Simufilam was no better than placebo but the fat lady hasn’t sung yet

Overall Simufilam was no better than placebo in a large (403 given the drug 401 placebo) study but there is one more piece of data that I’d like to see before writing it off completely.

From the beginning, I was impressed with the results on 10% of a sample of 50 at 9 months, showing unprecedented clinical improvement — for details please see — https://luysii.wordpress.com/2024/09/27/cassava-sciences-9-month-data-is-probably-better-than-they-realize-2/

But we’ve been trapped by semantics, namely by calling Alzheimer’s disease, a disease we are unconsciously saying that Alzheimer’s is a single disease — https://luysii.wordpress.com/2024/02/29/what-if-our-most-common-assumption-about-alzheimers-disease-is-wrong/.  All sorts of diseases have been split once we knew more.  Just look at breast cancer.  This is the hope of customized medicine.

So what I’d like to see is the distribution of ADAS-COG12 and ADAS-ADL changes at study’s end (e.g. plot the score vs. the number of people showing that score, for both placebo and Simufilam.  400+ data points for each should produce a decent looking histogram/distribution plot.  If there is a bump in the Simufilam distribution (not present in the placebo) for people doing well, then there still might be something to the drug.  A good effect in 10% would certainly be washed out when averaged into the other 90%.

Bumps in distributions have a long history.  This is how some elementary particles were found.

Imagine that you had given an antibiotic to 400 people infected with all sorts of  different bugs and compared it to placebo.  There likely would be no difference in response.  A similar experiment with immune checkpoint blockade in 400 people with randomly selected cancers would have shown no effect.

Cassava Sciences has the data.  Hopefully they will allow the distributions to be published.  Alzheimer’s is a terrible illness, and no stone (distribution) should be left unturned.