In September 2022, Food and Drug Administration (FDA) officials arrived at an imposing, glass-dominated research complex at the City University of New York (CUNY). They planned to review records and practices in a lab run by pharmacologist Hoau-Yan Wang, who had been involved in clinical tests for an experimental Alzheimer’s drug. Two years earlier Wang and his colleagues had analyzed samples of patients’ blood and cerebrospinal fluid from a key trial of the drug, called simufilam, developed by Wang’s longtime collaborator Cassava Sciences.
At first a CUNY official denied entry to the inspectors and security guards escorted them off the premises. When CUNY granted them access to the lab 2 days later, the FDA staff found a litany of problems with that old work.
Wang, the inspectors noted, never performed routine calibration of his equipment or completed verification experiments to ensure the tests were “accurate, sensitive, and [conducted] with appropriate precision.” And, they reported, he used improper statistical tests that “resulted in inaccurate determination of sample concentrations.” On a 10-point scale, with 10 being the worst score, the findings amounted to an eight, says Jerry Chapman, a senior quality expert at Redica Systems who reviewed the inspection report for Science. (Redica has collected and cataloged nearly 1 million such reports and guides clients on FDA compliance.)
“Essentially, the FDA said, ‘You can’t trust any of this.’ But the company used the results” nevertheless as evidence of the drug’s efficacy, says University of Southern California neuroscientist Lon Schneider, who also reviewed the report for Science. Indeed, Cassava publicly reported the trial’s finding as positive and the results helped pave the way for phase 3 trials of the drug, which are now underway.
The FDA report, disclosed under the Freedom of Information Act, raises new questions about the credibility of claims by Wang and Cassava about simufilam, a drug that has long been under scrutiny. (The report was requested by a person trying to profit if Cassava’s share price declines, and shared with Science by his colleague. They are among several so-called short sellers whom the company has sued for defamation.) Because of previous questions about basic and clinical research supporting the drug, the company faces ongoing federal investigations by the Department of Justice and, reportedly, the Securities and Exchange Commission, class-action lawsuits by investors, and calls by some in the scientific community for ongoing simufilam trials to be halted.
FDA and CUNY declined to comment about the report. So did Wang, through his attorney. Cassava said in a statement that the CUNY lab was not required to be compliant with FDA’s Good Laboratory Practices (GLPs) for its “exploratory research.” (The analysis of trial samples is considered clinical, not exploratory, research under FDA’s definition, however; moreover, GLPs don’t apply to clinical work.) The company also said the work from Wang’s lab had been “validated” by subsequent academic work elsewhere.
ADVERTISEMENT
Wang’s earlier preclinical work underpinning plans to test simufilam in people was already under fire. For example, last year, a CUNY panel concluded that 20 of his papers—some co-authored by Lindsay Burns, a leading Cassava scientist—contained “evidence highly suggestive of deliberate scientific misconduct by Dr. Wang,” according to a report obtained by Science. University investigators could not prove their suspicions, according to the report, because of “long-standing and egregious misconduct in data management and record keeping by Dr. Wang.”
The newly released FDA inspection report suggests Wang’s work on clinical trial samples was also flawed. His measurements were crucial to Cassava’s hopes for simufilam, a compound that is supposed to counter the effects of a misfolded protein in Alzheimer’s disease. In May 2020, the company had announced disappointing results for a phase 2 study of the drug. It failed to lower levels of key markers of the disease: tau, a protein linked to brain cell “tangles,” and beta amyloid, which forms infamous sticky “plaques” in the brain. The company’s stock lost three-quarters of its value overnight.
Hoau-Yan Wang CITY UNIVERSITY OF NEW YORK
In a press release a few months later, Cassava rejected those disappointing results, suggesting a contractor had analyzed patient samples improperly. The company offered better news from a 28-day study of 64 Alzheimer’s patients, including a group given a placebo. This time, simufilam sharply reduced tau, beta amyloid, and other biomarkers. Cassava’s share price immediately more than doubled, and some biotech analysts became more optimistic about the drug’s prospects.
It is not clear why FDA waited 2 years to investigate how Wang handled those samples—or why they only looked at that study, given that Wang tested samples for other Cassava studies. Whatever the reason, the inspectors found concerns that went beyond poor laboratory practice. A key problem, they wrote, involved how Wang handled “outlier” data. For each patient sample, he ran three tests for levels of tau and other biomarkers, then discarded any reading that diverged sharply from the others. That practice is accepted, but the criteria for doing so should be set in advance, the FDA inspectors noted. Wang told investigators that he didn’t use clearly defined criteria. Instead, he “based it on his judgment.”
Wang agreed he had made errors, according to the FDA report, and rechecked the rest of the data, finding no more anomalies. “However, our independent review of the same spreadsheets identified several additional examples affected by similar errors,” the inspectors wrote, noting 15 such cases. Wang “agreed with our findings,” they added.
The CUNY scientist later told the officials that he corrected all the errors and emailed updated data to a person whose identity was redacted in the report—presumably someone at Cassava. Cassava declined to comment on why it had not updated reports on Wang’s work with the corrected figures, saying that they “resulted in no material change to the data.”
Even the corrected figures remain suspect, according to the report, because of Wang’s lack of regular equipment calibration and other lapses. Echoing the CUNY report on problems in Wang’s basic studies, FDA inspectors said the scientist had not retained original Western blot films, needed to audit the accuracy of some biomarker results. Wang said he discarded them as “cumbersome to maintain,” the report noted. Nor did he maintain records about how blood and fluid samples were stored, tracked, or tested. His lab was unlocked and computers unprotected by passwords.
CUNY didn’t reply to Science’s requests to explain why its guards initially denied FDA inspectors access to Wang’s lab. Michael de la Torre, chief executive of Redica, called the rejection “extraordinary” for a U.S. facility. Nor would the school say whether Wang had been reprimanded or penalized in any way for the findings by the FDA inspectors.
Cassava’s critics say the FDA findings reinforce concerns that human tests of simufilam are not well supported. “This report extends a pattern of data that lack rigor and reliability, and further undermines my confidence in any clinical trial results from this program,” says neuroscientist Matthew Schrag. The Vanderbilt University professor, who previously identified what he said were problematic data from the company and Wang, reviewed the FDA report for Science, working independently of his institution. “As I’ve said before, I think that the simufilam clinical trials should be shut down,” he concluded.
FDA would not say whether its inspection findings would affect the agency’s review of the phase 3 trials of simufilam, normally the last set of safety and efficacy tests before the agency decides whether to approve a drug. Schneider says the FDA document shows that nothing can be concluded from the tests run in Wang’s lab. It is “a damning report,” he says.
