The two-day slugfest that took place this week over the safety of the diabetes drug Avandia could be the start of something big. It could herald a new day in the rigor of the science FDA demands -- both before and after a drug's approval.
Or it could be just another episode in FDA's equivalent of Groundhog Day, endlessly looping a frustrating scenario: This drug appears dangerous, but there's not enough data to pull it from the market.
Everybody in the crowded Hilton ballroom in suburban Washington where the hearings took place knew the subject wasn't just Avandia, which the panel agreed causes heart attacks but recommended it stay on the market. No, it was about how the FDA deals with safety problems that crop up after drugs are allowed on the market.
A sign of that spillover occurred Thursday, when many of the same advisors balked at recommending approval for the new diet drug Qnexa.
One of them, Dr. Abraham Thomas of Henry Ford Hospital in Detroit, voiced the worry that hung over both proceedings. Thomas voted to take Avandia off the market, and recommended against green-lighting Qnexa.
"Qnexa is far superior to anything that's on the market," Thomas said. "But our concerns are with safety. They have to be addressed, and I think it's best if these are addressed before approval."
And listen to Dr. Lamont Weide of the University of Missouri in Kansas City. Weide voted Wednesday to keep Avandia on the market with restrictions. But yesterday he voted no on Qnexa.
Failing to identify safety problems before approving a drug "really is some of what has given the FDA a bad reputation outside in the public," Lamont said, according to Medscape. "We have a responsibility to protect the public at large."
Noble words…but how?
A big worry about Qnexa is that all the safety data so far has been from study periods too short (less than a year) to assess problems for a drug that many people could take for a lifetime.
So what's new? That's true of virtually all drugs prior to marketing.
What's new, maybe, is a heightened sensitivity among FDA and its expert advisors that these kinds of studies aren't adequate to truly calculate the balance of benefit and risk.
With Avandia -- until a few years ago the most popular diabetes drug in the world -- most of the pre-marketing trials were of the usual six-month duration. As FDA panel members lamented again and again, those don't generate enough "events" (heart attacks, strokes, cardiovascular deaths, etc.) to give a clear signal of safety -- or lack of it.
And lumping together 42 or 52, or 56 of these inadequately "powered" studies, as the Avandia panel discovered, doesn't overcome that basic flaw. Pooled studies can give a signal, but it's often obscured by static.
So what does that leave? Well, sometimes there are larger clinical trials done by the drugmaker, for other reasons –- demonstrating a drug's superiority, or at least non-inferiority, to competing drugs. But those aren't safety studies. So those data need torturing too.
There can't be a clearer demonstration of the perils of such company-sponsored trials than this week's savaging of GlaxoSmithKline's biggish trial, called RECORD. Whether you believed the critics who called it sloppy or the others who say it was fraudulent, the RECORD study came away from this week's proceedings thoroughly discredited – and by implication, its ilk.
And, in case there's any doubt, the Senate Finance Committee this week put out fresh evidence that companies like Glaxo have powerful incentives not to mount (or publish) trials that raise questions about the safety of their blockbuster drugs.
And consider the big post-marketing study of Avandia that's struggling to get underway, called TIDE. That shows what happens when regulators and companies belatedly get around to mandating a head-on safety study.
FDA advisors were distressed to hear that, due in part to the cloud over Avandia, only about 1,100 of the hoped-for 16,000 TIDE volunteers have signed up, and that results might come, at best, in 2016. (At worst: 2020.) "This study should have been done 10 years ago," said one exasperated panelist.
That leaves what the experts call "observational" studies – analyses of what happened, in retrospect, to people who actually took the drug in question, or didn't. Necessarily, those kinds of look-backs are going to be a mainstay of post-marketing surveillance.
Those types of studies got a big boost this week from the Institute of Medicine, which pointed out that observational studies have the great virtue of showing what happens in the "real world" – a different place than the insular world of randomized clinical trials, with their carefully selected volunteers.
But the experts weren't willing to hang their hats on a big observational study of Avandia – it encompassed virtually all Medicare recipients who took the drug over several years, compared with an equal number who didn't. It wasn't that the study, by FDA gadfly David Graham and colleagues, wasn't impeccably done, they said. These studies just contain too many opportunities for hidden bias.
That leaves the FDA, the experts, the drug companies, the doctors and all those patients in an unhappy place.
As Dr. Sanjay Kaul of Cedars-Sinai Medical Center put it, "Boy, have we seen enough ambiguity to last a lifetime!"
There's a lot more ambiguity where that came from, of course. Unless the "new" FDA can figure out how to get better, more timely data on a drug's safety, before and after it's unleashed.