Bruno Falissard, biostatistician: interview | Tangente

Tangente: The emergence of so-called personalized therapies represents an opportunity for patients, but it poses a real challenge for assessing the effectiveness of these drugs. How can statistics help meet the challenge of personalized medicine?**

Bruno Falissard: Over the past five or ten years, therapies have become more targeted, so the patient populations likely to receive a new drug have grown smaller and smaller. For statisticians, this is an almost insoluble problem: workable solutions are rare. There are ways around the difficulty. For example, with a fairly rare disease, the new drug can be tested on as many available patients as possible—and there will not be many. Its effect in that sample can then be compared with previously collected data from databases or registries. These are known as historical controls. This increases statistical power and avoids having to recruit and randomly assign hundreds of subjects when only a few dozen exist. It is done, but the drawback is that comparisons with historical controls are far less reliable. In statistics, we say that the level of evidence is lower, which creates difficulties when deciding whether to grant marketing authorization or reimbursement.

Biological plausibility -----------------------

To offset this decline in the level of evidence, scientists are now paying increasing attention to the mechanism of action and the biological plausibility of therapeutic effectiveness. With a homeopathic medicine, for example, in which the substance is highly diluted, the mechanism of action might be regarded as highly speculative. With a monoclonal antibody targeted precisely at the biological defect underlying the disease, the drug might be expected to be effective a priori. This is a genuine revolution because, over the past thirty or forty years, the mechanism of action carried little weight, for epistemological reasons: biological explanations were considered "not reliable enough" to be used in a public-health context. That is now changing and is the subject of debate. Are we entitled—and is it reasonable—to market a drug primarily on the basis of its mechanism of action?