April 2012

Sidebar: Rough patches

The first cautionary tale in being too hasty with -omics technologies harks back 10 years. Scientists from the U.S. Food and Drug Administration, National Cancer Institute and a bioinformatics company Correlogic Systems published a paper that described proteomic patterns in patients’ serum that seemingly indicated ovarian cancer even at early stages (1). Correlogic Systems licensed the technology to Quest Diagnostics and the Laboratory Corporation of America to develop a diagnostic test called OvaCheck.

But other scientists set off alarm bells, questioning the analytical validity of the study (2, 3). When Baggerly’s team analyzed the data from one of the sets in the paper, Baggerly says by “using electronic noise, we could separate cancers from controls. We should never be able to do that. The fact that we could was evidence they screwed up the [experimental] design.” The FDA, on hearing the reports, eventually stepped in and insisted on further validation before OvaCheck was commercialized.

The other cautionary tale involves outright fraud. In 2007 and 2008, Duke University launched three clinical trials based on research led by Joseph Nevins and Anil Potti that used microarrays to develop personalized treatments for breast and lung cancer patients based on genomic signatures (4). Baggerly once again was involved, along with his collaborator Kevin Coombes, in pointing out various mistakes in the data interpretation (5; to watch a lecture by Baggerly on this topic, go to http://videolectures.net/keith_baggerly/). But it soon appeared that Potti had lied about his qualifications on his curriculum vita, and the data from his experiments were riddled with errors. In 2010, Duke University halted the clinical trials. So far, nine of the Nevins and Potti publications, including reference 4, have been retracted (6).CBS’s “60 Minutes” aired a segment on the Duke case on Feb. 12 (http://www.cbsnews.com/8301-18560_162-57376073/deception-at-duke/).

  1. 1. Petricoin, E.F. et al. Lancet 359, 572 – 577 (2002).
  2. 2. Sorace, J.M. & Zhan, M. BMC Bioinformatics 4, 24 (2003).
  3. 3. Baggerly, K.A. et al. Bioinformatics 20, 777 – 785 (2004).
  4. 4. Potti, A. et al. Nature Medicine 12, 1294 – 1300 (2006).
  5. 5. Baggerly, K.A. & Coombes, K. R. Ann Appl Stat 3, 1309 – 1334 (2009).
  6. 6. Ninth Potti paper to date gets retracted. Duke Chronicle. http://dukechronicle.com/article/ninth-potti-paper-date-gets-retracted.

Raj_MukhopadhyayRajendrani Mukhopadhyay (rmukhopadhyay@asbmb.org) is the senior science writer for ASBMB Today and the technical editor for the Journal of Biological Chemistry.

found= true1764