Determinants of long-term disability in multiple sclerosis (MS) remain incompletely understood, with clinical course varying greatly among individuals. Recent integrative genetic models have identified genomic clusters associated with divergent disability trajectories. Here, we integrated clinical, genetic, and postmortem neuropathological data from 291 MS brain donors from the UK MS Society Tissue Bank to test whether genomic severity clusters converge on distinct tissue phenotypes. Consistent with prior stratification, donors in genomic cluster 2 had lower age-adjusted disease duration score (ADDS) than cluster 3 (β = − 0.195, P = 0.023) and a shorter interval from progression to death than cluster 1 (β = − 0.225, P = 0.049). However, this adverse clinical profile did not reflect greater active or chronic active lesion burden. Instead, chronic active broad-rim lesions were more frequent in cluster 1 than clusters 2 and 3 (37.3

Objectives Rapid advances in transcriptomics have driven efforts to identify deregulated pathways in multiple sclerosis (MS) tissues, though many detected differentially expressed genes are likely false positives, with only a small fraction reflecting actual pathological events. Robust, integrative methods are essential for accurately understanding the molecular mechanisms underlying MS pathology. Methods We conducted a gene prioritization analysis of MS white matter pathology transcriptomic studies. Articles were sought in Scopus and PubMed up to July 31, 2024. Potentially eligible publications were those that provided either transcriptomics datasets (deposited in GEO) or lists of differentially expressed genes comparing MS white matter to control white matter. Results Applying a vote‐count strategy to search for the intersection of genes reported in multiple independent studies with a consistent fold‐change direction, followed by a Monte Carlo simulation, we identified 528 highly significant differentially expressed multi‐study genes ( p < 0.0001; 10,000 simulations). Functional enrichment analysis revealed deregulation of the folate pathway in MS normal‐appearing white matter, and tumor necrosis factor (TNF) ‐related and complement‐related pathways in active and chronic active lesions, respectively. Network analysis identified 6 key signaling hubs: PTPRC, HLA‐B, MYC, MMP2, COL11A2, MAG. The major nodes identified revealed mechanistic concordance with published in vivo MS models, supporting their value as potential therapeutic targets. Interpretation Our strategy provides a robust framework for integrating gene expression data, effectively identifying the intricate pathways altered in human diseased tissues. This method holds potential for translating findings into drug development strategies. ANN NEUROL 2025

Disability worsening in multiple sclerosis (MS) is linked to neurodegeneration. Cholesterol homeostasis is essential for normal brain function. CYP46A1, crucial for brain cholesterol turnover and reduced in some neurodegenerative diseases, is a potential neuroprotective target. We hypothesized that CYP46A1 is downregulated in MS brains and linked to cholesterol dysbalance. Mass spectrometric analysis of sterols was performed from matched plasma and cerebrospinal fluid (CSF) in an all-female MS cohort (n = 32, mean age = 33). Disability status was recorded at baseline and follow-up. MS brain tissue samples (n = 11; 7 females; ages 38-67; 10 Secondary Progressive MS, 1 Primary Progressive MS; Disease Duration: 13-49 years) and control samples (n = 8; 3 females; ages 41-68) analysed for pathological regions using mass spectrometry and RNA expression using in-situ hybridization. Significant dysregulation in 25-hydroxycholesterol, 27-hydroxycholesterol and 3β-hydroxycholestenoic acid in CSF correlated with disability at baseline and follow-up in the patient population. In brain tissue, reduced cholesterol, 24S-hydroxycholesterol and 24S,25-epoxycholesterol were observed in white matter lesions (p < 0.05), linked to CYP46A1 activity. CYP46A1 expression was enriched in neurons, with reductions in MS grey matter lesions and non-lesions compared to controls (p < 0.01). Cholesterol metabolism is dysregulated in MS and is associated with reduced neuron-specific CYP46A1 expression. Modulating CYP46A1, a druggable target, may benefit progressive MS.

Compartmentalised inflammation is a poorly understood aspect of multiple sclerosis (MS) that is associated with worse outcomes and represents an important therapeutic target. To gain deep insight into compartmentalised inflammation, we have taken the approach of digital spatial profiling of the whole human transcriptome in areas of Central Nervous System (CNS) perivascular and meningeal inflammation and tertiary lymphoid-like structures (TLS) in MS. Critically, we had access to rare archival tissue obtained before the era of disease-modifying therapies, representing the natural history of disease. This analysis has identified differentially expressed genes in TLS compared to meningeal or perivascular inflammation. Pathway analysis highlighted that TLS signalling is dominated by B cell activity including active antibody secretion. Our data demonstrated the diversity of immunoglobulins and the prominence of IgG3-and IgG4-secreting cells in TLS. Intriguingly, our analyses suggest pathways of active viral mRNA translation and associated-cellular responses within TLS immune cells, suggesting TLS may be hubs for viral (re)activation. These findings provide insight into the function of TLS in MS disease pathogenesis and reveal unique immune signatures that may support biomarker development to predict which patients harbour TLS in life. ### Competing Interest Statement The authors have declared no competing interest. Public Health Agency, https://ror.org/03ek62e72, EAT/5496/18 Wellcome Trust, https://ror.org/029chgv08, 110138/Z/15/Z Nanostring Technologies (United States), co-funded European Spatial Biology Award British Society for Immunology, co-funded European Spatial Biology Award

BACKGROUND AND OBJECTIVES:The objective of this study was to determine, at the time of diagnosis, a CSF and MRI profile of intrathecal compartmentalized inflammation predictive of progression independent of relapse activity (PIRA) in early relapsing-remitting multiple sclerosis (RRMS). METHODS:This five-year prospective study included 80 treatment-naïve patients with RRMS enrolled at time of diagnosis. All patients underwent a lumbar puncture, regular neurologic evaluations including an Expanded Disability Status Scale (EDSS) assessment every 6 months, and an annual 3T brain MRI. PIRA was defined as having a confirmed disability progression independent of relapse activity. CSF levels of 68 inflammatory molecules were evaluated in combination with white matter and cortical lesion number (CLn) and volume, and regional gray matter thickness and volume. RESULTS:During the follow-up, 23 patients with RRMS (28.8%) experienced PIRA. At diagnosis, participants with PIRA were older (44.0 ± 10.7 vs 37.4 ± 12.4, p = 0.017) and with more disability (median EDSS score [interquartile range] of 3 [range 2-4] for PIRA vs 1.5 [range 1-2] for no PIRA group, p < 0.001). Random forest selected LIGHT, CXCL13, sTNFR1, sTNFR2, CCL7, MIF, sIL6Rbeta, IL35, CCL2, and IFNβ as the CSF markers best associated with PIRA. sTNFR1 (hazard ratio [HR] 10.11 [2.61-39.10], p = 0.001), sTNFR2 (HR 5.05 [1.63-15.64], p = 0.005), and LIGHT (HR 1.79 [1.11-2.88], p = 0.018) were predictors of PIRA at regression analysis. Baseline thalamus volume (HR 0.98 [0.97-0.99], p = 0.005), middle frontal gyrus thickness (HR 0.05 [0.01-0.72], p = 0.028), and CLn (HR 1.15 [1.05-1.25], p = 0.003) were MRI predictors of PIRA. DISCUSSION:A specific intrathecal inflammatory profile associated with TNF superfamily markers, CLn, and atrophy of several cortical and deep gray matter regions, assessed at time of diagnosis, is predictive of PIRA in early MS.

Characterizing human proteins remains a major challenge: approximately 29% of human proteins lack experimentally validated functions and even well-annotated proteins often lack context-specific phenotypic insights. To enable universal modeling of protein phenotypes, we present ProCyon, a multimodal foundation model that utilizes protein sequence, structure, and natural language for generating and predicting protein phenotypes across diverse knowledge domains. ProCyon is trained on our novel dataset, ProCyon-Instruct, with 33 million protein phenotype instructions. On dozens of benchmarking tasks, ProCyon performs competitively against single-modal and multimodal models. Further, ProCyon conditionally retrieves proteins via mechanisms of action of small molecule drugs and disease contexts, and it generates candidate phenotypic descriptions for poorly characterized proteins, including those implicated in Parkinson's disease that were identified after ProCyon's knowledge cutoff date. We experimentally confirm ProCyon's predictions in multiple sclerosis using post-mortem brain RNA-seq, identifying novel MS genes and elucidating associated pathway mechanisms consistent with cortical pathology. ProCyon paves the way toward a general approach to generate functional insights into the human proteome.

OBJECTIVE:Analysis of postmortem multiple sclerosis (MS) tissues combined with in vivo disease milestones suggests that whereas perivascular white matter infiltrates are associated with demyelinating activity in the initial stages, leptomeningeal immune cell infiltration, enriched in B cells, and associated cortical lesions contribute to disease progression. We systematically examine the association of inflammatory features and white matter demyelination at postmortem with clinical milestones. METHODS:In 269 MS brains, 20 sites were examined using immunohistochemistry for active lesions (ALs) and perivenular inflammation (PVI). In a subset of 22, a detailed count of CD20+ B cells and CD3+ T cells in PVIs was performed. RESULTS:ALs were detected in 22%, whereas high levels of PVI were detected in 52% of cases. ALs were present in 35% of cases with high levels of PVI. Shorter time from onset of progression to death was associated with increased prevalence and higher levels of PVI (both p < 0.0001). Shorter time from onset of progression to wheelchair use was associated with higher prevalence of ALs (odds ratio [OR] = 0.921, 95% confidence interval [CI] = 0.858-0.989, p = 0.0230) and higher level of PVI (OR = 0.932, 95% CI = 0.886-0.981, p = 0.0071). High levels of PVI were associated with meningeal inflammation and increased cortical demyelination and significantly higher levels of B lymphocytes within the PVI. INTERPRETATION:ALs, a feature of early disease stage, persist up to death in a subgroup with high levels of PVI. These features link to a rapid progressive phase and higher levels of meningeal inflammation and B-cell infiltrates, supporting the hypothesis that chronic inflammation drives progression in MS. ANN NEUROL 2024;95:706-719.

Background: Multiple sclerosis (MS) is a clinically heterogeneous disease underpinned by inflammatory, demyelinating and neurodegenerative processes, the extent of which varies between individuals and over the course of the disease. Recognising the clinicopathological features that most strongly associate with disease outcomes will inform future efforts at patient phenotyping. Aims: We used a digital pathology workflow, involving high-resolution image acquisition of immunostained slides and opensource software for quantification, to investigate the relationship between clinical and neuropathological features in an autopsy cohort of progressive MS. Methods: Sequential sections of frontal, cingulate and occipital cortex, thalamus, brain stem (pons) and cerebellum including dentate nucleus (n = 35 progressive MS, females = 28, males = 7; age died = 53.5 years; range 38–98 years) were immunostained for myelin (anti-MOG), neurons (anti-HuC/D) and microglia/macrophages (anti-HLA). The extent of demyelination, neurodegeneration, the presence of active and/or chronic active lesions and quantification of brain and leptomeningeal inflammation was captured by digital pathology. Results: Digital analysis of tissue sections revealed the variable extent of pathology that characterises progressive MS. Microglia/macrophage activation, if found at a higher level in a single block, was typically elevated across all sampled blocks. Compartmentalised (perivascular/leptomeningeal) inflammation was associated with age-related measures of disease severity and an earlier death. Conclusion: Digital pathology identified prognostically important clinicopathological correlations in MS. This methodology can be used to prioritise the principal pathological processes that need to be captured by future MS biomarkers.

Clinical, pathological, and imaging evidence in multiple sclerosis (MS) suggests that a smoldering inflammatory activity is present from the earliest stages of the disease and underlies the progression of disability, which proceeds relentlessly and independently of clinical and radiological relapses (PIRA). The complex system of pathological events driving "chronic" worsening is likely linked with the early accumulation of compartmentalized inflammation within the central nervous system as well as insufficient repair phenomena and mitochondrial failure. These mechanisms are partially lesion-independent and differ from those causing clinical relapses and the formation of new focal demyelinating lesions; they lead to neuroaxonal dysfunction and death, myelin loss, glia alterations, and finally, a neuronal network dysfunction outweighing central nervous system (CNS) compensatory mechanisms. This review aims to provide an overview of the state of the art of neuropathological, immunological, and imaging knowledge about the mechanisms underlying the smoldering disease activity, focusing on possible early biomarkers and their translation into clinical practice. ANN NEUROL 2024