Pill power: Gleevec's success has boosted hopes for the molecular approach to cancer therapy. Credit: NOVARTIS

Every few years a claim is made for a 'miracle drug' that will cure cancer. But time after time, compounds that have performed wonders in mice have failed miserably when faced with clinical reality.

Even cynics, however, have been taken aback by the performance of a drug called Gleevec, produced by the Swiss company Novartis, and approved by the US Food and Drug Administration (FDA) last year. It is not a cure-all, by any means. But against two particular types of cancer, Gleevec has achieved unprecedented results. For cancer researchers, the drug's remarkable success confirms that they are on the right track: understand which genes go wrong in cancer, design therapeutics to correct these defects, and the disease can be beaten.

“Gleevec is the proof of principle that the molecular strategy works,” says Richard Klausner, until last September director of the National Cancer Institute (NCI) in Bethesda, Maryland, which funds most of the fundamental research into the disease carried out in the United States (see 'Box 1 Survivor at the helm').

Slow progress: US mortality rates for the major cancer killers have changed little in 10 years.

That proof is long overdue. Conventional cancer chemotherapy, after decades of fine-tuning since it was introduced in the 1950s, has turned around the dismal outlook for childhood leukaemias — up to 90% of cases are now curable. But against the big cancer killers — including breast, lung, prostate and colon cancers — there has been little progress (see figure); sufferers usually experience only a brief period of remission. Even then the price is high, because the drugs are so toxic. Most current chemotherapy agents target dividing cells, for example by blocking the synthesis of new DNA required for cell division — and so hit many healthy organs as well as tumours. In particular, they damage bone marrow, where blood cells are produced.

With more than $46 billion spent on cancer research by the US federal government alone since President Richard Nixon launched his 'war on cancer' in 1971, a minority of experts has even begun to suggest that cancer has become science's Vietnam. In a cutting essay in the February issue of Prospect magazine, for instance, cancer surgeon Michael Baum of University College London claimed that the fight against the disease was bogged down by “slavish adherence to outdated paradigms”.

Until Gleevec, promises of kinder and more effective therapies had proved empty. But 95% of cases of chronic myelogenous leukaemia (CML) respond to the drug, with the cancer being completely eliminated half of the time1,2. “It was extraordinary for clinicians to see such rapid and dramatic results,” says Brian Druker, director of the Leukemia Center at the Oregon Health and Science University Cancer Institute in Portland, who conducted many of the clinical trials. These stunning results, combined with a growing realization among cancer researchers that they have to start delivering the goods, have altered the outlook. The current catchphrase is 'translational research' — aiming to convert molecular insights into effective drugs.

Nixon launched his war on cancer as a successor to President John F. Kennedy's 1960s dream of putting a man on the Moon. But according to some experts, the goal of defeating cancer by 1980 was always unrealistic. “We now understand that cancer is not a simple target that can be approached with high-tech hardware alone,” says Klausner. Cancer, it is now realized, is a wily, shifting target — a battery of many different diseases, with a range of underlying causes.

In the vast majority of cases, cancer is an acquired genetic disease. Cells accumulate a series of mutations that allow them to escape, with ever-greater freedom, the body's normal constraints on their proliferation. “But even mutations promoting unchecked growth are not actually enough to fire a fully fledged cancer,” says Douglas Hanahan, a biochemist at the University of California, San Francisco.

Robert Weinberg believes cancer's complex molecular interactions will give way to simplicity.

In a recent review article3, Robert Weinberg of the Whitehead Institute for Biomedical Research in Cambridge, Massachusetts, together with Hanahan, described six 'hallmarks' of cancer — the acquired characteristics needed to turn a few wayward cells into an aggressive tumour.

Marks of malignancy

First, the cells must mutate so that they can dodge the cellular signals that suppress growth. Then they must acquire their own growth-signalling pathways, independent of external signals. Cells must also evade apoptosis, the system of programmed death under which abnormal cells trigger their own destruction. And they have to develop limitless potential to proliferate: normal cells can divide only about 70 times before their telomeres — the protective caps at the end of chromosomes — become so shortened that the chromosomes are damaged and the cell dies. But cancer cells exploit an enzyme called telomerase to rebuild their telomeres and so escape this constraint.

The other two cancer hallmarks apply only to solid tumours. Growing tumours must create their own networks of blood vessels to deliver the food and oxygen they need. This complex affair, known as angiogenesis, requires a multitude of special growth factors. Finally, the most dangerous tumours are those that have developed mechanisms to allow cells to detach from the main tumour and enter the bloodstream or the lymphatic system. From here they can reach distant tissues, where they grow into secondary tumours, or metastases. Nine out of ten cancer deaths result from metastases.

Very few cancers are caused by a single mutation. “There are so many things that need to go wrong, so it is not surprising that, in a lifetime, cancer is actually rare,” says Weinberg. This is why tumours caught early are usually easier to treat: mutations tend to accumulate as the cancer progresses. The fact that childhood leukaemias are caused by relatively few mutations also helps to explain why they have proved amenable to chemotherapy.

But the multiple changes needed to initiate and support a solid tumour offer a wide potential source of specific targets for scientists trying to develop anticancer drugs. First to be studied were the proteins produced by oncogenes — genes that, when activated, promote cell growth and division. Many, such as ras, which in 1982 became the first human oncogene to be cloned4,5,6, do this by stimulating signalling pathways normally activated by growth factors such as epidermal growth factor (EGF). In about 30% of human cancers, ras is mutated so that it is permanently switched on, providing a constant growth signal to the cell7.

Unfortunately, the discovery of ras did not translate into a major clinical advance. Inhibitors targeting the ras system were developed by several pharmaceutical companies, but the first generation did not perform well in clinical trials. “We were too excited,” admits Mariano Barbacid, director of the Spanish National Cancer Center (see 'Box 2 The real deal in Madrid'), who led one of the three groups that independently cloned the gene. “Some people had thought we had opened the door to curing cancer.”

The proteins produced by tumour suppressor genes, meanwhile, normally prevent cancerous growth. If the genes are damaged or lost, cells are more likely to become cancerous. The p53 and Rb tumour suppressors, for instance, are inactivated in most tumours8. But proteins themselves do not make good drugs — they are hard to administer, and tend to get broken down in the body — so scientists are now trying to block key proteins in the molecular signalling pathways given free rein when tumour suppressor genes are inactive.

Many genes associated with cancer interact with several signalling pathways. For example, p53 promotes apoptosis and activates DNA repair — and probably also inhibits angiogenesis9. Many oncogenes encode enzymes called tyrosine kinases, which add a phosphate group to a protein in a molecular pathway as a means of propagating the signal. Again, this activity can operate in several different signalling pathways.

Initially, drug development lagged behind the explosion of studies of cancer genes and the pathways that they influence. “The flood of identified oncogenes was very interesting,” says Barbacid. “But we got to a point where it was just another oncogene, another kinase, another pathway — and it was definitely time to put the knowledge in the service of the patient.”

When drug companies moved in, they were particularly interested in cancer genes encoding tyrosine kinases, because the pharmaceutical industry has decades of experience in finding small molecules to block specific enzyme targets. Dozens of tyrosine kinase inhibitors are now being tested in the clinic, and Gleevec is the first to make it through.

Toxic twist

Fresh ammunition: could drugs targeted at specific molecular pathways provide a weapon against the 'big' killers such as breast (left) and prostate cancer? Credit: O. BRAWLEY/NCI; NCI

Gleevec was originally designed as a specific inhibitor for the platelet-derived growth factor receptor, which acts as a tyrosine kinase and has been implicated in some cancers. But the compound was later found to interact with proteins produced by two other oncogenes, BCR–ABL and c-KIT. The former — a mutation caused by the fusion of sequences in the ABL and BCR genes — is the trigger for CML10. It causes a recognizable chromosomal defect called the Philadelphia chromosome, in which chromosome 22 is shorter than normal.

Initially, pharmacologists were worried that Gleevec would have dangerous side effects. They were particularly concerned about its interaction with c-KIT, which, when functioning normally, is involved in regulating the immune system. “The fact is that Gleevec should have been toxic,” says Barbacid, “but it wasn't.” No one understands exactly how, but the immune system seems to be able to switch on other pathways to compensate for the blocking of the tyrosine kinase encoded by c-KIT.

Since its approval by the FDA in May 2001, positive clinical results1,2 have led to Gleevec's licence being extended in February this year to a rare stomach cancer, gastrointestinal stromal tumour (GIST) — a disease that is caused by the mutation of c-KIT (ref. 11).

Because they are caused by single mutations, CML and GIST are 'easy' diseases. But tyrosine kinase inhibitors are also being tested in messier situations — in cancers where many mutations have accrued. Even Gleevec does not perform quite so well in these circumstances. In a later and frequently fatal stage of CML called blast crisis, where many mutations have appeared, only around two-thirds of patients respond to the treatment1. Resistance to the drug also develops quite quickly as BCR–ABL mutated its way free of Gleevec's effects, or was overexpressed.

But at least one other tyrosine kinase inhibitor has performed well in early clinical trials against a notoriously difficult cancer. AstraZeneca's Iressa, directed against the tyrosine kinase activity of the EGF receptor, has excited oncologists by prompting a 10% response rate in patients with non-small-cell lung cancer who had failed to respond to other therapy. Although this might sound like a low response, the outlook for these patients is usually extremely bleak. “This is miracle-like,” says Charles Sawyers of the Jonsson Comprehensive Cancer Center at the University of California, Los Angeles, who ran some of the Gleevec trials.

Road blocks

Other classes of drug have so far not performed as well, although efforts are continuing. A few years back, for instance, inhibiting angiogenesis by blocking the action of proteins such as vascular endothelial growth factor was thought to hold great promise. One front-page story in The New York Times, published in May 1998, infamously touted angiogenesis inhibitors as the long-sought cancer 'cure'. But clinical trials of the drugs have so far disappointed — cancer cells seem to find it easy to get around the blockage of one angiogenic pathway.

There is also less excitement now about the potential for telomerase inhibitors, mostly because of concerns about their toxicity to stem cells in bone marrow, which also require the enzyme. And attempts to disrupt metastasis — using drugs called matrix metalloproteinase inhibitors to block an enzyme used by cells to chew their way out of the extracellular matrix that usually keeps them in place — have so far disappointed in the clinic12.

Given previous experience, are researchers getting too excited about Gleevec? No, argue enthusiasts for the molecular approach. First, Gleevec is performing well where it matters — in the clinic. Second, no one is pinning their hopes on one drug alone. Gleevec is merely the proof of principle that you can block a pathway that cancer cells depend on. In most cancers, it may be necessary to block pathways at several points, or even to target several pathways. So, repeating the success of Gleevec in other cancers may require cocktails of drugs — perhaps including some of those, such as the angiogenesis inhibitors, that have performed poorly in isolation.

“It is naive to think that in solid tumours we will get dramatic results by targeting one gene,” says José Baselga, a clinical pharmacologist at the Vall d'Hebron University Hospital in Barcelona, who is organizing 58 clinical trials of tyrosine kinase inhibitors. “There will be many mutations, and so in future we can reckon on using combinations of drugs to hit many targets.”

“This is what happened with AIDS treatments,” agrees Edward Sausville, associate director of the NCI's developmental therapeutics programme. “Individual drugs did little on their own — but they worked very powerfully when they were put together.”

Specific drug combinations will probably need to be tailored to particular tumour types — which is why efforts to profile individual tumours to find out what, exactly, has gone wrong at the molecular level form an important new front in the war on cancer. “The first step is to understand the molecular profiles of cancers — then we'll need to identify targets,” says Robert Strausberg, head of the Cancer Genome Anatomy Project at the NCI.

Biological bull's-eyes

In the past few years, huge investments have been made in new genomic technologies to do just this. Strausberg's project is the largest of many attempts to build up databases storing information on gene and protein expression, gene mutation and silencing, and clinical information such as the rate of disease progression and response to drugs. According to Sausville, these projects will lead to the recognition that types of cancer currently viewed by pathologists as identical are in fact distinct molecular diseases. “The most important thing we will learn is how to classify cancers better,” he says.

Two recent papers, using DNA microarrays to look at gene expression in a lymph-node cancer called diffuse large B-cell lymphoma13 and in breast cancer14 have illustrated the point: within each disease, cancers can be grouped into subsets with distinctive gene-expression profiles that correlate with how fast the disease progresses. Sawyers also suspects that the 10% response to Iressa seen in the lung cancer trials indicates that the tumours that responded had a distinct molecular profile. “My hunch is that the EGF signalling pathway is the driving force in 10% of lung cancers, and the other 90% were different diseases,” he says.

Molecular profiling should also allow scientists to identify markers for early diagnosis of cancer — proteins that could be measured in blood, faeces, urine or even in shed skin. “Early diagnosis is very important because the cure rate is very high for early-stage disease and very low for late-stage disease,” says Lee Hartwell, director of the Fred Hutchinson Cancer Research Center in Seattle, who last year shared the medicine Nobel for his work on the cell cycle of growth and division.

Bruce Ponder, head of the University of Cambridge's Department of Oncology, and co-director of a new cancer centre being established at the university by the Medical Research Council and the charity Cancer Research UK (see 'Box 3 Two into one'), hopes to find molecular indicators of whether new drugs are working in clinical trials. “At the moment, tumour shrinkage is used as an endpoint and this is not only crude, but also delayed,” he says.

Molecular profiling might also help clinicians to use currently available drugs more efficiently. “By comparing accumulating data on molecular profiles with outcomes of clinical treatment, we'll be able to identify, for example, which patients will not benefit from chemotherapy after surgery and spare them from useless, and very unpleasant, treatment,” says Ponder.

Informed choice

In parallel with gathering data on tumours, some scientists are profiling cell lines used to study cancer in the lab. John Weinstein, a molecular pharmacologist at the NCI, is analysing gene and protein expression in the NCI's 60 standard cancer cell lines. Over the years, these cells have been used to test more than 70,000 different drugs. “There is a mine of pharmacological information there which we will correlate with the changes in gene and protein expression, to help us work out what types of drug will work best in a cancer with a particular expression profile,” Weinstein says.

Already, Weinstein's research has turned up interesting pharmacological insights. For example, the enzyme L-asparaginase is used to treat acute lymphoblastic leukaemia (ALL) because it destroys the amino acid L-asparagine in the blood. This works because ALL cells are unable to make their own supply of the amino acid, so once their access to it in the blood is blocked, they die. Weinstein has found that ovarian cancer cell lines have similar patterns of gene expression to ALL cells, which suggests that they might also be sensitive to L-asparaginase15.

Molecular profiling will yield masses of data to add to our existing understanding of the signalling pathways that influence cancer. But from this complexity, leading researchers are convinced that simple insights will emerge. “Cancer biology and treatment...will become a science with a conceptual structure and logical coherence that rivals that of chemistry or physics,” Hanahan and Weinberg argued in their 2000 review3. They claimed that within two decades, cell biologists will have derived a complete integrated circuit of the cell's signalling pathways, allowing us to model how specific genetic perturbations cause cancer, and to predict how to correct the problem using drugs acting on key points in the circuit.

Other experts agree that marrying this 'systems biology' approach with weapons such as Gleevec holds great promise — not of defeating cancer within a few years, as Nixon once promised, but hopefully of seeing real progress in the ongoing war on cancer over the next decade or two. “History shows that therapy comes when you understand the system of the disease,” says Bert Vogelstein, a leading cancer researcher at Johns Hopkins University in Baltimore, Maryland.