Does Low Vitamin D Cause Multiple Sclerosis? What Genetics Reveals
In their research article Vitamin D and Risk of Multiple Sclerosis: A Mendelian Randomization Study, Lauren E. Mokry and colleagues address a deceptively difficult question: does low vitamin D contribute to the development of multiple sclerosis, or is it merely associated with the disease? Multiple sclerosis is a chronic autoimmune disorder in which immune-mediated damage to the central nervous system can produce sensory, motor, cognitive, and visual impairment. Epidemiological observations had long suggested a connection with vitamin D. MS prevalence tends to be higher at greater latitudes, where ultraviolet exposure and cutaneous vitamin D synthesis are generally lower. Lower circulating 25-hydroxyvitamin D, or 25OHD—the standard biomarker of vitamin D status—had also been associated with higher relapse rates, greater MRI activity, and greater disability. Yet these observations could not establish causality. People with early or established MS may spend less time outdoors, lowering their vitamin D concentrations, while lifestyle, socioeconomic, dietary, genetic, or geographic factors could influence both vitamin D status and disease risk. This combination of confounding and reverse causation is precisely what makes conventional observational evidence difficult to interpret.
Mendelian Randomization as a Natural Experiment
To move from correlation toward causal inference, the investigators used Mendelian randomization, or MR. This method employs genetic variants that influence an exposure—in this case circulating 25OHD—as instrumental variables. Because alleles are allocated at meiosis before disease develops, their distribution is generally less affected by the behavioral, socioeconomic, and clinical factors that confound observational studies. A vitamin-D-lowering allele can therefore function as a proxy for lifelong exposure to slightly lower vitamin D concentrations. If several independent vitamin-D-lowering variants also predict greater MS susceptibility, and if they influence MS principally through vitamin D rather than unrelated biological pathways, the result supports a causal interpretation. MR is sometimes described as analogous to a randomized trial, although the analogy is incomplete: genetic studies examine small differences operating across an entire lifetime, whereas clinical trials administer a specific intervention, at a defined dose, during a limited period. The approach was especially relevant to MS because biological onset may precede clinical diagnosis by years, making it difficult to determine vitamin D status at the truly relevant etiological stage.
Four Genetic Instruments Across the Vitamin D Pathway
The analysis began with the SUNLIGHT genome-wide association study, which included 33,996 individuals of European ancestry. Four independent single-nucleotide polymorphisms reached genome-wide significance for circulating 25OHD: rs2282679 in GC, rs12785878 near DHCR7, rs10741657 near CYP2R1, and rs6013897 in CYP24A1. These loci were biologically compelling because they represented distinct parts of vitamin D physiology. GC encodes the principal vitamin D-binding protein; DHCR7 affects the availability of 7-dehydrocholesterol, the substrate used for cutaneous vitamin D production; CYP2R1 participates in hepatic 25-hydroxylation; and CYP24A1 helps inactivate vitamin D metabolites. The investigators estimated the variants’ effects on 25OHD in 2,347 participants from the Canadian Multicentre Osteoporosis Study and evaluated their associations with MS using data from genetic consortia containing as many as 14,498 cases and 24,091 controls. Importantly, the four variants were not in meaningful linkage disequilibrium, reducing the likelihood that they were simply tagging the same genetic signal. Together, their vitamin-D-decreasing alleles formed an instrument representing genetically determined, lifelong differences in vitamin D status.
Genetically Lower Vitamin D Predicted Substantially Higher MS Risk
The genetic score behaved as expected: participants carrying more 25OHD-decreasing alleles had progressively lower circulating 25OHD concentrations. In the Canadian cohort, the allele score was strongly associated with lower 25OHD, with an F-statistic of 49.7 and a trend-test p-value of 3.3×10−19 3.3×10−19, indicating that the combined instrument was not weak. In the principal fixed-effects MR analysis, each one-standard-deviation decrease in natural-log-transformed 25OHD was associated with approximately twice the odds of MS: odds ratio 2.02, 95% confidence interval 1.65–2.46, and p-value 7.72×10−12 7.72×10−12. A random-effects analysis produced a similar estimate, with an odds ratio of 2.07. The estimated heterogeneity was moderately high at 63%, although its confidence interval was wide because only four variants were included. The central finding was therefore not that a particular vitamin D measurement predicts MS, but that several inherited variants producing lifelong reductions in vitamin D status collectively predict markedly higher susceptibility. That distinction substantially reduces the possibility that low vitamin D is merely a consequence of reduced mobility, less outdoor activity, or undiagnosed disease.
Stress-Testing the Causal Interpretation
A valid MR analysis requires more than strong genetic associations: the variants should affect the outcome through the exposure of interest rather than through independent pathways, a problem known as horizontal pleiotropy. The authors therefore conducted multiple sensitivity analyses. The DHCR7 variant was potentially affected by population stratification because its frequency varied geographically, while the GC locus raised concern because vitamin D-binding protein may have immunological activities beyond transporting vitamin D. Removing the DHCR7 variant weakened but did not eliminate the association: the fixed-effects odds ratio remained 1.72, with a 95% confidence interval of 1.34–2.21. Excluding the GC variant likewise left the main result essentially intact. The authors also divided the instruments into variants related to 25OHD synthesis and variants related to metabolism. Both groups independently predicted higher MS risk: the synthesis estimate was 2.08 and the metabolism estimate was 1.86 per standard-deviation decrease in log-transformed 25OHD. Convergence across distinct biological pathways makes it less likely that one anomalous locus entirely generated the association, although no sensitivity analysis can completely exclude residual pleiotropy.
From Genetic Effect Sizes to Biological and Clinical Meaning
The pathway diagram on page 8 helps explain why the selected loci form a coherent instrument: they span vitamin D production, hepatic conversion, transport, and metabolic inactivation rather than clustering around a single molecular process. This biological distribution strengthens the argument that circulating vitamin D is the shared mediator. The authors also translated a one-standard-deviation increase in log-transformed 25OHD into approximate concentration changes. Starting at 25 nmol/L, the calculated corresponding level was 36.86 nmol/L; starting at 50 nmol/L, it was 73.72 nmol/L; and starting at 75 nmol/L, it was 110.6 nmol/L. Under the MR model, these changes corresponded to approximately 50% lower odds of MS. These values should not be interpreted as treatment targets or evidence that raising an individual’s vitamin D to a particular concentration will halve their risk. They are model-based translations of a genetically estimated lifetime association. Supplementation produces a comparatively abrupt biochemical change, and its effect may depend on age, baseline deficiency, immune-developmental timing, genotype, adherence, dose, and whether treatment begins before the pathological process has started.
A Strong Causal Signal, but Not Yet a Prescription
The study provides persuasive evidence that chronically reduced vitamin D status contributes to MS susceptibility, particularly because it combines large genetic datasets, biologically distinct instruments, a strong allele score, and sensitivity analyses that preserve the main direction of effect. Nevertheless, its scope is clearly bounded. The participants were predominantly of European descent, so generalizability to other ancestries was not established. The analysis concerned the risk of developing MS and could not determine whether vitamin D modifies relapse frequency, disability accumulation, MRI activity, or treatment response after disease onset. Residual pleiotropy and compensatory developmental processes—sometimes called canalization—also remain possible, even though the authors argued that such compensation would probably attenuate rather than create the observed association. Most importantly, Mendelian randomization is not a substitute for a long-term randomized controlled trial of supplementation. The findings justify prevention studies focused on appropriate populations and biologically relevant life stages, but they do not demonstrate that high-dose vitamin D prevents or treats MS. The scientifically defensible conclusion is therefore precise: genetically lower 25OHD is associated with increased MS risk, providing a strong rationale—rather than final clinical proof—for testing vitamin D interventions prospectively.
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:
Mokry, L. E., Ross, S., Ahmad, O. S., Forgetta, V., Smith, G. D., Leong, A., ... & Richards, J. B. (2015). Vitamin D and risk of multiple sclerosis: a Mendelian randomization study. PLoS medicine, 12(8), e1001866.
