The Causal Effect of Obesity on the Risk of 15 Autoimmune Diseases: A Mendelian Randomization Study

Obesity represents a pressing global health concern, exhibiting a high prevalence and contributing to increased mortality rates [1]. The incidence of obesity has experienced an alarming surgeon a global scale, reaching pandemic proportions in recent decades [2]. Estimates from 2015 indicated that approximately 609 million adults, accounting for around 39% of the world’s population, were either overweight or obese [3]. Disturbingly elevated body mass index (BMI) alone was responsible for 4 million deaths in that same year [4]. Extensive population-based studies have identified obesity as a risk factor for various autoimmune diseases, including multiple sclerosis [5], type 1 diabetes [6], rheumatoid arthritis [7], and psoriasis [8]. However, limited and inconsistent data exist regarding the association between obesity and other autoimmune diseases [9]. Most available evidence stems from observational studies, which are susceptible to confounding factors and reverse causality, leaving the causal link between obesity and autoimmune diseases uncertain. Therefore, establishing a causal relationship between obesity and autoimmune diseases is crucial as it can provide valuable insights for informing future public policies and clinical interventions.

Mendelian randomization (MR) is an influential epidemiological technique that enhances the strength of causal inference when examining associations between exposures and outcomes [10]. Unlike traditional observational studies, MR minimizes the impact of confounding factors as genetic variants are randomly inherited during conception and remain unaffected by environmental or self-adopted factors that often confound other study designs. Additionally, MR addresses concerns of reverse causality as genetic variants remain constant throughout an individual’s lifespan and are not influenced by the onset or progression of the disease under investigation [11]. Previous MR studies have explored the connections between obesity and autoimmune diseases [12‒17]. However, a comprehensive investigation examining the influence of obesity on a wide range of autoimmune diseases has yet to be conducted. In this study, we employed a two-sample MR analysis to comprehensively investigate the causal effects of obesity on the risk of 15 autoimmune diseases.

In this two-sample MR study, we aimed to examine the causal effects of obesity on the risk of 15 autoimmune diseases. Our primary analysis focused on BMI, while waist circumference, hip circumference, and waist-to-hip ratio were considered complementary analyses (Fig. 1). To conduct this study, we utilized publicly available summary-level data from various sources, including genome-wide association studies (GWAS), the FinnGen study, the UK Biobank study, and other large consortia (online suppl. Table. 1; for all online suppl. material, see https://doi.org/10.1159/000534468). Using data from separate sources ensured that the study populations did not overlap. We independently estimated the associations of genetically predicted obesity measures (BMI, waist circumference, hip circumference, and waist-to-hip) with each autoimmune disease in the FinnGen and UK Biobank studies. Finally, we combined the estimates using a meta-analysis with a fixed-effects model. All the studies included in this research received approval from the corresponding Institutional Review Boards and Ethical Committees, and all participants provided informed consent through signed consent forms.

We obtained single-nucleotide polymorphisms (SNPs) associated with obesity measures (BMI, waist circumference, hip circumference, and waist-to-hip ratio) from the Genetic Investigation of Anthropometric Traits (GIANT) consortium’s GWAS data. SNPs were extracted based on their genome-wide significance levels (p < 5 × 10⁻⁸). The BMI GWAS data included information from 234,069 individuals of European descent and incorporated covariates such as sex, age, age squared, and principal components [18]. The GWAS data for waist circumference, hip circumference, and waist-to-hip ratio involved 210,088 European participants and adjusted for age, age square, and study-specific covariates as needed [19] (online suppl. Table 1). We assessed the linkage disequilibrium among SNPs associated with each obesity measure using PLINK clumping. SNPs with a linkage disequilibrium of r²>0.001 and clump distance <10,000 kb were excluded from the analysis.

We collected genetic associations for 15 autoimmune diseases, namely, asthma, celiac disease, Crohn’s disease, Graves’ disease, hypothyroidism, irritable bowel syndrome, multiple sclerosis, myasthenia gravis, primary biliary cirrhosis, primary sclerosing cholangitis, psoriasis, rheumatoid arthritis, systemic lupus erythematosus, type 1 diabetes, and ulcerative colitis, from two sources: the UK Biobank study and the FinnGen study. The UK Biobank study is a large-scale prospective cohort study that enrolled over 500,000 individuals aged 40 and above between 2006 and 2010. Autoimmune diseases were diagnosed based on the International Classification of Diseases 9th Revision (ICD-9) and ICD-10 codes, surgical records, and self-reported information. For the MR analysis, we utilized GWAS data from the Lee Laboratory for Statistical Genetics and Data Science (Seoul National University, Seoul, Republic of Korea; https://www.leelabsg.org/resources). The analysis was adjusted for birth year, sex, genotyping batch, and the first four principal components. The FinnGen consortium incorporated health and genetic data from Finnish health registries. GWAS analyses for each trait were conducted using the R8 data release of FinnGen, adjusting for sex, age, genotyping batch, and the first ten genetic principal components. The confirmation of autoimmune diseases was based on ICD-8, ICD-9, and ICD-10 codes, surgical records, and medication purchase codes. Additional information regarding the outcome data sources and definitions can be found in online supplementary Tables 1 and 2.

We acquired the GWAS summary statistics for smoking and drinking from the alcohol and nicotine use of GWAS and Sequencing Consortium [20]. The smoking data encompassed 249,752 individuals of European descent, while the drinking data comprised 335,394 European participants. Smoking status was determined based on the average number of cigarettes smoked daily while drinking status was defined by the average weekly alcohol consumption reported. Covariates included in the analysis were age, sex, age-sex interaction, and the first ten principal genetic components. The GWAS analysis implemented genomic control to account for potential population stratification and confounding factors.

For the primary MR analysis, we employed the random-effects multiplicative inverse-variance weighted method to estimate the associations between genetically predicted obesity measures (BMI, waist circumference, hip circumference, and waist-to-hip ratio) and the risk of autoimmune diseases. We combined outcome data from various sources using the fixed-effects meta-analysis approach to generate MR estimates. To assess heterogeneity among the outcome data from different sources, we utilized the I² statistics. Sensitivity analyses were conducted to evaluate the robustness of the results and detect possible horizontal pleiotropy. We employed the weighted median method, MR-Egger regression, and MR Pleiotropy Residual Sum and Outlier (MR-PRESSO) method for these analyses. The weighted median method provides valid MR estimates when more than 50% of the weights come from valid SNPs. The MR-Egger regression includes an intercept test to detect and correct horizontal pleiotropy, providing estimates after adjusting for pleiotropic effects. The MR-PRESSO method and MR-PRESSO global test were used to identify and correct for possible outliers and evaluate horizontal pleiotropy, resulting from heterogeneity among SNPs estimates. The Cochran Q test examined heterogeneity among SNP estimates for each MR association. We assessed imbalanced horizontal pleiotropy that may distort causal inference using all three heterogeneity tests and the MR-Egger intercept test. Additionally, multivariate MR analyses were conducted, adjusting for established risk factors for autoimmune diseases, such as drinking and smoking [21]. To assess the instrument strength, we estimated the F statistic, with an F statistic >10 indicating a sufficiently powerful instrument. To correct for multiple testing of the 15 autoimmune diseases, we applied the Benjamini-Hochberg method to estimate the false discovery rate. Associations were considered suggestive if the nominal p value <0.05, but the Benjamini-Hochberg adjusted p value >0.05. Conversely, associations were deemed significant if the Benjamini-Hochberg adjusted p value <0.05. All analyses were conducted using the “TwoSampleMR” and “MRPRESSO” packages in R 4.2.2.

In this study, we utilized genetic instruments for the obesity measuring indices. Specifically, we employed 68, 52, 41, and 29 conditionally independent SNPs for BMI, hip circumference, waist circumference, and waist-to-hip ratio, respectively. These genetic instruments explained 2.3%, 1.8%, 1.4%, and 0.8% of the variance in BMI, hip circumference, waist circumference, and waist-to-hip ratio, respectively. To assess instrument validity, we calculated the F statistics for each SNP, and all of them exceeded the empirical threshold of 10, indicating sufficient instrument strength. For detailed information on the specific SNPs used in this study, please refer to online supplementary Tables 3 and 4.

The meta-analysis of all outcome sources (Fig. 2) revealed a significant association between genetically predicted BMI and an increased risk of five autoimmune diseases. These associations remained significant even after correcting for multiple testing (online suppl. Tables 5, 6). Specifically, for every one-unit increase in the odds ratio (OR) of BMI, the combined OR estimates from FinnGen and UK Biobank were as follows: 1.28 (95% CI, 1.18–1.38; p < 0.001) for asthma, 1.37 (95% CI, 1.24–1.51; p < 0.001) for hypothyroidism, 1.52 (95% CI, 1.27–1.83; p < 0.001) for psoriasis, 1.22 (95% CI, 1.06–1.40; p = 0.005) for rheumatoid arthritis, and 1.55 (95% CI, 1.32–1.83; p < 0.001) for type 1 diabetes. However, there was no strong association observed between genetically predicted BMI and the risk of celiac disease, Crohn’s disease, Graves’ disease, irritable bowel syndrome, multiple sclerosis, myasthenia gravis, primary biliary cirrhosis, primary sclerosing cholangitis, systemic lupus erythematosus, and ulcerative colitis (Fig. 2).

The sensitivity analyses yielded consistent results regarding the direction of association between genetically predicted BMI and the risk of autoimmune diseases (online suppl. Table 7). Furthermore, most autoimmune diseases showed no significant heterogeneity across the analyzed SNPs (online suppl. Table 7). The MR-Egger intercept analysis indicated the absence of horizontal pleiotropy (online suppl. Table 7). Although MR-PRESSO, the associations and significance detected one to two outliers persisted even after removing these outlier SNPs (online suppl. Table 7). Except for rheumatoid arthritis, the observed associations remained stable in the multivariable MR analysis, which adjusted for genetically predicted drinking and smoking (online suppl. Table 8).

Genetically predicted levels of hip circumference, waist circumference, and waist-to-hip ratio were associated with the risk of several autoimmune diseases. Specifically, after correcting for multiple testing, a 1 unit increase in OR for hip circumference was associated with a greater risk of asthma (OR, 1.17; 95% CI, 1.07–1.27; p = 0.001), Crohn’s disease (OR, 1.59; 95% CI, 1.17–2.17; p = 0.003), hypothyroidism (OR, 1.40, 95% CI 1.27–1.54; p < 0.001), psoriasis (OR, 1.30; 95% CI, 1.06–1.58; p = 0.011), rheumatoid arthritis (OR, 1.41; 95% CI, 1.22–1.62; p < 0.001), and type 1 diabetes (OR, 1.43; 95% CI, 1.17–1.75; p = 0.001). Additionally, there were suggestive associations with Graves’ disease (OR, 1.36; 95% CI, 1.03–1.78; p = 0.029) (Fig. 3). Furthermore, after correcting for multiple testing, a 1 unit increase in OR for waist circumference was associated with decreased risk of asthma (OR, 1.26; 95% CI, 1.13–1.39; p < 0.001), Graves’ disease (OR, 1.72; 95% CI, 1.23–2.41; p = 0.002), hypothyroidism (OR, 1.47, 95% CI 1.29–1.68; p < 0.001), psoriasis (OR, 1.59; 95% CI, 1.24–2.04; p < 0.001), rheumatoid arthritis (OR, 1.25; 95% CI, 1.06–1.47; p = 0.007), and type 1 diabetes (OR, 1.97; 95% CI, 1.55–2.49; p < 0.001). As for waist-to-hip ratio, a 1 unit increase in OR was associated with an increased risk of type 1 diabetes (OR, 1.58; 95% CI, 1.18–2.11; p = 0.002), while there was a suggestive association with an increased risk of asthma (OR, 1.20; 95% CI, 1.04–1.38; p = 0.015). Notably, sensitivity analyses confirmed these associations, and most outcomes showed no heterogeneity or horizontal pleiotropy (online suppl. Tables 9–11).

We conducted a comprehensive study using MR to investigate the causal relationship between different measures of obesity (such as BMI, hip circumference, waist circumference, and waist-to-hip ratio) and 15 autoimmune diseases. The findings, summarized in Figure 3, reveal that genetically predicted BMI is associated with an increased risk of asthma, hypothyroidism, psoriasis, rheumatoid arthritis, and type 1 diabetes. These associations were independent of genetic susceptibility to drinking and smoking, except for rheumatoid arthritis. Additionally, we observed significant or suggestive associations between genetically predicted waist circumference, hip circumference, and waist-to-hip ratio with certain autoimmune diseases.

Our MR investigation confirms and expands upon the results of previous observational studies, providing further evidence of a possible association between BMI and the risk of several autoimmune diseases, including asthma [22], hypothyroidism [23], psoriasis [8], rheumatoid arthritis [7], and type 1 diabetes [6]. Similarly, our study demonstrates that genetically predicted BMI may be positively associated with an increased risk of asthma [12], psoriasis [16], rheumatoid arthritis [14], and type 1 diabetes [24]. Notably, we found that the association between genetically predicted BMI and the risk of rheumatoid arthritis lost significance when considering drinking and smoking, indicating potential bias from the pleiotropic effects of these factors. Furthermore, our findings provide novel evidence of a possible positive association between genetically predicted BMI and the risk of hypothyroidism. Epidemiological studies suggest that the prevalence of hypothyroidism in obese patients (newly diagnosed or treated) is 14.0% [23]. Consequently, the European Society of Endocrinology recommends thyroid function screening in all obese patients due to this population’s high incidence of hypothyroidism [25]. Our results regarding the new association between BMI and hypothyroidism support early screening for hypothyroidism in obese individuals. Moreover, we identified a possible positive association between the genetically predicted hip circumference and waist circumference with the risk of several autoimmune diseases, including asthma, hypothyroidism, psoriasis, rheumatoid arthritis, and type 1 diabetes, highlighting the relationship between obesity and these diseases from different perspectives.

Although we did not include Crohn’s disease in the association between genetically predicted BMI and autoimmune disease analysis, previous studies have provided inconclusive evidence. A meta-analysis based on observational studies suggested an association between obesity and an increased risk of developing Crohn’s disease [26]. However, another prospective cohort study found no association between obesity, measured by BMI, and the progression of Crohn’s disease [27]. As the association between obesity and Crohn’s disease has mainly been derived from observational studies, the true causal relationship remains uncertain. While we did not find a significant association between genetically predicted BMI and Crohn’s disease, the point estimates (OR = 1.29) were higher than 1. Additionally, we observed that genetically predicted hip circumference was associated with an increased risk of Crohn’s disease. Given the inconsistent results and the nonsignificant association between genetically predicted BMI and Crohn’s disease, further investigation is warranted.

Observational studies have shown that being overweight during childhood and adolescence is positively associated with the risk of multiple sclerosis, increasing the risk as BMI increases [28]. However, current studies have not found a strong association between genetically predicted BMI and the risk of multiple sclerosis. Nonetheless, a previous MR analysis of 14,498 multiple sclerosis cases revealed a positive association between genetically predicted BMI and the risk of multiple sclerosis, suggesting that our negative finding may be due to insufficient power, resulting from the relatively small sample size [29].

Several potential mechanisms may influence the association between obesity and autoimmune diseases. Excess body weight can lead to the over-activation of intracellular nutrient and energy-sensing pathways, resulting in a metabolic overload of peripheral tissues involved in immune responses [30]. Adipose tissue also secretes inflammatory cytokines, such as interleukin-1 (IL-1), IL-6, IL-17, tumor necrosis factor-alpha, and interferon-gamma, as well as leptin, which may increase the risk of peripheral tissue damage and autoimmunity [31]. Additionally, it has been proposed that a Western diet can alter the gut microbiome, leading to significant modulation of immune responses outside the gut and increasing the risk of autoimmune diseases [32].

This study has several notable strengths, including its MR study design, which helps reduce confounding and reverse causality bias, enabling more accurate causal inferences regarding the association between obesity and autoimmune diseases. Using summary-level data from a large genetic study of a European population reduces the potential for demographic bias to influence the results. Furthermore, the associations were estimated using independent data sources combined through meta-analysis, ensuring statistical efficacy and robust findings.

However, this study also has limitations. First, there may be other causal pathways, known as horizontal pleiotropy, through which obesity-related SNPs affect autoimmune diseases. Although we minimized the impact of pleiotropic effects by not detecting any signs of horizontal pleiotropy in the MR-Egger analysis, obtaining consistent results across various sensitivity analyses, and observing robust associations in multivariate MR analyses with mutual adjustment, other causal pathways cannot be completely ruled out. Second, while combining data from different sources increases the statistical power, weak associations should still be interpreted cautiously due to the rarity of certain outcomes, such as myasthenia gravis and primary biliary cirrhosis. Additionally, the power calculation was limited as the variance of BMI phenotypes explained by SNPs could not be precisely estimated from pooled-level data. Finally, the findings of this analysis are limited to European populations, and the generalizability of these associations to other populations requires further investigation.

In summary, this study suggests that obesity may be associated with an increased risk of several autoimmune diseases, such as asthma, hypothyroidism, psoriasis, rheumatoid arthritis, and type 1 diabetes. The genetic evidence obtained from our research could serve as a basis for implementing focused lifestyle interventions aimed at individuals with obesity, particularly in preventing autoimmune diseases.

The authors thank the researchers and participants of the UK Biobank and FinnGen consortium.

This study protocol was reviewed and approved by Ethics Committees at each of the participating site. Signed informed consent was obtained from all participants. This full list of participating site and Ethics Committees can be found at: http://portals.broadinstitute.org/collaboration/giant/index.php/GIANT_consortium_data_files; https://conservancy.umn.edu/handle/11299/201564; https://www.finngen.fi/fi; https://www.leelabsg.org/resources.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work was supported by grants from the Natural Science Foundation of Anhui Province (2008085MH244), and Anhui Medical University 2021 Clinical and Pre-disciplinary Co-Construction (2021lcxk032). No funding bodies had any role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conceptualization: Xunliang Li, Jie Zhu, Haifeng Pan, Deguang Wang. Methodology: Xunliang Li, Jie Zhu. Formal analysis and investigation: Xunliang Li, Jie Zhu, Wenman Zhao, Rui Shi, Yuyu Zhu, Zhijuan Wang. Funding acquisition and Supervision: Haifeng Pan, Deguang Wang. Xunliang Li and Jie Zhu contributed equally to this work. All authors read and approved the final manuscript.

The data analyzed in the current study are publicly available GWAS abstract-level data. Specific information and links can be found in online supplementary Table 1. Further inquiries can be directed to the corresponding author.