Cancer researcher calls for trial process changes

Fewer animal studies, more human trials

Current methods for studying new cancer drugs are too inefficient and costly, resulting in unnecessary delays in the development of new therapies, a noted cancer researcher told colleagues at recent international meeting of surgeons in Australia.

During a presentation at the 2004 Annual Scientific Congress of the Royal Australasian College of Surgeons, held May 2-7 in Melbourne, Australia, Jonathan Lewis, MD, PhD, a former cancer researcher at Memorial Sloan-Kettering Cancer Center in New York City and the New Haven, CT-based Yale University School of Medicine, called for changes in oncology clinical trial design.

"The methods by which we clinically develop drug candidates are dismally out of step with the tremendous progress in basic science and our ability to discover promising new compounds," Lewis told attendees. "The issues are cost and time. The current approach results in clinical strategies that prolong the time it takes to get to a possible yes’ answer. Instead, we need to shorten the time it takes to produce a no.’ The ability to quickly rule out unpromising compounds or ill-conceived indications could cut 50% or more off the typical time and money devoted to cancer drug development."

According to Lewis, research has shown that studies indicating a compound’s efficacy in treating cancer in animals do not reliably indicate how well humans will respond. Yet, researchers still waste valuable time and money conducting many laboratory studies before attempting to move into early clinical trials.

"The current way of having a very rigorous process of step-by-step of cancer drug development is scientifically correct — we have no argument with that," Lewis says "I think, though, what has become very clear is that the use of the lab to understand how the drugs are going to work in humans has been shown in cancer to be worth very little. Animal models are much less predictive of what is going to happen in humans."

Laboratory research should be focused on determining whether a particular drug is too toxic for use in people. Once those questions are answered, however, he believes smaller, focused human trials will yield better information.

"The main issue, in the lab, is to make sure of the absence of any major toxicity. You have to be very sure of that in the research lab," Lewis explains. "But I think worrying about efficacy and so on, we put too much emphasis on that. If we have something that is valid, biologically, getting it into human studies sooner makes a lot of sense."

Once a drug is moved from the laboratory to clinical trials, scientists should use genotyping to identify certain people who would be expected to benefit from treatment, allowing investigators to conduct faster, smaller early trials that would yield more specific information.

Laboratory scientists currently do this with compounds by studying them in homogenous populations of laboratory mice, Lewis points out.

Genetics role 

Research is increasingly showing that genetics plays a large role in predicting how patients will respond to different treatments. Patients with the same type of cancer will often respond very differently to the same treatment, he says, and genetics are a large part of the reason why.

For example, researchers at Dana Farber Cancer Institute and Harvard Medical School1 recently discovered that certain patients with non-small cell lung cancer (NSCLC) respond dramatically to the drug gefitinib (Iressa). These patients’ cancers harbor a malfunctioning version of a protein known as EGFR. Researchers in Japan have reached similar conclusions. Patients with NSCLC whose tumors do not contain the gene mutation that produces the altered protein do not respond as well.

The Dana Farber researchers are trying a new approach to studying cancer therapies. Scientists scan the DNA of cancer cells for mutated genes that instruct cells to produce abnormal versions of growth proteins known as tyrosine kinases. The theory is that drugs known to block such proteins could slow tumor grown while leaving normal cells intact.

To test the theory, researchers analyzed tumor samples from patients who had been successfully treated with Iressa and four patients whose tumors did not respond. All of the responders had the mutation; none of the nonresponders did.

In the future, scientists should be able to use such genotyping to target which patients are most likely to respond to an experimental treatment and then study the drug in these patients first.

Lewis is the founder and chief executive of a new company, ZioPharm Inc., which plans to develop and commercialize anticancer compounds. ZioPharm intends to embrace this new approach, he says.

"Of course, there is no question that the gold standard, prior to [a drug’s] approval, is and has to be the large, randomized Phase III clinical trials," Lewis says. "But if one moves to those too quickly, it proves to be inefficient. What we are going to be doing, is trying very hard to get a more homogenous group of patients, first."

Lewis contends that too many ineffective drugs get moved too quickly from Phase II to Phase III studies, which are larger and very costly to conduct. When a drug shows little or no promise in Phase III trials it is often abandoned, when, if problems had been discovered earlier during smaller studies, the compound could either be revised or abandoned early.

"We are seeing too many bad therapies going too far," he notes. "We do see things fail after several trials. At the same time, there is often an emotional thing in clinical research that goes on that keeps pushing things forward at tremendous cost and effort, only to fail in the end."

Such a system increases the costs of research and exposes many subjects, unnecessarily, to drugs that won’t help them, he contends.

Statistics from the U.S. Department of Health and Human Services indicate the low productivity of cancer research dollars, Lewis says. The statistics show a 5.3% drop in cancer mortality between the years 1995 to 2002, but heart disease mortality dropped 14.4% during the same time period.

Not as simple as it looks

Changes do need to be made in the clinical trial process for cancer treatment, agrees Mark J. Ratain, MD, a specialist in hematology and oncology and chairman of the committee on clinical pharmacology at the University of Chicago Hospitals. And too many drugs get to Phase III trials.

However, Lewis’ proposals have some flaws, he notes.

First, while animal efficacy studies may not be very beneficial for small-molecule drugs, the same cannot be said for other therapies, he states. He would want to see some demonstration in animal models that the drug would be active in a living being before moving into human trials.

"Just because it works in cell culture doesn’t mean it is going to work in humans; but, then, I don’t know what he specifically he is proposing," he says. "And there are too many drugs in clinical trials now to say that we ought to have more drugs move to the clinic faster."

It is important to keep the Dana Farber research in perspective, he adds. The scientists there were studying the effects of Iressa in patients after the drug had already been approved by the U.S. Food and Drug Administration, and the researchers knew ahead of time which patients had responded and which had not responded to treatment.

It would be much more difficult for scientists to scan the DNA of prospective trial participants before they really knew what they were looking for, he adds. "That is great if you know the answer before you start. I didn’t hear anyone saying we should look for EGFR somatic mutations before we give patients Iressa. That correlation was discovered after the drug was approved. If we knew the answer before we started, we would have done it that way."

Operating on assumption

Even if you suspect there are certain patients who will respond to the drug better, you are still dealing with a hypothesis that must be tested. And in oncology drug development, history has shown that many drugs do not work the way they are initially expected to.

Researchers often proceed using many assumptions about what a particular drug will target and how it will work, only to have later clinical trials show the drug works, but against other targets — either instead of or in addition to its intended target — and in different ways, Ratain says.

For example, drug developers initially did not expect Cisplatin to be effective against testicular cancer, and the drug was nearly abandoned because of the high rates of renal failure among trial subjects taking the drug. However, a patient with advanced testicular cancer was included in a Phase I trial and unexpectedly responded well to treatment.

Several other drugs have been shown to work in ways that scientists did not originally expect, and in populations they did not expect.

"I don’t think we know anywhere near as much as we’d like to think we know, and there are so many assumptions that are being made that people accept without thinking about," he says. "Assumptions can range from targets, to dosing, to paradigms, to statistics. There is a lot of hand waving out there that is costing a lot of companies a lot of money."

Ratain says pharmaceutical companies and other drug developers should be more reluctant to take compounds to Phase III trials before there is solid evidence that the drug works.

However, he contends that more people, with different conditions, should be included in Phase II trials, in order to more quickly determine how well the drug will work.

"Sometimes you just get lucky. That has been the history of cancer drug development," he continues. "There are few examples where the research was carefully planned and the careful plan worked. There have been numerous examples where it was carefully planned and it didn’t. I think your chances of being right are often no better than your chance of just being lucky. I would say cast the net far and wide on top of your data and follow your leads and follow your gut."


1. Lynch TJ, Bell DW, Sordella R, et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small cell lung cancer to gefitinib. N Engl J Med. Published on-line April 29, 2004. In press. To be published in the May 20 print edition of the journal.