A group of cancer researchers whose work has been questioned by sleuths has been hit with their third retraction in less than a year.
Today, Science Translational Medicine (STM) withdrew a 2021 breast cancer study by former Indiana University researcher Yujing Li and 12 other authors for image falsification. The immunotherapy study had been described by senior author Xiongbin Lu as a “game-changer” for triple negative breast cancer in a 2021 IU press release.
The paper’s April 15 retraction notice states that a joint research misconduct investigation involving Indiana University, The Ohio State University, and the University of Maryland, College Park determined “falsification occurred during creation of figure S9C.” The institutions alerted the American Association for the Advancement of Science of the misconduct late last year and requested the paper’s retraction, according to Meagan Phelan, a spokesperson for AAAS, which publishes STM.
IU spokesperson Mark Bode told us “research integrity is paramount” and that “any allegation of misconduct is investigated thoroughly.” He declined to answer further questions about the misconduct findings but noted that Lu is no longer employed by IU. Lu, a nationally-recognized cancer biologist and professor of medical and molecular genetics, was named to the role in 2017, according to the university’s Facebook page.
Representatives from The Ohio State University and the University of Maryland did not return messages.
The STM paper was supported in part by U.S. National Institutes of Health grants.
The retraction is at least the third for Lu and three coauthors to result from the universities’ investigation. In January, The Journal of Clinical Investigation retracted two papers by Lu for data falsification. The JCI notices both state the universities informed the journal of figure manipulation in the papers. One of the notices specifies misconduct findings by the universities against Lu and coauthor Hanchen Xu, while the other notice cites misconduct by Lu and coauthor Yunhua Liu.
Three authors on the retracted STM paper are also authors on both retracted JCI papers, including Yujing Li, Kevin Van Der Jeught, and Xinna Zhang. Zhang is Lu’s wife.
When reached by Retraction Watch, Lu said he was unaware of the retraction, despite the notice indicating he disagreed with the decision.
“To my knowledge, the journal has not reached or communicated a final decision,” he told us. “Reporting on this matter prior to an official determination would be inappropriate and premature.” Li sent a similarly worded response. Van Der Jeught and Zhang did not return messages seeking comment.
Kevin Patrick is one of several sleuths who have called out problematic data in Lu’s research over the past five years. Patrick, who goes by the known pseudonyms Cheshire and Actinopolyspora biskrensis, started looking at papers from the research group in 2021 after a researcher asked him to examine several articles. During his analysis, Patrick found overlapping and repeated figures in the group’s work, including in both papers retracted by JCI in 2026.
Patrick shared his concerns about the JCI articles on PubPeer in April 2021. He also posted on X and commented on PubPeer about overlapping images in a 2018 paper by Lu and colleagues in ACS Central Science. ACS retracted the article in 2023.
Based on his past discoveries, Patrick told us he was “not surprised” by the figure issues found in the STM paper. He noted that another sleuth who goes by Dendrodoa grossularia on PubPeer pointed out problems with the STM paper in March 2021.
In an exchange on PubPeer at the time, Lu apologized for the “unintended” error.
The images in the associated treatment group in Figure S9C were “taken by mistake from one mouse in the control group, which resulted in duplicated figures,” he wrote.
“While this mistake does not affect the conclusion of this mouse toxicity experiment, we deeply apologize for our sloppiness,” Lu wrote on PubPeer. “This has resulted in a change in our lab standard operation. In our future studies, we will name all the image files with detailed information instead of simple numbers.”
The STM paper has been cited 34 times, according to Clarivate’s Web of Science.
The retraction notice states that authors Kathy D. Miller and Bryan P. Schneider agreed with the retraction, while Lu and six other authors disagreed and four did not respond or could not be reached.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on X or Bluesky, like us on Facebook, follow us on LinkedIn, 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].
I am a computational biologist, not a lab biologist. Maybe someone with more experience with these photos can explain:
It’s easy to imagine how you get the same photo in there twice. But how do you get the northeast corner of a photo in one slot and the southwest corner of the same photo in a different slot? Why do you have those files in the first place? What’s the purpose of making multiple partially overlapping crops of the same photo? (Other than disguising the fact you’re re-using it, of course.)
Frequently, the original piece of tissue is substantially larger than the amount that can be displayed at a useful magnification in a figure (especially in a single panel of a figure).
The researcher has to choose a suitable ‘representative’ (ahem) region of interest when composing a figure. They might take multiple photos of the same tissue in different positions and magnifications, planning to select the clearest one to include in an eventual publication. They might not even know what the manuscript will look like or what size of image they’ll need when they take their photos, so they’ll generate a stack of pictures from different areas of the same slide.
Or they’ll just give the entire slide to the scanning facility in the microscopy/pathology core, and get a giant image file that has the whole section on it. Data storage is cheap. They then cut out an appropriate region to paste into the figure. If you’re sloppy about recordkeeping or file naming in either of those workflows, you can end up with multiple images from the same section inadvertently.
Rotations can be carried out ‘honestly’ when a feature is inconveniently positioned (where a scale bar or caption would cover something important), or when a long feature (like a blood vessel) has to be oriented a particular way to fit into a non-square figure panel.
So in principle it’s possible to make errors like this inadvertently. It’s not a good look when it happens multiple times, across multiple publications from the same lab.