Skip to Content

A Reply on Academic Alzheimer’s Research

One of the authors of a paper I commented on has shown up in the comments section to that post, and I wanted to highlight his reply out here on the front page of the blog. Here’s J. R. Brender, from the Michigan side of the authorship:

Hi. I appreciate the comments the given about the paper. As one of the authors of the paper (with Ramamoorthy on the NMR part), I would like to clear a few things as time permits.
@ Derek An uncharitable view would be that they have also taken aim at the year 1995, which is about when all three of these ideas were also being worked on for AD.
All three are still be working on and are in (mostly mixed or unsuccessful) clinical trials. Vitamin E in particular went through a phase III clinical trial for mild to moderate Alzheimer’s with mixed results http://www.alzforum.org/news/research-news/trial-suggests-vitamin-e-protects-function-mild-alzheimers
To be fair, none of the other hypotheses have much support either.
@19 from Bob “The paper only uses the word drug once, in the context of including “drug-likeness” as a designed property, and therapeutics once in the conclusion.”
Correct. I wasn’t aware that at any point we claimed that this was a therapeutic or even a lead compound for a therapeutic. The discussion about drug discovery in academia vs. industry, while interesting, is in my opinion somewhat off-topic. A more relevant question is whether it is worth investigating one compound with a detailed approach (which you are going to have do if you want in any kind of mechanism based inhibitor) or try a high-throughput non-mechanistic approach phenotypic screening. I’m agonist on this point and i think both are viable (or a maybe both non-viable options). Large scale phenotypic screening for Alzheimer’s is going to exceed the resources of academic lab. Based on the amount of money spent on pharma and the current success rate, I suspect its been tried on some level and failed at relatively early stage.
@21 from JSR “If the end result of months or years of work by 14 authors and almost as many sources of funding…
The non mass spec work (the bulk of the paper) was supported by a single R21 and a private foundation grant of which this paper is a small part.
@21 from JSR “not ready to publish, especially not in the once hallowed pages of JACS.”
“MedChem journals likely would have asked that more work be done to answer some of the same questions Derek raised.”
@35 “I’d add ‘who partners with someone who knows how to build / run relevant screening assays”
There are no relevant high-throughput screening assays for amyloid inhibition in common use. This point in particular I would like to stress and is the reason (as one of the commenters guessed) we left some of the expected the out of the paper. A very high percentage of the papers in JACS and J. Med. Chem on amyloid inhibitors consist of a set of compounds with only three sets of data. A high-throughput thioflavin T assay to measure amyloid inhibition, a set of EM images to show amyloid disappearing, and an MTT assay. There is very rarely any kind of pharmokinetics often not even to the extant of calculating drug-likedness (if you don’t believe me look up amyloid inhibitor on basically any journal including the med chem ones). Though usually not acknowledged, ThT assay has a very high false positive rate since ThT generally binds at the same site as the inhibitor. Although not in the paper, we have shown this is true for the compound in the paper and many others. EM images suffer from multiple issues due to bias in binding to the grid, selection bias in sampling etc. The MTT assay has a sensitivity problem as suggested, and is not ideal for amyloid for a variety of other reasons.
The conformational antibodies sometimes used are also pretty non-specific, although this is only occasionally acknowledged in the literature. The end result is a lot of compounds with apparently quantifiable information that really isn’t. There is no information on where the compound binds and what it binds to (amyloid beta is a mixture of many different, rapidly equilibrating species even when it is claimed to be in a single form).
If you have experience in high-throughput screening, I urge you to team up with an amyloid person (there are many amyloid specific factors that need to be considered). The field desperately needs you. Also, if you know of compounds for which reliable PK data has been obtained let me know (jbrender at umich.edu). I am compiling a database of amyloid inhibitors and an discouraged at what I am finding.
Our goal in the Ramamoorthy NMR lab in particular was to take a single compound and analyze its binding on low MW and fibrillar Abeta , using a labor intensive approach with the aim of developing a future high throughput fluorescence based approach to isolate specific interactions with different Abeta species (some unpublished progress has been made on the fluorescence work).
The study is only one of handful that have identified specific interactions in terms of a structure of Abeta (the new structure we have is the only high-resolution structure not in detergents in organic solvents). ML binds at a specific site on the structure, and looking back at the literature, you can see a similar binding site for many of the compounds in the literature. That to me at least is interesting.
In conclusion, it is not a complete story by any means, just a progress report. But a complete story with Abeta and Alzheimer’s is going to take a very long time.

Note: I’m turning off comments here, so they can continue to thread in the previous post. I may have some more to say on this myself, but I’ll leave that to another entry.