Multiple Sclerosis–Specific Reference Curves for Brain Volumes to Explain Disease Severity
David R. van Nederpelt, Len Bos, Rozemarijn M. Mattiesing, Eva Strijbis, Bastiaan Moraal, Joost P.A. Kuijer, Jeroen Hoogland, Henk Mutsaerts, Bernard M.J. Uitdehaag, Joep Killestein, Lizette Heine, Bas Jasperse, Frederik Barkhof, Menno M. Schoonheim, Hugo Vrenken
Abstract
BACKGROUND AND OBJECTIVES: Brain atrophy is relevant for understanding disease progression and treatment response in people with multiple sclerosis (pwMS). Automatic brain volume-reporting tools often rely on healthy control (HC) reference curves to interpret brain volumes, whereas brain volume loss is different in pwMS. This observational study aimed to develop an MS-specific reference model for brain volumes and evaluate its performance compared with HC-based curves, as a proof-of-concept. METHODS: Participants, pwMS and HCs, from the Amsterdam MS cohort were included based on the availability of T1-weighted MR scans. Normalized brain volumes (NBVs) were obtained using commercially available software. The software program also provides NBV percentiles, based on age-specific and sex-specific HC curves, grouped into NBV quartiles, describing deviation from expected NBVs. Disease severity was determined with the MS severity score (MSSS), Symbol Digit Modalities Test (SDMT), and 9-Hole Peg Test (9HPT). An MS-specific model was developed by regressing NBVs against age, sex, disease duration, and MS phenotype. The resulting MS model was also used to classify pwMS into quartiles describing deviation from expected NBV, given the modeled patient characteristics, with leave-one-out predictions. Quartile classification from HC-based and MS-based reference curves was compared with MSSS using analysis of variance (ANOVA). RESULTS: = 0.01, respectively). DISCUSSION: NBV values derived from an MS-specific reference model offer improved relevance for assessing disease severity compared with curves derived from age-specific and sex-specific HC reference models. Improving the model toward application in individual people could enhance clinical implementation.