Memory Games, Part III: "Notes From a Molecular Biologist"
Cassava Substitutes Activism for Data
There’s a troubling pattern that has emerged from the wreckage of the attempts by Cassava Science’s two chief research scientists—Dr. Lindsay Burns and Dr. Hoau-Yan Wang—to explain signs of potential fabrication in their published work. As a sort of corollary to what we saw in the last part of this series, where we got an inside look into the inner workings of an editorial retraction of Wang’s work, this post will focus on a nexus of all things pro-Cassava: the investor blog ad-science. Ad-science is maintained by an enterprising group of Cassava shareholders as a way to push back against the skepticism of Cassava’s research and cast whatever aspersions they feel are necessary in order to protect their collective investment thesis. In a vacuum, it’s a perfectly fine diversion by a group of people who desperately want to believe that they’ve stumbled upon the next penicillin. But (thankfully), we aren’t in a vacuum.
We get a glimpse of the initial exchange between these investors and Cassava researchers in a FOIL document, which shows that they reached out to Wang shortly after the initial fraud allegations documented in the FDA citizen petition. They ask a series of clarifying questions about the possible mechanism of action for simufilam, acknowledging that Wang may not be able to respond, assumedly due to the backdrop of investigations into his work. That’s all perfectly fine, and Wang decides to forward the email on to Burns for handling—something that happened quite a few times in the FOIL document.
Dr. Burns decides to reply to the investors, which again is not troublesome on its own, as it’s mostly just context and reflections on the published research. If the interaction ended here then it would hardly be worth mentioning, let alone devoting an entire post to.
It does not, however, end there. After this exchange, ad-science actually becomes the primary line of defense against fraud allegations for Wang and Burns. We saw a glimpse of this in Wang’s last-ditch attempt to dissuade Alzheimer’s Research and Therapy from retracting his article, where he makes representations to the editors that the authors and contributors to ad-science, in particular Dr. Charles Spruck, are independent experts who can speak to the quality of the research with no confounding interest. Of course, we’ve seen already that this is not the case, and Wang knows it. And it’s not a one-off occurrence.
In November 2021 the Editor in Chief of IBRO’s Neuroscience, Dr. Juan Lerma Gomez, was planning to publish an Expression of Concern (EoC) for an article by Wang and Burns after reviewing concerns raised by various scientists and image sleuths. The EoC was actually so close to publication that it had been assigned a DOI by the journal, but it was pulled at the eleventh hour. We do not know with certainty all the factors that led to this decision, but we do know via FOIL documents that the ad-science blog was wielded by Burns as a direct response to being notified that an EoC was pending publication.
Recall the bombshell report by Charles Piller covered in the first part of this series, which broke the story about alleged fraud in the work of Alzheimer’s researcher Sylvain Lesné, contained the same signatures of manipulation that Matthew Schrag found in work by Wang and Burns. The allegations around Lesné’s work produced a chorus of disappointment and consternation from established Alzheimer’s researchers (see comments from researchers at the link), without, to my knowledge, any serious suggestions that the allegations may be unfounded. The case remains under investigation, and there is a distinct—if seemingly tiny—possibility that the issues could be resolved, but there appears to be no doubt in the community that what Schrag discovered is in fact a series of serious issues that need to be addressed by the author(s). What Burns is telling Lerma in her email is that there is a reliable independent source which contends that the issues raised do not even need to be taken seriously in the first place.
An editorial note published by Neuroscience shortly after this interaction contains some “original” data that made its way from Wang to Lerma at some point, which apparently satisfied the journal. These “originals,” however, contain the same background correlations that caused Alzheimer’s Research and Therapy to retract Wang’s article. To date, the issues found in the editorial note have not been acknowledged by Neuroscience.
There’s another place where ad-science makes a surprise appearance, this time a bit more out in the open. Schrag’s work implicated an influential article published by Wang and a handful of co-authors in The Journal of Clinical Investigation (JCI), though an authorship note explains that Wang and Dr. Konrad Talbot should both be considered lead authors on the paper. Wang has yet to respond to any issues raised by sleuths on PubPeer, but Talbot recently responded after Piller’s article shined a light on the 2012 paper. To summarize Talbot’s posts succinctly—JCI looked at the data when they published the paper; the data is no longer available; but concerns are invalid because of the theoretical explanations covered in the ad-science blog. For its part, JCI has consistently maintained that they are and unable to verify the findings internally. You can check high resolution images from the paper yourself on PubMed Central and compare with what is documented on PubPeer; generally, you’d need to do some forensic operations in order to spot the kinds of issues documented, but you can actually spot some of these issues by eye when zoomed in on the high-res images!
To be clear, there isn’t anything wrong a priori with investors or scientists coming to someone’s defense when accusations of fraud are made. The big, big problem here is that all three of the cases covered to this point require production of original data by the authors, and no actual raw data has been produced. Instead, the authors are choosing to try and discredit the accusations using an “independent” source that they’ve been in direct communication with. And hey, it appears to be working! For example, a closer look at internal publications from The Journal of Clinical Investigation finds them acknowledging that reviewers and editors at the journal have been missing cases of likely fraud, while citing papers by the very same authors who raised the concerns about this article in the first place—yet their response to Piller’s article (and to my emails to them directly) finds them flummoxed and unconcerned. There’s even a JCI paper from 2006 that presents erasure of background noise in Western blots as evidence of fraud—precisely what we’re seeing in the Wang and Talbot paper! Honestly, it boggles the mind a bit. Perhaps these journals will eventually explain how they landed at their decisions, but in the meantime it seems we’ll have to keep searching for raw data somewhere else.