Disease Modification in Migraine: Study Design and Sample Size Implications

Abstract

Describe a clinical trial design that is capable of detecting disease modifying effects (DME) of anti-migraine drugs, and to explore the effect on study size of the likely variability in initial active- and placebo-response rates. An orthodox power calculation and typical response rates in studies of prophylactic drugs for migraine were used to construct the initial model. The interactions between variability in response rates for active and placebo treatments (ARR and PRR, respectively), as well as required sample size, was then analyzed as a 3-dimensional problem. Overall, the putative study design requires 4 times as many patients as in each analyzable treatment group, all with allowances for loss to follow up, when both positive and negative control groups are available for comparison with the group being tested for DME. For an example prophylactic agent with ARR = 50% and PRR = 25%, the model suggests that the study might need 260-270 patients. Sample sizes increase in quasi-exponential manner as placebo-response rates rise, and active-response rates fall, and these are plotted for the reported variability of ARR and PRR in migraine prophylaxis. Variations on the study design can be modeled for the likely variability in outcome using a straightforward technique. Investigating DME activity for drugs with the efficacy that is typical of approved prophylactic agents might not require impractically large numbers of patients. Such models can be used not only in study design, but also in assessing the overall feasibility of whether or not to pursue DME detection within a given development program.