9219_C007.

fm Page 157 Thursday, October 26, 2006 7:34 AM

7 Tumor-Targeting
Strategies
7.1 INTRODUCTION
One of the greatest challenges in the discovery and development of new therapeutic
agents and strategies for the treatment of cancer is the achievement of selectivity
between tumor cells and healthy tissues. With few exceptions, all agents in clinical
use today are associated with varying degrees of side effects (i.e., toxicities) due to
their collateral action on healthy cells and tissues. Therefore, many ingenious meth-
ods have been devised to target drugs to tumor cells or tumor masses.
One obvious way to achieve targeting is through antibodies directed towards
markers on the surface of tumor cells. This method has given rise to four strategies.
First, an antibody can be used as a single entity; this approach has been validated
through the development of such agents as trastuzumab (Herceptin™) and bevaci-
zumab (Avastin™). Second, a drug can be directly attached to a tumor-specific
antibody to guide it to tumor cells, an example of which is gemtuzumab ozogamicin
(Mylotarg™). A third antibody strategy involves the use of antibodies to guide an
attached radioactive element or ligand to the tumor. This approach was exemplified
by pemtumomab (Theragyn™), an antibody labeled with yttrium-90 (no longer in
development). A fourth strategy involves the use of antibodies to guide a pro-
drug/drug-releasing enzyme to tumor cells, such as in antibody-directed enzyme
prodrug therapy (ADEPT).
A different approach to achieving selectivity involves targeting tumor vascula-
ture. Anti-angiogenic agents work by blocking the ability of tumors to develop new
vasculature vital for their growth and survival; healthy tissues are relatively unaf-
fected by these drugs because their vasculature is already in place. This approach
has been recently validated by the success of bevacizumab (Avastin™) in the clinic.
An alternative strategy involves targeting the existing blood supply of tumors, which
can be functionally distinct from the vasculature of healthy tissue. Drugs of this
class are known as vascular disrupting agents (VDAs). In particular, because tumor
vasculature is usually less mature than that of other tissues and organs, the cells are
more prone to drugs such as combretastatin due to their lack of actin and relatively
exposed tubulin.
Other triggers, such as oxygen concentration, are also being used to selectively
release drugs at tumor sites. For example, it is well established that the centers of
most tumor masses are hypoxic compared to healthy tissue due to an abnormal
vasculature. Therefore, prodrugs that are activated under these hypoxic (i.e.,
bioreductive) conditions have been designed. This approach is exemplified by AQ4N,

157

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 158 Thursday, October 26, 2006 7:34 AM

158 Chemistry and Pharmacology of Anticancer Drugs

a prodrug that produces a DNA–interactive agent (AQ4) upon bioreduction. In

addition, it has been observed that polymers such as polyethylene glycol (PEG) tend
to accumulate in tumor masses due to their abnormal vasculature, and this
phenomenon has been exploited in the form of experimental polymer-drug
conjugates that accumulate at the tumor site and then release active drug. More
elaborate two component systems based on this phenomenon involving both
polymer-bound drugs and polymer-bound releasing enzymes (i.e., polymer-directed
enzyme prodrug therapy [PDEPT]) have also been investigated but not yet fully
developed.
Enzymes have also been utilized in various tumor-targeting strategies. For exam-
ple, the experimental biphasic X-DEPT therapies involve a two-component system
whereby an enzyme is localized at the tumor site via an antibody, viral, or polymer
vector. Once there, the enzyme can selectively activate a prodrug administered
separately. The X-DEPT approaches are still being evaluated in the clinic, and none
has been successfully commercialized to date. An alternative enzyme-based strategy,
which is also still being evaluated in the clinic, relies on the use of an enzyme
constitutively expressed by tumor cells (but not healthy cells) to activate a prodrug
either without further intervention or after systemic administration of an enzyme
cofactor. Yet another approach involves the use of the enzyme asparaginase (Erwi-
nase™) to break down asparagine in the bloodstream and tissues. The targeting is
based on the fact that asparagine is required for survival by leukemic cells which,
unlike healthy cells, are unable to synthesize their own supply. This treatment has
been successfully used to manage certain types of leukemia for many years.
The concept of activating systemically administered prodrugs by a light source
selectively at the tumor site has been established for some time and is based on a
similar approach used for the treatment of psoriasis (i.e., psoralen ultraviolet A
[PUVA] therapy). Although originally restricted to phototreatment of the skin (e.g.,
melanomas), this area has developed rapidly in recent years due to advances in laser
technology and the introduction of “key hole” surgery techniques that allow flexible
laser light sources to be introduced into body cavities such as the gastrointestinal
(GI) tract and trachea. A related strategy, boron neutron capture therapy (BNCT),
uses a neutron beam to activate a prodrug specifically at the tumor site (usually in
the brain).
Finally, in the last decade, significant advances have been made in drug delivery
techniques, novel approaches, such as those involving gene therapy, nanotechnology,
intracranial wafers, and ultrasound, are being used experimentally to release estab-
lished anticancer drugs specifically at the site of tumors without harming healthy
tissues. All of these approaches are described below.

7.2 ANTIBODY-BASED APPROACHES

Monoclonal antibodies (MAbs) have been developed for both the diagnosis and
treatment of cancer, and several MAbs are already commercially available as cancer
therapies. MAbs can be used as single agents, paired with powerful cytotoxics or
radiopharmaceuticals to create tumor-specific agents, or they can be used in an
X-DEPT approach such as ADEPT. Therapies based on MAbs have been slow to

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 159 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 159

emerge, partly due to the technical challenges involved but also because early studies
with vinblastine-antibody conjugates were disappointing. However, the clinical
potential of MAbs has been recently validated by the success of such agents as
trastuzumab (Herceptin™) and bevacizumab (Avastin™). Even so, in practice, these
MAbs tend to be most efficacious when used in combination with traditional cyto-
toxic agents. Also, for solid tumors, debulking may be initially required.
Rather than use MAbs alone, an alternative approach to enhance efficacy is to
attach a cytotoxic agent through a chemical linker. The linker can be designed to
cleave specifically at the tumor site, thus releasing the cytotoxic agent. For example,
conjugates have been reported that contain linkers designed to cleave on exposure
to the enzyme cathepsin, which is overexpressed in some tumor cell types. With this
type of construct, in which exposure to the cytoplasm within the cell is important
for drug release, it is necessary to demonstrate that, once bound to the tumor cell,
the drug-antibody conjugate is internalized. If internalization does not occur, then
the full cytotoxic effect may not be achieved, although some drug may still be
released in the vicinity of the tumor. If some release occurs externally to the cancer
cell, then the drug is free to diffuse into neighboring cells, a phenomenon known
as the bystander effect. A further consideration is that, for maximum efficacy, as
many linker-drug units as possible should be attached to a single antibody without
affecting its antigen-recognition or pharmaceutical properties, a technically chal-
lenging feat. For example, in the case of gemtuzumab ozogamicin (Mylotarg™), a
loading of calicheamycin of 4 to 6 moles per mole of antibody over 50% of the
antibody’s surface is achieved. Another approach is to ligate the MAb to a radiop-
harmaceutical to create an agent for use in radioimmunotherapy. This combines the
advantages of targeted radiation therapy and specific immunotherapy.

7.2.1 ANTIBODIES AS SINGLE AGENTS

Antibodies used as single agents to target markers on tumor cells are discussed in
detail in other sections of this book. For example, trastuzumab (Herceptin™) is a
humanized recombinant anti-P185 monoclonal antibody targeted to human epider-
mal growth factor receptor 2 (HER2)/neu receptors (see Chapter 5). It inhibits the
HER2/neu signaling pathway and was developed for use in breast cancer. Cetuximab
(Erbitux™), a chimerized monoclonal antibody that is part mouse and part human,
targets and inhibits epidermal growth factor receptor (EGFR), and has proven useful
in EGFR-positive colorectal cancer (see Chapter 5). Similarly, bevacizumab (Avas-
tin™), which targets the vascular endothelial growth factor receptor (VEGFR), was
the first angiogenesis inhibitor to reach the market (in 2004) for first-line use together
with 5-fluorouracil (5-FU) for patients with metastatic colorectal cancer (see
Section 7.3.1.1).
Many other types of antibody-based single-agent therapies are under develop-
ment. For example, IGN311 (Aphton) is a humanized monoclonal antibody currently
in clinical evaluation that targets the tumor-associated antigen of Lewis Y, a marker
expressed in as many as 90% of all epithelial cancers, including pancreatic, gastric,
colon, and breast tumors. Data from Phase I studies suggest that this antibody
significantly decreases the number of tumor cells circulating in peripheral blood.

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 160 Thursday, October 26, 2006 7:34 AM

160 Chemistry and Pharmacology of Anticancer Drugs

Furthermore, IGN311 has favorable safety and tolerability profiles and a serum half-
life of more than 20 days. Similarly, a CD4-targeted antibody called HuMax-CD4
(Genmab AS/Serono SA) is being evaluated for the treatment of T-cell lymphomas.
This antibody has been granted fast-track status by the U.S. FDA and is currently
being evaluated in a Phase III clinical trial.

7.2.2 ANTIBODY-DRUG CONJUGATES

Historically, many attempts have been made to conjugate MAbs to cytotoxic “war-
heads” in order to achieve greater tumor selectivity. The first studies in this area
involved conjugation of vinca cytotoxic agents to MAbs; however, the results were
disappointing, with no significant improvement in therapeutic index. Only recently
have more encouraging results been obtained with the experimental agent gemtu-
zumab ozogamicin (Mylotarg™), which was approved by the FDA in 2000.

7.2.2.1 Gemtuzumab Ozogamicin

Gemtuzumab ozogamicin (Mylotarg™), originally developed by American Home

Products, was the first licensed chemotherapeutic agent to use MAb technology
targeted directly at tumor cells (Structure 7.1). Gemtuzumab was given accelerated
approval by the U.S. FDA for use in CD33-positive acute myeloid leukemia (AML)
in patients age 60 years or older in first relapse who are not considered suitable for
cytotoxic chemotherapy. This agent, which is comprised of the cytotoxic antitumor
antibiotic calicheamycin conjugated to a humanized recombinant IgG4 kappa anti-
body, has a molecular weight of 151 to 153 kDa. Calicheamycin is a natural product
isolated from the bacterium Micromonospora echinospora sp. calichensis discovered
in caliche clay (a type of soil found in Texas) by Wyeth-Ayerst researchers. The

hP67.6

HN

O O
O H3C CH3
O
HO
CH3 O H3C NNH S H
S N
H3C O
l O
S O H3C OCH3
N O
H O H
O OCH3 OH OH
O
OCH3 H3CH2C
H3C O
HO H3C N O
H3CO OH H3CO
O

STRUCTURE 7.1 Structure of gemtuzumab ozogamicin (Mylotarg™).The antibody is

depicted as a circle (top left) labeled as “hP67.6.”

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 161 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 161

CD33 antigen to which gemtuzumab binds is a sialic acid–dependent adhesion

protein. It is an attractive target for antibody therapy because it appears on the surface
of immature normal cells of myelomonocytic lineage and the surface of leukemic
myeloblasts but not on normal hematopoietic stem cells.
Approximately 98.3% of the amino acid sequence of the anti–CD33 antibody
fragment of gemtuzumab is human in origin, including the framework and constant
regions. However, the CD33-recognizing complementarity-determining regions
come from a murine p67.6 antibody. The antibody component is made in a mam-
malian myeloma NS0 cell suspension culture and then purified under conditions that
either remove or inactivate viruses. The early development work on the antibody
component was carried out by researchers at the Fred Hutchinson Cancer Research
Center (Seattle, WA), who licensed it to Wyeth-Ayerst. It was subsequently human-
ized by the Celltech Group, and a bifunctional linker was used to join it to N-acetyl-
gamma calicheamycin, with a loading of 4 to 6 moles of calicheamycin per mole
of antibody over 50% of its surface.
Gemtuzumab works by targeting and binding to the CD33 antigen, which is
expressed on the surface of hematopoietic cells but not on the surfaces of pluripotent
hematopoietic stem cells or nonhematopoietic cells. Thus, CD33 is expressed on
normal and leukemic myeloid colony-forming cells, including leukemic clonogenic
precursors. For example, more than 80% of patients with AML express CD33 on
their leukemic blasts. When gemtuzumab binds to the CD33 antigen, the complex
formed is subsequently internalized, which is followed by release of calicheamycin
inside the lysosomes of the cells. After reaching the nucleus, calicheamycin binds
in the minor groove of DNA, causing DNA double-strand breaks that lead to cell
death. (See Chapter 3 for more details.)
The most serious side effect of gemtuzumab is substantial myelosuppression
through a cytotoxic effect on normal myeloid precursors. However, because pluri-
potent hematopoietic stem cells are spared, this toxicity is reversible. Other side
effects reported include allergic reactions, nausea, vomiting, flu-like symptoms, high
or low blood pressure, sore mouth, taste changes, lowered resistance to infection,
bruising or bleeding, anemia, and liver toxicity.

7.2.3 ANTIBODY-RADIONUCLIDE CONJUGATES

Rather than conjugating an antibody to a cytotoxic agent, an alternative strategy is
to attach a radionuclide as the “warhead.” One of the best-known examples of this
approach is pemtumomab (Theragyn™), which gave encouraging results in early
clinical trials but was not developed any further due to disappointing results in
Phase III evaluations.

7.2.3.1 Pemtumomab

Pemtumomab (Theragyn™) is an yttrium-90 labeled antibody developed by Anti-

Soma Ltd., targeted to the aberrantly glycosylated polymorphic epithelial mucin
(PEM) in ovarian cancer. In this tumor type, a number of antigenic targets have been

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 162 Thursday, October 26, 2006 7:34 AM

162 Chemistry and Pharmacology of Anticancer Drugs

exploited, including the folate receptor CA-125 and PEM. MAbs to PEM were
developed in the early 1980s; one of the best known is HMFG-1, the antibody that
forms the basis of pemtumomab.
Research into the tumor-associated mucin molecule led to the discovery of a
gene on chromosome 1 of the human genome, coding for the core protein of PEM.
This gene, called the MUC-1 gene, codes for the MUC-1 antigen (a 20 amino-acid
antigen) repeated in tandem, which forms the core protein of PEM. In normal
epithelial tissue, the MUC-1 gene product is fully glycosylated (covered in carbo-
hydrates) and is therefore “hidden.” However, in many tumors of epithelial origin,
including ovarian cancer, the glycosylation is incomplete, which results in exposure
of amino acid sequences of the MUC-1 gene product, which was exploited by
pemtumomab.
During development, the fact that ovarian cancer is normally confined to the
peritoneal cavity was also exploited, and so the radiolabeled antibody was admin-
istered intraperitoneally. This strategy, known as regional immunotherapy, results
in increased localization of the antibody compared to that achievable when given
systemically, thus increasing the radiation dose to the tumor. Initially, iodine-131
was used as the isotope because it had been successfully used itself in the treatment
of other tumor types (e.g., thyroid cancer). However, the major disadvantage of
iodine-131 was that, being a γ-emitter, patients had to be isolated and kept in the
hospital for at least 5 days. Exposure of hospital staff to the radiation was also a
potential hazard. To overcome these disadvantages, the pure β-emitting isotope
yttrium-90 was used instead. This isotope had already been used on its own to treat
liver cancer (hepatoma) and raised none of the same concerns regarding exposure,
isolation, or long hospitalization. However, in order to use high amounts of this
radioactive metal, new chemical chelates had to be developed that could bind
yttrium to the antibody in an almost nonreversible manner.
The initial clinical trials of the yttrium-90 labeled MAb (pemtumomab) gave
promising results. Patients underwent surgery followed by chemotherapy and then
waited for some time before receiving pemtumomab. Those who received the agent
4 to 6 weeks after their chemotherapy, and at a time when they had no evidence of
disease, survived longer than patients who only received surgery and chemotherapy.
Five-year survival data on patients in a Phase II trial surpassed any previously
reported therapy for the treatment of ovarian cancer, which led to a multinational,
multicentered Phase III clinical trial to evaluate pemtumomab in a much wider setting
using a greater number of patients. Unfortunately, the results of this trial, published
in 2004, suggested that the benefits to patients were not significant enough to warrant
continued development.

7.3 VASCULAR-TARGETING STRATEGIES

The concept that solid tumors require a functioning network of blood vessels to
sustain growth through the supply of oxygen and nutrients and to avoid a buildup
of toxic by-products of cellular metabolism has been under consideration for more
than 50 years. The impetus to target tumor angiogenesis (the formation of new blood
vessels from the endothelium of existing vasculature) arose from key observations

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 163 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 163

by Folkman in the early 1970s. It is now recognized that for any tumor to grow
beyond a volume of 1 to 2 mm3, a so-called “angiogenic switch” must be present,
prompting the formation of new vasculature (i.e., neovascularization) (Figure 7.1
and Scheme 7.1). Since Folkman’s original observations, key molecules in the
angiogenesis process have gradually been identified, such as VEGF and its receptors,
culminating in the recent clinical proof of the concept of targeting VEGF in colorectal
cancer with the humanized MAb bevacizumab (Avastin™). Many small-molecule
inhibitors of VEGF receptors are also now in clinical development (e.g., SU11248
and PTK787/ZK22854).
An alternative but complementary strategy to anti-angiogenic agents is that of
targeting established tumor vasculature (Scheme 7.1). This approach was first sug-
gested as long ago as the 1920s, but research in this area was given impetus by
Denekamp in the early 1980s and has led to a class of drugs now known as Vascular
Disruptive Agents (VDAs), whose role is to cause a rapid and selective shutdown
of existing tumor blood vessels. The first known agent of this type was combretastatin
(and its derivatives), which was followed by related prodrug forms, such as CA4P
and Oxi4503. 5,6-Dimethylxanthenone-4-acetic acid (DMXAA), an agent with a
similar effect on existing tumor vasculature, is presently in late-stage clinical trials.
One attraction for clinicians is that agents such as combretastatin and DMXAA
cause a rapid and selective vascular shutdown in tumors in a matter of minutes to
hours, and so can be given in intermittent doses to potentiate the activity of conven-
tional chemotherapeutic agents (see below). Many other potential drugs of this type
are at the discovery and development stage; examples include ZD6126 and
AVE8062A.

FIGURE 7.1 Photomicrograph showing the process of angiogenesis. The image shows the
formation of numerous new blood vessels in the area of a metastasis on the surface of a
human lung.

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 164 Thursday, October 26, 2006 7:34 AM

164 Chemistry and Pharmacology of Anticancer Drugs

Anti-Angiogenic Agents Vascular Disrupting Agents (VDAs)

CA4P

DMXAA
Bavacizumab

AVE8062A
ZD6126
SUII248 PTK787/ZK22854
Small solid tumors with new blood Larger solid tumors with
vessels. Anti-angiogenic agents established blood vessels.
have major effect on tumor VDAs have a major effect on the
periphery inhibiting endothelial central part of the tumor causing
proliferation and migration. vessel occlusion and necrosis.

SCHEME 7.1 Mechanism of action of anti-angiogenic agents and VDAs. (Adapted with
permission from Kelland, L.R., Curr. Cancer Ther. Rev., 1:1-9, 2005).

7.3.1 ANTI-ANGIOGENIC AGENTS

The concept of blocking the growth of new tumor vasculature was first described
in the early 1970s, although a practical application has only just emerged, more
than 30 years later. Early enthusiasm for angiogenesis inhibitors was lost after a
number of promising agents were unsuccessful in increasing survival in pivotal
Phase III clinical trials. Only more recently have the colorectal cancer trials of
bevacizumab established a survival benefit through this mechanism of action. This
has not only demonstrated proof of concept, but has also provided more treatment
options for colorectal cancer patients with metastatic disease. As a result, many
more anti-angiogenic agents, both antibody-based and small molecules, are now in
development.

7.3.1.1 Bevacizumab

Bevacizumab (Avastin™) was hailed as a major advance in cancer chemotherapy

when it became the first angiogenesis inhibitor to gain U.S. FDA approval in early
2004 for first-line use, in combination with 5-FU, for the treatment of metastatic
colorectal cancer. Developed jointly by Roche and Genentech, it is a humanized
MAb directed against VEGF, a growth factor important in both tumor angiogenesis
and in the maintenance of existing tumor blood vessels. By binding to VEGF,
bevacizumab blocks its interaction with receptors such as Flt-1 and KDR on the
surface of endothelial cells, a process that normally triggers the proliferation of
endothelial cells and the formation of new blood vessels. This interferes with the
blood supply to tumors, hence reducing growth and potentially inhibiting the process
of metastasis. Human tumor xenograft studies in a colon model initially confirmed
that bevacizumab was able to reduce microvascular growth at the tumor site and
also inhibit metastasis. A large, randomized, placebo-controlled Phase III clinical
trial involving colorectal cancer patients then demonstrated a clear extension of
median survival of approximately 5 months when bevacizumab was given together

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 165 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 165

with an IFL regimen (5-FU/CPT-11/leucovorin) compared to the IFL alone (20.3

versus 15.6 months). This was one of the greatest improvements in survival ever
reported for colorectal cancer patients in a randomized Phase III trial, and led to
FDA approval in early 2004 for first-line use of an intravenous bevacizumab/5-
FU–based combination for patients with metastatic colon or rectal cancer. European
approval was obtained in early 2005 for a similar indication.
Bevacizumab is normally administered as an intravenous infusion once every
14 days, at a dose level of 5 mg/kg. Administration is halted when disease progression
occurs. As anticipated for an agent that inhibits the actions of VEGF, wound healing
complications and GI perforations are the most serious side effects and, in some
cases, can prove fatal. Other potentially serious adverse effects may include con-
gestive heart failure, nephrotic syndrome, hypertensive crises, and hemorrhage.
Hemoptysis has also been observed in non-small-cell lung cancer (NSCLC) patients
treated with both bevacizumab and other types of chemotherapy. Less serious adverse
effects include proteinuria; exfoliative dermatitis; anorexia; asthenia; pain, including
abdominal pain; stomatitis; headache; hypertension; nausea; vomiting; diarrhea;
constipation; upper respiratory infection; leukopenia; epistaxis; and dyspnea.
Bevacizumab is also being evaluated for use in other types of solid tumors,
including renal cell, breast, non-small-cell lung, prostate, and ovarian cancers; mel-
anoma; and some hematologic malignancies. Based on the proof-of-concept that
bevacizumab has provided for the anti-VEGF approach to cancer therapy, approxi-
mately 50 novel anti-angiogenesis inhibitors are presently in various stages of
development worldwide, demonstrating the high level of interest in this new area.

7.3.2 VASCULAR DISRUPTIVE AGENTS (VDAS)

Another group of drugs that is currently of significant interest is the VDAs, not only
because of their potential as single agents but also because they can be used for so-
called adjunct or “complementary” therapy. This involves using a VDA, such as
DMXAA, to target the central regions of a tumor via its blood supply in combination
with cytotoxic agents or radiotherapy which preferentially affect the well-oxygenated
outer rim of the tumor. In addition, there is the possibility of pharmacokinetically
“trapping” cytotoxic drugs in the tumor by scheduling a VDA to close the tumor
blood vessels after cytotoxic agents have been administered. Bioreductive agents
(see Section 7.3.3) should be particularly useful in this context because they them-
selves target the centers of larger tumors, and so a synergistic effect may be obtained.
Combretastatin and its derivatives were the first VDAs to be identified, followed by
the prodrugs Oxi4503 and CA4P. Flavone-8-acetic acid (FAA) and the related
DMXAA were discovered later, and DMXAA is presently in late-stage clinical
development.

7.3.2.1 Combretastatins

The combretastatins are a group of cis-stilbene compounds (Structure 7.2) isolated

from the bark and stem wood of the African bush willow tree (Combretum caffrum)
(Figure 7.2) that inhibit tubulin polymerization. They were identified in the early

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 166 Thursday, October 26, 2006 7:34 AM

166 Chemistry and Pharmacology of Anticancer Drugs

1980s by George R. Pettit at Arizona State University (U.S.A.) based on the

observation that extracts of the tree had been used in folk medicine and shown to
possess anti-leukemic properties. The highly attractive African bush willow only
grows on the banks of rivers in the Eastern Cape Province of South Africa, and
historians believe that Arabians traded for the bark with the San people (Bushmen)
for more than 2000 years. At the time, the bark was probably used as a general tonic
because, apart from its anticancer properties, it allegedly produces a feeling of
improved general well-being. However, it is also known that Zulu warriors used this
shrub to prepare a substance that they used both as a poison for their arrow tips and
also as a “charm” to protect themselves against enemies. Structurally, the
combretastatin molecules are biaryls connected in the cis(z) configuration by an
ethylene bridge, and structure activity relationship (SAR) studies have shown that
the Z-configuration is essential for their antitumor activity. Furthermore, restricted

(a) (b)

FIGURE 7.2 (a) African bush willow tree; (b) mohair tapestry produced in the Zulu village
of Howick in Kwa-Zulu (Natal, South Africa) showing the gently sweeping elegance of the
bush willow along the Mooi River in Northern Natal.

H3CO H3CO O−Na+

H3CO O
P
R − +
H3CO H3CO
H3CO O O Na
O
OCH3 OCH3 O
OH O P O−Na+ OCH3 O P O−Na+
OCH3 OCH3 O−Na+
OCH3 O−Na+
Combretastatin A-1 (Oxi4500): R = OH CA4P Oxi4503
Combretastatin A-4: R = H

STRUCTURE 7.2 Structures of the combretastatins A-1 and A-4, and related prodrugs CA4P
and Oxi4503.

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 167 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 167

rotation about the olefinic bridge is crucial for biological activity, as is the distance
between the two rings.
The interesting feature of the combretastatins is their selectivity for tumor
vasculature rather than healthy tissue. Combretastatin A-1(CA-1) is known to bind
to tubulin with a higher affinity for β-tubulin than colchicine at or near colchicine-
binding sites, thus destabilizing the tubulin cytoskeleton. When this tubulin structure
is disrupted, the endothelial cells change from a flat streamlined profile to a round
swollen shape because the internal tubulin skeleton is no longer able to mechanically
support their elongated shape. Due to this shape change, the cells effectively plug
the capillaries, thus restricting tumor blood flow. In addition, a process of apoptosis
is initiated in the cells of tumor blood vessels. This appears to first kill cancer cells
in the core of the tumor but with the apoptotic effect then radiating out, thus starving
the tumor of nutrients which, in turn, leads to exposure of the basement membrane,
hemorrhage, and coagulation (i.e., tumor necrosis). The selectivity arises because,
compared to mature endothelial cells that line blood vessels in normal tissues, tumor
blood vessels are usually newly formed and immature, and only vessels of this type
can be affected by combretastatin. Actin, a protein not present in immature endo-
thelial cells, protects the tubulin in mature endothelial cells from the effects of
combretastatin, and hence their shape is maintained. Interestingly, actin does not
appear for a number of days after the formation of new endothelial cells, and so this
provides a window of opportunity to achieve selectivity toxicity toward the tumor
microvasculature.
In addition to these fundamental differences between immature and mature
endothelial cells, characteristics of the tumor microcirculation, such as high inter-
stitial fluid pressure, procoagulant status, vessel tortuosity, and heterogeneous blood
flow distribution, assist the selectivity process. Reduced blood flow results in sig-
nificant hypoxia within the tumor, leading to cell death and regression. The study
of these agents in the clinic has been facilitated by the development of techniques
using positron emission tomography (PET) and magnetic resonance imaging (MRI)
that allow tumor blood flow to be measured in patients during treatment.
It is noteworthy that neither of the two side effects commonly associated with
cytotoxic agents (i.e., leukopenia and alopecia) occur during or after treatment with
combretastatin. However, one of the most serious side effects is tumor pain and, at
high doses (more than 90 mg/m2), some patients show signs of lung and heart
toxicity. Other more minor adverse effects observed in clinical trials include faint
flush, transient blood pressure changes, abdominal pain, nausea, vomiting, fever,
light-headedness, headache, diarrhea, bradycardia, and tachycardia. Finally, on an
electrocardiogram, nonspecific ST-T wave changes have been observed that peak
between 3 and 5 hours after infusion.
Despite extensive clinical evaluation of the original combretastatin compounds,
efficacy proved disappointing. The poor aqueous solubility of these molecules was
thought to be partly responsible for their poor performance in the clinic, and so a
number of phosphate prodrugs have prepared, such as Oxi4503 and CA4P (see
Section 7.3.2.2). These prodrugs are presently being evaluated in the clinic. Other
types of tubulin depolymerizing agents are also in clinical development (e.g., ZD6126
and AVE8062), along with DMXAA, which has a different mechanism of action.

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 168 Thursday, October 26, 2006 7:34 AM

168 Chemistry and Pharmacology of Anticancer Drugs

7.3.2.2 Combretastatin Prodrugs: CA4P and Oxi4503

CA4P and Oxi4503 are monophosphate and diphosphate prodrugs of combretastatin

A-4 and A-1, respectively. Oxi4503 is 10 times more potent than the monophosphate
CA4P and has superior in vivo activity to A-4. Also, it is active in various cell lines
in vitro, including those expressing the multidrug resistant (MDR) phenotype.
Oxi4503, which is presently in clinical trials, was designed to improve the pharma-
ceutical properties of the parent molecule, particularly by enhancing its water sol-
ubility. After injection, it hydrolyzes to the parent A-1 by enzymes in the blood-
stream, which then rapidly penetrates the endothelial cells that line the tumor blood
vessels to produce an antitumor effect.

7.3.2.3 Flavone-8-Acetic Acid (FAA) and DMXAA

Interest in this nontubulin class of VDAs began in the mid-1980s, with the seren-
dipitous discovery of the antitumor properties of flavone-8-acetic acid (FAA), a
molecule originally synthesized as a nonsteroidal anti-inflammatory agent by Lyon-
naise Industrielle Pharmaceutique (Structure 7.3). FAA was found to be active (even
curative) against a wide variety of preclinical solid tumor models, and its mechanism
of action appeared to be related to the production of tumor necrosis factor (TNF).
Initial Phase I studies using an ester prodrug (LM985) of FAA showed that the dose-
limiting toxicity was acute, reversible hypotension occurring during drug infusion
at doses of 1500 mg/m2. Later Phase I studies using FAA itself (LM975) showed
that hypotension was again dose-limiting but at the significantly higher dose of 10
g/m2. However, no evidence of efficacy was apparent from any of these trials,
including a Phase II study in NSCLC. A Phase I trial of FAA in combination with
recombinant interleukin-2 gave one complete and two partial responses, but severe
hypotension (grade 3 or 4) was observed after the third dose of FAA, with no sign
of increased tumor necrosis in several tumor biopsies. This led to the clinical
abandonment of FAA on the grounds that it did not elicit similar antivascular effects
in humans as those seen in many murine tumor models.
A synthetic program to identify a more potent inducer of TNF-α was initiated
by Denny, Baguley, and colleagues at the University of Auckland, New Zealand,

O
O

O
H3C O
CH2COOH
CH3 CH2COOH
Flavone-8-acetic acid (FAA) DMXAA

STRUCTURE 7.3 Structures of flavone-8-acetic acid (FAA) and DMXAA.

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 169 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 169

which resulted in DMXAA (AS1404, Antisoma) (see Structure 7.3). Encouragingly,

DMXAA was shown to induce TNF-α (mRNA) in both murine and human cells
(whereas FAA only induced TNF in murine cells). However, the mechanism of
action of DMXAA appears to be more complex than simply induction of TNF-α,
and data support an early direct effect on endothelial cells (possibly through involve-
ment of nuclear factor κB, although the exact target remains to be identified), leading
to rapid apoptosis. This is followed by indirect effects involving the release of
vasoactive agents within tumor tissue, such as serotonin (from platelets) and TNF-α,
possibly other cytokines and, even later, nitric oxide. A comparison of blood perfu-
sion effects in a murine tumor (C3H) for DMXAA and CA4P showed that blood
flow reduction with CA4P was more rapid (maximum effect only 1 hour after
administration for CA4P versus 6 hours for DMXAA) but more transient, as perfu-
sion had returned to normal 24 hours later with CA4P but not with DMXAA.
DMXAA showed single-agent antitumor activity, including cures in some mod-
els such as the syngeneic colon 38, especially when administered as a loading dose
of 25 mg/kg followed by two supplementary maintenance doses of 5 mg/kg 4 and
8 hours later. As with CA4P, DMXAA showed at least additive, even synergistic
antitumor effects in combination with a variety of treatment modalities, including
radiation, radioimmunotherapy, and hyperthermia. In addition, marked potentiation
was observed with a variety of chemotherapeutic drugs in a syngeneic mouse
mammary tumor model. The therapeutic gain was most striking with paclitaxel but
was also apparent with vincristine, etoposide, carboplatin, cyclophosphamide, dox-
orubicin, and cisplatin. In mice, the therapeutic index for DMXAA is rather narrow;
with vascular targeting/tumor necrosis effects typically being observed at doses
greater than 15 mg/kg, relatively close to the maximum tolerated dose of approxi-
mately 25 mg/kg. However, fortunately, there appears to be a wider therapeutic index
in man.
DMXAA (AS1404) entered clinical trials in 1996 both in the U.K. (weekly
schedule) and in New Zealand (3-weekly schedule), using a 20-minute intravenous
infusion. In the U.K. study, 46 patients received DMXAA over a dose range of 6
to 4900 mg/m2 and, unlike with conventional chemotherapeutic agents, dose-limiting
toxicities included urinary incontinence, visual disturbance (as was reported for
FAA), and anxiety. The New Zealand study reported similar findings, with the rapidly
reversible toxicities of confusion, tremor, slurred speech, visual disturbance, anxiety,
urinary incontinence, and possibly left ventricular failure (seen at the 4900 mg/m2
dose level). A transient prolongation of cardiac QTc interval was also observed in
13 patients administered doses of 2000 mg/m2 and greater. Dose-dependent plasma
increases in the serotonin metabolite 5-hydroxyindoleacetic acid were seen at doses
above 650 mg/m2, thus providing a possible pharmacodynamic marker of vascular
targeting for this agent. A number of patients in these trials had unconfirmed partial
responses.
Based on the findings that DMXAA is well tolerated using both weekly and
3-weekly schedules of administration, and that adverse affects appear manageable,
a number of combination trials were planned, particularly involving coadministration
with taxanes. These trials are still underway.

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 170 Thursday, October 26, 2006 7:34 AM

170 Chemistry and Pharmacology of Anticancer Drugs

7.3.3 BIOREDUCTIVE AGENTS

The importance of oxygen in cancer therapy (particularly radiotherapy) has been
understood for many years. Low oxygen levels (hypoxia) make cells resistant to
being killed by radiation and chemotherapy, thus reducing the efficacy of treatment.
Furthermore, oxygen levels are lower in solid tumors (especially their centers) than
in healthy tissues, and tumors over a certain size (i.e., greater than 2 mm3) suffer
from disordered vasculature and contain hypoxic fractions (Scheme 7.2). In vitro
data indicate that, compared to normally oxygenated (i.e., “oxic”) cells, when
hypoxic fractions are present in a tumor, a much lower response (approximately
one third the normal) to radiotherapy and conventional chemotherapeutic agents is
obtained. One explanation for this is that, compared to their oxic counterparts,
hypoxic cells are quiescent (i.e., not replicating) and are thus less vulnerable to
antiproliferative agents such as DNA-damaging drugs and antimetabolites. There-
fore, if drugs could be designed to be optimally effective at low oxygen levels, then
they would be useful for killing tumor cells in the centers of larger tumors and those
that remain after radiotherapy. They may also be less toxic towards “oxic” normal
cells, thus providing a greater therapeutic index (TI).

Blood capillaries
Invasive zone

Zone of cell
stasis low in oxygen

Necrotic center
Proliferating zone with well-
of dead or
developed capillary network
dying cells
and high oxygen level

SCHEME 7.2 Section through a solid tumor showing the necrotic center and the development
of a network of small blood vessels.

One approach to the design of hypoxia-targeted agents is to incorporate a “biore-

ductive trigger” as part of a prodrug so that a linker can be cleaved and an active
drug is released only under hypoxic conditions (see Scheme 7.3). Such bioreductive
prodrugs can be made to work preferentially at low oxygen levels if an intermediate

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 171 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 171

Pro-drug
Cellular enzyme
Prodrug
Low oxygen intermediate High oxygen Trigger Effector
(tumors) (normal tissues)
Activated Restored Superoxide
drug prodrug radical Link broken
This part on activation Effective
May be toxic Potentially
reduced (reduction) drug
directly or release damaging but cell
by cells
an ‘effector’ has defenses
A B

SCHEME 7.3 (a) Schematic diagram showing the bioreductive prodrug activation pathway;
(b) design of bioreductive prodrug.

in the activating process is oxygen-reactive. In this case, the relatively higher oxygen
level in healthy cells and tissues protects the prodrug from reduction by reacting
with an intermediate (usually a short-lived “free radical”), thus restoring the drug
to its original (prodrug) form.
A number of examples of bioreductive drugs exist, some of which are either in
clinical use or are currently undergoing preclinical or clinical studies. The first group
consists of molecules containing quinone moieties; best-known members include
mitomycin C (which contains an indolequinone nucleus) and the experimental agent
EO9. The mechanism of action of mitomycin is described in detail in Chapter 3.
The second group of bioreductive agents includes molecules based on aromatic N-
oxides (e.g., tirapazamine) and aliphatic N-oxides (e.g., AQ4N). The latter is pres-
ently being evaluated in the clinic, and preliminary pharmacodynamic and pharmaco-
kinetic results convincingly demonstrate that it is working through a bioreductive
mechanism (see Section 7.3.3.1). Another group comprises molecules containing
nitroheterocycles and includes experimental agents such as CB-1954, SN-23862,
NITP, SR4554, and pimonidazole, the latter of which is also being evaluated in the
clinic for use as a hypoxia marker.

7.3.3.1 AQ4N

Apart from the natural product mitomycin C, AQ4N is the first example of a
bioreductive agent to provide encouraging results in the clinic (see Structure 7.4).
It is a water-soluble anthracene derivative that, when administered intravenously,
targets the hypoxic compartments of tumors. Although AQ4N enters all cells freely,
in theory it is only converted to a charged (at physiological pH) tertiary amine species
in hypoxic tumor cells. This amine species becomes trapped in the hypoxic cells,
killing them through DNA intercalation and inhibition of topoisomerase II. Clini-
cians are generally enthusiastic about this mechanism of action because treating the
hypoxic fractions of tumors has the potential to significantly enhance selective
toxicity and thus improve therapeutic index.
AQ4N itself is a bis(N-oxide) which, because it is uncharged, crosses cell
membranes with ease. However, in hypoxic tumor cells, it is selectively and irre-
versibly reduced to AQ4, the charged quaternary amino form, where it remains

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 172 Thursday, October 26, 2006 7:34 AM

172 Chemistry and Pharmacology of Anticancer Drugs

CH3 CH3
O−
N NH
OH O HN + CH3 OH O HN + CH3
Bioreductive
Conditions

CH3 CH3
OH O HN O− OH O HN
N NH
+ CH + CH
3 3
AQ4N AQ4
(N-Oxides are not charged overall and (After reduction, tertiary amines cause
allow molecule to cross cell membrane) molecule to become “trapped” in cell
due to charge)
STRUCTURE 7.4 Structures of AQ4N and its biologically active metabolite (AQ4).
localized. The bioreduction process is enhanced in the absence of oxygen but inhib-
ited strongly in its presence. It is thought that when radiotherapy or chemotherapy
kills oxygenated cells on the outside of a tumor, the more-central quiescent AQ4-
containing cells become reoxygenated. When these cells resume replication, they
are rapidly killed by the antitopoisomerase II activity of AQ4. As demonstrated in
a recent Phase I clinical trial, systemic toxicity is thus reduced to a very low level
(i.e., the agent has a high maximum tolerated dose) because the prodrug form
(AQ4N) is relatively nontoxic and is only transformed into the active agent (i.e.,
AQ4) when it is inside hypoxic tumor cells.
Phase I clinical trials involving patients with non-Hodgkin’s lymphoma and
advanced solid tumors are presently underway. AQ4N is also being evaluated in the
treatment of esophageal cancer in combination with radiation therapy. During these
trials, it has been possible to confirm by studying biopsy material that the drug is
converted into its tertiary amine form (AQ4) in hypoxic cells. One curious but
nonproblematic side effect observed with AQ4N in these clinical trials was that the
mucous membranes of patients acquired a blue coloration, as did body fluids such
as urine and saliva. Interestingly, this effect was welcomed by many patients, who
felt that the unusually visible sign of the drug in their bodies might be associated
with a beneficial effect on their tumors.

7.3.4 POLYMER-DRUG CONJUGATES

Polymer-drug conjugates are based on the principle that a tumor’s blood supply and
lymphatic drainage capability are usually poor, and so polymer-bound drugs can
become trapped and accumulate. The linker between the drug and polymer may then
be cleaved enzymatically, releasing the drug specifically at the tumor site and thus
avoiding toxicities associated with wider distribution of the active agent. Polyethyl-
ene glycol (PEG) is a popular choice of macromolecule for studies of this type
because it is highly water soluble, nontoxic, and nonimmunogenic, and it has been
previously used in the modification of proteins to increase their biological half-life.
There has been a significant amount of research carried out on the biodistribution
of inert macromolecules and polymers in humans in relation to their hydrodynamic
radius. For example, early studies demonstrated a greater uptake of macromolecules

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 173 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 173

in tumor cells than in healthy ones, a phenomenon ascribed to a higher rate of

constitutive pinocytosis in tumor cells. However, a more general effect is now
recognized in which circulating macromolecules accumulate passively in solid
tumors. This phenomenon, known as the enhanced permeation and retention (EPR)
effect, is thought to be the result of three properties of solid tumors: (1) enhanced
vascular permeability, (2) limited macromolecular recovery via postcapillary
venules, and (3) poor lymphatic drainage. The optimum size range of macromole-
cules for this effect is thought to be 40 to 70 kDa. Conjugation of cytotoxic drugs,
such as doxorubicin, to polymers has been shown to result in an increased concen-
tration of the drug in tumors, and a number of clinical trials of such conjugates have
been undertaken during the past decade. Although PEG is an ideal pharmaceutically-
compatible polymer with which to form drug conjugates, research is also ongoing
to identify polymers with more favorable properties and to attach other types of
cytotoxic agents to PEG and related polymers. Important areas of concern with new
types of conjugates include stability in plasma, and the extent of selective uptake
by tumors compared to normal tissues, and the efficiency of drug release at the
tumor site. Although a number of Phase I and Phase II clinical trials have been
carried out with various polymer-drug conjugates, none has yet demonstrated any
clear benefits in the clinic.
One ingenious variant of this approach, PDEPT, involves the use of a two-
component system of polymer-linked drug and polymer-linked enzyme, which are
both administered and subsequently accumulate in the tumor. The linker attaching
the drug to the polymer is designed to be cleaved by the polymer-bound enzyme,
thus providing drug release specifically at the tumor site (see Section 7.4.3).
A related polymer uptake effect can be utilized by entrapping anticancer drugs
in liposomes of an appropriate size. For example, Caelyx™ is a liposomal form of
doxorubicin that is licensed for advanced AIDS-related Kaposi’s sarcoma and for
advanced ovarian cancer when platinum-based chemotherapy has failed. A similar
product, Myocet™, is licensed for use with cyclophosphamide for metastatic breast
cancer. Due to the different pharmacokinetic profiles of these liposomal preparations
compared to doxorubicin itself (i.e., accumulation at the tumor site), the incidence
of cardiotoxicity is also lowered, as is the potential for local necrosis at the site of
administration (see Chapter 3).

7.4 X-DEPT (BIPHASIC) STRATEGIES

The usefulness of traditional cytotoxic chemotherapeutic agents is usually restricted
by their low therapeutic indices. One approach to improving this situation has been
the development of strategies that allow the conversion of an inactivated form of an
anticancer drug (i.e., a prodrug) to an active agent specifically at the tumor site. One
such tactic relies on the activation process being carried out by an enzyme that has
been targeted to the surface of tumor cells via a suitable antibody. This therapy,
known as antibody-directed enzyme prodrug therapy (ADEPT), is presently in
Phase I evaluation. An alternative tactic, also being evaluated in Phase I, involves
activation of a prodrug by an enzyme caused to be expressed (i.e., not naturally
expressed) within the tumor cells. This therapy, known as gene-directed enzyme

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 174 Thursday, October 26, 2006 7:34 AM

174 Chemistry and Pharmacology of Anticancer Drugs

prodrug therapy (GDEPT) or virus-directed enzyme prodrug therapy (VDEPT),

involves the introduction of enzyme-producing genes through viral or nonviral vec-
tors. Finally, the concept of using separate polymers to concentrate both the enzyme
activity and prodrug in the tumor mass simultaneously led to the experimental
PDEPT approach, although this is no longer being developed.

7.4.1 ADEPT
Bagshawe and co-workers first introduced ADEPT in the early 1990s. In this two-
or three-component approach, an antitumor antibody linked to an enzyme is first
used to target the enzyme to tumor cells. A second antibody selective for the
antibody-enzyme conjugate may then be introduced to clear the conjugate from
general circulation, thus reducing systemic toxicity. A prodrug form of a cytotoxic
agent is then administered, which is transformed into the active agent by the anti-
body-bound enzyme selectively at the tumor site (Scheme 7.4 and Scheme 7.5).
Collateral damage to healthy tissue is theoretically avoided because the enzyme
chosen has no human equivalent (e.g., a bacterial enzyme). Variations of this
approach (GDEPT and VDEPT) were later established in which the activating
enzyme is introduced through organ-specific gene therapy or other means rather than
through an antibody-enzyme conjugate.
The ADEPT system originally proposed by Bagshawe utilizes the bacterial
enzyme carboxypeptidase G2 (CPG2), which is a metalloenzyme isolated from a
Pseudomonas species. It has no mammalian counterpart and catalyzes the conversion
of both reduced and nonreduced folates into pteroates and L-glutamic acid. It is also
known that CPG2 is capable of cleaving the amidic bond of methotrexate and folic
acid at the benzoylglutamate linkage. Thus, a conjugate of CPG2 with an appropriate
monoclonal antibody was designed for the selective activation of chemically
“masked” nitrogen mustards. The original experimental mustard prodrugs were
bifunctional alkylating agents incorporating either chlorine- or sulfonate-leaving
groups in which the activating effect of the ionizable carboxyl group had been
inhibited by protection with a glutamic acid moiety connected through an amide
linkage that formed the substrate for CPG2 (Scheme 7.4).

X CPG2 cleavage site

X
O
O
N COOH
N
HN
OH
Y
Y
COOH Free Mustard
Experimental ADEPT prodrugs Agent
1: X = Y = Cl
2: X = Cl, Y = OSO2CH3
3: X = Y = OSO2CH3

SCHEME 7.4 Examples of ADEPT prodrugs and their conversion to active agents. The
original experimental mustard prodrugs were bifunctional alkylating agents incorporating
either chlorine or sulfonate leaving groups (1-3).

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 175 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 175

To improve the clinical effectiveness of these prodrugs, extensive mechanistic

and structure-activity studies were carried out, with three factors being identified as
important: (1) the electronic character of the substituent para- to the aromatic ring
of the nitrogen mustard, (2) the electronic character of other substituents in the
aromatic ring (i.e., positions 2- and 3-), and (3) the nature of the leaving groups. As
a result, further novel nitrogen mustards were synthesized and evaluated in order to
optimize their properties based on the influence of these three factors. This led to
the di-iodomustard prodrug, ZD2767P, which entered two Phase I clinical trials that
have produced encouraging results (Scheme 7.5).

Cancer Cell

Ab CPG2

O
H
HO N O
HO
O l
HO O N
l
N

l
ADEPT Prodrug ZD2767P l
Active Agent:
Di-iodophenyl mustard

SCHEME 7.5 The principle of ADEPT, showing the conversion of the di-iodomustard pro-
drug ZD2767P, presently in Phase I clinical trials, into the active di-iodophenyl mustard at
the tumor site by the Ab-CPG2 conjugate.

In both trials, patients with colorectal carcinoma expressing carcinoembryonic antigen

received an A5B7 F(ab')(2) antibody conjugated to CPG2, followed later by admin-
istration of a galactosylated antibody directed against the active site of CPG2 (SB43-
gal) to clear and inactivate circulating enzyme. The prodrug (ZD2767P) was then
administered after plasma of enzyme conjugate levels had fallen to a predetermined
safe level. It proved possible to measure tumor enzyme levels through quantitative
gamma camera imaging (using labeled antibody conjugated) and from direct mea-
surements in plasma and tumor biopsies, which demonstrated that the median tumor-
to-plasma ratio of enzyme exceeded 10,000:1 prior to prodrug administration. More-
over, enzyme concentrations in the tumor were shown to be sufficient to generate
cytotoxic levels of active drug, and so the concentration of prodrug needed for optimal
conversion was achieved. Drug release was confirmed by establishing detectable
levels in plasma; one patient was observed to have a partial response, and six had
stable disease for a median of 4 months after previous tumor progression. Therefore,
conditions for effective antitumor therapy were met, and evidence of tumor response
in colorectal cancer obtained.
DNA interstrand cross-links, the cytotoxic lesions produced by these types of
mustard prodrugs, can be observed in individual cells by a modification of the

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 176 Thursday, October 26, 2006 7:34 AM

176 Chemistry and Pharmacology of Anticancer Drugs

COMET assay, and the extent of genomic damage at any time can be defined for a
population of tumor or normal cells. Experiments with the ZD2767/ADEPT system
in tissue culture showed that cross-links were produced in colon carcinoma cells in
a dose-dependent fashion that related to the degree of cell kill. However, the cross-
links were repaired within 24 hours in a sufficient proportion of cells, suggesting a
high probability of tumor regrowth in the clinic. This finding was supported by
similar findings in nude mice bearing xenografts of human colon carcinoma. There-
fore, a third clinical trial utilizing ZD2767 focused on studying DNA cross-links.
Tumor biopsies and peripheral blood lymphocytes were obtained on a small number
of patients for COMET assay. One patient with evidence of tumor response had
extensive cross-linking in tumor cells but no increase over controls in lymphocytes.
However, a patient with no evidence of response had no significant cross-links in
either his or her tumor or lymphocyte cells. The fact that tumor progression followed
in the patient with evidence of response is consistent with human tumor xenograft
model findings of repair of the cross-links induced by ZD2767. Therefore, research
is currently underway to develop novel nonmustard prodrugs suitable for ADEPT
that produce DNA adducts resistant to repair.

7.4.2 GDEPT (OR VDEPT)

GDEPT, also known as VDEPT, is a two-phase polymer-leased strategy to treat solid
tumors. First, a gene coding for a nonhuman enzyme is targeted to the tumor in a
vector suitable for expression. In the second phase, a nontoxic prodrug capable of
being activated by the tumor-associated enzyme is administered systemically. The
important feature of this concept is that the enzyme genes are expressed exclusively,
or with a relatively high abundance, in the tumor cells compared to healthy cells.
Unfortunately, current vectors for gene delivery are incapable of providing
expression of a foreign enzyme specifically in all tumor cells. Therefore, a bystander
effect (BE) is helpful whereby active agents, once produced, can diffuse and kill
neighboring tumor cells not expressing the foreign enzyme. In addition, as with
ADEPT, the prodrugs used need to be tailored to respond to the particular nonhuman
enzyme being used and, preferably, the enzyme should have no human homologs
so that prodrug conversion occurs only at the tumor site.
A large number of enzyme systems have been developed for GDEPT. Ideally,
they should have high catalytic activity toward their respective prodrugs without the
need for cofactors, which could become rate-limiting in target tumor cells. Also,
these enzymes must achieve concentrations sufficient to activate the prodrugs under
physiologic conditions. Enzymes suitable for GDEPT can be characterized into two
major classes. The first are those of nonmammalian origin, although they may have
human counterparts. Examples include viral TK, CPG2, and bacterial or yeast CD.
The second class comprises enzymes of human or other mammalian origins that are
found only in low concentrations in tumor cells or that may be completely absent,
such as CYP450, β-Glu, or CPA.
The enzyme genes can be manipulated for either intracellular or extracellular
expression in the recipient tumor cells. A key point concerning intracellular expres-
sion is that not only must the prodrug be able to penetrate a cell to become activated,

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 177 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 177

but the activated agent must be able diffuse back across the cellular membrane in
order to elicit a BE. This is not a requirement when the enzyme is expressed on the
outer cell surface because it can activate the prodrug extracellularly. Although the
latter mechanism can enhance a BE, the disadvantage is that activated drug can
potentially leak back into the general circulation and lead to toxic effects, thus
lowering the therapeutic index.
The prodrugs used for GDEPT and VDEPT should ideally possess such char-
acteristics as limited cytotoxicity until activated, good chemical stability under
physiological conditions, a favorable ADME profile, the ability to kill both prolif-
erative and quiescent cells after activation, and the capacity to induce a BE. Also,
for intracellular activation, the prodrugs should have sufficient lipophilicity to pen-
etrate cell membranes or otherwise utilize an active transport mechanism. Many
prodrugs in current use, such as the nucleoside analogs (e.g., 5-FC, CP, and CMDA)
and the alkylating agent CB-1954, penetrate cells by passive diffusion.
The precise mechanism of activation of prodrugs is also important. The prodrugs
used in GDEPT are commonly based on antimetabolites that require cycling cells
(e.g., in S phase) for cytotoxicity and are not active in quiescent cells. Therefore, it
has been suggested that alkylating prodrugs may have an advantage over, for exam-
ple, purine nucleosides or 5-FC in that they are cytotoxic to noncycling as well as
proliferative cells. In non-self-immolative prodrugs, the active agent is formed
directly following a one-step process. Most prodrugs used in clinical trials to date
have been based on licensed anticancer drugs whose characteristics are already well
known. Examples of activation mechanisms exploited include scission reactions
(e.g., CPG2, CE, CPA, PNP, PGA, MDAE, TP, β-GAL, β-L, and β-Glu), reductions
(e.g., nitroreductase [NR] and DT-diaphorase), phosphorylation (e.g., HSV-TK,
VZV-TK, and dCK), hydroxylation (e.g., CYP4B1), functional group substitution
(e.g., NH2 to OH; CD), deoxyribosylation (e.g., XGPRT), or oxidation (e.g., DAAO).
A challenge related to the choice of prodrug is that the enzymes used in GDEPT
impose rigid structural requirements on the prodrug substrates, which limits the
choice of anticancer drugs employed One approach to this problem is to design a
self-immolative prodrug that, when activated, forms an unstable intermediate that
extrudes the active agent through a series of subsequent degradative steps. For self-
immolative prodrugs, the initial activation step is generally enzymatic in nature and
distinct from the extrusion step, which is usually chemical and relies on spontaneous
fragmentation. Examples of self-immolative mechanisms include 1,4- or 1,6-elimi-
nations or cyclizations. A major advantage of this approach is that the structure of
the active drug is independent of the substrate requirements of the enzyme being
used in the system. Therefore, a broad range of drugs of various structural classes
can be converted to self-immolative prodrugs.
It is also an advantage if the expressed enzyme can activate the prodrug directly
without multiple catalytic steps because the requisite host endogenous tumor
enzymes can become defective or deficient in some tumor cells. For example,
alkylating agent prodrugs such as CP and CB-1954 (5-[aziridin-1-yl]-2,4-dini-
trobenzamide) have an activation cascade that depends on endogenous enzymes. In
the latter case, nitroreductase (NR) activates CB-1954 to the 5-aziridinyl-4-hydrox-
ylamino-2-nitrobenzamide intermediate, which is followed by a further enzymatic

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 178 Thursday, October 26, 2006 7:34 AM

178 Chemistry and Pharmacology of Anticancer Drugs

N N N

O2N HOHN HOHN

DT (Rat) 4

CONH2 2
CONH2 CONH2

NO2 NO2 NHOH

CB-1954

NR (E. Coli) NR (E. Coli)

SCHEME 7.6 Activation of prodrug CB-1954 by rat DT diaphorase [DT (rat)] or E. coli NR
(NR [E. coli]).

step that converts it to a powerful electrophile that alkylates DNA. An optimal half-
life for the active agent produced is also critical; it should be short enough to prevent
drug leakage from the tumor site but long enough to allow a local BE.
A recent GDEPT clinical trial involving NR to activate the prodrug CB-1954
used a replication-deficient adenovirus vector expressing NR from the CMV pro-
moter (CTL102). Patients with operable primary or metastatic liver tumors were
treated with increasing doses of CTL102 given as a single intratumor injection prior
to radical surgery. This trial demonstrated the tolerability of the vector and the ability
of CTL102 to induce a dose-dependent NR expression in the tumor. At the highest
dose of 5 × 1011 virus particles, NR expression occurred in up to 50% of the tumor
cells and was detectable in more than 50% of the specimen slides.
A detailed knowledge of the relevant enzymology is crucial when developing
X-DEPT approaches. For example, it is known that reduction of the prodrug CB-
1954 by rat DT-diaphorase results in conversion of the 4-nitro group to a hydroxyl-
amine moiety, leading to an agent capable of cross-linking DNA. Human DT-
diaphorase can also perform this reduction but at a slower rate, which explains why
degrees of activation are reported to be up to 10,000-fold in rodent cell lines, whereas
an activation of only 670-fold is observed in human cell lines. This finding probably
explains, in part, why CB-1954 is devoid of antitumor activity in humans. Interest-
ingly, Escherichia coli NR is able to reduce both the 2- and 4-nitro groups efficiently
and thus provides a good choice of enzyme for GDEPT studies (Scheme 7.6).
Another recent example of GDEPT is a clinical trial involving prostate cancer,
in which researchers are using CTL102 delivered via multiple intraprostatic injec-
tions in combination with intravenous CB-1954. These trials have demonstrated the
safety of the treatment and have provided preliminary evidence of biological activity.

7.4.3 PDEPT
PDEPT is an experimental two-phase antitumor strategy that employs a combination
of polymer-prodrug and polymer-enzyme conjugates to generate a cytotoxic species
selectively within the tumor mass. The treatment is based on the principle that
polymeric materials are known to accumulate in tumors due to their poor vasculature
(the so-called enhanced permeation and retention (EPR) effect).

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 179 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 179

A degree of proof-of-concept was demonstrated in preliminary preclinical in

vivo experiments in which polymeric prodrug conjugate PK1 (N-[2-hydroxy-
propyl]methacrylamide [HPMA] copolymer-Gly-Phe-Leu-Gly-doxorubicin) and the
HPMA copolymer-cathepsin B were studied in combination. Following polymer
conjugation (with a yield of 30% to 35%), HPMA copolymer-cathepsin B retained
approximately 20% to 25% of native enzymatic activity in vitro. To investigate
pharmacokinetics in vivo, 125I-labeled HPMA copolymer-cathepsin B was adminis-
tered intravenously to B16F10 tumor-bearing mice. It was found to exhibit a longer
plasma half-life (free cathepsin B t1/2α = 2.8 h; bound cathepsin B t1/2α = 3.2 h) and
a 4.2-fold increase in tumor accumulation compared to the free enzyme. Also, when
PK1 (10 mg kg-1 doxorubicin equivalent) was injected intravenously into C57 mice
bearing B16F10 tumors, followed after 5 hours by HPMA copolymer-cathepsin B,
a rapid increase in the rate of doxorubicin release occurred within the tumor (3.6-
fold increase in the AUC compared to that seen for PK1 alone). Furthermore, when
this PDEPT combination was used to treat established B16F10 melanoma tumors
(single dose; 10 mg kg-1 doxorubicin equivalent), the antitumor activity observed
was 168% compared to 152% seen for PK1 alone and 144% for free doxorubicin.
The PDEPT combination also showed activity against a COR-L23 xenograft, which
PK1 did not.
These preclinical results provide some limited evidence that PDEPT may work
in the clinic. This approach also has certain advantages compared to ADEPT and
GDEPT. For example, the relatively short plasma residence time of the polymeric
prodrug allows subsequent administration of polymer-enzyme without fear of pro-
drug activation in the circulation. In addition, polymer-enzyme conjugates have
reduced immunogenicity compared to enzyme alone. Although the PDEPT strategy
is not being further developed at present, advances in polymer chemistry and drug
delivery technologies may see this approach revisited in the future.

7.5 ENZYMATIC TARGETING

In addition to the use of enzymes in the X-DEPT therapies previously described,
three other distinct strategies involving enzymes are discussed in this section. The
first relies on the use of an enzyme constitutively expressed by tumor cells (but not
healthy cells) to activate a prodrug without further intervention. The second is a
similar approach but requires systemic administration of an enzyme cofactor in order
to activate the enzyme. The third approach involves use of the enzyme asparaginase
(Erwinase™) to break down asparagine in the blood and tissues. Asparagine is
required for survival by leukemic cells.

7.5.1 PRODRUG ACTIVATION BY CONSTITUTIVELY OVEREXPRESSED

TUMOR ENZYMES
There is growing evidence that certain tumor cells constitutively overexpress some
enzymes relative to healthy cells. For many years, it has been a goal to use these
enzymes to activate a systemically administered prodrug selectively at the tumor
site. To realize this goal, the first challenge is to identify a suitable enzyme or enzyme

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 180 Thursday, October 26, 2006 7:34 AM

180 Chemistry and Pharmacology of Anticancer Drugs

family, and the second challenge is to design prodrugs that are not toxic to normal
tissues but that can be converted to cytotoxic species by the relevant enzyme.
In terms of the choice of enzyme, recent attention has focused on the cytochrome
P450 family which, in addition to endogenous substances, metabolizes many drugs
and other xenobiotics (i.e., foreign) substances. The P450s are mixed function
oxidases displaying selective aromatic hydroxylase activity that can be exploited by
medicinal chemists to activate prodrugs. One particular family member, CYP1B1,
is thought to be overexpressed in a wide variety of human tumors but not in normal
tissues, and so is presently being studied more than any other family members in
the context of drug activation.
It is worth noting that subtle genetic modifications (i.e., base pair changes) in
the gene coding for the P450 drug metabolizing enzymes are often used to explain
differences between the response of individuals to cancer chemotherapy agents.
Thus, it is thought that relatively modest genetic changes can have an impact on the
pharmacodynamic and pharmacokinetic characteristics of any anticancer drug and
may have an especially significant effect on those agents with narrow therapeutic
indices. The detection of variations in the P450 genes is therefore growing in
importance as a means to select the most beneficial anticancer drugs for an individual,
and the most appropriate doses and dose schedules to use. This approach has led to
the identification of many different types of P450 enzymes, such as those in the
CYP2C and CYP3A families, as well as CYP1B1 and CYP2D6. In particular, studies
on the CYP2C and CYP3A families have shown that polymorphisms (i.e., subtle
changes of base-pair sequence) in individual genes can lead to significant changes
in the cytotoxic potential of certain drugs. In addition, the etiology of a range of
human cancers is now thought to be related to polymorphisms in the CYP1B1 gene.
Prodrugs based on this approach are still at the discovery stage and so are only
just beginning to reach the clinical evaluation. A number of biotech companies have
been set up to exploit this new therapeutic strategy, and so the structures of novel
prodrugs are not always freely available. However, prodrugs based on the duocar-
mycin DNA minor-groove alkylating agents are in development that require hydrox-
ylation in order to become DNA reactive. Similarly, Phortress, a novel agent that
has just reached Phase I trials, is activated by the CYP family of enzymes to give
an intermediate that spontaneously fragments into DNA-damaging species.
The major criticism of this therapeutic approach is that, given the heterogeneity
of tumors, it is highly unlikely that all tumor cells in a tumor mass will robustly
overexpress the chosen enzyme and will therefore succumb to the drug released.
However, this problem may be somewhat alleviated by a bystander effect. A second
criticism is that it is unlikely that the chosen enzyme would not be expressed
elsewhere in the body. Therefore, taken together, these points suggest that the
therapeutic index of such a therapy is likely to be narrower in practice than expected.

7.5.2 COSUBSTRATE-MEDIATED PRODRUG THERAPY

This therapeutic strategy is similar to the one previously described but requires a
systemically administered cofactor to activate an enzyme constitutively expressed
by tumor cells. One such activation system, presently being evaluated in Phase I

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 181 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 181

clinical trials, involves the systemic administration of a small synthetic cosubstrate

molecule, dihydronicotinamide ribose (NRH), which can activate the latent enzyme
NAD(P)H quinone oxidoreductase 2 (NQO2). The enzyme is a flavin-containing
oxidoreductase that catalyzes two- or four-electron reductions of a variety of quinone
compounds. Crucially, it has a unique activation mechanism that relies on the
synthetic electron donor cosubstrate NRH.
The prodrug used in this system is CB1954 (see Scheme 7.6), which is activated
by conversion to a 4-hydroxylamine derivative thus generating a potent DNA cross-
linking agent (see Section 7.4.2) when administered together with NRH (see Struc-
ture 7.5). Interestingly, although activation has been shown to occur in some rat
tumors, human cancer cells are inherently resistant to CB1954 because they cannot
efficiently catalyze this conversion. Also, the expression of DT-diaphorase (NQO1),
an enzyme known to be expressed in high concentrations in certain human tumors,
appears to be related to NQO2 activity.

STRUCTURE 7.5 Molecular model showing the flavin-containing NQO2 oxidoreductase

enzyme that catalyzes reduction of the prodrug CB1954 (bound in left-hand side of image;
arrowed) in the presence of the electron donor NRH (bound in right-hand side of image;
arrowed).

In preclinical studies, the presence of NRH has been shown to provide a very
large (i.e., 100- to 3000-fold) increase in the cytotoxicity of CB1954 in both non-
transfected human tumor cell lines and NQO2-transfected rodent cells. Other
reduced pyridinium compounds have been shown to serve as cosubstrates for NQO2,
and Knox and co-workers have identified key SAR features through a medicinal
chemistry program. One derivative in particular, 1-carbamoylmethyl-3-carbamoyl-
1,4-dihydropyridine (Structure 7.6), has been identified as a good cosubstrate for
NQO2 but with several key advantages over NRH, such as greater stability and
cellular penetration properties, along with the ability to potentiate the cytotoxicity
of CB1954.

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 182 Thursday, October 26, 2006 7:34 AM

182 Chemistry and Pharmacology of Anticancer Drugs

CH2 CONH2

CONH2

STRUCTURE 7.6 Structure of the synthetic NQO2 cofactor 1-carbamoylmethyl-3-carba-

moyl-1,4-dihydropyridine.
Importantly, 1-carbamoylmethyl-3-carbamoyl-1,4-dihydropyridine is easy to
synthesize, and Phase I clinical trials are now underway with patients being treated
with both CB1954 and this cofactor to establish whether selective activation of the
prodrug occurs in tumor cells.

7.5.3 ASPARAGINASE
Asparaginase (Cristanaspase™; Erwinase™) therapy takes advantage of the fact that
certain types of tumor cells (in particular lymphoblastic leukemia cells) are unable
to synthesize their own asparagine. Systemic administration of the enzyme aspara-
ginase causes a significant reduction in systemic concentrations of asparagine, thus
starving the cancer cells of this nutrient and leading to cell death while healthy cells
continue to synthesize their own. This process is described in detail in Chapter 8
(Biological Agents).

7.6 PHOTOACTIVATED DRUGS (PHOTODYNAMIC

THERAPY)
Photodynamic therapy (PDT) involves the administration of a nontoxic prodrug that
can be activated selectively at the tumor site by the light of a specific wavelength.
This general strategy has been in use for many years for the treatment of psoriasis
using 8-methoxypsoralen (PUVA treatment) (Structure 7.7). This agent is relatively
nontoxic until exposed to UV light, when it then cross-links DNA at thymine sites,
causing distortion of the DNA helix with consequent toxicity toward the psoriatic cells.

OCH3

O
O

CH3

STRUCTURE 7.7 Structure of 8-methoxypsoralen used in the phototherapy of psoriasis

(PUVA treatment).

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 183 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 183

This prodrug photoactivation concept was extended into cancer therapy in

the early 1990s, when it was discovered that porphyrin-type molecules are selec-
tively taken up by some tumors. This led to the development of porfimer sodium
(Photofrin™) and temoporfin (Foscan™), which are now used in the photodynamic
treatment of various tumors (Structure 7.8 and Structure 7.9). After systemic admin-
istration, these drugs accumulate in malignant tissue and can then be activated by
laser light to produce cytotoxic effects. Indications are presently limited to obstruct-
ing esophageal cancer and NSCLC, and to situations in which a tumor manifests
near the skin surface (e.g., advanced head and neck cancers). However, recent
progress in the development of surgical lasers with flexible optical fibers has allowed
experimental use of these agents for the treatment of tumors in inaccessible places,
such as parts of the GI tract and the ovaries. An intense nonlaser light source has
also been developed for this type of therapy, and many new types of prodrugs are
under investigation. However, one challenge to progress in broadening the use of
this therapy is that the shorter wavelengths of light required to activate the present
generation of photoactivated agents do not penetrate tissues very effectively. There-
fore, new agents that can be activated at longer wavelengths are required so that
deeper tissue penetration of light sources can be achieved.

7.6.1 PORFIMER SODIUM

Porfimer sodium (Photofrin™), introduced in the early 1990s for the treatment of
esophageal cancer, is not a single entity but a dark reddish-brown mixture of oligo-
mers formed by up to eight porphyrin units joined through ether and ester linkages
(Structure 7.8). As part of a two-stage treatment process, it is administered by
intravenous infusion over 3 to 5 minutes, followed by illumination of the tumor with
laser light at 630 nm after 40 to 50 hours. Porfimer sodium is licensed for use in
obstructing esophageal and endobronchial NSCLC that can be reached with laser
endoscopy methods. It is also in use experimentally for other tumors of the skin and
body cavities (e.g., stomach, colon, ovarian) that can be reached with flexible lasers.

NaO2C(CH2)2 CH3 R CH3 H3C (CH2)2CO2Na

CH3
NaO2C(CH2)2 N CH3 H3C N O H3C N (CH2)2CO2Na

NH HN H NH HN H H NH HN
CO(CH3)2

H3C O C CH3
N N H2 N
CH3 CH3

R CH3 NaOOC(CH2)2 CH3 H3C R

n

H
R= HO CH and/or C CH3 n = 0-6
H
CH3

STRUCTURE 7.8 Structure of porfimer sodium (Photofrin™).

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 184 Thursday, October 26, 2006 7:34 AM

184 Chemistry and Pharmacology of Anticancer Drugs

The cytotoxicity and consequent antitumor effect of porfimer sodium are light
and oxygen dependent. After intravenous injection, the agent clears from most tissues
in 40 to 72 hours. However, organs of the reticuloendothelial system (including the
liver and spleen), the skin and, most importantly, tumor tissue, retain the agent for
longer periods of time. Therefore, selectivity occurs through a combination of
preferential uptake and retention of the agent by the tumor and accurate delivery of
the laser light, which induces a photochemical rather than a thermal effect.
At the molecular level, propagation of radical reactions, initiated after porfimer
sodium absorbs light to form singlet oxygen, is thought to be responsible for the
cellular damage caused by this agent. Hydroxyl and superoxide radicals, lethal to
cells, are thought to form during subsequent radical reactions. Ischemic necrosis
secondary to vascular occlusion, thought to be partly mediated through thromboxane
A2 release, can also cause tumor shrinkage. From a patient-care perspective, it is
important to note that the necrotic reaction and associated inflammatory responses
may develop over several days.
Side effects of porfimer sodium include severe photosensitivity, with sunscreens
offering no protection. Although some individuals are photosensitive to porfimer
sodium for 90 days or more after administration due to residual drug in all areas of
the skin, in general, patients are advised to avoid bright indoor light or direct sunlight
for at least 30 days after treatment. Therefore, careful monitoring of treated patients
is required, and it is important to expose small test areas of skin to sunlight imme-
diately after treatment to assess the degree of photosensitivity likely to occur. Con-
stipation can also be a problematic but treatable side effect.

7.6.2 TEMOPORFIN
Temoporfin (Foscan™) is a second-generation photosensitizing agent for PDT intro-
duced in the mid- to late-1990s (Structure 7.9). The main advantage of temoporfin
over porfimer sodium is that it is a discrete chemical moiety rather than a polymeric
mixture. It is licensed for the PDT of advanced head and neck cancer.
Temoporfin is administered by intravenous injection over at least 6 minutes. Side
effects include photosensitivity (as with porfimer sodium, sunscreens offer no pro-
tection), and so exposure of eyes and skin to bright indoor light or direct sunlight
should be avoided for at least 15 days after treatment. Other side effects include
constipation, local hemorrhage, facial pain and edema, dysphagia, and possible
scarring near the treatment site. Temoporfin is contraindicated in patients with
porphyria or other diseases exacerbated by light. In addition, ophthalmic slit-lamp
examination cannot be carried out for 30 days after administration.

7.7 BORON NEUTRON CAPTURE THERAPY (BNCT)

Boron Neutron Capture Therapy (BNCT) is a targeted biphasic approach to cancer
therapy in which boron-10 (10B)–enriched delivery agents are first administered
intravenously and are taken up by tumors to varying extents. Once maximal tumor
uptake has been achieved, the target area is then irradiated with low-energy neutrons
(epithermal neutrons in the 1 to 10,000 electron volt energy range), which become

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 185 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 185

OH

HO
N HN

NH N
OH

HO

STRUCTURE 7.9 Structure of temoporfin (Foscan™).

thermalized at depth in the tissue and are captured by the 10B atoms, with the resultant
reaction producing α-particles (4He) and lithium-7 (7Li) ions that destroy the tumor
tissue:

10 B + neutrons → 13B → 7Li + 4He + 2.79 MeV

These particles are energetic but have a limited range of less than 9 μm in tissue,
thereby preferentially targeting the radiation to the tumor while sparing surrounding
tissues. They also have high linear energy transfer (LET) characteristics compared
to X-rays and gamma rays and, as a consequence, the level of tissue oxygenation
is less important, which is an additional advantage if the tumors contain hypoxic
cells.
A major advantage of BNCT over conventional radiotherapy is that it enables
relatively large volumes of normal tissue to be irradiated with a reduced risk of
adverse effects. This is because the presence of a neutron capture agent ensures that
the radiation dose to the tumor is significantly higher than that to the surrounding
normal tissue, despite the fact that the delivery of low energy neutrons cannot be
focused. However, this advantage is dependent upon successful selective uptake of
the neutron capture agent by the tumor. A major disadvantage is the lack of avail-
ability of expensive nuclear reactor or linear accelerator facilities (required to pro-
duce neutron beams) in most hospitals, and a paucity of effective BNCT delivery
agents.
Experimental and clinical studies relating to the use of BNCT have largely, but
not exclusively, focused on brain tumors— glioblastomas, in particular. High-grade
gliomas are frequently inoperable and, without treatment, can prove fatal within a

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 186 Thursday, October 26, 2006 7:34 AM

186 Chemistry and Pharmacology of Anticancer Drugs

few months of diagnosis, with death being due to extensive proximal disease and
local metastatic spread. Gliomas rarely metastasize via the blood and are only
occasionally dispersed along cerebrospinal fluid pathways. Therefore, a treatment
capable of controlling the primary disease and local metastatic spread could poten-
tially result in patient cures. Unfortunately, high-grade gliomas do not respond well
to conventional radiation or traditional chemotherapeutic agents. Enhanced radio-
therapy methods, including hypoxic cell sensitizers and fast neutrons, have also
proved ineffective. Therefore, BNCT has the potential to address this challenging
clinical problem provided that suitable 10B delivery agents can be found, which is
an active area of research.
A number of boron delivery agents have either been studied or are in develop-
ment. For example, borocaptate sodium (Na2B12H11SH) has been shown to be effec-
tive in limited clinical trials in Japan (but in fewer than 150 patients). In these trials,
patients were irradiated with thermal neutrons after the administration of BSH and,
in those with relatively small superficial (less than 6-cm deep), high-grade brain
tumors, 5-year survival rates approached 60%. This compared well with a 5-year
survival rate of less than 3% in patients with comparable tumors treated with
conventional radiotherapy. However, the use of thermal neutrons with limited pen-
etration characteristics in tissue was a major limitation of this Japanese study.
More recently, much effort has focused on the development of more-deeply
penetrating epithermal neutron beams. At present, nuclear reactor based epithermal
beam facilities are available at the Brookhaven National Laboratory (BNL) in the
U.S., at the European Joint Research Center (Petten) in the Netherlands, and at the
University of Helsinki in Finland. In addition, the development of a new British
epithermal beam facility has been funded at the University of Birmingham. This
facility is unique in that it is based on a linear accelerator.

OH

B
HO NH2

OH

STRUCTURE 7.10 Structure of the BNCT agent p-boronophenylalanine (BPA).

Clinical BNCT studies using an epithermal beam have been undertaken at BNL
using the delivery agent p-boronophenylalanine (BPA) dissolved in a fructose solu-
tion and administered intravenously (Structure 7.10). Preliminary results have been
encouraging in terms of tumor responses observed, and no evidence of radioinduced
brain necrosis has been observed.
BNCT also has the potential to treat tumors other than those of the central
nervous system. For example, clinical trials in Japan involving the treatment of
malignant melanoma using BPA as the neutron capture agent have provided encour-
aging results. The development of improved neutron capture agents could potentially

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 187 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 187

enable a wider range of tumor types to be treated. Recently, a compound called

boronated porphyrin, or BOPP, has been developed that accumulates in tumors in
a manner similar to the PDT agents.

7.8 NOVEL DRUG DELIVERY APPROACHES

Effective drug delivery remains a challenge in the management of cancer. Existing
drugs could be significantly more effective if techniques could be developed to
deliver them selectively to the tumor site while avoiding healthy tissues. Therefore,
there is a focus on the development of sophisticated targeted delivery systems that
will not only supplement conventional chemotherapy and radiotherapy but may also
prevent the occurrence of drug resistance. Knowledge and experience from areas
such as nanotechnology, advanced polymer chemistry, and electronic engineering
are being drawn upon to help develop these novel approaches. Examples from the
areas of gene therapy, nanotechnology, novel polymers, and ultrasound are high-
lighted below.

7.8.1 GENE THERAPY

A significant research effort is underway to establish methodologies to deliver viral
and other vectors to specific organs and tissues to enable therapeutic strategies such
as GDEPT. There is also interest in the use of gene therapy strategies in combination
with radiotherapy, including the use of radiation-sensitive promoters to control the
timing and location of gene expression specifically within tumors. In particular, it
is thought that gene therapy may enhance the effectiveness of limited-dose radio-
therapy, which falls short of destroying a cancer. Yet another possibility is radiopro-
tective gene therapy using transgenes coding for antioxidants that may ameliorate
the effects of radiation-induced reactive oxygen species and could be used to spare
normal tissues. Tumor-specific vectors leading to the expression of anti-angiogenic
proteins are also of interest as a means to shut down blood flow specifically in tumors.
Therefore, not surprisingly, most research activity in the gene therapy area
involves the development of have viral vectors and techniques to “wrap-up” DNA
vectors in microparticles, such as liposomes or nanoparticles, to protect them from
degradation by enzymes in the bloodstream and to target them to specific organs or
tissues. In the latter case, the DNA must then be released, incorporated into the
cancer cell genome, and efficiently expressed, which is also an active area of
research.
Finally, it should be noted that the long-term safety of gene therapy techniques
is a significant concern. In practice, it is highly unlikely that vectors can be targeted
completely specifically to tumors, organs, or tissues, and some degree of collateral
delivery to healthy cells is always likely. This could lead to cancer in these healthy
tissues after a period of time, a phenomenon already observed in one clinical trial
of children who received gene-replacement therapy for a noncancer disease. The
viral vector used caused DNA to be added to the genome of healthy cells, and some
patients developed a form of leukemia.

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 188 Thursday, October 26, 2006 7:34 AM

188 Chemistry and Pharmacology of Anticancer Drugs

7.8.2 NANOTECHNOLOGY-BASED DRUG DELIVERY

Nanobiotechnology is a research area being applied to the improvement of drug
delivery in various cancer therapies. It has been known for some time that encap-
sulation of cancer drugs in particles such as liposomes can modify their behavior
after administration. Advantages of polymeric micellar drug delivery systems
include: (1) long circulation time in the blood and stability in biological fluids; (2)
appropriate size (10 to 30 nm) to escape renal excretion but to allow for extravasation
at the tumor site; (3) simplicity in incorporating the drug compared to covalent
bonding of the agent to a polymeric carrier; and (4) drug delivery that is independent
of drug characteristics. Some micellar systems are dynamically stable because their
solid-like cores dissociate slowly at concentrations below their critical micelle con-
centration. Others are not so stable and require additional stabilization that may be
achieved, for instance, by cross-linking the micelle core. In a study of the pharmaco-
kinetics and distribution of doxorubicin in micelles formed by drug-polymer conju-
gates, the micelles circulated much longer in blood than did free agents. Furthermore,
the uptake of the conjugated drug by various organs proceeded much more slowly
than that of free drug, and lower levels of conjugate were found in the heart, lungs,
and liver than in the tumor. These findings were the basis for developing Caelyx™,
a liposomal form of doxorubicin that is licensed for advanced AIDS-related Kaposi’s
sarcoma and for advanced ovarian cancer when platinum-based chemotherapy has
failed (see Chapter 3). Due to the different pharmacokinetic profile of this liposomal
preparation (i.e., accumulation at the tumor site), the incidence of cardiotoxicity is
reported to be lower, as is the potential for local necrosis at the site of administration.
A similar product, Myocet™, is licensed for use in combination with cyclophos-
phamide for metastatic breast cancer (see Chapter 3). Combined delivery of drugs
and surfactants in micelles and liposomes, and the use of polymer-drug conjugates
have also been studied as a means to overcome drug resistance by bypassing the
P-Glycoprotein (P-Gp) pump system. The potential for liposomes and polymers to
accumulate in tumors has also led to therapies such as PDEPT.
Various other long-circulating colloid drug delivery systems have been studied
for use in cancer chemotherapy. Only a few known block copolymers form micelles
in aqueous solution, and these include AB-type block copolymers (e.g., poly[L-
amino acid]-co-polyethylene oxide). In particular, the PLURONIC family of ABA-
type triblock copolymers having the structure PEO-PPO-PEO is well known, where
PPO is polypropylene oxide, the hydrophobic central PPO block forms a micelle
core, and the flanking PEO blocks form the shell or corona that protects the micelles
from recognition by the reticuloendothelial system (RES).
A common structural motif of all these long-circulating systems, whether they
are nanoparticles, liposomes, or micelles, is the presence of polyethylene oxide (PEO)
on their surfaces. The dynamic PEO chains prevent opsonization and render the
particles unrecognizable by the RES. This advantage has encouraged extensive
research into the development of new techniques (ranging from physical adsorption
to chemical conjugation) to coat particles with PEO. From a technological perspective,
polymeric micelles formed by hydrophobic-hydrophilic block copolymers, with the
hydrophilic blocks comprised of PEO chains, are very attractive drug carriers. These

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 189 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 189

micelles have a spherical, core-shell structure, with the hydrophobic block forming
the core of the micelle and the hydrophilic block or (blocks) forming the shell. They
have promising properties as drug carriers both in terms of their size and architecture.
The hydrophobic drug molecules partition inside the micelles, which are 15 to 35 nm
in diameter. It is thought that these structures enter cells by phagocytosis or endo-
cytosis. The drug is thus delivered inside the cells by local delivery or by fusion of
the particle with the cell membrane, which destabilizes the micelle.
There is now significant ongoing research into new types of nanoparticles and
the nanoencapsulation of drugs for targeted delivery to tumors of various organs
both as a single therapy and in combination with other treatments, such as radio-
therapy. In particular, there is growing interest in attaching entities such as antibody
fragments to the surface of drug-containing nanoparticles in order to guide them
specifically to a tumor, where they can release the active agent. Nanoparticles are
also being used in gene therapy strategies as previously described. Finally, nano-
technology-based diagnostics can be combined with therapeutics, which is likely to
be important in the future for the growing emphasis on the personalized management
of cancer.

7.8.3 INTRACRANIAL DELIVERY

The improvement of cancer drug delivery to the brain is now a major area of research.
One of the current limitations of the treatment of brain tumors is the lack of a suitable
method to deliver therapeutic agents directly to the lesion. The challenge for systemic
therapy is to develop methods to allow drugs to cross the blood–brain and
brain–tumor barriers in order to allow higher concentrations to be obtained within
the tumor bed. There are also opportunities in local drug delivery, and one commer-
cially successful product of research in this area, Gliadel™, is based on local,
controlled delivery of carmustine by a biodegradable polymer implanted at the tumor
site after surgical resection (see Chapter 3). This allows effective concentrations of
carmustine to reach any remaining tumor cells, a situation not usually achieved with
systemic administration due to dose-limiting bone marrow toxicity. It also avoids
some of the other adverse effects of carmustine, such as cumulative renal damage
and delayed pulmonary fibrosis. Many other promising examples of both polymer
and nonpolymer based drug delivery systems are under investigation.

7.8.4 ULTRASOUND TARGETING

The efficacy of cancer chemotherapy is often limited by the toxic effects of the
current generations of drugs. One approach to this problem is to sequester a drug
into a “package” that interacts minimally with healthy cells, keeping the drug
contained until release at an appropriate time at the tumor site. Recently, an exper-
imental method has been developed that involves ultrasound to achieve this objective.
It involves systemic administration of a micellar drug carrier with a hydrophobic
core containing an effective amount of an anticancer drug. Ultrasonic energy is then
applied to the tumor site to release the drug from the hydrophobic core of the
micelles. Examples of polymers studied include ABA triblock copolymers, AB

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 190 Thursday, October 26, 2006 7:34 AM

190 Chemistry and Pharmacology of Anticancer Drugs

diblock copolymers, mixtures of these, and mixtures of such polymers with PEGy-
lated diacylphospholipids.
Ultrasound has been used extensively for both medical diagnostics and physical
therapy, and is widely regarded to be safe when exposed to healthy tissues. Advan-
tages of ultrasound include the fact that it is noninvasive, and that its energy can be
controlled and focused easily with a capability of penetrating deep into tissues to
predetermined depths. In particular, large amounts of ultrasonic energy can be
accurately deposited deep into tumors. The depth of penetration and the shape of
the energy deposition pattern may be controlled by varying the ultrasound frequency
and the type and shape of the transducer. Optimal power densities at the target site
may be obtained by adjusting the output power of the ultrasonic transducer.
Two possible mechanisms for the ultrasound enhancement of chemotherapy have
been proposed. The first relates to enhanced drug release from micelles (as previously
described). The other is associated with the apparent enhanced uptake of the micellar-
encapsulated drug by cells. For example, there are literature reports of ultrasound-
induced hypersensitization of anthracycline-sensitive cell lines, which would appear
to support the enhanced drug-uptake mechanism. However, other experiments on
the effect of low-frequency ultrasound pulse duration on drug uptake suggest that
both mechanisms may work in concert.
Both low-frequency and high-frequency ultrasound have proved effective in
triggering drug release from micelles. Low-frequency ultrasound is more effective
but does not allow sharp focusing. In contrast, high-frequency ultrasound allows
sharp focusing but does not penetrate as deeply into the interior of the body. There-
fore, optimal design of ultrasound treatment will depend on tumor size and location.
For example, for tumors 2 cm in diameter or larger, application of 100 kHz (or a
lower frequency) ultrasound is feasible, with optimal power densities achieved by
controlling output energy. For smaller tumors that are not so deep, a higher-frequency
ultrasound can be used because it provides sharper focusing. In addition, power
densities produced at the focal site by existing hyperthermia devices appear sufficient
to cause drug release from micelles. Importantly, in vivo experiments have demon-
strated improved survival rates for ovarian carcinoma–bearing mice treated with 3
mg/kg of doxorubicin in PLURONIC micelles delivered intraperitoneally in com-
bination with ultrasound (1 MHz, 30 s, 1.2 W/cm2) applied 1 hour after injection
of the drug. In the same experiment, no effect was observed in untreated control
mice or mice treated with 3 mg/kg of doxorubicin administered intraperitoneally in
a physiological solution.
The current findings suggest that ultrasound targeting is a promising new area
of research, particularly as high-frequency ultrasound is widely used in clinical
practice for imaging purposes (though at much lower power densities than used in
these experiments). The ideal scenario would be to ultimately combine imaging and
therapeutic ultrasound transducer arrays in one instrument that could be used initially
to image the tumor but, followed by automatic focusing of the ultrasound beam for
therapeutic purposes. Finally, the combination of micellar drug delivery carriers and
ultrasound is especially promising for treating MDR-positive tumors that do not
respond to conventional treatment regimens.

© 2007 by Taylor & Francis Group, LLC

9219_C007.fm Page 191 Thursday, October 26, 2006 7:34 AM

Tumor-Targeting Strategies 191

FURTHER READING
Algire, G.H., and Chalkley, H.W. “Vascular Reactions of Normal and Malignant Tissue In
Vivo,” J. Nat. Cancer Inst., 6:73-85, 1945.
Denekamp, J. “Endothelial Cell Proliferation as a Novel Approach to Targeting Tumour
Therapy,” Br. J. Cancer, 45:136-139, 1982.
Folkman, J. “Tumor Angiogenesis: Therapeutic Implications,” NEJM, 285:1182-86, 1971.
Folkman, J. “Angiogenesis in Cancer, Vascular, Rheumatoid, and Other Disease,” Nat. Med.,
1:27-31, 1996.
Hahnfeldt, P., et al. “Tumor Development under Angiogenic Signaling: A Dynamical Theory
of Tumor Growth, Treatment Response, and Postvascular Dormancy,” Cancer Res.,
59:4770-75, 1999.
Hua, J., et al. “The Role of New Agents in the Treatment of Colorectal Cancer,” Oncology,
66(1):1-17, 2004.
Hurwitz, H., et al. “Bevacizumab Plus Irinotecan, Fluorouracil, and Leucovorin for Metastatic
Colorectal Cancer,” NEJM, 350:2335-42, 2004.
Presta, L.G., et al. “Humanization of an Anti-Vascular Endothelial Growth Factor Monoclonal
Antibody for the Therapy of Solid Tumors and Other Disorders,” Cancer Res.,
57:4593-99, 1997.
Saaristo, A., et al. “Mechanisms of Angiogenesis and Their Use in the Inhibition of Tumor
Growth and Metastasis,” Oncogene, 19:6122–29, 2000.
Salgaller, M.L. “Technology Evaluation: Bevacizumab, Genentech/Roche,” Curr. Opinion
Mol. Ther., 5(6):657-67, 2003.
Sheng, Y., et al. “Combretastatin Family Member OXI4503 Induces Tumor Vascular Collapse
through the Induction of Endothelial Apoptosis,” Int. J. Cancer, 111:604-10, 2004.

© 2007 by Taylor & Francis Group, LLC