Scientist in a laboratory analysing a western blot result on a transparent film

Images showing the results of experiments in molecular-biology research papers are often scrutinized on websites such as PubPeer. Credit: Shutterstock

Several research articles co-authored by Nobel-prizewinning geneticist Gregg Semenza are being investigated by publishers after internet sleuths raised concerns about the integrity of images in the papers. Journals have already retracted, corrected or expressed concerns about 17 papers over the past decade, and others are investigating image- and data-integrity issues in further studies.

Semenza, who works at Johns Hopkins University in Baltimore, Maryland, shared the 2019 Nobel prize in physiology or medicine with two other scientists for discovering how cells sense and adapt to oxygen availability in the body. He published his Nobel-prizewinning work in the 1990s; the latest concerns focus on related molecular-biology research published since.

Image integrity in scientific papers has come under increasing scrutiny in recent years, as digital tools have made it easier for scientists to manipulate their results. There can be legitimate reasons to alter images — to make results clearer by raising contrast or colour balance, for example. Figures might also be mistakenly mislabelled or become distorted while the paper is being prepared. But image-editing tools can also be used to create fraudulent results.

Elisabeth Bik, a prominent image-integrity consultant in San Francisco, California, who is among those who have pointed out irregularities in work co-authored by Semenza, says that the number of corrections seems reasonable for a 20-year period in a successful lab and many of the concerns potentially fall into “sloppy science territory”. But “five retractions for papers with image manipulation is much more than one should expect”, she adds.

Semenza did not respond to requests for comment from Nature’s news team.

Image concerns

Commenters on the website PubPeer — where users scrutinize published research, often anonymously — have questioned images in 52 articles co-authored by Semenza that were published between 2000 and 2021. Since 2011, 17 of these papers have been retracted, corrected or had an expression of concern issued on them. The editorial notices cite the potential alteration, reuse or incorrect labelling of images showing experimental results. Another 15 of the papers are currently under investigation at their respective journals, Nature’s news team has learned.

Across the 32 papers that have so far drawn publisher scrutiny, all list Semenza as an author, but there are many combinations of different co-authors. Semenza is the corresponding or co-corresponding author on 14 of these papers, which cover research related to the molecular mechanisms of oxygen sensing in different types of cancer, and the function and dysfunction of blood vessels, among other topics. No wrongdoing has been proven, and with a lack of clarity about who contributed what to the papers, it is unclear who might have been responsible for any errors or problems with images. Corresponding authors do, however, carry responsibility for ensuring a paper’s overall integrity.

In Bik’s opinion, “the fact that there are multiple papers now retracted for manipulated images, and several others still under investigation suggests an intention to mislead”.

Retractions and corrections

The first post about Semenza’s work appeared on PubPeer in 2015, but most posts are from 2020 and 2021. Journals had issued one retraction in 2011 and corrections in 2013 to two papers Semenza co-authored, but the rest of the editorial notices have appeared in the past two years.

In 2021, five journals issued corrections on five papers because of errors including mislabelled data and the apparent re-use of images. In March, the journal Cancer Research corrected one paper and issued an expression of concern on another after an investigation found that the authors inadvertently presented the same data as results of different experiments, and reused data from an earlier publication.

And last month, the Proceedings of the National Academy of Sciences USA retracted four cell biology papers co-authored by Semenza and corrected three others. The editorial notices describe concerns about figures including possible data duplication — where one set of results are used for more than one experiment — and ‘splicing’ of immunoblot images (when specific parts of an image are cut out and relocated). In three of the four retraction notices, the authors say that updated figures or confirmatory experiments are detailed in new articles uploaded to the bioRxiv preprint server. And in all of the retraction notices, the authors say they believe the overall conclusions of the work remain valid but they are retracting the work because of concerns over the figures.

A spokesperson for Johns Hopkins University says that the institution “maintains the highest standards for accuracy and integrity in research” and takes allegations of impropriety seriously. It adds that there are “strict protocols and processes in place to vet any such allegations and to determine an appropriate path forward, if necessary”. They declined to disclose details of these review processes, or to comment on whether there were any specific allegations against Semenza or his group.

Further investigations

The dozen papers currently under investigation include one paper in Nature Genetics and one paper in Oncogene, which are published by Springer Nature (Nature’s news team is independent of its publisher). Science Signalling, which is looking into two papers, has concluded an investigation into one and says that it will publish an erratum soon.

Another title, The Journal of Physiology, says that it is reconsidering its position on two papers that it had previously investigated in light of the recent retractions. A spokesperson for the journal says that it didn’t take action after the first investigation because the original data were unavailable and “the resolution of the figure in the published paper was too poor”.

Seven journals that published 20 of the papers co-authored by Semenza that have received comments on PubPeer told Nature that they were aware of the criticisms raised but declined to comment on them. Three other journals made no comment, two said they were not aware of the allegations and were not investigating the papers and one did not respond to requests for comment.

A researcher in the field, who wished to remain anonymous, says that they are awaiting the results of the investigations “with a mixture of concern and interest”. (Several other researchers in the field contacted by Nature’s news team declined to discuss the irregularities in papers co-authored by Semenza.)

The researcher says that Semenza’s most influential contribution to research on oxygen-sensing — the identification of a protein complex called HIF-1 — has stood the scientific test of being reproduced and built upon by others. “The work under discussion [on PubPeer] does not have the same broad significance, although its total extent is large.” It remains to be seen whether the problems with images affect the papers’ conclusions, the scientist says.