Cassava Sciences is under fire for irregularities in its research results for a new treatment for Alzheimer's disease. Photo / Ilana Panich-Linsman for The New York Times
Studies linked to Cassava Sciences, once a stock market favourite, have been retracted or challenged by medical journals.
A small biotech company that trumpeted an exciting new treatment for Alzheimer's disease is now under fire for irregularities in its research results, after several studies related to its work were retracted or questioned by scientific journals.
Cassava Sciences, based in Austin, Texas, announced last summer that its drug, simufilam, improved cognition in Alzheimer's patients in a small clinical trial, describing it as the first such advance in treatment of the disease. Cassava later initiated a larger trial.
The drug's potential garnered enormous attention from investors. Alzheimer's disease affects roughly 6 million Americans, a number expected to double by 2050, and an effective treatment would be lucrative. Cassava's stock soared, by more than 1,500 per cent at one point. The company was worth nearly US$5 billion last summer.
But many scientists have been deeply sceptical of the company's claims, asserting Cassava's studies were flawed, its methods opaque and its results improbable.
Families of some trial participants have said they see improvements. But critics noted that the trial reporting better cognition due to simufilam lacked a placebo group, and asserted that the Alzheimer's patients were not followed long enough to confirm that any improvements in cognition were genuine.
Some experts went further, accusing the company of manipulating its scientific results.
In response to the allegations, in December, The Journal of Neuroscience published "expressions of concern" regarding two brain studies authored by the company's chief collaborator, Hoau-Yan Wang, a professor at the City University of New York. One was co-written by Lindsay Burns, chief scientist at Cassava. The journal editors also noted errors in the images accompanying the latter study.
(An "expression of concern" indicates that the editors have reason to question the integrity and accuracy of a paper.)
In March, the journal Neurobiology of Aging attached an expression of concern to another publication about Alzheimer's authored by Wang and Burns, among others, that has been key to the company's treatment hypothesis.
The editors "did not find compelling evidence of data manipulation intended to misrepresent the results" but went on to list a variety of methodological errors. Both journals said they would wait to take further action pending an investigation by Wang's employer, CUNY.
Wang did not respond to requests for comment. The university declined to confirm or deny that an investigation was underway.
On March 30, another scientific journal, PLoS One, retracted five papers by Wang after a five-month investigation into "serious concerns about the integrity and reliability of the results," according to a spokesperson for the journal. Two of the papers, co-written by Burns, were about a brain protein targeted by Cassava's drug.
The New York Times contacted nine prominent experts for comment about the scientific underpinnings of Cassava's trials. All said they did not trust the company's methods, results or even the premise underlying the drug's supposed effectiveness.
Dr. Roger Nicoll, a neuroscientist at the University of California, San Francisco, said he was particularly angered that Cassava's work is partly funded by taxpayers. In all, the company has received more than US$20 million from the National Institutes of Health.
"This drug should not be put into patients. It should never have been. Never," he said. "The longer this goes on, the more outraged I am."
Cassava founder and CEO Remi Barbier said many of its critics were "bad actors" with financial conflicts of interest. He said the allegations of data manipulation were false, and noted that some were publicised by short sellers trying to drive down the company's stock price.
"They have gone, and continue to go, to unreal extremes to halt our progress," Barbier wrote in an email. "The effort to besmirch Cassava Sciences appears endless."
Cassava is a newcomer to Alzheimer's research. Until 2019, it was called Pain Therapeutics, and its product Remoxy — a gel form of the opioid oxycodone that was intended to deter drug abuse — was a favourite among investors.
But the Food and Drug Administration repeatedly rejected the drug and reprimanded the company for promoting Remoxy before approval.
According to Cassava, simufilam restores the normal shape and functioning of a protein called filamin A that becomes warped in the brains of people with Alzheimer's disease, and thereby slows dementia and improves cognition.
But Alzheimer's experts said they knew of no independent studies that supported this hypothesis or would explain the results.
"The overall conclusions with regard to Alzheimer's disease make no sense to me whatsoever," said Dr. Thomas Südhof, a Nobel laureate and neuroscientist at Stanford.
Cassava's theories "are not in the mainstream of the field, and to me, they seem implausible and contrived," he said.
In its papers, Cassava has said the theory is supported by strong evidence. "But, in fact, all the evidence seems to be from this lab," said Dr. Lawrence Honig, an Alzheimer's disease expert at Columbia.
The criticisms and retractions have dented Cassava's reputation. This month, the company told investors it had enrolled only 60 of the 1,750 participants it needed for its Alzheimer's trial. As of Monday, the company's stock was trading at US$25 per share, down from a high of US$135 last summer.
Cassava has indeed been the object of short-selling tactics. In August, two scientists filed a citizen's petition to the FDA, using a law firm that represents whistleblowers and citing "grave concerns about the quality and integrity" of Cassava's research. Its stock dropped by about 30 per cent.
Both scientists held a short position in Cassava's stock and profited from its decline, which undercut the credibility of their petition. Both have since divested themselves of those financial positions.
The FDA denied the petition "solely on the grounds that your requests are not the appropriate subject of a citizen petition," according to its statement.
But FDA officials also acknowledged the issues raised and said its response did not represent "a decision by the agency to take or refrain from taking any action."
A spokesperson told the Times that "the FDA, by policy, does not discuss possible or ongoing investigations." The scientists, Dr. Geoffrey Pitt of Cornell and Dr. David Bredt, a neuroscientist, declined to comment, as did the lawyer who represented them.
The Securities and Exchange Commission is investigating allegations that the company manipulated research results, The Wall Street Journal has reported. The National Institutes of Health is also investigating, Barbier acknowledged.
Cassava said in a securities filing in November that it was cooperating with "certain government agencies" that had asked the company to provide them with corporate information and documents for ongoing investigations.
"No government agency has informed us that any wrongdoing has occurred by any party," the company said.
The most striking evidence of simufilam's effectiveness was said to be an improvement in cognition in two-thirds of patients after taking the drug for a year. Some of the 50 patients had no neurological or psychiatric symptoms at all by that time, the company reported.
But this study did not compare the medication to a mock treatment or placebo, other researchers noted. All of the participants in this trial received the drug and were aware of it, and their reported improvement could have been a result of their positive expectations — a so-called placebo effect, which has been particularly strong in Alzheimer's trials.
For a drug to improve cognition scores, it would have to have the "extraordinary effect of actually reversing neurodegeneration," said Dr. Robert Howard, professor of old-age psychiatry and an expert in clinical trial design at University College London.
"Generally, in this area, we would regard slowing or stopping of cognitive decline as a win," he said. To claim that patients are actually improving, on the basis of small numbers, "at the very least is implausible," he added.
Cassava reported that initial analysis of a trial that did include a placebo group showed "highly anomalous" data, so it had the data reanalysed by an "outside" lab. The company later acknowledged that the second analysis was, in fact, conducted by its collaborator, Wang.
"If the data is suspect in key papers, and not just minor mistakes, you can't trust anything," said George Perry, a neuroscientist at the University of Texas at San Antonio and editor-in-chief of The Journal of Alzheimer's Disease, who serves on the boards of companies with other drug candidates. "It'll have to be independently validated."
Experts also pointed to several methodological oddities in the company's work. For example, Cassava has pointed to changes in the levels of certain molecules in cerebrospinal fluid, or CSF, as evidence of simufilam's effectiveness.
But the levels that the company reported are out of range for the testing method that was used, said Dr. William Hu, an expert on spinal fluid markers at Rutgers who works with a number of companies that develop such assays.
One marker of inflammation was much higher than is typically reported for Alzheimer's disease patients, he said. "There's a clear discrepancy there for those of us who work with CSF biomarkers," Hu said. "That type of discrepancy really raises questions in terms of the rigour as well as the reliability of these results."
Honig said the company's reliance on an outdated lab technique called Western blotting, used to detect proteins, was unusual because it does not have "the sensitivity or reproducibility or accuracy" of newer methods.
In another set of experiments, Cassava reported data suggesting that its drug was able to restore the shape of most of the filamin A protein in the brain — a dramatic recovery that Südhof, who consults for drug companies and serves on the board of Sanofi, questioned.
"There's just no possibility, no rational way this could happen," he said.
Elisabeth Bik, a leading expert in image manipulation, became aware of the company last summer after short sellers filed the FDA petition. In a series of Twitter posts, posts on her blog and on the website PubPeer, Bik has pointed to signs that she said show some results had been copied and pasted from other experiments.
"Those were of severe concern," she said in an interview. Based on the pattern of irregularities in images in multiple papers, she believes "it is highly likely that there was some manipulation going on."
Charles Spruck, a cancer researcher at the Sanford Burnham Prebys Medical Discovery Institute in San Diego, who has more than 25 years of experience with the Western blot method, said he believed the anomalies in those images could be the result of simple mistakes or vagaries of the technique.
Irregularities or errors in one or two images could be due to chance, but "when you see it again and again, it makes it unlikely that you could do it accidentally," said Dr. David Vaux, deputy director of scientific integrity and ethics at the Walter and Eliza Hall Institute of Medical Research in Australia.
Barbier dismissed the criticisms, saying the irregularities did not impact the company's research findings. "These background pixels have no impact on the data or its interpretation," he said, adding, "We stand by Professor Wang 100 per cent."
Vaux and others bemoaned the limitations of peer review in identifying mistakes or manipulation and said many scientific journals are reluctant to retract papers because of their fear of being sued, or damage to their own reputations.
"It's time for regulatory bodies to step in, as it seems that the peer-review process has taken it as far as it can," Hu said. "If the tide has shifted such that the science seems to not be there, I just don't see how the clinical trials can proceed."
This article originally appeared in The New York Times.
Written by: Apoorva Mandavilli
Photographs by: Ilana Panich-Linsman
© 2022 THE NEW YORK TIMES
Russell Vought was a key architect of Project 2025.