Using biomarkers to enrich studies too early in the drug-development process, especially for cancer drugs, can prematurely narrow the patient population and lead to drug failure, cautioned Mark Ratain, professor of medicine at the University of Chicago.
“Should biomarkers be used for eligibility criteria in Phase I/II studies? My answer is, ‘No,’” Ratain said during a presentation at the American Association for Cancer Research’s annual meeting in Los Angeles last month.
In an effort to cut costs and attrition rates, drug makers are attracted to the idea of enriching Phase I/II trials clinical trials with patients with biomarkers linked to drug response before starting large-scale Phase III trials. However, the use of biomarkers in the earlier trials also requires sponsors to know a significant amount about the drug and the biomarker early in drug development.
“The downside is that you have to be right twice,” Ratain said. “The drug has to have activity and the biomarker has to be predictive.”
This downside presents an additional risk to drug makers at a time when they are challenged by soaring drug-development costs and high attrition rates in late-stage drug development.
“In this game of onco-roulette you don’t really want to hit double zero twice to try and win,” Ratain said. “You really have to be certain about the drug. Even if you are certain about the target, you may be wrong about the biomarker.”
Ratain’s views are supported by some in pharma circles. Eli Lilly, for instance, has said that in its experience such biomarker-based PGx studies are more expensive and complex than traditional clinical trials, particularly when using a biomarker with weaker power to target a subpopulation [see PGx Reporter 04-18-07].
At the AACR meeting, Ratain analyzed the experiences of several companies utilizing biomarkers to develop targeted oncologics, outlined the pros and cons of their various strategies, and discussed just the right time to delve into validated biomarker research.
The Biomarker Hypotheses
When using biomarkers to develop a targeted drug, there are several strategies at the sponsor’s disposal. The sponsor can study only the subpopulation with a particular biomarker as Genentech did with Herceptin and Bristol-Myers Squibb did with Erbitux. Or the drug maker can test the validity of a biomarker in the general population in a prospectively defined randomized trial. A company can also wait to study the biomarker until the drug is approved, as AstraZeneca did with Iressa.
Sponsors have had varying degrees of success with the first strategy. Bristol-Myers Squibb and Imclone limited Erbitux’s indication by only studying the drug in metastatic colorectal cancer patients who over-express the epidermal growth factor receptor protein. Ratain pointed out that a subsequent indication for head and neck cancer suggested that Erbitux did not need to be restricted to EGFR+ patients.
“There is clearly evidence that cetuximab is active in colorectal cancer patients with tumors that do not express EGFR,” Ratain said. In this way, BMS and Imclone “limited their market … because the drug is active outside the population that was studied in the pivotal trial.”
Wait for It
The second option, illustrated by AstraZeneca’s experience with Iressa, shows the difficulties of waiting too long to find a biomarker for a lackluster drug.
Iressa did not initially rely on early biomarker identification. Though the FDA put Iressa on a track for accelerated approval based on preliminary data from a Phase II study, the drug ultimately did not show a survival advantage in the Iressa Survival Evaluation in Lung Cancer trial in 2005. As a result, the agency limited its use only in patients who were already on the drug and deriving a benefit from it.
Eventually, researchers discovered from a subgroup analyses of ISEL that Iressa yielded statistically significant increases in survival compared to placebo in Asian non-smokers. Although clinical trials have established that EGFR status is not a predictor of response, analyses of biomarker data from ISEL have found that a high EGFR gene-copy number may be a strong predictor of benefit with Iressa in pre-treated advanced non-small cell lung cancer patients.
“In this game of onco-roulette you don’t really want to hit double zero twice to try and win. You really have to be certain about the drug. Even if you are certain about the target, you may be wrong about the biomarker.”
In 2006, Genzyme introduced its KRAS mutation-analysis test that identified NSCLC patients who test positive for specific KRAS mutations, which have been associated with resistance to the tyrosine kinase inhibitors Tarceva, made by Genentech, and Iressa.
“If you don’t study your biomarker until the drug is approved you have this battle between those selling predictive tests,” Ratain said.
This uncoordinated entry of drug and then the device may create a marketing tug of war. “So you will have people selling tests saying ‘I can predict who is going to benefit from the drug,’ and you have people selling the drug saying, ‘Don’t believe those people who think they know where our drug works.’”
Finally, if biomarkers aren’t validated prior to moving onto late-stage drug development, sponsors can risk a false negative Phase III trial. This threatens to shrink the market potential of the drug as it did with Erbitux — or worse, lead to a drug-candidate failure.
For example, 5- fluorouracil as an inhibitor of the thymidylate synthase enzyme would have not entered the market if clinicians had chosen to study the drug only in thymidylic synthase over-expressers.
“How badly would we screw up 5FU today if we brought it in into the clinic today as the first thymidylate synthase inhibitor?” Ratain posited. “We can say that low thymidylate synthase expression is a potential predictive marker for adjuvant 5-fluorouracil in this setting. Well, what if you said that thymidylic synthase is the target, therefore we’re only going to study patients who are thymidylic synthase over-expressers or high expressers. …you would conclude your drug doesn’t work.”
Big Phase I/II Trial Equals Big Success?
Ultimately, a successful strategy for employing biomarkers is to prospectively study all-comers to get a better sense of the biomarker, Ratain said. This type of trial promises to be large and more expensive, but it may pay off in a successful Phase III study and a marketable drug candidate.
The early validation of biomarkers in large studies in the general population “will facilitate Phase III testing and demonstrate the clinical utility,” of the biomarker, Ratain explained.
For instance, the Phase II studies for Bayer’s kidney RAF kinase inhibitor Nexavar initially looked at the drug’s efficacy in patients with any kind of cancer resistant to treatment, particularly colon cancer. However, after investigators noted that colon cancer patients were not responding, they switched their focus to kidney cancer patients who were responding to the drug.
“If in Phase II trials [Nexavar] had been limited to patients with RAF-dependent tumors, the drug would clearly have failed,” Ratain said.