The work of Sam W. Lee, a cancer biologist at Harvard and Massachusetts General Hospital, has come under fire at Science Fraud lately over concerns about the possible reuse of images in his group’s published studies.
Turns out there’s some there, there after all. The journal Current Biology has issued a pretty thorny correction for one of Lee’s 2006 articles, “RhoE Is a Pro-Survival p53 Target Gene that Inhibits ROCK I-Mediated Apoptosis in Response to Genotoxic Stress,” citing multiple issues with its figures:
It has been brought to the authors’ attention that incorrect images for Figure 2A and the left panel of Figure 4D were selected during the figure preparations for this article. We thus provide corrected versions of Figures 2 and 4here.
In the originally published Figure 2A “Expression of RhoE protein and mRNA in RhoE knockdown cells,” the northern blot panel for the loading control (36B4) does not match the experimental conditions described in the figure. It has been pointed out to us that this panel is very similar to Figure 4A in another of our publications (J. Biol. Chem. 279, 38597–38602; September 10, 2004). It is possible that this is an inadvertent duplication, but at this point we regrettably cannot confirm or disprove this. Therefore, the experiments have been repeated, and the new Figure 2A is shown here.
In addition, in the originally published Figure 4D “Effect of ROCK I inhibitor (Y-27632) on CPT-mediated apoptosis,” the panel for DMSO control (0.1% apoptosis) was misplaced in the next panel for Y-27632-treated control (0.3% apoptosis). In order to ensure integrity of the data, the experiments have been repeated, and the results reproducing the original data are shown in the new Figure 4D here.
Neither of these errors affects the results or conclusions of the article. The authors apologize for these errors and any confusion that may have resulted.
We were particularly drawn to this part of the notice:
It is possible that this is an inadvertent duplication, but at this point we regrettably cannot confirm or disprove this. Therefore, the experiments have been repeated, and the new Figure 2A is shown here.
It’s indeed regrettable that the lab cannot demonstrate the provenance of its data. On those grounds alone, we’d expect a retraction, not a correction. We have reached out to the editor for more information on this case and will update this post as we learn more.
The paper has been cited 46 times, according to Google Scholar.
Lee was not immediately available for comment. His secretary initially suggested we speak with a Dr. Bringhurst — whom we believe might be F. Richard Bringhurst, MGH’s senior VP for medicine and research management — about the paper, but we pointed out that his name doesn’t appear on list of authors.
The original Science Fraud post, dated August 13, does not mention the Current Biology paper, focusing rather on a 2009 article in Molecular Cell, “GAMT, a p53-inducible modulator of apoptosis, is critical for the adaptive response to nutrient stress,” and another in the Journal of Biological Chemistry, “DDR1 receptor tyrosine kinase promotes prosurvival pathway through Notch1 activation.”
But the Current Biology paper did crop up in a subsequent comment, and was among papers about which Clare Francis raised concerns in an August 15 to the journal and others — suggesting that, if that was roughly when the journal learned of the problem, it acted pretty swiftly.
We’ll keep an eye out for more corrections, and possibly retractions, in the near future from this lab.
Update, 12:15 p.m. Eastern, 11/20/12: Current Biology editor Geoffrey North tells us:
We first heard about the potential problems with the Lee paper in an email from a concerned anonymous reader.
With regard to the issue of duplication (or not), I am agnostic. We consulted with the relevant institutional RIO about the matter and there was an institution investigation which concluded that it was impossible to resolve conclusively if there was a duplication. As described in the Erratum in the latest issue of Current Biology, the authors have replicated the experiments and reproduced the published data.
Update, 3:45 p.m. Eastern, 11/20/12: We spoke with Lee this afternoon, and he told us, in a word, that the problem stemmed from the sloppiness of the first author —Pat Ongusaha, then a post-doc — rather than misconduct. Said Lee of the missing figure:
For some reason we don’t have that loading control plot. The post-doc lost it some time back. We looked through the notebooks and couldn’t find it.
As for Ongusaha, who is now at the University of Chicago, Lee said:
I think she’s a very thorough person. I don’t know how that happened.
We tried Ongusaha, but were told she was traveling this week.
Lee said his group has passed on all of its relevant lab materials to research integrity officials at Harvard and Mass. General, but he wouldn’t confirm the pending retractions or corrections of other papers.
Update, 4 p.m. Eastern, 11/20/12: It turns out this wasn’t Lee’s first mega-correction. Here’s one from Nature in January, as commenter michaelbriggs notes, for “Selective killing of cancer cells by a small molecule targeting the stress response to ROS.” Two things are worth pointing out. One, Ongusaha is not an author. Two, the correction includes this sentence, which seems a lot like the “we regrettably cannot confirm or disprove” language in the Current Biology notice:
We have also been unable to verify without doubt that the image in Supplementary Fig. 9b shows four different mice within the treated and untreated groups and therefore wish to replace this figure (see Fig. 2 of this Corrigendum).
If not already done so, I hope all of the publications from the authors of this paper are scrutinized in depth. I believe that re-production of figures is certainly grounds for retraction and am completely surprised that Current Biology has not moved in this direction.
“reaching out” to someone would probably constitute assault and certainly cunjures up deeply unpleasant images. I hope you mean you emailed/wrote to/contacted or some more appropriate verb?
Other than adopting this particularly horrible and recently popular phrase, thank you for your very interesting and useful blog!
Nature published one of its “megacorrections” on another paper from Sam W Lee. Original paper: Nature 475, 231–234 doi:10.1038/nature10167; megacorrection: Nature 481, 534 doi:10.1038/nature10789
Two items in the megacorrection were notable. They make the remarkable statement “We have also been unable to verify without doubt that the image in Supplementary Fig. 9b shows four different mice within the treated and untreated groups and therefore wish to replace this figure” and the pictures of the mice in the corrected figure have tumors of a size that are far, far, larger than any permitted by any animal welfare committee. While Dr Lee has many questions to answer, Nature should also explain why it publishes megacorrections rather than retracting papers, and why it publishes images that give tacit acceptance of animal cruelty, and thus potentially tarnishes the reputation of all researchers who use animals.
Ah, the old blame-it-on-the-postdoc scam.
If you take a look at the new new Figure 2A, part C, the DMSO scatter plots (upper paif of scatter plots) you will see that some of the outliers are the same. For example in the upper part of left sector, just above the main cloud. There are also other outliers that line up, such as those just to the right of the internal vertical division above the main cloud. I think the cells in the scatter plots are from the same population, i.e. did not have two treatments, perhpas from an overlapping time window.
Ah, the old Luk Van Parijs technique.
To have to correct a figure because of duplicated blots may be regarded as a misfortune; to include duplicated flow cytometry data in the correction looks like carelessness.
There was also an outright fraud (image duplication and manipulation) of a p21 panel in Fig. 1a of Mol Cell Biol 2000;20;7450-9; PMID11003642 and Fig. 1a of EMBO J 2003;22;1289-301; PMID12628922. The same p21 Western was spliced and pasted, and used to represent two completely different cell lines. These two papers have different first authors but the same last author (SW Lee). Go figure.
The last author (SW Lee) didn’t know what was going on is one possibility.
With so many questionable papers spanning so many years and the number of postdocs as first authors without SW (sloppy work) Lee knowing what’s going on. Yeah right.
I was trying to be sarcastic.
Quite.
Just seeing this belatedly. When you read the Science Fraud account of Lee’s work, you realise that internal investigations can choose not to see fraud even when it is obviously there, simply by failure to do the right thing and look at all the relevant papers. Clearly the system in the US still allows impunity for certain researchers.
The journal doesn’t help by saying that they contacted the ‘institutional ORI’ – let’s be clear the true Office of Research Integrity of the US does not seem to have been contacted.
Perhaps if you had said that Sam Lee had also put ice down your t-shirt you would have got a more rigorous response?
amw wrote “you realise that internal investigations can choose not to see fraud even when it is obviously there, simply by failure to do the right thing and look at all the relevant papers”
Sagan’s “Absence of evidence is not evidence of absence!” may well apply in many similar insitutional investigations. But they do seem to be acting in some cases, so it’s a start.
I do hope Lockes “argumentum ad ignorantiam” does not take centre stage in pending cases, as this will simply deter the inevitable and the quicker these issues are dealt with, the better.
New Andy Warhol (SW Lee really) exhibition.
http://retractionwatch.files.wordpress.com/2013/02/cancer-res-2001-yellow-arrow-1.ppt
In reply to fernando pessoa February 8, 2013 at 9:45 pm
Information on the new Andy Warhol exhibtion available at the pubpeer database.
http://pubpeer.com/publications/578F85A1DDAAA96599C26C88EA7BE1
Clare
I am sorry to say this as I am a great admirer of your work here and elsewhere but we really really need one thing in pointing out “errors” and pubpeer doesn’t cut the mustard on that count.
We need Diagrams with Big Red Arrows pointing out the “errors” clearly for all to see.
If pubpeer provided the possibility for that I for one may be interested but as it stands it takes onlookers too much work to see what is often quite challenging to the untrained eye.
For example:
http://web.archive.org/web/20120909024225/http://www.science-fraud.org/?tag=sw-lee
Stewart May 17, 2013 at 3:52 pm
Dear Stewart,
Thanks you for your comment. That was the point of my doing it.
People like Jörg Zwirner at abnormalscience blog and
the people behind science-fraud blog did provide the illustrated versions.
They did a lot of work because the journal editors would not see points.
Pubpeer is trying its best. It has only started and may only have few resources.
Perhaps the publishers (businessmen) should be taxed to provide the regulatory service?
Clare
That was my point. Thanks for reiterating it.
Ghosts in the machine:
http://web.archive.org/web/20120909024225/http://www.science-fraud.org/?tag=sw-lee
2016 correction figure 2C.
http://www.cell.com/current-biology/fulltext/S0960-9822(16)30866-1