South San Francisco biotech Genentech on Thursday announced that an internal review of misconduct allegations concerning a landmark 2009 paper co-authored by Marc Tessier-Lavigne, a former top executive at the company and Stanford University’s current president, did not find any evidence of fraud or intentional wrongdoing. But the review also points to another previously undisclosed case of scientific misconduct by a post-doctoral researcher in Tessier-Lavigne’s lab.
The investigation’s findings, detailed in a five-page document released by the company, note that none of more than 35 current or former employees interviewed reported observing or knowing of fraud related to the 2009 study in the journal Nature, whose lead author was a postdoc working for Tessier-Lavigne.
“I am not surprised by Genentech’s review, which directly and unequivocally refutes the false and hearsay rumors concerning the 2009 Nature paper and related research,” Tessier-Lavigne wrote in an email to STAT.
The review does acknowledge that researchers within and outside of the company struggled to reproduce certain findings before and after the paper’s publication, which led at least one senior leader within the company to say that the study should be retracted or corrected. Instead, the company ran further experiments to better understand the issue, and Tessier-Lavigne continued follow-up work after he became president of The Rockefeller University.
The review pointed to a complaint filed in mid-2010 concerning a case of scientific misconduct that until now had not been publicly known. The complaint, which involved a different postdoctoral researcher working under Tessier-Lavigne’s supervision, had to do with a project unrelated to the 2009 Nature paper. The results from this work were submitted for publication in a scientific journal, but the investigation triggered by the complaint led to the submitted study being withdrawn and the postdoc being fired in August 2010, according to the review. Tessier-Lavigne and others were co-authors of the manuscript.
In his email to STAT, Tessier-Lavigne declined to go into specifics about the 2010 case, other than to say he was the one who referred the issue to Genentech’s legal department.
The findings come after the Stanford Daily, the university’s student newspaper, had reported that former Genentech employees claimed that an internal review of the 2009 Nature paper based on work that Tessier-Lavigne supervised uncovered falsified data, and that the renowned neuroscientist tried to keep that information quiet.
The concerns raised in the Daily and later by scientific image experts, who found figures in the paper that appeared to have been duplicated, prompted Genentech’s legal team to take a close look at the work that led to the 2009 study by interviewing current and former employees. Some of them were part of the company’s research review committee, while others worked on the neuroscience team or in other roles that would have given them knowledge of the work that preceded and followed the Nature paper. The company also pored over lab notebooks, meeting minutes, and emails from the time, though it also acknowledged it was unable to find some of the original data and images from roughly 15 years ago.
The original study proposed a mechanism in which two molecules, death receptor 6 (DR6) and amyloid precursor protein (APP), interact in a way that can cause neurons to die and lose connections. It’s a process that is part of the development of a healthy brain, but one that scientists believed could also contribute to deadly diseases such as Alzheimer’s.
The company’s legal team found that none of the past and present employees it interviewed has seen or knew of any intentional wrongdoing related to the 2009 study. But an outside expert in analyzing images used in scientific publications did find instances of image duplication in the study, consistent with what Vanderbilt University Alzheimer’s expert Matthew Schrag had noted on PubPeer, a website that allows users to comment on published studies.
“We have not determined how these anomalies occurred. Following the receipt of these findings, Genentech reported them to Nature,” the company wrote in its report. On March 15, Nature added an editor’s note to the 2009 paper, warning that “concerns have been raised regarding some of the figures in this paper.”
The review also acknowledged that researchers struggled to reproduce one of the study’s original findings — an interaction between DR-6 and a cut fragment of APP known as N-APP. It’s an issue Tessier-Lavigne knew about, the report concluded, though in a mid-February letter to a Daily reporter, the university leader wrote that “the data were reproducible.”
Tessier-Lavigne addressed this issue in the email to STAT: “There was indeed discussion about differences seen in binding with various preparations of N-APP and DR6, but it is quite common for proteins prepared in different ways (for example in mammalian versus non-mammalian cells) to behave differently,” he wrote. “Importantly, at the time of publication, independent Genentech scientists were very clear that they saw binding of DR6 and N-APP made in mammalian cells (the relevant cell type), in line with the cell-based data produced in my laboratory.”
Some external research groups did reproduce the study’s key findings, but Genentech’s investigation summary notes that other scientists struggled for years to reproduce findings both internally and outside of the company. As a result, one senior leader at the biotech stressed that the paper needed to be retracted or corrected. That was not done, however; the review noted that correcting or retracting a paper is in the purview of the authors, but didn’t explain why they didn’t notify Nature of this issue with the paper.
Follow-up papers, including research that Tessier-Lavigne contributed to while at Rockefeller, suggested that some of the issues could have been related to the purity of the reagents researchers used in their experiments. And researchers eventually concluded that APP does not have to be cut by an enzyme to latch onto DR-6. Genentech’s research program around DR-6 was eventually discontinued in 2012 after findings from experiments in mice used to study Alzheimer’s dampened hopes that this line of research could lead to an effective treatment for the neurological disease.
“Many scientists who worked on the project were disheartened by having devoted substantial time and energy to a program whose underlying biology was ultimately proven wrong,” the company wrote in its review. “That sentiment gave rise to rumors about why the DR6 program failed.”
Genentech’s report comes just hours after STAT’s own reporting — based on interviews with 20-plus Ph.D. students, faculty, experts familiar with Tessier-Lavigne’s work, and former lab members from throughout his decades-long career — underscored the complexity of this case and how it shines an uncomfortable spotlight on the world of elite science.
All of those interviewed said it was highly unlikely Tessier-Lavigne would manipulate or knowingly present questionable results, but also acknowledged the underlying pressure in science to present perfect data that fit a coherent story to appeal to top-tier journals. And in high-achieving labs such as Tessier-Lavigne’s, defined by long hours and high ambitions, that pressure is ever-present.
This story has been updated with comments from Marc Tessier-Lavigne.
