Two Dutch researchers were preparing a review of preclinical animal models for hemorrhagic stroke last July when they stumbled across a disturbing pattern in the literature.
First, they found many more papers on the topic than the 50 or so they expected based on their experience: more than 600.
Also, nearly every study proposed a different intervention, which was “very unusual,” said René Aquarius, a neurosurgery researcher at Radboud University Medical Center in the Netherlands. “Why would you show a very beneficial effect and then say, ‘let’s do something else?’”
Aquarius and his colleague, Kim Wever, an assistant professor in the department of anesthesiology, pain and palliative medicine at the same institute, didn’t have experience with scientific sleuthing. But they recently had seen Elisabeth Bik speak about image manipulation, and her experience inspired them to dig deeper.
The pair decided to collect images of Western blots from a random sample of the papers, project them on a meeting room wall, and gather two biomedical students to help with the detective work. Within a few minutes of scrolling through the images, Wever recalls someone exclaiming, “Hey, I think I saw that one before!”
The researchers’ investigation revealed unexpected overlaps between data in three papers of a neurosurgical group in Taiwan. They kept digging, and found problems in a total of 24 of the group’s papers which reused the same data while purporting to show different experimental conditions.
One article at a time, Aquarius and Wever pieced together their findings, posted them on PubPeer, and contacted the journal editors. Their efforts so far have led to nine retractions and two corrections. But the problems in the field extend far beyond this one group in Taiwan, they say.
In each of Aquarius’ posts on PubPeer, he states he “noticed a considerable overlap between figures published in different studies.” So far, an author has responded to only one of his posts. Chih-Zen Chang of the Kaohsiung Medical University in Taiwan claimed the figures are “original” and apologized for not being able to offer the original films since they “were spoiled in a damaged box.” Chang did not respond to requests for comment from Retraction Watch.
Chang frequently appears as a co-author of Aij-Lie Kwan, a neurosurgeon at the same institute and the former president of the International College of Surgeons. Kwan did not respond to requests for comment.
All nine of the retractions posted so far note that the authors did not respond to correspondence from the journal. In contrast, they seem to have offered responses regarding articles that now have corrections. In one correction, the journal wrote “the error stems from a misclassification in the authors’ computer folder,” In another, the authors claim to have “mistakenly used the same Western blot.”
At the time of writing, the retracted papers linked to Aij-Lie Kwan are:
- Glycyrrhizin Attenuates Proinflammatory Cytokines through a Peroxisome Proliferator-Activated Receptor-γ-Dependent Mechanism and Experimental Vasospasm in a Rat Model, Journal of Vascular Research, cited 15 times, according to Clarivate’s Web of Science.
- 6-Mercaptopurine attenuates adhesive molecules in experimental vasospasm, Acta Neurochirurgica, cited 17 times.
- A purine antimetabolite attenuates toll-like receptor-2, -4, and subarachnoid hemorrhage-induced brain apoptosis, Journal of Surgical Research, cited 7 times.
- Magnesium lithospermate B alleviates the production of endothelin-1 through an NO-dependent mechanism and reduces experimental vasospasm in rats, Acta Neurochirurgica, cited 13 times.
- Preconditioning with pitavastatin, an HMG-CoA reductase inhibitor, attenuates C-Jun N-terminal kinase activation in experimental subarachnoid hemorrhage-induced apoptosis, Acta Neurochirurgica, cited 12 times.
- 6-Mercaptopurine reverses experimental vasospasm and alleviates the production of endothelins in NO-independent mechanism—a laboratory study, Acta Neurochirurgica, cited 14 times.
- Purine anti-metabolite attenuates nuclear factor κB and related pro-inflammatory cytokines in experimental vasospasm, Acta Neurochirurgica, cited 8 times.
- Valproic acid attenuates intercellular adhesion molecule-1 and E-selectin through a chemokine ligand 5 dependent mechanism and subarachnoid hemorrhage induced vasospasm in a rat model, Journal of Inflammation, cited 12 times.
- 6-Mercaptopurine exerts an immunomodulatory and neuroprotective effect on permanent focal cerebral occlusion in rats, Acta Neurochirurgica, cited 7 times.
Aquarius described his experience with journal editors and ethics teams as “hit or miss.” He said some are “very dismissive and say, well, I don’t see a problem.” One journal declined to take further action on a paper after relying “in good faith upon the authors’ representations,” according to Aquarius’ post on PubPeer about the journal’s response.
He also noted the inconsistency of the journals’ decision making. The same set of data led to a retraction of one paper, but resulted in a correction for another. “It’s super weird to me that a publisher comes to a different conclusion based on the same type of evidence,” he said.
Other publishers were helpful and immediately started investigating, Aquarius said. One journal, Acta Neurochirurgica, which published six of the now-retracted articles, declared the editor-in-chief “no longer has confidence in the presented data” because of duplications and other errors with the data.
Wever and Aquarius said they see the problems and flag them, but leave the judgment for the journals. “Whether the problem is an honest mistake or sloppy science or even worse, scientific misconduct,that’s not up to us to decide,” Aquarius said.
The 600 papers found by the pair suggest there is extensive research into potential treatments for early brain injury. However, “it just gives a completely false idea of the research field,” according to Wever. For Aquarius, he hears firsthand about the consequences of hemorrhagic stroke from patients in a clinical study, and stresses that “errors in the scientific record need to be resolved.”
Their analysis so far has revealed no common denominator for the problematic articles. “It’s all over the world, all different types of universities,” Aquarius said. “From top universities to not very known universities… from the top journals to the worst.”
Like Retraction Watch? You can make a tax-deductible contribution 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].