The phenotypic association between endometriosis and immunological conditions was investigated in the UKBB using both retrospective cohort and cross-sectional study designs (see Materials and methods). Supplementary Table S1 shows factors that were determined as potential confounders or mediators in the association analyses between endometriosis and immunological diseases. Adding factors significantly associated with both endometriosis and immunological diseases in a logistic regression model with endometriosis as exposure and any immunological disease as the outcome (see Materials and methods), none were found to be confounders or mediators that significantly affected the effect size of association (>5% change). Genetically determined ancestry and age at recruitment were included a priori as potential confounders.
In both the retrospective cohort (Table 1) and cross-sectional analyses (Supplementary Table S2), females with endometriosis versus those without had a significantly increased risk for all immunological diseases combined (HR: 1.32 (1.20–1.45); OR: 1.32 (1.25–1.39)), classic autoimmune diseases (HR: 1.41 (1.15–1.74); OR: 1.24 (1.13–1.36)), autoinflammatory diseases (HR: 1.29 (1.17–1.43); OR: 1.33 (1.26–1.41)), and mixed-pattern diseases (HR: 1.88 (1.25–2.81); OR: 1.23 (1.10–1.52)).
Immunological diseases significantly associated with endometriosis in both analyses were: rheumatoid arthritis (OR: 1.22 (1.04–1.41), P = 0.011; HR: 1.57 (1.18–2.10), P = 0.002); coeliac disease (OR: 1.35 (1.06–1.70), P = 0.011; HR: 1.99 (1.30–3.07), P = 0.002); and osteoarthritis (OR: 1.35 (1.27–1.43), P < 0.001; HR: 1.31 (1.19–1.44), P < 0.001). In addition, in the ‘gold standard’ cohort analyses, psoriasis (HR: 1.67 (1.05–2.65), P = 0.030) was significantly associated with endometriosis. Two immunological conditions significantly associated with endometriosis in cross-sectional analysis, systemic lupus erythematosus (OR: 1.62 (1.14–2.24), P = 0.005) and gout (OR: 1.66 (1.18–2.26), P = 0.002), could not be tested in a cohort study design due to insufficient case numbers (Supplementary Table S2). Overall, females with endometriosis compared to females without known endometriosis exhibited a 14% increased risk for at least having one immunological disease (OR = 1.14 (1.08–1.21)), a 21% increased risk for at least having two immunological diseases (OR = 1.21 (1.05–1.39)), and a 30% increased risk for having at least three immunological diseases (OR = 1.30 (0.92–1.78)) at any point in their lifetime (P < 0.001) (Supplementary Table S3).
When stratifying by menopausal status, gynaecological surgery (hysterectomy/oophorectomy), or HRT use, effect sizes for the association between endometriosis and overall immunological disease risk remained largely unchanged (Supplementary Table S4).
Genetic correlation between endometriosis and immunological diseases
We investigated whether the eight immunological diseases associated with endometriosis, either in cohort or cross-sectional analyses (ankylosing spondylitis, coeliac disease, inflammatory bowel disease, multiple sclerosis, osteoarthritis, psoriasis, rheumatoid arthritis, and systemic lupus erythematosus), shared a genetic basis with endometriosis through genetic correlation (rg) analyses (see Materials and methods). For this, we conducted GWAS focusing on female and combined-sex groups of European ancestry in the UKBB (Supplementary Table S5, see Materials and methods). To maximize statistical power, the combined-sex GWAS results from UKBB were meta-analysed with the largest available published GWAS summary statistics for each immune disease (Supplementary Figures S1, S2, S3, S4, S5, S6, S7 and S8).
Using these comprehensive GWAS meta-analyses, alongside the largest published endometriosis GWAS, we estimated the genetic correlations (rg) between endometriosis and these eight immunological diseases (see Materials and methods). Significant genetic sharing (P < 6.25 × 10−3) was observed for osteoarthritis (sex-combined rg = 0.29, P = 3.25 × 10−15, female-specific rg = 0.32, P = 1.76 × 10−14), rheumatoid arthritis (sex-combined rg = 0.26, P = 1.54 × 10−5, female-specific rg = 0.28, P = 0.001), and multiple sclerosis (sex-combined rg = 0.09, P = 4.00 × 10−3, female-specific rg = 0.25, P = 0.075) (Table 2). These results suggest a shared genetic basis between endometriosis and these conditions, potentially acting through common biological pathways.
Causal relationship between endometriosis and immunological diseases
To investigate whether endometriosis causally increases the risk of osteoarthritis, rheumatoid arthritis, and multiple sclerosis, MR analyses were conducted using 39 independent genetic variants associated with endometriosis (P < 5 × 10−8) as IVs. The primary results from the MR-IVW analysis (see Materials and methods) are shown in Table 3. Sensitivity analyses, including weighted median MR and MR-Egger regression, to validate the robustness of the results
For rheumatoid arthritis, the MR-IVW analysis suggested a potential causal relationship in females, with an odd ratio (OR) of 1.16 (95% CI: 1.02–1.33, P = 0.028). For osteoarthritis and multiple sclerosis, no significant causal relationship was identified in either combined-sex or female-only analyses (Table 3).
Multi-trait analysis of endometriosis and immunological diseases: osteoarthritis, rheumatoid arthritis, and multiple sclerosis
To identify additional genetic variants associated with endometriosis, we employed MTAG (Turley et al., 2018). This approach leverages shared genetic signals between endometriosis and conditions with which it showed shared genetic basis (osteoarthritis, rheumatoid arthritis, and multiple sclerosis) to enhance statistical power in detecting genetic associations. We conducted MTAG analysis for all four diseases simultaneously. Through this analysis, we identified 42 genome-wide significant (5 × 10−8) genetic variants significantly associated with endometriosis (Supplementary Table S8), 6 of which were not reported previously (Rahmioglu et al., 2023): ABHD1/2p23.3, TMEM131/2q11.2, XRCC4/5q14.2, PPP1R9A/7q21.3, XKR6/8p23.1, and TRPS1/8p23.3 (Supplementary Fig. S9a–f). These 6 novel variants are linked to genes (eQTLs, see Materials and methods) with diverse biological roles: (i) MSRA and PON2 protecting and repairing cells from oxidative stress in blood (Shin et al., 2014; Manco et al., 2021); (ii) BLK and ZAP70 encoding enzymes that belong to tyrosine kinase family with roles in cell proliferation and differentiation in particular B-cell and T-cell development and adhesion (Wang et al., 2010; Ichikawa-Tomikawa et al., 2023); (iii) ATRAID, SLC35F6, TMEM214, and XKR6 involved in apoptosis-related pathways (Li et al., 2013; Stelzer et al., 2016); and (iv) TRPS1 encoding a transcription factor that represses GATA-regulated genes involved in progesterone resistance and endometriosis progression in the pelvis (Dyson et al., 2014) (Supplementary Table S9).
In addition to endometriosis-specific results, MTAG analysis revealed 27 significant genetic variants for osteoarthritis (Supplementary Table S10), 28 for rheumatoid arthritis (Supplementary Table S11), and 64 for multiple sclerosis (Supplementary Table S12).
Functional annotation of identified genome-wide significant variants and pathway analysis
Functional analysis was performed to understand the biological roles of genetic variants identified in the MTAG analysis for endometriosis, osteoarthritis, rheumatoid arthritis, and multiple sclerosis. Using databases such as GTEx V8 (54 tissues) and eQTLGen (31 684 blood datasets), the genome-wide significant variants were mapped to genes associated with their expression (see Materials and methods). This analysis identified: 439 genes regulated by 42 endometriosis-associated variants, 379 genes regulated by 27 osteoarthritis-associated variants, 490 genes regulated by 28 rheumatoid arthritis-associated variants, and 1113 genes regulated by 64 multiple sclerosis-associated variants. Among the 439 genes linked to endometriosis variants, 192 were also regulated by a genetic variant associated with one or more of the other immune diseases, highlighting overlap in the genetic basis of these diseases. Figure 1 illustrates the shared genes regulated by variants associated with endometriosis and each of the three immune conditions: osteoarthritis, rheumatoid arthritis, and multiple sclerosis, respectively.
We conducted an unprecedentedly comprehensive investigation of the association between endometriosis and risk of a wide range of immune conditions, using a female-only study population from the large-scale UKBB dataset. Our findings reveal a significant increase in the risk of autoimmune and autoinflammatory diseases among endometriosis patients, particularly rheumatoid arthritis (HR: 1.57 (1.18–2.10), P = 0.002), coeliac disease (HR: 1.99 (1.30–3.07), P = 0.002), osteoarthritis (HR: 1.31 (1.19–1.44), P < 0.001), and psoriasis (HR: 1.67 (1.05–2.65), P = 0.030). The UKBB study population (aged 40–69 years) includes a relatively low proportion of diagnosed endometriosis cases (3% females), which is lower than the estimated population prevalence (up to 10% (Zondervan et al., 2020)). As a result, the presence of undiagnosed cases may have diluted estimated effect sizes, potentially driving associations towards the null (Zondervan et al., 2002; Shafrir et al., 2018). Despite this, our results align with findings from previous case/control and cohort studies, which suggested significant associations between endometriosis and rheumatoid arthritis (Relative Risk (RR): 1.46 (0.70–3.03), coeliac disease (RR: 1.39 (1.14–1.70)), multiple sclerosis (OR: 7.1 (4.4–11.3)) (Shigesi et al., 2019), and psoriasis (RR: 1.75 (1.10–2.78))(Harris et al., 2022). For systemic lupus erythematosus, although our cohort analysis was limited by sample size, cross-sectional data showed a significant association (OR: 1.62 (1.14–2.24)), consistent with prior longitudinal studies reporting an increased risk (HR: 2.03 (1.17–3.51)) (Harris et al., 2016). Furthermore, our findings indicate that endometriosis patients are at a significantly increased risk of suffering from multiple immunological diseases, a trend that was most pronounced for autoinflammatory conditions (one, OR: 1.15 (1.08–1.22); two, OR: 1.26 (0.94–1.64); three, OR: 3.75 (1.24–9.18), P < 0.001). This trend, previously observed in a cross-sectional analysis in an adolescent and early adulthood cohort (Shafrir et al., 2021), is now expanded to a broader age range in our study.
Our results strongly suggest a biological basis for the epidemiological associations observed between endometriosis and a variety of immune conditions. The genetic correlation analysis demonstrated that genetic factors contribute to the association between endometriosis and the increased risk of rheumatoid arthritis (rg-female = 0.28, P = 1 × 10−3, rg-combined-sex = 0.27, P = 1.54 × 10−5), osteoarthritis (rg-female = 0.32, P = 1.76 × 10−14, rg-combined-sex = 0.28, P = 3.25 × 10−15), and to a lesser extent, multiple sclerosis (rg-female = 0.25, P = 0.075, rg-combined-sex = 0.09, P = 4.00 × 10−3). These correlations could arise from several mechanisms: (i) endometriosis may causally lead to the development of these conditions; (ii) both conditions may share a common genetic cause; or (iii) multiple shared causes could be at play, and the direction of effect between them can be complex (Kraft et al., 2020). Genetic correlation between complex diseases is driven by polygenic genetic architectures with many causal SNPs of small effect that act cumulatively into aggregated effects, which is the case for endometriosis and the immune conditions studied here.
The MR analysis found limited evidence of causality (endometriosis directly causing an immune condition), with a suggestive causal effect of endometriosis on rheumatoid arthritis in females (OR = 1.16, 95% CI = 1.02–1.33, P = 0.028). A recently published MR-based study also illustrated a suggestive causal association between endometriosis and rheumatoid arthritis (OR = 1.005, 95 CI: 1.001–1.009, P = 0.014) (Tang et al., 2024). However, the power of MR analysis depends on using genetic variants that strongly predict the exposure (endometriosis). Even genome-wide significant variants often offer only modest prediction of exposure as they explain only a small proportion of the heritable variation, as is the case for endometriosis (Burgess, 2014). In our analysis, the 39 endometriosis-associated variants explained 4.8% of medically/surgically confirmed and 5.01% of stage III/IV disease risk (10% of disease risk) (Rahmioglu et al., 2023), limiting the power to detect causal relationships. Our MR instruments would have been weighted towards risk for stage III/IV disease, in particular ovarian endometrioma (Rahmioglu et al., 2023). Previous studies associating risk of autoimmune and inflammatory conditions with endometriosis included predominantly stage I/II cases (Harris et al., 2016; Shafrir et al., 2021), although some of this evidence was based on adolescents who may have been genetically predisposed to develop stage III/IV disease later in life (Shafrir et al., 2021).
A clear limitation in available data in our analyses was the lack of large, female-specific GWAS datasets for immune conditions, which restricted our ability to draw stronger conclusions about causal inferences in genetic overlaps. While we conducted female-specific GWAS analyses in the UKBB, the sample sizes were limited compared to sex-combined GWAS meta-analysis in the literature. This limitation is particularly relevant for immune conditions with higher prevalence in females. Sex-specific genetic signatures are well documented in conditions that show sex-based variability (Voskuhl, 2011), and female-specific GWAS for immune conditions could reveal stronger genetic correlations with endometriosis and increased opportunity for the discovery of shared genetic signals. Future studies with larger female-specific GWAS results for immune conditions are needed to better understand these relationships.
While power limitations hamper the interpretation of causal relationships between endometriosis and osteoarthritis, rheumatoid arthritis, or multiple sclerosis, results of the genetic correlation analyses highlight a shared genetic basis. Clinically, we recommend increased awareness of the risk of comorbidity in endometriosis patients and vigilance for early signs of these immune conditions. Understanding the basis of genetic sharing regardless of causality is important, as understanding of the shared biological mechanisms of pathogenesis and pathophysiology could open new avenues for treatment development. Leveraging the shared genetic basis between endometriosis, osteoarthritis, rheumatoid arthritis, and multiple sclerosis via MTAG analysis, we identified many shared genetic variants, effector genes, and pathways that could aid the discovery of novel treatment targets.
Future genetic comorbidity analyses should also explore results for different endometriosis subtypes. Recent GWAS analyses have suggested that ovarian endometriosis has a different genetic basis than peritoneal disease (Rahmioglu et al., 2023), but the sample sizes for which summary statistics were generated did not allow for sufficiently powered inclusion in the present analyses. Similarly, future analyses should explore signals for different subtypes of immunological diseases such as osteoarthritis (Boer et al., 2021), as well as specific immunophenotypes of rheumatoid arthritis (Kubo et al., 2024), once larger GWAS datasets become available. Lastly, genetic analyses were limited to European ancestry individuals, and larger GWAS across more diverse ancestry groups for all diseases are needed to add translational value.
In conclusion, our results show that females with endometriosis are at a modestly increased risk of both autoimmune (42%) and autoinflammatory (28%) conditions, and that comorbidity with osteoarthritis, rheumatoid arthritis, and to a more limited extent multiple sclerosis, is biologically underpinned. Clinically, our results highlight the importance of awareness among treating physicians about the increased risk of such comorbidities. Early detection of immunological conditions in individuals with endometriosis—and vice versa—could improve patient outcomes. While current clinical action is limited to increased vigilance, our results offer a wide range of novel avenues and targets for exploring mechanisms and potential cross-condition treatment development or drug repurposing.