Epigenetic Signatures Beyond Genetic Risk in Multiple Sclerosis: Insights from a Large-Scale EWAS

Multiple sclerosis (MS) is a chronic inflammatory and neurodegenerative disorder of the central nervous system whose risk reflects both inherited susceptibility and environmental exposures. While genome-wide association studies have identified more than 200 MS-associated loci—dominated by immune-related genes and led by the major histocompatibility complex (MHC) region—genetics alone does not fully explain disease onset, heterogeneity, or temporal dynamics. DNA methylation (DNAm), a key epigenetic mark typically occurring at CpG dinucleotides, provides a mechanistically plausible interface between inherited background and environment because it can shift with exposures (e.g., smoking, viral infection, vitamin D status) yet also be influenced by genotype. The article by Xavier and colleagues positions DNAm not as a secondary correlate but as a measurable molecular layer that may capture disease-relevant immune dysregulation earlier than clinical consolidation, and potentially beyond what polygenic models capture.

Study architecture: large-scale whole-blood EWAS with early-disease emphasis
The investigators performed an epigenome-wide association study (EWAS) using whole-blood DNAm profiles from three cohorts totaling 583 clinically definite MS cases and 643 controls, integrating both discovery and validation designs. A notable strength is the discovery cohort derived from the Ausimmune study: individuals sampled around the first demyelinating event and followed prospectively, enabling analyses that are less confounded by long disease duration and extensive treatment exposure. DNAm was measured using Illumina methylation arrays (EPIC and 450K platforms across cohorts), while genotyping (where available) enabled explicit modeling of the canonical HLA-DRB1 risk haplotype and a non-MHC polygenic risk score (PRS). The workflow (Figure 1 in the paper) also includes deconvolution to estimate immune-cell proportions and a dedicated cell-specific differential methylation strategy to attribute signals to particular leukocyte subsets rather than treating blood as a homogeneous tissue.

Core epigenetic signal: the MHC/HLA locus dominates and recapitulates prior MS biology
In the combined analysis, the strongest MS-associated methylation differences localized to the MHC class II region within HLA genes, consistent with earlier MS methylation studies and with the centrality of antigen presentation pathways in MS susceptibility. The authors report thousands of differentially methylated positions (DMPs) at stringent genome-wide significance, with a subset meeting both statistical and effect-size criteria, and a concentration of larger effects across HLA-D genes (including multiple differentially methylated regions, DMRs). Importantly, they also connect methylation to function by correlating methylation and expression at HLA-DRB1 in purified immune cell types; methylation at a highlighted region (DMR-2 in exon 2) showed a strong negative association with HLA-DRB1 expression, supporting a biologically coherent model in which hypomethylation aligns with higher expression of antigen presentation machinery. This integrative step strengthens interpretability because it links CpG-level variation to plausible downstream immune phenotypes rather than leaving findings at the level of statistical association alone.

Mediation and causality framing: genotype can drive methylation at HLA-DRB1
A key interpretive challenge in epigenetics is separating “genetically driven methylation” (where sequence variation determines DNAm) from methylation changes that may reflect environment, disease processes, or both. The authors explicitly address this at the HLA locus using mediation-oriented logic (causal inference testing) to evaluate a genotype → methylation → disease chain. Their analyses indicate that methylation differences observed between MS cases and controls at HLA-DRB1–linked regions are driven by genotype in both the discovery and first validation cohorts (where genotypes were available). This is an important nuance: the HLA methylation signal is real and functionally relevant, but a substantial component is consistent with methylation mediating—or at least reflecting—genetic risk rather than representing an independent exposure imprint. The result therefore refines, rather than diminishes, the role of HLA methylation: it becomes part of the molecular manifestation of inherited risk, potentially bridging HLA haplotypes to immune activation programs measurable in peripheral blood.

Beyond known genetics: methylation signatures persist after PRS and HLA adjustment and classify MS more strongly
The paper’s central claim is that substantial MS-associated methylation differences remain after controlling for known genetic risk (HLA-DRB1 haplotype plus a non-MHC PRS derived from established GWAS loci). In genotype-adjusted EWAS models (performed in the early-stage discovery cohort), the authors identify a set of significant DMPs that largely lack evidence of local genetic control (cis methQTLs) or direct association with PRS SNPs, implying that many signals may be more compatible with environmental/lifestyle influence or disease-proximal immune states than with inherited variation. They then translate these CpGs into a genotype risk-corrected methylation score (grcMethScore) intended to quantify epigenetic risk independent of known genetic risk. Notably, receiver operating characteristic analyses show that methylation-based classification outperforms genetic risk alone (reported AUC ~0.85 for the methylation score versus ~0.72 for combined genetic risk), with only marginal improvement when genetics and other covariates are added to methylation. Conceptually, this suggests that DNAm captures risk-relevant biology not adequately summarized by static genotype—whether representing exposure history, immune activation, or early prodromal processes.

Cell-of-origin matters: B cells and monocytes account for the dominant disease-associated methylation architecture
Because whole blood is a mixture of immune lineages, disease-associated differences can reflect either shifts in cell proportions or true within-cell-type methylation remodeling. The authors tackle this using deconvolution (EpiDISH) to estimate major leukocyte fractions and then apply a modified “per cell type” modeling strategy inspired by CellDMC to detect cell-specific DMPs (csDMPs). Two complementary findings emerge: first, MS cases show significant differences in estimated proportions for some subsets (e.g., lower NK and CD8+ T cell proportions and higher CD4+ T cell proportions), consistent with known immunologic alterations; second, and more consequentially for mechanism, the bulk of differential methylation signal is attributed to B cells and monocytes, with comparatively smaller contributions from CD4+ and CD8+ T cells and minimal signal in NK cells and neutrophils. The paper also highlights that the prominent HLA-region methylation effects observed in whole blood largely originate from antigen-presenting lineages (B cells and monocytes), a biologically coherent result given HLA-DR expression patterns. Finally, pathway enrichment suggests both shared and lineage-specific biology: pathways seen in B cells overlap with whole blood and monocytes, while monocytes uniquely capture certain cytokine and interferon-related pathway enrichments, reinforcing the premise that MS-associated epigenetic remodeling is not uniform across immune compartments.

Interpretation, limitations, and translational trajectory: from associative maps to actionable biology
The study provides a rigorous, multi-cohort characterization of MS-associated DNAm with several translational implications: (i) epigenetic variation—particularly when adjusted for known genetic risk—may offer improved disease-state discrimination and could inform future biomarker development; (ii) B cells and monocytes emerge as primary epigenetic “actors,” aligning with therapeutic realities such as the efficacy of CD20-targeting agents and emphasizing the value of finer-resolution profiling (e.g., memory B-cell subsets); and (iii) genotype-dependent and genotype-independent methylation signals should be interpreted differently, with the former potentially mediating inherited risk and the latter potentially reflecting exposure or early immune activation. Nevertheless, the paper is appropriately cautious: validation cohorts differ in age, disease duration, and treatment exposure; deconvolution and cell-specific inference, while informative, are still indirect compared with sorted-cell epigenomics at scale; and expression–methylation correlations were not always derived from the same individuals, limiting causal inference. The most compelling next steps are therefore prospective, treatment-naïve longitudinal designs with purified immune subsets and matched multi-omics to determine whether genotype-independent methylation marks precede clinical MS, track conversion risk, or respond to modifiable exposures—thereby moving from epigenetic association toward mechanistic and preventive precision medicine.

Disclaimer: This blog post is based on the provided research article and is intended for informational purposes only. It is not intended to provide medical advice. Please consult with a healthcare professional for any health concerns.

References:
Xavier, A., Maltby, V. E., Ewing, E., Campagna, M. P., Burnard, S. M., Tegner, J. N., ... & Lechner-Scott, J. (2023). DNA methylation signatures of multiple sclerosis occur independently of known genetic risk and are primarily attributed to B cells and monocytes. International journal of molecular sciences, 24(16), 12576.