A Genome-Wide Search for the Genetics of Multiple Sclerosis Relapse: Evidence Implicating LRP2 in Disease Activity

Genome-wide association studies (GWAS) in multiple sclerosis (MS) have historically prioritised susceptibility—that is, identifying variants that differentiate cases from controls—rather than variants that shape clinical trajectory after onset. Zhou and colleagues address a clinically consequential gap: relapse risk, a core feature of relapsing MS and a primary endpoint in many pivotal disease-modifying therapy trials. Because relapse histories are rarely captured prospectively and longitudinally at scale, earlier genetic studies have struggled to test relapse risk in a genome-wide manner. The present work therefore reframes the genetic question from “who develops MS?” to “who relapses sooner and more often?”, using prospectively adjudicated relapse outcomes and time-to-event modelling tailored to recurrent events.

Study Architecture: A Three-Stage Longitudinal GWAS Strategy
The investigators implemented a three-stage design intended to balance discovery with validation while leveraging independent, well-characterised cohorts. First, they performed GWAS in an adult relapsing–remitting MS cohort from the southern Tasmania MS Longitudinal Study (TasMSL). Second, they conducted GWAS in the Ausimmune Longitudinal Study (AusLong), which followed individuals after a classic first demyelinating event and subsequent conversion to MS. Third, they tested top signals from the first two stages in an independent US paediatric MS cohort, followed prospectively after onset. This staged approach is notable because it evaluates genetic predictors across distinct epochs of MS (early post-onset and established disease) and across age groups, increasing the relevance of any replicated signal to relapse biology rather than cohort-specific artefacts.

Cohorts and Phenotyping: Prospective, Neurologist-Verified Relapses
A major methodological strength is the prospective and clinically verified relapse ascertainment. In TasMSL, 141 genotyped participants with relapsing–remitting MS were followed for a mean of 2.3 years, with relapses reported via scheduled assessments and real-time notifications, and included only when diagnosed and verified by a neurologist. AusLong contributed 127 genotyped participants who had a relapse-onset first demyelinating event and converted to MS before five-year review, with relapses defined consistently and captured over a five-year horizon. The US paediatric cohort enrolled 181 children meeting criteria for paediatric MS or high-risk clinically isolated syndrome, tracking relapses prospectively and recording disease-modifying therapy exposure as a time-varying covariate. Across studies, relevant covariates—including age, sex, vitamin D (25(OH)D), HLA-DRB1*15 status, and treatment—were measured to support sensitivity analyses.

Genomics and Statistics: Imputation and Recurrent-Event Survival Modelling
Because genotyping platforms differed across cohorts, the authors used genotype imputation (Minimac3 with 1000 Genomes Phase 3 as reference) to maximise coverage, applying stringent filters on imputation quality and minor allele frequency. The relapse endpoint was analysed using a mixed-effects Cox model suitable for repeated events, an important distinction from conventional “time to first event” analyses. Here, each relapse defines a new “epoch,” resetting the time clock to evaluate subsequent relapse hazards while leveraging all observed events per participant. For synthesis across cohorts, an inverse-variance fixed-effects meta-analysis was used, with the authors explicitly noting the limited suitability of random-effects modelling when combining fewer than four studies. Sensitivity analyses compared covariate adjustment schemes and suggested that the principal signal was robust to differing adjustment sets.

Principal Finding: An LRP2 Intronic Variant Predicts Relapse Hazard
In the pooled meta-analysis across the three cohorts (total n=449), the single nucleotide polymorphism rs12988804 within LRP2 achieved genome-wide significance for relapse risk, with a hazard ratio of 2.18 and p=3.30×10⁻⁸, indicating that carriers of the risk allele experienced more than double the relapse hazard compared with non-carriers. Importantly, the direction of effect was consistent across all three cohorts, strengthening internal credibility despite modest sample size by GWAS standards. No additional loci reached genome-wide significance. The authors also examined a secondary outcome in AusLong—time to conversion to MS (operationalised as time to second relapse after the first demyelinating event)—and observed a similar magnitude of effect (HR≈2.10), reinforcing the notion that this locus may influence inflammatory disease activity early in the MS course as well as relapse propensity in established disease.

Biological Plausibility: LRP2 as a Neurodevelopmental and Repair-Relevant Node
LRP2 (megalin) is compelling not because it is an obvious immune gene, but because of its documented roles in central nervous system development and cellular signalling. The article highlights evidence from animal models in which complete LRP2 knockout produces profound brain malformations and severe vitamin D deficiency phenotypes, and expression studies showing LRP2 on choroid plexus and ependymal cells as well as on oligodendrocytes in spinal cord white matter. Mechanistically, LRP2 has been linked to neural development pathways (including Notch-related signalling contexts) and to axonal guidance (in conjunction with LRP1), which frames a plausible bridge from genetic variation to processes relevant for MS lesion dynamics, remyelination, and neurorepair. Notably, adjusting for serum 25(OH)D did not materially change the association, suggesting the relapse signal is not simply a proxy for measured vitamin D status in these cohorts.

Interpretation, Limitations, and Translational Implications
This study is best viewed as a proof-of-concept that prospective relapse phenotyping can enable discovery of relapse-modifying loci at genome-wide scale. Its principal limitation is power: even with three cohorts, the sample size remains small relative to typical GWAS, and the authors appropriately underscore the need for replication in additional longitudinal datasets. The lead SNP rs12988804 is intronic with no established direct functional annotation in the paper, raising the possibility that it tags other functional variants; the authors note a nearby rare missense candidate (rs754235034) predicted to be potentially damaging by in silico tools, but laboratory validation is required. If replicated and functionally clarified, LRP2 would represent an instructive departure from predominantly immunological susceptibility architecture, pointing instead toward neurodevelopmental or neurorepair pathways as modulators of relapse activity and, potentially, treatment response stratification. The work therefore motivates larger harmonised longitudinal consortia integrating genomics with dense clinical and biomarker time series.

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:
Zhou, Y., Graves, J. S., Simpson, S., Charlesworth, J. C., Van Der Mei, I., Waubant, E., ... & Taylor, B. V. (2017). Genetic variation in the gene LRP2 increases relapse risk in multiple sclerosis. Journal of Neurology, Neurosurgery & Psychiatry, 88(10), 864-868.