Alcobra’s Selective Inattention

Last week, in the October issue of NeuroPerspective, which covered ADHD, we noted the delay in Alcobra’s disclosure of Phase III data for MDX (metadoxine), and stated “ our suspicion is that the analysis of the primary and secondary endpoints for the overall patient population was negative, and that they are seeking evidence of benefit
in the ‘Primarily Inattentive’ subgroup in order to salvage something positive from a failed trial.” This turned out to be at least partly accurate, since they have not yet reported anything on secondary endpoints, but they are indeed working hard at the salvage process.

Basically, the results of the primary endpoint analysis was that the trial failed–though Alcobra termed the obtained p=.15 as a ‘trend’, which is generous, we generally reserve ‘trend’ for that grey area between .05 and .10. However, their emphasis was on a posthoc analysis that excluded four patients in the Combined Inattentive/Hyperactive subgroup who had received placebo, but shown dramatic ‘improvement,’ to an extent that pointed to invalid responses. Removing those four led to a change in the reported value, to p=.03, which is somewhat astounding in itself. This raises the reasonable question of whether the trial failed or the drug failed–it is one or the other.

Unfortunately, Alcobra doubled down on a selective emphasis upon the perceived nuggets of value in the trial, and did not adequately attend to the wider spectrum of implications:

1) Three of the four outlier placebo patients came from the same US site. Alcobra did not break out how many of the trial patients in total came from that site, but we would suggest that it is self-serving to just exclude the dramatic placebo outliers. Invalid inclusion and/or response patterns could be the case for some of the patients who received MDX, whose improvement might not be nearly as much of an outlier compared to historical drug-responders. Alcobra would likely state that since the overall change in the drug group was similar to previous testing, and it was the placebo group that diverged, that the problem is in the latter. Unfortunately, that cannot be safely assumed: If, for example, additional ‘ATM patients” entered the trial at that site and were not screened out, there is no reason to assume that all fraudulent patients received placebo, and that less dramatic improvement in the drug group can be safely attributed to MDX, not patient fakery.

2) Which means that there is absolutely no reason to think that the FDA is going to accept this as one of the two required pivotal trials. Alcobra’s statement that the FDA looks at the “totality of the data”, as if they might set aside those four patients, denies regulatory reality (e.g. Neurocrine Biosciences‘ first Phase IIb in Tardive Dyskinesia, where one site’s onsite ratings had zero congruence with the video ratings–Neurocrine still had to run another Phase IIb, they did not get the go-ahead to enter Phase III based on overtly distorted Phase IIb data). Alcobra kept saying that they would meet with the FDA to assess their next step: The next step would be to run another Phase III trial, while assuming the requirement of a second. Meeting with the FDA would offer the opportunity to finetune the protocol and enrollment criteria to reduce the risk of this being replicated, but not the necessity of repeating the trial.

3) Alcobra’s conference call began with the statement that they would ‘start with the bottom-line’–that the trial results support MDX as providing clinical effect in Adult ADHD. We would disagree: While the results do not prove the converse, that it does not have beneficial effect,  a failed Phase III whose data are suspect does not establish benefit. Their emphasis on finding positive trends in the Combined Inattentive/Hyperactive subgroup where none had been seen before, might reflect the greater statistical power of this larger sample, but given that all of the identified outliers came from that group, we would suggest that some other patients in that category may have received drug but also responded fraudulently.

Alcobra needlessly undercut its own credibility with its press release and conference call, where they selectively inattended to negative ramifications of the dataset as it currently stands. The stock sell-off, of over 50%, reflects disappointment with the results, but also with the spin operation that ensued: One fund manager told me he thought the press release was “disgusting” in its flagrant denial of trial failure, he clearly was not alone.

Metadoxine is not finished, but its all-Israeli pediatric ADHD PhII data due late this year will not undo the damage done, and Alcobra’s management should consider a reset of its PR stance, which cost their credibility dearly.