Not wanting to be left out of the early-2016 chaos, the neuroscience area chimed in today with the announcement that Alkermes’ first two Phase III trials of ALKS 5461 had failed in Major Depression. In one trial (FORWARD-4), the higher of the two doses used (2mg/2mg each of samidorphan and buprenorphine) showed a “clear trend” towards efficacy. In the other (FORWARD-3), which only used the 2mg/2mg dose, there was no mention of any trend, but the trial failure was attributed to a high placebo effect. Alkermes, which boasts one of biotech’s best spinmeisters as CEO (Richard Pops), stated that they will now increase the sample size for the other, ongoing Phase III (FORWARD-5), and suggested that success in that trial, along with the trend in FORWARD-4, and the results of the Phase II put together, “could provide substantial evidence of efficacy for ALKS 5461”: As if they could somehow cobble together an NDA filing from these cherry-picked datapoints. Right.
Some preliminary thoughts:
1) The Phase II data, which came from a 142pt trial, had achieved enough of an apparent treatment effect that we did expect a signal to emerge from the Phase III program. We had not anticipated what is not far from a complete failure in these first two Phase IIIs.
2) We had previously focused on the troubling side effect profile for ALKS 5461 as potentially off-setting modest therapeutic efficacy. After all, the PhII had revealed a high rate of nausea (34%), vomiting (17%), and dizziness (19%). Today’s press release noted that the “safety and tolerability profile of ALKS 5461 was consistent with that reported for the Phase 2 and FORWARD-1 studies.” That is not good news–it suggests that a marginal signal that they hope to bludgeon into significance with a larger trial population is accompanied by a much worse-than-standard tolerability profile. Add in sublingual administration and probable Schedule III classification, and one has a nausea-prone drug with mediocre clinical benefit and poorer ease-of-use than its competitors.
3) Coming back to the regulatory path ahead if FORWARD-5 ekes out a better than p=.05 signal: Even hinting that the FDA might consider a single Phase III success sufficient for approval, when accompanied by a Phase III trend and a Phase II success (in a four week trial) is pure fantasy. One law firm has already announced that it is looking at Alkermes’ behavior, this the kind of legal quackery we usually dismiss out of hand. But if Alkermes continues to disseminate PR blarney, they could generate some genuine legal exposure for themselves. Depression is not an untreated disorder, patients do have a multitude of alternatives, however imperfect, that they can try, thus we do not think that the FDA is going to twist itself into a regulatory pretzel in order to make way for ALKS 5461.
4) In terms of collateral damage: ALKS 5461 is thought to work primarily through a kappa opioid antagonist mechanism, and Cerecor had licensed such a compound from Lilly several months back. With these marginal-at-best ALKS 5461 results, doubt is cast on the viability of the mechanism–thus Cerecor shed 17% of its valuation today. It is not their lead program, but as their second-in-line, its perceived value has been shaken.
5) SPCD trial designs could be seen as taking a hit here, but that would be premature, perhaps completely incorrect. Alkermes had made much of their plan to use SPCD in the ALKS 5461 program, the premise being that one could thereby run smaller studies, with less risk of placebo effects arising and masking a signal of effect. But it is believed that Alkermes did not use SPCD with FORWARD-3, the two arm, 429pt trial for which the high placebo effect was reported by Alkermes. The much smaller treatment arm sizes for FORWARD-4 (385 patients, three arms) suggest that SPCD was used in this trial, and indeed the placebo effect was less salient, and this is where a ‘trend’ was seen. It is worth noting that FORWARD-5 appears to be structured like FORWARD-4, and thus may also be an SPCD protocol study. In sum, it is possible that what we see here is the kind of marginal, mediocre outcome that is not uncommon in antidepressant trials, and it was only in the conventional-design study that the placebo effect was problematic–which might actually lend support to the use of SPCD. Unless and until Alkermes confirms which trials utilized which protocol, this is all provisional.
6) Speed kills. FORWARD-3 was originally scaled to enroll 670 patients, and to finish mid-year, but only 429 patients were enrolled, and the results were revealed today. Rushing clinical trials is never a good thing, and it must be wondered why Alkermes accelerated this study, and cut enrollment off after just over 60% of the original target had been enrolled.
This does not necessarily eliminate ALKS 5461 completely as a novel antidepressant candidate, but it does mean that any efficacy that it could eventually demonstrate may not be commensurate with its tolerability issues. This also does not mean that SPCD does not work as a clinical trial protocol, in fact it may have done what it was intended to do in FORWARD-4. And it does not disenfranchise the kappa opioid receptor system as a potential antidepressant target–this combination pharmacotherapy may have been an indirect and inefficient vehicle for tapping that system. These results do raise questions about Alkermes’ self-promotion and its execution of this Phase III program. It also reduces ALKS 5461’s profile as a potential ‘new-generation’ alternative for the treatment of depression: JNJ’s esketamine and Allergan’s rapastinel/NRX-1074 are now at the front of that pack.
1) Acorda’s acquisition of BioTie further expands Acorda’s comprehensive Parkinson’s portfolio, and removes the pressure on BioTie to execute a Phase III program that would have strained its resources to the max.
2) The BIAL FAAH-inhibitor tragedy in France is a reminder of the risks courted in pharmaceutical development, but does not necessarily kill the mechanism. Pfizer and Vernalis have safely completed Phase II trials with their FAAH-inhibitors, and while JNJ suspended clinical testing for its compound, which reflects commendable caution, we suspect that this is not a class effect.