MRI lesions as a surrogate for relapses in multiple sclerosis: a meta-analysis of randomised trials

Background A meta-analysis of randomised trials in relapsing-remitting multiple sclerosis published in 2009 showed a quantitative relation between the treatment effects detected on Mill lesions and clinical relapses. We aimed to validate that relation using data from a large and independent set of clinical trials in multiple sclerosis. Methods We searched Medline for clinical trials that assessed disease-modifying drugs for relapsing-remitting multiple sclerosis published from Sept 1,2008, to Oct 31,2012. We extracted data for the treatment effects on MRI lesions and on relapses from each trial, and the correlation of log transformed relative measures of these treatment effects was assessed with a weighted linear regression analysis. The R-2 value was estimated to quantify the strength of the correlation, and we used an interaction test to test for a difference in slope from the previously estimated equation. We also ran several sensitivity analyses. Findings We identified 31 eligible trials, which provided data for 18 901 patients with relapsing-remitting multiple sclerosis. The regression equation derived using data from these studies showed a relation between the concurrent treatment effects on MRI lesions and relapses (slope=0.52; R-2=0.71), much the same as was previously estimated (P-interaction=0.45). Analysis of trials that tested the same drugs in phase 2 and phase 3 studies showed that the effects on MRI lesions over short follow-up periods (6-9 months) can also predict the effects on relapses over longer follow-up periods (12-24 months), with reported effects on relapses that were within the 95% prediction intervals in eight of nine trials. Interpretation Our findings indicate that the effect of a treatment on relapses can be accurately predicted by the effect of that therapy on MRI lesions, implying that the use of Mill markers as primary endpoints in future clinical trials of treatments for multiple sclerosis can be considered, in specific situations, such as in trials testing generics or biosimilars of drugs with a well known mechanism of action or in paediatric trials testing drugs already approved for adults.