A Duke researcher is in the process of retracting a foundational cancer genomics paper, signaling a lack of confidence in some of its findings.
Dr. Joseph Nevins is requesting the retraction as part of an ongoing review of experiments conducted by Dr. Anil Potti, the cancer researcher under investigation for research misconduct who resigned Friday. Potti was a principal author of the paper, which describes ways of using genomic information to predict how individuals would respond to cancer-fighting drugs.
“This process has been initiated due to concerns about the reproducibility of reported predictors, and their possible effect on the overall conclusions in this paper,” according to a Friday Duke Medicine news release.
Nevins’s decision to withdraw the paper comes three years after outside researchers first raised questions about it. At the time, Potti and Nevins both defended their work. Nevins did not respond to a request for comment Friday.
The paper, titled “Genomic signatures to guide the use of chemotherapeutics,” was published in the journal Nature Medicine in October 2006. The paper seemed to help answer long-standing questions in oncology concerning how doctors should choose cancer treatments for their patients. It has since been cited by 334 other articles, according to Google Scholar.
Checking Potti’s work
Huntington Willard, director of Duke’s Institute for Genome Sciences and Policy, said he hopes the retraction of this influential paper will not impede progress in cancer genomics research.
“My hope is that this will, in the fullness of time, be seen as a very modest setback, that the field will continue to thrive and move forward,” he said. “The concept behind the science is completely solid, and no one is really challenging that.”
The retraction is the second Nevins has requested as he reviews Potti’s research. Nevins, Barbara Levine professor of breast cancer genomics and director of the Center for Applied Genomics and Technology, was Potti’s mentor at IGSP and is listed as a co-author on much of his research, including the Nature Medicine paper. Last week, the Journal of Clinical Oncology retracted a paper based on Potti’s research at Nevins’s request.
Willard said several researchers at IGSP are reviewing Potti’s work, but it is too soon to say whether other papers will need to be retracted. He said they are paying particular attention to papers whose conclusions have been disputed and to research related to ongoing experiments at IGSP.
“We’ll go back as far as we have to and as deeply as we need to to get a good sense of what the state of the science really is,” he said. “If we find other papers that need to be retracted, then those will be retracted. If we find papers where everyone stands by the data, then we or others will make that clear.”
“These trials should not have been done”
The consequences of the first two retractions are already being felt.
Three clinical trials that were based on the retracted papers were terminated earlier this month. The trials were suspended in July, when allegations that Potti falsified portions of his resume led to scrutiny of his work.
Dr. Michael Cuffe, Duke University Health System vice president for medical affairs, said he does not think patients in the trials were harmed because all received high quality cancer treatment.
“In retrospect, these trials should not have been done,” he said. “Given that they were based on his science, I think that the best that we can say right now is that we’re thankful that the therapies were either standard of care, widely used or widely accepted.”
Cuffe said the doctors responsible for the trials were instructed to inform patients about the research problems. He said he told the physicians to discuss other cancer treatments the patients might want to pursue.
Some researchers had raised questions about the Nature Medicine paper for several years. In 2007, three University of Texas MD Anderson Cancer Center biostatisticians wrote in a letter to Nature Medicine that they could not reproduce the paper’s original findings.
“The idea of using... cell lines to predict patient response to chemotherapy is exciting,” they wrote. “Our analysis, however, suggests that it did not work here.”
In fall 2009, the MD Anderson biostatisticians published an article in Annals of Applied Statistics further criticizing the research and saying that clinical trials based on it should be halted. After the publication of this paper, the National Cancer Institute contacted Duke to express concern about the clinical trials. The trials were then suspended and the research underlying them reviewed. Ultimately, a committee appointed by Duke’s Institutional Review Board approved re-opening the trials to patients.
“The deeper issues of data provenance and the deeper work that Dr. Nevins is evaluating now were not in the scope of that review,” Cuffe said. “They were simply not addressed at that depth and level. That’s why, and this is difficult work, we’re so deliberately looking at the science right now.”
But one of the MD Anderson researchers, Keith Baggerly, believes the trials should never have been restarted.
“We don’t think Duke needed to do any further deep investigation to find these problems,” he said. “They knew enough in November or December of last year that the decision to restart the trials in January was wrong.”
Get The Chronicle straight to your inbox
Signup for our weekly newsletter. Cancel at any time.