Replies to post #387865 on Anavex Life Sciences Corp (AVXL)

[frrol]

I suspect the company will provide follow up commentary either today or Monday to affirm/clarify/correct the PR'd and CTAD numbers. At some point we'll certainly be getting more tables, and time study charts.

I'm glad we were able to present at CTAD. Unfortunately, instead of favorable ADAS-Cog comparison to Lecanemab, the main narrative coming out is "positive but incomplete and possibly erroneous data". That can be remediated. Let's see.

[peeved]

Why not allow for phase 4?

Id subgroups are great and primary endpoin met assuming they post other slide that was omitted 3.5 one

[GD]

Street agrees with you, in the end, Missling is trying to massage the data,
good subgroup data for P3, we are right after all about mixed topline data,
and AVXL may see AD P3 data by end of 2026 the earliest if AVXL starts
the P3 before end of 2023, AVXL dreamers can forget about any type of
approval with this data set, IMO.

[Doc328]

Yep that is exactly what they did. One mystery solved. Now show us the ITT data

From initial presentation slide 21:
footnote says

a 
 Analysis method: logistic regression 
 ITT: Intent-to-treat

and in revised deck:

Ref.: Reference 
 a 
 Analysis method: t-test on change from baseline at the end of treatment (week 48) on subjects with available scores at week 48. 
 Mean change from baseline obtained from an average of ADAS-Cog score change for each subject. 
 Mean of change from baseline not equivalent to subtracting mean baseline scores from mean end of treatment scores when all subjects do not have both measures. 
 bRelative Reduction in Decline = (1- [mean change in active/mean change in placebo])*100

This is not the standard ITT population . It is the protocol completers.

Likewise slide 22 old slide:

Ref.: Reference 
 a 
 Analysis method: t-test on change from baseline at the end of treatment (week 48)

New slide:

Ref.: Reference 
 a 
 Analysis method: t-test on change from baseline at the end of treatment (week 48) on subjects with available scores at week 48. 
 Mean change from baseline obtained from an average of CDR-SB score change for each subject. 
 Mean of change from baseline not equivalent to subtracting mean baseline scores from mean end of treatment scores when all subjects do not have both measures. 
 bRelative Reduction in Decline = (1- [mean change in active/mean change in placebo])*100

They also added this to slide 25

a Maintenance phase of trial

What about the poor patients who could not tolerate the drug during titration?? Does their safety data not count?

[growingpain]

There are several ways to get a p-value of 0.033. I will give you a very far-fetched example here which I am sure is far from reality. It is based on two irrational assumptions:

Assumption 1: They presented a one-tail p-value. Remember the head-scratching change in the endpoints from "change in decline" to "reduction in decline." A few of us here were speculating that this would enable them to report one-sided p-values. Not that they did.

Assumption 2: They made a minor mistake and they assumed a sample size of 301 for the placebo too. Such mistakes can happen sometimes, especially when you only have one day to analyze the data and prepare a presentation. But I am sure that was not the case here.

Let's see what happens when we use the above fictional assumptions to compute the p-value for the ADAS-Cog score change:



It seems we get a p-value that is almost identical to that reported by the company (any small differences could be justified on the basis of rounding).

As I said the above is far-fetched, but just for fun, let's apply the same process to the data we have for CDR-SB. The company reported a p-value of 0.04 in this case. Under the same assumptions as above, here is what we get:



Strangely, this is again similar to the p-value reported by the company with a tiny difference that could be attributed to rounding again.

In any case, the above are all product of my imagination and I am sure the next PR will ease all our concerns about the TLR.