Causality of immune cells and endometriosis: a bidirectional mendelian randomization study

We undertook a two-sample Mendelian Randomization (MR) analysis to explore the causal relationship between immune cell profiles and endometriosis, with the analytical process outlined in Fig.  1 . Commencing with the extraction of genetic variants associated with immune cells from GWAS summary statistics as instrumental variables (IVs), we employed a sequential application of five MR methods to rigorously evaluate this relationship. Significant findings were then subjected to a battery of sensitivity analyses, including tests for heterogeneity, pleiotropy, a leave-one-out validation and reverse MR, to confirm the robustness of the associations. To ensure the validity of our MR inferences and mitigate potential biases, we diligently aimed to meet the following three canonical assumptions: firstly, the Relevance Assumption, where IVs must be strongly associated with the immune cell profiles; secondly, the Exclusivity Assumption, stipulating that the IVs impact endometriosis solely through the immune cells without any alternate pathways; and thirdly, the Independence Assumption, ensuring that the IVs are not confounded by any other factors [ 16 ]. These assumptions are integral to the integrity of our MR analysis, underpinning the reliability of our results and the potential causal inferences drawn therefrom. Fig. 1 The flowchart of present MR analysis The flowchart of present MR analysis We accessed summary-level GWAS data for an array of 731 immunophenotypes [ 17 ], encompassing absolute cell counts (AC; n  = 118), median fluorescence intensities (MFI; n  = 389) reflective of surface antigen levels, morphological parameters (MP; n  = 32), and relative cell counts (RC; n  = 192). This dataset included metrics for B cells, dendritic cells (CDCs), stages of T cell maturation, monocytes, myeloid cells, as well as TBNK (T, B, natural killer cells) and Treg (regulatory T cells), with MP covering CDC and TBNK. The foundational GWAS utilized a cohort of 3,757 European individuals, avoiding cohort overlap, and was based on approximately 22 million SNPs genotyped and imputed with a Sardinian sequence-based reference panel [ 18 ], with adjustments for sex, age, and age-squared. In the Mendelian randomization (MR) framework, candidate instrumental variables (IVs) were carefully selected based on a set of stringent criteria to ensure their validity and strength. Single nucleotide polymorphisms (SNPs) exceeding a genome-wide significance threshold ( p  < 1 × 10 − 5) were considered initial IVs. These were further refined by applying rigorous filters to mitigate confounding biases: SNPs were required to exhibit low linkage disequilibrium (LD, r2 ≤ 0.001) over a 10,000 kb window, have a minor allele frequency (MAF) of at least 0.01, and palindromic SNPs with ambiguous allele assignment were excluded. SNPs with an F-statistic below 10 were also discarded to avoid weak instrument bias. To address horizontal pleiotropy, which can invalidate MR findings, the MR-Pleiotropy Residual Sum and Outlier (MR-PRESSO) test—optimal when horizontal pleiotropy affects less than 50% of instruments—was employed to detect and remove outliers, as recommended by Verbanck et al. [ 19 ]. The adherence to these comprehensive selection parameters is vital for the integrity of MR analyses, ensuring that the IVs used robustly reflect the genetic associations with the immune traits under study, thereby enabling credible and reliable causal inference. For our analysis of endometriosis, we meticulously chose the most pertinent genome-wide association study (GWAS) from the Integrative Epidemiology Unit (IEU) GWAS database, accessed in October 2023. Prioritizing GWAS with large sample sizes and European ancestry, we utilized the extract_outcome_data function in R to acquire detailed summary statistics from the FinnGen cohort (GWAS ID: finn-b-N14_ENDOMETRIOSIS) [ 20 – 22 ]. The FinnGen repository, part of the Finnish Biobank’s national network, provided a robust genomic dataset on endometriosis, comprising 8,288 cases and 68,969 controls, totaling 16,377,306 SNPs. Within FinnGen, endometriosis was meticulously classified using specific ICD codes: N80 for ICD-10, 617 for ICD-9, and 625.3 for ICD-8 [ 23 ]. Such detailed and standardized categorization ensures the precision of case identification, substantially augmenting the outcome data’s reliability for in-depth analytical pursuits. The relevant ethics committee authorized initial GWAS, and all subjects supplied informed permission. Statistical analyses in our investigation were carried out using the “TwoSampleMR” ( https://mrcieu.github.io/TwoSampleMR ), “MRPRESSO”, and “gwasglue” packages within the R computational environment (version 4.3.1). The synthesis of results across multiple genetic instruments was visualized in forest plots, generated via the “forestplot” package (version 2.0.1) in R. Throughout our analyses, a p-value < 0.05 was set to denote statistical significance. In our study, we utilized Mendelian Randomization (MR) analysis to investigate the causal relationship between immune cell types and endometriosis. We selected single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) based on their prior association with immune cell traits, adhering to stringent inclusion criteria. In cases where immune cell features comprised multiple IVs, the Inverse Variance Weighted (IVW) test served as the primary analysis method [ 24 ], supplemented by MR-Egger, simple mode, weighted median, and weighted mode methods. IVW, by transforming the outcome effects of IVs into a weighted regression with a zero intercept, enabled a comprehensive assessment of immune cells’ impact on endometriosis risk, providing unbiased estimates in the absence of horizontal pleiotropy [ 25 ]. Associations with a p  < 0.05 were deemed suggestive. While MR-Egger is prone to distortion from outlier genetic variables, it could still yield unbiased estimations even with invalid IVs [ 26 ]. Simple mode, though less potent than IVW, offers robustness against pleiotropy. The weighted median method provides precise effect estimates if at least 50% of the data from valid instruments are available, and the weighted mode method is adaptable for genetic variables challenging the pleiotropy hypothesis [ 27 ]. To uphold the integrity of our Mendelian Randomization (MR) results amidst potential biases from heterogeneity and pleiotropy, we performed thorough sensitivity analyses. Heterogeneity among instrumental variables (IVs) was assessed using Cochran’s Q statistic, while funnel plot asymmetry visually represented this heterogeneity. To address horizontal pleiotropy, we employed MR-Egger intercept and MR-PRESSO global test. A leave-one-out sensitivity analysis was conducted to test the robustness of our effect size estimates, recalculating MR estimates by excluding each SNP sequentially to prevent undue influence from any single SNP. Our bidirectional MR analysis explored the causal link between endometriosis and immune cell population regulation, applying a stringent IV selection with SNPs meeting a genome-wide significance threshold of p  < 5 × 10 − 8. We examined endometriosis’s impact on immune cell regulation to pinpoint potential disruptions in immune homeostasis. This bidirectional framework also facilitated investigation into the reverse causal relationship, especially how endometriosis might influence statistically significant immune cells pathways, enriching our understanding of the endometriosis-immune system interaction. Mendelian Randomization (MR) analysis was selected for its ability to infer causality by leveraging genetic variants as instrumental variables, thereby addressing key limitations of observational studies and experimental designs. Unlike observational studies, MR analysis can mitigate biases such as reverse causation and confounding, providing more reliable estimates of causal relationships between immune cell profiles and endometriosis risk. Traditional randomized controlled trials (RCTs), while effective in testing interventions, are often impractical for studying long-term exposures and complex disease pathways like those in endometriosis. Additionally, MR analysis circumvents ethical concerns and logistical challenges associated with interventional studies by using genetic data from large-scale consortia, ensuring robustness and generalizability. By systematically evaluating the relevance, exclusivity, and independence assumptions, MR analysis strengthens the validity of causal inferences, offering unique insights into the pathophysiological mechanisms underlying endometriosis. This methodological approach not only enhances precision in estimating causal effects but also contributes significantly to understanding disease etiology and informing targeted interventions.

We conducted a two-sample Mendelian Randomization (MR) analysis to investigate the causal relationships between immune cell profiles and endometriosis risk, employing the Inverse-Variance Weighting (IVW) method as the primary analytical approach. Our study encompassed 16 immune cell subcategories, revealing significant associations with endometriosis across multiple panels (Fig.  2 and Supplementary Table S1 ). Several immune cell phenotypes exhibited a protective effect against endometriosis: CD11c + HLA DR + + monocytes %monocyte (cDC panel): OR = 0.954, 95% CI: 0.919 to 0.989, p  = 0.012. CD25 on CD39 + CD4 + T cells (Treg panel): OR = 0.967, 95% CI: 0.936 to 0.998, p  = 0.037. CD25hi CD45RA + CD4 non-Treg AC (Treg panel): OR = 0.966, 95% CI: 0.935 to 0.999, p  = 0.041. HLA DR on HLA DR + CD8bright T cells (TBNK panel): OR = 0.933, 95% CI: 0.882 to 0.988, p  = 0.018. Naive DN (CD4 − CD8−) %DN T cells (Maturation stages of T cell panel): OR = 0.935, 95% CI: 0.878 to 0.995, p  = 0.035. CD11c + HLA DR + + monocytes %monocyte (cDC panel): OR = 0.954, 95% CI: 0.919 to 0.989, p  = 0.012. CD25 on CD39 + CD4 + T cells (Treg panel): OR = 0.967, 95% CI: 0.936 to 0.998, p  = 0.037. CD25hi CD45RA + CD4 non-Treg AC (Treg panel): OR = 0.966, 95% CI: 0.935 to 0.999, p  = 0.041. HLA DR on HLA DR + CD8bright T cells (TBNK panel): OR = 0.933, 95% CI: 0.882 to 0.988, p  = 0.018. Naive DN (CD4 − CD8−) %DN T cells (Maturation stages of T cell panel): OR = 0.935, 95% CI: 0.878 to 0.995, p  = 0.035. Conversely, several immune cell phenotypes showed associations predisposing towards endometriosis: CD127 on T cells (Treg panel): OR = 1.063, 95% CI: 1.001 to 1.130, p  = 0.047. CD24 + CD27+ %lymphocytes (B cell panel): OR = 1.024, 95% CI: 1.001 to 1.048, p  = 0.037. CD25 on CD28 + CD4 + T cells (Treg panel): OR = 1.149, 95% CI: 1.026 to 1.286, p  = 0.016. CD28 on CD39 + activated Treg cells (Treg panel): OR = 1.048, 95% CI: 1.015 to 1.081, p  = 0.004. CD33bright HLA DR + CD14− %CD33bright HLA DR + myeloid cells (Myeloid cell panel): OR = 1.044, 95% CI: 1.009 to 1.081, p  = 0.013. CD45 on lymphocytes (TBNK panel): OR = 1.058, 95% CI: 1.010 to 1.108, p  = 0.017. CD45 on NK cells (TBNK panel): OR = 1.042, 95% CI: 1.001 to 1.086, p  = 0.046. CM CD8bright AC T cells (Maturation stages of T cell panel): OR = 1.060, 95% CI: 1.001 to 1.122, p  = 0.046. CX3CR1 on monocytes (Monocyte panel): OR = 1.071, 95% CI: 1.025 to 1.120, p  = 0.002. Granulocyte AC (TBNK panel): OR = 1.060, 95% CI: 1.004 to 1.120, p  = 0.035. HLA DR + NK %NK cells (TBNK panel): OR = 1.044, 95% CI: 1.000 to 1.090, p  = 0.049 CD127 on T cells (Treg panel): OR = 1.063, 95% CI: 1.001 to 1.130, p  = 0.047. CD24 + CD27+ %lymphocytes (B cell panel): OR = 1.024, 95% CI: 1.001 to 1.048, p  = 0.037. CD25 on CD28 + CD4 + T cells (Treg panel): OR = 1.149, 95% CI: 1.026 to 1.286, p  = 0.016. CD28 on CD39 + activated Treg cells (Treg panel): OR = 1.048, 95% CI: 1.015 to 1.081, p  = 0.004. CD33bright HLA DR + CD14− %CD33bright HLA DR + myeloid cells (Myeloid cell panel): OR = 1.044, 95% CI: 1.009 to 1.081, p  = 0.013. CD45 on lymphocytes (TBNK panel): OR = 1.058, 95% CI: 1.010 to 1.108, p  = 0.017. CD45 on NK cells (TBNK panel): OR = 1.042, 95% CI: 1.001 to 1.086, p  = 0.046. CM CD8bright AC T cells (Maturation stages of T cell panel): OR = 1.060, 95% CI: 1.001 to 1.122, p  = 0.046. CX3CR1 on monocytes (Monocyte panel): OR = 1.071, 95% CI: 1.025 to 1.120, p  = 0.002. Granulocyte AC (TBNK panel): OR = 1.060, 95% CI: 1.004 to 1.120, p  = 0.035. HLA DR + NK %NK cells (TBNK panel): OR = 1.044, 95% CI: 1.000 to 1.090, p  = 0.049 Fig. 2 Forest plots showed the causal associations between endometriosis and immune cell traits. IVW: inverse variance weighting; CI: confidence interval Forest plots showed the causal associations between endometriosis and immune cell traits. IVW: inverse variance weighting; CI: confidence interval Table  1 summarizes the sensitivity analyses, confirming the robustness of our MR findings. Heterogeneity tests (Cochran Q statistics) across instrumental variables showed no significant heterogeneity (all p  > 0.05). MR-Egger intercept and MR-PRESSO global tests indicated no evidence of horizontal pleiotropy (all p  > 0.05). MR-PRESSO outlier analysis found no outliers affecting the reliability of instrumental variables. Forest plots and leave-one-out cross-validation reaffirmed the stability of our results, with no single SNP exerting undue influence (Supplementary Figure S1 ). Table 1 Evaluation of heterogeneity and directional pleiotropy using different methods Panel Trait Heterogeneity Cochran’s Q p Horizontal pleiotropy R -Egger intercept p MR-PRESSO global test p Treg CD25 on CD39 + CD4+ 0.431 0.430 0.503 CD25hi CD45RA + CD4 not Treg AC 0.553 0.387 0.594 CD127 on T cell 0.863 0.401 0.886 CD25 on CD28 + CD4+ 0.823 0.538 0.869 CD28 on CD39 + activated Treg 0.578 0.112 0.601 TBNK HLA DR on HLA DR + CD8br 0.684 0.986 0.713 CD45 on lymphocyte 0.959 0.925 0.935 CD45 on NK 0.386 0.850 0.449 Granulocyte AC 0.385 0.531 0.389 HLA DR + NK %NK 0.364 0.760 0.412 Maturation stages of T cell Naive DN (CD4 − CD8−) %DN 0.092 0.960 0.167 CM CD8br AC 0.568 0.510 0.546 cDC CD11c + HLA DR + + monocyte %monocyte 0.425 0.303 0.584 B cell CD24 + CD27+ %lymphocyte 0.614 0.698 0.714 Myeloid cell CD33br HLA DR + CD14− %CD33br HLA DR+ 0.612 0.394 0.586 Monocyte CX3CR1 on monocyte 0.911 0.882 0.922 MR, mendelian randomization; IVW, inverse variance weighted; OR, odds ratio; P-het, P value for heterogeneity using Cochran Q test; P-intercept, P value for MR-Egger intercept; MR-PRESSO, Mendelian randomization-pleiotropy residual sum outlier; SNP, single-nucleotide polymorphism The table should be inserted in line 263, after the first paragraph of the second section of the Results Sect. (2. Sensitivity Analysis Results) Evaluation of heterogeneity and directional pleiotropy using different methods MR, mendelian randomization; IVW, inverse variance weighted; OR, odds ratio; P-het, P value for heterogeneity using Cochran Q test; P-intercept, P value for MR-Egger intercept; MR-PRESSO, Mendelian randomization-pleiotropy residual sum outlier; SNP, single-nucleotide polymorphism The table should be inserted in line 263, after the first paragraph of the second section of the Results Sect. (2. Sensitivity Analysis Results) In reverse MR analyses, exploring endometriosis as the exposure and immune cell phenotypes as outcomes, significant associations were observed, notably CD25 expression on CD39 + CD4 + T cells (OR = 0.864, 95% CI: 0.760 to 0.983, p  = 0.026). These findings highlight specific immune cell phenotypes potentially influencing endometriosis risk. The results underscore the complex interplay between immune regulation and endometriosis pathogenesis, offering insights into novel therapeutic targets and diagnostic approaches (Supplementary Table S2 , Supplementary Figure S2 ).

Our Mendelian Randomization analysis identifies specific immune cells that may causally affect endometriosis risk. These findings highlight the potential for immune-targeted diagnostic and therapeutic strategies and emphasize the role of immunity in endometriosis pathology. This could advance biomarker discovery and disease understanding.

Our Mendelian Randomization (MR) analysis provides new insights into the complex interplay between immune cell phenotypes and endometriosis risk. By investigating potential causal relationships, we identified immune phenotypes that exhibit both protective and predisposing associations with endometriosis, highlighting the diverse roles of immune regulation in disease pathogenesis. Specifically, immune cell subsets such as CD11c + HLA-DR + + monocytes and distinct T cell populations demonstrated inverse associations with endometriosis risk, suggesting a protective effect. Conversely, phenotypes like CD24 + CD27 + B cells and CD25 + CD28 + CD4 + T cells showed positive correlations, potentially indicating an increased risk of developing endometriosis. These findings underscore the complexity of immune responses in endometriosis and underscore the need for targeted therapeutic approaches. The intricate immunological landscape revealed by our study highlights the heterogeneous functions of immune cell phenotypes in endometriosis. For instance, T regulatory cells (Tregs), known for their role in immune evasion and angiogenesis promotion in endometriosis, exhibit complex dynamics [ 28 – 31 ]. Our findings identify specific Treg subsets, such as CD25 + CD39 + CD4 + Tregs, which may exert a protective effect, contrasting with CD25 + CD28 + CD4 + Tregs that could potentially contribute to disease progression. This dual role underscores the nuanced functions of Tregs in endometriosis and suggests potential avenues for targeted therapeutic interventions. Our study advances understanding of systemic immune dysregulation in endometriosis beyond localized tissue findings. While previous studies have noted altered T cell subsets in ectopic endometrial tissues( 32 – 33 ), our study reveals systemic immune cell composition alterations across various phenotypes. Notably, the protective effect observed with HLA-DR on HLA-DR + CD8br T cells contrasts with the increased risk associated with elevated CD45 expression on NK cells. These insights integrate disparate findings and enhance our understanding of immune dysregulation in endometriosis, opening avenues for targeted therapies. Environmental factors, such as endocrine disruptors activating plasmacytoid dendritic cells (pDCs), interact intricately with endometriosis pathogenesis ( 34 – 35 ). Our findings suggest that increased CD11c + HLA-DR + + monocytes may mitigate environmental triggers, adding a new dimension to disease understanding. Importantly, endometriosis can manifest in various anatomical locations such as the intestinal walls, abdominal cavity, and ovaries, each potentially leading to different clinical consequences. Due to data limitations, we were unable to distinguish between these different anatomical sites and their specific associations with immune cell phenotypes in endometriosis. Despite the robust sensitivity analyses conducted, our study has limitations. Potential horizontal pleiotropy cannot be entirely ruled out, despite rigorous methodology. The use of summary-level data limits our ability to control for individual-level confounders, including lifestyle and environmental factors. Furthermore, our findings may not generalize beyond European ancestry populations, emphasizing the need for diversity in study cohorts to enhance the robustness and applicability of our results. Future research should encompass diverse cohorts and incorporate individual-level data to refine causal inferences and elucidate the biological pathways underlying immune cell phenotypes in endometriosis. Mechanistic studies are crucial for developing targeted therapies and identifying novel diagnostic biomarkers. In conclusion, our study unveils the intricate interplay between immune cell profiles and endometriosis, laying a foundation for deeper investigation into its immunological underpinnings. Identifying novel biomarkers and developing innovative treatments holds promise for improving outcomes in this debilitating disease.

Endometriosis, a benign gynecological condition, shares characteristics with malignant tumors like hyperplasia, metastasis, and recurrence [ 1 ]. It affects about 10% of reproductive-age women, rising to 50% among those who are infertile, posing a significant global health issue [ 2 ]. The hallmark of endometriosis is the presence of endometrial-like tissue outside the uterus, accompanied by diverse disease phenotypes and varying symptom severity. Its exact cause remains unclear, though chronic inflammation and high estrogen levels are implicated. The condition’s complexity involves genetic, hormonal, environmental, and immunological factors [ 3 , 4 ]. Current research aims to uncover its underlying pathophysiology for preventive measures [ 5 ]. During menstruation, retrograde endometrial cells attach to pelvic structures, triggering inflammation, fibrosis, and pain [ 6 ]. Cytokines are pivotal in this process, influencing immune responses and contributing to lesion development [ 7 ]. Altered immune function and cytokine expression play crucial roles in endometrial lesion formation [ 8 – 10 ]. Recent research underscores immune cells’ pivotal role in endometriosis. Neutrophils and peritoneal macrophages secrete mediators promoting angiogenesis, aiding endometriotic cell growth and invasion [ 11 , 12 ]. However, in endometriosis, peritoneal macrophages and NK cells struggle to eliminate ectopic endometrial cells, exacerbating the condition. Imbalances in Th1/Th2 cells drive abnormal cytokine secretion, fostering inflammation and lesion progression [ 13 ]. Despite numerous observational studies, causal links between immune cells and endometriosis remain unclear due to confounding and reverse causality issues in conventional research methods. Mendelian Randomization (MR) analysis represents a robust methodology in epidemiology that utilizes genetic variants as instrumental variables to explore causal relationships between exposures and health outcomes [ 14 , 15 ]. Unlike traditional observational studies, MR provides a stronger framework to mitigate confounding biases and reverse causation, thereby offering clearer insights into complex biological mechanisms. By applying MR to investigate the relationship between immune cell dynamics and endometriosis, we aim to delineate causal pathways and potentially identify novel biomarkers. This approach holds promise for refining diagnostic and therapeutic strategies in managing gynecological conditions like endometriosis, contributing to advancements in clinical practice and patient care.

Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2 Supplementary Material 3 Supplementary Material 3 Supplementary Material 4 Supplementary Material 4 Supplementary Material 5 Supplementary Material 5