The Proceedings of the National Academy of Sciences (PNAS) has published the seventh retraction for former Duke researcher Anil Potti, who now faces a lawsuit in the midst of an ongoing investigation into his work:
Retraction for “A genomic approach to colon cancer risk stratification yields biologic insights into therapeutic opportunities,” by Katherine S. Garman, Chaitanya R. Acharya, Elena Edelman, Marian Grade, Jochen Gaedcke, Shivani Sud, William Barry, Anna Mae Diehl, Dawn Provenzale, Geoffrey S. Ginsburg, B. Michael Ghadimi, Thomas Ried, Joseph R. Nevins, Sayan Mukherjee, David Hsu, and Anil Potti, which appeared in issue 49, December 9, 2008, of Proc Natl Acad Sci USA (105:19432–19437; first published December 2, 2008; 10.1073/pnas.0806674105).
The authors wish to note the following: “We wish to retract this article because we have been unable to reproduce certain key experiments described in the paper regarding validation and use of the colon cancer prognostic signature. This includes the validation performed with dataset E-MEXP-1224, as reported in Fig. 2A, as well as the generation of prognostic scores for colon cancer cell lines, as reported in Fig. 4. Because these results are fundamental to the conclusions of the paper, the authors formally retract the paper. We deeply regret the impact of this action on the work of other investigators.”
The 2008 paper, which has been cited 27 times, according to Thomson Scientific’s Web of Knowledge, was already the subject of a minor 2009 correction:
Correction for “A genomic approach to colon cancer risk stratification yields biologic insights into therapeutic opportunities,” by Katherine S. Garman, Chaitanya R. Acharya, Elena Edelman, Marian Grade, Jochen Gaedcke, Shivani Sud, William Barry, Anna Mae Diehl, Dawn Provenzale, Geoffrey S. Ginsburg, B. Michael Ghadimi, Thomas Ried, Joseph R. Nevins, Sayan Mukherjee, David Hsu, and Anil Potti, which appeared in issue 49, December 9, 2008, of Proc Natl Acad Sci USA (105:19432–19437; first published December 2, 2008; 10.1073/pnas.0806674105).
The authors note that the term “prognostic score” should be substituted for the term “Recurrence Score,” which is a registered trademark of Genomic Health and is not associated in any way with the authors or the article. The online version of the article has been corrected accordingly as of April 7, 2009.
If we may, we’d like to say we also deeply regret the potential impact of “this action” — eg the retraction — on the work of one of the co-authors, Shivani Sud. Sud’s name was familiar to us, because Ivan edited a 2006 profile of her for The Scientist, when she was 16:
Sud, who lives in Durham, NC, has been doing cutting-edge cancer research since she was 14 years old. Getting started was difficult, she says; as an eighth grader she had trouble finding a scientific advisor. “No one trusts that you’re willing to work that hard,” she explains. “A lot of it comes down to maturity. They are not sure what it’ll be like having someone that young in the lab.”
That means she was no older than 18 when she had a paper published in PNAS. Impressive, to say the least. Then again, she also won the Intel Science Talent Search award the year the paper came out. But now she has a retraction on her CV, because a terrific and energetic kid ended up working for a lab that can’t seem to reproduce much of their data. We sincerely hope that Sud’s bright prospects aren’t dimmed in the slightest by “this action.” [See update at end.]
It turns out we also missed a correction of a 2009 paper by Potti and colleagues that ran on August 30, 2011:
Correction for “Characterizing the developmental pathways TTF-1, NKX2–8, and PAX9 in lung cancer,” by David S. Hsu, Chaitanya R. Acharya, Bala S. Balakumaran, Richard F. Riedel, Mickey K. Kim, Marvaretta Stevenson, Sascha Tuchman, Sayan Mukherjee, William Barry, Holly K. Dressman, Joseph R. Nevins, Scott Powers, David Mu, and Anil Potti, which appeared in issue 13, March 31, 2009, of Proc Natl Acad Sci USA (106:5312–5317; first published March 11, 2009; 10.1073/pnas.0900827106).
The authors note, “We wish to bring to your attention an issue regarding our PNAS publication referenced above. The focus of the paper was the characterization of the signaling pathways of TTF-1, NKX2–8, and PAX9 in lung cancer using a transcriptome-based approach. As such, gene expression signatures representative of the activation of individual transcription factors (TFs), TTF-1, NKX2–8, or PAX9 were developed and were found to have prognostic implication in a data set of clinically annotated lung cancer specimens (Figs. 1–3).
“A second validation was then performed in an independent data set of lung cancer specimens. However, the source publication of the gene expression dataset (1) was retracted by the paper’s authors in March 2011, calling into question the validity of the results using this dataset as described in Fig. 4, and with a subset of the data in Fig. S7. In addition, we have not been able to reproduce results reported in Fig. S6, which contains a subset of the data shown in Fig. 4. Given this finding, we believe that the affected figure in the paper (Fig. 4), as well as the data in Figs. S6 and S7 that report an analysis of a subset of the data in Fig. 4, are no longer valid.
“Finally, we also note that a specific portion of analysis (i.e., cisplatin sensitivity prediction) presented in Fig. 5 can no longer be supported, because the publication describing the cisplatin signature has been retracted (2). However, this issue does not alter the validity of the Fig. 5 data describing the unsupervised hierarchical clustering of 56 lung cancer cell lines using the three TF signatures (Fig. 5A Upper) and the experimental determination of cisplatin sensitivity (Fig. 5B Right). More importantly, the final conclusion of Fig. 5, which is that activation of the TF is correlated with sensitivity to cisplatin (Fig. 5B Right), remains unaltered. Importantly, we do not believe that this correction alters the primary conclusions of the paper.”
That’s quite a correction to not “[alter] the primary conclusions of the paper,” which has been cited 10 times. Reminds us of this.
Update, 6 p.m. Eastern, 10/4/11: Intel Science Talent Search tells us “this does not have any effect on Shivani Sud’s Intel Science Talent Search award and recognition.” To which we say: Good.
Hat tip: “phosphatase kinase”