An FDA advisory panel Thursday panned a new experimental drug for depression from Alkermes, meaning the agency is likely to reject the treatment by early next year.
By a 21 to 2 vote, the 23 experts decided that the data accrued by Alkermes (NASDAQ: [[ticker:ALKS]]) don’t support approval of the drug, currently known as ALKS-5461. They also voted 20 to 3 that Alkermes hasn’t provided substantial evidence that the drug is effective.
Though the results of these meetings aren’t binding, the FDA looks to its advisory panels for guidance and its approval decisions are often in line with the panel votes. The FDA will decide the fate of the Alkermes drug by Jan. 31.
The vote was based largely on criticisms of the drug’s effectiveness, which many panel members said seemed small at best, as well as Alkermes’s trial design, and the changing statistical analysis the firm employed.
“I don’t think there’s evidence that this drug works,” said one panelist, Martin Kulldorff, a professor of population medicine and a biostatistician at Harvard Medical School.
The decision sets up what looks to be a stinging setback for Alkermes, which started out as a drug delivery specialist but has since made a concerted effort to transform itself into a drugmaker. ALKS-5461 is a key part of that transformation.
Alkermes shares fell about 5.7 percent, to $38.50 apiece, in post-market trading Thursday.
The Alkermes drug is meant to help treat people with the most common form of depression, major depressive disorder, which affects an estimated 18.1 million Americans. ALKS-5461 is being developed for the large percentage of patients who don’t respond adequately to typical antidepressants and still show symptoms. Alkermes estimates that two-thirds of people who start treatment fit this description.
But the case for ALKS-5461 is complex for several reasons. The drug has had a mixed track record in clinical testing. Alkermes tested the drug in human studies that used a novel design, a two-stage trial known as a sequential parallel comparison design (SPCD), that the FDA has never reviewed before. And the drug combines an opioid drug for pain relief (buprenorphine, also used to treat opioid addiction) with a chemical (samidorphan) meant to counteract its addictive effects.
As Xconomy reported earlier this week, the FDA’s review team criticized Alkermes’ methods, including the use of SPCD, and highlighted the drug’s potential safety issues. Still, prior to today’s public hearing with the FDA advisory panel, there was reason for optimism. Placebo-controlled studies of psychiatric drugs often fail due to placebo effects, and the FDA, cognizant of the challenges, has a track record of approving such treatments even with a mixed record in human testing.
The SPCD design was invented to try to deal with the placebo effect. The idea is to randomize and split patients into a drug and placebo arm, and then take the placebo patients who don’t respond and randomize them a second time into either a placebo or treatment arm, to identify patients who consistently don’t show a placebo effect.
Speaking with Xconomy before the advisory panel hearing, Maurizio Fava, a psychiatrist with Massachusetts General Hospital in Boston—and also an investigator for ALKS-5461 and the co-inventor of the SPCD design—disagreed with the FDA’s read on the data.
The trials “have shown, in my mind, a consistent effect,” says Fava, who also spoke during a portion of the hearing that allowed comments from the public. “To me, in a field where it’s very hard to get consistent results, that suggests clearly that the compound is efficacious.”
But FDA reviewers felt otherwise, arguing both before and during today’s hearing that the drug wasn’t supported by the seemingly positive study results. The panel of outside experts agreed, saying that the results either weren’t “robust enough,” or didn’t provide substantial evidence that the drug worked. The panelists also pointed to a number of “methodological issues,” from changes Alkermes made along the way to the use of the SPCD design.
Speaking with Xconomy prior to the hearing, Fava defended the use of SPCD. He noted that there are a number of already approved drugs that were tested using another method of combating the placebo effect. The FDA knows that the method—called a “single blind placebo lead in,” which involves giving everyone a placebo to start the study, not informing them, and then weeding out the responders—hasn’t worked very well historically, he says. But “they’ve accepted the principle that [it] was not biasing the sample.”
SPCD, by comparison, more effectively weeds out placebo responders, Fava told Xconomy, so you can get a “more precise, better sense of the actual efficacy” of the drug.
The FDA disagreed. At the hearing, FDA representatives said the agency specifically told Alkermes not to analyze its data through an average, which it still did. And the agency hasn’t deemed SPCD trials statistically acceptable.
“Conceptually the design is appealing…but you’d expect the drug effect difference to be amplified,” said one FDA reviewer. But that didn’t happen, which means “either SPCD doesn’t work, or the drug doesn’t work.”
“It seems clear that FDA has essentially made up its mind,” wrote Stifel analyst Paul Matteis in a research note during the hearing.
