Brain disconnectome mapping and serum neurofilament light levels in multiple sclerosis

The pathophysiological mechanisms for classical plaque characteristics and their predictive value for clinical course and outcome in multiple sclerosis is unclear. Connectivity-based approaches incorporating the distribution and magnitude of the extended brain network aberrations caused by lesions may offer higher sensitivity for axonal damage. Using individual brain disconnectome mapping, we tested the longitudinal associations between putative brain network aberrations and levels of serum neurofilament light chain (sNfL) as a neuroaxonal injury biomarker. Multiple sclerosis patients (n = 328, mean age 42.9 years, 71 % female) were prospectively enrolled at four European multiple sclerosis centres, and reassessed after two years (n = 280). Post-processing of 3 Tesla (3T) MRI data was performed at one centre using a harmonized pipeline, and disconnectome maps were calculated using BCBtoolkit based on individual lesion maps. Global disconnectivity (GD) was defined as the average disconnectome probability in each patients white matter. Serum NfL concentrations were measured by single molecule array (Simoa). Robust linear mixed models (rLMM) with GD or T2-lesion volume (T2LV) as dependent variables, patient and centre as a random factor, sNfL, age, sex, timepoint for visit, diagnosis, and treatment as fixed factors were run. Robust LMM revealed significant associations between higher levels of GD and increased sNfL (t = 2.30, {beta} = 0.03, p = 0.02), age (t = 5.01, {beta} = 0.32, p < 5.5 x 10-7), and diagnosis progressive multiple sclerosis (PMS); t = 1.97, {beta} = 1.06, p = 0.05), but not for sex (t = 0.78, p = 0.43), treatments (effective; t = 0.85, p = 0.39, highly-effective; t = 0.86, p = 0.39) or sNfL change between base line and two-year follow up (t = -1.65, p = 0.10). Voxel-wise analyses revealed distributed associations in cerebellar and brainstem regions. In our prospective multi-site multiple sclerosis cohort, rLMMs demonstrated that the extent of global brain disconnectivity is sensitive to a systemic biomarker of axonal damage, sNfL, in patients with multiple sclerosis. These findings provide a neuropathological correlate of advanced disconnectome mapping and provide a platform for further investigations of the functional and clinical relevance in patients with brain disorders.