You bring up some interesting points from the SAP and how it might / might not be relevant to Anavex. I'll add a few comments to your notes.
They have a very detailed SAP. Many I've seen are more like 30 pages while the Adu SAP is 90 pages - likely reflecting the complexity of the study. It's a shame SAP's aren't required to be made public at the time of data release as many questions are answered in them - of course what would we talk about then.
Note 1: They defined the ITT population as "The Intent-to-treat population is defined as all randomized subjects who received at least one dose of study treatment (aducanumab or placebo)" That's a standard definition (the most rigid ITT definition includes a few patients who drop out before the first dose but few object to their exclusion). A similar definition (one dose) for Anavex 273 would lead to n= 335 for treated pooled and n = 168 for placebo. I think Anavex can argue the mITT definition for efficacy as being those successfully titrated but it's difficult to exclude those who have had one follow-up post-baseline evaluation (i.e 12 weeks). Hence only counting patients who have baseline and 48 weeks remains troubling to me (and possibly regulatory agencies). For slide 21 A273 "Analysis method: t-test on change from baseline at the end of treatment (week 48) on subjects with available scores at week 48." There is an additional comment for the Responder analysis definition of ITT. he traditional way of imputing subjects missing the final clinical analysis is to record them as non-responders. For Adu the language used was "All subjects with missing data at Week 78 will be classified as non-responders." Despite higher thresholds of response, this method of accounting is far more conservative and one reason why comparison is difficult.
Note 2: Yes, the standard for every regulatory agency for superiority trials is a 2 sided 0.05 significance. Of course, one is free to substitute a one-sided at 0.025 significance as they are identical. That's the main reason I sold most of my remaining shares last Monday (down to just a token 1000 shares now) when one vs two sided was not cleared up at the pre-market conference and when ADCS-ADL change from baseline to 48 weeks was still absent.
Note 3 and note 4: Multiplicity for closed dose testing procedure. You bring up a good point that other methods besides the most used Bonferroni method are acceptable. This is actually covered in the FDA guidance page 20 Fixed sequence testing as one of the valid ways of handling multiplicity. including different doses. Biogen stated they will first test the highest dose for significance at 0.05 and then test the low dose only if the high doe if significant. We don't know the wording in the Anavex SAP. It is possible that they could have said " We will first officially test combined dose, then 50 mg assigned only if significant, then 30 mg only if 50 is significant". However, I feel this is unlikely for 2 reasons: (1) Missling has often spoken about the 50 mg being a better response than 30 mg (based on concentration levels from the 2a and from the PDD study). Therefore I would assume he would be very reluctant to put his eggs in the 30 mg plus 50 mg pooled data basket vs the 50 mg. (2) The worsening in the CTAD slide 24 "Therefore, ANAVEX®2-73 treatment groups were combined in primary analyses (all exposed to ANAVEX®2-73)" really sounds post-hoc to me not prespecified Therefore, if Anavex did Fixed sequence testing I think they would have specified to statistically test 50 mg first.
Note 5: From the Adu SAP it is clear that the baseline to week 78 difference of means was the primary CDR-SB analysis and that the responder analysis was supplemental. Could Anavex have specified the responder analysis odds ratio as the primary and the difference of means as secondary? Theoretically, they could but that would have gone against (1) what other studies have done and what the FDA accepted in other drug decisions and (2) Wording of the primary endpoint on clinicaltrials.gov "Reduction in cognitive decline assessed from baseline over 48 weeks with ANAVEX2-73 compared to placebo using the Alzheimer Disease Assessment Scale-Cognition (ADAS-Cog)" ditto for ADCS-ADL. We won't know until the SAP is published on ct.gov. Note also that the Adu SAP lists both ways of looking at the CDR data but very clearly lists the primary endpoint vs supplemental -- hence Missling could 'honestly' state that the Responder analysis is in the SAP, even if it is supplemental and not primary. They also do extensive sensitivity analysis with different imputation and analysis methods - all spelled out in the SAP - to demonstrate the robustness of the data.
Market Data 
Markets 
