Alkermes late yesterday announced that the third of the ALKS 5461 antidepressant Phase III trials had yielded positive results for the high dose, utilizing an SPCD trial design and an alternative parsing of the MADRS as the primary endpoint. The good news was the evidence of efficacy; the bad news was that the side effect profile continues to be consistent, meaning that nausea/vomiting (PhII had revealed a high rate of nausea (34%), vomiting (17%), and dizziness (19%)) are not irrelevant adverse event concerns. The great unknowns going FORWARD involve the FDA, on a couple of points:
1) Alkermes is the first company to devise a pivotal trial program largely (but not completely, as they learned to their discomfort in FORWARD-3, devastated by a placebo effect) utilizing SPCD, the Maurizio Fava designed sequential trial process intended to reduce the impact of the placebo effect. There is as of yet no consensus in the field as to whether SPCD reliably delivers as promised, but it is difficult to see any reason why the FDA should not accept the results as valid, other than that their statisticians will have to adapt to a novel design. Psychiatric drug trials, especially depression, have been chronically undermined by placebo effects, and the trend appears to be worsening. The FDA would be foolish to reject this framework, and we do not think they will be foolish.
2) Alkermes hopes that this clearly successful SPCD FORWARD-5 trial, in conjunction with the almost-positive FORWARD-4 results (also SPCD), and the spectacular but nonreplicable results in Phase II (also SPCD), can be assembled into an NDA filing that will be accepted, and eventually approved by the FDA. This is where it becomes more opaque, and political. To approve based on such a package, the FDA would have to buy into four or five subcomponents:
a) A new trial protocol design (SPCD)
b) a new parsing of the MADRS using six items they believe most critical (but the high dose also hit on the MADRS-10)
c) this Phase III averaged MADRS-6 scores across several timepoints, instead of using a single time measure for establishing change. This is different from what they did with the previous trials.
d) reportedly, retrospectively analyzing FORWARD-4 this way would also have brought the high-dose outcome across the p=.05 threshold. The FDA’s Psychiatry Division does not have any history that we recall of embracing post hoc re-analyses of pivotal trials as constituting success.
the FDA would have to embrace the successful Phase II results as the ‘second confirmatory trial’, perhaps in conjunction with the almost-OK FORWARD-4. The Psychiatry Division has no history of redefining Phase II results as if they were Phase III. The FDA is deeply wedded to the idea of prospectively-established definitions and analyses.
3) Accepting the components in #2 would require a philosophical shift on the part of the FDA, an institutional move towards a degree of flexibility that has not been part of the mindset for Psychiatry or Neurology. The approval of Acadia Pharmaceuticals’ Nuplazid for Parkinsonian psychosis, on the basis of one successful Phase III after a couple of previous failures, has been interpreted as opening the door to such flexibility. Alkermes is not the only company to hold on to this view after mixed Phase III results: Intra-Cellular Therapies hopes that one successful Phase III, plus a very substantial Phase IIb that was successful, might provide critical mass for ITI-007’s filing in schizophrenia, despite the failure to separate in its second Phase III. Axovant hopes that a single Phase III success with RVT-101 in Alzheimer’s, in conjunction with all the data assembled by GSK during the drug’s previous clinical trials, would constitute an NDA filing. Alzheon hopes that a positive Phase III, along with the truckload of data from Neurochem’s failed program, might do the same for their Alzheimer’s drug. There are some false equivalencies that get tucked into these aspirations: Only the Acadia program addressed a relatively undertreated disorder, one that constituted a small to moderate size market, rather than mega-scale. Depression and schizophrenia have lots of treatment options, even Alzheimer’s has a few. The FDA was undoubtedly concerned that Nuplazid would look like a precedent heralding a more open door for drug approvals; they have a basis for conceptually narrowing that window once again, given the fact that none of these disorders was as undertreated as is PDP.
4) Whither the FDA? Sarepta’s drug approval in Duchenne was based on skeletal Phase II findings plus sentiment, and along with Nuplazid’s approval, may represent a Janet Woodcock/Robert Califf shift towards flexibility, and thus could in theory continue. But our expectation is that, when it comes to the FDA, progressive actions are often followed by an almost-equal and opposite regression, as they seek to reinforce their bona fides in safeguarding the public against the Pharma ‘barbarians at the gate’. With the transition to a Clinton Presidency, we would guess that the industry is in for at least a temporary period of heightened scrutiny and criticism (the only industry that damages itself as frequently with tone-deaf PR as much as Pharma does is the tobacco industry) and the FDA will want to look like it is playing its regulatory role. Hence our belief that Alkermes and Intra-Cellular will each be required to run an additional Phase III. This is not the decision we would make if it was up to us–we are satisfied that both drugs work, at least in some patients, and we would argue for adding to prescriber choices in areas that may not be undertreated in terms of choices, but where the choices are frequently unsatisfactory. Axovant and Alzheon do not have the kind of existing datasets that would afford them the benefit of the doubt in this regard.
5) We have been critical of the Alkermes PR style in the past, and hope that they do not taint their own prospects by getting too publicly cocky about them. The surest way to provoke the FDA would be to take them for granted.