Biologics production: impact of bioburden contaminations of non-
sterile process intermediates on patient safety and product quality. American Pharmaceutical
Review 2017; 20.
[84] Bowen RJE, Brunen M, Goldschmidt NA, et al. ISPE Baseline Guide: Volume 6 –
Biopharmaceutical Manufacturing Facilities, Second. ISPE | International Society for
Pharmaceutical Engineering, 2013.
[85] ISO/TC 209 Cleanrooms and associated controlled environments. ISO 14644-1:2015,
Cleanrooms and associated controlled environments – Part 1: Classification of air cleanliness
by particle concentration. 2015.
[86] ISO/TC 209 Cleanrooms and associated controlled environments. ISO 14644-4:2001,
Cleanrooms and associated controlled environments – Part 4: Design, construction and start-
up. 2001.
[87] Sun W. Cleanroom airlock performance and beyond. ASHRAE Journal 2018; 60.
[88] EudraLex – Volume 4 – EU Guidelines for Good Manufacturing Practice for Medicinal Products
for Human and Veterinary Use, Annex 15: Qualification and Validation. 2015. (Accessed Feb
28, 2021, at https://ec.europa.eu/health/sites/health/files/files/eudralex/vol-4/2015-10_
annex15.pdf).
[89] Biosafety in Microbiological and Biomedical Laboratories, 5th Edition. HHS Publication No.
(CDC) 21-1112 2009.
[90] Department of Health and Human Services, National Institutes of Health. NIH Guidelines for
Research Involving Recombinant or Synthetic Nucleic Acid Molecules (NIH Guidelines). 2019.
[91] Odum JN, Flickinger MC. Process Architecture in Biomanufacturing Facility Design. Hoboken,
NJ: John Wiley & Sons, Hoboken, NJ:, 2018.
[92] Langer ES. Outsourcing trends in biopharmaceutical manufacturing. Pharmaceutical
Outsourcing 2017; 18.
 References 317

[93] Otto R, Santagostino A, Schrader U Rapid growth in biopharma: Challenges and

opportunities. McKinsey & Company, 2014. (Accessed Nov 1, 2020, at https://www.mckin
sey.com/industries/pharmaceuticals-and-medical-products/our-insights/rapid-growth-in-
biopharma).
[94] Alonso W, Vandeberg P, Lang J, et al. Immune globulin subcutaneous, human 20% solution
(Xembify®), a new high concentration immunoglobulin product for subcutaneous
administration. Biologicals 2020; 64: 34–40.
[95] 21CFR211.25 – Code of Federal Regulations, Title 21, Vol. 4, Chapter I -- Food and Drug
Administration, Department of Health and Human Services. Subchapter C – Drugs: General.
Part 211 -- Current Good Manufacturing Practice for Finished Pharmaceuticals. Subpart B –
Organization and Personnel. Sec. 211.25 Personnel qualifications. 2020. (Accessed Jan 3, 2021,
at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=211.25).
[96] ICH Harmonised Tripartite Guideline Good Manufacturing Practice Guide for Active
Pharmaceutical Ingredients Q7. 2000.
[97] EudraLex – The Rules Governing Medicinal Products in the European Union – Volume 4 –
Good Manufacturing Practice (GMP) guidelines. (Accessed Feb 28, 2021, at https://ec.eu
ropa.eu/health/documents/eudralex/vol-4_en).
[98] People in Cleanrooms: Understanding and Monitoring the Personnel Factor 2014. Sandle T
(Accessed Jan 3, 2021, at http://www.ivtnetwork.com/article/people-cleanrooms-understand
ing-and-monitoring-personnel-factor).
[99] 21CFR211.28 – Code of Code of Federal Regulations, Title 21, Vol. 4, Chapter I -- Food and
Drug Administration, Department of Health and Human Services. Subchapter C – Drugs:
General. Part 211 -- Current Good Manufacturing Practice for Finished Pharmaceuticals.
Subpart B – Organization and Personnel. Sec. 211.28 Personnel responsibilities. 2020.
(Accessed Jan 3, 2021, at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/
CFRSearch.cfm?fr=211.28).
[100] 21CFR211 – Code of Code of Federal Regulations, Title 21, Vol. 4, Chapter I -- Food and Drug
Administration, Department of Health and Human Services. Subchapter C – Drugs: General.
Part 211 -- Current Good Manufacturing Practice for Finished Pharmaceuticals. Subpart D –
Equipment. (Accessed Jan 3, 2021, at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/
cfcfr/CFRSearch.cfm?CFRPart=211&showFR=1&subpartNode=21:4.0.1.1.11.4).
[101] Good Manufacturing Practices (GMP) Guidelines for Active Pharmaceutical Ingredients (API) –
(GUI-0104) 2014. Health Canada, Health Products and Food Branch Inspectorate. (Accessed
Jan 6, 2021, at https://www.canada.ca/en/health-canada/services/drugs-health-products/
compliance-enforcement/information-health-product/drugs/guidelines-active-pharmaceuti
cal-ingredients-0104.html#s5.C.02.005).
[102] 21CFR211.68 – Code of Code of Federal Regulations, Title 21, Vol. 4, Chapter I -- Food and
Drug Administration, Department of Health and Human Services. Subchapter C – Drugs:
General. Part 211 -- Current Good Manufacturing Practice for Finished Pharmaceuticals.
Subpart D – Equipment. Sec. 211.68 Automatic, mechanical, and electronic equipment. 2020.
(Accessed Jan 3, 2021, at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/
CFRSearch.cfm?fr=211.68).
[103] EudraLex – The Rules Governing Medicinal Products in the European Union – Volume 4 –
Good Manufacturing Practice, Medicinal Products for Human and Veterinary Use, Chapter 4:
Documentation. 2011. (Accessed Feb 28, 2021, at https://ec.europa.eu/health/sites/health/
files/files/eudralex/vol-4/chapter4_01-2011_en.pdf).
[104] U.S. Food and Drug Administration. Guidance for Industry, Part 11, Electronic Records;
Electronic Signatures – Scope and Application. 2003.
318 References

[105] 21CFR 211 – Code of Federal Regulations, Title 21--Food and Drugs, Chapter I -- Food and Drug
Administration, Department of Health and Human Services. Subchapter C – Drugs: General.
Part 211 -- Current Good Manufacturing Practice For Finished Pharmaceuticals. Subpart E –
Control of Components and Drug Product Containers and Closures. (Accessed Jan 4, 2021, at
https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?CFRPart=
211&showFR=1&subpartNode=21:4.0.1.1.11.5).
[106] 21CFR211.110 – Code of Federal Regulations, Title 21--Food And Drugs, Chapter I -- Food And
Drug Administration, Department of Health And Human Services. Subchapter C – Drugs:
General. Part 211 -- Current Good Manufacturing Practice For Finished Pharmaceuticals.
Subpart F – Production and Process Controls. Sec. 211.110 Sampling and Testing of In-
Process Materials and Drug Products. (Accessed Jan 4, 2021, at https://www.accessdata.fda.
gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?fr=211.110).
[107] 21CFR211.165 – Code of Federal Regulations, Title 21--Food And Drugs Chapter I--Food And
Drug Administration, Department Of Health And Human Services. Subchapter C – Drugs:
General. Part 211 -- Current Good Manufacturing Practice For Finished Pharmaceuticals.
Subpart I – Laboratory Controls. Sec. 211.165 Testing and release for distribution. (Accessed
Jan 4, 2021, at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?
fr=211.165).
[108] 21CFR820.3 – Code of Federal Regulations – Title 21--Food and Drugs Chapter I--Food And
Drug Administration, Department Of Health And Human Services. Subchapter H – Medical
Devices Part 820 -- Quality System Regulation. Subpart A – General Provisions. Sec. 820.3
Definitions. (Accessed Feb 28, 2021, at https://www.accessdata.fda.gov/scripts/cdrh/
cfdocs/cfcfr/CFRSearch.cfm?fr=820.3).
[109] 21CFR211.192. – Code of Federal Regulations – Title 21--Food and Drugs Chapter I--Food And
Drug Administration, Department Of Health And Human Services. Subchapter C – Drugs:
General. Part 211 -- Current Good Manufacturing Practice For Finished Pharmaceuticals.
Subpart J – Records and Reports. Sec. 211.192 Production record review. (Accessed Feb 28,
2021, at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=211.
192).
[110] Root Cause Analysis for Drugmakers. Washington Business Information Inc, 2013.
[111] What is Root Cause Analysis (RCA)? | ASQ. (Accessed Jan 6, 2020, at https://asq.org/quality-
resources/root-cause-analysis).
[112] EudraLex – The Rules Governing Medicinal Products in the European Union – Volume 4 – EU
Guidelines for Good Manufacturing Practice for Medicinal Products for Human and Veterinary
Use. Part 1. Chapter 8: Complaints, Quality Defects and Product Recalls. 2014. (Accessed Feb
28, 2021, at https://ec.europa.eu/health//sites/health/files/files/eudralex/vol-4/2014-08_
gmp_chap8.pdf).
[113] 21CFR211.198 – Code of Federal Regulations – Title 21--Food And Drugs. Chapter I--Food And
Drug Administration, Department Of Health And Human Services. Subchapter C – Drugs:
General. Part 211 -- Current Good Manufacturing Practice For Finished Pharmaceuticals.
Subpart J – Records and Reports. Sec. 211.198 Complaint files. 2020. (Accessed Feb 28, 2021,
at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?fr=211.198).
[114] 21CFR314.81 – Code of Federal Regulations – Title 21--Food And Drugs. Chapter I--Food And
Drug Administration, Department Of Health And Human Services. Subchapter D – Drugs for
Human Use. Part 314 -- Applications For FDA Approval to Market a New Drug. Subpart B –
Applications. Sec. 314.81 Other postmarketing reports. 2020. (Accessed Feb 28, 2021, at
https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?fr=314.81).
[115] Blanch HW, Yee L. Recombinant protein expression in high cell density fed-batch cultures of
escherichia coli. Bio/Technology (New York, N.Y.1983) 1992; 10: 1550–56.
 References 319

[116] Porro D, Sauer M, Branduardi P, Mattanovich D. Recombinant protein production in yeasts.

Molecular Biotechnology 2005; 31: 245–59.
[117] Gu MB, Todd P, Kompala DS. Foreign gene expression (β‐galactosidase) during the cell cycle
phases in recombinant CHO cells. Biotechnology and Bioengineering 1993; 42: 1113–23.
[118] Monod J. The growth of bacterial cultures. Annual Review of Microbiology 1949; 3: 371–94.
[119] Watanabe I, Okada S. Effects of temperature on growth rate of cultured mammalian cells
(L5178Y). The Journal of Cell Biology 1967; 32: 309–23.
[120] Charoenchai C, Fleet GH, Henschke PA. Effects of temperature, pH, and sugar concentration
on the growth rates and cell biomass of wine yeasts. American Journal of Enology and
Viticulture 1998; 49: 283–88.
[121] Russell JB, Diez-Gonzalez F. The effects of fermentation acids on bacterial growth. Advances
in Microbial Physiology 1998; 39: 205.
[122] Schmidt S. Drivers, opportunities, and limits of continuous processing. BioProcess
International 2017; 15: 30–37.
[123] Steam In Place (Technical Report No. 61): Parenteral Drug Association, Inc., 2013.
[124] U.S. Dept. of Health and Human Services, Food and Drug Administration, Center for Biologics
Evaluation and Research (CBER). Guidance for Industry – Characterization and Qualification
of Cell Substrates and Other Biological Materials Used in the Production of Viral Vaccines for
Infectious Disease Indications. 2010.
[125] ICH Topic Q 5 D – Quality of Biotechnological Products: Derivation and Characterisation of
Cell Substrates Used for Production of Biotechnological/Biological Products. 1998.
[126] Gancedo C, Gancedo JM, Sols A. Glycerol metabolism in yeasts. Pathways of utilization and
production. European Journal of Biochemistry 1968; 5: 165–72.
[127] Guo R, Gu J, Zong S, Wu M, Yang M. Structure and mechanism of mitochondrial electron
transport chain. Biomedical Journal 2018; 41: 9–20.
[128] Boron WF. Regulation of intracellular pH. Advances in Physiology Education 2004; 28: 160.
[129] Orij R, Brul S, Smits GJ. Intracellular pH is a tightly controlled signal in yeast. Biochimica et
biophysica acta. General Subjects 2011; 1810: 933–44.
[130] Booth IR. Regulation of cytoplasmic pH in bacteria. Microbiological Reviews 1985; 49:
359–78.
[131] Los DA, Murata N. Membrane fluidity and its roles in the perception of environmental signals.
Biochimica et biophysica acta. Biomembranes 2004; 1666: 142–57.
[132] Beales N. Adaptation of microorganisms to cold temperatures, weak acid preservatives, low
pH, and osmotic stress: a review. Comprehensive Reviews in Food Science and Food Safety
2004; 3: 1–20.
[133] Han Y, Liu X, Liu H, et al. Cultivation of Recombinant Chinese hamster ovary cells grown as
suspended aggregates in stirred vessels. Journal of Bioscience and Bioengineering 2006;
102: 430–35.
[134] Yeast Cell Architecture and Functions. In: Feldmann H, ed. Yeast: Molecular and Cell Biology.
2nd completely rev. and greatly enl. ed. Weinheim: John Wiley & Sons, Inc, 2012: 5–24.
[135] Orlean P. Architecture and Biosynthesis of the saccharomyces cerevisiae cell wall. Genetics
(Austin) 2012; 192: 775–818.
[136] Reshes G, Vanounou S, Fishov I, Feingold M. Cell shape dynamics in escherichia coli.
Biophysical Journal 2008; 94: 251–64.
[137] Components and Structure | Boundless Biology. (Accessed Nov 18, 2020, at https://courses.
lumenlearning.com/boundless-biology/chapter/components-and-structure/).
[138] Van Der Rest ME, Kamminga AH, Nakano A, Anraku Y, Poolman B, Konings WN. The plasma
membrane of Saccharomyces cerevisiae: structure, function, and biogenesis. Microbiological
Reviews 1995; 59: 304–22.
320 References

[139] Erickson HP. How bacterial cell division might cheat turgor pressure – a unified mechanism
of septal division in Gram-positive and Gram-negative bacteria. Bioessays 2017; 39:
1700045.
[140] Petsch D, Anspach FB. Endotoxin removal from protein solutions. Journal of Biotechnology
2000; 76: 97–119.
[141] Singh SM, Panda AK. Solubilization and refolding of bacterial inclusion body proteins.
Journal of Bioscience and Bioengineering 2005; 99: 303–10.
[142] Baneyx F, Mujacic M. Recombinant protein folding and misfolding in Escherichia coli. Nature
Biotechnology 2004; 22: 1399–408.
[143] Yang Z, Zhang L, Zhang Y, et al. Highly efficient production of soluble proteins from insoluble
inclusion bodies by a two-step-denaturing and refolding method. PLoS One 2011; 6: e22981.
[144] Jalalirad R. Selective and efficient extraction of recombinant proteins from the periplasm of
Escherichia coli using low concentrations of chemicals. Journal of Industrial Microbiology &
Biotechnology 2013; 40: 1117–29.
[145] Harcum SW, Lee KH. CHO cells can make more protein. Cell Systems 2016; 3: 412–13.
[146] Chen R. Bacterial expression systems for recombinant protein production: E. coli and beyond.
Biotechnology Advances 2012; 30: 1102–07.
[147] Vandenberghe LH, Xiao R, Lock M, Lin J, Korn M, Wilson JM. Efficient serotype-dependent
release of functional vector into the culture medium during adeno-associated virus
manufacturing. Human Gene Therapy 2010; 21: 1251–57.
[148] Berman HM, Westbrook J, Feng Z, et al. The protein data bank. Nucleic Acids Research 2000;
28: 235–42.
[149] Salazar O, Asenjo JA. Enzymatic lysis of microbial cells. Biotechnology Letters 2007; 29:
985–94.
[150] Garcia FAP. Cell Wall Disruption. In: Flickinger M, ed. Encyclopedia of IIndustrial
Biotechnology: Bioprocess, Bioseparation, and Cell Technology. John Wiley & Sons, Inc,
2009: 1–12.
[151] Kampen I, Kwade A. Cell Disruption, Micromechanical Properties. In: Flickinger M, ed.
Downstream Industrial Biotechnology: Recovery and Purification. John Wiley & Sons, Inc,
2013: 49–64.
[152] Middleberg A. The Release of Intracellular Bioproducts. In: Bioseparation and Bioprocessing.
Weinheim; New York: Wiley-VCH, 1998: 131–63.
[153] Hetherington P, Follows M, Dunnill P, Lilly M. Release of protein from baker's yeast
(saccharomyces cerevisiae) by disruption in an industrial homogeniser. Transactions of the
Institution of Chemical Engineers 1971; 49: 142–48.
[154] Kleinig AR, Mansell CJ, Nguyen QD, Badalyan A, Middelberg APJ. Influence of broth dilution
on the disruption of Escherichia coli. Biotechnology 1995; 9: 759–62.
[155] Ariete Series, Homogenizers. Ariete Series Homogenizers. GEA Mechanical Equipment, Italia,
S.P.A.
[156] Theory of Separation. Sweden: Alfa Laval, Industrial Separation Division.
[157] Richardson A, Walker J. Continuous solids-discharging centrifugation – a solution to the
challenges of clarifying high–cell-density mammalian cell cultures. BioProcess International
2018; 16: 38–47.
[158] Kubitschek HE. Buoyant density variation during the cell cycle in microorganisms. Critical
Reviews in Microbiology 1987; 14: 73–97.
[159] Axelsson H. Cell Separation, Centrifugation. In: Flickinger MC, ed. Downstream Industrial
Biotechnology: Recovery and Purification. Hoboken, New Jersey: John Wiley & Sons, Inc,
2013: 27–48.
 References 321

[160] Holdich RG, Rushton A. Centrifugal Separation. In: Rushton A, Ward AS, and Holdich RG, eds.
Solid-Liquid Filtration and Separation Technology. Weinheim; New York: Wiley-VCH, 1996.
[161] Bell DJ, Hoare M, Dunnill P. The Formation of Protein Precipitates and their Centrifugal
Recovery. In: Downstream Processing, Advances in Biochemical Engineering/ Biotechnology,
Vol. 26. Berlin: Akademie-Verlag, 1983: 1–72.
[162] Harrison RG, Todd PW, Rudge SR, Petrides DP. Bioseparations Science and Engineering.
New York: Oxford University Press, 2015.
[163] Wang DIC, Cooney CL, Demain AL, Dunnill P, Humphrey AE, Lilly MD. Fermentation and
Enzyme Technology. New York: John Wiley & Sons, Inc, 1979.
[164] CARR® Centritech Separation Systems – CARR® UniFuge® Pilot. Pneumatic Scale Angelus.
Stow, OH.
[165] CARR® Centritech Separation Systems – U2k® Single-Use Centrifuge. Pneumatic Scale
Angelus. Stow, OH.
[166] Hebb MH, Smith FH. Encyclopedia of Chemical Technology, Vol. 3. In: Hebb MH, Smith FH,
and Re K, eds. New York: Interscience Encyclopedia, inc, 1948: 501–05.
[167] Ambler CM. The evaluation of centrifuge performance. Chemical Engineering Progress 1952;
48: 150–58.
[168] Ambler CM. The theory of scaling up laboratory data for the sedimentation type centrifuge.
Journal of Biochemical and Microbiological Technology and Engineering 1959; 1: 185–205.
[169] Harrison RG, Todd PW, Rudge SR, Petrides DP. Bioseparations Science and Engineering
(Topics in Chemical Engineering). New York: Oxford University Press, 2003.
[170] Aiba S, Kitai S, Heima H. Determination of equivalent size of microbial cells from their
velocities in hindered settling. Journal of General and Applied Microbiology 1964; 10: 243–56.
[171] Shukla AA, Kandula JR. Harvest and recovery of monoclonal antibodies from large-scale
mammalian cell culture. BioPharm International 2008; 21.
[172] Sullivan FE, Erikson RA. De Laval’s “KQ value” spelled out. Design factors with the continuous
disk centrifuge as a case in point. Industrial and Engineering Chemistry 1960; 53: 434–38.
[173] Baker RW. Membrane Technology and Applications. Oxford: Wiley-Blackwell, 2012.
[174] O'Brien TP, Brown LA, Battersby DG, Rudolph AS, Raman LP. Large-scale, single-use depth
filtration systems for mammalian cell culture clarification. BioProcess International 2012; 10:
50–57.
[175] Quigley GT. Prefiltration in Biopharmaceutical Processes. In: Meltzer TH and Jornitz MW, eds.
Filtration and Purification in the Biopharmaceutical Industry. Baton Rouge: CRC Press LLC,
2007: 1–21.
[176] Millistak+® HC Pro Fully synthetic depth filters for clarification and downstream filtration
applications. Millipore Data Sheet No. DS1240EN00 Ver. 3.0 2017-08673, Merck KGaA,
Darmstadt, Germany and/or its affiliates. 2017.
[177] Provantage® Services – Millistak+® Pod Filter Holder qualification protocol and services.
Merck Millipore publication Lit No. SDS1014EN00 Rev. C PS-SBU-12-06403, EMD Millipore
Corporation, Billerica, MA. 2015.
[178] Millistak+® family and Clarisolve® depth filters at a glance. Millipore publication
PB1900EN00 Ver. 3.0 EMD Millipore Corporation. 2017.
[179] AllegroTM MVP Single-use System. Pall Life Sciences publication GN13.5711 USD 2881, Pall
Corporation, Port Washington, NY. 2013.
[180] Goldrick S, Joseph A, Mollet M, et al. Predicting performance of constant flow depth filtration
using constant pressure filtration data. Journal of Membrane Science 2017; 531: 138–47.
[181] Lutz H. Rationally defined safety factors for filter sizing. Journal of Membrane Science 2009;
341: 268–78.
322 References

[182] Frahm G, Clarke T, Johnston M. Evaluation of microflow digital imaging particle analysis for
sub-visible particles formulated with an opaque vaccine adjuvant. PLoS One 2016; 11.
[183] Prashad M, Tarrach K. Depth filtration: cell clarification of bioreactor offloads. Filtration &
Separation 2006; 43: 28–30.
[184] Russell E, Wang A, Rathore AS. Harvest of a Therapeutic Protein Product from High Cell
Density Fermentation Broths: Principles and Case Study. In: Shukla AA, Etzel MR, and Gadam
S, eds. Process Scale Bioseparations for the Biopharmaceutical Industry. Boca Raton: CRC/
Taylor & Francis, 2007: 1–58.
[185] Yavorsky D, Blanck R, Lambalot C, Brunkow R. The clarification of bioreactor cell cultures for
biopharmaceuticals. Pharmaceutical Technology ((2003)) 2003; 27: 62–76.
[186] Pmax™/Tmax™ Constant Flow Rate Test for Depth Filter Sizing. Millipore Lit. No.
MS_BR4506EN Ver. 2.0 33211, Merck KGaA, Darmstadt, Germany and/or its affiliates. 2020.
[187] Millistak+® Pod disposable depth filter system. Millipore Data Sheet Lit. No. DS0217EN00
Ver. 11.0 2018-10419, Merck KGaA, Darmstadt, Germany and/or its affiliates, Burlington, MA.
2019.
[188] Yigzaw Y, Piper R, Tran M, Shukla AA. Exploitation of the adsorptive properties of depth filters
for host cell protein removal during monoclonal antibody purification. Biotechnology
Progress 2006; 22: 288–96.
[189] Zhou JX, Solamo F, Hong T, Shearer M, Tressel T. Viral clearance using disposable systems in
monoclonal antibody commercial downstream processing. Biotechnology and Bioengineering
2008; 100: 488–96.
[190] Lutz H. Ultrafiltration for Bioprocessing. Development and Implementation of Robust
Processes. Waltham, MA, USA: Elsevier, 2015.
[191] Hughes DJ, Cui Z, Field RW, Tirlapur UK. In situ three-dimensional characterization of
membrane fouling by protein suspensions using multiphoton microscopy. Langmuir 2006;
22: 6266–72.
[192] Chitale KC, Presz W, Ross-Johnsrud BP, Hyman M, Lamontagne M, Lipkens B Particle
manipulation using macroscale angled ultrasonic standing waves. Proceedings of Meetings
on Acoustics Jun 25, 2017; 30.
[193] Cadence™ Acoustic Separator. Pall Life Sciences GN16.9974 USD3133(2), Pall Corporation,
Port Washington, NY. 2017.
[194] MilliporeSigma Becomes the First to Use Acoustic Technology for Cell Therapy Manufacturing
[press release]. 2019. (Accessed Dec 2, 2020, at https://www.emdmillipore.com/US/en/
20191009_172025).
[195] Aunins JG, Wang DIC. Induced flocculation of animal cells in suspension culture.
Biotechnology and Bioengineering 1989; 34: 629–38.
[196] pDADMAC flocculant reagent for use with Clarisolve® depth filters. Merck Millipore
Application note, Lit. No. AN33330000 Version 4.0 PS-14-09928, EMD Millipore Corporation,
Billerica, MA. 2015.
[197] Eichholz C, Silvestre M, Franzreb M, Nirschl H. Recovery of lysozyme from hen egg white by
selective magnetic cake filtration. Engineering in Life Sciences 2011; 11: 75–83.
[198] Wang A, Lewus R, Rathore AS. Comparison of different options for harvest of a therapeutic
protein product from high cell density yeast fermentation broth. Biotechnology and
Bioengineering 2006; 94: 91–104.
[199] U.S. Dept. of Health and Human Services, Food and Drug Administration, Center for Drug
Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER).
Guidance for Industry – Q6B Specifications: Test Procedures and Acceptance Criteria for
Biotechnological/Biological Products. 1999.
 References 323

[200] Rosenberg AS. Effects of protein aggregates: an immunologic perspective. The AAPS Journal
2006; 8: E501–7.
[201] Vlasak J, Ionescu R. Fragmentation of monoclonal antibodies. Monoclonal Antibodies 2011; 3:
253–63.
[202] Murphy SL, High KA, Maus MV, et al. CD8 + T-cell responses to adeno-associated virus capsid
in humans. Nature Medicine 2007; 13: 419–22.
[203] Patel J, Kothari R, Tunga R, Ritter NM, Tunga BS. Stability considerations for
biopharmaceuticals, Part 1 – overview of protein and peptide degradation pathways.
BioProcess International 2011; 20–31.
[204] Annex 1 – Requirements for the use of animal cells as in vitro substrates for the production of
biologicals (Requirements for Biological Substances No. 50): World Health Organization,
1998.
[205] Dept. of Health, Education, And Welfare, Public Health Service, Food And Drug
Administration, *ORA/ORO/DEIO/IB*. FDA Inspection Technical Guide Number 40, Bacterial
Endotoxins/Pyrogens. 1985.
[206] U.S. Pharmacopeia (39th Revision) and National Formulary (34th Edition), USP 39 NF 34,
(1132) Residual Host Cell Protein Measurement in Biopharmaceuticals. Rockville, MD: The
United States Pharmacopeial Convention, 2016.
[207] Chon JH, Zarbis-Papastoitsis G. Advances in the production and downstream processing of
antibodies. New Biotechnology 2011; 28: 458–63.
[208] Remington KM. Fundamental strategies for viral clearance, Part 2: technical approaches.
BioProcess International 2015; 13: 10–17.
[209] Boström M, Tavares FW, Bratko D, Ninham BW. Specific ion effects in solutions of globular
proteins: comparison between analytical models and simulation. The Journal of Physical
Chemistry B 2005; 109: 24489–94.
[210] Carta G, Jungbauer A. Protein Chromatography: Process Development and Scale-Up.
Hoboken: Wiley-VCH Imprint], Hoboken: John Wiley & Sons, Incorporated, 2020.
[211] Ion Exchange Chromatography – Principles and Methods. GE Healthcare handbook 11000421
AC, General Electric Company. 2006.
[212] Ion Exchange Chromatography | LSR | Bio-Rad. (Accessed Jan 30, 2021, at bio-rad.com).
[213] POROS™ CaptureSelect™ AAV Resins: AAV8, AAV9, AAVX User Guide. Pub. No. 100038399
Rev. E 2017.
[214] Hydrophobic Interaction and Reversed Phase Chromatography – Principles and Methods. GE
Healthcare handbook 11-0012-69 AA, General Electric Company. 2006.
[215] Sartobind® Membrane Adsorbers Chromatography as Easy as Filtration. Sartorius Stedim
Biotech GmbH. Publication No.: SL-1533-e180509 · Order No.: 85037-544-14 · Ver. 05 | 2018.
[216] Mustang® Chromatography Products. Pall Corporation. (Accessed Jan 30, 2021, at https://
shop.pall.com/us/en/biotech/chromatography/membranes/zidgri78lef).
[217] Products for preparative applications. BIA Separations. (Accessed Jan 30, 2021, at https://
www.biaseparations.com/en/products/monolithic-columns/products-for-preparative-
applications).
[218] Subramanian G. Biopharmaceutical Production Technology. Weinheim: Wiley-VCH Verlag
GmbH & Co. KGaA, 2012.
[219] Shukla AA, Hubbard B, Tressel T, Guhan S, Low D. Downstream processing of monoclonal
antibodies – Application of platform approaches. Journal of Chromatography. B, Analytical
Technologies in the Biomedical and Life Sciences 2007; 848: 28–39.
[220] Janson J-C. Protein Purification: Principles, High Resolution Methods, and Applications.
Hoboken, N.J.: John Wiley & Sons, 2011.
324 References

[221] Karger BL, Snyder LR, Horvath C. An Introduction to Separation Science. New York: John Wiley &
Sons, 1973.
[222] Van Deemter JJ, Zuiderweg FJ, Klinkenberg A. Longitudinal diffusion and resistance to mass
transfer as causes of nonideality in chromatography. Chemical Engineering Science 1956; 5:
271–89.
[223] Boi C, Malavasi A, Carbonell RG, Gilleskie G. A direct comparison between membrane
adsorber and packed column chromatography performance. Journal of Chromatography.
A 2020; 1612: 460629.
[224] Goyon A, Excoffier M, Janin-Bussat M, et al. Determination of isoelectric points and relative
charge variants of 23 therapeutic monoclonal antibodies. Journal of Chromatography.B,
Analytical Technologies in the Biomedical and Life Sciences 2017; 1065–1066: 119–28.
[225] Gottschalk U. Process Scale Purification of Antibodies. Hoboken, NJ: John Wiley & Sons, Inc.,
2017.
[226] Bemberis I, Noyes A, Natarajan V. Column packing for process-scale chromatography:
guidelines for reproducibility. Biopharm International 2003; 2003 Supplement.
[227] GE Healthcare Application note 28-9372-07 AA – Column efficiency testing. 28-9372-07 AA
01/2010. General Electric Company. 2010.
[228] ÄKTA ready Chromatography Systems. 2020. CY12285-14Apr20-DF, Cytiva. 2020.
[229] Sartorius Product Datasheet – Allegro™ Fully Automated, Single-Use Chromatography
System – For control, flexibility and ease of use in manufacturing. 2020.
[230] Mobius® FlexReady Solution with Smart Flexware™ Assemblies for Chromatography and TFF –
For Purification of Clinical and Process Scale Biologics. Millipore Data Sheet DS2251EN00 Ver.
7.0 2019-26771, Merck KGaA, Darmstadt, Germany and/or its affiliates. 2020.
[231] Mirasol F. Single-use for downstream chromatography: benefit or hindrance? BioPharm
International 2019; 32: 34–37.
[232] Unparalleled flexibility – OPUS® Pre-packed Chromatography Columns. Repligen Corporation.
2018.
[233] Fahrner RL, Knudsen HL, Basey CD, et al. Industrial purification of pharmaceutical antibodies:
development, operation, and validation of chromatography processes. Biotechnology &
Genetic Engineering Reviews 2001; 18: 301–27.
[234] Hanke AT, Ottens M. Purifying biopharmaceuticals: knowledge-based chromatographic
process development. Trends in Biotechnology (Regular ed.) 2014; 32: 210–20.
[235] Shukla AA, Etzel MR, Gadam S. Process Scale Bioseparations for the Biopharmaceutical
Industry. Boca Raton: CRC/Taylor & Francis, 2007.
[236] PreDictor 96well filter plates and Assist software – HighThroughput Process Development.
CY13663-29Jul20-DF, Cytiva. 2020.
[237] OPUS® RoboColumn® Pre-packed Chromatography Columns User Guide Repligen
Corporation. 01-08R-E02. 2018.
[238] Welsh JP, Petroff MG, Rowicki P, et al. A practical strategy for using miniature
chromatography columns in a standardized high-throughput workflow for purification
development of monoclonal antibodies: a practical strategy for using miniature
chromatography columns. Biotechnology Progress 2014; 30: 626–35.
[239] Jensen OE, Kidal S. Using volumetric flow to scaleup chromatographic processes. BioPharm
International 2006; 19.
[240] Bracewell DG, Pathak M, Ma G, Rathore AS. Re-use of protein a resin: fouling and economics.
BioPharm International 2015; 28.
[241] Zhang M, Miesegaes GR, Lee M, et al. Quality by design approach for viral clearance by
protein a chromatography. Biotechnology and Bioengineering 2014; 111: 95–103.
 References 325

[242] Strauss DM, Cano T, Cai N, et al. Strategies for developing design spaces for viral clearance
by anion exchange chromatography during monoclonal antibody production. Biotechnology
Progress 2010; 26: 750–55.
[243] U.S. Department of Health and Human Services, Food and Drug Administration, Center for
Drug Evaluation and Research (CDER), Center for Biologics Evaluation and Research (CBER).
Guidance for Industry – Q5A Viral Safety Evaluation of Biotechnology Products Derived From
Cell Lines of Human or Animal Origin. 1998.
[244] MilliporeSigma– Protein Concentration and Diafiltration by Tangential Flow Filtration.
Millipore Technical Brief Lit. No. TB032 Rev. C 06/03 03-117, Millipore Corporation, Billerica,
MA. 2003.
[245] Challener CA. Evolving UF/DF capabilities. BioPharm International 2018; 31: 24–27.
[246] Allegro™ Fully Automated, Single-Use Tangential Flow Filtration Systems. Pall Life
GN16.6449 USD3120, Pall Corporation, Port Washington, NY. 2016.
[247] Mobius® FlexReady Solution for TFF. Millipore Data Sheet Lit. No. DS3125EN00 Rev. D 09/11
DP SBU-09-01518, Millipore Corporation, Billerica, MA. 2011.
[248] Sartorius – Flexact® Modular | Single-use Automated Solutions. (Accessed Dec 11, 2020, at
https://www.sartorius.com/en/products/process-filtration/flexact-single-use-automated-
solutions).
[249] Latulippe DR, Molek JR, Zydney AL. Importance of biopolymer molecular flexibility in
ultrafiltration processes. Industrial & Engineering Chemistry Research 2009; 48: 2395–403.
[250] Van Reis R, Zydney A. Bioprocess membrane technology. Journal of Membrane Science 2007;
297: 16–50.
[251] Spectrum Hollow Fiber Membranes. (Accessed Dec 16, 2020, at https://www.repligen.com/
technologies/spectrum-hollow-fibers/hollow-fiber-membranes).
[252] Selection Handbook – Hollow fiber cartridges and systems for membrane separations. GE
Healthcare literature 28-9409-67 AD 10/2008, General Electric Company. 2008.
[253] Pellicon®3 Cassettes with Ultracel® Membrane. Millipore Data Sheet Lit. No. MS_DS1285EN
Ver. 3.0 2019 – 23065, Merck KGaA, Darmstadt, Germany and/or its affiliates. 2019.
[254] Yee Lau S, Pattnaik P, Raghunath B. Integrity testing of ultrafiltration systems for
biopharmaceutical applications. BioProcess International 2012; 10: 52–66.
[255] ProCell™ large volume process-scale cartridges. Data file 18-1168-06 AD, GE Healthcare Life
Sciences. 2013.
[256] MaxCell™ process-scale hollow fiber cartridges. 2013. Data file 18-1168-02 AD GE Healthcare
Life Sciences. 2013.
[257] Ultracel® Membranes – The membrane of choice for ultra-low protein binding and robust
process performance. Lit No. PF1401EN00 Ver. 2.0 10/2015. EMD Millipore Corporation,
Billerica, MA. 2015.
[258] Pellicon® Capsules with Ultracel® Membrane – Innovative single-use tangential flow
filtration (TFF) devices for streamlined bioprocessing. Millipore Data Sheet Lit. No.
MS_DS1285EN Ver. 3.0 2019 – 23065, Merck KGaA, Darmstadt, Germany and/or its affiliates.
2019.
[259] Biomax® Membranes – The membrane of choice for fast processing and exceptional
chemical resistance. Lit No. PF1402EN00 Ver. 2.0 10/2015. EMD Millipore Corporation,
Billerica, MA. 2015.
[260] Prostak™ Microfiltration Modules – For Convenient and Economical Perfusion and
Clarification/Concentration Applications. Millipore Data Sheet DS4775EN00 Ver. 4.0 2016 –
00991, Merck KGaA, Darmstadt, Germany and/or its affiliates. 2018.
[261] Microfiltration Cassettes with Supor® Membrane. USD 3253, GN17.10044, Pall Life Sciences,
Port Washington, NY. 2017.
326 References

[262] Suspended-Screen Ultrafiltration Cassettes with Omega™ Membrane. USD 3252,

GN17.10042, Pall Life Sciences. Port Washington, NY. 2017.
[263] Pall Tangential Flow Filtration Cassettes. (Accessed Feb 13, 2021, at https://shop.pall.com/
us/en/biotech/tangential-flow-filtration/cassettes).
[264] Flat Sheet Membranes. (Accessed Feb 13, 2021, at https://www.repligen.com/technologies/
tangenx-tff/membranes).
[265] Hydrosart® Ultrafiltration Cassettes – Protein purification, concentration and diafiltration.
Publication No.: SPC2063-e190705 · Order No.: 85034-534-90 · Ver. 07 | 2019. Sartorius
Stedim Biotech GmbH. 2019.
[266] Hydrosart® Microfiltration Cassettes – Cell harvest and bacteria concentration. Publication
No.: SPC2062-e12033 Order No.: 85034-534-89 Ver. 03 | 2012. Sartorius Stedim Biotech
GmbH. 2012.
[267] Polyethersulfone Ultrafiltration Cassettes – Protein purification, concentration and
diafiltration. Publication No.: SPC2059-e12034 Order No.: 85032-532-86 Ver. 03 | 2012.
Sartorius Stedim Biotech GmbH. 2012.
[268] Polyethersulfone Microfiltration Cassettes – Cell removal and mycoplasma reduction.
Publication No.: SPC2058-e12034 Order No.: 85032-532-85 Ver. 03 | 2012. Sartorius Stedim
Biotech GmbH. 2012.
[269] Pellicon ® 2 Cassettes Installation, User, and Maintenance Guide. MK_UG2996EN Rev 2 01/
2019 Millipore. 2019.
[270] Korson L, Drost-Hansen W, Millero FJ. Viscosity of water at various temperatures. The Journal
of Physical Chemistry 1969; 73: 34–39.
[271] Pellicon® Capsules for Single-Use Tangential Flow Filtration – Enhancing ease-of-use, batch
turnaround, and process flexibility. Millipore Application Note Lit. No. MS_AN2608EN 2017-
08175, Merck KGaA, Darmstadt, Germany and/or its affiliates. 2018.
[272] Vilker VL, Colton CK, Smith KA, Green DL. The osmotic pressure of concentrated protein and
lipoprotein solutions and its significance to ultrafiltration. Journal of Membrane Science
1984; 20: 63–77.
[273] Binabaji E, Rao S, Zydney AL. The osmotic pressure of highly concentrated monoclonal
antibody solutions: Effect of solution conditions. Biotechnology and Bioengineering 2014;
111: 529–36.
[274] Van Reis R, Goodrich EM, Yson CL, Frautschy LN, Dzengeleski S, Lutz H. Linear scale
ultrafiltration. Biotechnology and Bioengineering 1997; 55: 737–46.
[275] Parenteral Drug Association Technical Report No. 15 (Revised 2009): Validation of Tangential
Flow Filtration in Biopharmaceutical Applications. 2009. (Accessed Feb 28, 2021, at https://
store.pda.org/TableOfContents/TR1509_TOC.pdf).
[276] De Los Reyes G, Mir L. Method and apparatus for the filtration of biological solutions. 2008;
11/615(028): 1–56.
[277] Cadence™ Systems Employ New Single-Pass TFF Technology to Simplify Processes and Lower
Costs – Understanding the New Technology. Pall Life Sciences publication Application Note
GN11.4802 USD 2789. Pall Corporation. 2011.
[278] BioContinuum™ Ultrafiltration Platform – Pellicon® Single-Pass Tangential Flow Filtration.
Millipore Data Sheet Lit. No. MS_DS2735EN Ver. 2.0, Merck KGaA, Darmstadt, Germany and/
or its affiliates. 2019.
[279] Cadence™ Single-Pass TFF Modular Kit. USD3284 GN18.07131. Pall Biotech. Port Washington,
NY, 2018.
[280] Cadence™ Inline Diafiltration Module. USD3248 GN17.0912. Pall Life Sciences. Port
Washington, NY, 2017.
 References 327

[281] Langer ES, Rader RA. Biopharmaceutical manufacturing is shifting to single-use systems. Are
the dinosaurs, the large stainless steel facilities, becoming extinct? American Pharmaceutical
Review 2018; 21.
[282] Whitford W, Nelson D. From interest in intensification to a factory of the future. BioProcess
International 2019; 17: 24–30.
[283] U.S. Department of Health and Human Services Food and Drug Administration Center for Drug
Evaluation and Research (CDER) Center for Veterinary Medicine (CVM) Office of Regulatory
Affairs (ORA). Guidance for Industry: PAT – A Framework for Innovative Pharmaceutical
Development, Manufacturing, and Quality Assurance. 2004.
[284] Jones S. BioPhorum operations group technology roadmapping, Part 2 – efficiency,
modularity, and flexibility as hallmarks for future key technologies. BioProcess International
2017; 15: 14–19.
[285] Jiang M, Severson KA, Love JC, et al. Opportunities and challenges of real-time release testing
in biopharmaceutical manufacturing. Biotechnology and Bioengineering 2017; 114: 2445–56.
[286] Glossary – EudraLex – Volume 4 – Good Manufacturing Practice (GMP) guidelines. 2004.
(Accessed Feb 28, 2021, at https://ec.europa.eu/health/sites/health/files/files/eudralex/
vol-4/pdfs-en/glos4en200408_en.pdf).
[287] ISO/TC 209 Cleanrooms and associated controlled environments. ISO 14644-2:2015
Cleanrooms and associated controlled environments – Part 2: Monitoring to provide evidence
of cleanroom performance related to air cleanliness by particle concentration. 2015.
[288] ICH Harmonised Tripartite Guideline, Pharmaceutical Development Q8(R2). 2009.
[289] Good Manufacturing Practice (GMP) Resources. (Accessed Feb 24, 2021, at https://ispe.org/
initiatives/regulatory-resources/gmp).
[290] U.S. Food & Drug Administration, Office of Regulatory Affairs Center for Drug Evaluation and
Research Center for Biologics Evaluation and Research. Guidance Document, Sterile Drug
Products Produced by Aseptic Processing – Current Good Manufacturing Practice, Guidance
for Industry.
Index
adeno-associated virus 15, 24, 134, 213, 272 bioreactor mixing
adenovirus 15 – dynamics 118, 125
aeration efficiency 121 – fluid dynamic model 123
affinity chromatography – strategy 106
agitation in a bioreactor bioreactor modes of operation
– agitator shaft 105 – batch 100
– impeller tip velocity 120 – continuous 102
– impellers 105, 118 – fed batch 101
air locks 52–53, 55 – perfusion 102
ALCOA 78 bioreactor production run 99
A-Mab case study 47 bioreactor types
amino acids 3, 212 – single-use 103
animal cells – traditional (reusable) bioreactors 103
– lysis of 137 bulk filling 29
– structure of 130–131
antibody-drug conjugates 11, 13, 256 calibration of equipment 72
aseptic technique 70 CAPA. See corrective and preventive action
autoclaves 35 (CAPA) under current Good Manufacturing
Avastin® 11, 13, 20, 27 Practice
cell culture definition 27
bacteria structure 130–131, 147 cell division 94
batch records 42, 75, 77 cell growth
batch review process 87 – auxotroph 114
biofilms 109 – comparison among cell types 93
biological products 7 – doubling time 102
biopharmaceuticals 1, 8, 43 – doubling 94
– classification of 10 – effect of oxygen on 114
– compared to traditional pharmaceuticals 8 – effect of pH on 115
– definition of 1 – effect of temperature on 115
– examples of 9–10, 12, 14, 16, 20 – generations 102
– impact of 17 – genetic mutation 102
– life cycle of 43 – kinetics 92
bioreactor design – phases 93
– design criteria 103 – process parameters 95–96, 116
– drive motor 106 cell lysis 133
– feed ports 108 – by enzymatic treatment 137, 139
– flow paths 109 – by freeze/thaw 137, 139
– hygienic design 108 – by high-pressure homogenization 138
– material of construction 103 – by organic solvent treatment 137
– multi-layer films 104 – by osmotic shock 137, 139
– rocker bags 106 – by sonication 138–139
– sanitary seal 106 – by surfactant treatment 137
– sensors 37 – different methods for 148
– single-use bioreactors 103, 105–106 – of animal cells 137
– soft parts 103 – of bacteria 137
– sparger 107 – of E. coli 137, 139, 145
– spray device 110 – of S. cerevisiae 137, 139

https://doi.org/10.1515/9783110616880-013
330 Index

– of yeast 137 – binding capacity 225, 227

– using a Microfluidizer® 138–139 – breakthrough of BSA on Q Sepharose® FF
– using bead mills 138–139 column 227
cell proliferation 93, 95, 114 – dynamic 225
cell structure – equilibrium 225
– animal cells 131 – breakthrough studies 226
– bacteria 130–132, 147 – chromatogram 210, 236
– cell membrane 129, 131 – column volumes (CVs) definition 231
– cell wall 129 – columns 236
– Chinese hamster ovary (CHO) cells 130 – Cytiva AKTAprocess skid 38, 235
– E. coli 130 – Cytiva BPG 300/500 column 235
– eukaryotic cells 129 – expanded bed chromatography 152
– gram-negative bacteria 130–131 – flow-through mode 229
– gram-positive bacteria 130, 132 – height equivalent of a theoretical plate
– plant cells 148 (HETP) 223, 238
– prokaryotic cells 130 – high-performance liquid chromatography
– S. cerevisiae 130, 147 (HPLC) 207
– yeast 131, 147 – hydrodynamic dispersion in 224
cell therapies 16, 24 – hydrophobic interaction chromatography 214
– CAR-T 17 – intrabatch cycling 232
– examples of 16 – ion exchange chromatography 214
centrifugation 134, 152–176, 198 – linear gradient elution 232
– applications in biopharmaceutical – membrane adsorbers 218
processes 152 – monoclonal antibody (mAb) purification 240
– centrifugal force 153 – monoliths 218
– decanter centrifuges 165 – multimodal chromatography 215
– disc-stack centrifuges. See disc-stack – packed beds 217
centrifuges – porosity 217
– multi-chamber bowl centrifuges 165 – principles (basic) of 210, 248
– principles of 153 – procedures for production 230, 249
– production centrifuges other than disc- – process and performance parameters for 240
stack 164 – process design 249
– relative centrifugal force 153 – process development studies for 242
– single-use centrifuge examples 164 – productivity 229
– tubular centrifuges 165 – purification of protein from an E. coli
cGMP. See current Good Manufacturing Prac- lysate 237
tice (cGMP) – residence time of liquid in a packed bed 222
Chinese hamster ovary (CHO) cells 58, 130, – resins 213, 218
158, 188, 200 – resolution 225
– depth filtration of 188 – reversed phase chromatography 215
– flocculation of 200 – scale-up 244
– settling velocity 158 – sensors 37, 233, 235
– stucture of 130 – conductivity measurement 235
chromatography 206–252 – UV absorbance measurement 235
– affinity chromatography 214 – single-use systems 239
– asymmetry measurement 238 – size exclusion chromatography 215
– axial molecular diffusion 224 – skids 233
– bind-and-elute 229 – step elution 232
 Index 331

– step yield calculation for 228 – disposition of batches 87

– superficial velocity in a packed bed 221 – documentation practices 75
– systems for production 234 – documents 74
– transfer of solute to stationary phase 214, – guidelines and guidance documents 66
218–219, 224 – personnel gowning 69
– ultra-high-performance liquid – post-marketing surveillance 88
chromatography (UHPLC) 207 – testing requirements 79
– validation considerations for 246–248 – validation 80
– blank runs 247
– lifetime studies 247 dead legs 109
– viral clearance 248 delivery of oxygen in a bioreactor 121
CIP. See clean in place (CIP) 35 denaturation of proteins 6, 140
clean compressed gases 36 depth filtration 136, 152, 178, 180–195, 198
clean in place (CIP) 35, 72, 110, 167 – 0.2 µm filter post depth filter 185
– of bioreactor 110 – applications in biopharmaceutical
– of disc-stack centrifuges 167 processing 152, 178
clean out of place (COP) 35 – capacity definition 188
clean steam 36 – depth filters 182
clean utilities 35 – feed flux definition 187
cleaning validation 81 – filter fouling 189
cleaning – for removal of soluble impurities 194
– by flow 109 – module holders for 182
– CIP 35 – modules for 182
– definition 35 – of CHO culture 190
– performance evaluation 81, 110 – of P. pastoris broth 190
– validation 110 – particle retention mechanisms in 185
cleanrooms 52–53, 70 – performance parameters for 190
– EU classification of 55 – procedures for 186
– ISO classification of 55 – process parameters for 190
clinical trials 43 – production equipment for 185–186
closed processing 51 – single-use skids 186
computerized systems 73 – screening and sizing studies for 193
conductivity 235 design of experiment (DoE) 47, 117
contaminants 208 design space 117
control strategy 49 deviations 84
corrective and preventive action (CAPA) 85 diafiltration (DF). See diafiltration (DF) under ul-
COVID-19 14 trafiltration (UF)
CQA. See critical quality attributes (CQAs) diaphragm valves 109
critical quality attributes (CQAs) 21, 43 disc-stack centrifuges 159–164
cross contamination 51 – Alfa Laval LAPX 404 38, 161
current Good Manufacturing Practice – CIP of 167
(cGMP) 22, 43, 64–65 – performance parameters for 168
– as it applies to equipment 71 – procedure for cGMP operations 164
– as it applies to personnel 67 – process parameters for 168
– change control 84 – sigma analysis and scale up 170
– cleanroom behaviors 70 disposition of batches 87
– corrective and preventive action (CAPA) 85 DNA 3
– deviations 84 documentation 74
332 Index

– controlled document process flow 75 – classification of 15

– procedures 74 – definition of 15
– records, reports 74 – examples of 16
– specifications 74 genetic engineering 3
DoE. See design of experiment glycosylation 92
drug definition 7 good documentation practice 75
drug product 80, 87 – corrections 76
– definition of 18 – electronic records 79
– disposition of 87 Good Manufacturing Practice. See current Good
– liquid solution 18 Manufacturing Practice (cGMP)
– lyophilized 18 gowning 52–53, 69
– quality of 20 Gram stain test 132
– testing 80 green fluorescent protein (GFP) 143, 236
drug product from drug substance 31 growth medium 95, 114, 117
drug substance 21, 29, 32, 80, 87 – carbon source growth medium 113
– disposition of 87 – nitrogen source 114
– testing 80 – optimal formulation 117
– spent medium analysis 117
E. coli 2, 58, 133–134, 137, 158 – standardized media 117
– cell structure of 130 – vitamins and cofactors 114
– lysis of 137
– settling velocity 158 harvest processes for intracellular and
– source of endotoxin 132 extracellular products 136
endotoxin 132, 208 headspace in a bioreactor 107
entrained foam in a bioreactor 122 heat in a bioreactor 122
environmental monitoring 80 – as a waste product 122
excipients 19 – cooling fluid for heat removal 122
height equivalent of a theoretical plate
fermentation definition 27 (HETP) 223, 238
filtration. See filtration HEPA filters. See high-efficiency particulate air
– applications in biopharmaceutical (HEPA) filters
processing 152, 178 HETP. See height equivalent of a theoretical
– definition of 177, 200, 254 plate (HETP) under chromatography
– depth filtration 178 high purity water (HPW) 35
– normal flow (NFF) 178 high-efficiency particulate air (HEPA) filters 52
– reverse osmosis 195 high-pressure homogenizers 138, 140, 149
– solids retention 180 – for lysis of E. coli 143
– surface vs. depth 180 – heat generation in 140
– tangential flow (TFF) 178 – homogenizer valve parts 140
– transmembrane pressure definition 180 – Niro Soavi Type NS3006H 141
– types of membrane separations 196 – performance parameters for 143
– ultrafiltration (UF) 253 – procedure for 147
Flucelvax® 2, 8–9, 14 – process parameters for 143
freeze drying. See lyophilization host cell DNA 208
host cell proteins (HCPs) 209
gas feeds to a bioreactor 107 host selection 92
gene therapies 24
 Index 333

HPLC. See high-performance liquid chromatog- media prep 34

raphy (HPLC) under chromatography metabolic processes in cells 93
Humira® 8–10, 13, 23–24, 40 microbial contamination 51
hydrophobic interaction chromatography mixed mode chromatography 215–216, 248
214, 216, 220, 248 monoclonal antibodies (mAbs) 8–9, 13, 17, 27,
32, 240
IgG 8–9, 27 – process for production of 32
immunoglobulin G (IgG). See IgG – purification by chromatography 240
impurities 207, 248 – structure of 9
– definition of 207 Monod model of cell growth 94
– process-related 207–209, 248
– endotoxin 208 open processing 51
– host cell DNA 208 OQ. See operational qualification (OQ) under
– host cell proteins (HCPs) 209 qualification
– leachables 209 osmotic pressure 273
– product-related 207, 248 outsourcing 58
inclusion bodies 133, 152 oxygen transfer in a bioreactor
influenza virus 2, 9, 24, 58 – gas to liquid 121
in-process measurements in upstream – interfacial surface area 122
processes 116 – specific mass transfer coefficient 122
insulin 1, 6
– history of 1 parenteral delivery 8, 18
– Humulin® 2 parts washers 35
– structure of 6 passaging cells 99
International Council for Harmonization of peptides 7
Technical Requirements for percent recovery 41
Pharmaceuticals for Human Use (ICH) 66 percent yield 41
ion exchange chromatography 214, 215, performance parameter
220–221, 248 peristaltic pump 109
IQ. See installation qualification (IQ) under personnel gowning. See gowning 69
qualification personnel hygiene 68
platform processes 50
leachables 39, 209 post-marketing surveillance 88
lentivirus 15–17, 58 power/volume for a bioreactor 119
lyophilization 18 precipitation 152
process design 42
mAbs. See monoclonal antibodies process development upstream 116
Madin-Darby canine kidney (MDCK) cells 2, process parameter
8–9, 28 process qualification 43
manufacturing facility design 50 process validation 43, 82–83, 284
master cell bank – chromatography steps 246
– characterization 98 – process qualification 43
– generation 98 – process verification 43
– storage 112 – studies 83
– testing 112 – tangential-flow filtration steps 284
material attributes 46 – three stages 82
MDCK cells. See Madin-Darby canine kidney process yield 41
(MDCK) cells processes for biopharmaceutical production 26
334 Index

– batch 33 – for S. cerevisiae 158

– continuous 33 shear rates during mixing 106
– downstream 31 single-use equipment 37
– objectives of 26 – advantages and disadvantages 37
– stages of 26 – history 39
– support activities 34 SIP. See steam in place (SIP)
– upstream 31 size exclusion chromatography 215, 216, 248
product formation 96 skid definition 37
product yield in bioreactors 96 solid-liquid separation by gravity 154
productivity 41, 100, 117 solid-liquid separations
protein therapeutics 10, 24 – acoustic wave 199
– classes of 12 – cell flocculation 199
– definition of 10 – centrifugation 152
– examples of 12 – comparison among methods 198
proteins 3 – examples in biopharmaceutical
– definition of 3 processes 152
– hydrophobicity of 213 – flocculation of CHO cells 200
– isoelectric point of 212 – magnetic cake filtration 200
– properties of 213 solution prep 34
– structure of 4 SOP (Standard Operating Procedure) 75
specifications 22, 33, 74
QTPP. See quality target product profile (QTPP) – for monoclonal antibody drug product 22
qualification 80 standard operating procedures (SOPs) 75
– documentation 80 steam in place (SIP) 35
– installation qualification (IQ) 81 step yield 41
– operational qualification (OQ) 81 sterilization 35, 111
– performance qualification (PQ) 81 substrate consumption by cells 95
quality target product profile (QTPP) 21, 43 subsurface gas addition to a bioreactor 107

recombinant medicines 3 tangential-flow microfiltration (MF) 152–153, 178,

reversed phase chromatography 215, 216, 248 185, 195–199, 201, 260–265, 271–279, 288
Reynolds number 118 – applications in biopharmaceutical
RNA 3 processing 152
– cassettes for 261
S. cerevisiae 130, 137, 147, 158 – commercially available membranes 264
– cell structure for 147 – development of steps 279
– lysis of 137 – equipment setup for 195
– settling velocity 158 – hollow-fiber modules for 261
– structure of 130 – membrane integrity test 271
sanitization 35 – membrane modules for 261, 288
scale-down model for bioreactors 123 – membrane pore size 180, 195
scale-out of bioreactors 123 – membranes 260, 288
scale-up definition 49 – particle retention 185
seed train 27, 99 – permeate flux vs. TMP 274
setting velocity (centrifugal force) 159 – procedures for 270
settling velocity (gravity) – scale-up 280
– expression for gravity settling 156 – step yield calculation 271
– for CHO 158 – validation 284
– for E. coli 158 testing in biopharmaceutical processes 79
 Index 335

TFF. See tangential flow (TFF) under filtration – procedures 265

titer 40 – process configurations for 256, 258
training of personnel 67 – process development 279
tubing welder 111 – process parameters 271, 288
turbidity 189 – scale-up 280
– single-pass diafiltration (DF) 288
UHPLC. See ultra-high-performance liquid chro- – single-pass TFF 289
matography (UHPLC) under – transmembrane pressure (TMP)
chromatography definition 258
ultrafiltration (UF) 177, 196, 253–291 – validation 284, 289
– applications in biopharmaceutical upstream production campaign 113
processing 178, 254, 256 UV absorbance 235
– batch concentration setup 257
– cassettes for 261 vaccines 2, 11, 24
– commercially available membranes 264 – definition of 11
– concentration factor 275 – examples of 14
– diafiltration (DF) 254, 257, 275, 288 – types of 14
– extent of buffer exchange 275 validation documentation 80
– setup 257 validation. See process validation and cleaning
– diavolumes (DVs) 275 validation
– equipment for 257 viral clearance 29, 247
– hollow-fiber modules for 261 virus 209
– integrity testing of membranes 267 virus structure 14
– membrane modules for 261, 288
– membranes 195, 260, 288 waste production during cell growth 98
– molecular weight cutoff (MWCO) 260 water for injection (WFI) 19, 35
– normalized water permeability (NWP) 267 WFI. See water for injection (WFI)
– performance parameters 271, 288 working cell bank 98
– permeate flux vs TMP 274
– permeate flux 272 yield of cells on substrate 96
– polarization 272 yield of product on cells 96
– principles of 254 yield of product on substrate 96