THE DESIGN AND DEVELOPMENT OF ANTI?CANCER DRUGS

THE DESIGN AND DEVELOPMENT OF
ANTI-CANCER DRUGS

Cancer is one of the most widespread and feared diseases in the Western world today -
feared largely because it is known to be difficult to cure. The main reason for this
difficulty is that cancer results from the uncontrolled multiplication of subtly modified
normal human cells. One of the main methods of modern cancer treatment is drug therapy
(chemotherapy).

The majority of drugs used for the treatment of cancer today are cytotoxic (cell-killing)
drugs that work by interfering in some way with the operation of the cell's DNA.
Cytotoxic drugs have the potential to be very harmful to the body unless they are very
specific to cancer cells - something difficult to achieve because the modifications that
change a healthy cell into a cancerous one are very subtle. A major challenge is to design
new drugs that will be more selective for cancer cells, and thus have lesser side effects.
Initially the specificity of drugs was worked out simply by testing on animals, but now it is
possible to use our knowledge of cancer cell biology to actively design drugs to be more
specific. However, animal tests still need to be carried out at some point.

As with any pharmaceutical, new anticancer drugs are developed in a three-step process.

Step 1 - Initial discovery
A wide range of compounds, both natural and synthetic, are tested in high-capacity screens
to discover molecules with useful properties.

Step 2 - Molecular modification of a known compound
A molecule that shows suitable properties is chemically altered to give it the best
combination of properties to make the most effective anti-cancer drug.

Step 3 - Development into a useful pharmaceutical
Because the above process is very time-consuming and expensive, the new discovery is
usually patented at this time so that the discoverers can eventually recover some of these
costs. The most effective route for synthesising the molecule is then worked out. A long
process of advanced testing is then begun, ending up with tests on patients in specialised
hospitals. If the results are favourable, the drug is then able to be released for use.

The process of drug development is very long and involved, with maybe only one in ten
thousand of the molecules originally tested finally being clinically used. This article
describes different types of drug interactions and discusses the development of the
cytotoxic drug asulacrine from the less cytotoxic amsacrine.

INTRODUCTION

Cancer is a major disease. About one in four people will get it in some form during their
lifetime, and at the present time about one in five of all deaths are due to cancer. Currently
there are three major ways of treating cancer: radiation therapy, surgery and cytotoxic drugs.
All of these have significant limitations, but drugs offer the only approach to treat cases
where the cancer has spread (metastasised) through the body. Other less well established


XII-Biotech-J-Cancer Drugs-1

---

options include drugs that can stimulate the immune system to assist the body itself to fight
the disease, and non-cytotoxic drugs that can prevent cancer cells from multiplying.

This article focuses on the development of drugs to combat cancer. Over the last fifty years
about 500 000 natural and synthetic chemical compounds have been tested for anticancer
activity, but only about 25 of these are in wide use today. This gives an indication of the
difficulty of this problem. Currently drugs are available that significantly reduce the
mortality rates for some cancers (e.g. leukemia and testicular and ovarian cancer), and give
longer overall patient survival times. However, there is a long way to go before truly
curative drugs are available for most cancers. The reason for this is simple: cancer cells are
not foreign to the body but are simply subtly mutated forms of normal human cells, and it is
very different to synthesise drugs that can tell the difference.

The origins of cancer
Cancer is a term which describes a group of perhaps 120 different diseases which share some
broad similarities. In these diseases, a single cell begins to divide uncontrollably forming a
tumour, and eventually bits of this tumour break off and form new tumours (this is known as
metastasis). Normal cells do not divide in this fashion, being kept under tight control by a
number of different biological mechanisms that are still being explored. We do know that
cell division is controlled by a relatively small group of enzymes. Some of these operate to
form a communication network, relaying growth signals from the surface of the cell to its
DNA and telling it when to begin dividing. Others work as a surveillance team, preventing a
cell with damaged DNA from reproducing by first repairing the damage or by instructing it to
die. However, sometimes the damage (mutation) occurs in the DNA that codes for these
enzymes, so that they are themselves defective. Such a cell will divide uncontrollably, and
produce daughter cells that do the same.

A human cell contains approximately 100 000 genes, of which about 50 are known as proto-
oncogenes1. Many of these code for the enzymes that make up the communication and
surveillance systems described above. If a cell accumulates critical mutations in five or six
of these proto-oncogenes, the resulting multiple but subtle changes are likely to result in a
fully malignant cell, capable of forming a tumour.

DESIGNING DRUGS TO COMBAT CANCER

The drugs used to combat cancer belong to one of two broad categories. The first is
cytotoxic (cell killing) drugs and the second is cytostatic (cell stabilising drugs). Both
categories lead to a reduction in the size of the tumour because cancer cells (for various
reasons) have such a high mortality rate that simply preventing them from dividing will lead
to a reduction in the population.

Cytotoxic drugs
Cytotoxic drugs work by interfering with DNA replication. Because cancer cells are rapidly
dividing they are rapidly synthesizing new DNA - and if this is damaged the cell will die.
There are three main groups of molecules that can be used to interfere with DNA replication:

• antimetabolites: molecules that appear to be nucleotides and so are incorporated into
DNA, leading to non-functional DNA.

1Protooncogenes are genes that, if mutated, predispose the cell to becoming a cancer cell.


XII-Biotech-J-Cancer Drugs-2

---

• alkylating agents: molecules that permanently attach to the DNA, distorting its shape.
Unfortunately these also attach to many other molecules in cells.

• DNA-binding agents: molecules that attach to the DNA chain, break it, disengage from
the chain and then attach to another chain to repeat the process. These usually function in
conjunction with an enzyme.

DNA-binding agents are currently the most effective drugs used, but usually a patient is
given a combination of drugs from several of these groups to take advantage of the different
ways in which they work.

None of these drugs are 'cancer cell-specific' — they are all simply 'quickly dividing cell-
specific'. This is the explanation of the side effects associated with chemotherapy (nausea,
immunosuppresion, ulceration and hair loss): the drugs not only attack cancer cells but also
any other quickly dividing cells such as those in bone marrow or the gut. For this reason,
work is currently being done into targeting drugs more specifically for cancer cells by giving
the patient a precursor of the drug (a 'prodrug') which is only activated in the cancerous
tumour. Two possible methods for doing this are outlined below.

Prodrug activation by hypoxic cancer cells
It has been noticed that cells in solid tumours have chronic hypoxia (oxygen deprivation) and
that this property is unique to these cells. In the body, the oxygen needed by cells for
respiration is carried over long distances in the bloodstream complexed with hemoglobin, but
has to get from the bloodstream to cells by diffusion. Because it is consumed by living cells,
oxygen concentration decreases with distance from the nearest blood vessel, and at about
150 m falls essentially to zero. In normal tissue the blood vessel network is so well-
developed that all cells are well supplied with oxygen, but virtually all solid tumours larger
than about 1mm in diameter possess a proportion (usually a few percent) of chronically
hypoxic cells. Much work has recently gone into the concept of developing drugs that are
activated only in hypoxic cells, thus leaving healthy cells intact. Molecules designed as
hypoxia-activated prodrugs have a variety of different chemical structures, and three
examples (compounds 1 - 3) are shown below. Two of these (tirapazamine and EO-9) are
currently in clinical trials.
-
O
O
OH
N
+
N
Br
N
N
NO2
N
NH
+
N
NH
N
2
-
O
CH
OH
OH
O
3



1 2 3
(tirapazamine) (EO-9)

Regardless of their structure, hypoxia-activated prodrugs function by the inactive form of the
drug being converted into the active drug by a mechanism that can only occur in hypoxic
(and therefore cancer) cells. The initial prodrug must be nontoxic and able to diffuse to the
hypoxic cells in a tumour, and is designed so that it can easily be reduced (i.e. pick up an
electron from cellular enzymes). While this reduction will happen in all cells, in oxygenated


XII-Biotech-J-Cancer Drugs-3

---

cells it will be immediately reversed to regenerate the parent prodrug. However, in hypoxic
cells this step is unable to be reversed, allowing the reduced drugs to react further, becoming
cytotoxic.

A specific example of a class of hypoxia-activated prodrugs being developed in the Cancer
Research Laboratory in Auckland, and the mechanism of their function, is given in Figure 1.
This prodrug (4), designated SN (screen number) 25246, contains the N(CH2CH2Cl)2 group
known as a mustard group. This is usually a very cytotoxic alkylating agent, able to
permanently attach to the DNA, distorting its shape and preventing it from separating. In the
prodrug form this mustard is deactivated by the positive charge on the nitrogen (the
reactivity and toxicity of mustards depends on their having an electron-rich nitrogen). This
prodrug is rapidly reduced by enzymes present in all cells, resulting in an intermediate radical
anion (Figure 1). In normal, oxygenated cells this radical anion, once formed, is rapidly re-
oxidised by the free oxygen present to reform the non-toxic prodrug. However, in the
absence of oxygen the radical anion eventually fragments, releasing the mechlorethamine
mustard. This is a cytotoxic drug that has been used in patients, but resulted in side effects.
Such side effects will be lessened if the drug is released only in hypoxic (i.e. cancer) cells.

reduction
O
O
2N
O2N
2N
reoxidation
CH
+
Cl
+
Cl
3
N
N
H3C
H3C
+
N
Cl
Cl
Cl
Cl

4
(SN 25246) futile cycling in radical anion fragmentation in
oxygenated cells hypoxic cells
Figure 1

Gene-Directed Enzyme-Prodrug Therapy (GDEPT)
Alternatively, the prodrug can be activated by an enzyme that has been produced in only the
cancer cells. A gene coding for a non-human enzyme is integrated into a retrovirus2, and the
engineered retrovirus is used to infect cancer cells. Using a retrovirus makes the therapy
dividing cell-specific, but still not cancer-cell specific. However, the treatment can be made
cancer cell-specific by using a gene that is only coded for if activating proteins that are much
more abundant in cancer cells than in healthy cells are used. The enzyme is then produced in
these cells, meaning that the prodrug is activated only in cancer cells.

One disease for which this therapy could be useful is colon cancer. Compound 6 (5-
fluorourasil) is currently widely used in the treatment of colon tumours, but has serious side
effects. Cells in colon tumours over-express the gene coding for a particular transcriptional
protein. A potential colon cancer therapy could involve inserting a gene coding for the

2A retrovirus is a virus that is able to incorporate its DNA into that of dividing cells.


XII-Biotech-J-Cancer Drugs-4

---

bacterial enzyme cytosine deaminase into rapidly dividing cells. This gene is only
transcribed in cells with a high level of the protein which is over-expressed in colon cancer
cells, so cytosine deaminase is only produced in these cells. The patient is then given the 5-
fluorourasil prodrug 5-fluorocytosine (5), which is only converted by cytosine aminase to the
cytotoxic 5-fluorourasil in tumourous cells (Figure 2).
NH2
O
F
F
N
cytosine
HN
deaminase
O
NH
O
NH

5 6
(5-fluorocytidine) (5-fluorourasil)
Figure 2 - Enzymatic conversion of prodrug to active drug in GDEPT

If the foreign enzyme can be generated selectively in this way, the non-toxic prodrug will
then only be converted to the cytotoxic drug in the target cells. In addition, because the
active drug is released catalytically by the enzyme, and can be designed to diffuse short
distances from where it is released, only a few percent of the target cells need to express the
enzyme for the tumour to be completely destroyed. Many combinations of enzymes and
prodrugs are currently under study for GDEPT. A project in the Cancer Research Laboratory
in Auckland is concerned with the development of prodrugs able to be activated specifically
by a bacterial nitroreductase enzyme to generate cytotoxic nitrogen mustards. The
dinitrophenyl-carboxamide mustard (7) has been shown to be a suitable prodrug. It is
essentially non-toxic because the three electron-withdrawing groups on the molecule remove
electrons from the mustard nitrogen. It is activated by the bacterial nitroreductase (but not by
cellular enzymes), via reduction of the 2-nitro group to give 8, which is much more reactive
(Figure 3).

NO2
NHOH
CNH2
CNH2
O
O
nitroreductase
O2N
O2N
N
N
Cl
Cl
Cl
Cl

7 8
Figure 3 - Activation of a nitrogen mustard by nitroreductase

Cytostatic drugs
While the prodrug approach is being developed to try and improve the specificity of the
common "cytotoxic" anticancer drugs, another new approach is to use "cytostatic" drugs.
Many of these specifically target the altered biochemical pathways that enable cancer cells to
reproduce quickly. These drugs are designed to deactivate the altered enzymes that result
from changes in the oncogene involved. These drugs are not designed to kill the cell
involved, but simply to prevent it from reproducing. However, because cancer cells have


XII-Biotech-J-Cancer Drugs-5

---

very high death rates, due largely to the multiple mutations they possess, simply preventing
reproduction is expected to lead to a reduction in tumour size. These drugs are theoretically
'cancer cell-specific' in that they target processes occurring only in cancer cells.

One of the targets for these drugs are the enzymes making up the communication networks
by which growth signals instructing a cell to divide are transmitted form the outside of a cell
to the nucleus. The growth signal pathways begin at the cell surface, with receptor enzymes
that protrude through the cell membrane. Polypeptide growth factors bind (attach) to these,
and this binding induces the receptor enzyme to change shape. This shape change turns on
the enzyme, allowing it to phosphorylate (add a phosphate group to) tyrosine amino acids in
other enzymes (ligands) that associate with its internal sections. This tyrosine
phosphorylation activates the ligand, which then transmits the growth message to other
enzymes in the pathway.

One of the most important of these receptor enzymes is called the epidermal growth factor
receptor (EGFR). This does not usually occur in normal cells, but is present on the surface of
a large proportion of human tumour cells. EGFR has a binding site for the ligand that it
phosphorylates (to pass on the signal), and one for the cofactor ATP (which donates the
phosphate). The role of the EGFR enzymes is really to bring these two components together.
There is a great deal of interest in drugs that can selectively block the tyrosine kinase activity
of EGFR (but not of similar, useful enzymes such as the insulin receptor). In theory this
could be achieved by drugs which bind either at the ATP site, or at the ligand site.

Because most enzymes have sites of similar structure, where the ATP binds, it was felt until
recently that drugs that worked by binding at this site would not be very selective. However,
recent work in the Cancer Research Laboratory (in conjunction with the Parke-Davis
Company) has discovered a novel class of drugs, the anilinoquinazolines, which inhibit
EGFR by binding at the ATP site. The lead molecule, 4-anilinoquinazoline (9), was itself a
quite potent inhibitor, but it was found that changes at three positions on the molecule
increased the activity about a million-fold. Thus the related molecule (10) can inhibit the
enzyme at a concentration of 25 pM (0.000025 M). The related tricyclic molecule 11 is
even more potent, inhibiting at a concentration of 8 pM, i.e. about 2 mg dissolved in an
Olympic-sized swimming pool. More detailed studies with 10 also show that it is a very
selective drug. It inhibits the EGFR enzyme about a million-fold more effectively than it
inhibits related tyrosine kinase enzymes. This work has shown for the first time that the
extremely potent and selective capacity of the EGFR enzyme can be developed.
H
H
H
N
N
Br
N
Br
MeO
N
N
N
N
N
MeO
N
N
N

9 10 11



XII-Biotech-J-Cancer Drugs-6

---

THE DEVELOPMENT OF THE CYTOTOXIC DRUG ASULACRINE
AN EXAMPLE OF PHARMACEUTICAL DEVELOPMENT

The development of a new pharmaceutical is a complex process, but can be broken down to
three main steps:
• Discovery of a new potentially useful molecule.
• Appropriate molecular modification to produce a molecule with the best combination of
properties.
• Development of this molecule into a safe and affordable drug.

This process is outlined below for the anticancer drug CI-921 (asulacrine), which was
discovered and mainly developed by the Cancer Research Laboratory (CRL) in Auckland.
The CRL was established in 1956 for the discovery and development of drugs useful for the
treatment of cancer. The drug asulacrine was developed using funding from the Auckland
Division of the Cancer Society of New Zealand, the Health Research Council of New
Zealand and Warner-Lambert/Parke-Davis (an American pharmaceutical company).

Step 1 - Initial Discovery
The first phase, the discovery of new classes of active compounds, is especially difficult in
the case of anticancer drugs because (at least until recently) there were few identifiable
targets to aim for. For this reason, many of the drugs used today were discovered from the
random testing of compounds isolated from natural sources or made for other purposes. Not
surprisingly, such an approach gives a very poor return, with less than one compound in ten
thousand proving even slightly useful. The early tests for usefulness used human or animal
tumour cells grown in culture. These measure the cell killing ability (cytotoxicity) of a
molecule.

However, to measure how specific a given drug is for cancer cells it is necessary to see
whether it will work against tumours in an animal model (usually mice). Such tests are
expensive to carry out and can sometimes be misleading, since human and mouse tumours
are quite different. By contrast, antibacterial drugs can be evaluated in a test tube against the
very same bacterium which they are designed to kill in humans. However, limited animal
testing is necessary to ensure that new drugs are useful and reasonably safe before being used
in human patients.

Work in the CRL in the 1970s on drugs that bound reversibly to DNA resulted in the class of
molecules known as the acridinylaminomethanesulfonanilides (AMSA compounds). One
member of this class (amsacrine, 12), after reciving advanced testing both in the CRL and by
the American National Cancer Institute, was approved for human trials in 1978. After two
years of clinical testing in America, Europe and Australasia (including in Auckland),
amsacrine was approved for hospital use. It is particularly useful against leukemias, and is
now one of the drugs of choice thoughout the world for the treatment of this disease. In 1984
it became the first pharmaceutical developed in New Zealand to be registered for use in this
country. Amsacrine is produced and marketed by the Warner-Lambert/Parke-Davis
pharmaceutical company, with which the CRL has a close association.

Amsacrine itself is useful (in combination with other drugs) in the treatment of leukemia, but
has little activity against the more common solid tumours. However, it is a novel molecule,
working by causing DNA breaks in conjunction with an enzyme called typoisomerase II. It
was thought that by changing its structure in certain ways a new and better drug might be


XII-Biotech-J-Cancer Drugs-7

---

developed. If the drug properties important for activity could be determined, other molecules
of this general structure but which were active against a wider range of tumours might be
found.
H O
CH3O
N
CH3
S
H
O
N
N

12
Amsacrine

Step 2 - Molecular modification of a known compound
It was known that the biological effects of the AMSA molecules was greatly affected by
substituents placed on the parent structure. For example, the two compounds 13 and 14 differ
in their biological effectiveness (potency) by more than 50 000-fold.
NH2
H O
N
CN
S
H
O
H
N
N
N
N


13
14

The goal of the drug development work, begun in the early 1980s, was to obtain analogues
that retained potency and effectiveness against leukemia, but also had a wider range of
activity against solid tumours. The next step was to make compounds of a similar structure
to amsacrine, and test them for the physical and biological properties. Their ability to bind to
DNA, their ability to kill leukemia and lung cells in culture, their effects on leukemia and
lung cancer cells in culture and their effects on leukemias and lung tumours in mice were
evaluated. The lung tumour test was considered more relevant for selecting compounds with
possible activity against human solid tumours. Most of the drugs already in clinical use
against solid tumours had some activity in this mouse lung tumour, but amsacrine (not
effective against human lung cancer) did not.

How were the target molecules to be selected? Modifications could not be usefully made
randomly: the science of organic chemistry is too vast for that to be practical. Even if only
20 different substituents were used, placed two at a time at available positions of the
amsacrine structure, more than 109 different possibilities would result. Instead, the molecule
has to be looked at as an entity, and the effect that an attached substituent will have must be
predicted as far as possible. For example, if the 4-methyl analogue is made (15), the added


XII-Biotech-J-Cancer Drugs-8

---

methyl group will alter the overall solubility of the molecule and the base strength of the
acridine nitrogen. Its steric bulk will affect the ability of the molecule to bind to DNA, and it
may provide a new site for metabolic breakdown. All of these factors, and others, will
influence the biological activity of 15 in different ways, but we will see only the overall
result. Rational design requires an appreciation of the molecular properties that might
improve biological activity, and a knowledge of how the addition of substituents to a
molecule will contribute to all these properties. This is the science of molecular design.
H O
CH3O
N
CH
H
3
S
O
CH3O
N
CH3
H
O
S
N
H
O
N
N
CH3
CH3
N
O
N
CH3
H


15
16

Previous work had shown that high antitumour activity was generally associated with strong
binding to DNA, and also with slow binding (where the drug attaches to one particular site
for a long time, rather than coming on and off very quickly). It was also thought that the drug
would be able to diffuse throughout the body better if it were only weakly basic, because this
would mean that more of the drug was in the neutral (uncharged) form. Finally, for ease of
administration a drug somewhat more water-soluble than amsacrine would be very desirable.
Modifications to the acridine portion of amsacrine were sought that would achieve all these
effects, and in so doing enhance its antitumour activity and usefulness.

Considering accumulated experience about how these drugs interacted with DNA,
modifications concentrated on 3-, 4- and 5-substitution patterns. A large number (upward of
200) compounds were made and evaluated for both the above desired physical properties and
for antitumour activity in both the leukemia and lung tumour models. Several analogues
showed excellent activity against lung tumours, and three of these were sent for additional
testing in overseas laboratories. On the basis of all the results, the 4-methyl-5-
methylcarboxamide (16, CI-921, asulacrine) was chosen for clinical trial.

A comparison of a number of important physical properties of asulacrine (16) with those of
the parent amsacrine (12) shows that these modifications have largely achieved the specific
goals (Table 1). Placing two substituents, one of them the electron-withdrawing
carboxamide, adjacent to the acridine nitrogen has lowered the basicity by over one pK unit.
At the normal body pH of 7.2 this means that 69% of the asulacrine is in the diffusable
neutral form compared with only 12% of the amsacrine. At the same time, DNA binding has
increased 16-fold, due mainly to the carboxamide (placed on the molecule so that it can make
specific binding contact in the minor groove of the DNA). The average residence time of the
molecule on the DNA has also been increased 20-fold, and the water solubility is improved.


XII-Biotech-J-Cancer Drugs-9

---

However, these changes have not significantly affected the oxidizability of the aniline portion
of the molecule, as shown by the similar oxidation potentials.

Table 1 - A comparison of the physicochemical properties of amsacrine and asulacrine




amsacrine
asulacrine



Solubility of hydrochloride salt in water / mg mL-1
0.12
0.71



Strength of DNA binding / log K
5.11
0.32



Average residence time at a DNA site / milliseconds
5.5
108



% ionised (i.e. protonated) at pH 7.2
88
33



Oxidation potential / V (c.f. SCE)
0.282
0.243



Basicity of acridine nitrogen / pKa)
7.93
6.90

It is reasonable to suppose that these alterations in physical properties have contributed to the
greatly improved experimental antitumour activity of asulacrine compared to amsacrine
(Table 2). While the potency of both drugs is similar, asulacrine shows a significant
proportion of complete cure in animals bearing either leukemia or lung tumours, whereas
amsacrine under the same conditions achieves no cures.

Table 2 - A comparison of anticancer activity of amsacrine and asulacrine in mice




Leukemia
Lung tumour




amsacrine
asulacrine
amsacrine
asulacrine





Dose of drug for best
13.3
20
13.3
20
effect / mg kg-1)





Average % of animals
0
50
0
82
cured

Step 3 - Development to a useable pharmaceutical
The above improved physical properties and biological activity of asulacrine compared to
amsacrine resulted in the drug being selected as a second-generation analogue of amsacrine
for trial in human patients. This meant very extensive toxicology studies, and the production
of large quantities of the drug. Such work is very time-consuming and expensive, and mostly
beyond the scope of the CRL. A commercial partner was therefore needed. However, before
a commercial company would agree to spending the many millions of dollars necessary to
develop the drug for human trials, it was necessary to obtain a patent on it. A patent is
granted if, in the view of the examiners, an invention is novel, not obvious from existing
knowledge, and useful. Once a patent is granted in a particular country only the organisation
to which the patent is assigned can make or market the product for a period of 17 years. In
the case of asulacrine, patents were successfully obtained in a number of countries, and
further development was then undertaken by the Warner-Lambert/Parke-Davis Company.


XII-Biotech-J-Cancer Drugs-10

---