Cassava Science Results vs. Anavex Life Sciences Projections $SAVA $AVXL

Do not trade this blog as 4 billion years of evolution is against me and possibly even God laughs at it, He has a great sense of humor!

Baseline in $SAVA data and the same in $AVXL Phase 2a data.

Picture first.

There seems to be a discrepancy from calculated ADAS-Cog/MMSE baseline numbers. By ADAS-Cog11 numbers the patients in $SAVA’s trial are a bit better off than the data on MMSE tests would suggests. It is difficult to interpret the data as these two scores can give quite varied results, nevertheless it is worth to mention it here, especially in the context of comparison with $AVXL data. As the sample data here is 50 one would suggest that the discrepancy would be smaller. The calculated data comes from this graphical calculator.

Source bilbliography [1]

In the $AVXL results there is a bifuraction of results in the same time frame. There is a group of 4 patients exhibiting therapeutic effect and group of patients declining as if there would be no therapeutic effect for them. Of course, we are talking about the high concentration cohort. Those who showed improvement are well documented by the company to have average baseline MMSE scores of 23 MMSE. This is compatible with the $SAVA numbers. The high concentration cohort has rather small sample size but the pattern exhibited by the bifurcation is telling of undoubfull therapeutic effect, though limited to certain population. In the illustration I have tried to determine the mean of the second latter group of patients but this is more or less impossible, that is why there is a question mark.

Dispertion and Therapeutic Effect

$SAVA data shows the dispersion to narrow by so few points on ADAS-Cog scale that it is insignificant (or due to my error in calculations and reading the slide). Calculated results are average of Delta 3.2+/-6.1 vs baseline score 16.6+/-7.7. It seems that Simufilam just rises all boats with well distributed terapeutic effect but nevertheless milder than $AVXL. I wonder how it would perform had it have the same mix of MMSE scores as Phase2a of Blarcamesine (MMSE 21). Let’s see the slide for $AVXL.

the same illustration as above.

The combined average and standard deviation for both groups of high concentration cohort is Delta MMSE -.46+/-4.3 or recalculated ADAS-Cog11 +2.7+/-19.2. Notice the very wide dispesion. This is due to bifurcation of response to Blarcamesine. When we calculate these for two separate groups they turne to be much different. The therapeutic effect group have Delta MMSE +3.5+/-0.7, notice very narrow dispersion about the average. This suggests very strong and uniform therapeutice effect. The other group basically follows the decline in the placebo. Reference [2] gives the rate for annual decline for large group of patients to be -3.4+/-3.7 MMSE. If we compare the other group’s declining annual scores ( -4.05+/-3.2 MMSE), the deteriration does not stand out from the annualized fall in scores in reference [2]. Recalculating the results for the group with therapeutic effect we get Delta ADAS-Cog=-9.7+/-1.8 (vs Simufilam’s Delta ADAS-Cog=-3.2+/-6.1)


Source: Bibliography [3]

The group who carries the therapeutice effect up to 3 years without decline is defined in “Group1”, let’s estimate probaility for a patient to be in this group.

Source bibliography [4]

The carriers of APOE3 alleles are 87% of the stricken with Late On-set Alzheimer’s Disease. There is certain amount of ambiguity here. Does it include e3e4 carriers, or these are e4 carriers and therefore excluded from the group. It seems that they are excluded, as we know that 75% of high concentration group is APOE4 carriers. This would make 35% of patients to be protected from farther decline over 3 years (combined probability of SIGMAR1 WT and APOE other than e4). If we include APOE e3e4 carriers we can see that that would cover additional 35% patients with considerably slowed decline after 70 weeks (~1.5 years).

There is always the question of interpretation what the companies say or what info they release really mean. $SAVA has reported that after 6 months of dosing, if memory saves me well, 93 or 96% of patients responded to Simufilam. At 9 months there is a visible deterioration in these numbers, if I interpret them correctly, as 33% still decline. This would indicate deterioration in the quality or strength of response. $AVXL has the data on 3 years of dosing, though on very small sample size, but then it has been partially validated with the genetic and other biomarkers.

The reference [2] concludes that due to distribution and dispersion of the testing scores for large number of participants, to have a truly validated insight into efficacy of a Alzheimer’s drug, at least 3 years of data collecting is necessary. Let’s have a look at their data supporting this conclusion.

For the first 3 years, the scores as they are presented in the graph can spontaneously become positive or increase vs. the baseline scores. By about 3 years time, this phenomenon atenuates so much that the judgement on the efficay of a drug can be made without excessive ambiguity.


  • Simufilam produces widely spread but mild therapeutic effect (vs. $AVXL)
  • Simufilam widely spread therapeutic effect becomes narrower with time (?)
  • There is very litle information on dropouts from Simufilam trial (<10%?)
  • $AVXL presents narrower therapeutic effect with greater strength peristing for 3 years
  • $AVXL has problem with uncertainty due to small sample size but holds data for 3 years
  • Combination of biomarkers validate the $AVXL data, creating groups whose members respond similarly
  • Simufilam needs to report at least for 18 months and $AVXL needs to release data on Phase2b/3






I have to confess that it took me some time to reach these conclusion. The number cruching takes few hours so please take it into considerations.


Make a one-time donation

Make a monthly donation

Make a yearly donation

Choose an amount


Or enter a custom amount


Your contribution is appreciated.

Your contribution is appreciated.

Your contribution is appreciated.

DonateDonate monthlyDonate yearly

4 thoughts on “Cassava Science Results vs. Anavex Life Sciences Projections $SAVA $AVXL

  1. You ask:
    Does it include e3e4 carriers, or these are e4 carriers and therefore excluded from the group.?

    The answer is in the paper by Anavex “A precisionmedicine framework using artificial intelligence for
    the identification and confirmation of genomic biomarkers of
    response to an Alzheimerโ€™s disease therapy: Analysis of the
    blarcamesine (ANAVEX2-73) Phase 2a clinical study”

    See section 2.5.0 where it says:
    When using a binary descriptor for genomic variants, the
    presence of one copy of the allele was considered sufficient to define
    the descriptor. For example, apolipoprotein E gene (APOE) genotype
    ๐œ€4 (APOE ๐œ€4/-) was considered โ€œTRUEโ€ if at least one copy of ๐œ€4 was

    Note it says the presence of even “ONE COPY” was considered sufficient. So the answer is e3e4s were included in Group1, as were e2e3s (though there may not have been any of these).

Leave a Reply