Notes on a Scandal

As part of my ASA membership, I get a gratis copy of Signficance magazine, which was also the first I'd heard of the scandal surrounding Dr. Potti at Duke University.  Sadly, the Significance magazine article is not publicly available, but the writer's highlighted conclusion is that "Universities should support statistics as they support computing: specialist units should give advice on each."  The sad thing is that the Statistics department at Duke University has a 20+ year long history of working with the Medical Center; in fact, the Statistics department is housed in the Old Chemistry building right next to the West Campus entrance of the Medical Center (while I was there, the grad students used to go to the Med Center cafeteria for lunch all the time).  Don Berry, in particular, was a strong liaison with the Medical Center and did a ton of work on clinical trials at Duke in the 90's before moving to, ironically, to be the head of the Division of Quantitative Sciences at M.D. Anderson, where Baggerly and Coombes (no website?!), the professors who unearthed the problems in Potti's research, currently work.


But it's not like after Berry left there were no connections between the Duke Statistics department and the Medical Center.  Dalene Stangl (former chair of the department) has been doing clinical trials research with the Medical Center since before Berry left Duke, and Mike West (chair of the department during the 90's and the driver behind the department's rise from obscurity to prominence among Bayesian departments) and Ed Iversen have been doing work in genomics for ten years.  All this is to say that there is absolutely no reason why Potti should not have known about the statistical resources at his disposal.  So why didn't he use them?


Well, looking at the list of papers at the Duke Institute for Genome Sciences and Policy (IGSP), West appears as coauthor on three papers with Potti, though thankfully not the Nature Medicine article, which is not (no longer?) listed on the IGSP list of papers.  The problem is that many of the remaining papers in the list that include Potti as coauthor have titles like "Oncogenic pathway signatures in human cancers as a guide to targeted therapies" and "An integrated genomic-based approach to personalized treatment of patients with advanced stage ovarian cancer", very similar in title to the problematic "Genomic signatures to guide the use of chemotherapeutics" Nature Medicine article.  Excising the one tumor won't really give confidence in the health of the  other papers, especially any that are based on the same data, until a thorough independent biopsy is conducted on each of them.