Abstract
Evidence-Based Medicine (EBM) and Personalized Medicine (PM) share a common goal: reducing the gap between the results of biomedical research and their clinical application. PM is, however, often presented as a “new paradigm” for medicine, just as EBM was in the 1990s. It covers a wide variety of projects but the core idea that generally unites them is the ambition of better taking account of individual specificities than did EBM with its statistical and population-centred approach. In this article, I concentrate on PM in cancerology, the essence of which is to target treatments based on the molecular profile of the patient. This targeting is made possible by gaining better knowledge about the molecular mechanisms of cancers. The classification of patients as a function of their molecular profile entails the creation of patient sub-groups. This creates a problem for the traditional evaluation of therapeutic treatment promoted by EBM, in particular the use of randomized trials using sizeable cohorts. But a better understanding of the mechanisms and the greater precision of treatments could reduce the need for these trials. Does PM thus represent the revenge of a physio-pathological and mechanistic culture in clinical research against the statistical and empirical one of EBM? My objective is to show how current practices of PM leads to epistemological changes in our estimation of what count as relevant types of information and proof in medicine, in particular in the field of therapeutic evaluation. I defend the idea that PM, far from obviating the need for statistical approaches and the search for correlations, ultimately poses new challenges for EBM. PM drives EBM to strengthen the articulation and the integration of statistical and mechanistic data with a view to providing a better service for each patient.