The FDA’s decision to strip the drug Avastin of the indication for treating metastatic breast cancer is of a piece with the agency’s aversion to risk—not just the risk of safety problems, but the risk of benefits.
For years, the FDA’s risk aversion was confined mostly to the agency’s focus on drug safety. But now, this culture is applying equally to its evaluation of drug efficacy, as the agency shows an increasing unwillingness to take a chance that the magnitude of the benefit that it observes in clinical data may not be as strong as it seems.
To guard against this “risk,” the FDA has pushed sponsors harder to conduct “perfect experiments” where patients are neatly randomized to artificial treatment groups for purposes of getting a pure statistical answer. That’s the case even if it means that the clinical trial, and in turn the information it yields, has less correlation to the kinds of practical information that doctors and patients are going to need to make prescribing decisions. Clinical trials, especially in cancer, increasingly randomize patients to treatment arms that don’t approximate how doctors treat patients in the real world. For this reason, more and more of these trials must be run in Eastern Europe, where patients are more likely to enroll in trials that randomize to outdated treatment regimens just to get access to the drugs.
The most vivid example of this regulatory approach was evidenced with the FDA’s recent handling of drugs for the treatment of Hunter’s Syndrome and Gaucher’s Disease. The agency required rigorous, and, some argued, unethical trials where babies with these degenerative diseases were randomized for up to a year to receive infusions of a promising medicine or an inert placebo. Many of the babies on the placebo arm were permanently impaired by forgoing active therapy. The drugs worked. Having run a perfect experiment, FDA had incontrovertible evidence of the drug benefits. Its final regulatory decision was an easy one for the FDA. But at what cost?
There are statistical constructs that would allow FDA to have its cake and eat it too—to get firmer answers to questions of a drug’s benefits, while enabling these decisions to be reached perhaps more quickly, and in clinical trials that don’t require patients to subject themselves to hard clinical or ethical choices to be randomized to sham treatments. But the agency has been painfully slow to embrace new science when it comes to clinical trial design. The FDA commissioner recently committed to put out guidance on these “enriched” trial designs. FDA first promised those guidance documents in 2005.
In the case of Avastin, these enriched trial designs might allow doctors to more easily glean which subgroups of patients are getting a benefit from the drug. Even the FDA commissioner acknowledged that some patients seemed to be responding to the medicine. But when these patients are averaged across all of those included in the trials, this evidence of benefit starts to get diluted. In announcing its decision on Avastin, FDA made an odd distinction between the information it uses to guide its regulatory decision and information issued by institutions such as the National Comprehensive Cancer Network, that “provide clinicians with ready access to synthesized information they can use in making patient decisions.” It begs the question what FDA thinks its role is in clinical medicine—and in this case, how much practical value the agency believes its own judgment has for patients.
