With seven retractions on the books so far, the process of sifting through the research publications of Dr. Anil Potti will soon draw to a close, though the misconduct investigation may last several more months.
Since July 2010, when it was discovered that Potti falsely claimed that he was a Rhodes Scholar, the former Duke cancer researcher’s work has been under review. Researchers who have co-authored papers with Potti are re-evaluating the work to determine if it should be retracted. In a separate process, a committee appointed by the Research Integrity Office of the Duke University School of Medicine is investigating whether or not Potti committed research misconduct. The review of publications will reach a conclusion sometime in November, but there is no clear end date for the investigation.
The journal, Proceedings of the National Academy of Sciences, published the seventh retraction of a paper co-authored by Potti Oct. 3. The authors, including Potti, signed the letter petitioning the journal to retract the work, which noted that the group was unable to reproduce results from certain vital experiments regarding a colon cancer prognosis gene signature.
Approximately 40 papers with original data contributed by Potti are currently under review by their co-authors, under the oversight of Dr. Robert Califf, vice chancellor for clinical research and director of the Duke Translational Medicine Institute. Of those papers, roughly one-third will be retracted, another third will be corrected and one-third will be left alone, Califf said.
“The numbers are roughly holding steady,” he said. “We’re not quite there yet, but we’re near the end of the road.”
Although Duke’s retraction activities may be over in less than a month, Califf said the journals must then evaluate the retraction requests, meaning it might take longer for the results to become public.
“We’re almost done,” Califf said. “But making a decision and having everything work its way through the journals to a public decision, that’s up to the journal, so it could take several months.”
Califf noted that no formalized method exists to guide co-authors when deciding to retract a paper.
“There are no rules written down about how you deal with retractions when there are different points of view among the authors,” he said.
Califf said the lack of rules is partly due to the fact that establishing these rules is not a top priority for scientists who are engaged in research.
“Who wants to spend their career looking at the rules for retraction?” Califf said. “I’d rather spend my career doing good research and answering questions about biology and treatment.”
Reproducibility is the key to a paper’s validity, said Sayan Mukherjee, associate professor of statistics for the Institute for Genome Sciences and Policy. The authors work together to see if they are able to replicate the main findings of an experiment, said Mukherjee, who worked on statistical analysis for several of Potti’s papers—including the one most recently retracted. If they are unable to reproduce the results, they can petition the editor of a journal for a retraction.
“The [Potti] retractions have been, I understand, unanimous,” Mukherjee said. “We tried to the best of our ability to reproduce [key results,] and we couldn’t—and at that point, there’s really not much more you can say.”
Due to the nature of original scientific research, the retractions may not have a substantial effect on existing papers that cite the retracted research, Califf said. This is because other researchers apply Potti’s genetic concepts to their own research instead of building on Potti’s specific findings.
Potti’s most recently retracted paper has been cited 27 times, according to the blog Retraction Watch.
These retractions set an example that may have positive effects on the field, Califf added.
“The discussions that have emanated from the retractions have probably made everyone’s research better—kind of like, when you see somebody speeding and get caught, you slow down,” he said.
Duke’s research misconduct investigation of Potti’s work is ongoing, said Donna Cookmeyer, the research integrity officer for School of Medicine. Cookmeyer, who administrates the research misconduct investigation process, declined to comment on the specifics of the Potti case because scientific misconduct proceedings remain confidential until a finding is made public in order to protect those involved. The Chronicle previously reported that the research misconduct investigation started during the summer of 2010.
In order to conclude that a case of research misconduct has occurred, an investigating committee—proceeding under strict federal guidelines—must find that a researcher plagiarized, fabricated data or falsified data.
This standard is a high bar to meet, Cookmeyer said.
“You can’t just find the data was wrong,” she said. “The data has to be wrong and someone had to make it wrong, and making it wrong had to be a serious deviation from accepted practice.”
The investigation passes through several defined phases—starting with assessment of allegations, then inquiry into whether an investigation should be launched, to the investigation itself, Cookmeyer said. Federal guidelines set desired time frames for each step in the process, but an investigation can last one year to 18 months without extensions, she added.
“It would be a rare case where you wouldn’t need to extend at least one of the deadlines because they tend to be very scientifically complicated,” she said.
An investigative committee is generally composed of several of the researcher’s peers, chosen for their expertise on the subject. The logistical complexities of the investigation often impede a quick resolution, Califf said.
“These are people that have full-time research careers,” he noted. “They are busy people, and I’m sure it takes time to go through all the evidence and hear from everybody.”
In order to reduce the possibility of another Potti-like scandal and improve overall quality of research, Califf said his office will begin to implement a Translational Medicine Quality Framework over the next few months.
“This kind of thing... could happen anywhere,” he said. “There’s no fail-safe system. But the goal is to reduce the chances of it happening.”
Get The Chronicle straight to your inbox
Signup for our weekly newsletter. Cancel at any time.