GAITHERSBURG, Md., Dec 16 (Reuters) - Companies looking to tailor use of cancer drugs to people with certain genetic profiles may have to run larger studies to gather enough data, U.S. Food and Drug Administration advisers said on Tuesday.
Proponents of such so-called personalized medicine say it will curb costs by treating only to those most likely to benefit and help others avoid toxic side effects.
But the FDA, concerned about whether a retrospective look at earlier data is sufficient to alter prescribing information for the drugs, sought advice on how much evidence companies need to target their therapies based on genetics.
Panelists said companies must have a well-organized plan to review their data. They will also need enough patients and tissue samples to analyze genetic issues.
"It's going to make life more complicated, costly ... it's going to require larger clinical trials," said panel member Richard Simon of the National Cancer Institute.
At the meeting, ImClone Systems and Amgen Inc. said they have enough data from past studies to show that their colon cancer drugs, Erbitux and Vectibix respectively, best helped patients with a normal version of the gene Kras.
The Kras gene helps regulate cell division.
"The data are ... nothing less than transformational," said Hagop Youssoufian, a senior vice president at ImClone, which was recently acquired by Eli Lilly and Co .
ImClone and Amgen want FDA permission to tell doctors that patients with mutant Kras should not be treated with their drugs. About 40 percent of patients have mutant Kras tumors, according to Bristol-Myers.
While treating only patients with normal Kras would cut into sales of the expensive drugs, it would expand the field of personalized medicine. Some breast cancer therapies, Genentech Inc's Herceptin and GlaxoSmithKline Plc's Tykerb, are used in HER-2 positive patients.
Despite ImClone and Amgen's request, the FDA did not seek advice on Erbitux and Vectibix but cited them as examples.
Early data may guide drug development. "However, it cannot directly address whether a particular biomarker test ... distinguishes patients who will or will not benefit from the drug," said Robert Becker, an FDA device medical officer.
Analyzing data from earlier studies can lead to false conclusions or misinterpretation, FDA officials said. It also may be unclear whether patients have a normal gene or not.