Captivity of Information

Still Searching for Cassava's Fabled Data

The tempest surrounding Cassava Sciences has quieted somewhat—their full Phase 2 trial results raised more questions than answers, but a rudimentary subgroup spin and a somewhat baffling analysis by a consulting statistician (who is apparently sworn to secrecy) have served to keep things afloat while we await news from the CUNY investigation that is now approximately 500 days beyond the dictated timeline. Setting aside for the moment that Cassava’s Phase 2 trial essentially failed its own flawed, pliant design (falling short of statistical significance against multiple external control groups of Cassava’s own choosing), or that its primary endpoint continues to be mischaracterized by the company as “safety only” despite clearly listing three primary efficacy endpoints on the clinical trial website, there really isn’t a whole lot for Cassava to say between now and the release of the maintenance study data later this year.

While the Phase 3 trials and the maintenance study press on, and patients continue to ingest an experimental medication for which the sponsor did not deign to make any concrete assessment about whether or not the Phase 2 trial results met the pre-specified endpoints—simply calling the results “positive” and omitting claims of statistical significance—critical questions about the preclinical research linger. If you aren’t already familiar with the nature of these questions, then I encourage you to read the four part series that I wrote here called Memory Games. Suffice it to say that there is a whole lot of suspiciously unavailable and undisclosed data, representations of which made their way to a platoon of scientific journals, with each seeming to decide on a completely distinct course of action from the others. Among the decisions were forced editorial retraction; retraction by agreement; clearance with problematic data published leading to eventual reversal of course; clearance with data published containing blatant issues defiantly ignored by the journal; clearance with refusal to publish raw data; expressions of concern that confusingly condone admitted critical errors in every aspect of the experiment; and finally outright refusal to acknowledge that there are any issues to begin with.

The public has, for the most part, been kept in the dark about how these journals reached their dissonant conclusions. We’ve also been kept in the dark about the status of the ongoing investigation at CUNY—which is, again, over 500 days past its deadline—or at HHS ORI, which is likely the nexus of the federal investigations that Cassava has disclosed in their SEC filings. I have already covered in some detail the reasons why sound preclinical research is so critical for any evaluation of which drug candidates are worth the inherent risks that come with ingesting an experimental substance. There is a somewhat technical argument for this involving Bayesian inference and the base rate fallacy (see prior article), but I hope that people’s ethical intuition would also help them recognize how unacceptable this situation really is. If you’re a fan of Matt Levine and his excellent newsletter about modern finance, then it has probably been impressed upon you that public markets operate primarily on the basis of disclosure—a public company is tasked with disclosing anything and everything material about its business and the risk factors the business faces. “Everything is securities fraud,” Matt likes to say, but only insofar as it wasn’t disclosed to investors. Medicine works in a similar way, except in medicine it’s called informed consent. It is similarly broad and fungible, but the core tenet remains—you are allowing people to take very real risks that will likely have significant effects on their livelihoods one way or another, and you have a responsibility to these people to disclose as much information as possible to help them arrive at a sound decision.

Let’s be clear, there are likely not any laws being explicitly broken here. And there would be questions about whose responsibility it is to disclose what even if there were. But there are impartial(ish) third parties who have reviewed the data—much like, say, an auditing firm would review a company’s finances—and that one might expect to feel a sense of ethical responsibility for disclosing, or at the very least making attestations about, the science underlying a drug with over 1,000 people actively participating in clinical trials. They apparently do not feel this responsibility, or at least not enough to outweigh considerations regarding the researcher’s and company’s reputation and good standing, or the stock price of the litigious sponsor.

There is, in this circumstance, a potential remedy to this unfortunate situation that we’ve been mired in for the better part of two years—Freedom of Information Law. Before we dive in, I should note that I am not a lawyer, and that everything that follows involves my interpretation of the law and of lawyers’ communications as a layperson reading it in plain English.

FOIA Requests at the Federal Level

The first thing that we can do is ask the myriad federal agencies for records pertaining to investigations of Cassava Sciences. I asked the SEC, and was met with what’s called a 7(A) exemption, meaning that they have investigation records that they will not release because they could reasonably be expected to interfere with potential enforcement action. That has been a consistent response by the SEC over the last six months and there’s likely little more we can learn from them at the moment.

I also asked HHS, and was issued a Glomar response—the existence of an investigation into research misconduct can be neither confirmed nor denied. As the HHS FOIA Officer explained, there are competing considerations of privacy and public interest that must be balanced when deciding whether or not to release records that pertain to official investigations of individuals (or even to acknowledge that an official investigation is underway). It may not feel quite right that the Public Health Service would weigh the “substantial privacy interest in not being associated with an allegation of research misconduct” more heavily than the public interest in knowing whether or not the science forming the foundation of an experimental drug currently in active clinical trials is under formal investigation by the agency. But, as the officer explains, the FOIA “focuses on the citizens’ right to be informed about ‘what their government is up to.’” They continue that “in this case, it is unclear how, even if records responsive to your request existed, information regarding a single research misconduct investigation would ‘contribute significantly to the public understanding of the operation or activities of the government.’” I mean, I suppose I get it—the (theoretical) investigation is about determining whether or not there was research misconduct, and it’s not the job of HHS to regulate clinical trials, the FDA is the one that is supposed to care about clinical trials. So the HHS can sort of turn a blind eye to how disclosure of an ORI investigation might serve the public’s interest insofar as federal investigations reinforce the legitimacy and severity of the allegations, because regulating about the impact that their investigation may have on public health is not explicitly part of their job description. I guess?

Anyway, there are other federal agencies that almost certainly have pertinent records, but none are likely to have the data and are even less likely to be disclosed (or acknowledged) via a FOIA request. These will all undoubtedly be fascinating records to peruse when the investigations conclude, but they’re a dead end until then.

New York Freedom of Information Law Records—Getting Somewhere

My requests to CUNY for documentation pertaining to their ongoing investigation of Dr. Hoau-Yan Wang were met with similar denials based on privacy considerations and the argument that implications regarding public health and safety stemming from their investigation are not in their purview. There does appear to be a foundation of New York FOIL case law that is quite clear about exempting disclosure for investigations that have not yet reached a conclusion, though I’m not sure that any of these precedents deal with an investigation into something that may constitute an ongoing risk to the public which in a sense requires the expertise and access of the investigators to potentially mitigate. Still, I am most certainly not a lawyer and challenging exemptions based on this theory would require hiring one and filing a lawsuit.

There is one last thing, though, and it is no small thing. Dr. Wang is an employee of the government in New York, and the email records of government employees pertaining to all official duties fall under the domain of New York Freedom of Information Law. You may simply write the City College of New York Records Access Officer an email and ask them to search the email records of any employee at the college. There are some guidelines about how specific you need to be in describing the records you’re seeking, and these records may still be exempt from disclosure for various reasons, but the New York Freedom of Information Law is built upon the presumption of access to all official records, including emails. Emails pertaining to the internal CUNY investigation will almost certainly all still be exempt, but emails exchanged with journal editors that asked Dr. Wang to provide them raw data so that they could decide whether further action was warranted at the journal (e.g. Erratum, Expression of Concern, retraction) are definitely part of the official job description of a public research faculty member. These records are not part of ongoing investigations by a government body in the State of New York (or, they may be, but that isn’t where they originated). They should, by and large, be accessible.

We have seen a trove of emails disclosed from these types of FOIL requests already, which is part of why we know that, for example, The Journal of Neuroscience spent about an hour signing off on “raw data” in the form of PowerPoint slides that they changed their mind about after image sleuths were allowed to see them. Or that the inquiry phase of the CUNY investigation found the allegations substantial enough to recommend a full investigation (it’s surprising that we even know this, and that that particular email was disclosed at some point by CUNY). Or that Dr. Wang was using a Cassava investor blog—while representing it as impartial—to lobby journal editors against punitive action. Or that the hard drive with the original data “melted.” On and on.

I have personally filed a series of requests for additional records that I believe exist and should contain either the raw data that was provided to various journals, or the reasoning that the journal accepted in the data’s stead. I don’t know what, if anything, will end up being disclosed, but I was informed that a large volume of responsive records are being examined and a final disposition is expected within the next couple weeks. Meanwhile, there is one journal in particular for which I have received records already, and it raises some serious concerns.

Missing Records at Neurobiology of Aging?

I requested all emails exchanged between Dr. Wang and the Editor-in-Chief of Neurobiology of Aging, and received a document which I believe had likely been prepared during prior FOIL requests (given that these emails made the rounds months ago). You can read them at the link. They are a short exchange that begins with the Editor-in-Chief asking Dr. Wang to address various concerns, and ends a few emails later with “Dear Dr. Wang: Thank you very much for this quick reply. I look forward to resolving the issues raised.” And that’s the end of it. Now, we know from the Expression of Concern attached to this paper that “these issues were conveyed to the authors, who provided a detailed response, including images of relevant uncropped western blots and photomicrographs, as the editor requested.” So the data was definitely transmitted to the journal. And we have email records of the Editor-in-Chief awaiting said data. After asking the CCNY Records Access Officer to do multiple searches for additional records, no additional emails were located.

It is quite clear, I believe, that the Editor-in-Chief received official records from Dr. Wang, whose lab is the place where these records would reside and who was already in correspondence with the journal, and that these records should exist and be accessible via New York Freedom of Information Law. So where are they? If they were sent from personal email, that would be an issue—New York officials have famously run afoul of this guideline while trying to keep certain interactions off of the public record. Government business, even if conducted through personal email, needs to be entered into the FOIL system. I asked the current Editor-in-Chief at Neurobiology of Aging (the one that originally communicated with Dr. Wang is now “emeritus”) if they could provide clarity about who provided the data, but they have not responded to my requests.

This is eerily reminiscent of my exchange with the Editor-in-Chief of The Journal of Neuroscience more than a year ago, where I was informed that they “contacted Dr. Wang directly and were referred to a co-author,” and that “Cassava sent [The Journal of Neuroscience] the blots.” And yet, Cassava was very clear in stating that “Cassava Sciences does not have its own laboratory facilities. We use other people’s labs. For this reason, we don’t have the original films or images for the Western blots in question.” None of these things add up. They should, and I will continue to do the math until they do.