Testing devices or experimental systems? Cancer clinical trials take the genomic turn.

Clinical trials are often described as machine-like systems for generating specific information concerning drug safety and efficacy, and are understood as a component of the industrial drug development processes. This paper argues that contemporary clinical trials in oncology are not reducible to mere drug testing. Drawing on ethnographic fieldwork and interviews with researchers in the field of oncology from 2010 to 2013, we introduce a conceptual contrast between trials as testing machines and trials as clinical experimental systems to draw attention to the ways trials are increasingly being used to ask open-ended scientific questions. When viewed as testing machines, clinical trials are seen as a means to produce answers to straightforward questions and deviations from the protocol are seen as bugs in the system; but practitioners can also treat trials as clinical experimental systems to investigate as yet undefined problems and where heterogeneity becomes a means to produce novel biological or clinical insights. The rise of "biomarker-driven" clinical trials in oncology, which link measurable biological characteristics such as genetic mutations to clinical features such as a patient's response to a particular drug, exemplifies a trend towards more experimental styles of clinical work. These transformations are congruent with changes in the institutional structure of clinical research in oncology, including a movement towards more flexible, networked research arrangements, and towards using individual patients as model systems for asking biological questions.