University suggests journal correct diabetes paper. Publisher retracts it.

After a publisher learned there may be issues with a 2008 diabetes paper, it asked the author’s university to investigate. The university found evidence of image duplication, and asked the journal to consider correcting the paper.

Instead, the journal has retracted it.

The backstory involves diabetes researcher Kathrin Maedler, who has one previous retraction, as well as multiple corrections. Even the paper in question, published in Diabetes, received an erratum in 2014 regarding a duplicated image, as well as an expression of concern last August after the American Diabetes Association questioned “the reliability of the data” in both the article and erratum.

According to the expression of concern, the ADA asked the University of Bremen to investigate the issues in the paper. In October, the University of Bremen concluded that several duplications present in her work were the result of negligence, not misconduct.

To resolve the issues, the university’s rector recommended that the authors publish a second erratum with corrected figures. But after conducting its own review, the ADA overruled the university, opting instead to retract the paper:

At the recommendation of the Rector of the University of Bremen, the authors proposed to publish an erratum to correct “the wrongly placed western blot band.” This recommendation is also expressed in the Rector’s statement about the results of the university’s investigation (6), which states that the Rector “obliged Dr. Mädler to communicate the image duplications identified in the investigations to the publishers of the concerned journals and possibly publish Errata in consultation with the co-authors.”

On the basis of its review, however, ADA’s Panel on Ethical Scientific Programs (ESP) has concluded that the multiple confirmed duplications involving this article, one of which was introduced with the publication of the 2014 erratum (2), compromise the overall reliability of the data. Furthermore, the ESP has determined that such concerns regarding the reliability of the data cannot be sufficiently mitigated by publishing another erratum with another image; therefore, the Panel recommends to retract the article, and ADA has accepted this recommendation.

Maedler told us that she strongly objected to the retraction of the 2008 paper, and “there is no question on the reliability of our data.” The experiments have been repeated by independent students in the lab, she said, and the data are fully reproducible.

The paper, “Transcription Factor 7-Like 2 Regulates β-Cell Survival and Function in Human Pancreatic Islets,” has been cited 169 times, according to Clarivate Analytics’ Web of Science, formerly part of Thomson Reuters.

A spokesperson at the University of Bremen declined to comment on the publisher’s decision to retract the paper.

The past few years have been rocky for Maedler, a rising star in diabetes research. In 2014, commenters on PubPeer flagged more than a dozen of her papers, raising questions about the images. Both the University of Zurich, where Maedler conducted her doctoral research, and the University of Bremen, where she is now affiliated, began investigating her work. Both institutions found no evidence of misconduct, though the University of Bremen concluded that Maedler acted negligently.

After the University of Bremen’s investigation, the German Research Foundation decided to revoke the prestigious Heisenberg professorship it gave Maedler in 2014.

This isn’t the first time the ADA has disagreed with a university’s investigation of one of its researchers. In 2015, the publisher went head-to-head against the University of Campinas over its conclusions about images produced by diabetes researcher Mario A. Saad, who sued the ADA to stop retraction of his papers. (He lost.)

We’ve also uncovered another correction issued in April for one of Maedler’s papers, published in the Journal of Clinical Investigation (JCI).  According to the notice, three images in the 2002 paper were “inadvertently incorporated” from a 2002 PNAS paper.

Last author of the JCI paper Marc Y. Donath, from University Hospital Basel in Switzerland, told us that the duplication “is very simple, a mistake that could happen to everyone.” Essentially, according to Donath, some figures from the PNAS experiments ended up in the JCI folder because they were given the same number code. Donath added he is confident that the data in the PNAS paper are correct because:

…all figures are in the same original folder from 12/2001 and were presented with its original data set during our weekly laboratory meetings at that time.

But, because of the coding error, some of the figures from the PNAS paper ended up in the JCI folder:

We did not recognize this at the time of data assembling, but we see now, that these figures cannot belong here, because of their different date of origin (April/2001).

Donath noted that, despite the figure duplication, the results of the JCI are accurate:

… the main findings of this publication have been reproduced by around 50 independent follow up publications. I am aware that 2-3 groups claimed that they cannot reproduce it, but the technical issues underlying these differences can be easily addressed and have been partly published…. Last but not least, more than 12 clinical studies have confirmed the findings

Here’s the correction notice:

The editors recently became aware that three images in Figure 4F of this article are duplicated in a 2002 Proceedings of the National Academy of Science of the United States of America (PNAS) publication by this group (1). The specific panels of Figure 4F that were duplicated are the Fas-stained islets cultured in 5.5 mM glucose + IL-1[beta], the insulin-stained islets cultured in 5.5 mM glucose, and the insulin-stained islets cultured in 5.5 mM glucose + IL-1[beta]. The images appear in the PNAS publication as representing different treatment conditions. The authors were able to provide the original source data for both the JCI and PNAS figures. They determined that the correct images appear in the PNAS paper; however, the same images were inadvertently incorporated into the JCI paper due to similarities in the blinded code file numbers assigned to the correct images. The authors also determined that the incorrect image was used for the Fas/insulin double-stained islets cultured in 33.3 mM glucose. The corrected panel appears below.

“Glucose-induced β cell production of IL-1β contributes to glucotoxicity in human pancreatic islets” has been cited 626 times.

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 new daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.