MS is characterized by pronounced inter-individual heterogeneity. MS lesion characterization has demonstrated substantial variability in lesion burden, composition, and inflammatory activity between and within donors, and previous studies by us and others have linked this variability to differences in clinical progression [18, 27, 29, 32]. However, the biological mechanisms of this heterogeneity between individuals still remain incompletely understood.

In this study, we extend these previous MS lesion-focused analyses by examining whether donor-specific pathological features (perivascular cuffs, microglial nodules, BRLs, and remyelination efficiency) associate with inter-individual differences in lesion patterns and disease course in a large MS autopsy cohort of the NBB. We show that these features reflect inter-individual differences in key pathological processes that are linked to genetic background, relate to distinct relative lesion type abundance and neuropathological dimensions and associate with clinically relevant differences in disease course.

Perivascular cuffs have been associated with lesion formation and demyelinating lesion activity in MS [9, 14, 37, 39], and with a more severe clinical disease trajectory [14, 32]. Our findings suggest that the presence of perivascular cuffs are genetically influenced, as they are enriched among carriers of variants associated with both MS susceptibility and disease severity. This is consistent with a role for perivascular cuffs in both lesion initiation and sustained inflammatory activity. Pathologically, this feature is associated with increased predominance of foamy microglia/macrophages in active and mixed lesions, which is consistent with ongoing myelin phagocytosis [4, 29, 41, 42, 47]. This inflammatory profile is further reflected in neuropathological dimensions indicative of ongoing demyelinating activity with limited progression toward chronic scarring. Our data align with previous research [14, 32] linking perivascular cuffs to clinical severity. Overall, these findings support the interpretation of perivascular cuffs as a marker of sustained inflammatory activity across the disease course.

Microglial nodules have been proposed to represent early inflammatory events and possible starting points of MS lesions [9, 43]. Their enrichment among carriers of the HLA-DRB1*15:01 susceptibility-associated allele supports a link between genetic risk and early innate immune activation within the CNS, and their sex bias parallels known sex differences in MS susceptibility [8]. The association of nodules with increased proportions of active lesions further supports their role in early lesion dynamics [3, 9, 43]. Despite the genetic and pathological associations with nodules, their presence was not associated with clinical course. Together, these findings are consistent with the interpretation that the microglial nodules reflect permissive or early-stage pathological processes rather than direct drivers of disability. The dissociation between pathological activity and clinical outcome highlights the importance of considering multiple biological processes when interpreting MS heterogeneity.

BRLs have emerged as a pathological marker of rapid disease progression [25], and our findings further reinforce their clinical and biological relevance. BRLs were enriched among carriers of the severity-associated SNP at the DYSF–ZNF638 locus, suggesting a genetic contribution to chronic lesion activity. Pathologically, BRLs were associated with increased proportions of active and mixed lesions, increased microglia activation, higher BRS lesion rate, and increased cortical lesion rate. Clinically, donors with BRLs exhibited a more aggressive disease course starting at a younger age. These findings indicate that BRL associated pathology extends beyond individual lesions and is consistent with high inflammatory activity across the CNS.

Donors with cuffs and BRLs were enriched for the severity-associated common genetic variant, consistent with a previous observation linking this locus to prolonged inflammation and chronic lesion activity [25]. While these associations are observational and mechanistic interpretations remain speculative, this variant has been associated with increased neuro-axonal loss in MS [5, 13, 19, 35], suggesting increased immune activation associating with genetically higher susceptibility to neuro-axonal damage [41]. Mechanistically, this locus has been linked to altered regulation of flanking gene ZNF638, encoding NP220, a component of the HUSH complex involved in transcriptional silencing of viral DNA, including EBV-derived sequences [13]. Reduced suppression of latent EBV activity could enhance antigen presentation and may contribute to sustained compartmentalized immune activation, potentially supporting chronic lesion activity rather than directly determining disease susceptibility. In addition, the second flanking gene, DYSF, encodes dysferlin, a mediator of membrane repair following cellular injury [12, 21, 36]. Dysferlin expression is increased in normal-appearing grey matter in MS [13], and impaired dysferlin-mediated repair mechanisms may further contribute to persistence of tissue damage within chronically inflamed lesions.

Remyelination efficiency captures donor-specific repair capacity [7] and reflects another process that can contribute to inter-individual variability. In contrast to inflammatory features, remyelination efficiency was not associated with the genetic variants examined, suggesting that repair capacity may be more strongly influenced by local tissue environment and cumulative disease burden. Poor remyelination efficiency aligned with a pathological profile dominated by mixed and inactive lesions, increased BRS- and cortical lesion rate, and the dimension associated with gliosis and chronic scar formation. Clinically, these donors exhibited a less favorable disease course. These findings support the idea that a less permissive tissue environment, marked by ongoing inflammation and gliosis, may limit effective remyelination.

In contrast to donor-specific pathological features, the presence of active lesions primarily reflects a transient inflammatory disease state, associated with ongoing relapse-related activity [32]. Although active lesions showed overlapping pathological and clinical associations with several donor-specific features, the donor-level features described here likely reflect broader and more persistent biological processes that shape lesion evolution across the CNS. While these features define subgroups, they frequently co-occur, reflecting overlapping pathological processes that may together influence lesion development and persistence within individual donors. Donors lacking all four features did not exhibit an entirely benign disease course, underscoring the multifactorial nature of MS.

Our findings on post-mortem tissue raise important questions regarding the translation of these pathological features into clinical practice. If identifiable, ante-mortem donor-specific features such as perivascular cuffs, BRLs, and remyelination capacity could help predict disease course and guide personalized therapeutic strategies targeting distinct biological pathways. Achieving this will require reliable in vivo biomarkers. Recent advances in imaging, including translocator protein 18 kDa positron emission tomography (TSPO-PET), have demonstrated that BRLs can be detected in patients with MS [25]. Detection of BRLs using routinely applied magnetic resonance imaging (MRI) techniques would further enhance their clinical applicability. In contrast, visualization of microglial nodules and perivascular cuffs will likely require imaging modalities with higher spatial resolution. Although remyelination can also be assessed using MRI and PET [38], these methods currently provide indirect measures of myelin repair. The identification of imaging, fluid, or extracellular vesicle biomarkers associated with these pathological features may therefore facilitate longitudinal studies into their temporal dynamics and biological significance. Candidate biomarkers include established markers of axonal injury, myelin damage, and astrocyte activation, including NFL, MOG, and GFAP [1, 41], as well as molecules that have been previously associated to pathological processes, including CHI3L1, OPN, and C1Q in nodules, CHIT-1 and GPNMB in mixed lesions, and TGFB, EGF, and IGF1 in remyelinating environments [7, 23, 43].

Beyond the four donor-specific pathological features analyzed here, additional variables may further refine our understanding of inter-individual heterogeneity in MS pathology. For example, iron-containing lesions identified using Turnbull blue staining are clinically relevant, as their in vivo correlates, paramagnetic rim lesions on MRI, are associated with progression independent of relapse activity [46]. However, Turnbull blue staining is not routinely performed within the NBB cohort and was therefore not included in the present analyses. Moreover, lesion sampling may be biased toward white matter pathology, as white matter lesions are more readily identified macroscopically and by MRI than grey matter lesions. In addition, remyelination efficiency likely differs between white and grey matter, whereas cortical remyelination remains challenging to assess. While MRI-guided tissue sampling was available for the majority of donors, variability in sampling procedures may still have modestly favored detection of active lesions in some cases. Future studies integrating broader genetic datasets may further clarify the contribution of genetic variation to pathological heterogeneity beyond the two SNPs studied here. Finally, quantification of total lesion burden measured by area affected and spatial lesion distribution may provide additional clinically relevant dimensions of pathology.

In conclusion, this study demonstrates that perivascular cuffs, microglial nodules, BRLs, and remyelination efficiency reflect biologically relevant pathological processes that vary between individuals with MS. These donor-specific features provide insight into how disease biology is expressed at the individual level and help contextualize variability in lesion type distribution and clinical trajectories. Integrating donor-specific pathology with genetic background, lesion classification, and clinical data, offers a more biologically grounded framework for interpreting post-mortem MS tissue and studying disease heterogeneity. Such approaches may improve the design and interpretation of future neuropathological and translational studies.