When Does Reproducibility Become an Ethical Issue?

• Activity Number: 102
• Type: Invited
• Date/Time: Monday, July 30, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Committee on Professional Ethics
• Author(s): Keith Alan Baggerly*+
• Companies: MD Anderson Cancer Center
• Keywords: reproducibility ; genomics ; clinical trials

Abstract

On Nov 9, 2010, Duke University terminated three cancer clinical trials using genomic "signatures" to direct patient therapy. Duke officials acknowledged that "in retrospect, the trials should not have been run", because the analytical reproducibility of the signatures had never been adequately verified.

While the core underlying problems were basic, identifying them took a good deal of time. Such time is not always available. Critiquing published results is also not easy, and is unavoidably adversarial. When we suspect there are problems in the literature, how do we decide which ones to fight about, and how hard? When complex datasets are involved, making interpretation difficult, what level of evidence needs to be provided before these interpretations are used to guide therapy or policy? These latter issues are currently being discussed by the National Cancer Institute (NCI) and the Institute of Medicine (IOM) with regards to clinical trials; we will briefly touch on the major issues raised.