Journal Retracts Fraudulent Oral Cancer Studies
New England Journal of Medicine articles didn't affect patient care, experts say
THURSDAY, Nov. 2, 2006 (HealthDay News) -- The editors of the New England Journal of Medicine this week added another chapter to the story of disgraced Norwegian researcher Dr. Jon Sudbo, as they formally retracted two reports on oral cancer his team published in the journal in 2001 and 2004.
By his own admission -- issued through his lawyer earlier this year -- much of Sudbo's published data was fabricated. A recent report from an investigative commission formed by Sudbo's former employers, the Rikshospitalet-Radiumhospitalet Medical Center and the University of Oslo, has since confirmed the fraud. The NEJM editors based their retraction largely on the report's findings.
"Given the weight of evidence offered in the commission's report and the requests of most of the authors of the articles, we retract both articles," NEJM editors Dr. Gregory Curfman, Dr. Jeffrey Drazen and Stephen Morrissey wrote in the Nov. 2 issue of the journal.
They noted that all of the two studies' authors, with the exception of Sudbo, have agreed with the Oslo commission's findings and the journal's subsequent retraction of the article. Sudbo's various co-authors have been cleared of any wrongdoing.
Both articles focused on patients (typically long-term tobacco users) who were at high risk of oral cancer due to the presence of precancerous white growths in their mouths or on their tongues known as oral leukoplakia. Sudbo's April 1, 2004, NEJM article suggested that a cellular aberration called "aneuploidy" -- the presence of disordered cells marked by overabundant DNA -- might predict which patients would benefit most from surgical removal of these white spots.
But the Oslo report found glaring errors in Sudbo's work. Most notably, they point out that in the 2004 study, "the same patient appears several times," so the total number of patients listed is bumped up from 141 to 150.
Even more concerning, "69 of the 141 patients included in the study should have been excluded because they had been diagnosed with oral cancer before or at the same time as the leukoplakia was diagnosed," the report's authors noted. "For these patients, it was not possible to study the future development of cancer, since they already had cancer. This error alone is so serious that the results and the conclusions are invalid."
In an "Expression of Concern" published in the journal earlier this year, the NEJM editors also noted that, in the 2001 article, two photographs from a microscope reportedly representing two different patients at different stages of precancerous mouth lesions were "in fact, different magnifications of the same photomicrograph."
Sudbo also admitted faking data for studies published in two other leading journals, The Lancet (a 2005 article linking over-the-counter pain relievers with lowered oral cancer risk) and the Journal of Clinical Oncology (another oral cancer study, published in 2005). Editors at those journals also published special "Expression of Concern" letters in the following months.
One of the lingering questions is whether the findings had any negative impact on cancer patient care. Experts say that, in the case of the NEJM articles, the answer is probably "no."
The high-tech molecular test needed to spot aneuploidy "is a very complex method, and it's very expensive," and so it is out of the reach for most clinicians who might treat patients at risk for oral cancer, explained Dr. Deborah Greenspan, a professor of dentistry and clinical oral medicine at the University of California, San Francisco.
Sudbo's findings were also generally regarded as preliminary, she added. "To my knowledge, there have not been any changes in clinical practice, or indeed in any recommendations, based as a result of his findings," said Greenspan, who authored a NEJM "Perspectives" piece on Sudbo's work when it was published in 2004.
She said she was "completely surprised" by the revelation that Sudbo had faked his data. "I'm always troubled when we see scientific investigators fake their results," Greenspan added.
Still, she said, the basic premise behind the study -- that aneuploidy might help doctors better tailor treatment for oral cancer patients -- is still a valid one. "In fact, as a result of some of the things he started to do, other people have become very interested in exploring aneuploidy," Greenspan noted.
Dr. Len Lichtenfeld, deputy chief medical officer at the American Cancer Society, agreed that the NEJM research "did not gain a lot of traction" in terms of its impact on oncological practice. "I remember commenting at the time years ago that this kind of approach was 'not yet ready for primetime,' " he said.
Still, both experts were troubled that this type of fraud made its way into some of the world's leading medical journals.
"I can't make a definitive statement on what happened here," Lichtenfeld said, "but it appears that, based on the report, there was a lack of real oversight and an overabundance of trust by various investigators as [Sudbo's work] went forward. This wasn't an isolated, one-step issue, this was a systemic problem."
Indeed, the Oslo report cites Sudbo's relative isolation -- he was sole guardian of the data he claimed to have collected -- as one reason his fraud went undetected for so long.
Lichtenfeld did not point any fingers at the peer-review process at the NEJM.
"There's really nothing that the New England Journal could've done about this," he said. "To have found this, they would have had to have gone to every lab, sort through every piece of information, every approval, every memo. It's simply not possible."
Karen Pedersen, a spokeswoman for the journal, said NEJM is limited in what it can do to minimize the threat.
"Our graphics department is using PhotoShop to examine figures more carefully, but this practice only works on certain types of images and would not have found the fraudulent images in the Sudbo case," she said. "While we continue to work hard at being vigilant and skeptical, and have instituted additional safeguards in how we review images, at this point there is not an ironclad, foolproof way to prevent fraud by a single, determined author."
Lichtenfeld said Sudbo's case, and others like it, remain very rare and should not shake the public's trust in medical research and the journals that publish that research.
"When you have an unfortunate event like this, it certainly raises questions," he said. "But it should not bring discredit to the thousands of reports that researchers who are honest and hardworking put out every year."
There's more on oral cancer at the American Cancer Society.