On Tuesday, we broke the news of the retraction in Circulation of a paper on cardiac stem cells by a group of researchers being investigated by Harvard Medical School and the Brigham and Women’s Hospital. Today, The Lancet has issued an Expression of Concern about another paper led by Piero Anversa, the last author of the Circulation paper.
Here’s the notice:
On March 25, 2014, Gretchen Brodnicki, Harvard Medical School’s Dean for Faculty and Research Integrity, wrote to The Lancet to inform us that “Harvard Medical School (HMS) and Brigham and Women’s Hospital (BWH) are reviewing concerns about the integrity of certain data generated in a laboratory at BWH and included in the following published paper…[the] 2011 Lancet SCIPIO paper”.1 The focus of this investigation is on two supplemental figures published online (figures 2A and 3). As far as we are aware, the investigation is confined to the work completed at BWH. Gretchen Brodnicki continues, “Because review of this paper is ongoing, we cannot provide additional details at this time.” In further discussions with Harvard Medical School, we have been told that the current investigation into this aspect of the work reported in the SCIPIO trial is likely to take several months. As soon as The Lancet receives further information, we will inform readers accordingly.
The paper has been cited 270 times, according to Thomson Scientific’s Web of Knowledge. Anversa is the last author, and the University of Louisville’s Robert Bolli is the corresponding author.
Bolli was quoted at the time the study came out saying that the work might be “the biggest advance in cardiology in my lifetime.” Here he is discussing work based on the results of the study, known as SCIPIO, in 2012.
Update, 6:20 p.m. Eastern, 4/11/14: Anversa tells Retraction Watch:
As stated in the retraction notice in Circulation, there in an institutional review at Harvard Medical School and Brigham & Women’s Hospital regarding this matter that is ongoing. I am fully cooperating with that process, and cannot comment on it due to its ongoing nature. I continue to believe, based on research in our laboratory and elsewhere, that human cardiac cells regenerate at a higher rate than has been stated in other published research.
I welcome further study of these important issues.
I am, unfortunately, not able to comment or discuss the matter further.
Also, Larry Husten at CardioBrief reports:
According to knowledgeable sources, the Harvard letter requested a full retraction of the article by the Lancet. But without more details about the concerns raised by the Harvard investigation the editors thought an expression of concern was the most they could justify.
I say this again, if you commit fraud or allow it to be committed in your lab, the default action should be that you get fired and your position is replaced by an outside candidate. Plenty of young scientists that won’t commit fraud ready to take those jobs.
I disagree with this suggestion, as it would provide a crass incentive for bogus accusations. Furthermore, why would outsiders (“young” or not) be less likely to commit fraud? By definition, we don’t know who is committing fraud! These things are highly situational, and I can imagine plenty of young scientists getting corrupted by the rules in the system, but having had no intention whatsoever to commit fraud when they started out.
Let’s say 5% commit fraud and 95% do not commit fraud. The second person on the list of candidates for that job would have been less likely to commit fraud.
I appeal to the Lancet to retain this study in the literature. When I go to conferences I habitually bet fellow attendees the cost of dinner, that they can’t guess the lowest number of randomized controls needed to publish a randomized trial in the Lancet. This trial shows that it is currently 7 minus 4 minus 1 (see legend to Figure 1 on thelancet.com, and if further clarification needed Chugh A R et al. Circulation 2012;126:S54-S64). This comes to 2.
If this landmark paper is lost from the literature I shall have to return to buying my own meals.
One warning to those who may try to cash in on this easily won bet: make sure you show them the official article in the Lancet, which does include the word “randomized” in the title and a very helpful legend for Figure 1. When I once (out of laziness to enter passwords etc) used the version on EuroPMC, the title and figure legend turned out to be subtly different, which cost me the bet, against the sharp-eyed stickler Dr Michael Bellamy.
I found that people randomize their mice differently than I do.
I had 22 mice in 5 cages and so I set about splitting them into 2*(4,4,3) group.
I set out 6 fresh cages and took the first pooled animal an placed in the first new cage, the second in the second new cage and so on. In the end I had 4*4 and 2*3. I grabbed 2*4 and 1*3 and made that Group A and the remainder Group B.
My animal technicians were given the choice of which group would be given drug and which would be given saline. I thought that I was randomizing.
However, I was told that I should have weighed each animal, ranked, and then split in ranked pairs, to that the weights of the two groups had the same mean and SD.
So have I screwed up? Is my random-randomizing wrong and I should have done a pairwise ranking to make sure that my two groups had identical weights?
I think it needs to be said that the work from Bolli and Anversa has been highly controversial for a very long time. Half of the cardiovascular community has doubted the authenticity of their work and the other half have feared their influence in both Circ and Circ Res as well as their influence on NIH grant panels. For people in the field, this is not at all shocking. I will admit to being glad that it is finally out.
Sounds like his work has long contained the seeds – or ‘stems’ if you will – of this scandal.