The independent news organization of Duke University

Co-author in Potti paper retracts work

Joseph Nevins

As investigations into Dr. Anil Potti’s research continue, Dr. Joseph Nevins, Potti’s mentor and collaborator, has now acknowledged errors in a paper whose results have been questioned for nearly a year.

The paper described a method of assigning patients to cancer treatments and was the basis of two cancer clinical trials being conducted at Duke. Two biostatisticians brought concerns about data in this paper to Duke officials last Fall, prompting Duke to conduct an investigation of the research. At that time, members of the investigation committee identified only minor problems that did not affect the validity of the research.

“I wonder if we should have caught it earlier, or if [Nevins and Potti] should have caught it earlier,” Dr. Victor Dzau, chancellor for health affairs and president and CEO of the Duke University Health System, told The Chronicle. “I think [it took] a deep dive, particularly by Dr. Nevins and new statisticians, to go through point by point and line by line to look at this issue.”

Nevins, Barbara Levine professor of breast cancer genomics and director of the Center for Applied Genomics and Technology, wrote in an e-mail that he is asking the Journal of Clinical Oncology to retract the paper because of faulty research. He sent the e-mail Oct. 22 to the 13 co-authors of the paper. The e-mail was obtained by The Cancer Letter and published Friday.

Nevins wrote that the research, published in 2007, “cannot show a capacity to distinguish responders and non-responders when the correct clinical information was used, contrary to what was reported in the paper.”

Potti and Nevins had previously claimed that they could predict patient response to certain chemotherapy drugs using genomic models.

Potti is currently under investigation for research misconduct and Dzau said that Nevins’ retraction will be considered by the investigation committee.

Nevins is not under investigation. He did not respond to several requests for comment.

Old problems resurface

The errors identified by Nevins call into question the scientific basis for two clinical trials involving patients at Duke.

Enrollment in those trials in addition to one other was temporarily suspended a year ago after Keith Baggerly and Kevin Coombes, biostatisticians at the M.D. Anderson Cancer Center pointed out problems with Potti and Nevins’ research. The trials resumed after Duke commissioned a review of the scientific work. Enrollment in the trials was stopped again this summer after allegations surfaced that Potti falsified portions of his resume, including his claim of being a Rhodes Scholar.

Baggerly and Coombes noted that the problems they identified last year are the same as those now recognized by Nevins.

“The errors that [Nevins lists] now in October 2010 in stating a rational for withdrawing the paper, are qualitatively identical to errors we identified and reported to Duke in November 2009, almost a year ago,” Baggerly said.

Coombes noted that the data was mislabeled in a way that scrambled those who responded to the drugs and those who did not.

However, Dzau said that such mistakes occur at institutions across the nation, noting recent cases at Harvard and Stanford universities.

“There are cases like this because science is by nature an exploration and an innovation,” Dzau said. “It’s important for us to see this as an opportunity to advance in science.... We become leaders because we learn from it, and we can become a better university for it.”

“I’m starting to worry”

In light of Nevins’ request for a retraction, questions have risen about whether cancer patients enrolled in the trials were harmed.

“I’m starting to worry,” Dr. Otis Brawley, chief medical officer of the American Cancer Society, said in an interview with The Chronicle. “[Potti and Nevins] might have caused some harm, I don’t know yet. I’m waiting for the end of the [Institute of Medicine] investigation.”

ACS partially funded Nevins’ and Potti’s research through a $729,000 grant awarded to Potti in 2007. Brawley said he is responsible for protecting both the ACS’s research money and patients involved in clinical trials. Brawley added that he is not convinced patients were harmed, but if investigations turn up fraud in Potti’s research, the IOM may ask that the grant money be returned.

Investigations continue

Paul Goldberg, editor of The Cancer Letter, reported the problems with Potti’s resume that led to serious investigations into Potti’s research.

“This is the ultimate tragedy of the whole thing. Had I not caught Potti claiming to be a Rhodes scholar, which has nothing to do with his research... this thing might have lingered on and kept on going, and patients could be harmed,” he said. “Even if they weren’t harmed, and I think they were, just the idea of wasting their time on the clinical trial of garbage is harm. It’s harm because it might give them false hope, it’s harm because it’s wasting the resources of Duke staff. It’s harmful in 100 different ways.”

After Goldberg identified issues regarding Potti’s resume in July, the University launched an investigation into his credentials and a research misconduct inquiry. The Institute of Medicine later agreed to lead an external review of both Potti’s research and broader issues related to certain predictive tests.

The University found “issues of substantial concern” in Potti’s resume but declined to give specific details. The research misconduct inquiry is still underway.

The IOM review committee is in its initial stages and will not issue a report until spring 2012. All the information related to the retraction will be provided to the IOM, Doug Stokke, assistant vice president of communications for Duke University Health system, told The Cancer Letter.

Dzau said that although the errors Nevins has identified are an important piece of the evidence being reviewed by the Duke committee, they do not necessarily show that research misconduct was committed. Other factors must also be considered, particularly whether the errors were intentional.

“The data is not valid,” Dzau said. “But... there is more to misconduct than just looking at the data.”

Taylor Doherty contributed reporting.


Share and discuss “Co-author in Potti paper retracts work” on social media.