SfN journal retracts paper, bans UPenn researchers over “data misrepresentation”

journal of neuroscienceThe Journal of Neuroscience has yanked an Alzheimer’s paper and banned three University of Pennsylvania authors from publishing there temporarily, following conflicting investigations by the university and the publisher, the Society for Neuroscience, into the data.

The 2011 paper looked into the cellular makeup of the characteristic plaques that develop in the brains of people with Alzheimer’s disease. It’s been cited 64 times, according to Thomson Scientific’s Web of Knowledge.

While the notice states that Penn’s investigation “supports the journal’s findings of data misrepresentation,” last author Virginia Lee said she asked the journal to simply issue a correction of the faulty data, since the findings are “extremely important” for the field and have been affirmed by a later paper. According to author John Trojanowski (who is married to and publishes regularly with Lee), he and Lee have been barred from publishing in Journal for Neuroscience for several years. Senior Co-author Edward Lee is out for a year [see update at the bottom of this post].

Lee provided us with a letter Vice Dean of Research Glen Gaulton sent to the journal (click here to read), in which he says an investigation found “no evidence of research misconduct” and the “errors…do not detract from or otherwise alter the conclusions of the manuscript.”

Ultimately, however, the journal decided to retract the paper. Here’s the notice for “Intraneuronal APP, not free Aβ peptides in 3xTg-AD mice: implications for tau versus Aβ-mediated Alzheimer neurodegeneration”:

The Journal of Neuroscience has received notification of an investigation by the Perelman School of Medicine at the University of Pennsylvania, which supports the journal’s findings of data misrepresentation in the article “Intraneuronal APP, Not Free Aβ Peptides in 3xTg-AD Mice: Implications for Tau Versus Aβ-Mediated Alzheimer Neurodegeneration” by Matthew J. Winton, Edward B. Lee, Eveline Sun, Margaret M. Wong, Susan Leight, Bin Zhang, John Q. Trojanowski, and Virginia M.-Y. Lee, which appeared on pages 7691–7699 of the May 25, 2011 issue. Because the results cannot be considered reliable, the editors of The Journal are retracting the paper.

We spoke to Trojanowski for more details:

Last April, we got an email about an inquiry into figures that I would call erroneously used. An error was made by [first author] Matt Winton, who was leaving science and in transition between Penn and his new job. He was assembling the paper to submit it, there were several iterations of the paper. One set of figures was completely correct – I still don’t know what happened, but he got the files mixed up, and used erroneous figures to illustrate points that are valid and have been confirmed by others. SfN noticed they were erroneous and launched an investigation.

Without any communication, [the SfN ethics committee] came back and said, these are the sanctions. There’s never been disclosure about their process. What’s quite puzzling is the fact that they didn’t abide by the Penn committee’s investigation, and we have no idea what investigation they did. SfN has been completely opaque about their process. So we’re completely unaware why they dismissed the Penn assessment and went ahead with the sanctions and the retraction.

It’s not the American way to have a secret process investigating a potential wrongdoing. Penn showed there was no ill intent, it was a mistake and mistakes happen. I’m not excusing it – we take full responsibility and intended to publish a corrigendum to correct our mistakes. SfN foreclosed on that option.

Here’s more from Lee:

In a nut shell, Dean Glen Gaulton asserted that the findings in the paper were correct despite mistakes in the figures. I suggested to J. Neuroscience that we publish a corrigendum to clarify these mistakes for the readership of J Neuroscience since the findings of this paper are extremely important for the Alzheimer’s disease field because it provided convincing evidence pointing out that a previous report claiming accumulation of intracellular Abeta peptide in a mouse model (3XFAD) is wrong (Oddo et al., Neuron 2003), as evidenced by the fact that this paper has been cited by others for 62 times since publication.   Subsequent to our 2011 J. Neuroscience paper, others also have found no evidence of intracellular Abeta in the 3XFAD mice (e.g. Lauritzen et al., J. Neurosci, 2012).

SfN president Steve Hyman declined to comment on the case, citing confidentiality.

Hat tip: Rolf Degen

Update 3:40 p.m. EST 3/2/15: We spoke with Edward Lee, who was not a senior author but a post-doc at the time of the publication. He told us he received a letter from SfN in early February overturning his publication ban, and that he didn’t join the team until a year after the figures were assembled.

Like Retraction Watch? Consider supporting our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, and sign up on our homepage for an email every time there’s a new post.

20 thoughts on “SfN journal retracts paper, bans UPenn researchers over “data misrepresentation””

  1. “An error was made by [first author] Matt Winton, who was leaving science, …” and “’m not excusing it – we take full responsibility”

    When people in positions of power say “we take full responsibility”, what do they really mean? Is it just an empty phrase? The prototypical example is, to my mind, Governor Chris Christie who also “took full responsibility” for the bridge closure, and then, in the same breath, claimed innocence and pointed the finger at his subordinates. Taking full responsibility means that you accept that you, personally, are at fault.

  2. I wonder at what stage of the publication process the error was made. If the erroneous figures were in the initial submission, I’d say this is a huge failure of the review and editorial process (of course, this does not dismiss the authors either). As a reviewer, I once encountered a manuscript that was so badly mixed up (mislabeled or missing figures and tables), that I sent it back to the editor without review and requested a “do-over” from the authors. SfN should consider investigating themselves.

  3. I completely agree with eeke.
    Lotsa things dont add up and finger pointing back to a student “leaving science” seems a tad too convenient.

  4. The institution’s letter raises an interesting question. Dean Gaulton’s letter stated that the institution reached an investigative conclusion but conducted only a formal Inquiry. (I hope I am reading that correctly.) If there was PHS jurisdiction in this matter (and I have not looked to see), technically Penn’s process would be in potential violation of the PHS regulations governing those matters. This is because the purpose of an inquiry governed by the PHS Regs is only to see if the evidence warrants the kind of fact finding that occurs only in formal investigation, and only if the latter is found to they have to notify ORI. Once you start drawing detailed conclusions, and are engaging expert opinions, you are really doing an investigation.

  5. This is a prime example while retraction notices must be transparent and as specific possible. This one was not.
    Here, the journal applies such enormous measure of punishment, while the authors and their institution claim it all was a simple misunderstanding. I personally tend to believe the less biased party here, i.e., the journal and its publisher, the Society for Neuroscience, but whether the retraction and ban were justified, the facts should be there for the readers to make their own judgements.

  6. I agree with the 2 previous posts that it appears very convenient to have the first author gone from science to place the responsibility on him. SFN should conduct an independent investigation. Trojanowski and wife Lee are very powerful people at Perelman. It sounds convenient (if not disingenious) for them to say: “It’s not the American way to have a secret process investigating a potential wrongdoing”. Maybe there are situations (in which the risk of potential obstruction) require a secret investigative process for the truth to be known.

  7. Dear Dr Hyman,
    Let me congratulate on your principled decision. The letter from U Penn is without substance and borders on the insulting. Work that is full of errors is inferior by definition and has no place in a quality journal. It is surprising how difficult it is for some to understand this simple truth.

  8. Lee: “as evidenced by the fact that this paper has been cited by others for 62 times since publication” Dr. Lee and Dr. Trojanowski, could you be so kind, as soon as you can, and now that your paper is no longer officially published, i.e., retracted, to please contact the editors of the 62 journals that have referenced your paper, and request them to kindly add an erratum to those papers’ reference lists. That would now be the logical and responsible thing to do to maintain the integrity of the downstream academic record that has widely referenced your paper.

    1. Can you please cite an example of a correction published solely to delete a reference after it was retracted? I have never seen this, whether at request of the authors of the retracted papers or anyone else. Is this expected in any field?

  9. Here is an unfortunate example of how an apparently minor (I won’t say honest) mistake erupts into a conflagration. The underlying mechanisms: an opaque journal investigation, an opaque institutional investigation, and senior author egos that appear to be Himalayan.

    The history of JoN regarding its inability to properly handle retractions is well documented on Retraction Watch, as are the many failures of institutions to conduct fair, conflict-free, and transparent investigations of their own research grant cash-cows. I’ll focus instead on the latter, because a clear lesson is here provided on how senior authors should NOT respond when mistakes of this type are revealed.

    First, even if it is richly deserved, do not publicly throw a student/postdoc/underling under the bus, as senior author Trojanowski did to Winton. The only result, unsurprisingly, is additional suspicion placed upon the senior author. Second, do not take the stance of senior author Lee, who believes that citation number equates with scientific validity. Such a petulant and absurd response only serves, as above, to place additional suspicion upon the senior authors. Third, along with talking the talk (Trojanowski: “we take full responsibility”), walk the walk: when a mistake becomes evident to senior author(s) immediately and decisively request an unconditional retraction. By truly owning the mistake a corrected paper could have quickly been prepared and submitted, the weird journal sanctions (which are themselves worthy of further investigation by the publisher) would likely have been avoided, the institutional investigation would likely have never seen the light of day, and the entire incident would have quickly been forgotten.

    But no. Time and again we see on Retraction Watch examples of researchers who allow their egos to trump proper professional behavior, with painful results. Perhaps Trojanowski and Lee truly are the most important scientists on the face of the planet, upon whom Nobel Prizes should regularly shower. It doesn’t signify. The fact is, if they had behaved from the outset not as politicians with bruised egos but as scientists interested solely in correcting a minor mistake, this entry on Retraction Watch would probably not exist.

  10. With a little digging I was able to find this webinar:
    http://www.alzforum.org/webinars/intraneuronal-av-was-it-app-all-along
    The comments in particular seem to all say that Winton’s conclusions are not so, due to inadequate sample preparation for staining– specifically the use of paraffin vs frozen sections prepared with formic acid etc. Without the right samples and preparation, they couldn’t make the inference that there was no Abeta inside the cells (IMHO.)
    I think, dumb little me, that the conclusions of Winton’s paper were highly controversial from the beginning (June 2011) and that the figures in particular were a problem.
    Now does this mean the paper is wrong? What do I know? But at least there is/was significant skepticism about it, leading me to suspect that, if the figures can’t be trusted, it should definitely have been retracted– regardless of whether the authors claim that “the conclusions were correct anyway.”
    Any comments on a basic level as the the science behind this??

    1. The “fun fact” is that : indeed other (Lauritzen et al for instance) have found that there is no intra Ab42 but they forgot to say that no full lenght App was found either ! (just look at the paper). I like how Virginia Lee just drop a rather cryptic “thers also have found no evidence of intracellular Abeta in the 3XFAD mice”, missing half of the point. This paper is indeed kind of big deal, 3XFAD mice from Laferla are one of the most widely used mouse model in Alzheimer research so there is lot of paper builded on the fact that there is intraneuronal Abeta accumulation. But of course, I don’t think it will block publication of a paper with strong evidence. Laferla, at the time of his paper on the 3XFAD model, may not have the background/antibodies knowledge to know exactely what were accumulated in neuron, Ab42 fited just well with the pathology and the evidence we got at this time. So it’s an “honnest” mistake, even if it may have driven scientist on wrong way.

      For the one who are not familiar with AD pathology and Abeta, there is also a lot of antibodies used for detected Abeta 40/42 or C-terminal fragment or APP and some can detect both or all of the fragment, or only the agreggated fragment form (or only the free fragment form). So it’s very easy, depending on the preparation or the antibody specificity, to be wrong. And some antibodies were believed to only detect one form or another and later it was shown that theire were not so specific.

      I think the Virigina Lee paper may be wrong on some aspect and kind of right on other (like Laferla first paper) but with today knowledge the mistake are way less forgivable and sound way more like a rushed paper.

    2. I also used to work in the Alzheimer’s disease modelling field and that webinar was widely followed. In the lab I worked in and from conversations with others in the field there was skepticism about the reliability of the 3xFAD mice and a general frustration that the model was not as described in the original paper (http://www.ncbi.nlm.nih.gov/pubmed/12895417?dopt=Citation). This is by no means unique to this model; around the time there was a proliferation of genetically modified models for studying Alzheimer’s disease and many models were poorly derived (the breeding and back crossing procedures followed were frequently questionable, reducing the reliability of the models in the hands of others) and added little to the field. This particular model was unique in recapitulating features of Alzheimer’s that other models did not, particularly the “tangle” part of the “plaque and tangle” pathology and so was deemed extremely significant and high-profile. However, despite widespread usage of the model, it proved difficult to replicate findings outside the LaFerla lab (from my own experience with it and from conversations with others) and prior to the Winton paper, work demonstrating the lack of intracellular APP in this model remains unpublished, most likely through self-censorship (when people thought they were doing something wrong) and difficulty in getting work contradicting a fairly established finding published. So in this regard the Winton paper was extremely well regarded in the field, not least because it confirm researchers’ observations but also because it came from a lab with an reputation for producing good science. At the same time there were problems with animal breeding procedures reported in the original paper (compare the reporting in the original paper to the guidelines here http://www.ncbi.nlm.nih.gov/pubmed/18778401 and chapter 2 of this book http://books.google.co.uk/books?id=CTES09IWtm0C&lpg=PR7&ots=HP2etyhw9g&dq=behavioural%20genetics%20transgenic%20guidelines&lr&pg=PA16#v=onepage&q&f=false).

      With this background in mind, LaFerla’s comments were seen (again from my conversations with others in the field) as being an example of his petulance and annoyance that others’ findings did not agree with his (most of the papers he cites in defense of his arguments have his name on them). In 2008 my PhD supervisor and a post doc in our lab attended the International Conference on Alzheimer’s Disease (ICAD, the Alzheimer’s Association conference and original launching point for Alzforum) where they reported some rather vigorous, argumentative and downright angry questioning of LaFerla after his presentation. LaFerla was dismissive of researcher’s concerns and willing to put his name on papers using the model, regardless of their quality (e.g., 1: http://www.ncbi.nlm.nih.gov/pubmed/21419194; 2: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2893673/), causing mounting frustration from people unable to publish contradictory findings and wasting time on a model that was not as described.

      Brief explanation of example paper problems:
      1. Out of focus microscope images that are at a uselessly low magnification, dreadful Western blots, no allegation of fraud but it’s work that would have been rejected at a lab meeting anywhere that I have worked.

      2. Less bad but similarly, uselessly low-res microscope images and terrible Western blots).

  11. People have argued previously that an important question is “Would this paper have been accepted for publication in the absence of the borked Figures?” And if the answer is “No” (because the Figures were a crucially persuasive part of the evidence), then it doesn’t matter whether the authors’ conclusions turn out in retrospect to be correct; the authors are not entitled to priority over other authors who reached the same conclusions but put more effort into correct illustrations.

  12. For the people interested in the matter, Alzforum have posted on the topic (http://www.alzforum.org/news/research-news/mistakes-prompt-retraction-controversial-paper-and-publication-ban)
    It is stated that SfN have uncover image manipulation (rotated image or image magnification of one same image represented as different caption) but they do not state if it was a ‘routine check’ or if someone pointed the matter to the journal.
    The senior author claim that they also have been restricted to publish in the journal for an untermined number of years “In addition, the authors said they have been banned from publishing in the journal for several years”
    There is some more detail on upenn inquiry too.

  13. A puzzle in another paper from same lab (different from the mistake they published in the Erratum): Yoshiyama et al. 2007 Neuron 53, pg 337–351
    Worth looking at Fig. 7 I and J panels.

    1. Nope… Panels E and F were explained as mistakes in an Erratum. The panels below (I and J) are where the overlap is a bit more difficult to find, so it has never been noticed, it seems.

      1. Missed the erratum. But yeah, it’s concerning that there’s two errors in one figure and then the J. Neuroscience bit. Especially since in both of these cases, the color seems ot have been enhanced as well.

Leave a Reply to Harsha Radhakrishnan Cancel reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.