Replies to post #390932 on Anavex Life Sciences Corp (AVXL)

[Anshu2]

Yes, it’s not a failure. FDA doesn’t consider 0.05 to be hard cut-off.

[lifescienceDD]

CONTINUED Statistical Issues and Conflation of P-Value Amounts
No one on this board is disagreeing on the academic niceties of significance level of 0.05. In fact, I think that many promising clinical therapies in early P1 or P2 trials (not just in Alzheimer's, lots in inflammation / dermatology) don't receive enough attention because the studies were underpowered from low N and obtained p-values just on the cusp of 0.05. No one is arguing that. But, the issue at hand with AVXL - concrete data in the context of regulatory approval - is different.

A p-value of say .055 (1 out of 18) is NOT a failure.

Less than or equal to 0.05 (1 out of 20) is an arbitrary number for significance.

To begin to make the implication that, in a pivotal setting for regulatory approval, the FDA would not care about breaching a set significance level of 0.05 for pre-specified primary analyses would incorrect, especially for large indications (e.g., Alzheimer's).

We're talking the FDA's Neuro division, right?
Ok - why do you think studies like solanezumab's EXPEDITION 2, where the sponsor achieved p=0.06 in a pivotal setting (on the cusp of significance, as you make the argument), not pursued for regulatory approval in Alzheimer's Disease?

Let's even go the other way. I point out Reata's current case, where there is high unmet need in a severe orphan indication and the sponsor's pre-specified primary endpoint analysis clearly succeeded p<0.05 with p=0.014; yet, they have been in regulatory limbo for ages.

Ok, let's say your study is barely p>0.05. Can you suddenly switch and say, "yes, CDR-SB is my primary endpoint now because I unblinded it and it worked, haha!"? First of all, that would be a clear violation of multiplicity. That's the equivalent of saying "yes, that horse got first place and I'm betting on that horse" after the race is over, and the FDA obviously knows that.

Second, if you breach the pre-specified significance level (here it is 0.05), your primary analysis has already failed. Any subsequent secondary or exploratory endpoint analysis p-values are considered to be only supportive in nature.

Ask yourself - are you someone who takes words at face value? Or are you an actual investor, who values concrete evidence, including the data to support misleading claims such as "MET PRIMARY AND KEY SECONDARY ENDPOINTS"?

ANAVEX has been extremely truthful in its revelations of trial results. They have been successful MEETING all endpoints of all the trials: