Retraction Watch

Tracking retractions as a window into the scientific process

High-profile Science paper retracted for misconduct

with 10 comments

Science has retracted a high-profile immunology paper after a probe concluded the corresponding author had committed misconduct.

The paper — which initially caught media attention for suggesting a protein could help boost the immune system’s ability to fight off tumors — has been under a cloud of suspicion since last year, when the journal tagged it with an expression of concern, citing a university investigation.

That investigation — at Imperial College London — has concluded that the paper contained problematic figures that were the result of research misconduct. All were prepared by last and corresponding author Philip Ashton-Rickardt, who took full responsibility. Even though the paper was published in 2015, some original blots and accompanying details have disappeared.

Today, the journal released a retraction notice:

AN INVESTIGATION BY Imperial College into the Science Research Article “The protein LEM promotes CD8+ T cell immunity through effects on mitochondrial respiration” (1), which was the subject of an Editorial Expression of Concern in December 2015 (2), has now concluded that duplications and use of incorrect Western blots occurred during the preparation of several figures in the paper. The investigation also found that examples of the original Western blots and accompanying experimental details had been lost. The investigation found that the problematic figures had been prepared solely by corresponding author Ashton-Rickardt and he accepted full responsibility for them. In agreement with the recommendation of the investigation, Science is therefore retracting the Research Article.

The 2015 paper, “The protein LEM promotes CD8+ T cell immunity through effects on mitochondrial respiration,” has been cited 17 times, according to Clarivate Analytics’ Web of Science, formerly part of Thomson Reuters.

At the time of its publication, the paper received some press coverage. Ashton-Rickardt called the protein a “game-changer” in The Telegraph, noting:

This is exciting because we have found a completely different way to use the immune system to fight cancer.

We contacted Ashton-Rickardt, who referred us to Imperial College’s press officer. A spokesperson told us:

The College investigation found that there was inadequate record keeping and the loss of original experimental data. There were also repeated errors in compiling figures in the paper, some of which were duplicated and presented as different experiments which supported other data. It was concluded that these errors and the loss of experimental data and the presentation of duplicated figures in the paper fell below acceptable scientific standards, and therefore constituted research misconduct.

We asked about Ashton-Rickardt’s status at the university, and the spokesperson declined to comment, noting:

The College will take the matter forward according to its policy on research misconduct. It would not be appropriate to comment further on an individual case.

The paper has been discussed on PubPeer since October, 2015. In November, the authors issued a correction, noting that some blots had been cropped and mistakenly swapped, which then-editor-in-chief Marcia McNutt attributed to “carelessness.” That month, she told us the journal didn’t have sufficient evidence to request a misconduct investigation; however, as the journal continued to see more public allegations of misconduct, it requested the university conduct a probe in December, 2015. But by then, the probe was already underway — so Science tagged the paper with an EOC. 

Jeremy Berg, Science’s current editor-in-chief, explained that McNutt’s initial reason not to request an investigation

should not be misconstrued as a policy statement. As more information became available after the Erratum appeared, we did request an investigation.

He added:

Our goal in such cases is to follow up on concerns raised by the community in a manner that supports a fair process for everyone involved, quickly alerts readers to potential problems and ensures that the literature is appropriately corrected. Sometimes multiple steps are required to accomplish these goals.

Berg noted:

…we do not base our judgments on whether or not a study’s conclusion will change, but rather on the integrity of the content we have published. In some cases, an erratum is all that is necessary to uphold the paper, while at other times, a full retraction is needed.

Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.

Comments
  • rfg December 15, 2016 at 2:34 pm

    If the journal bends over backward to allow a dubious correction, then finds that the correction is [even may be] a cover-up they should IMMEDIATELY publish an EoC.

    Here we find that as the investigation started the journal EoC. Better late than never.

    “Berg noted:

    …we do not base our judgments on whether or not a study’s conclusion will change, but rather on the integrity of the content we have published.”

    OTHER JOURNALS SHOULD FOLLOW THIS LEAD. Especially egregious are journals that think that it’s ok to retract one contrived figure or a fabricated panel or two of from a figure if it doesn’t “change the conclusions.” That perverts the whole review process.

  • Sir Duke December 15, 2016 at 5:26 pm

    Good job Science and Jeremy Berg! This is exactly how it has to be done for the sake of science.

  • Mary Kuhner December 15, 2016 at 7:02 pm

    If a single figure is falsified, as opposed to simply incorrect, the whole paper needs to be retracted: why should we believe anything in it, since at least one author shows willingness to commit fraud? I am deeply frustrated with “the results did not change.”

    • fernandopessoa December 20, 2016 at 1:19 pm

      Well there are no results now.

  • Farid Elasmer December 16, 2016 at 3:08 am

    This is a very positive behavior from Science. I hope that it could be the same behavior from others. It is important to see a real action from the institute to whom the researcher belongs

  • TL December 16, 2016 at 8:32 am

    There is nothing courageous about retracting a paper after an institutional investigation has issued a verdict of scientific misconduct. This is the minimum that should happen.

    • Anonymous December 16, 2016 at 10:24 am

      Maybe we are overreacting because there are too many cases in which that minimum is not even reached…

  • Neuroskeptic December 16, 2016 at 3:07 pm

    The Imperial spokesperson said: “There were also repeated errors in compiling figures in the paper, some of which were duplicated and presented as different experiments which supported other data.”

    Were these truly ‘errors’ or were they fraud?

  • herr doktor bimler December 19, 2016 at 2:19 am

    some blots had been cropped and mistakenly swapped, which then-editor-in-chief Marcia McNutt attributed to “carelessness.”

    The accompanying PubPeer thread was educational. I had not previously realised that when you crop a scanned image digitally (rather than physically) and embed the isolated lanes in a PDF to illustrate particular conditions, the entire image can later be recovered, and matched with other copies of the same image cropped and manipulated to illustrate different experimental conditions. This is embarrassing if you have claimed that the re-use was due to the images being physically cropped of identifying context and therefore easily confused.

    • John Krueger December 20, 2016 at 11:57 am

      These technics, and more useful ideas for examining image evidence, are outlined in the second of the “readme” files that I created to accompany the Forensic tools (see “Examining Image Evidence…”. Fortunately, they are still on the ORI website; but (ruefully), they may be somewhat dated.

  • Post a comment

    Threaded commenting powered by interconnect/it code.