A group of exercise researchers at the University of Rome Tor Vergata has lost their May 2010 paper in BMC Physiology on the effects of marathon running on blood cells, because of figure irregularities.
The article, “The effect of marathon on mRNA expression of anti-apoptotic and pro-apoptotic proteins and sirtuins family in male recreational long-distance runners,” purported to find that marathoning arrested apoptosis, or programmed cell death. It has been cited 13 times, according to Thomson Scientific’s Web of Knowledge.
According to this press release:
Gabriella Marfe from the University of Rome ‘Tor Vergata’ led a team of researchers who studied ten amateur athletes after a 42km run. Marfe said, “Apoptosis is a normal physiological function dependent on a variety of signals, many of which can be modulated by strenuous exercise. Here, we’ve shown for the first time that exercise modulates expression of the sirtuin family of proteins, which may be key regulators of training.”
Or, as the Telegraph put it:
Strenuous exercise can temporarily halt a natural process which causes cells to die, they found.
This effect could help the body to maintain bone and heart muscle, they believe.
Here’s what the study reported, in more detail:
Background
A large body of evidence shows that a single bout of strenuous exercise induces oxidative stress in circulating human lymphocytes leading to lipid peroxidation, DNA damage, mitochondrial perturbations, and protein oxidation.
In our research, we investigated the effect of physical load on the extent of apoptosis in primary cells derived from blood samples of sixteen healthy amateur runners after marathon (a.m.).
Results
Blood samples were collected from ten healthy amateur runners peripheral blood mononuclear cells (PBMCs) were isolated from whole blood and bcl-2, bax, heat shock protein (HSP)70, Cu-Zn superoxide dismutase (SOD), Mn-SOD, inducible nitric oxide synthase (i-NOS), SIRT1, SIRT3 and SIRT4 (Sirtuins) RNA levels were determined by Northern Blot analysis. Strenuous physical load significantly increased HSP70, HSP32, Mn-SOD, Cu-Zn SOD, iNOS, GADD45, bcl-2, forkhead box O (FOXO3A) and SIRT1 expression after the marathon, while decreasing bax, SIRT3 and SIRT4 expression (P < 0.0001).
Conclusion
These data suggest that the physiological load imposed in amateur runners during marathon attenuates the extent of apoptosis and may interfere with sirtuin expression.
Alas, the authors had a stumble. Per the retraction notice:
The authors have retracted the article [1]. Following publication of the article irregularities within the Northern blot figures were brought to the attention of the editors and subsequently confirmed by an investigation by the University of Rome Tor Vergata. We apologize to all affected parties.
We’ve asked the authors for more details, and will update with anything we learn.
Figure 1, actin controls, lanes 2 through 8 seem to be duplicates. I didn’t look any further.
This is very crude, very obvious. It is difficult to see how any of the coauthors could not have noticed it. And reviewers, if they had given even a cursory glance at the data should have been able to spot it.
I don’t have such clinical eyes, but this case really surprises me. For example, Dan claims that the actin control lanes appear to be replicates. However, if you expand the PDF file to about 200 or 500% in Adobe Reader, you will clearly see that lanes 3, 7 and 8 have, at the far right lower corner, slightly different signals that make them distinct from lanes 2, 4, 5 and 6, which may be duplicates. This highlights the risk about possibly incorrect or false claims made during post-publication peer review. To be honest, looking at the other blots, as an amateur, I couldn’t distinguish the problems, so it will be interesting to learn the exact details about these “figure irregularities”. If a decision to retract was based on such “irregularities”, then why can’t BMC simply publish the detailed list? The lack of such detail actually raises more concern and doubt (if not even suspicion) about BMC. The biggest concern I have, as I see this paper, which at first glance would to the untrained eye, appear as legitimate, is that there are hundreds of papers in quite high level plant and agricultural science journals that seem to carry blots of similar quality. In order for the plant science community to be able to actually differentiate what is true from fake banding and blots, we really need some specialists in our post-publication peer review (PPPR) efforts. Bloggers with detailed techniques of detection, or practical advice as to how to analyze such blots when one only has the PDF file in hand, please add links or references here at RW. In plant and agricultural science, the PPPR bell has been rung. Now we need to develop guidelines as to how scientists should make reports, and act on them. Also, we need short-lists of what aspects to look out for when looking for fraud.
” This highlights the risk about possibly incorrect or false claims made during post-publication peer review.”
You are absolutely right about this. I have seen a number of examples where people complain about a figure and I can’t see where the difficulty is. For myself, you will note that I said “seem to be duplicates”. I am always open to the possibility that other people have different opinions.
“you will clearly see that lanes 3, 7 and 8 have, at the far right lower corner, slightly different signals that make them distinct from lanes 2, 4, 5 and 6, which may be duplicates.”
For me, and I have looked at many thousands of gel and blot images, the obvious problem is the little blobs on the right side of each of the bands 2-7 in the actin control. This can happen, and is usually due in my experience to the presence of a bubble in the side of the loading well. A small fraction of the sample gets trapped in the bubble and separated from the rest of the lane. It forms it’s own mini-band in the space between lanes. Needless to say, the chance that this happens identically in a group of lanes is essentially nil.
You are quite correct that the blobs in lanes 3, 7, and 8 look different. My judgement is that for these lanes there was either an attempt at touching up the blobs or else they are different due to a cut-and-paste artifact. All the main bands in lanes 2-8 look identical on close-up and if you adjust the contrast and brightness of the original image (downloadable from the journal) you can see that the mottled background above the bands is identical too.
To prove I am wrong the authors could show the original image file from the device they used to collect the primary data.
Dan, your last sentence was actually the most important one, and not so much my challenge to what you had observed. Whenever there is reasonable doubt, or proof that can cast a shadow of doubt, then yes, I thoroughly agree with you that the original gel image files should be submitted. As I say, let’s see what additional information comes from BMC and/or from the authors. Perhaps this discussion may too premature. But, I would join your call to see the original images in high resolution so that anybody in the community can make a judgment call. I am also very curious to see if the University of Rome Tor Vergata will provide publically accessible data about its investigation. Maybe the 16 “healthy” runners that were sampled were not that helathy. Actually, how does one go about scientifically identifying a “healthy” runner? I am more interested to learn this information than the blobs on the gel lanes… (kidding!).