Editors' Note: Blood Neurofilament Light in Progressive Multiple Sclerosis: Post Hoc Analysis of 2 Randomized Controlled Trials.

Abstract

In a post hoc blinded analysis of 2 placebo-controlled, phase 3 trials, Dr. Leppert and colleagues investigated the association of plasma neurofilament light (pNfL) levels with disability progression, cognitive decline, and brain atrophy in 1,452 patients with secondary progressive multiple sclerosis (SPMS; EXPAND trial of siponimod) and 378 patients with primary progressive multiple sclerosis (PPMS; INFORMS trial of fingolimod). They found that higher pNfL levels were independently associated with higher risks of confirmed 3- and 6-month disability progression, earlier wheelchair dependence, cognitive decline (in SPMS), and higher rates of 24-month brain atrophy. pNfL levels were lower in patients treated with siponimod or fingolimod vs placebo, and the authors concluded that pNfL may be a meaningful outcome measure in PMS studies. In response, Dr. Barro raises concerns about whether the pNfL quantification in the 2 trials were performed with the same batches of reagents, given the potential for batch-to-batch variability, and questions the use of a fixed threshold for high vs low pNfL levels in the analyses instead of age-adjusted thresholds, given the association of older age with both higher pNfL levels and disability progression. Responding to these comments, the authors agree that batch-to-batch variability may be an important assay confounder, but note that the coefficient of variation was less than 15% for each of the 3 native serum control samples with high, medium, and low pNfL used in the study. They argue that whereas the assay used in their study did not allow direct application of the age-adjusted cutoffs, age was included as a covariable in all statistical models, so the higher risk of progression in patients with a higher baseline pNfL in their study cannot be attributed to age differences. They also agree that z scores or percentiles would be preferable to fixed cutoffs. This exchange indicates the difficulties of standardization of assays for pNfL measurement. Although pNfL measurements now permeate the literature on numerous neurologic conditions, harmonization of study methods continues to be a challenge. In a post hoc blinded analysis of 2 placebo-controlled, phase 3 trials, Dr. Leppert and colleagues investigated the association of plasma neurofilament light (pNfL) levels with disability progression, cognitive decline, and brain atrophy in 1,452 patients with secondary progressive multiple sclerosis (SPMS; EXPAND trial of siponimod) and 378 patients with primary progressive multiple sclerosis (PPMS; INFORMS trial of fingolimod). They found that higher pNfL levels were independently associated with higher risks of confirmed 3- and 6-month disability progression, earlier wheelchair dependence, cognitive decline (in SPMS), and higher rates of 24-month brain atrophy. pNfL levels were lower in patients treated with siponimod or fingolimod vs placebo, and the authors concluded that pNfL may be a meaningful outcome measure in PMS studies. In response, Dr. Barro raises concerns about whether the pNfL quantification in the 2 trials were performed with the same batches of reagents, given the potential for batch-to-batch variability, and questions the use of a fixed threshold for high vs low pNfL levels in the analyses instead of age-adjusted thresholds, given the association of older age with both higher pNfL levels and disability progression. Responding to these comments, the authors agree that batch-to-batch variability may be an important assay confounder, but note that the coefficient of variation was less than 15% for each of the 3 native serum control samples with high, medium, and low pNfL used in the study. They argue that whereas the assay used in their study did not allow direct application of the age-adjusted cutoffs, age was included as a covariable in all statistical models, so the higher risk of progression in patients with a higher baseline pNfL in their study cannot be attributed to age differences. They also agree that z scores or percentiles would be preferable to fixed cutoffs. This exchange indicates the difficulties of standardization of assays for pNfL measurement. Although pNfL measurements now permeate the literature on numerous neurologic conditions, harmonization of study methods continues to be a challenge.