Personalized medicine—giving the right treatment to the right patient at the right time—is pushing modern medicine ever further. It is driving the development of new treatments involving precision drugs for increasingly specific patient populations, for which demonstrating a therapeutic benefit poses a real challenge. Why are statistics so important in proving a drug's efficacy? How can they help us meet the challenge of personalized medicine?
A long road --------------
Before a new drug can be tested in humans, the molecule's activity must be investigated during a preclinical phase in the laboratory, using cells (in vitro studies) and animals (in vivo studies). Once these tests suggest that the molecules may have a beneficial effect in humans, they undergo three successive phases of clinical development (phases I, II and III). These three phases aim to quantify and compare the new drug's benefits and risks: in other words, to assess whether its efficacy outweighs the side effects caused by its toxicity and therefore suggests that it will improve the health of the patients who receive it.
The crucial step in demonstrating the efficacy of a new drug is the phase III clinical trial. This examines whether the new drug (the experimental treatment) is superior to another treatment (the control) regarded as the best therapy currently administered to patients with this profile—or to a placebo if no such therapy exists. Large numbers of patients are therefore recruited to take part in the study. How can they be allocated fairly between the treatment groups?
The statistician's first task is to define the method for assigning treatments (known as randomization): the procedure used to allocate patients randomly to each group. This ensures, for example, that the patients receiving the new drug have the same mean age and level of disease severity as those in the control group; the resulting groups are said to be comparable. Although software can handle this randomization step efficiently, it relies on principles of probability and patient stratification that must be thoroughly understood in order to identify and define the optimal method. By introducing chance and ensuring balance between the groups being compared, randomization provides the foundation for the statistical analysis used to assess and compare the new drug's efficacy.