Rectitude question for a Nobel winner

A scandal is brewing about faked data in papers by Nobel-winning genetic researcher Gregg Semenza. The 67-year-old Dr Semenza (MD, PhD) shared the 2019 Nobel Prize for medicine “for discoveries of how cells sense and adapt to oxygen availability”. He is director of the vascular programme at The Johns Hopkins University School of Medicine.

Since October last year, he has retracted at least 10 papers, with the earliest dating back to 2005. This was after researchers pointed out falsified and photoshopped data. His work on how cancer cells adapt to oxygen-poor environments is considered crucial. The retracted papers had been cited over 750 times.

The Nobel citation reads in part “How cells adapt to changes in levels of oxygen has long been unknown. William Kaelin, Peter Ratcliffe, and Gregg Semenza (the three winners) discovered how cells can sense and adapt to changing oxygen availability. The discoveries may lead to new treatments of anemia, cancer and many other diseases”.

A Nobel Prize places a halo around the recipient’s head. More than the prize money (which is substantial) it is recognition. Moreover, since such awards are usually made years after the work itself, there’s usually a strong consensus on the lasting value. That consensus is now open to review.

Scientific research has a supposedly rigorous, multi-step validation process. Few papers are single-author — all the retracted papers in question are multi-author. But Dr Semenza is the last author (by tradition, the group leader).

There’s “internal” review in that authors examine one another’s contributions. There’s peer review before a paper is accepted for publication. Researchers then try to replicate results, and design experiments to extend the findings.

Many Nobel winners have retracted papers, of course. Frances Arnold (2018 chemistry) retracted a 2019 paper when she was unable to replicate results. Linda Buck (2004 medicine) has retracted papers she published in 2005 and 2006. Michael Rosbash (2017 medicine) and Jack Szostach (2009 medicine) have also retracted papers. In all these cases, however, the withdrawals came with apologies from the scientists. The retractions occurred fairly soon after publication, citing flaws discovered by the authors themselves.

Ten retractions in papers published over the period of 2005-20 is quite unusual. The spate of retractions also started in October 2022 after these papers were flagged by a pseudonymous researcher “Claire Francis”. Another seven papers where Dr Semenza is the last author have led to “expressions of concern” and could lead to more retractions.

So, many of the papers were “live” for over a decade. When a top-echelon scientific team publishes, millions may be spent on follow-up research. For example, according to Reuters, the US National Institutes of Health spent $588 million to research the use of stem cells to regenerate heart tissue. But this was a blind alley based on 31 falsified papers published by Dr Piero Anversa and his team at Harvard.

There are differences between wrong data, wrong interpretation of data, and falsified data. Science deals with measurements to micro-levels of accuracy and sophisticated statistical methods are employed to tease signals out of noise. Mistakes happen.

But under pressure to publish, researchers may also be tempted to produce deliberately faked results. When the data includes images and diagrams, and the images are photoshopped or otherwise tampered with, the evidence swings towards deliberate falsification.

How rampant is this cheating? Dr Elizabeth Bik, a Stanford microbiologist, has spent a lot of time looking into this. She claims she’s examined over 100,000 papers in her domains of competence, and found apparent image falsification in 4,800 papers and other signs of fabrication in an additional 1,700. Thus far, her reports have led to 950-odd retractions and corrections published in many more.

The US Office of Research Integrity, a watchdog under the Federal Department of Health & Human Services, is now examining complaints against Dr Semenza. But image and data manipulation appears to be very, very widespread if you trust the conclusions of “detectives” like Dr Bik. Given tools like Ai, it could easily proliferate. This could lead to even greater mistrust of science across an already suspicious civil society.