HPLC Method Validations:

Navigating the Pitfalls

Joe Page, Ph. D., President

Eurofins Advantar, San Diego

www.AdvantarLabs.com
Method Validation and Pitfalls Outline

• Documents and Validation Parameters

• System Suitability
• Accuracy
• Sample and Standard Stability
• Precision
• Intermediate Precision
• Linearity
• Specificity
• LOQ/LOD
• Robustness
• Range

2
Method Validation Guidance Documents

ICH Harmonized Guideline:

▪ Validation of Analytical Procedures: Text and

Methodology Q2(R1)

U.S. Department of Health and Human Services FDA, CDER, CBER:

▪ Analytical Procedures and Methods

Validation for Drugs and Biologics

3
Method Validation Parameters

4
Internal Method Validation Documents

Prior to starting a method validation, you need two documents:

▪ QA approved validation protocol with

acceptance criteria

▪ A detailed written method with version

control
Lack of version control can create confusion and
potentially cause the validation to fail

5
System Suitability

System Suitability testing is an integral part of a GMP HPLC Method

Typical Data:
Standard injections (n=6), NMT 2% RSD.
%Recovery of Check Standard 98.0 to 102.0% (assay)
Resolution between two key peaks r ≥ 2.0
Tailing of main peak NMT 2.0

System suitability should be run at the start of every validation

sample set. It’s the only way to know that the system is suitable for
testing. Its a validation parameter that is sometimes overlooked, but
it can call your results into question (or failure) if it’s not performed.

6
Accuracy

Also known as “Spike and Recovery”. Your method should be

able to quantitatively recover a known amount of standard or API
spiked into your placebo
▪ Typical Assay Data: Spiking is typically performed at 80% (n=3),
100% (n=3) and 120% (n=3) of your drug’s label claim
▪ Typical Assay Acceptance Criteria: % Recovery is within 98.0% to
102.0% of the amount spiked into the placebo.

▪ Typical Impurity Data: Spiking is typically performed near the LOQ,

Specification, and 120% of the Specification
▪ Typical Impurity Acceptance Criteria: % Recovery is within 95.0% to
105.0% (or 90.0 to 110.0%) of the amount spiked into the placebo.
Ranges vary depending on the capability of the method and
toxicology results

7
 Chromatogram
Accuracy Pitfalls
SAMPLE I NFO RM ATI O N
Sample Name: Ref Std, Prep 1 Acquired By: pnguyen
Sample Type: Unknown Sample Set Name: CPR20518_050718_01_GMP37
Reasons for not Achieving 100% Recovery
Vial: 5 Acq. Method Set: APL_GTM11_LC44
Injection #: 1 Processing Method: APL_PM_PDN
Injection Volume: 8.00 ul Channel Name: A1100 VWD AU
Run Time: 37.0 Minutes Proc. Chnl. Descr.: VWD AU 220 nm

Method lacks
Date Acquired: specificity
5/7/2018 4:34:12 PM PDT and peaks are not fully resolved
Date Processed: 5/9/2018 7:38:18 AM PDT

Auto-Scaled Chromatogram
0.20

20.111
0.18

0.16

0.14
AU

21.391

21.742
20.884
0.12

0.10

0.08

0.06

18.00 19.00 20.00 21.00 22.00 23.00 24.00 25.00

Minutes
Auto-Scaled Chromatogram
0.15
APL-2 - 20.111

21.742
20.884
6 Imp - 21.391

0.10
AU

8
0.05
Accuracy Pitfalls

Method matrix interferes with recovery

50.00

45.00

40.00
Standard spiked without Sample Matrix
35.00

30.00

25.00
pA

20.00
Standard spiked in Sample Matrix
15.00

10.00
Sample Matrix component
5.00

0.00

16.80 17.00 17.20 17.40 17.60 17.80 18.00 18.20 18.40 18.60 18.80 19.00 19.20 19.40 19.60
Minutes

Date Acquired: 5/8/2018 4:48:04 AM PDT

Date Acquired: 5/9/2018 2:09:54 AM PDT

9
Accuracy Pitfalls

Reasons for not Achieving 100% Recovery (continued)

More on method matrix interference

▪ Method diluent is slowly degrading the analyte (TFA, formic acid)

▪ Method diluent does not fully solubilize the analyte
▪ Sample has limited stability in the method diluent

Diluent evaporation results in over recovery (methanol).

Not accounting for differences in response factors (impurities)

10
Accuracy Pitfalls

Reasons for not Achieving 100% Recovery (continued)

Method lacks solubility for the drug at the 120% spike.

Low level impurity spikes are adhering to the glassware.

Filtering of the sample results in a loss of analyte.

Spiking the standard into placebo is not equivalent to drug product

which may affect its solubility. This is especially true for:
▪ Homogenized suspension formulations
▪ Nanoparticle formulations

11
Sample and Standard Stability

The test samples and working standard solutions must be

demonstrated to be stable over a defined period of time that the
assay is run.

▪ Typical Data: Measure sample and standard at T=0, 1, 3, 5 days

▪ Typical Acceptance Criteria:
• Longest period in which the Assay sample and standard is within 98.0%
to 102.0% of the T=0 result.
• Longest period in which the Impurity sample and standard is within
95.0% to 105.0% (or 90.0 to 110.0%) of the T=0 result.

12
Sample and Standard Stability Pitfalls

Run this test early in your validation!

I’ve seen sections of validations that had to be repeated

because a parameter was determined with a 3-day old
sample, and later it was determined that the sample was
only stable for 1 day.

13
Precision (repeatability)

Precision of a method is the closeness of agreement between a

series of measurements obtained from multiple samplings of a
homogeneous sample

▪ Typical Assay Data: Precision is from the n=3 spiking replicates at

80%, 100% and 120% or n=6 at the 100% level
▪ Typical Assay Acceptance Criteria: RSD is NMT 2.0% at each level

▪ Typical Impurity Data: Precision is from the n=3 spiking replicates at

LOQ, specification and 120% of the specification
▪ Typical Impurity Acceptance Criteria: RSD is NMT 10% at LOQ and
NMT 5% to 10% at specification and 120% of the specification levels

14
Precision Pitfalls

Reasons for Failing Precision Criteria

Precision repeatability results are usually tied to the accuracy of

the method. Problems with accuracy are also manifested in the
precision results.
▪ For example, if your method poorly resolves the main analyte
this is likely to show up during precision too as it may be
difficult to reproducibly integrate the main analyte peak.

15
Intermediate Precision

Within the same lab, perform the accuracy experiment (assay at

80, 100, and 120% or impurities testing near LOQ, Spec, 120% of
Spec) with a different analyst, with different preparations, different
instruments, on different days.

▪ Typical Acceptance Criteria:

• Must meet accuracy criteria
• Each analyst must achieve precision criteria
• Determine the precision of all results according to analysts, instruments,
and days.
• Compare the precision of each analyst, each instrument, and days.

16
Intermediate Precision Pitfalls

Intermediate precision failures are often the hardest to resolve due

to the human component. Method directions may be vague or
interpreted differently between Analyst 1 and Analyst 2

▪ Example 1:
Analyst 1 aliquots the sample preparation into a glass HPLC vial
whereas Analyst 2 aliquots into a plastic HPLC vial. The analyte in
question has different compatibility with glass compared to plastic
which skews the results.

Often times the only way to resolve intermediate precision issues is to

closely observe both analysts in the lab.

17
Intermediate Precision Pitfalls individual chromatogram
SAMPLE I NFO RM ATI O N
Sample Name: 100% 1 Acquired By: mpham
▪ Example 2: Sample Type:
Vial:
Unknown
13
Sample Set Name: CPR17293_092117_01_MTP
Acq. Method Set: MRA_GTM0023_LC0004_MS
Injection #: 1
• Intermediate Precision Testing Processing
Injection Volume:
Run Time:
for Assay
10.00 ul using the same DP on a column
Method: CPR17293_PM02
5.5 Minutes
Channel Name: A1100 DAD AU Ch1
Proc. Chnl. Descr.: DAD AU Ch 1 Sample 260, Bw 10 ,
lots X Date
and Y. System
Acquired: suitability
9/22/2017 2:50:26 AM PDT passed on both
Reference 360, systems,
Bw 100 but the assay
Date Processed: 9/22/2017 11:20:59 AM PDT
failed every time column lot Y was used. individual chromatogram
Auto-Scaled Chromatogram
0.60
SAMPLE I NFO RM ATI O N

2.736
Sample
0.40 Name: Inter. Precision 100% 1 Acquired By: mpham
Sample Type: Unknown Sample Set Name: CPR17293_092117_02_MTP
AU

Vial: 5 Acq. Method Set: MRA_GTM0023_LC0044_MTP_MS

0.20
Injection #: 1
Injection Volume: 10.00 ul Processing Method: CPR17293_PM01
Run Time: 5.5 Minutes Channel Name: A1100 VWD AU
0.00
Proc. Chnl. Descr.: VWD AU 260 nm
0.00 Acquired:
Date 0.50 1.00
9/22/2017 1.50 2.00
2:02:13 AM PDT 2.50 3.00 3.50 4.00 4.50 5.00 5.50
Minutes
Date Processed: 9/22/2017 1:20:31 PM PDT
SampleName 100% 1; Injection 1; Date Acquired 9/22/2017 2:50:26 AM PDT; Result Id 7696
Auto-Scaled
Auto-ScaledChromatogram
Chromatogram
0.60

mRNA - 2.736
0.02
0.40
2.613
AU
AU

0.00
0.20

-0.02
0.00
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5.50
Minutes
Minutes
SampleName 100% 1; Injection 1; Date Acquired 9/22/2017 2:50:26 AM PDT; Result Id 7696
SampleName Inter. Precision 100% 1; Injection 1; Date Acquired 9/22/2017 2:02:13 AM PDT; Result Id 7748
Peak Results
Auto-Scaled Chromatogram
SampleName Name RT Area Int Type
A - 2.613

1 100% 1 mRNA 2.736 486350 BB

0.02
18
Intermediate Precision Pitfalls

▪ Example 2 (continued)

Investigations showed that all the solutions preparations would pass on

column X, but never with column lot Y. It was concluded that the method
was sensitive to changes in column lots and that a control drug product
standard would be needed to screen new column lots.

19
Specificity

The ability of the method to resolve the analytes of interest from

impurities, degradation products, excipients, and matrix
components.

▪ Typical Assay Data: Chromatographic resolution and peak purity of

the peaks of interest in aged and force degraded (heat, acid, base,
oxidation, light) samples.

▪ Typical Assay Acceptance Criteria:

• All peaks of interest are resolved with a resolution ≥ 1.0
• All peaks of interest are homogeneous as demonstrated by PDA or
LCMS peak purity analysis

20
Specificity Pitfalls

Some separations are difficult and you may not be able to resolve
all the impurities in a single method.

Example: A client in early phase developed an RP-HPLC

assay method with seemingly good resolution and a
homogeneous peak by PDA peak purity analysis. An
orthogonal strong cation exchange (SCX) method was
developed that showed the co-elution of related substances
that was missed by the PDA analysis (due to similar UV
spectra)

21
Specificity Pitfalls

Single RP-HPLC Peak, seemingly homogeneous by PDA Analysis

Multiple peaks under the main peak as resolved Ion Exchange-HPLC

Arrows are resolved impurities

22
Specificity Pitfalls

For compounds that have impurities with closely related

structures, the UV spectra are often the same which renders peak
purity analysis by PDA useless. In these cases, you must go to
LCMS for peak purity.

PEGylated and polymeric compounds have heterogeneous

structures so even peak purity by LCMS may be impossible.
PEG spectra

PEG+drug spectra

23
Linearity

The ability of the method to obtain results that are directly

proportional to the concentration of the sample

▪ Typical Data:
• Assay: 5 concentrations from 80% to 120% of label claim
• Impurities: 5 concentrations from LOQ to 150% of specification.

▪ Typical Acceptance Criteria:

Assay and Impurities: r2 > 0.98
Report the slope and intercept
Show plot of linear regression line

24
Linearity Pitfalls

An intercept value that is not close to zero is an indicator of bias in

the method. Some validations require the intercept to be NMT 2%
of the peak area of the 100% or assay sample.

Slope values in the impurity range may differ from slope values in
the assay range.

Non-linear detectors should not use linear regression, but should

use a polynomial model.

25
Limit of Quantitation

The lowest amount of a substance that can be measured with

accuracy and precision.

▪ Typical Data: Spiking into placebo is typically performed at 2 or 3

levels that approach a signal to noise ratio of 10:1.

▪ Typical Acceptance Criteria:

• Average S/N ≥ 10
• % Recovery is within 90.0% to 110.0% of the amount spiked into the
placebo. Ranges vary depending on the capability of the method and
toxicology results.
• The lowest concentration with a S/N ≥ 10 defines the LOQ. This can be
reported as a percentage of the label claim.

26
Limit of Quantitation Pitfalls

Reasons for not Achieving Targeted LOQ Level

Method noise is decreasing the sensitivity of the method:
▪ Different system/detector being used than in development.
▪ Sample preparation adding noise.

An aged column is causing band-broadening.

At low levels the analyte is sticking to glass or plastic.

27
Limit of Quantitation Pitfalls

Reasons for not Achieving Targeted LOQ Level (continued)

Method lacks specificity and peaks are not fully resolved.

Not accounting for differences in response factors (if standard is

different than analyte)

28
Limit of Detection

The lowest amount of a substance that can be detected but not

necessarily quantitated

▪ Typical Data: Spiking into placebo is typically performed at 2 or 3

levels that approach a signal to noise of 2:1 or 3:1.

▪ Typical Acceptance Criteria:

• Average S/N ≥ 2
• % Recovery is within 70.0% to 130.0% of the amount spiked into the
placebo.
• The lowest concentration with a S/N ≥ 2 defines the LOD. This can be
reported as a percentage of the label claim.

The same pitfalls as LOQ apply to LOD

29
Robustness

Robustness demonstrates that the method is reliable with respect

to deliberate variations in the method parameters
▪ Typical Data: Monitor retention time, peak area, resolution with
respect to changes in:
• Mobile phase pH (higher and lower pH)
• Mobile phase composition (+/- % organic phase)
• Different column lots
• Column temperature (+/-)
• Flow rate (+/-)
▪ Typical Acceptance Criteria:
• Retention time remains within X% or X min of the prescribed conditions.
• Peak area of select analytes remains within X% of the peak area under
the prescribed conditions.
• Resolution of key peak separations is unaffected

30
Robustness Pitfalls

When a Robustness variation fails, one can narrow the range of

the variation, or lock the parameter at a set value (e.g. a set pH, a
set % organic, a set flow rate, etc.)

Setting ranges too wide can “over-challenge” the method forcing

extra robustness testing, or forcing the method to be locked into
set value that may not be necessary.

31
Range

The Range of a method covers sample concentrations where the

method has been demonstrated to be linear, accurate and precise.

Typical Data: The range of an HPLC method is typically taken as a

composite of the linearity, accuracy and precision results.

32
Eurofins Advantar Laboratories, San Diego

Thank You!

33