Computing absolute numbers for ADAS-Cog and ADCS-ADL responders referenced on pages 19 & 20, Redux: I never drilled down the exact populations "n"s used on these pages for Blarcamasine/Placebo (Thanks much, Anavex). However, if these numbers are ANYWHERE CLOSE to 338 Blarcamasine/170 Placebo, the results below will be exceptionally close to those referred to by the company on pages 19 and 20. If the "n"s are closer to Doc's 277/152, then the results will still be close to those below, but require an adjustment of 1-5 downward, particularly in the number that received Blarcamasine.
I believe there were (or were very close to) 60 Blarcamesine and 18 Placebo to achieve -.5 on ACAS-Cog
I believe there were (or were very close to) 25 Blarcamesine and 5 Placebo to achieve +3.5 on ADCS-ADL
Plugging the above into the Odds Ratio Calculators, with populations in the vicinity of 338 Blar/170 Placebo will give Odds Ratios, P-Values and 95% CIs very close to those on pages 19 and 20.
Of note: I'm no expert, but achieving +3.5 (clinically meaningful) on the ADCS-ADL seems more difficult than -.5 (stabilization) on the ACAS-Cog, which would account for the smaller numbers of both Blar and Placebo for the former. Anavex picked the threshold numbers (and presumably did so with purpose), so I can't draw too many inferences, but obviously it does superficially look promising that Blar did "better" on the ADCS-ADL requirement (5x placebo, rather than 3.33x on ADAS-Cog) Again, this doesn't mean too much and is an apples to oranges comparison on the already incomplete information we were provided.
The 60 Blarcamasines that equalled or bested -.5 on ACAS-Cog, averaged a clinically meaningful -4.03 (contrast to placebo's overall +4.11), which is incredible, absolutely incredible. I don't know the definition of "super-responder," but if it equals clinically significant improvement from baseline, then those 60 folks had a good number of stabilizers as well as some "SUPER-DUPER-responders" to achieve that mean. Such a performance was likely integral in achieving the pooled p-value of .033. If form from 2A holds, the 50 mg numbers will be bonkers, and with an impressive p-value, despite halving n.
I would hope the management team & lawyers know not to play with the verbiage and mislead investors regarding the ADCS-ADL results, but even the most ardent believer must admit the data release was a bit cagey, and without a logical explanation given it predictably encouraged FUD and kneecapped the SP....again. If not for Missling and the PR's assurances, and working under the assumption that the company would want to pick a threshold for page 20 (+3.5 on ADCS-ADL) to put the treatment cohort in the best possible light, I'd say it's possible that the pooled results would be > .05, but I'd be absolutely shocked if it weren't very very very close. All you'd have to do is power it up a little bit, and it'd be a stone cold lock. But you probably wouldn't even have to do that as the 50 mg results are likely to be fantastic.
Even if you needed to run another trial (and I maintain that a BP partner could push these results through with our safety profile, ease of administration, and likely earth-shattering results on the other endpoints 50 mg dose), BP would fall over themselves bidding to partner for this kind of data. We'd double the amount of ihub AVXL millionaires the second the ink on that contract dried, even if we had to wait a couple years for approval. So zero need whatsoever to mislead. Got that, Chris?
Alternatively, if Missling and co saw the pooled results had a p-value of say .055, but a few quick calculations showed the 50 mg dose cohort would have no issues whatsoever besting the threshold despite the halved n, is it good faith for the company to then claim it met the ADCS-ADL endpoint with statistical significance? Or is that determination only made from the pooled treatment cohorts? Come to think of it, that might be the best explanation for this whole rigamarole.
Anyone know the reason to have co-primary endpoints? If the inclusion of a second means both NEED to be satisfied for approval, regardless of how overwhelmingly positive 1 of them is, it seems an unnecessarily higher burden to meet. Obviously we didn't know this when we started the trial, but the PDD trial pretty much guaranteed ADAS-Cog would be very good. Would that have been an FDA-sanctioned lone primary endpoint? If ADCS-ADL joined CDR-SB as a secondary endpoint, would either or both need to be met successfully for approval?
