Exactly ten years ago today, on January 29, 2014, I wrote about two new Nature papers on so-called STAP cells. The papers claimed that stress alone could convert regular non-stem cells into some of the most powerful stem cells. More specifically, the authors claimed to make pluripotent stem cells similar to iPS cells this way. Maybe even totipotent cells. They called the cells “STAP cells” or STAP stem cells.
I reviewed the papers journal club style here on The Niche the day they came out. I was already skeptical then. In the weeks after that I felt increasingly convinced that the papers were not only wrong, but also there may have been misconduct. The whole STAP cells idea just didn’t add up.
Ultimately, the papers were retracted and this was an awful, even if temporary situation for the stem cell field and biology more generally.
Today, a decade later on the STAP anniversary, are things better in science related to research misconduct? What perspectives are there now looking back 10 years?
What was the scandal over STAP cells?
Before we get into that, let’s briefly go over what happened with the STAP cell situation.
As I wrote back then, an international team from Harvard and Riken reported:
“the astounding finding of reprogramming differentiated cells back to a pluripotent or even totipotent state simply by exposing the cells to extreme environmental stress, creating cells they called STAP cells.”
The key stressor was low pH. Other stressors like pressure may work too, the people involved said. However, it turned out that STAP cells were not real. The work was never replicated. Evidence of major misconduct surfaced. Much of the blame fell on first author Haruko Obokata. She or someone else on the team may have mixed real pluripotent stem cells into the STAP cell cultures too to try to get the desired functional outcomes. It’s not clear to what extent other authors on the teams may have committed misconduct.
There was an international media storm over this. Part of the reason was that this was a team from two famous institutions, Riken in Japan and Harvard here in the U.S. One of Obokata’s mentors was Harvard professor Charles Vacanti, well known for the Vacanti mouse.
Ultimately, Nature retracted the two STAP papers.
This all had huge negative impact on many people and institutions. One of the senior authors Yoshiki Sasai committed suicide, probably because of STAP.
Thinking about research misconduct 10 years later
Is science doing better now in 2024 a decade after the STAP cell situation? Perhaps the one main area where research has improved is greater awareness of misconduct problems. Sadly, misconduct itself is still probably just as common. However, sleuths and misconduct experts like Elisabeth Bik have raised awareness and documented problems, which is positive and important.
Every year the paper problems and research misconduct headlines just keep coming. For example, 2023 was the year that paper issues took down Stanford President Marc Tessier-Lavigne and Harvard President Claudine Gay. In the latter case, which involved extensive plagiarism, national politics played a major role too, which is unfortunate.
Just this week news broke on a major series of problematic papers from various Dana Farber Cancer Institute labs.
It’s like a never-ending stream. Why does this keep happening?
The ongoing problems exemplified by STAP cell situation
Part of the problem is at a cultural level where research papers are often expected to have results that perfectly fit a model. Usually data aren’t that simple but some scientists seem to expect an unnatural level of consistency and perfection in data.
For instance, kind of ugly, but accurate and interesting Western blot results can draw criticism from reviewers. Some people give in to the temptation to doctor the results, such as Western blot data, to fit their hypotheses and other’s expectations. The same goes for countless other research readout assays like pictures of cells or tissues or flow cytometry histograms. Purely quantitative data could easily be manipulated as well but there are fewer obvious indications of that so we just don’t know how often that happens.
Beyond reviewer expectations, there are also some labs where trainees feel intense pressure from the PI to produce data that look a certain way and fit the lab’s model. It shouldn’t be that way but it happens too often.
Cell biology research, for instance, should be about finding out the reality of biological systems, which are beautiful but can seem messy, sometimes counterintuitive to us humans, and don’t always fit into our models. Some scientists try to make the data fit the model anyway. Beyond data re-use or manipulation, there is also plagiarism, which is another form of misconduct.
Journals are now turning to computational approaches, perhaps even AI (more on this below) to scan submitted manuscripts for problems that could be associated with misconduct. A better but harder solution is changing the culture of science so misconduct is just not something people turn to so often. Is this even possible?
Some personal reflection on The Niche and STAP cells
Getting back to the STAP cell mess, the 10th anniversary is a good time for some reflection on what happened then. A decade ago, admittedly I was more idealistic and somewhat naive than now in doing this blog. In hindsight, I probably focused too much on the STAP cell mess here on The Niche in early 2014. I think part of it was that the concept seemed so intriguing but simple. At the same time for two Nature papers to be so wrong kind of blew my mind at first. Too good to be true. The people involved in STAP were also interesting characters in some cases.
It also took me a while to fully realize how intensely STAP unfolded specifically in Japan. It was a national scandal making TV news headlines every night for a while. At one point that spring Japanese reporters I didn’t know were somehow calling my home number in the middle of the night asking for comments. That was a sign to me that things were out of control in the media. Later in 2014 when Yoskihi Sasai committed suicide that August it struck home even more just how serious this whole thing had become.
Now looking back on the STAP cell situation a decade later, I still think of it in a somber light. I also just don’t know how far we’ve come in science since then on misconduct. Maybe not very far, which is discouraging.
It’s also unclear whether AI could make research misconduct worse in the future, for example by allowing researchers to quickly make fake data from scratch or manipulate existing data. Conversely AI might make research misconduct easier to spot. It’s likely going to be a combination of both facilitation of misconduct and making its detection easier.
Finally, looking ahead, unfortunately it seems likely there will be more situations like STAP. Maybe we can deal with them better. Make them less common?