Two Clinical Trial Designs to Examine Personalized Treatments for Psychiatric Disorders

Abstract

The National Institute of Mental Health Strategic Plan calls for the development of personalized treatment strategies for mental disorders. In an effort to achieve that goal, several investigators have conducted exploratory analyses of randomized controlled clinical trial (RCT) data to examine the association between baseline subject characteristics, the putative moderators, and the magnitude of treatment effect sizes. Exploratory analyses are used to generate hypotheses, not to confirm them. For that reason, independent replication is needed. Here, 2 general approaches to designing confirmatory RCTs are described that build on the results of exploratory analyses. These approaches address distinct questions. For example, a 2 × 2 factorial design provides an empirical test of the question, "Is there a greater treatment effect for those with the single-nucleotide polymorphism than for those without that polymorphism?" and the hypothesis test involves a moderator-by-treatment interaction. In contrast, a main effects strategy evaluates the intervention in subgroups and involves separate hypothesis-testing studies of treatment for subjects with the genotypes hypothesized to have enhanced and adverse response. These designs require widely disparate sample sizes to detect a given effect size. The former could need as many as 4-fold the number of subjects. As such, the choice of design impacts the research costs, clinical trial duration, and number of subjects exposed to risk of an experiment, as well as the generalizability of results. When resources are abundant, the 2 × 2 design is the preferable approach for identifying personalized interventions because it directly tests the differential treatment effect, but its demand on research funds is extraordinary.