Inverse dependence between patient population and correlation of composite MRI scores with EDSS in Multiple Sclerosis

INTRODUCTION. In multiple sclerosis (MS), patient disability is assessed according to the qualitative, ordinal Expanded Disability Status Scale (EDSS). In order to monitor treatment efficacy and disease progression, quantitative MRI metrics are sought that will correlate with EDSS [1]. In the relapsing-remitting MS (RRMS) literature, r-values for correlation with EDSS of various metrics have ranged from 0.31 – 0.61 [2-9], with a strong inverse trend with population sample size (Figure 1) implying that studies with fewer patients have biased correlation results. It is well-known that using small samples overestimates the true correlation, but a rigorous analysis would provide useful insight into whether some metrics are more resistant to the degrading effect of population size than others, what minimum population size is needed for robust results, and whether any fundamental limitations affect correlation with EDSS overall.