Obesity and Multiple Sclerosis Severity: A Mendelian Randomization Study

Fatema Alzamanan; Yuan Ding; Adil Harroud

doi:10.1002/ana.27112

. 2024 Oct 21;97(1):90–94. doi: 10.1002/ana.27112

Obesity and Multiple Sclerosis Severity: A Mendelian Randomization Study

Fatema Alzamanan ^1,^2,^†, Yuan Ding ^1,^3,^†, Adil Harroud ^1,^2,^3,^✉

PMCID: PMC11683183 PMID: 39431322

Abstract

Obesity is implicated in the development of multiple sclerosis (MS), but its effect on disability is less well‐established. This study examined the effects of various obesity measures on MS severity in 12,584 MS cases, using Mendelian randomization to mitigate confounding. Results showed a significant association between higher genetically‐determined body mass index (N = 806,834) and increased MS severity (P = 0.02). This finding was supported by additional measures of general obesity but not adiposity distribution. The convergence of this genetic evidence with prior observational studies strengthens the association between obesity and adverse long‐term disability in MS, suggesting weight management as a potential therapeutic strategy. ANN NEUROL 2025;97:90–94

Despite therapeutic advancements, people with multiple sclerosis (MS) continue to accumulate disability across disease stages. ¹ The specific factors driving this disability progression, particularly in the absence of relapses, remain elusive and likely vary among patients. ² In addition to disease‐specific factors, concurrent conditions such as obesity have emerged as potentially key factors influencing outcomes. ³

Obesity is increasingly implicated in various autoimmune diseases, ⁴ and epidemiological evidence indicates that obesity during adolescence and early adulthood nearly doubles the risk of MS. ⁵ However, the role of obesity in MS prognosis is less well established and findings are more variable. ⁶ , ⁷ This is further complicated by the fact that obesity is intertwined with numerous socioeconomic determinants and prone to confounding. Moreover, common obesity measures like body mass index (BMI) are known to be affected by MS, ⁵ introducing a potential for reverse causality (whereby MS and its severity might alter body size).

To help overcome these limitations, we utilized Mendelian randomization (MR), a method that uses genetic variants as proxies for obesity measures to assess their influence on MS severity. While MR has provided orthogonal evidence supporting a causal role for obesity in MS development, ⁸ a lack of robust genetic associations with disease outcomes has limited similar analyses of MS severity. Recently, the largest genome‐wide association study (GWAS) of MS severity to date provided the necessary data, yet initial MR analyses did not confirm an effect of BMI. ⁹ Here, we update this MR analysis with improved statistical power, revealing a positive correlation between obesity and MS severity, and extend our investigation to include additional obesity measures.

Methods

In a 2‐sample MR design, we leveraged large‐scale genetic studies on anthropometric or physiological measures of obesity (exposures) together with the most recent GWAS of MS severity (outcome). As genetic variants related to each exposure are randomly assigned at conception independent of potential confounders, MR lessens bias from unobserved confounding. It also prevents reverse causality, as genotypes are fixed at conception and thus precede the outcome. A key assumption required for causal inference is that the genetic variants related to the exposure are not related to the outcome other than through their association with the exposure (i.e., no horizontal pleiotropy). ¹⁰ The data sources used in this study obtained informed consent from all participants; no additional ethical approval was required.

We first updated our recent MR analysis on the effect of BMI on MS severity by using a larger GWAS of BMI involving 806,834 participants ¹¹ (vs. 681,275 previously ⁹ ). Mean BMI was 27.4 (SD 4.8). ¹¹ We also considered the following additional obesity measures (Table 1): whole‐body fat mass (WBF), body fat percentage (BFP), trunk fat percentage (TFP), trunk fat mass (TFM), waist hip ratio adjusted for BMI (WHRadjBMI), and visceral adipose tissue (VAT). By including these, we aimed to provide a more comprehensive assessment of body composition, including fat distribution (WHRadjBMI, VAT) as opposed to general obesity (WBF, BFP, TFP, and TFM). Trunk fat free mass (TFFM) was used as a negative control.

TABLE 1.

Genome Wide Association Datasets Used in the Mendelian Randomization Analyses

Phenotype	Description	Population	Sample Size	Data Source	Mean F‐Statistic ^a
Body mass index	Body weight (in kilograms) divided by the squared height (in meters)	GIANT+UK Biobank	806,834	PMID: 30239722	72.9
Waist‐hip ratio adjusted for body mass index	Ratio of waist to hip circumference regressed on BMI among other covariates	GIANT+UK Biobank	694,649	PMID: 30239722	85.8
Visceral adipose tissue	Estimated using demographic, anthropometric, and bioelectrical impedance data, based on a model derived from dual‐energy X‐ray absorptiometry measurements in a subset of participants	UK Biobank	325,153	PMID: 31501611	54.5
Body fat mass	Body composition estimated by bioelectrical impedance via Tanita BC418MA	UK Biobank	354,628	Neale Lab	55.9
Body fat percentage	Bioelectrical impedance; assessment of fat percentage	UK Biobank	354,628	Neale Lab	53.4
Trunk fat mass	Bioelectrical impedance; specific to trunk region	UK Biobank	354,628	Neale Lab	55.7
Trunk fat percentage	Bioelectrical impedance; segmental reading for trunk	UK Biobank	354,628	Neale Lab	53.0
Trunk fat free mass	Bioelectrical impedance; assessment of trunk fat‐free composition	UK Biobank	354,628	Neale Lab	77.0

Open in a new tab

^{^a}

The F‐statistic for each genetic variant was calculated by dividing the squared variant‐exposure estimate by the squared corresponding standard error.

The effect on MS severity was determined using a GWAS of 12,584 people with MS (mean age and disease duration 51.7 and 18.2 years, respectively, 71.1% female). ⁹ Severity was measured using the Age‐Related Multiple Sclerosis Severity Score (ARMSS) at last follow‐up. ¹² A rank‐based inverse‐normal transformation was applied to ensure normality. To provide an interpretable effect estimate, we repeated the MR analysis using genetic associations with untransformed ARMSS scores on the same participants. All participants were of European ancestry to limit confounding from population structure. Accordingly, the European subset of the 1,000 Genomes Project Phase 3 was used to estimate linkage disequilibrium.

For each exposure, we selected genome‐wide significant (P < 5 × 10⁻⁸) and linkage disequilibrium‐independent (r ² < 0.001) variants (Tables S1–S8). For variants missing from our MS severity GWAS, we attempted to identify correlated proxies (r ² > 0.8). However, for all traits, variants were either entirely represented or absent from the reference panel.

The primary analysis was the inverse variance weighted (IVW) MR method. ¹⁰ We indirectly assessed evidence of horizontal pleiotropy using Cochran's Q heterogeneity statistic. For sensitivity analysis, we used various pleiotropy‐robust methods, including MR‐Egger, the weighted median, and the Robust Adjusted Profile Score (RAPS). ¹⁰ Additionally, we used the Bayesian Causal Analysis Using Summary Effect (CAUSE) method, ¹³ which accounts for both correlated and uncorrelated pleiotropy. Variants were filtered using a P‐threshold of 10⁻³ and clumped at an r ² of 0.01. ¹³

Given the potential influence of cigarette smoking on obesity and MS severity, we examined BMI‐associated variants for smoking phenotypes in the GWAS Catalog (>667,000 genotype–phenotype associations) and excluded these variants in a sensitivity analysis.

Results

Using the IVW method, elevated BMI was associated with higher MS severity (β = 0.08 per SD increase in BMI, P = 0.02; Fig 1), with little evidence of heterogeneity (Cochran's Q P = 0.8) and no directional pleiotropy (MR‐Egger intercept −0.001 to 0.005, Table S9). Estimates from RAPS (β = 0.09, P = 0.02) and weighted median methods (β = 0.07, P = 0.30) were concordant with the main result, although the latter was not significant. The CAUSE estimate (γ) further supported a positive association between BMI and MS severity (95% credible interval 0.02 to 0.13). In addition, model comparison using the expected log pointwise posterior density favored the causal model (δ_ELPD = −1.5), although a sharing model allowing for horizontal pleiotropy could not be excluded (one‐sided P = 0.11). After excluding smoking‐associated variants (n = 12, Table S10), the effect of BMI on MS severity remained unchanged (β = 0.09, P = 0.02). Using untransformed scores, each standard deviation increase in BMI corresponded to a 0.21 increase in ARMSS.

Association of BMI with MS severity. MR estimates for the effect of a standard deviation increase in BMI on rank‐based inverse‐normal transformed ARMSS scores. The CAUSE analysis employs a Bayesian framework and frequentist P‐values are not applicable (NA). CAUSE = Bayesian Causal Analysis Using Summary Effect; CI = confidence interval (for CAUSE, credible interval); MR = Mendelian randomization; MS = multiple sclerosis; SNVs = single‐nucleotide variants.

Additional general obesity measures (WBF, BFP, TFP, and TFM) showed a consistent positive effect on MS severity (Fig 2). In the IVW analysis, only WBF had a significant association (β = 0.08, P = 0.04). Similar to BMI, we observed no heterogeneity, a null MR‐Egger intercept, and a significant RAPS estimate (β = 0.08, P = 0.039). CAUSE revealed a comparable effect for BFP (γ = 0.08, 95% credible interval 0.003–0.16). Adiposity distribution measures had no significant association with MS severity, and as expected, no association was observed for TFFM. Results from sensitivity analyses are detailed in Table S9. Power varied among exposures, with borderline power for BMI (73%) and lower power for others (median 62%; Table S11).

Association of additional obesity measures with MS severity. MR estimates using the main analysis (inverse‐variance weighted method). BMI = body mass index; CI = confidence interval; MS = multiple sclerosis; SNVs = single‐nucleotide variants.

Discussion

This MR study provides supporting evidence that obesity is associated with increased long‐term disability in MS. The lack of heterogeneity and overall concordance in sensitivity analyses mitigate the likelihood of bias from horizontal pleiotropy and suggest a potential causal effect. These results were corroborated by associations between MS severity and both WBF and BFP. Variations in obesity measure associations may reflect biological differences (e.g., general obesity vs. adiposity distribution) but may also be partly due to limited power. This underscores the need for larger genetic studies of MS severity to more precisely dissect obesity effects, including mediation analyses with other risk factors like educational attainment.

Our findings align with a previous prospective longitudinal study involving 1,066 MS participants, which found that BMI ≥30 kg/m² at onset was associated with higher EDSS and a shorter time to EDSS 3. ⁶ Conversely, another study in 468 BENEFIT trial participants with clinically isolated syndrome found no association between baseline BMI and sustained EDSS progression. ⁷ However, obesity was associated with higher rates of MS diagnosis and relapse, a driver of early‐stage disability. ¹ Emerging research further supports obesity's impact on multiple MS disability outcomes. ¹⁴

A key contribution of the present study is the use of the MR design to mitigate the risks of confounding and reverse causality. The convergence of our genetic approach with observational evidence, each with different potential biases, strengthens the argument for a causal effect of obesity on MS severity.

The mechanisms driving the association between obesity and MS remain unclear. Obesity is characterized by low‐grade chronic inflammation, including in the CSF of people with MS, ¹⁵ and is associated with brain atrophy and dementia. ¹⁶ In the OPERA I/II studies, higher BMI also reduced the benefits of ocrelizumab compared to interferon β‐1a. ¹⁷

A limitation of this study is that genetic associations with MS severity were adjusted for but not stratified by sex, precluding the examination of sex‐differential effects of obesity. Similarly, the MR design did not allow for assessing potential interactions with specific disease‐modifying therapies. However, as patients received a range of therapies, ⁹ the observed effect is unlikely to depend on any single therapy. Pleiotropy cannot be completely excluded, and some pleiotropy‐robust methods, such as the weighted median and MR‐Egger in the BMI analysis, were not significant, although this may be explained by their lower statistical power. ¹⁸ Also, collider bias may occur when considering risk factors impacting both disease onset and progression, ¹⁹ yet this concern is shared by MR and observational studies. Last, MR estimates the average lifelong impact of a risk factor but might not fully capture time‐dependent or non‐linear effects, such as those arising from body sizes at either end of the distribution.

In conclusion, this study provides MR evidence that strengthens the association between obesity and greater long‐term disability in MS, identifying obesity as a potentially modifiable risk factor whose management could alleviate the severity of MS. To this end, the emergence of drug therapies targeting obesity ²⁰ presents a potential strategy for people with MS and co‐morbid obesity.

Author Contributions

F.A. and A.H. contributed to the conception and design of the study; F.A., Y.D. and A.H. contributed to the acquisition and analysis of data; F.A., Y.D. and A.H. contributed to drafting the text and preparing the figures.

Potential Conflicts of Interest

Nothing to report.

Supporting information

Table S1. Included variants and harmonized genetic associations for body mass index and MS severity. MS = multiple sclerosis.

Table S2. Included variants and harmonized genetic associations for body fat percentage and MS severity. MS = multiple sclerosis.

Table S3. Included variants and harmonized genetic associations for trunk fat free mass and MS severity. MS = multiple sclerosis.

Table S4. Included variants and harmonized genetic associations for trunk fat mass and MS severity. MS = multiple sclerosis.

Table S5. Included variants and harmonized genetic associations for trunk fat percentage and MS severity. MS = multiple sclerosis.

Table S6. Included variants and harmonized genetic associations for visceral adipose tissue and MS severity. MS = multiple sclerosis.

Table S7. Included variants and harmonized genetic associations for whole body fat mass and MS severity. MS = multiple sclerosis.

Table S8. Included variants and harmonized genetic associations for waist‐hip ratio adjusted for BMI and MS severity. BMI = body mass index; MS = multiple sclerosis.

Table S9. Summary of causal effect estimates for all traits and MR methods. MR = Mendelian randomization.

Table S10. BMI‐associated variants influencing smoking phenotypes. BMI = body mass index.

Table S11. Power calculations.

ANA-97-90-s001.xlsx^{(219.6KB, xlsx)}

Acknowledgments

A. Harroud is supported by a Clinical Research Scholarship (349722) and a Clinician‐Researcher Establishment Grant (358114) from the Fonds de Recherche du Québec Santé (FRQS), as well as a Future Leaders in Canadian Brain Research Grant from the Brain Canada Foundation. This research was made possible using data collected by the International Multiple Sclerosis Genetics Consortium, MultipleMS, the GIANT consortium, and the UK Biobank.

Data Availability

The data necessary to replicate the present study are publicly available. The GWAS summary statistics for MS severity can be accessed through the International Multiple Sclerosis Genetics Consortium website (https://imsgc.net/). We used publicly available data from the GIANT consortium for BMI and WHRadjBMI (https://portals.broadinstitute.org/collaboration/giant/index.php/GIANT_consortium_data_files). GWAS summary statistics for visceral adiposity were obtained from the GWAS Catalog (https://www.ebi.ac.uk/gwas/publications/31501611). Additional obesity measures were made available by the Neale lab (http://www.nealelab.is/uk-biobank).

References

1. Lublin FD, Häring DA, Ganjgahi H, et al. How patients with multiple sclerosis acquire disability. Brain 2022;145:3147–3161. [DOI] [PMC free article] [PubMed] [Google Scholar]
2. Kuhlmann T, Moccia M, Coetzee T, et al. Multiple sclerosis progression: time for a new mechanism‐driven framework. Lancet Neurol 2023;22:78–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
3. Marrie RA. Comorbidity in multiple sclerosis: implications for patient care. Nat Rev Neurol 2017;13:375–382. [DOI] [PubMed] [Google Scholar]
4. Versini M, Jeandel P‐Y, Rosenthal E, Shoenfeld Y. Obesity in autoimmune diseases: not a passive bystander. Autoimmun Rev 2014;13:981–1000. [DOI] [PubMed] [Google Scholar]
5. Munger KL, Chitnis T, Ascherio A. Body size and risk of MS in two cohorts of US women. Neurology 2009;73:1543–1550. [DOI] [PMC free article] [PubMed] [Google Scholar]
6. Lutfullin I, Eveslage M, Bittner S, et al. Association of obesity with disease outcome in multiple sclerosis. J Neurol Neurosurg Psychiatry 2023;94:57–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
7. Manuel Escobar J, Cortese M, Edan G, et al. Body mass index as a predictor of MS activity and progression among participants in BENEFIT. Mult Scler 2022;28:1277–1285. [DOI] [PMC free article] [PubMed] [Google Scholar]
8. Harroud A, Mitchell RE, Richardson TG, et al. Childhood obesity and multiple sclerosis: a Mendelian randomization study. Mult Scler 2021;27:2150–2158. [DOI] [PubMed] [Google Scholar]
9. Harroud A, Stridh P, McCauley JL, et al. Locus for severity implicates CNS resilience in progression of multiple sclerosis. Nature 2023;619:323–331. [DOI] [PMC free article] [PubMed] [Google Scholar]
10. Sanderson E, Glymour MM, Holmes MV, et al. Mendelian randomization. Nat Rev Methods Primers 2022;2:1–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
11. Pulit SL, Stoneman C, Morris AP, et al. Meta‐analysis of genome‐wide association studies for body fat distribution in 694 649 individuals of European ancestry. Hum Mol Genet 2019;28:166–174. [DOI] [PMC free article] [PubMed] [Google Scholar]
12. Manouchehrinia A, Westerlind H, Kingwell E, et al. Age related multiple sclerosis severity score: disability ranked by age. Mult Scler 2017;23:1938–1946. [DOI] [PMC free article] [PubMed] [Google Scholar]
13. Morrison J, Knoblauch N, Marcus JH, et al. Mendelian randomization accounting for correlated and uncorrelated pleiotropic effects using genome‐wide summary statistics. Nat Genet 2020;52:740–747. [DOI] [PMC free article] [PubMed] [Google Scholar]
14. Wu J, Olsson T, Hillert J, et al. Obesity negatively affects disease progression, cognitive functioning, and quality of life in people with multiple sclerosis. In: MSMilan2023–Oral presentations. Mult Scler 2023;29:4–136. [Google Scholar]
15. Stampanoni Bassi M, Iezzi E, Buttari F, et al. Obesity worsens central inflammation and disability in multiple sclerosis. Mult Scler 2020;26:1237–1246. [DOI] [PubMed] [Google Scholar]
16. Morys F, Potvin O, Zeighami Y, et al. Obesity‐associated neurodegeneration pattern mimics Alzheimer's disease in an observational cohort study. J Alzheimers Dis 2023;91:1059–1071. [DOI] [PMC free article] [PubMed] [Google Scholar]
17. Turner B, Cree BAC, Kappos L, et al. Ocrelizumab efficacy in subgroups of patients with relapsing multiple sclerosis. J Neurol 2019;266:1182–1193. [DOI] [PMC free article] [PubMed] [Google Scholar]
18. Burgess S, Foley CN, Allara E, et al. A robust and efficient method for Mendelian randomization with hundreds of genetic variants. Nat Commun 2020;11:376. [DOI] [PMC free article] [PubMed] [Google Scholar]
19. Munafò MR, Tilling K, Taylor AE, et al. Collider scope: when selection bias can substantially influence observed associations. Int J Epidemiol 2018;47:226–235. [DOI] [PMC free article] [PubMed] [Google Scholar]
20. Chakhtoura M, Haber R, Ghezzawi M, et al. Pharmacotherapy of obesity: an update on the available medications and drugs under investigation. EClinicalMedicine 2023;58:101882. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials