After more than a decade of controversy surrounding a 2001 clinical trial led by a Brown professor on the effects of the antidepressant Paxil, a reanalysis of the original data claims the drug is no more effective than a placebo and that the initial study underplayed the drug’s potential to increase suicidal thoughts among adolescents.
Since its publication, the initial paper has been alleged by some scientists to be fraudulent and ghostwritten, and both Professor Emeritus of Psychiatry and Human Behavior Martin Keller and the University have been criticized for their involvement in the study.
The reanalysis, published in the British Medical Journal Sept. 16, contradicts the results of the initial clinical trial — led by Keller and commonly referred to as Study 329 — which concluded Paxil was effective and safe for the treatment of adolescent depression. The study’s critics have cited the new data as evidence in support of their long-standing doubts.
But in an email to The Herald, Keller maintained the integrity of his team’s original conclusions about Study 329, citing two experts unaffiliated with the study who say the new analysis does not render the initial findings obsolete.
“This is probably the most heavily investigated incident of academic research corruption since 2000,” said Paul Thacker, a former staffer for Sen. Chuck Grassley, R-Iowa, who helped lead investigations into Paxil and Study 329 in 2008. “This thing has been turned upside down and inside out.”
Back to the beginning
In July 2001, Keller and 21 listed co-authors published the results of Study 329 in the Journal of the American Academy of Child and Adolescent Psychiatry. The article disclosed that GlaxoSmithKline, the manufacturer of Paxil, sponsored the study.
Though Paxil had been approved by the Food and Drug Administration for adults, GSK wished to expand the market to adolescents through off-label prescriptions — a practice in which physicians prescribe an FDA-approved drug for other non-approved uses, said Leemon McHenry, an early critic of Study 329 who worked with Jon Jureidini, a co-author on the reanalysis, to publish four papers on the topic.
In 2002, the year after the study’s publication, 2 million prescriptions for Paxil were given to children and adolescents, said David Egilman MD ’78, clinical professor of family medicine, adding that the study is likely responsible for the increase in the prescription of antidepressants to adolescents, which occurred during this time period.
But throughout the following decade, GSK was sued by numerous families with children who were prescribed Paxil, as well as by the attorney general of New York and the U.S. Department of Justice. In 2012, after being sued by the Department of Justice, GSK settled for $3 billion — the largest settlement in corporate history at the time.
The series of cases brought about the release of a host of internal GSK documents, based on which critics alleged that Study 329 was ghostwritten and that the study’s results were misreported to demonstrate the drug’s efficacy.
The litigation also led to the release of the study’s raw data in 2013.
“They selected only the positive data, and they ignored all the negative data,” McHenry said.
Study 329 was not the first GSK study to test Paxil’s effect on adolescent depression. Two prior studies — Study 377 and Study 701 — failed to show that Paxil was more effective than a placebo, according to GSK’s report synopses.
In a 1998 internal GSK “position piece” on Studies 329 and 377, the company states that its “target” was “to effectively manage the dissemination of these data in order to minimize any potential negative impact.”
In February 2001, five months before the publication of Keller’s paper, a GSK internal email chain reads, “Originally we had planned to do extensive media relations surrounding this study until we actually viewed the results. Essentially the study did not really show Paxil was effective in treating adolescent depression, which is not something we want to publicize.”
But five months later, a July 2001 letter from Zachary Hawkins of Paxil Product Management to all Paxil sales representatives states the drug “was significantly more effective than placebo,” adding that the drug “demonstrates remarkable efficacy and safety in the treatment of adolescent depression.”
Critics allege the paper was ghostwritten by Sally Laden, an employee of the writing company Scientific Therapeutics Information, which was hired by GSK.
Letters between Keller, Laden and GSK released during court proceedings show that the three communicated about drafts of the manuscript.
“You did a superb job with this. Thank you very much. It is excellent. Enclosed are rather minor changes from me,” Keller wrote in a letter to Laden, who was mentioned in the paper’s acknowledgments for her “editorial assistance” but was not listed as one of the 22 authors.
Regardless of who technically wrote the paper, GSK’s influence on the analysis and writing process is unethical due to its corporate interest in the product, McHenry said.
Keller and his colleagues have denied both the accusations of misreporting and ghostwriting since they were first made. In a letter responding to the reanalysis’ publication, Keller and eight co-authors maintain that their study was accurate in reporting the drug’s efficacy, adding that “the goal of the study designers was to do the best study possible to advance the treatment of depression in youth, not primarily as a drug registration trial” and that the report was “authored mainly by the academic investigators with industry collaboration.”
This process is often the norm, wrote Steven Rasmussen ’73 MD ’77, chair of psychiatry and human behavior, in an email to The Herald. These types of studies “are almost always initiated by companies that have a vested interested in bringing new treatments to the market.”
Judgments about the match between data and conclusions are part of a journal’s peer-review process, Rasmussen wrote. “The paper went through a peer-review process and was accepted at a major journal in the field,” he wrote.
A second time around
In June 2013, the reanalysis began as part of the British Medical Journal’s Restoring Invisible and Abandoned Trials initiative, a project that “calls for third party authors to publish or republish unpublished and misreported clinical trials,” The Herald previously reported.
Upon the release of Study 329’s data in 2013, a team of psychiatrists and researchers began reanalyzing the 77,000 pages of data.
The published reanalysis differs from the original paper in two major ways. First, it concludes that Paxil is not effective for treating depression in adolescents, and second, it places a heavier emphasis on Paxil’s side effects, the harms of which the team claims the initial paper ignores.
In the initial paper, researchers included suicidal thoughts under the term “emotional lability,” effectively reducing the rate of suicidal ideation reported, said Mickey Nardo, a member of the team and author on the new paper.
In their response letter to the reanalysis, Keller and his colleagues raise a number of objections. The advancement of science and methodology over the 14-year gap between the two papers should not be ignored, they wrote, adding that the reanalysis team analyzed Paxil’s side effects using methodology that was not available in 2001.
The original authors also take issue with what they see as discrepancies in the reanalysis team’s approaches, including a lack of an explicit explanation of what the Restoring Invisible and Abandoned Trials initiative “recommendations” for reanalyzing data are.
In an email to The Herald, Keller included a quote from a Sept. 16 New York Times article, which refers to the fact that the reanalysis team includes an expert who was a paid witness in lawsuits against GSK.
“Of course we’re biased. Everyone’s biased,” Jureidini said in reference to the reanalysis team’s history. Bias is exactly why making data public is so crucial, he added.
The reanalysis “is not just another journal article which you are forced to trust at face value,” Thacker said. This is the open source software movement of research, he added.
In a response to Keller and his colleagues’ criticisms, David Healy, a co-author on the reanalysis, writes that the team disagrees with the criticisms but agrees there is always room for reinterpretation.
“If people can demonstrate we’re wrong, we’re open to being proven wrong,” Jureidini said.
‘Party to the fraud’
When Study 329 was published, Keller was chair of the Department of Psychiatry and Human Behavior, a position he had held since 1989. By the time Keller stepped down as chair in 2009, he had come under fire for failing to disclose consulting payments from GSK and Pfizer, The Herald previously reported.
Then-Provost David Kertzer ’69 P’95 P’98 told The Herald at the time that Keller’s resignation was not related to the Paxil controversy or the Grassley investigation.
In 2011, Jureidini and McHenry wrote a letter to the University, asking administrators to write to the journal that published the original paper and support its retraction.
The University at first cited confidentiality concerns, but after Jureidini and McHenry clarified they were not seeking any information, the University subsequently notified them that it would not write to the journal. The University’s reference to confidentiality rules — both at the time and when asked for comment since — is “a red herring,” McHenry said, adding that the policy itself is not objectionable but it “is wrong is when they try to use that to escape responsibility.”
Marisa Quinn, then-vice president for public affairs and University relations, wrote in an email to The Herald in 2013 that “a thorough and impartial review” of Study 329 took place.
But in all subsequent responses to questions about Study 329, Quinn and Vice President for Communications Cass Cliatt declined to comment on whether the University has reviewed Keller’s research, citing the need for confidentiality in all individual cases.
Roy Poses ’74 MD ’78, clinical associate professor of medicine and president at the Foundation for Integrity and Responsibility in Medicine, said he understands the need for confidentiality but urged the University to make a statement now that any investigation would presumably have concluded.
“Nobody at Brown has responded in a substantive way,” Poses said. “I’ve never seen anyone dealing with the specifics.”
The University is “an enabler” of the study being published in Keller’s name, McHenry said, and several sources suggested the University’s silence can be attributed to its desire to protect its reputation.
The University received money from GSK during the years Study 329 was ongoing, according to a 2006 deposition of Keller.
“Brown always has adhered strictly to all issues of scientific integrity and compliance with conflict of interest policies laid out by the federal government,” Cliatt wrote, adding that the University takes these types of concerns seriously.
The reanalysis is not only significant in regard to the history of Study 329, or the study’s possible retraction, multiple sources said.
“It’s really not about Study 329,” Nardo said. “It’s about integrity in research.”
Nardo added that he is dismayed by the role of advertising in psychiatry. “It’s fine for them to do it with toothpaste, but not with our medicines. They’re too dangerous,” he said.
Jureidini said he and his colleagues hope the reanalysis demonstrates the importance of data being public. Without viewing the data of each individual, it is impossible to be confident in the conclusions papers draw about benefits and harms, he said. The reanalysis shows that companies commissioning their own studies does not work, he added.
As the first Restoring Invisible and Abandoned Trials initiative study published, the reanalysis of Study 329 is “just the tip of the iceberg,” McHenry said, with numerous other studies listed in the BMJ document as ones that should be reanalyzed.
“What we’re basically saying to the manufacturers is, ‘If you’re not right, we’re gonna come back at you,’” Nardo said.
- Additional reporting by Elena Weissmann