Replies to post #476713 on Anavex Life Sciences Corp (AVXL)

[boi568]

It also appears from the posted abstract that there is a larger SD among the ADL groups, given the slightly higher p value recorded despite a slightly high delta between the two groups. ADL results seem to remain noisier even after a period of years.

[frrol]

Thanks, I was looking at Sabbagh's CTAD presentation which omits the ADL slide. The ADL shows clear decline (regardless of volatility) on a 9-point graph scale for both cohorts. It's nice to hear there is arguably separation in the OLE, but it's only separation, and only stat-sig at week-192. This is why it's important to see the OLE's graph; hopefully there is earlier positive trajectory change in both the early and delayed cohorts, which is a far more compelling outcome in a long-term study. New guidelines or not, ADL will be a source of contention for skeptical evaluators (with trial size and dropouts) so we need a strong argument. As the guidelines state, "positive trend" for a functional endpoint will be supportive.

As for CGI-I, that's a clinician score, not by the patient/care-giver. It's exploratory, and rightly so.

[boi568]

This is an interesting ADL discussion from a statistical standpoint, but it's really a transposed version of the argument that "because the 2b/3 failed the ADL co-primary endpoint, the regulatory agencies are unlikely to approve an application."

This is not how policy is applied. As far as we can tell, the EMA as a matter of policy does not require the ADL co-primary endpoint for early AD, and the FDA is well on the road to the same place. (Anavex 2024 CTAD slide note; FDA draft guidance.)

Policies need to be reasonable, but they are necessarily somewhat blunt instruments. There is always an occasional instance where you can pretty easily argue against a policy on the merits. For the sake of argument, perhaps this is one of those times -- perhaps. It wouldn't be the case that the EMA, say, would now need to make a policy exception to allow for an approval because of the stats; it's the reverse, the EMA would need to make a policy exception to justify a disapproval.

When you argue that a trial has failed for regulatory purposes for missing an unnecessary co-primary endpoint (even though it fully met the required regulatory primary endpoint), you have lost the plot. It's comparable to the circumstance where a heavy underdog in a sport easily beats the spread and loses, achieving a "moral victory." That's something to feel good about, but it was still a loss. Flip it, and here's a circumstance where we you have a win -- an approval -- but there's something to feel bad about. It's still a win.

What matters is what the regulators want to see at the time they make an approval/disapproval call, not what they or the applicant wanted to see a few years before when the trial was designed.

In other words, it will be an uphill battle here to argue that the 2b/3 trial failed under the terms the EMA uses today. You can argue that it failed on its own terms, or that the EMA's terms are wrong. Both interesting arguments -- to a point -- but irrelevant and a waste of time. The EMA knows how to apply its own regulatory standards, and if it has accepted the argument that the 2b/3 had failed because the ADL co-primary endpoint was not hit, the CHMP would never have green lit Anavex submitting an MAA. Full stop.