Obesity and Multiple Sclerosis: A Mendelian Randomization Study

Abstract

Background: Observational studies have reported an association between obesity, as measured by elevated body mass index (BMI), in early adulthood and risk of multiple sclerosis (MS). However, bias potentially introduced by confounding and reverse causation may have influenced these findings. Therefore, we elected to perform Mendelian randomization (MR) analyses to evaluate whether genetically increased BMI is associated with an increased risk of MS.

Methods and findings: Employing a two-sample MR approach, we used summary statistics from the Genetic Investigation of Anthropometric Traits (GIANT) consortium and the International MS Genetics Consortium (IMSGC), the largest genome-wide association studies for BMI and MS, respectively (GIANT: n = 322,105; IMSGC: n = 14,498 cases and 24,091 controls). Seventy single nucleotide polymorphisms (SNPs) were genome-wide significant (p < 5 x 10-8) for BMI in GIANT (n = 322,105) and were investigated for their association with MS risk in the IMSGC. The effect of each SNP on MS was weighted by its effect on BMI, and estimates were pooled to provide a summary measure for the effect of increased BMI upon risk of MS. Our results suggest that increased BMI influences MS susceptibility, where a 1 standard deviation increase in genetically determined BMI (kg/m2) increased odds of MS by 41% (odds ratio [OR]: 1.41, 95% CI 1.20-1.66, p = 2.7 x 10-5, I2 = 0%, 95% CI 0-29). Sensitivity analyses, including MR-Egger regression, and the weighted median approach provided no evidence of pleiotropic effects. The main study limitations are that, while these sensitivity analyses reduce the possibility that pleiotropy influenced our results, residual pleiotropy is difficult to exclude entirely.

Conclusion: Genetically elevated BMI is associated with risk of MS, providing evidence for a causal role for obesity in MS etiology. While obesity has been associated with many late-life outcomes, these findings suggest an important consequence of childhood and/or early adulthood obesity.

Fig 1. Schematic representation of an MR analysis.

This diagram shows that SNPs associated with BMI were selected from the GIANT consortium. Corresponding effect estimates for these SNPs upon risk of MS were obtained from the IMSGC. Because of the randomization of alleles at meiosis, SNPs are not associated with confounding variables that may bias estimates obtained from observational studies.

Fig 2. MR-Egger regression scatterplot for BMI on MS analysis.

The red line shows the results of standard MR analysis (inverse-variance weighting [IVW]), and the blue line shows the pleiotropy-adjusted MR-Egger regression line. The estimated slope of the MR-Egger regression, expressed as an OR, was 1.35 (95% CI 0.91–2.02). The estimated MR-Egger intercept term was 0.0013 (95% CI −0.010–0.013)

Fig 3. MR-Egger regression funnel plot for BMI on MS analysis.

Each SNP’s MR estimate is plotted against its minor-allele frequency corrected association with BMI. A minor allele frequency (MAF) correction proportional to the SNP-BMI standard error is used since a low-frequency allele is likely to be measured with low precision. Similar to the use of funnel plots in the meta-analysis literature, this plot can be used for visual inspection of symmetry, where any deviation can be suggestive of pleiotropy. We note that our plot appears generally symmetrical.

Fig 4. Forest plot of MR estimates.

Forest plots of all main and sensitivity analyses. ORs for MS are reported for a 1 SD increase in BMI. Estimates were obtained using a fixed effects model. MAF refers to minor allele frequency.