8 years after three papers are flagged — and after losing original correspondence — PLOS ONE retracts

A group of nutrition researchers in Canada led by the prominent diabetes scientist Emile Levy has lost three papers in PLOS ONE over concerns about the integrity of the data. 

The concerns were raised nearly eight years ago by Elisabeth Bik, early in her career as a data sleuth.  

In May 2014, Bik told us, she contacted the journal to point out problems with images in the articles, as well as a fourth paper that has received an expression of concern:

I reported these four papers to the journal in May 2014 – a whopping 8 years ago! -, but from the wording in the notices it appears the journal lost the correspondence. So did I when I left Stanford, unfortunately.  If the journal had addressed these concerns a bit earlier, the authors might still have had the original data to show, since these papers were published in 2010-2013. Also note that some notices mention that the author(s) provided originals that did not always seem to match the published panels. 

The three retracted papers are:

• Modulatory Role of PYY in Transport and Metabolism of Cholesterol in Intestinal Epithelial Cells
• Oxidative Stress and Mitochondrial Functions in the Intestinal Caco-2/15 Cell Line
• Apple Peel Polyphenols and Their Beneficial Actions on Oxidative Stress and Inflammation

Although none of the papers would be considered highly cited, they did garner a modest amount of attention from other scholars – the vast bulk of which occurred since Bik first flagged them for the journal. 

The first article in that list has been cited 17 times, and 12 since May 2014, according to Clarivate Analytics’ Web of Science. The second: 33 cites, with 24 since May 2014. And the third: 87 cites, 82 of which came since Bik’s alarm.  

Here’s the retraction notice for the apple peel paper, which, as with the others (but unlike some at other journals), does not mention Bik by name: 

Following the publication of this article [1], concerns were raised about results presented in Figs 4, 8, 9, and 10. Specifically, the following results appear more similar than would be expected from independent samples:

The Fig 4D Occludin panel and the Fig 10A β-actin panel.

Lanes 1–2 of the Fig 8A COX2 panel, lanes 1–2 of the Fig 8B COX2 panel, and lanes 1–2 of the Fig 8C COX2 panel.

Lanes 3–4 of the Fig 8A COX2 panel and lanes 3–4 of the Fig 8B COX2 panel.

The Fig 8A β-actin panel, the Fig 8B β-actin panel, the Fig 8C β-actin panel, and lanes 2–5 of the Fig 8D β-actin panel.

Lanes 4–5 and lanes 6–7 of the Fig 9A NF-κB panel.

Lanes 1–3 of the Fig 9A IκB panel and lanes 1–3 of the Fig 9B IκB panel.

Lanes 1–5 of the Fig 9B NF-κB panel and lanes 1–5 of the Fig 9C NF-κB panel.

Lanes 4–7 of the Fig 9B IκB panel and lanes 4–7 of the Fig 9C IκB panel.

Lanes 2–3 and lanes 4–5 of the Fig 10A NRF2 panel.

Lanes 1–5 and lanes 6–10 of the Fig 10C β-actin panel.

The corresponding author stated that the similarities between reference (housekeeping) protein blots are due to the stripping and re-probing of the same blot, and commented that some blot images were spliced to remove extra lanes and/or duplicate or triplicate samples from the panel.

The corresponding author provided image data to support the results presented in Figs 8A, 8B and 9A. The underlying data for Fig 8B confirm that the panel has been spliced, but the underlying data provided for Fig 8A and Fig 9A do not appear to match the published panels. Overall, the data did not resolve the concerns about these figures.

The corresponding author stated that the underlying data for other figures were provided to PLOS ONE in 2014 but are no longer available in the laboratory records. PLOS is unable to access the journal’s 2014 correspondence records for this case. We sincerely regret that this case was not resolved much sooner after the prior correspondence.

Although we have been unable to review the primary data for several figures of concern, the nature and extent of unresolved image data reporting concerns call into question the reliability of the article’s results and conclusions. Therefore, the PLOS ONE Editors retract this article.

EL did not agree with the retraction. MCD, AF, SD, AM, CG, YD, and ED either did not respond directly or could not be reached.

Levy told us the controversy “has been terribly upsetting to me and my colleagues.”

We have been dealing with a recent PLOS ONE Ethics Officer who has been relentless, closed to any solution, coloured by partisanship and a lack of understanding, with no ethical sensitivity to the authors of the publications, some of whom are deceased, retired, or unreachable by email, and with no coherent analysis of PLOS ONE’s historical responsibility in this ethical issue. I leave you to read the exchanges with her and the journal’s editor.

We are in the process of organizing an appeal to the Committee on Publication Ethics (COPE) …

In correspondence with PLOS ONE, which we’ve made available here, Levy tells the journal’s Maria Zalm, senior editor of publication ethics, that his team sent all of the material the journal requested by the end of 2014:

Despite our insistence, the response from PLoS One did not arrive, and we had to wait until February 20, 2015 to receive the message below from Mrs. Sarah Wade, thanking us for the documents made available to PLoS One, and informing us that the internal editor and the academic editors will take note of it. Inexplicably, all our attempts until the end of 2015 to obtain PloS ONE feedback failed. At no time did PLoS One consider it useful to advise us that the information provided is correct. Also, all the phone calls made by my assistant to the Journal Office (San Francisco) yielded no results.

Dear Mrs. Zalm, you come back to this subject again 8-12 years after the articles were published and after the departure of my students, and 8 years after having done my duty with PLoS One with conviction, honesty and conscientiously. PLoS ONE as a respected journal should have finalized this sensitive issue when it was time. Currently, I no longer have my laboratory head coordinator (in knowledge of all the data from the various projects) because she retired four years ago; I no longer have contact with the students who had their protocols and results on their computer; moreover, most of the old files and laboratory notebooks from the last 40 years of research were eliminated during the move to the new Research Center a few years ago.

Levy asked the journal to stop the retraction process on March 17. The retractions appeared on April 14.

The expression of concern involves a 2013 paper titled “Iron-Ascorbate-Mediated Lipid Peroxidation Causes Epigenetic Changes in the Antioxidant Defense in Intestinal Epithelial Cells: Impact on Inflammation.” It has been cited 25 times in total, with 21 since May 2014.

Per the notice: 

The corresponding author stated that the underlying data for other Fig panels in question were provided to PLOS ONE in 2014 but are no longer available in the laboratory records. PLOS is unable to access the journal’s 2014 correspondence records for this case. We sincerely regret that this case was not resolved much sooner after the prior correspondence.

In the absence of supporting image data to confirm the image results presented in Figs 4B, 6A, and 6C, and the underlying individual level data to support the results presented in the graphs of Figs 3, 4, 6A, and 6B, we cannot clarify whether the reported claims based on these Figs are reliable. Given the concerns regarding the preparation of the data presented in Figs 3, 4, and 6, the PLOS ONE Editors issue this Expression of Concern to notify readers and relay the supporting data provided by the corresponding author.

David Knutson, a spokesperson for PLOS, told us: 

As we discussed in a recent blog post, PLOS is working through a backlog of cases that dates back to before we had a dedicated Publication Ethics team. The case involving these four articles is from this historical backlog. PLOS began discussing the concerns with the authors in 2014, and we reactivated the case in January 2022. Unfortunately, some of the 2014 records for this case are no longer accessible to PLOS staff, as we acknowledged in the public notices.

We retracted pone.0011817, pone.0040992, and pone.0053725 because the extensive figure concerns cannot be resolved without the original primary data. According to the corresponding author’s statements and PLOS’ current records, the primary data are not available at this time.

PLOS ONE is still investigating pone.0063456. Due to the anticipated time it will require to complete this work, we published an Expression of Concern in the meantime to notify readers of the concerns and provide the available supporting data.

This is not the first time the journal has cited a backlog of cases involving Bik’s work. PLOS ONE created a research integrity team in January 2018. By 2019, the journal had become a “major retraction engine.”

Like Retraction Watch? You can make a one-time tax-deductible contribution by PayPal or by Square, or a monthly tax-deductible donation by Paypal to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].