Retraction Watch

Tracking retractions as a window into the scientific process

Publisher and uni go head to head over disputed images in diabetes papers

with 12 comments

Mario Saad, via unicamp.br

Mario Saad, via unicamp.br

The State University of Campinas University of Campinas and the American Diabetes Association disagree strongly over how to handle disputed images from faculty member Mario Saad, who is suing the ADA to prevent retraction of his papers.

While the State University of Campinas University of Campinas (Unicamp) acknowledges that 2 of Saad’s papers contain “mistakes”, it concluded there was “not an intention in the actions of the authors,” and the mistakes did not have a negative impact on the scientific community. Ultimately: “the studies published have their own strength, are healthy and were not artificially strengthened by the incorrect images.”

In response, however, Saad’s lawsuit says the ADA asked the school to reinvestigate the articles, and refused to accept any papers from Unicamp faculty in any ADA journals until the issues are resolved.

The ADA has issued four expressions of concern in Saad’s research published in its flagship journal, Diabetes. Saad’s lawsuit aims to prevent the journal from retracting those papers, and asks for monetary compensation.

Retraction Watch has obtained a report of the investigation by Unicamp and Saad’s lawsuit. You can read the full lawsuit here, and the investigation report here.

Here is an excerpt from the Unicamp’s findings:

-Problems related to identification, storage, and manipulation of the gel images in the laboratory of Prof Dr. Mario Saad’s group, where the studies were conducted. It was also considered that some of the image manipulation options adopted by the researchers and technicians of the laboratory during that period of the study were not adequate, which resulted in the mistakes described in the articles mentioned.

-Although the image manipulation practices adopted in that laboratory over that period were of a common use in molecular biology laboratories around the world, they are no longer acceptable currently, having been deemed as inadequate in referential journals around the world, such as the Nature, Journal of Cell Biology and Diabetes.

-Despite identification, storage, and manipulation mistakes of the images, evidences indicate that there was not an intention in the actions of the authors of the articles.

-The articles repercussions were positive in the scientific community with numerous quotes in this regard, which allows the commission to affirm that the studies published have their own strength, are healthy and were not artificially strengthened by the incorrect images.

Based upon such conclusions, it recommends (i) that the identification, storage and manipulation protocols of scientific images are reviewed, (ii) monitoring of manuscripts by more than one experienced researcher, at independent environments and times, as an additional assurance that mistakes as such are not repeated, and (iii) periodical training of students and researchers in ethics and good research practices, specially as far as the manipulation of images and research data.

The lawsuit describes the journal’s reaction to these findings:

By e-mail dated December 2, 2014 [Christian Kohler, of the ADA’s ethics committee] informed Dr. Saad that the ADA’s [Subcommittee on Ethical Scientific Publications] completed its review of the University’s report regarding the 2011 and 2007 Articles as well as Dr. Saad’s and Dr. Carvalheirais recent responses related to the 1997 and 2006 Articles.

In the December 2nd correspondence Mr. Kohler informed Dr. Saad that based on the ESP’s review and recommendations, Diabetes will publish a digital expression of concern online on December 15, 2014 and the expression of concern will appear in the March print issue of Diabetes, which releases on February 24, 2015.

In the December 2nd correspondence Mr. Kohler informed Dr. Saad that the ADA’s ESP will ask the University to reinvestigate the data published in the 2011 and 2007 Articles and the information presented in Dr. Saad’s point by point response.

Most detrimental, the ADA stated that it will not consider, for publication in any ADA journal any submissions authored by any of the faculty of the University until the issues described in the expression of concern have been appropriately reviewed and addressed by the University.

According to Saad’s lawyer, Steven Brooks, it’s unclear how the case will proceed. They’ve drawn the judge presiding over the Boston Marathon bombing trial; thanks to snow shutting down the city for several days, the bombing trial has been delayed.

We contacted Diabetes after Saad filed his lawsuit, and will update if we hear back.

Comments
  • Leonid Schneider February 12, 2015 at 11:14 am

    One has to admire the lawyers’ logic:
    “The articles repercussions were positive in the scientific community with numerous quotes in this regard, which allows the commission to affirm that the studies published have their own strength, are healthy and were not artificially strengthened by the incorrect images.”
    For further reference in all such cases: If a paper was cited, it cannot be wrong. QED.

  • Leonid Schneider February 12, 2015 at 11:23 am

    “Although the image manipulation practices adopted in that laboratory over that period were of a common use in molecular biology laboratories around the world….”
    When again was this practice of a common use and previously acceptable, in which molecular biology laboratories and around which world, can Unicamp please specify:
    https://pubpeer.com/publications/FC6FB9DF824662F15BF5D764904480#fb16930
    https://pubpeer.com/publications/FC6FB9DF824662F15BF5D764904480#fb17050

  • Sylvain Bernès February 12, 2015 at 12:33 pm

    The last quote from Unicamp’s report is the most baffling argument that I’ve seen for a long time:
    ” […] the studies published have their own strength, are healthy and were not artificially strengthened by the incorrect images”.
    That means that we are allowed to draw conclusions based on “incorrect” images? This is indeed good news for researchers, and it will save me a lot of time.

    My 2¢ suggestion to the crystallographers tired of having to waste their time on beamlines: measure just enough frames to get an image of your protein (don’t worry about quality at this step). Then, complete your diffraction data to a publishable resolution through a Fourier inversion of your image, merge all data, and finally, compute the “incorrect” image with the full data set. Just remember the Unicamp’s gold rule: your protein structure should not be “artificially strengthened” by a too good data resolution. Be modest, and avoid inventing, say, more than 80% of the diffraction intensities.

  • AMW February 12, 2015 at 12:48 pm

    To be clear Diabetes has not retracted, or threatened to retract, the articles – only issued expressions of concern about them all, and maintained this EoC given the failings of the institutional investigation. As such, the journal is safe.

    But what is the evidence that the University’s investigation and final decision was flawed?

    Most fundamentally, the investigation panel completely failed to see the basic point that these experiments were always done as replicates e.g. in the Diabetes 2011 paper, there were 10 rats per bar shown. Those actually shown in the paper are simply a representation from one animal; there should have been 10 gel bands for each bar. So where are the other nine gels for each condition? The University investigation accepted without comment Saad’s ‘original version’ of Figure 7D of the 2011 Diabetes paper. But apparently not a single person considered that there must have been NINE OTHER GELS that the lab had undertaken. These were never asked for, or produced by Saad at his hearing. Surely he would have brought them all out since he was fighting for his career?

    As outsiders I don’t think you can conclude anything other than that the experiments were never done. Strangely, the image manipulation may not actually be the primary problem; rather, it is just a visible sign of gels being fabricated which the parties involved assume will never be detected. Fabricating gels can’t be easy – as Mark Twain said : ‘If you tell the truth, you don’t have to remember anything’. People make mistakes which others can then detect because of telltale artefacts in the bands.

    A couple of other signs that the investigation was somewhat farcical. The panel accepted without question the explanation that the image problems were an accident, partly due to the ‘internal technical sector of his faculty, created to give teaching staff assistance with their publication, which is not made up by researchers’. In this model, the researchers run the gels and the technicians make up the Figures? Who can take that seriously?

    The investigation made no attempt to consider whether OTHER papers might also be afflicted. The RW article from 3 days ago contains links (from Paul Brookes) to the PubPeer website where Paul and another contributor point out obvious image manipulation in a Plos Biology paper and an Endocrinology paper.

    Finally the university officials suggest that the large number of citations of the articles indicates ‘validate the results’. The number of citations is entirely irrelevant – it just means people are quoting the paper and only that. Even if people find results that agree with Saad’s articles’ conclusions, that doesn’t mean the papers are not fabricated.

    Good journals have always been vindicated when threatened in this way, but it will still be interesting to see how this plays out.

  • Paul Brookes February 12, 2015 at 1:56 pm

    Let me go out on a limb (because this particular limb looks pretty darn thick to me) and dispute these two statements from the University…
    1. “Evidences indicate that there was not an intention in the actions of the authors of the articles”
    2. “The studies published have their own strength, are healthy and were not artificially strengthened by the incorrect images”

    From just a single PLos One paper alone, we have the following examples, including 3 separate instances of blot images with the wrong number of bands!
    http://i.imgur.com/JvSuSn9.jpg
    http://i.imgur.com/bgOmUak.jpg
    http://i.imgur.com/nf60fQi.jpg
    http://i.imgur.com/F7PfovR.jpg
    http://i.imgur.com/4DmR4it.jpg
    http://i.imgur.com/1Z42k5g.jpg
    http://i.imgur.com/wjEmuW0.jpg
    Here’s the PubPeer thread: https://pubpeer.com/publications/22162948

    Regarding the first statement, something that appears to have escaped the (admittedly limited) collective reasoning power of the University committee, is that mistakes are rare. When mistakes cluster – both within a single paper and around a single author – there’s usually another explanation, and it’s not bad luck.

    Regarding the second point, I utterly disagree. The counter-argument is quite simple… if the study can stand alone without the strength afforded by the problem images, then why were they included in the first place? Surely there was a motivation to include these problem images, and that motivation may have led to “mistakes” being made.

    ADA is on solid ground here, IMO.

    • AMW February 12, 2015 at 9:42 pm

      Agreed – the ‘mistakes’ definitely seem to cluster around Professor Saad. He has papers going back well over a decade with questions hanging over them. Given his failure to produce any of the multiple gels apparently forming the 2011 Diabetes paper (either to Diabetes or his own institution’s investigation), it seems likely that none of the experiments reported in his papers were ever done.

      On entering a lab where people are very successfully fabricating articles and gaining from it, with the whole operation condoned by the laboratory head, I would think 99 out of 100 people would decide to continue in the same vein. It would be extraordinary for a whistleblower to speak out against what was going on.

      The suing has just put the whole case out on the world stage.

  • herr doktor bimler February 12, 2015 at 1:58 pm

    The articles repercussions were positive in the scientific community with numerous quotes in this regard, which allows the commission to affirm that the studies published have their own strength, are healthy and were not artificially strengthened by the incorrect images

    If I may paraphrase: The frequency with which these papers were cited proves conclusively that the unusual manipulations of the pictorial component of the papers did not make them more convincing or increase the frequency of their citations.

  • Narad February 12, 2015 at 3:12 pm

    I’d put the odds of getting an injunction in time to prevent publication in the March issue at exactly zero. (Actually, I’d say the same thing overall; “equity will not enjoin a defamation”).

  • qqq February 15, 2015 at 2:41 pm

    I’m going to go on a big limb here and guess that the journal expressly forbids this sort of image manipulation. If so, whether the science holds up will be moot. I can’t imagine a world where this doesn’t get summarily dismissed as the university clearly admits that he broke the rules. It’s like saying, “Yes, your honor, I shot the sheriff and it was clearly illegal, but it shouldn’t have been.”

  • Nota bene February 15, 2015 at 10:19 pm

    Mario Saad (syn. Mario JA Saad, Mario A Saad, MJA Saad, Mário J. Saad, or Mario José Abdalla Saad)

    https://pubpeer.com/publications/85181AC5DFD5F8204FE1BA8326E076#fb25015
    Figures 2A, 2B, 2C, 2D, 2F, 2G, 2I, 2H, 3A, 3E, 4B, 4C, 4H, 5A, 5B, 5C, 6F, 6H, 6I in PLOS ONE 2010

    https://pubpeer.com/publications/6874355737C6787D83BE14179B69C4
    Fig 3 and 4 in Endocrinology 2008

    https://pubpeer.com/publications/16919274
    Fig 4 in FEBS Letters 2006

    https://pubpeer.com/publications/17604977
    Fig 5, 6 and 7 in Diabetologia 2007

    https://pubpeer.com/publications/12960006
    Fig 3, 4, 6 and 7 in Endocrinology 2003

    https://pubpeer.com/publications/19332486
    Fig 2 and 3 in J Physiology 2009

    https://pubpeer.com/publications/22897821
    Fig 4 in Critical Care 2012

    https://pubpeer.com/publications/17717055
    Fig 2 in Endocrinology 2007

    https://pubpeer.com/publications/23519983
    Fig 1 vs 2 in Obesity 2013

    https://pubpeer.com/publications/22828956
    Fg 4 in Diabetologia 2012

    https://pubpeer.com/publications/22677195
    Fig 1, 2 and 5 in Gastroenterology 2012

    https://pubpeer.com/publications/19888448
    Fig 2 and 3 in PLOS ONE 2009

    https://pubpeer.com/publications/16002548
    Fig 4 in J Endocrinology 2005

    https://pubpeer.com/publications/16638016
    Fig 4 in J Neurochemistry 2006

    The State University of Campinas was the affiliation Saad listed in this paper:
    http://ccforum.com/content/16/4/R158
    (is this not the correct affiliation?)

  • AMW February 17, 2015 at 5:01 am

    The list is beginning to look scary (with the Diabetes papers it looks like at least 20 papers are involved). The scale looks now to be in line with some of the biggest stories covered in the last five years (Boldt, Stapel etc..)

  • Liem February 19, 2015 at 2:57 pm

    Investigation was carried and signed by Paulo Mazzafera, Ronaldo Pili, Anibal Vercesi and Licio Velloso. They all currently share a grant as principal investigators with Professor Saad, according to FAPESP:

    http://www.bv.fapesp.br/pt/auxilios/87736/centro-de-biologia-quimica-de-proteinas-quinases-alavancando-desenvolvimento-de-farmacos-atraves-de-/

  • Post a comment

    Threaded commenting powered by interconnect/it code.