Gene–Environment Interactions in Multiple Sclerosis: How Polygenic Susceptibility Amplifies Early-Life Risk
Multiple sclerosis (MS) is a complex neuroinflammatory disorder in which susceptibility arises from both inherited genetic architecture and non-genetic exposures accumulated across the life course. While many environmental risk factors (e.g., smoking and obesity) and genetic determinants (notably within the major histocompatibility complex, MHC) have been independently associated with MS, a central unresolved question is whether genetic liability modifies the effect size of environmental exposures—that is, whether there are meaningful gene–environment interactions that could illuminate pathobiology and sharpen prevention strategies. Jacobs and colleagues (2021) address this question using the scale of the UK Biobank, explicitly testing statistical interaction between polygenic risk for MS and established environmental risk factors.
Cohort, Case Ascertainment, and Analytical Framework
The investigators implemented a case–control design within UK Biobank, identifying MS cases via ICD-10 coding and/or self-report, and comparing them to a very large control population. Environmental associations were quantified using multivariable logistic regression, adjusting for key confounders (including age, sex, ethnicity, socioeconomic deprivation, and birth latitude). To manage multiplicity across candidate exposures, they applied a Bonferroni correction for the primary risk-factor screen. This framework is important because apparent gene–environment interactions can be artifacts of confounding, selection, or model misspecification; therefore, careful covariate control and conservative inference thresholds are essential when interrogating interaction structure in observational biobank data.
Environmental Exposures Implicated in MS Susceptibility
Among the environmental factors evaluated, the study reports robust associations between MS and (i) childhood obesity (captured as larger body size in childhood), (ii) smoking (including smoking at young ages), and (iii) earlier age at menarche. These findings align with a broader literature suggesting that immune-metabolic state in early life and inflammatory toxicant exposures may contribute to MS risk, while reproductive timing may reflect underlying endocrine and adiposity-linked biology relevant to immune regulation. A key methodological point is that the authors did not treat these exposures in isolation: exposures showing evidence of association were subsequently considered together (and alongside strong genetic factors such as HLA alleles) to evaluate whether the observed effects were independent or potentially overlapping through shared pathways (for example, adiposity influencing pubertal timing).
Constructing Polygenic Risk Scores With and Without the MHC
To quantify genome-wide inherited susceptibility, Jacobs et al. derived polygenic risk scores (PRS) using a clumping-and-thresholding strategy with external weights from the largest MS genome-wide association study meta-analysis available at the time. Critically, they built two PRS variants: one including the MHC region (PRS_MHC) and one excluding it (PRS_non-MHC). This separation matters because the MHC harbors the strongest common genetic effects in MS; if interaction signals were driven solely by that region, they might not generalize to broader polygenic liability. The PRS were tuned in a training subset and validated in an independent testing subset, and both PRS versions showed strong association with MS case status in validation, with the MHC-inclusive score performing better—as expected given the magnitude of MHC effects.
Core Finding: Interaction Between Childhood Obesity and Polygenic Risk
The central result is evidence that childhood obesity and genetic susceptibility interact such that the combined effect exceeds what would be expected from simply adding their independent contributions. The authors quantify interaction on an additive scale using the attributable proportion due to interaction (AP), reporting a positive AP for childhood obesity with both PRS_MHC and PRS_non-MHC—indicating that a non-trivial fraction of risk among jointly exposed individuals is attributable to interaction itself rather than to the sum of separate effects. Importantly, the persistence of the interaction signal when excluding the MHC suggests that this phenomenon is not merely an HLA-driven artifact; instead, it is compatible with a model in which broad polygenic background amplifies the risk consequences of early-life adiposity.
Interpretation in Immunobiological Terms and Relationship to Prior HLA-Focused Literature
Prior gene–environment work in MS has often centered on specific HLA haplotypes (e.g., DRB115:01 and protective A02:01), with reports that smoking, adolescent obesity, and certain infections confer disproportionate risk in genetically predisposed individuals. Jacobs and colleagues extend this concept by shifting from single-locus interaction to polygenic modification, which is biologically plausible given that MS risk alleles are enriched in immune regulatory pathways. One mechanistic interpretation is that childhood obesity may establish a chronic pro-inflammatory milieu, alter antigen presentation dynamics, and reshape T-cell/B-cell activation thresholds; in an individual with high polygenic risk—already nearer to an autoimmunity “activation boundary”—the same obesogenic exposure could more readily precipitate CNS-directed immune dysregulation. The fact that interaction is detectable even beyond the MHC is consistent with multi-pathway synergy, rather than a single-gene explanatory model.
Limitations, Replication Needs, and Prevention Implications
Despite its scale and rigorous approach, the study remains observational and therefore vulnerable to residual confounding, measurement error in retrospectively reported early-life exposures, and selection biases intrinsic to UK Biobank participation (which may distort exposure distributions and associations). Interaction estimates are also statistically delicate: they can be sensitive to model specification, prevalence assumptions, and correlation between exposures. Accordingly, the authors emphasize the need for replication in independent cohorts and, ideally, triangulation using complementary designs (including longitudinal data and causal inference frameworks). If confirmed, however, the translational implication is consequential: childhood obesity would not only be a modifiable MS risk factor but also a risk factor whose harm is magnified in those with high polygenic susceptibility. That framing supports the rationale for targeted prevention—particularly early-life adiposity reduction—while also motivating deeper mechanistic studies of immunometabolic programming across genetic risk strata.
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:
Jacobs, B. M., Noyce, A. J., Bestwick, J., Belete, D., Giovannoni, G., & Dobson, R. (2021). Gene-environment interactions in multiple sclerosis: a UK Biobank study. Neurology: Neuroimmunology & Neuroinflammation, 8(4), e1007.
