Identifying environmental risk factors for inflammatory bowel diseases: a Mendelian randomization study

Inflammatory bowel diseases (IBD), including Crohn’s disease (CD) and ulcerative colitis (UC), are chronic disorders driven by dysregulated immune responses to intestinal dysbiosis in genetically susceptible individuals. The incidence is stabilizing in developed regions but rising in transitioning economies, implicating westernized lifestyle factors such as smoking, obesity, physical inactivity, and diet. Observational studies and umbrella reviews have reported associations between these environmental exposures and IBD risk, but findings are often conflicting and may be confounded or affected by reverse causation (e.g., divergent effects of smoking on CD vs UC; inconsistent associations for BMI and folate). To assess causality more robustly, this study applies Mendelian randomization (MR), leveraging the random allocation of genetic variants associated with putative risk factors to estimate their effects on IBD and its subtypes. The aim is to test whether lifestyle and nutrient exposures previously linked to IBD in meta-analyses causally influence IBD risk using a two-sample MR framework with large-scale GWAS data.

Prior epidemiologic research links environmental factors to IBD, with smoking consistently associated with higher CD risk and lower UC risk, obesity and western-type diet implicated in dysbiosis and IBD pathogenesis, and physical activity suggested as protective. An umbrella review identified risk and protective factors with varying evidence strength; however, subgroup analyses revealed heterogeneity and potential confounding (e.g., folate, smoking). Previous MR work on vitamin D with smaller case numbers (~2,000 cases) did not support a protective effect. Observational studies of fats suggest omega-3 intake may reduce, and omega-6 increase, IBD risk, though therapeutic efficacy of omega-3 supplementation remains unclear. Given these inconsistencies, MR offers a strategy to infer causality while minimizing confounding and reverse causation.

Design: Two-sample Mendelian randomization (MR) using summary-level GWAS data. Exposure instruments: Genome-wide significant (P < 5×10^-8), largely European-ancestry SNPs identified from large GWAS/meta-analyses, primarily including UK Biobank and major consortia. Uncorrelated SNPs were selected (LD R^2 < 0.01); ambiguous A/T or C/G SNPs were replaced with high-LD proxies (R^2 > 0.8) using proxysnps (European panel). Exposure effect sizes were standardized to SD units where necessary. Exposures and instruments (n SNPs; approximate explained variance): smoking status, ever vs never (331; 2.3%); cigarettes per day (44; 1.1%); body-mass index (BMI) (816; 6.0%); waist-to-hip ratio (403; 3.0%); body fat percentage (378; 3.5%); overall physical activity (accelerometry; 2; 0.08%); vitamin D (59; 3.4%); vitamin B9/folate (2; 1.0%); vitamin B12 (10; 6.3%); omega-3 fatty acids (4; 2.4%); omega-6 fatty acids (8; 4.6%); monounsaturated fatty acids (4; 2.4%); total fatty acids (8; 3.7%). Outcomes: IBD, CD, UC from the largest European-ancestry GWAS meta-analysis (International IBD Genetics Consortium and UK IBD Genetics Consortium): 25,042 IBD cases (12,194 CD; 12,366 UC) and 34,915 controls. Primary MR analysis: Likelihood-based MR combining variant-specific Wald ratios (βoutcome/βexposure), assuming linear exposure–outcome relation and bivariate normality of genetic effects. Heterogeneity between CD and UC effects assessed via Q statistic (Pdisease het). Multiple testing control via Bonferroni correction across 13 exposures (α = 0.0038). Sensitivity analyses: MR-PRESSO to detect horizontal pleiotropy and outliers (PGlobal; PDistortion for bias in estimates), MR-Egger intercept for directional pleiotropy, weighted median MR, and modal-based MR for robustness to invalid instruments. Visualization via scatter plots of SNP-exposure vs SNP-outcome effects. Power: A priori power calculations indicated adequate power for most exposures to detect OR ≥ 1.20; physical activity instrument had limited power (minimum detectable OR ~2.27 for IBD) due to very low variance explained (0.08%). Software: R (v3.4.1), TwoSampleMR (v0.4.18), MRPRESSO (v1.0), ggplot2 (v3.1.0).

• Smoking: Ever vs never smoking showed a modest association with CD (OR 1.13, 95% CI 1.02–1.25; P=0.02) but not UC (OR 0.99, 95% CI 0.90–1.10; P=0.88; Pdisease het=0.08). Smoking intensity (cigarettes/day) was not significantly associated with IBD (OR 1.04, 95% CI 0.88–1.22), CD (OR 1.17, 95% CI 0.94–1.44), or UC (OR 0.90, 95% CI 0.73–1.10). Sensitivity analyses detected pleiotropy for smoking intensity (MR-Egger intercept P=0.004 for CD), with estimates around null in other robust methods.
• Adiposity and body composition: Per SD increases were associated as follows. BMI (SD 4.6 kg/m^2): CD OR 1.11 (95% CI 1.02–1.20; P=0.02); UC OR 0.85 (95% CI 0.78–0.92; P=5.2×10^-5); heterogeneity between CD and UC Pdisease het=5.3×10^-6. Waist-to-hip ratio (SD 0.1): CD OR 1.22 (95% CI 1.09–1.37; P=6.0×10^-4); UC OR 0.90 (95% CI 0.81–1.01; P=0.07); Pdisease het=2.1×10^-4. Body fat percentage (SD 6.6%): IBD OR 1.26 (95% CI 1.14–1.39; P=3.8×10^-6); CD OR 1.50 (95% CI 1.32–1.71; P=3.0×10^-10); UC OR 1.11 (95% CI 0.98–1.25; P=0.11); Pdisease het=7.3×10^-4. Horizontal pleiotropy/outliers were detected but did not distort estimates (MR-PRESSO PDistortion > 0.08); robust MR methods yielded similar results.
• Physical activity: Higher objectively measured activity (per SD ~8.1 mg acceleration) associated with lower IBD risk (OR 0.51, 95% CI 0.27–0.98; P=0.04), with no CD-UC heterogeneity (Pdisease het=0.51). However, the instrument comprised only 2 SNPs with very low explained variance, indicating limited power and potential instability.
• Vitamins: Vitamin D (SD 34.7 nmol/L) showed no association with IBD (OR 0.96, 95% CI 0.87–1.06; P=0.38; Pdisease het=0.85). Vitamin B9/folate (SD 8.44 nmol/L) suggested a non-significant protective trend for IBD (OR 0.83, 95% CI 0.67–1.03; P=0.09). Vitamin B12 (SD 150.11 pmol/L) had marginal associations (IBD OR 1.07, P=0.06; CD OR 1.10, P=0.04) but MR-PRESSO identified pleiotropy and outlier influence (PDistortion=0.02 for CD), and other sensitivity methods yielded null effects (OR ~0.92–1.09).
• Fatty acids: Omega-3 fatty acids (per SD ~0.20 mmol/L) were inversely associated with IBD (OR 0.78, 95% CI 0.68–0.88; P=1.3×10^-4) with stronger effect for CD (OR 0.67, 95% CI 0.56–0.79; P=2.3×10^-6) than UC (OR 0.88, 95% CI 0.75–1.03; P=0.11; Pdisease het=0.02). Omega-6, monounsaturated, and total fatty acids showed no consistent associations with IBD overall (e.g., omega-6 IBD OR 0.96, 95% CI 0.88–1.05). Sensitivity analyses did not indicate problematic pleiotropy for fatty acids.

Using a large two-sample MR framework, this study addresses whether key lifestyle and nutrient exposures causally influence IBD risk. Results do not support a robust causal role of smoking in IBD etiology, contrasting with observational findings suggesting increased CD risk and decreased UC risk in smokers. The MR findings imply that previously reported associations may be confounded (e.g., by adiposity) or subject to reverse causation. In contrast, higher adiposity, especially greater body fat percentage, shows a strong causal association with increased CD risk and contributes to IBD risk overall. The divergent associations of BMI with CD (positive) and UC (negative) align with cohort-based observations and suggest reverse causation may explain inverse associations seen in cross-sectional designs. Omega-3 fatty acids exhibit a protective causal effect for CD, consistent with anti-inflammatory and pro-resolving roles in intestinal inflammation, while other fatty acid classes do not show clear effects. Physical activity appears protective for IBD, but limited genetic instrument strength tempers confidence. Overall, the findings refine causal inference on environmental risk factors for IBD, highlighting adiposity and omega-3 levels as potential targets for prevention, and challenging the presumed causal role of smoking.

This MR study finds no robust causal effect of smoking on IBD or its subtypes, supports a causal risk effect of higher body fat proportion (particularly for CD), and identifies higher circulating omega-3 fatty acids as protective for CD. These results help clarify causal contributions of modifiable factors to IBD etiology and support preventive strategies focused on healthy body composition, physical activity, and dietary patterns that increase omega-3 intake. Future research should develop stronger genetic instruments for physical activity and additional environmental exposures (e.g., perinatal factors, medications, surgeries), and identify reliable microbiome-related instruments to evaluate causal roles of microbial traits. Larger and more diverse ancestry GWAS will improve generalizability and instrument strength, and further work should continue to assess and mitigate pleiotropy in MR analyses.

Key limitations include: (1) weak instruments for some exposures, particularly physical activity (2 SNPs explaining ~0.08% variance), limiting power and stability; (2) unavailability of valid genetic instruments for many relevant environmental factors (e.g., living environment, prenatal/postnatal factors, surgeries, drugs, detailed microbiome traits), restricting the scope of analyses; (3) potential violations of MR assumptions due to horizontal or directional pleiotropy, although extensive sensitivity analyses (MR-PRESSO, MR-Egger, weighted median, mode-based) were applied and generally did not indicate biased estimates; (4) reliance on European-ancestry GWAS for instruments and outcomes, which may limit generalizability to other populations; and (5) assumption of linearity between exposures and outcomes inherent to the primary MR method.