The International Journal of Cancer, a Wiley title, has retracted a pair of articles from a group at UT Southwestern Medical Center in Dallas, for image manipulation.
The papers, from the lab of Adi Gazdar, the W. Ray Wallace Distinguished Chair in Molecular Oncology Research who is known for his massive collection of human cancer cells, were published in 2005.
The first was titled “Aberrant methylation of Reprimo in human malignancies.” According to the retraction notice:
The following article from International Journal of Cancer, “Aberrant Methylation of Reprimo in Human Malignancies” by Takao Takahashi, Makoto Suzuki, Hisayuki Shigematsu, Narayan Shivapurkar, Chinyere Echebiri, Masaharu Nomura, Victor Stastny, Meena Augustus, Chew-Wun Wu, Ignacio I. Wistuba, Stephen J. Meltzer, and Adi F. Gazdar, published online on 7 February 2005 in Wiley Online Library (wileyonlinelibrary.com), has been retracted by agreement between the authors, the journal Editor in Chief, Professor Peter Lichter and John Wiley & Sons Ltd. The retraction has been agreed due to presentation of an improperly manipulated figure (Figure 1A, page 506) in the article.
The work has been cited 41 times, according to Thomson Scientific’s Web of Knowledge. It also caught the attention of Science News, which wrote a story about the findings.
The second article, which appeared about six months later, was titled “Aberrant promoter methylation of multiple genes during multistep pathogenesis of colorectal cancers.” Again:
The following article from International Journal of Cancer “Aberrant promoter methylation of multiple genes during multistep pathogenesis of colorectal cancers” by Takao Takahashi, Hisayuki Shigematsu, Narayan Shivapurkar, Jyotsna Reddy, Yingye Zheng, Ziding Feng, Makoto Suzuki, Masaharu Nomura, Meena Augustus, Jing Yin, Stephen J. Meltzer and Adi F. Gazdar, published online in February 2006 in Wiley Online Library (wileyonlinelibrary.com), has been retracted by agreement between the authors, the journal Editor in Chief, Professor Peter Lichter and John Wiley & Sons Ltd. The retraction has been agreed due to manipulation of Figure 1A and Figure 2A (page 926).
The paper has been cited 23 times.
We’ve asked Gazdar for comment, and will update with anything we learn.
Update, 3:30 p.m. Eastern, 11/23/12: We spoke with Gazdar, who filled in some blanks for us. He said the papers came to the attention of UT officials when a figure turned up in the work of a Spanish investigator named José Román-Gómez, whom we covered back in April (see this comment from “amw” putting some pieces together). According to Gazdar, investigators identified problems with the plagiarized image — problems they dismissed as inconsequential but which which prompted him to conduct his own inquiry.
The internal review committee
had not reached a conclusion as to whether it was accidental or deliberate and were mulling over whether to launch a further investigation when I decided to launch my own investigation. Based on my findings they launched their own investigation that confirmed my findings. They also investigated an unstated number of randomly selected publications of mine to determine that similar occurrences were not present in manuscripts not coauthored by Takao Takahashi. Presumably they found no evidence as there was no further mention.
(Gazdar stressed that Takao Takahashi should not be confused with another occasional co-author of his, Tak Takahashi.)
Gazdar said he eventually reviewed nearly 20 papers, identifying five that had suspect images. Takao Takahashi — the first author on the two retracted articles listed above, who now works in Japan — acknowledged manipulating images in accepted “sole responsibility” for four of the articles, Gazdar said, but denied doing so refused to do so for a fifth. Another former post-doc “accepted responsibility” for the last paper, Gazdar said. but did not acknowledge Neither researcher has acknowledged wrongdoing.
In addition to the two retractions in the International Journal of Cancer, a third paper by Takahashi has been retracted, from Anticancer Research. The notice is behind a paywall — tsk, tsk. Here it is:
The article entitled: “Aberrant methylation to heparan sulfate glucosamine 3-O-sulfotransferase 2 genes as a biomarker in colorectal cancer” by Tokuyama Y, Takahashi T, Okumura N, Nonaka K, Kawaguchi Y, Yamaguchi K, Osada S, Gazdar A and Yoshida K, published in Anticancer Research, 2010, 30(12): pp. 4811-4818 is retracted.
This action, requested by the authors, is based on the findings and recommendations of an Internal Review Committee at the University of Texas Southwestern, Dallas, TX, USA, as the GAPDH loading control in Figure 1A of the article (p.4813) appears to be a modified version of a figure previously published elsewhere.
Gazdar said he has requested the retraction of two more articles, including this one from the British Journal of Cancer and one, which we’re not sure about, from Clinical Cancer Research.
Gazdar — who by all accounts is going above and beyond what many in his position might have — said he reported the image problems to the Office of Research Integrity “months ago” but has not heard anything about an investigation. The agency would neither confirm nor deny an inquiry when we asked about it today.
We’ve updated the headline of this post to reflect all five retractions.
Update 2:10 p.m Eastern, 11/26/12: This post has been updated to include corrections and additional information from Adi Gazdar.
Regular Retraction Watch followers may recall that it was in the process of discussing another RW article several months ago (http://www.retractionwatch.com/2012/04/05/jco-expresses-concern-over-western-blots-from-spanish-group-that-had-aroused-earlier-concern/) that apparent image fraud by the Gazdar lab was also noticed. That article initially concerned a researcher from University of Cordoba, José Roman-Gomez, who has been poaching entire or partial images from other papers, leading to several (at least 4 I believe) retractions in Blood and other journals, several of which have also featured on RW. One of these poached images came from a Gazdar lab paper, and it was while looking at that image that it became obvious that this image was itself duplicated within different figures of Gazdar lab publications. The trails and links were laid out at some length in that RW article (see link above) with several readers examining the image problems and agreeing that many Gazdar papers appeared to be affected by image fraud.
It would be fascinating to know how the Gazdar issues have now come to the attention of IJC. It’s 6-7 years since publication so it’s reasonable to suspect that the recent uncovering of the Gazdar problems in the RW article has acted as the trigger (? mechanism). It’s also interesting that these were straight retractions with no attempt to correct the articles with new gels (something Nature seems to be particularly adept at arranging).
Incidentally, sadly, and very predictably, this is probably just the tip of the iceberg. The Gazdar lab appears to have produced more image fraud than just these two IJC papers described so far. When we reviewed this before there were clear concerns about the following articles (see the earlier RW article link above for more detailed discussions and image links) and there may well be more…
Takahashi T, Shivapurkar N, Riquelme E, Shigematsu H, Reddy J, Suzuki M, Miyajima K, Zhou X, Bekele BN, Gazdar AF, Wistuba, II. Aberrant promoter hypermethylation of multiple genes in gallbladder carcinoma and chronic cholecystitis. Clin Cancer Res. 2004 Sep 15;10(18 Pt 1):6126-33. (internal duplication of gel images).
Takahashi T, Shivapurkar N, Reddy J, Shigematsu H, Miyajima K, Suzuki M, et al. DNA methylation profiles of lymphoid and hematopoietic malignancies. Clin Cancer Res. 2004 May 1;10(9):2928-35. (internal duplication of gel images)
Suzuki M, Shigematsu H, Shames DS, Sunaga N, Takahashi T, Shivapurkar N, Iizasa T, Frenkel EP, Minna JD, Fujisawa T, Gazdar AF. DNA methylation-associated inactivation of TGFbeta-related genes DRM/Gremlin, RUNX3, and HPP1 in human cancers. Br J Cancer. 2005 Oct 31;93(9):1029-37.
Suzuki M, Shigematsu H, Shivapurkar N, Reddy J, Miyajima K, Takahashi T, Gazdar AF, Frenkel EP.
Cancer Lett. 2006 Oct 28;242(2):222-30. Methylation of apoptosis related genes in the pathogenesis and prognosis of prostate cancer (duplication of bands between these two papers)
Suzuki M, Shigematsu H, Shames DS, Sunaga N, Takahashi T, Shivapurkar N, Iizasa T, Minna JD, Fujisawa T, Gazdar AF: Methylation and gene silencing of the Ras-related GTPase gene in lung and breast cancers. Ann Surg Oncol 2007, 14(4):1397-1404 (splicing artefact, Fig. 3b).
Thanks for pointing them out. A lead is always important.
amw suggests that the improper images in the above mentioned five articles were discovered independently by the scientific community. This is not so, as explained in detail in my own comment. The original observation was made during an investigation conducted at UT Southwestern Medical Center into another unrelated matter. The other four examples were found by a very detailed and laborious examination conducted by myself and my coworkers. A detailed report was provided to the Dean of Scientific Research who convened another Investigation that basically confirmed my findings. A copy of the Investigative Committee’s report was provided to all 40 or so coauthors of the five retracted articles and to other involved parties. Presumably amw obtained his/her information from this report, as it is extensively quoted.
Hopefully and predictably amw is wrong and that this is the entire iceberg, not just the tip. The Investigative Committee examined an unspecified number of randomly selected articles authored by myself and found no further examples of malfeasance.
Adi Gazdar
Just to clarify – I am unaware of this Committee’s Report; I do not work in the US or Japan, and have no connection to this field of study. My motivation for commenting externally on such cases as this arises from first-hand experience of how research fraud can be swept under the carpet, eroding science’s integrity and severely disrupting the lives of researchers, both locally (post-docs and students) and in the wider field, who subsequently work on areas contaminated by the fraud. I believe that highlighting the indisputable examples that image fraud provides is at least one mechanism by which the science-literate public can exert some influence on authors, institutions and journals whose tendency is, unfortunately, to act in ways that prevent such issues ever seeing the light of day.
Adi Gazdar’s acceptance of responsibility for the problems described in this article provides a startling exception to this pattern. For what it is worth, I would suggest that his comments below constitute the most frank and honest description of a problem with research misconduct that I have read from someone involved in such a case. As Robert L Park says, many scientists will come across situations in their careers where they discover that they have published flawed work – it is how they react that is key. In my experience, and from RW’s many articles in this area, denials, threats, obfuscations and use of legal resources are common reactions, and even though such an approach may be successful in avoiding a finding of misconduct, I don’t think such people can ever really contribute to science again in a meaningful way – once you sacrifice the purity of science for your own advancement, what would motivate you to set up another single experiment?
Those whose overriding concern is for finding out the truth, rather than their own glory, can admit that human mistakes have been made, and everyone can, in their own way, move on to the next challenge that science provides.
Good response, amw. But questions still remain. Prof Gazdar response is appreciated.
Re:
Suzuki M, Shigematsu H, Shames DS, Sunaga N, Takahashi T, Shivapurkar N, Iizasa T, Frenkel EP, Minna JD, Fujisawa T, Gazdar AF. DNA methylation-associated inactivation of TGFbeta-related genes DRM/Gremlin, RUNX3, and HPP1 in human cancers. Br J Cancer. 2005 Oct 31;93(9):1029-37.
Suzuki M, Shigematsu H, Shivapurkar N, Reddy J, Miyajima K, Takahashi T, Gazdar AF, Frenkel EP.
Cancer Lett. 2006 Oct 28;242(2):222-30. Methylation of apoptosis related genes in the pathogenesis and prognosis of prostate cancer (duplication of bands between these two papers)
Amw:
I will greatly appreciate it if you would inform us which figures from the two published manuscripts that you mentioned above contained bands that were duplicated. We tried searching for any similarities but were unsuccessful.
Thank you for your help.
Kuntal
Kuntal2012
Sincere thanks for taking these concerns seriously.
Firstly – context is important – please look at the PDF file for:
Suzuki M, Shigematsu H, Shames DS, Sunaga N, Takahashi T, Shivapurkar N, Iizasa T, Minna JD, Fujisawa T, Gazdar AF: Methylation and gene silencing of the Ras-related GTPase gene in lung and breast cancers. Ann Surg Oncol 2007, 14(4):1397-1404.
Figure 3b top panel contains clear evidence of lane splicing; a moderate zoom onto the PDF image reveals several abrupt changes in background between lanes. This indicates that one gel image was assembled from multiple independent gels. To my understanding this is a form of data falsification because the reader and reviewer believe the image to have come from the running of a single gel when in fact multiple gels were run.
It is therefore important to examine other images from this period for similar issues on the assumption that other gel images may have been produced by this approach. Please now look at the PDF file for:
Suzuki M, Shigematsu H, Shivapurkar N, Reddy J, Miyajima K, Takahashi T, Gazdar AF, Frenkel EP.
Cancer Lett. 2006 Oct 28;242(2):222-30. Methylation of apoptosis related genes in the pathogenesis and prognosis of prostate cancer.
In Figure 1 the background for these DNA gels is generally low. The DCR2 M panel, lane 5, looks different. Zoom in on the PDF again and a splice artefact is clearly visible on the right-hand side indicating that the individual lane image has been artificially pasted in from another source. There is also another splice artefact in the panel above (between lane 6 & 7).
Clearly it is extremely difficult for an outsider to know where an individual lane image was obtained from, but in the context of this situation where panels were being reused between figures and papers, and images spliced, it is reasonable to look at other papers from the group for candidates. The Br J Cancer 2005 paper has no apparent problems in isolation but Figure 4, RUNX3 M panel, POS lane looks very similar to the spliced-in band from CL 2006. The two bands aren’t identical at the pixel level but it is still quite possible that they are the same original band given that image compression (legitimate) occurs during figure preparation.
Thanks again
amw
I have no real understanding of image manipulation, so please help me out. Is it possible that, in an effort to clean up an image and make it more presentable, an investigator could inadvertently cross the line and clean up the data (not just the image)? Where does one draw the line?
Is it OK to make an image easier to understand? Or is ANY image manipulation necessarily bad? Are standards changing? Is it possible that people are now convicted of actions that were not criminal at the time they were done?
The sands are shifting, so the line is blurred. Scientists used to take a gel to the art department, for image preparation for a paper submission. Art departments are gone, scientists do their own image preparation with whatever tools are available in Powerpoint, Acrobat and related software. Some of these software tools include bells and whistles to “clean up” the image, and new guidelines have to get articulated so it becomes clear where the line is once again. When making an image “more presentable” distorts the quantitative nature of the image, and the data that one would get from the “more presentable” image poorly matches the data one would get from the “less presentable” original, a line has been crossed.
At a minimum, an unadulterated image of the original blot or gel that was quantified should be included in the supplemental materials of a paper. Journals need to step up and implement this requirement. A recent post on Retraction Watch (“The Nature paper that required three corrections” Nov 19, 2012) shows an example of this. The third corrigendum to this Nature paper shows complete gel images, along with rectangles showing which portions of the gels were cut for use in the paper. This really should become a standard requirement for all journals for papers presenting evidence quantified from gels and blots. This would allow others to see the whole picture, and judge whether such a gel could have provided reliable scientific evidence. It would also allow others to re-quantify a gel, possibly to study other questions of interest. This is one of the hallmarks of science, the ability for someone else to look at your data with fresh eyes and a different perspective, and possibly learn something new. Hiding gel data because it seems “less presentable” is when scientists become marketing and sales people instead of scientists.
What I wonder is why scientists think an image needs to be “cleaned up” to make it “more presentable”? The image acquired is the experimental observation. If the image as acquired needs to be “cleaned up”, what was wrong with the experiment that yielded data that needed to be “cleaned up”? If you botch an experimental run, and the data is obviously compromised, you throw that run out and do it again. “Cleaning up” an image in any fashion is tantamount to making up data. Throwing out data just because it does not look presentable amounts to cherry-picking data, which introduces obvious biases. What is happening in science that scientists are becoming marketing and sales people? This is the issue that needs to be studied and understood.
Very true, Steven McKinney. Some journals are practising this already as Nature Cell Biology. Here is my observation…
I was searching for publications of a person whose papers had issues raised by abnormal science earlier on pubmed . I bumped into two papers in Nature Cell Biology ( Ghosh A et al. 2012 Nat Cell Biology Nov. 18 doi: 10.1038/ncb2621 andTeo et al. 2010 Nat Cell Biology 12(8): 758 – 767). In fact, supposedly original blots are deposited as supplementary materials.
I compared the figures in the 2010 NCB paper with the full scans (as the authors call this) in supplementary figures (S8) and the original figures 2 – 4. I find that the full scans (S8) don’t appear to be ‘’strictly” original– you will again see the cut and paste of rows from other blots and it looks like there is lot of splicing (not alternate) going. Another thing to notice: figure 4 p Rap1 bands are made to show lowered intensity whereas for p65(s536), the intensity was enhanced in the FIgure 4 b of main article (NCB 2010 paper). I am not an expert in looking at these blots and I had emailed this to sciencefraud.org. Any one else has anything to offer to this observation?
I wonder what is going on here?
What about “Ghosh A et al. 2012 Nat Cell Biology Nov. 18 doi: 10.1038/ncb2621”. Did you find any error in this paper as well?
This is in response to ressci.
A critical problem of your allegations made here is that inferences are made without any basic background knowledge of the methods. The splicing that I see is only horizontal rather than splicing between experimental samples in the same blot which is not insidious as you believe. As far as I can see, I also didn’t find gel duplications if that is what you meant by “you will again see the cut and paste of rows from other blots”. May be you could point out the specifics?
To blot for multiple proteins without having to stripe and reprobe many times over, membranes can be cut horizontally into however many pieces as it is practical. This allows you to minimize the amount of antibody needed given that you are working with a smaller piece of membrane, and allows you to probe for many proteins of enough mw difference from one membrane. In practice, I don’t think it is useful to strip and reprobe more than once, and unless you use the licor system, you can’t mix antibodies when probing a membrane because you won’t be able to determine the specificity of each antibody.
Differences in “intensity” between proteins of interest is expected if they were exposed separately. Exposure time would depend on the antibody, concentration of the protein, quality of the samples, etc…
This is in response to Fernando Pessoa. Again, I think people should really understand the basics of methods before commenting.
I’m not sure what you meant by “people don’t really know what the bands really represent. You may very well get some band that comes and goes, but that’s all it is.”
In western blot when using a validated antibody, the most that you do to confirm specificity of the antibody is to see if the band corresponds to the correct MW size (which is the purpose of the ladder). How would using a smaller piece of membrane change that? Non specific bands due to an antibody would exist on both an uncut and cut membrane.
Also, I mentioned specifically that you can really only horizontally cut membranes to a practical approximation. You wouldn’t be able to cut thin slices for fear of losing your protein of interest.
You also mention: “another point is that the thin strips are themselves evasive as you do not see enough of the picture to know if the gel has run according to physics (i.e. they are easy to reuse, fake).”
I disagree as there is no evidence that people who use horizontal splicing are trying to commit fraud.
RW covered an article a few months ago in which it was stated that images like western blots are to be considered DATA and not DATA REPRESENTATIONS. It follows that any beautification attempts are not legit.
In response to blotman November 27, 2012 at 12:48 pm
These are general points.
It is true that you can cut the blot into horizontal strips to avoid reprobing. One point is that people don’t really know what the bands really represent. You may very well get some band that comes and goes, but that’s all it is. You cannot deliver a band to anybody. Another point is that the thin strips are themselves evasive as you do not see enough of the picture to know if the gel has run according to physics (i.e. they are easy to reuse, fake).
Very often you see papers where the only method is westen blotting. It does not show function.
Splicing between samples means that you cannot compare them.
Often you see one panel claiming to be a loading control and then 10 other panels stacked above. From simple inspection you can see that the loading control is probably the loading control for 2 or 3 the panels above it. The point is that the assay is not up to working out the relationships of several proteins, certainly not up to supporting those silly diagrams with arrows pointing to wherever you want (add a minus sign), until you get some explanation that your grrandmother might come up with about pregnancy.
In reply to blotman November 27, 2012 at 2:31 pm
“I disagree as there is no evidence that people who use horizontal splicing are trying to commit fraud.”
Example, SW Lee: http://www.science-fraud.org/?tag=sw-lee
blotman: thanks for your justification. I did mention that I am not an expert. May be I am wrong. Let us wait and see.
Fernando: thanks for bringing the other case up….
this is a breakthrough…AFG is on the editorial board of many cancer journals…It is better to check on others who are collaborating with him…there are quite a few of them in that building, I guess.
The investigation by University of Texas Southwestern, Dallas and the acceptance by Gazdar that papers from his lab needed to be retracted are indeed developments that should be applauded, and substantially better than the denials and obfuscations that we typically discuss on RW.
Nevertheless potential concerns remain. To get to the bottom of these issues it is necessary to actually examine the original laboratory records (as opposed to the papers only) and determine if the experiments for all papers during this period of time were actually done. It must be remembered that the image manipulations we can see simply provide visible signals that there was fraud; image duplications, splice artefacts and other irregularities occur because people are fabricating / falsifying their images, and this can leave behind telltale signs. But most of the time those falsifying or fabricating their results will avoid leaving such externally visible signs and the paper will look perfect to an external view. So only a detailed and fundamental investigation of the raw data entering these 20 or so papers from that period could actually produce a full list of papers that require retraction. Clearly such investigations take a lot of resources – and are difficult when several years have passed – but that would be the right way to do it. Although there was an internal review by the university, it sounds like Gazdar himself decided what to do based on examining the papers alone. Such an investigation could easily ignore images which might actually be produced by fraud but appear normal or nearly normal (actually three additional papers with definite problems are listed above which haven’t entered Gazdar’s list of retractions). This lab was churning out about a paper per month to the international cancer journals in 2004 – 2007, and it’s reasonable to suspect that many more papers from this period contain falsified data.
There is also an imbalance in terms of responsibility. If two researchers in a lab falsify results in a large number of papers, there has to be something more general wrong in the lab’s culture over a long period, and this is fundamentally the responsibility of the PI. Gazdar has published hundreds of papers, and he can’t control the detailed content of them all, but it still would have been nice to see some expression of personal contrition from Gazdar himself as the senior author, regarding his failure to prevent these papers from contaminating the scientific literature, rather than simply blaming his post-docs.
Finally, all this work is likely to have directly or indirectly supported Gazdar’s reputation and funding. Were any grants funded on the basis of this work i.e. were these papers used as references in any grant application? Likewise those falsifying the papers would have advanced their careers, ahead of other more honest researchers. What can be done to redress these injustices and wasted resources?
Nevertheless, RW’s editors are probably right to consider the glass half full here. Readers now know that this period of work from Gazdar’s lab contains many unreliable papers, and that is fundamentally the most important thing in global terms.
amw has hit the nail on the head. One of the most famous cases of research misconduct was John Darsee, a cardiologist from Harvard who worked in Eugene Braunwald’s lab. What got Darsee unstuck was that he was too productive, publishing a major paper every few months. I mention this because it should also have reflected on his boss, who is arguably the most prestigious cardiologist in the world, and has authored over 1000 primary papers. The point is that if someone publishes an average of 15 papers per year, as Gazdar does, this alone should be sufficient to raise questions about lab culture – are those who actually do the work adequately supervised, and do the PI’s claim honorary authorship on papers they have not read carefully, if at all, let alone carefully scrutinized the figures or seen the primary data?
To reduce the amount of research misconduct will require a restructuring if the incentives and power structures, in addition to penalizing not only those who fake the data but also those who take credit for the papers before they are found out but do not take responsibility when they are, as well as making it easier to report and retract papers that contain falsified data.
I think it is unlikely that the Brauwalds, Gazdars, Aggawals, Das’s etc. etc. fake data themselves, but they create a culture where misconduct flourishes, and take credit for the falsified papers until they are discovered.
I wonder if we need an additional category on journal articles so that people who amass large amounts of research real estate and employ dozens of post-docs and graduate students can get credit for being program directors without getting credit for doing research. The people currently known as principal investigators (although often they are the only people in the lab who do not investigate scientific questions) need credit for doing what they do — writing the grant proposals that bring in the money, hiring and firing, managing the budget, coordinating the final reports, making decisions about directions, dealing with the university bureaucracy, and so forth. That is administration, not science. But it is necessary if a large program is to stay coherent and contribute to a university’s reputation.
The people who handle these administrative tasks want to be rewarded with tenure and promotion and high salaries. Since universities have a reward system based partly on getting one’s name on publications, it is understandable that lab directors want their names on papers no matter how distant they are from the science. It is also understandable that such lab directors may be so distant from the science that they have no idea whether the people working for them are cheating. Maybe technology has advanced a great deal since they were graduate students. Maybe they have redirected their efforts toward new questions and new areas, so that they don’t understand enough of the details to spot fraud. Maybe they drop by the lab to say hello whenever they are in town, but their schedules of scientific meetings keep them away for long periods during which much occurs that you have to be there to appreciate.
It is unfair to give credit to and demand accountability from people who realistically cannot be involved in the scientific work. They need some other way to get recognition. How about an additional line below the real authors? Something like Project Director(s).
This is why many journals require that the contribution of each author to the article submitted is stated. However, this is probably not a requirement for the majority of journals. Maybe it’s time it should be. [Although one wonders how you address the issue of assigning a role to the ‘sleeping’ professor (aka as Demiigod) in whose lab you just happened to be carrying out work.]
I would like to address some of the points raised by the commentators and to further clarify certain matters.
1) It is stated that I have not accepted responsibility for the action of my coauthors. This is not true. I have informed the Dean of UT Southwestern Medical Center and the Internal Review Committee investigating these issues that a) I was totally unaware of the problem images and did not participate knowingly in any malfeasance; b) however, as lab chief, the buck stops there and I have to accept responsibility for the actions of my lab members; c) Only some of the retracted work was performed in my laboratory – in fact the entire work published in two of the five retracted articles was performed by co-authors after their return to Japan; d) some of the problem illustrations were obvious and should have been detected by me (and by my coauthors and the reviewers), while others were skillfully manipulated and were only detected after laborious and detailed comparison with every illustration published by the coworkers.
2) As part of my accepting responsibility, I examined all of the multiple articles that I coauthored with the authors of the disputed articles. While the first problem article was noted by a UT Southwestern committee investigating an unrelated fraudulent manuscript, my lab coworkers and myself spent several weeks investigating the articles and discovered four further cases of improper reproduction and/or manipulation of images. I provided the Dean of Basic Research with a detailed report and he convened another Investigative Committee that basically confirmed the findings in my report. The Committee also examined an unstated number of randomly selected articles authored by myself and found no further evidence of malfeasance. The Committee concluded that there was no evidence of my direct participation in the improperly manipulated/reproduced images, but pointed out my ultimate responsibility.
3) As pointed by a commentator, the original blots should have been examined, not just the resultant illustrations. I agree, and I have taken this and other measures to ensure that similar incidents do not arise again in my published work.
4) The Dean of UT Southwestern Medical Center has provided certain instructions to me, and I have carried them out promptly, including a) notifying all coauthors; b) promptly withdrawing the disputed manuscripts; c) informing the Office of Scientific Integrity at the NIH, and d) instituting measures to prevent such occurrences in the future.
5) Retraction Watch provides a unique and invaluable service. When contacted by Adam Marcus of Retraction Watch for comment and further detail, I promptly responded and assured him of my complete co-operation. He was kind enough to publish a more detailed version reflecting the information I provided.
I deeply regret that these incidents occurred on my watch. I also realize that the 40 or so coauthors of the five retracted manuscripts have suffered greatly. I wish to assure the readers of this blog and the scientific community that I remain dedicated to the importance of scientific integrity. Without integrity there is no science.
Adi Gazdar
In the original blog post “and one (retraction), which we’re not sure about, from Clinical Cancer Research.”
This may it.
Retraction: DNA Methylation Profiles of Lymphoid and Hematopoietic Malignancies
Clin Cancer Res Published OnlineFirst December 17, 2012; doi:10.1158/1078-0432.CCR-12-3679
http://clincancerres.aacrjournals.org/content/early/2012/12/17/1078-0432.CCR-12-3679.full?sid=583afc58-640a-45f8-8c5b-33feebde4616