P-values Fears And Jesus Visits The Placebo Arm of AVATAR Rett Syndrome Anavex2-73 $AVXL

This Blog Is Only For Educational Purposes. Do Not Trade On This Blog Alone. Do Your Due Diligence or Consult a Professional.

I was informed that what market fear is that …..

On Stocktwits I got following message.

Let us examine this claim….

We have to asks ourselves what is the p-vlaue and how is it calculated? For the null hypothesis to be true is that large enough number of the values of the sample (Dose cohort) should have high probability of falling under the distribution of the Palcebo arm. It is asumed that is this calculated probability below 5% or p-value .05 than the null hypothesis is false.

How is it calculated?

• First you calculate Z value for the Placebo distribution
• Z is proportional to the difference between Sample Mean (Dosed Cohort) and Hypothesis Mean (Placebo)
• Inversly proportional to Sample Standard Deviation and proportional to Square Root of Sample size n number.

Z=(Sample Mean – Placebo Mean)/(Sample Standard Deviation/Square Root(n))

The Object here is to get magnitude of Z large enough to so that when you check the value of probability of Z value in the Probability Tables of Normal Distribution the probability shall fall below .05 (5%). It references to the “extreme data” in the Placebo distribution .

Nota bene, the inclusion of the square root of as multiplier of means difference in consequence explains how increasing the sample size (dosed patients) increases value of Z so that p-value becomes smaller, ego with large enough n many more things become statistically significant. LOL.

Why RSBQ-AUC and CGI-I p-values are the same?

I have reread the PR and immediately was struck by the fact that CGI-I p-values and RSBQ-AUC p-values are the same. CGI-I has not been altered in any way, it is the raw data that had been used the same ways as in RS-001. RSBQ-AUC is transformed by multiplication from raw RSBQ data. If the increase between the means of Placebo arm and Dosed arm takes place and corresponding increase in Standard Deviation for the Dosed then the p-values can be unaffected by inclusion of time into the mix. Of course, if those increases are somehow proportional they might give the same results for CGI-I and RSBQ-AUC p-values. I prepared this illustrations to clarify the effect of time inclusion in the data by going to area under the curve for both the Placebo Arm and RSBQ raw. My assuption was that both of these distribution would be subject to this treatment.

As I said if these transformations make the means difference amd standard deviation of the sample somehow proportional than the p-values can stay true to raw data.

This transformation of raw data had the effect of increasing the values of the Effect Size in case of RS-002 and decreasing the Effect Size for RS-00 because it is a completely different calculation than p-value. I am not possesing any knowledge how the RSBQ-AUC accounts for the negative area under the curve showing progressive deterioration, or were there any girls who in the few weeks it took to run the trial deteriorated even further? If we take the Effect Size of the RSBQ-ACU for both RS-001 and RS-002 as the gold standard the separation between them suggests almost 4 times that large therapeutic signal in RS-002 Avatar. Is it credible to expect such increase? I think, yes, with 6 times increased dose almost 4 times increase in the therapeutic effect can be credibly expected. We are lacking the details how the area under the curve calculations were conducted hence the uncertainty in the market. In general, I can venture to say that this exercise increases the sensitivity to positive outcomes, their immediacy and duration. It might not alter the calculations for Statistical Significance but it does alter the Effect Size calculations. With that, I think that the fears of the market participants can be assuaged, if not entirely made baseless.

The PR informed us that ~72% in the Dosed Cohort were responders. This agrees with RS-001 results and the knowledge about the SIGMAR1 receptors and protein, and its interaction with Blarcamesine. We know now, that the most important metric determining resonse to Blarcamesine is the level of SIGMAR1 protein in the blood, the other, the lesser one albeit, being carring the wild type SIGMAR1 (unmutated). The blood levels determin the magnitude of the responce. Individuals with low level of SIGMAR1 protein might not even respond to the drug. Fortunately, these people are not numerous. This might be the reason why we need 50mg dose to have a therapeutic effect in Alzheimer’s where these patients produce the least of this protein, relatively. The ability of Blarcamesine to prevent deterioration into Alzheimer’s before the blood levels go significantly down with the incipient disease might be the key for having it the status of general health preventative. The recognition of the fact that the SIGMAR1 protein is the most important factor in the response is a recent addition to the pracitce of precision medicine by Anavex.

The 38.5% of the Placebo arm having both higher scores of RSBQ and CGI-I are surprising as if Jesus visited the arm and girls suddenly recovered on their own. The anchoring the RSBQ to CGI-I was somehow to prevent the possibility of false RSBQ scores. Nevertheless, it seems the subjective nature of these measures of progress and healing got better of the investigators. At least, so it seems on the surface. It is possible to have erroneous RSBQ or CGI-I scores somewhere around the baseline but the higher incorrect improvement scores are rather hard to be accepted even in very subjective questionnaires as the gap between the scores and the progress of a patient becomes too conspicuous. This was the purpose of anchoring the RSBQ scores to CGI-I, a check on too hopeful caregivers.

It is now known that it was FDA who directed Anavex to change the description of RS-002 Avatar to “phase 3”. Notwithstanding the knee-jerk reaction of the market public to new measure, the change in the status is adding an upbeat story to the prospects of Blarcamesine (Anvex2-73) for the approval.

In conclusion I can only point to the EffectSize in RSBQ-ACU between RS-001 and RS-002 AVATAR as the measure of increased efficacy. But, I also believe that the p-values scores has not been made even lower by the fact that ~38% patients in the Placebo arm has shown improvent. Was this improvement all due to subjectivity of the process of collecting data. Possibly, but that is nature of medicine where the caregivers and practitioners evaluate change through the lense of their experiences and biases. Did the news of Anavex success with Blarcamesine against Rett Syndrome affected these scores? Good news carries far.