One of the biggest challenges in managing multiple sclerosis is the heterogeneity of clinical manifestations and progression trajectories. It still remains to be elucidated whether this heterogeneity is reflected by discrete immune signatures in the blood as a surrogate of disease pathophysiology. Accordingly, individualized treatment selection based on immunobiological principles is still not feasible. Using two independent multicentric longitudinal cohorts of patients with early multiple sclerosis ( n = 309 discovery and n = 232 validation), we were able to identify three distinct peripheral blood immunological endophenotypes by a combination of high-dimensional flow cytometry and serum proteomics, followed by unsupervised clustering. Longitudinal clinical and paraclinical follow-up data collected for the cohorts revealed that these endophenotypes were associated with disease trajectories of inflammation versus early structural damage. Investigating the capacity of immunotherapies to normalize endophenotype-specific immune signatures revealed discrete effect sizes as illustrated by the limited effect of interferon-β on endophenotype 3–related immune signatures. Accordingly, patients who fell into endophenotype 3 subsequently treated with interferon-β exhibited higher disease progression and MRI activity over a 4-year follow-up compared with treatment with other therapies. We therefore propose that ascertaining a patient’s blood immune signature before immunomodulatory treatment initiation may facilitate prediction of clinical disease trajectories and enable personalized treatment decisions based on pathobiological principles.

A concern for researchers planning multisite studies is that scanner and T1-weighted sequence-related biases on regional volumes could overshadow true effects, especially for studies with a heterogeneous set of scanners and sequences. Current approaches attempt to harmonize data by standardizing hardware, pulse sequences, and protocols, or by calibrating across sites using phantom-based corrections to ensure the same raw image intensities. We propose to avoid harmonization and phantom-based correction entirely. We hypothesized that the bias of estimated regional volumes is scaled between sites due to the contrast and gradient distortion differences between scanners and sequences. Given this assumption, we provide a new statistical framework and derive a power equation to define inclusion criteria for a set of sites based on the variability of their scaling factors. We estimated the scaling factors of 20 scanners with heterogeneous hardware and sequence parameters by scanning a single set of 12 subjects at sites across the United States and Europe. Regional volumes and their scaling factors were estimated for each site using Freesurfer's segmentation algorithm and ordinary least squares, respectively. The scaling factors were validated by comparing the theoretical and simulated power curves, performing a leave-one-out calibration of regional volumes, and evaluating the absolute agreement of all regional volumes between sites before and after calibration. Using our derived power equation, we were able to define the conditions under which harmonization is not necessary to achieve 80% power. This approach can inform choice of processing pipelines and outcome metrics for multisite studies based on scaling factor variability across sites, enabling collaboration between clinical and research institutions.

Disability in multiple sclerosis is generally classified by sensory and motor symptoms, yet cognitive impairment has been identified as a frequent manifestation already in the early disease stages. Imaging- and more recently blood-based biomarkers have become increasingly important for understanding cognitive decline associated with multiple sclerosis. Thus, we sought to determine the prognostic utility of serum neurofilament light chain levels alone and in combination with MRI markers by examining their ability to predict cognitive impairment in early multiple sclerosis. A comprehensive and detailed assessment of 152 early multiple sclerosis patients (Expanded Disability Status Scale: 1.3 ± 1.2, mean age: 33.0 ± 10.0 years) was performed, which included serum neurofilament light chain measurement, MRI markers (i.e. T₂-hyperintense lesion volume and grey matter volume) acquisition and completion of a set of cognitive tests (Symbol Digits Modalities Test, Paced Auditory Serial Addition Test, Verbal Learning and Memory Test) and mood questionnaires (Hospital Anxiety and Depression scale, Fatigue Scale for Motor and Cognitive Functions). Support vector regression, a branch of unsupervised machine learning, was applied to test serum neurofilament light chain and combination models of biomarkers for the prediction of neuropsychological test performance. The support vector regression results were validated in a replication cohort of 101 early multiple sclerosis patients (Expanded Disability Status Scale: 1.1 ± 1.2, mean age: 34.4 ± 10.6 years). Higher serum neurofilament light chain levels were associated with worse Symbol Digits Modalities Test scores after adjusting for age, sex Expanded Disability Status Scale, disease duration and disease-modifying therapy (B = −0.561; SE = 0.192; P = 0.004; 95% CI = −0.940 to −0.182). Besides this association, serum neurofilament light chain levels were not linked to any other cognitive or mood measures (all P-values > 0.05). The tripartite combination of serum neurofilament light chain levels, lesion volume and grey matter volume showed a cross-validated accuracy of 88.7% (90.8% in the replication cohort) in predicting Symbol Digits Modalities Test performance in the support vector regression approach, and outperformed each single biomarker (accuracy range: 68.6–75.6% and 68.9–77.8% in the replication cohort), as well as the dual biomarker combinations (accuracy range: 71.8–82.3% and 72.6–85.6% in the replication cohort). Taken together, early neuro-axonal loss reflects worse information processing speed, the key deficit underlying cognitive dysfunction in multiple sclerosis. Our findings demonstrate that combining blood and imaging measures improves the accuracy of predicting cognitive impairment, highlighting the clinical utility of cross-modal biomarkers in multiple sclerosis.

Objective: Fatigue is a frequent and severe symptom in multiple sclerosis (MS), but its pathophysiological origin remains incompletely understood. We aimed to examine the predictive value of subcortical gray matter volumes for fatigue severity at disease onset and after 4 years by applying structural equation modeling (SEM).Methods: This multicenter cohort study included 601 treatment‐naive patients with MS after the first demyelinating event. All patients underwent a standardized 3T magnetic resonance imaging (MRI) protocol. A subgroup of 230 patients with available clinical follow‐up data after 4 years was also analyzed. Associations of subcortical volumes (included into SEM) with MS‐related fatigue were studied regarding their predictive value. In addition, subcortical regions that have a central role in the brain network (hubs) were determined through structural covariance network (SCN) analysis.Results: Predictive causal modeling identified volumes of the caudate (s [standardized path coefficient] = 0.763, p = 0.003 [left]; s = 0.755, p = 0.006 [right]), putamen (s = 0.614, p = 0.002 [left]; s = 0.606, p = 0.003 [right]) and pallidum (s = 0.606, p = 0.012 [left]; s = 0.606, p = 0.012 [right]) as prognostic factors for fatigue severity in the cross‐sectional cohort. Moreover, the volume of the pons was additionally predictive for fatigue severity in the longitudinal cohort (s = 0.605, p = 0.013). In the SCN analysis, network hubs in patients with fatigue worsening were detected in the putamen (p = 0.008 [left]; p = 0.007 [right]) and pons (p = 0.0001).Interpretation: We unveiled predictive associations of specific subcortical gray matter volumes with fatigue in an early and initially untreated MS cohort. The colocalization of these subcortical structures with network hubs suggests an early role of these brain regions in terms of fatigue evolution. ANN NEUROL 2022;91:192–202

Background: Multiple sclerosis is a chronic inflammatory central nervous system disease diagnosed by clinical presentation and characteristic magnetic resonance imaging findings. The role of cerebrospinal fluid (CSF) analysis has been emphasized in particular in the context of differential diagnosis in patients with a first episode suggestive of multiple sclerosis. Objective: We investigated here the potential additional value of analysis of CSF cellularity by fluorescence activated cell sorting (FACS) in the setting of a routine diagnostic work-up in our inpatient clinic. Methods: CSF cells from back-up samples from patients with suspected chronic inflammatory central nervous system disorder were analyzed by FACS and correlated with clinical data, magnetic resonance imaging findings and oligoclonal band status. Results: We found distinct changes of T cell/monocyte (CD4/CD14) and B cell/monocyte (CD20/CD14) ratios between clinically isolated syndrome (CIS)/multiple sclerosis and other neurologic diseases or other inflammatory neurologic diseases. In particular, patients with a rapid transition from CIS to multiple sclerosis had an elevated CD4/CD14 ratio. A subgroup analysis showed diagnostic value of CD4/CD8 ratio in the differential diagnosis of CIS/multiple sclerosis to neurosarcoidosis. Conclusion: The diagnostic and prognostic accuracy of autoimmune neuroinflammatory diseases can be improved by FACS analysis of CSF cells.

Background: Network science provides powerful access to essential organizational principles of the brain. The aim of this study was to investigate longitudinal evolution of gray matter networks in early relapsing–remitting MS (RRMS) compared with healthy controls (HCs) and contrast network dynamics with conventional atrophy measurements. Methods: For our longitudinal study, we investigated structural cortical networks over 1 year derived from 3T MRI in 203 individuals (92 early RRMS patients with mean disease duration of 12.1 ± 14.5 months and 101 HCs). Brain networks were computed based on cortical thickness inter-regional correlations and fed into graph theoretical analysis. Network connectivity measures (modularity, clustering coefficient, local efficiency, and transitivity) were compared between patients and HCs, and between patients with and without disease activity. Moreover, we calculated longitudinal brain volume changes and cortical atrophy patterns. Results: Our analyses revealed strengthening of local network properties shown by increased modularity, clustering coefficient, local efficiency, and transitivity over time. These network dynamics were not detectable in the cortex of HCs over the same period and occurred independently of patients’ disease activity. Most notably, the described network reorganization was evident beyond detectable atrophy as characterized by conventional morphometric methods. Conclusion: In conclusion, our findings provide evidence for gray matter network reorganization subsequent to clinical disease manifestation in patients with early RRMS. An adaptive cortical response with increased local network characteristics favoring network segregation could play a primordial role for maintaining brain function in response to neuroinflammation.