Why retraction notices matter: Group’s repeated misuse of figures gets different play from five journals

For some journals, thorough retraction notices are the rule — and, when misconduct is involved, the price authors pay for abusing the trust of the editors and the readers. Others seem to take a more casual approach. Guess which we think is best.

Consider the case of a group of researchers in China led by Tan Jinquan, an immune system expert at Wuhan University. Over the past two years or so, Jinquan and colleagues have lost no fewer than a half-dozen papers containing evidence of image manipulation. But, depending on the journal pulling the articles, you might not know it.

We learned, but did not first hear, about the problems with Jinquan’s papers (note: he seems to be the senior author, and we have no evidence that he was anything more than a bystander in any fabrication) by reading a retraction notice from the June 2011 issue of the American Journal of Transplantation:

The following article from American Journal of Transplantation 2008; 8: 1401–1412, doi: 10.1111/j.1600-6143.2008.02275.x, Essential Role of Sphingosine-1-Phosphate Receptor 1-Bearing CD8+CD44+CCR7+ T Cells in Acute Skin Allograft Rejection, by H. Yuling, X. Ruijing, J. Xiang, X. Luokun, Y. Wenjun, C. Feng, H. Baojun, Y. Hui, Y. Guang, Y. Chunlei, Z. Jixin, C. Lang, Q. Li, A. Chang, B. Zhuan, J. Youxin, G. Feili, T. Jinquan, published online on June 28, 2008 in Wiley Online Library (http://www.onlinelibrary.wiley.com), has been retracted by agreement between the authors, the journal’s Editor-in-Chief, Allan D. Kirk, and Wiley Periodicals, Inc. The retraction has been agreed upon due to concerns relating to the data contained in Figures 1 and 7. The authors, upon presentation with the figures in question, were unable to satisfactorily regenerate the data or explain the areas in question.

Then we found this notice, from PLoS Pathogens, for a 2010 paper titled “EBV Promotes Human CD8+ NKT Cell Development”:

After an internal investigation by the Wuhan University Academic Ethics Committee, the authors of this paper requested that it be retracted. Several of the dataplots in Figure 8 contain regions that appeared to be identical, and concerns have been raised about other figures. The internal investigation could not account for the apparent duplications within Figure 8 and so the provenance and accuracy of these, and potentially other, figures cannot therefore be guaranteed. As a result, the overall integrity of this work is uncertain, and the paper was retracted from PLoS Pathogens on 03 Dec 2010.

So far, so good — at least, if good is the transparent communication of relevant information to readers and the community of scientists.

But we were disappointed to see several more retraction notices involving Jinquan’s papers that, for whatever reason, said nothing of value. Indeed, two of these were what we might call the index cases that got us interested in the matter, but which, as will become clear in a moment, required some dot-connecting that most readers wouldn’t have had the time or inclination to pursue.

Two of the retractions involved the journal Oncogene, a Nature Publishing Group title. The first of these, from 2011, refers to a 2005 paper (with a 2004 online pub date), “Selectively frequent expression of CXCR5 enhances resistance to apoptosis in CD8+CD34+ T cells from patients with T-cell-lineage acute lymphocytic leukemia.”

The blindingly illuminating notice:

This paper has been retracted.

The second Oncogene notice, for the 2007 article “Androgen activates PEG10 to promote carcinogenesis in hepatic cancer cells,” which also appeared in 2011, is equally revealing:

This paper has been retracted.

Not to be outdone, the Journal of Immunology has retracted two Jinquan papers with a rhetorical reserve that would make the Stoics nod in approval. Here’s one in the June 2012 issue, for a 2005 article titled “CD226 Expression Deficiency Causes High Sensitivity to Apoptosis in NK T Cells from Patients with Systemic Lupus Erythematosus”:

We wish to retract the article titled “CD226 Expression Deficiency Causes High Sensitivity to Apoptosis in NK T Cells from Patients with Systemic Lupus Erythematosus” by Deng Tao, Liu Shangwu, Wu Qun, Liu Yan, Ju Wei, Liu Junyan, Gong Feili, Jin Boquan, and Tan Jinquan, The Journal of Immunology, 2005, 174: 1281–1290.

Also in its June 2012 issue, the journal has retracted a 2005 paper by the group with the following notice:

We wish to retract the article titled “Vα24-Invariant NKT Cells from Patients with Allergic Asthma Express CCR9 at High Frequency and Induce Th2 Bias of CD3⁺ T Cells upon CD226 Engagement” by Yang Sen, Bi Yongyi, He Yuling, Xie Luokun, He Li, Xiong Jie, Deng Tao, Zhou Gang, Liu Junyan, Hu Chunsong, Xuejun Zhang, Jin Youxin, Gong Feili, Jin Boquan, and Tan Jinquan, The Journal of Immunology, 2005, 175: 4914–4926.

In addition to being functionally useless as retraction notices — who, for example, is “We?”, the authors? the editors?; are bad images the reason? — both of them were behind a paywall when we first saw them. After we contacted the journal, which is published by the American Association of Immunologists, the statements moved to a more accessible location.

We attempted to contact the JI‘s editor, Jeremy Boss, to find out why the notices were so uninformative. So far, no reply.

This isn’t the first time we’ve been able to take advantage of a sort of natural experiment to see how various journals dealt with retractions due to the same cause, and found some of the notices wanting. That was the case in the Silvia Bulfone-Paus series of retractions, too.

In fairness to the JI, it does not appear to be, ahem, allergic to informative retraction notices — at least in its past. A search of the archives turns up several examples of statements that say quite a bit about the problems with the affected papers.

For example, this 2010 notice is a reasonable exposition on the travails of a 2003 article:

We wish to retract the article titled “Neutrophil Serine Proteinases Activate Human Nonepithelial Cells to Produce Inflammatory Cytokines Through Protease-Activated Receptor 2,” by Akiko Uehara, Koji Muramoto, Haruhiko Takada, and Shunji Sugawara, The Journal of Immunology, 2003, 170: 5690–5696.

This retraction follows an investigation by Tohoku University into scientific misconduct. The investigation pointed out the following:

1. Figs. 1B and 2A: Total RNA was extracted from different cells, and cDNA was prepared and analyzed for the expression of SLPI and PARs and GAPDH by RT-PCR. However, the patterns of GAPDH in these figures and those in the figures of The Journal of Immunology, 2002, 169: 4594–4603 and in Fig. 2A of Clinical and Diagnostic Laboratory Immunology, 2003, 10: 286–292 are the same.

2. Fig. 2B: Two panels of the expression of PAR-2 with HLE and Cat G are the same.

The first author, who conducted these experiments, could not counter the argument by adducing raw data at the investigation, and the investigation recognized them as scientific misconduct. Therefore, we wish to retract the article.

We deeply regret these errors and apologize to the scientific community for the need to retract the article.

And the JI in 2011 retracted a 2006 paper whose authors had ordered the wrong batch of mice, and told readers about it.

We’re hopeful that the most recent bursts of indifference are simply an oversight and not a reversion to a less communicative adolescence.