From ‘Nature News’ at Nature.com, 5 October 2011 (story by Heidi Ledford).
Paper linking chronic fatigue syndrome to a virus suffers another blow as duplicate figures surface.
The scientist behind a study(1) that linked chronic fatigue syndrome to a virus has lost her job and is now facing accusations that she has misrepresented data.
Judy Mikovits, formerly research director at the Whittemore Peterson Institute for Neuro-Immune Disease in Reno, Nevada, was fired on 29 September after she clashed with the institute’s president and co-founder, Annette Whittemore, over the work of another researcher.
The following day, in what seems to be a separate development, a blogger posted a figure from a 2009 paper(1) that Mikovits co-authored in Science alongside one that Mitkovits used in a recent presentation. The two figures, which are used to describe different results, look identical, except for the labelling.
No known cause
The paper in Science raised hopes among those who have chronic fatigue syndrome, because it identified a possible cause of the mysterious debilitating condition. The researchers reported that a retrovirus called XMRV, which stands for xenotropic murine leukaemia virus-related virus, was present in 68 out of 101 people with chronic fatigue syndrome, compared with only 8 out of 218 healthy controls(1)(see Nature 471, 282–285; 2011).
But attempts to reproduce the results in other laboratories have failed. On 22 September, the authors took the unusual step of retracting only part of the paper, including the first figure, after learning that some of the samples were contaminated(2). On the same day, Science published a large study conducted by nine independent laboratories, including Mikovits’, that failed to confirm the presence of XMRV in patients with chronic fatigue syndrome(3).
Mikovits has since acknowledged that XMRV may not be behind the syndrome, but says that another related retrovirus is likely to be responsible.
Despite the growing criticism over her work, Mikovits seemed to enjoy the support of Whittemore and her institute. According to a termination letter that Whittemore sent to Mikovits it was a laboratory power struggle unrelated to the XMRV paper that soured the relationship. Whittemore wrote that she had ordered Mikovits to allow another researcher in the lab to work with a particular cell line but Mikovits refused. “Your actions have shown a complete lack of respect for your colleague and for my authority,” Whittemore wrote.
Mikovits counters that the experiments would have been beyond the scope of the federal grants that funded the work. Whittemore could not be reached for comment on the matter but the Institute released a statement that it would, in the wake of Mikovits’ departure, re-evaluate its research projects.
Mikovits also came under fire when an apparent duplication of data was brought to light by Abbie Smith, a graduate student in virology at the University of Oklahoma in Norman and author of the blog ERV. In a post on 30 September, Smith overlayed a figure from the 2009 Science paper and a figure presented by Mikovits at a meeting of the The International Association for Chronic Fatigue Syndrome/Myalgic Encephalomyelitis in Ottawa, Canada, last month.
The image, Figure 2c in the paper, tracks the expression of a viral protein called p30 Gag in people with and without chronic fatigue syndrome. The corresponding slide shown at the meeting was said to show activation of the protein’s expression following treatment by a compound called azacytidine. Furthermore, some samples labelled as ‘normal’ in the Science paper were given patient numbers in the presentation. And the patient sample numbers differed between the two figures.
Commenters on the blog also pointed out that the cropped figure used in the presentation can be expanded to show more of the gel, including the original handwritten labelling. The labels on that original image differed from how the slide was described in the paper and the presentation.
Such discrepancies are difficult to cast as unintentional errors, argues Jonathan Stoye, head of virology at the MRC National Institute for Medical Research in London. “There is no consistency in the way these things are labelled,” he says.
Mikovits says that the Science paper didn’t mention the azacytidine treatment because that detail was not necessary for the publication. Instead, the authors referred to the samples as being ‘activated’. But for her presentation, she wanted to emphasize that the addition of methyl groups to viral nucleic acids may have prevented the virus from being detected using standard assays. Azacytidine strips off those methyl groups and, Mikovits argues, can boost viral gene expression.
As for the variability in patent sample numbers, she says it was common practice for her to change the numbers before any public presentation to prevent patients from identifying their own samples. Mikovits maintains a database of the original numbers and their corresponding new labels. But in the rush to prepare for her talk in Ottawa, she says she neglected to switch the numbers on the slide in question.
And she chose to relabel a ‘normal, untreated’ portion of the image with a patient number because the data were the same. “It simplified the slide primarily for a patient audience,” she says. “This is not in anyway inappropriate for a presentation as long as the data are correct, and they are.”
The Whittemore Institute responded to the allegations by saying: “It is our understanding that some patient ID numbers may have been changed to a new set of coded numbers during the research to protect their privacy before publication.” The institute says it has requested information from Mikovits and that it will work with Science “to gain a full understanding of the cause of any potential discrepancies”.
Critics have also argued that the original, handwritten labels, which include the letters ‘RT’ suggest that the blot had been stained for reverse transcriptase, another viral protein altogether. But Mikovits says the ‘RT’ on the blot simply meant ‘room temperature’.
Mikovits’ collaborator, Frank Ruscetti of the National Cancer Institute’s Center for Cancer Research in Frederick, Maryland, whose lab prepared the blot, offers an alternative explanation: The antibody used to detect p30 was originally made to detect reverse transcriptase, he says, but instead recognizes p30 and its precursor so it was used for this purpose.
“This is the problem of people trying to interpret lab jargon,” says Ruscetti.
Science says it is reviewing the allegations. The journal has requested the original images used to make the published figure; Mikovits says the data were sent on 4 October.
Those who have been critical of the Science article1 say all of this simply strengthens the argument in favour of retracting the entire paper. “I have a copy of that paper on my desk,” says Stoye. “It has a line through Fig 1. It has a line through Fig 2c. Where is the next line going to be? Is there going to be anything left?”
(1) Lombardi, V. C. et al. Science 326, 585-589 (2009).
(2) Silverman, R. H. et al. Science advance online publication http://dx.doi.org/10.1126/science.1212182 (2011).
(3) Simmons, G. et al. Science advance online publication 10.1126/science.1213841 (2011).