$AVXL Phase 2b/3 Spoiler Alert, or is it?

The Criteria for Phase 2b/3

On July 25, 2018 AVXL has given genetic and performance criteria resulting from KEM analysis study of 26 patients still in the Phase 2b at 57 weeks for best improvement.  I did not know which patients had the right genetic material but I knew that some of them had initial scores above MMSE 20 points and their improvement or decrease in score was also known and could be matched to the given average.  It would be only matter of inspection to find patients who would fulfil such criteria.  My previous post dwelled on it. avxl-preview-of-phase-2b-3for-anvex-2-73

Bigger Picture of “Placebo” Alzheimer

Some maintain that there is no placebo effect and some don’t.  Nevertheless, the natural progress of the disease was given in the reference 775209 from which I derived the plot of the fate of all patients divided into 4 classes.  The classes are given in the legend.

The area in red (primarily) and area in yellow are of most interest for us and determine the therapeutic effect in prospective plot of any trial of a Alzheimer drug.  It is highly unlikely that placebo effect can be found in the sufferers of Alzheimer so this can be taken as the placebo arm of Phase 2b/3.

 

Is it Spoiler Alert for Phase 2b/3?

Disclaimer:  This is only projection based on logic, some 3rd grade math and extrapolation of data, so relax and take it with grain of salt – don’t buy Tequila yet!

I took the liberty for extend the rial to 6 years, we have data for almost all patient who fulfil the selection criteria for 2 years only, yet from the rates of decline/improvement one can make extrapolations into the future.   Everybody knows how unreliable extrapolation are in real world but the temptation is hard to resist.  The class of patients who have No Decline are up 2 years consistently improved their scores so we see it happening for 4 next years.  Small subset of 2 patients create the model’s 33% who Decline<3 MMSE and are projected to keep doing it further.  This plot can explain this much better:

Of course this projection/model applies to the 80% of population who have the right genetic variations.  So to get the general population data multiply 3 classes by .80 and add to Dropout the missing from the picture 20% of population.

 

if you belive I did a good job.  Buy me a beer!