Replies to post #391539 on Anavex Life Sciences Corp (AVXL)

[Warby 3]

You think the market is right about anavex or biogen

Your posts are long and sometimes hard for me to follow
Thanks

[boi568]

My assumption is that the 30mg and 50 mg doses versus placebo has been the trial design all along. It's what was used in the PDD Phase 2 that Missling asked his audience to remember at the Guggenheim conference. The 30mg and 50mg doses were used in the AD 2a, and that small sample clearly indicated that the 50mg dose was worth pursuing over a combined dose.

You note that Missling had the dose breakdown data by CTAD. That has been my assumption. I believe it is his basis for announcing the trial had met all of its endpoints when, by looking only at the combined numbers, that was certainly debatable and not something that would, from a legal perspective, be advisable to do.

So I think all the critiques of the combined dose versus placebo outcomes are pretty much besides the point. We will shortly see the dose breakdown outcome. As noted on this board, a 50mg versus placebo trial will have to overcome its minimum size by great clinical effect. There have been great clinical effects with this drug in Rett and PDD episodic memory, so history is on our side.

[frrol]

Re ADAS, it was -.5 change from baseline at 48 weeks, which is not just a reduction in decline, but (slight) improvement. Why -.5 and not -1 or -2, not sure.

[catdaddy]

Some aspire to be vexations to the spirit.

[12x]

Since you brought up ADUHELM 2-dose arm p3 trial, ADUHELM used the same 2-dose arm trial, very relevant to Anavex trial. Here is the SAP at https://clinicaltrials.gov/ProvidedDocs/00/NCT02477800/SAP_001.pdf

Unless otherwise specified, baseline value is defined as the most recent non-missing measurement collected prior to the first dose. Change from baseline will be defined as post- baseline value minus baseline value.
Unless stated otherwise, all statistical tests will be 2-sided with statistical significance level of 0.05

Note 1: since ADUHELM trials don’t have titration period, I think ITT/mITT defined as post titration but prior to treatment period is reasonable for Anavex. So p3 308/163 as baseline is valid ITT n.
Note 2: no 1-side p

Considerations for multiple comparison adjustments
A sequential (closed) testing procedure will be used to control the overall Type I error rate due to multiple comparisons for the primary endpoint. The order of treatment comparisons is as follows: aducanumab high-dose versus placebo and aducanumab low-dose versus placebo. All comparisons after the initial comparison with p > 0.05 will not be considered statistically significant.
Secondary endpoints have been rank prioritized, in the order shown in Section 1.2. In order to control for a Type I error for the secondary endpoints, a sequential closed testing procedure will be used and will include both the order of the secondary endpoints and treatment comparisons. Specifically, for each of the secondary endpoints, a sequential (closed) testing procedure, as for the primary endpoint, will be used to control the overall Type I error rate due to multiple treatment comparisons. If statistical significance is not achieved for 1 or 2 treatment comparisons, all endpoint(s) of a lower rank will not be considered statistically significant for that 1or 2 treatment comparisons, respectively.There will be no multiple comparison adjustments for the sensitivity and supplementary analyses for the primary and secondary efficacy endpoints, the tertiary efficacy endpoints, the subgroup analyses or the additional analyses.

Note 3: no multiplicity adjustment required for multi-dose p3 trial if closed testing procedure is followed.
Note 4: the procedure calls for high-dose test first, then lose dose, then all. Did Anavex SAP call a pooled first and high-dose next?
Note 5: following list is ADUHELM test procedure. The responders analysis is way down the list. Did Anavex have a different order list? Can Anavex put responders analysis on top to replace the score delta?

CDR-SB Primary: Analysis of change from baseline at Week ITT 78 (MMRM)
Sensitivity: Pattern mixture model (ANCOVA) ITT
Sensitivity: Copy increment from reference method ITT (ANCOVA)
Sensitivity: Imputation by natural disease progression ITT (ANCOVA)
Sensitivity: Tipping point analysis (ANCOVA) ITT
Supplementary: Censoring after intercurrent events ITT (MMRM)*
Supplementary: Per-protocol analysis (MMRM) Per-protocol
Supplementary: Responder analysis (Logistic ITT regression)
Supplementary: Slope analysis (MMRM) ITT
Supplementary: Divergence effect analysis (MMRM) ITT

[12x]

what does the pre-specified subgroup analysis do for p3 trials? Does it enable a drug approval for that subgroup (e.g. WT with drug label) in the event the whole group not meeting primary endpoints? Or only the subgroup analysis only serves as basis for the next p3 trial?

[lifescienceDD]

Doc328 is a reasonable voice of sanity. It would benefit us to read through his comments.

Yes, this is an issue on ADAS-COG. Even if I give AVXL an extra boost by using all patients enrolled (n=170 vs. n=338 pooled, again this is BENEFITTING AVXL with more N than using a smaller ITT # which they did not disclose) and plug in the direct #s of 4.11 (0.86 SEM) and 2.26 (0.51 SEM), I get 2-sided p>0.05. I encouraging everyone on this board to take 10 seconds and do it yourself - https://www.graphpad.com/quickcalcs/ttest1/?format=SEM. Plug in n=170, 4.11 with 0.86 SEM vs. n=338 pooled, 2.26 with 0.51 SEM from Slide 21. Yet somehow AVXL legal team greenlighted press releasing "MET PRIMARY AND KEY SECONDARY ENDPOINTS".

The problem with A273 is that it looks like the main cognitive data did not pass the one sided 0.25 (or two sided 0.05) threshold. And that was with a very liberal modified ITT first ignoring patients unable to get titrated to enter the maintenance phase and then excluding patients who did not have a 48 week measure .

Yes, exactly. It is mindnumbingly bizarre that the two pre-specified primary analyses from the CRO were the first things AVXL received (and AVXL has already shown us one of them on ADAS-COG), but is withholding the ADCS-ADL mean difference.

We don't even know the ADCS-ADL mean change.

I have given AVXL the benefit of doubt that pooling is acceptable (pooling can only help a a sponsor because it gives you increasing powering) and that they can present 30/50mg N splits later. Even with pooling, I came to the same conclusion as you.

Was the pooled data the pre-specified cohort in the SAP? If so, the analysis to date is ok to present as the cohort though insufficient for an NDA (IMO) (due to one sided > 0.025 and lack of co-primary ADCS-ADL mean change data).

Amen, I have zero clue why on earth N was not disclosed. I encourage anyone to give an example of a P3 pivotal readout where N for an analysis was not spelled out.

Why was n not included in the clinical endpoint slides?

I'll go even further - folks on this board liberally talk about a surrogate biomarker for AA for 2-73. Aduhelm (via ab plaques) and Sarepta (via % dystrophin) are clearly linked to the biological mechanism of action so it was straightforward to at least make the case for their biomarkers. Why hasn't anyone on this board pointed out an AA biomarker for 2-73 yet? It should have been prospectively identified like these two examples.

So I put the likelihood at near zero. I do think you came up with possibilities that together add up to the 1-2% chance that I give for AA

I look forward to PDD/Rett's updates to help us shareholders out in the coming year. But this lack of disclosure on the most basic pieces of the AD trial perplexes me.

[MycroftHolmes]

Re: Doc Analysis

Umm.. huh?

By your own analysis:

If instead 50 vs placebo and 30 mg vs placebo was the primary endpoint pre-specified endpoint then we have no data to offer an educated opinion as dose breakdown data is missing (missing for us, Missling has had it concurrent with receiving the initial analysis --- you don't not get the primary data)

and

So I put the likelihood at near zero. I do think you came up with possibilities that together add up to the 1-2% chance that I give for AA

So you give 1-2% chance that Anavex both 1) pre-specified a primary endpoint of 50mg vs placebo and that the 2) resulting benefit (which we don't have but know the overall mean is high) is sufficient for Accelerated Approval?

I do appreciate your analysis on this board, but I don't think I am being a Polly Anna when I say that this assessment seems unnecessarily pessimistic.

Cheers

Mycroft

[DoTheRightThing]

Appreciate your posts Doc. One of the few realists on the board and obviously you have a ton of knowledge to share. I wish I had followed you and taken more off the table pre CTAD. Anyone who followed you would've saved a ton of money.