The wonder of the US FDA advisory committee meeting to consider Alkermes's depression drug ALKS 5461 is that the project even got as far as this. Two indisputably negative trials and two "positive" studies the regulators regard as failures should have marked the end of development, but Alkermes obviously thought it could persuade enough people that ‘5461 reduced symptoms of depression.
The 20-3 vote against recommending ‘5461, along with the agency’s clear exasperation with the ways Alkermes ignored its scientific advice, virtually guarantees that the project is headed for rejection. The only question left for Alkermes is whether to forge ahead as a developer of mental health agents or pivot to early-stage immuno-oncology projects – where, surprisingly, it already has an asset.
If the shift is on, Alkermes is giving no evidence of it yet. The group said it would “continue to work with the FDA” on the ‘5461 submission and that it remained committed to drugs treating mental illness. As its marketed mental health products are forecast to bring in close to $1bn in 2024, according to EvaluatePharma’s consensus of sellside analysts, the company probably does not want to walk away from the field altogether.
In addition, Alkermes is awaiting phase III readout of another CNS asset, ALKS 3831, in schizophrenia, due by the end of the year. That candidate, a combination of samidorphan, which was also part of ‘5461, and the generic antipsychotic olanzapine, needs to show that it is better than existing antipsychotics on the side effect of weight gain (Upcoming events – A double date with destiny for Alkermes, October 5, 2018).
But it looks like game over for ‘5461. Adcom members were no more convinced than FDA reviewers that the two trials that Alkermes reported as wins were actually positive. In one, study 202, a lack of significance on the HAMD scale in the group taking the higher dose of 8mg each of buprenorphine hydrochloride and samidorphan cast doubt on the findings. Additionally, the statistically significant findings for the lower 2mg/2mg group were not adjusted for multiplicity and, if the results from a single outlier patient were removed from the final analysis, even that dose lost significance.
The other study Alkermes claimed as evidence of efficacy was called 207. The reviewers pointed out that in this a 2mg/2mg dose showed significance on the MADRS scale only when averaged across timepoints during treatment, not at the end of treatment.
The FDA reviewers also were displeased with Alkermes’s decision to use the novel sequential parallel comparison design in spite of the fact that they had not endorsed it, and disagreed with the statistical analysis of study 207 using two different versions of the MADRS scale and using the average of timepoints.
Committee members obviously agreed. The vote was 20-3 against ‘5461 on efficacy, and 21-2 against the project having proved a sufficient risk-benefit profile for approval. Alkermes shares fell 4% yesterday.
Another placebo-controlled trial of ‘5461 is ongoing, but realistically the project's chances here must be seen as slim.
Whether Alkermes chooses to remain a CNS developer could therefore rest on the readout of the Enlighten-2 trial of ‘3831. Alkermes seems to believe that it has proved itself on efficacy, with its candidate looking at least the equal of olanzapine monotherapy on the PANSS measure. For commercial purposes, however, the trial needs to show that patients on ‘3831 do not see the weight gain of the established generic, something that could persuade payers that it is worth the extra money.
Mental health drugs are a tricky area of development, of course, as many effective products have lost patent protection. In addition, the placebo effect tends to make proving efficacy difficult, and there are concerns over safety issues such as suicidality.
Alkermes might already have a new avenue to follow: as is frequently the case these days, it is dabbling in immuno-oncology with the IL-2 agonist ALKS 4230. Data from a first-in-human study in solid tumours are due at next week's SITC meeting.
Should Enlighten-2 turn out to be a dud, executives will still have something to talk about at investor meetings – and one sure way to create continued excitement is to have an early-stage cancer agent.