Sample-size calculations for short-term proof-of-concept studies of tissue protection and repair in multiple sclerosis lesions via conventional clinical imaging

Background: New multiple sclerosis (MS) lesion activity on magnetic resonance imaging (MRI) can test immunomodulatory therapies in proof-of-concept trials. Comparably powerful endpoints to assess tissue protection or repair are lacking. Objective: The objective of this paper is to report sample-size calculations for assessment of new lesion recovery. Methods: In two sets of six active MS cases, new lesions were observed by monthly MRI for approximately 12 months. Averages and quartiles of normalized (proton density/T1/T2 weighted) and quantitative (T1/T2 and mean diffusivity maps for dataset 1, T2 and magnetization transfer ratio maps for dataset 2) measures were used to compare the lesion area before lesion appearance to afterward. A linear mixed-effects model incorporating lesion- and participant-specific random effects estimated average levels and variance components for sample-size calculations. Results: In both datasets, greatest statistical sensitivity was observed for the 25th percentile of normalized proton density-weighted signal. At 3T, using new lesions 3/415 mm³, as few as nine participants/arm may be required for a six-month placebo-controlled add-on trial postulating a therapeutic effect size of 20% and statistical power of 90%. Conclusion: Lesion recovery is a powerful outcome measure for proof-of-concept clinical trials of tissue protection and repair in MS. The trial design requires active cases and is therefore best implemented near disease onset.