Leveraging Autoimmune Genetic Correlation to Expand Multiple Sclerosis Risk Loci: Meta-analysis and Proxy-Phenotype Discovery
Multiple sclerosis (MS) is a complex immune-mediated disorder in which inflammatory processes contribute to demyelination and neuroaxonal injury in the central nervous system. Genome-wide association studies (GWAS) have established that MS risk is highly polygenic and overlaps substantially with other autoimmune diseases (ADs), yet many true susceptibility variants remain below conventional genome-wide significance thresholds. The article by Olafsson and colleagues addresses this discovery gap by combining two complementary strategies: (i) expanding statistical power through meta-analysis across large MS cohorts, and (ii) increasing prior probability for association by leveraging genetic correlations—i.e., shared polygenic architecture—between MS and other ADs using polygenic risk scores (PRS) and a proxy-phenotype approach.
Cohorts, genotyping architecture, and dual meta-analysis design
The investigators assembled MS association evidence from an international Immunochip-based dataset (IMSGC discovery phase) and three Nordic MS case–control cohorts (Sweden, Norway, Iceland). Because Immunochip content is enriched for immune loci rather than uniformly genome-wide, the authors performed two inverse-variance weighted meta-analyses: one focused on variants well-covered by Immunochip (maximizing power for immune-enriched loci), and a second genome-wide analysis using imputed genotypes within the Nordic cohorts to detect associations outside Immunochip tagging. This two-track design explicitly acknowledges platform ascertainment (dense immune coverage vs. genome-wide imputation) while still pursuing both sensitivity (for immune loci) and breadth (for non-Immunochip signals).
Seven MS susceptibility variants identified by direct meta-analysis
Excluding the major histocompatibility complex (MHC) region—where extensive linkage disequilibrium complicates fine mapping—the meta-analysis yielded seven previously unreported MS associations reaching genome-wide significance. These signals map to loci including MTHFR (rs1801133), ETS1 (rs4245080), LPP (rs11707807), NCOA2 (rs13260060), and a region spanning immune-relevant genes and non-coding RNAs near PTPRC / MIR181A1HG (rs9427431), alongside intergenic signals near CD6/CD5 (rs175126, conditional) and the DLEU1/DLEU2 microRNA cluster (rs806321). Notably, rs1801133 is a missense variant in MTHFR (Ala222Val) with known functional consequences for enzyme activity and homocysteine metabolism, standing out from the predominantly intronic/intergenic architecture typical of many GWAS hits. The authors further interrogated functional plausibility via expression quantitative trait locus (eQTL) resources (e.g., GTEx and Icelandic whole-blood eQTLs) and regulatory annotations to connect association signals to gene regulation where possible.
Quantifying autoimmune overlap using PRS and identifying an MS–PBC axis
A central contribution of the study is its systematic PRS-based mapping of shared common-variant risk across multiple autoimmune phenotypes in an Icelandic population sample, with the extended MHC region removed to avoid dominance by HLA effects. This analysis recapitulated a broader structure in which several ADs cluster according to serological features (roughly, “seronegative” inflammatory bowel/psoriasis-related conditions vs. “seropositive” autoantibody-associated diseases). Strikingly, primary biliary cirrhosis (PBC) and MS did not fit neatly into either cluster; instead, the MS PRS showed its strongest correlation with PBC, and the PBC PRS showed notable predictive capacity for MS. The authors replicated the key PRS-based MS–PBC relationship in an independent Swedish MS cohort, strengthening the inference that this is not a cohort-specific artifact but reflects meaningful shared genetic etiology outside the HLA region.
Proxy-phenotype (PBC-informed) discovery yields seven additional MS variants
Having established PBC as the most informative proxy phenotype for MS in their framework, the authors applied a proxy-phenotype strategy: they selected variants contributing to a predictive PBC PRS, tested those variants for association with MS across available MS datasets, and corrected for multiple testing within this candidate set. This yielded seven additional novel MS-associated variants that were not explained by established MS loci. These include intronic or regulatory signals in IL12RB2 (rs72678531), TXK/TEC (rs17674224), TNPO3/IRF5 (rs35188261), an additional independent signal at the DLEU region (rs12871645; conditional on rs806321), an upstream variant near EDC4 (rs2271293; conditional on a known MS signal), an intronic variant in TEF (rs2073167), and a missense protective variant in TYK2 (rs35018800; conditional on rs34536443). A key methodological nuance is that the authors removed variants with discordant direction of effect between PBC and MS before pursuing discovery, thereby enriching for biologically concordant pleiotropy.
Biological interpretation: from folate metabolism to cytokine signaling circuitry
The combined set of 14 newly reported variants underscores two mechanistic themes. First, the presence of a functional coding association in MTHFR suggests that metabolic pathways affecting methylation potential and homocysteine handling may intersect with MS susceptibility, either directly (e.g., immune cell function) or indirectly (e.g., neuroinflammatory vulnerability). Second, and more prominently, the proxy-phenotype hits converge on cytokine and tyrosine kinase signaling relevant to T-cell differentiation and activation. IL12RB2 implicates IL-12 receptor biology and Th1 polarization; TXK is a Th1-associated transcriptional regulator influencing cytokine production; and TYK2 is a pivotal Janus kinase family member mediating signaling downstream of multiple cytokine receptors. The enrichment of such loci is consistent with MS immunopathogenesis and illustrates how cross-disease genetic sharing can prioritize immune circuitry that may be underpowered in single-phenotype GWAS.
Implications, limitations, and next steps for MS genomics
Scientifically, this work demonstrates a pragmatic path to discovery in highly polygenic disease: augment sample size via meta-analysis while using genetic correlation to reshape the multiple-testing problem by testing better-justified hypotheses. Clinically, the findings do not imply immediate diagnostic utility, but they sharpen the mechanistic map of MS risk—particularly highlighting an MS–PBC shared component beyond HLA—and reinforce the relevance of cytokine signaling pathways that are already therapeutically tractable. Important limitations remain: PRS-derived proxy approaches depend on the quality and comparability of summary statistics, may preferentially recover pleiotropic immune loci (potentially underrepresenting CNS-intrinsic mechanisms), and require careful conditioning to separate independent signals in dense regions. The logical next steps are fine-mapping, cell-type-specific functional validation (especially in relevant immune subsets and CNS-resident cells), and integrative analyses linking these variants to gene expression, chromatin state, and cytokine-response phenotypes—work that would convert statistical associations into causal biology and, ultimately, more precise therapeutic hypotheses.
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:
Olafsson, S., Stridh, P., Bos, S. D., Ingason, A., Euesden, J., Sulem, P., ... & Stefansson, K. (2017). Fourteen sequence variants that associate with multiple sclerosis discovered by meta-analysis informed by genetic correlations. NPJ genomic medicine, 2(1), 24.
