UDP-Glucose Ceramide Glucosyltransferase Inhibition, Immune Cell Mediation, and Endometriosis Risk: A Mendelian Randomization Study

Endometriosis (EM) is a chronic inflammatory gynecological disorder that affects approximately 190 million women worldwide. 1 , 2 Current treatments for EM include medical therapies (eg, hormone treatments) and surgical interventions (eg, hysterectomy); 3–5 however, EM symptoms often persist or recur. 1 , 6 The pathogenesis of the disease remains incompletely understood, and it is believed to involve the interplay of multiple factors, including genetic susceptibility, immune dysfunction, hormonal imbalances, and environmental influences. 1 , 2 , 5 , 6 Several etiological hypotheses proposed by researchers suggest that immune system abnormalities may be a key factor in the growth of ectopic endometrial tissue. Immune dysfunction may impair the normal immune surveillance mechanisms, preventing the clearance of these ectopic tissues. Additionally, hormonal imbalances, particularly abnormal estrogen levels, may promote the proliferation and survival of ectopic endometrial cells. Furthermore, genetic susceptibility and environmental factors may also play a synergistic role in the onset of the disease, further complicating the pathological process of EM. Understanding the pathogenesis of EM and subsequent development of effective treatments are essential for improving therapeutic outcomes. 1 Previous research focusing on identifying diagnostic biomarkers for EM has highlighted various proteins and genes with potential as diagnostic, therapeutic, and prognostic markers. 3 , 7 , 8 Elevated expression of UDP-glucose ceramide glucosyltransferase ( UGCG ) has emerged as a promising therapeutic target for EM. 9 Notably, dysregulated immune cell activities, compositions, and regulatory functions in patients with EM may contribute to disease progression. The immune alterations include reduced cytotoxicity of CD8 + T cells, distinct macrophage profiles from those in normal endometrial tissues, and impaired natural killer cell function. 10–12 UGCG, which plays a critical role in cancer pathogenesis, affects immune cell functions, and inhibition of UGCG expression reportedly enhances CD8 + T cell immunogenicity and potentially inhibits tumor growth. 13–15 Although UGCG inhibitors, such as eliglustat and migalastat, are approved by the United States Food and Drug Administration (FDA) for other conditions, their application in EM remains unexplored. 13 , 16 , 17 Moreover, existing research on EM predominantly relies on observational studies, limiting causal inference. 18 Considering the potential of UGCG inhibitors, such as eliglustat and migalastat, in modulating immune responses and their approval for other conditions by the FDA, investigating their role as drug targets in EM is warranted. Therefore, robust research approaches are needed to investigate the causal relationships between UGCG inhibition, immune cells, and EM onset. Mendelian randomization (MR) is a powerful tool for addressing complex questions in the fields of human biology and epidemiology. MR analytical methods adopt statistical techniques from economics, allowing researchers to analyze the effects of the environment, drug treatments, and other factors on human biology and disease. 19 Drug target MR offers a promising approach to assess the causal effects of UGCG inhibition on immune cell functions and the onset of EM. MR utilizes natural genetic variations to establish causal relationships, minimizing the effects of confounding factors and enhancing the validity of research findings. 20 , 21 This method enables the exploration of the long-term effects of pharmacological interventions on EM risk, akin to simulated clinical trials targeting the action mechanisms of various drugs. 22 , 23 Davies et al 24 highlighted that advancements in genome-wide association studies (GWASs) and molecular studies have strengthened the basis for MR research. 24 In the present study, we aimed to investigate the causal relationships between UGCG inhibition, immune cells, and EM, particularly, the mediating role of immune cells, using two-sample and two-step MR analyses. The findings of this study may deepen our understanding of the pathological mechanisms linking UGCG inhibition to EM development.

A two-sample, two-step MR design was employed ( Figure 1a–c ). Drug target MR uses single nucleotide polymorphisms (SNPs) restricted to the gene locus encoding the drug target as indicators for a specific drug. 25 To ensure the validity of potential causal effects in the MR analysis of drug targets, the following key assumptions should be met: (i) Correlation assumption: the instrumental variables should exhibit a strong association with UGCG inhibition, with a significance level of P < 1 × 10 –5 indicating a robust correlation. Instrumental variables with an F-statistic ≤ 10 were therefore excluded from further analysis; (ii) Independence assumption: the selected instrumental variables must not be influenced by confounding factors; (iii) Exclusion restriction assumption: the instrumental variables should influence EM solely via UGCG inhibition and do not exhibit direct association with EM, ensuring the absence of genetic pleiotropy, and non-significant differences ( P > 0.05) in the intercept term of the MR-Egger regression analysis indicating the absence of genetic pleiotropy; (iv) Target gene relevance: instrumental variables should fall within the cis-regulatory range of the target gene (kb = 300). 26 This study adhered to the principles specified in the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for MR ( Supplementary Material 1 ). 27 Figure 1 Outline of the study structure. ( a ) Flowchart depicting the assessment of the role of immune cells in mediating the influence of UGCG inhibition on EM. ( b ) In the two-step MR analysis, β1 represents the effect of UGCG inhibition on immune cells, β2 represents the effect of immune cells on EM, and β3 represents the direct effect of UGCG inhibition on EM. ( c ) The flow diagram of conducting the two-step MR step by step. Abbreviations : UGCG , UDP-glucose ceramide glucosyltransferase; EM, endometriosis; MR, Mendelian Randomization. Outline of the study structure. ( a ) Flowchart depicting the assessment of the role of immune cells in mediating the influence of UGCG inhibition on EM. ( b ) In the two-step MR analysis, β1 represents the effect of UGCG inhibition on immune cells, β2 represents the effect of immune cells on EM, and β3 represents the direct effect of UGCG inhibition on EM. ( c ) The flow diagram of conducting the two-step MR step by step. Expression quantitative trait loci (eQTLs) associated with the target gene UGCG were obtained from the IEU OpenGWAS platform ( https://gwas.mrcieu.ac.uk/;accessionnumber:ENSG00000148154 ). This study used 731 immune cell trait data from the GWAS catalog (numbered from GCST0001391 to GCST0002121). 28 The immune cell characteristics of this dataset can be divided into seven panels: B cells; conventional dendritic cells (cDC); maturation stages of T cells; monocytes; myeloid cells; B cells, natural killer cells, T cells (TBNK); and T regulatory cells (Tregs). GWAS data for EM (N14_ENDOMETRIOSIS) were sourced from the FinnGen project website ( https://www.finngen.fi/en ), accessed on March 13, 2024 ( Table 1 ). By using distinct datasets from separate sources, we minimized the risk of sample overlap and enhanced the robustness of our findings. Table 1 Summary of the eQTL and GWAS Databases Used in MR Analyses Data Source Phenotype Sample Size Cases Population Adjustment IEU Open GWAS project (eqtl-a-ENSG00000148154) UGCG 31,684 - European Sex IEU Open GWAS project Immune cells 3757 - European - FinnGen (N14_ENDOMETRIOSIS) EM 128,171 16,588 European - Abbreviations : eQTL, expression quantitative trait loci; GWAS, genome-wide association study; MR, Mendelian randomization; UGCG, UDP-glucose ceramide glucosyltransferase; EM, endometriosis. Summary of the eQTL and GWAS Databases Used in MR Analyses Abbreviations : eQTL, expression quantitative trait loci; GWAS, genome-wide association study; MR, Mendelian randomization; UGCG, UDP-glucose ceramide glucosyltransferase; EM, endometriosis. Genetic variants involved in drug target genes were identified through the following four steps ( Figure 1a ): (i) SNPs were selected based on P < 1 × 10 –5 , linkage disequilibrium coefficient (r²) 0.01. Recent studies widely endorse these as standard parameters, promoting SNP independence and reducing the influence of linkage disequilibrium. 29–31 (ii) Confounding SNPs were removed using PhenoScanner ( http://www.phenoscanner.medschl.cam.ac.uk/ ), which identifies and excludes SNPs linked to confounding factors and outcomes. (iii) eQTL data were used to extract instrumental variables within 300 kb of the cis-regulatory region associated with drug target genes. (iv) For data filtering, relevant SNPs from GWAS data of 731 types of immune cells and EM ( Table 1 ) were selected, excluding SNPs with palindromic structures and MAF > 0.42. Outlier SNPs were removed using MR-PRESSO. The use of at least 10 independent SNPs as instrumental variables ensures sufficient statistical efficiency in MR analyses; therefore, a threshold of P < 1 × 10 –5 was used based on previous findings. 29 , 32 (i) To ensure SNP independence and minimize the effect of linkage disequilibrium, a linkage disequilibrium coefficient of r² < 0.001 and region width of 10,000 kb were established. (ii) PhenoScanner was used to exclude SNPs linked to confounding factors and outcomes. (iii) For data extraction and filtering, SNPs that met the above criteria were extracted from the summary GWAS data of EM ( Table 1 ). SNPs with palindromic structures and MAF > 0.42 were excluded. (4) To remove outliers, abnormal SNPs were eliminated using MR-PRESSO. 33 Five regression models were used in this study: MR-Egger regression, 34 the inverse-variance weighted (IVW) method, 35 , 36 weighted median estimator, 37 weighted mode, 38 and a simple model. We performed a two-sample MR analysis, utilizing eQTLs as instrumental variables, to examine the possible causal association between UGCG inhibition and EM risk. The selected models were designed to enhance both the reliability and robustness of the findings by mitigating the effects of various biases and offering complementary perspectives on the potential causal relationship. There are various ways to interpret results in drug target MR analysis. Notably, when UGCG is used as an indicator for MR analysis of drug targets, the OR is > 1, indicating elevated UGCG as a risk factor for the outcome. However, the UGCG inhibitor is the exposure factor, which acts on the drug target and inhibits the increase in UGCG. Therefore, the result should be interpreted in reverse, that is, the actual effect of the UGCG inhibitor on the EM outcome is the inverse of the original OR value \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document}${1 \over {OR}}$\end{document} . \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document}${1 \over {OR}} < 1$\end{document} . This suggests that the use of UGCG inhibitor drugs may reduce the risk of adverse outcomes. The IVW method was primarily used for causal estimation because of its precision and efficiency. The IVW results were interpreted as the main causal estimates. MR-Egger regression was additionally employed to identify and correct pleiotropy, ensuring that the causal estimates were not biased by pleiotropic effects of the instrumental variables. The Wald ratio method was used to evaluate the effects of individual SNPs on the outcome when there were three or fewer SNPs, whereas the fixed-effect IVW method was applied for the other analyses. The random-effects IVW method was used for analyses involving more than three SNPs. This approach calculates causal effects by weighting each SNP’s estimate with its inverse variance (R²) before summing the weighted estimates to determine the final causal influence. MR-Egger estimates the causal effect using a weaker assumption, termed the InSIDE assumption, based on IVW. Incorporating a regression intercept helps identify and adjust for bias caused by the pleiotropic effects of instrumental variables, facilitating the estimation of the causal relation between exposure and outcomes. MR-Egger results are particularly valuable when horizontal pleiotropy is present. 34 , 39 The causal effect estimates for UGCG inhibition on EM were evaluated using an online efficacy calculator ( https://shiny.cnsgenomics.com/mRnd/ ), with the threshold for statistical power set at 0.8. 40 Given the key role of immune cells in EM development, we utilized a two-step MR analysis to explore their role in mediating between UGCG inhibition and EM risk ( Figure 1b ). First, we used SNPs associated with UGCG inhibition to determine their causal effects on immune cells. Second, we employed SNPs related to immune cells to evaluate their causal effects on EM outcomes. Subsequently, we examined how UGCG inhibition directly influences EM development and the indirect role of immune cells in modulating this effect. The mediation ratio was calculated to quantify this relationship ( Figure 1c ). Notably, β1 represents the influence of UGCG inhibition on immune cell behavior, whereas β2 reflects the role immune cells play in EM. The indirect effect of UGCG inhibition on EM, or the mediation effect, was calculated as the product of β1 and β2 (β1 × β2). β3 indicates the total causal effect of UGCG inhibition on EM. The mediation effect (β1 × β2) was divided by the total effect (β3) to calculate the proportion of the indirect effect (E%). When the total, direct, and indirect effects were consistent in direction and the mediation proportion surpassed 5%, immune cells were identified as likely mediators, indicating a notable contribution of the indirect effect. We assessed SNP heterogeneity using Cochran’s Q test and the I-squared (I²) statistic. 36 Cochran’s Q test indicated heterogeneity when P 50% indicated notable variability in the IVW findings. The formula for I 2 is \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document}${I^2} = {{Q - Q\_df} \over Q} \times 100\rm\% $\end{document} . We performed heterogeneity analysis utilizing the intercept from MR-Egger regression alongside the MR-PRESSO test. Sensitivity analysis was conducted using the leave-one-out approach, where each SNP was individually removed, and the remaining data were reanalyzed to evaluate the effect of excluding that SNP on the overall results. An intercept close to zero in the MR-Egger regression with a non-significant P -value ( P > 0.05), along with an MR-PRESSO P > 0.05, suggested the absence of pleiotropy among the selected SNPs. All analyses were performed using the TwoSampleMR package v0.5.10 in R v4.1.0, with a significance threshold of α = 0.05.

We identified 24 SNPs linked to EM from the UGCG eQTL data, all exhibiting an F-statistic value of >21 ( Supplementary Table 1 ). The statistical power of the study was 1.0, indicating a very high probability that the study correctly rejected the null hypothesis if there was a true effect ( Supplementary Table 2 ). MR analysis revealed that UGCG inhibition was associated with a decreased risk of EM, as indicated by the IVW results (odds ratio [OR] = 0.915, 95% confidence interval [CI]: 0.859–0.975, P = 0.006). The IVW results revealed no heterogeneity among the eQTLs associated with UGCG inhibition and EM (I² = 20%, Cochran’s Q = 28.798, P = 0.187). The MR-Egger results indicated that the intercept term and deviation from zero were not significant ( P = 0.773), suggesting the absence of horizontal pleiotropy. Additionally, MR-PRESSO did not detect substantial horizontal pleiotropy ( P = 0.236), corroborating the reliability of the MR findings ( Table 2 ). Sensitivity analyses further validated the results. Table 2 Causal Association Analysis Between UGCG Inhibition and EM Exposure Outcome Method Nsnp MR Heterogeneity Horizontal Pleiotropy MR-PRESSO OR (95% CI) P I 2 (%) Cochran’s Q P Egger Intercept SE P P UGCG inhibition EM IVW 24 0.915 (0.859–0.975) 0.006 20 28.798 0.187 0.236 MR-Egger 24 0.930 (0.822–1.053) 0.264 23 28.687 0.154 0.002 0.008 0.773 Weighted median 24 0.934 (0.858–1.016) 0.111 Simple mode 24 0.946 (0.816–1.097) 0.472 Weighted mode 24 0.921 (0.820–1.034) 0.178 Abbreviations : MR, Mendelian randomization; UGCG, UDP-glucose ceramide glucosyltransferase; EM, endometriosis; Nsnp, number of single-nucleotide polymorphisms; OR, odds ratio; CI, confidence interval; SE, Standard Error of β; P, P-value. Causal Association Analysis Between UGCG Inhibition and EM Abbreviations : MR, Mendelian randomization; UGCG, UDP-glucose ceramide glucosyltransferase; EM, endometriosis; Nsnp, number of single-nucleotide polymorphisms; OR, odds ratio; CI, confidence interval; SE, Standard Error of β; P, P-value. The leave-one-out analysis showed that the calculated effect sizes did not significantly differ after excluding individual SNPs, indicating the stability of the results. A funnel plot suggested no apparent bias, as the points on both sides of the symmetry axis were approximately balanced. A scatter plot indicated that the risk of developing EM increased with increasing UGCG expression, implying a positive association between UGCG activation and the risk of EM. A forest plot showed that the 95% CI for the overall effect size was situated to the right of 0, confirming that UGCG activation was positively associated with an increased risk of developing EM ( Figure 2a–d ). Figure 2 Sensitivity analysis. ( a ) Sensitivity analysis using a leave-one-out approach for UGCG and EM. ( b ) Funnel plot for UGCG and EM. ( c ) Scatter plot for UGCG and EM. ( d ) Forest plot for UGCG and EM. Abbreviations : UGCG , UDP-glucose ceramide glucosyltransferase; EM, endometriosis. Sensitivity analysis. ( a ) Sensitivity analysis using a leave-one-out approach for UGCG and EM. ( b ) Funnel plot for UGCG and EM. ( c ) Scatter plot for UGCG and EM. ( d ) Forest plot for UGCG and EM. We obtained 15,033 SNPs associated with 731 immune cells from the eQTL data for the UGCG inhibition drug target gene ( Supplementary Table 3 ). Using eQTLs linked to UGCG suppression to assess their causal impact on immune cells, the IVW method demonstrated a statistically significant causal relationship between UGCG suppression and 156 immune cells. UGCG inhibition exerted dual effects: it acted as a protective factor for 84 immune cells (OR < 1, P 1, P < 0.05) ( Supplementary Table 4 ). The IVW results showed no significant heterogeneity between UGCG inhibition and most immune cell-associated SNPs. MR-Egger analysis validated the MR findings, indicating no considerable heterogeneity or horizontal pleiotropy. However, IVW analysis revealed significant heterogeneity for T/B cells ( P = 0.035) and B cell %lymphocytes ( P = 0.043). Nonetheless, the MR-Egger analysis confirmed the robustness of the results without significant horizontal pleiotropy, and MR-PRESSO analysis revealed no significant levels of horizontal pleiotropy. In the second step, 10,496 EM-associated SNPs were identified in 731 types of immune cells ( Supplementary Table 5 ). The IVW analysis revealed significant causal relationships between 41 immune cells and EM. These immune cells had dual effects on EM, with 21 immune cells acting as protective factors (OR < 1, P 1, P < 0.05) ( Supplementary Table 6 ) Consistent with the results obtained in the first step, the IVW findings showed no evidence of heterogeneity between the SNPs associated with these immune cell types and EM. MR-Egger analysis confirmed the stability of the MR findings, with no significant heterogeneity or horizontal pleiotropy. Combining the results of the first two steps, 12 immune cell types exhibited significant causal relationships with both UGCG inhibition and EM ( Figure 3 and Table 3 ). The final 12 immune cells obtained from the intersection included: B cell panel: IgD+ CD24+ %B cell, CD20 on CD20- CD38-, CD24 on IgD+ CD38-, IgD on IgD+ CD38 dim , B-cell activating factor receptor (BAFF-R) on CD20-; Maturation stages of T cell panel: terminal differentiation (TD) CD4+ %T cell, CD4 on naive CD4+; Myeloid cell panel: HLA DR on CD33 br HLA DR+ CD14 dim ; TBNK panel: HLA DR++ monocyte %leukocyte, CD4+ %T cell, CD45 on B cell; Treg panel: CD127- CD8 br %T cell. Notably, eight of these immune cells could serve as potential mediators of UGCG inhibition and EM. The mediators were as follows: B cell panel: IgD+ CD24+ %B (15.499%), CD24 on IgD+ CD38- (16.627%), IgD on IgD+ CD38 dim (7.816%), BAFF-R on CD20˗ (14.379%); Maturation stages of T cell panel: TD CD4+ %T cell (26.727%); Myeloid cell panel: HLA DR on CD33 br HLA DR+ CD14 dim (13.946%); TBNK panel: CD45 on B cell (26.081%); Treg panel: CD127- CD8 br %T cell (14.005%) (details of the leave-one-out sensitivity analysis are illustrated in Supplementary Figures 1 and 2 ). TD CD4+ %T cells exhibited the highest mediation proportion, at 26.727% (95% CI: 25.198–28.255%). This indicates that UGCG inhibition indirectly reduces the risk of EM by impacting TD CD4+ T cells. IgD on IgD+ CD38 dim cells accounted for the lowest proportion of mediating cells at 7.816% (95% CI: 6.33–9.303%). The mediating effects of these eight immune cell types were negative, indicating that UGCG inhibition indirectly decreased the risk of EM by affecting these immune cells ( Figure 4a–h ). Table 3 Immune Cells as Mediators of the Effects of UGCG Inhibition on EM Mediator β1 β2 Mediating Effect Direct Effect Total Effect Proportion Mediated (%) IgD + CD24 + %B cells –0.207 0.066 ˗0.014 –0.075 –0.088 15.499 (13.938–17.06) HLA DR ++ monocyte %leukocytes –0.147 –0.109 0.016 –0.104 –0.088 –18.138 (–19.467–16.809) TD CD4 + %T cells 0.202 –0.117 –0.024 –0.065 –0.088 26.727 (25.198–28.255) CD4 + %T cells –0.146 –0.067 0.01 –0.098 –0.088 –11.056 (–12.363–9.749) CD127 – CD8 br %T cells 0.17 –0.073 –0.012 –0.076 –0.088 14.005 (12.533–15.476) CD20 on CD20 – CD38 – cells –0.158 –0.117 0.018 –0.107 –0.088 –20.945 (–22.295–19.595) CD24 on IgD + CD38 – cells –0.168 0.087 –0.015 –0.074 –0.088 16.627 (15.223–18.031) IgD on IgD + CD38 dim cells –0.188 0.037 –0.007 –0.081 –0.088 7.816 (6.33–9.303) BAFF-R on CD20 – cells 0.155 –0.082 –0.013 –0.076 –0.088 14.379 (13.05–15.708) CD45 on B cells –0.193 0.119 –0.023 –0.065 –0.088 26.081 (24.553–27.609) CD4 on naive CD4 + 0.207 0.081 0.017 –0.105 –0.088 –18.87 (–20.545–17.195) HLA DR on CD33 br HLA DR + CD14 dim –0.236 0.052 –0.012 –0.076 –0.088 13.946 (11.835–16.057) Abbreviations : UGCG, UDP-glucose ceramide glucosyltransferase; EM, endometriosis; β1, represents the effect of UGCG inhibition on immune cells; β2, represents the effect of immune cells on EM. Figure 3 Forest plot depicting the effects of inhibiting UGCG activity on immune cells and the influence of immune cells on EM. Abbreviations : UGCG , UDP-glucose ceramide glucosyltransferase; EM, endometriosis. Figure 4 Mediation diagrams. Mediation effects of inhibiting UGCG activity on EM exerted via various immune cell populations: ( a ) TD CD4 + %T cell, ( b ) IgD + CD24 + %B cell, ( c ) CD127- CD8 br %T cell, ( d ) CD24 on IgD + CD38-, ( e ) IgD on IgD + CD38 dim , ( f ) BAFF-R on CD20 ˗ , ( g ) CD45 on B cell, and ( h ) HLA DR on CD33 br HLA DR + CD14 dim . Abbreviations : UGCG , UDP-glucose ceramide glucosyltransferase; EM, endometriosis. Immune Cells as Mediators of the Effects of UGCG Inhibition on EM Abbreviations : UGCG, UDP-glucose ceramide glucosyltransferase; EM, endometriosis; β1, represents the effect of UGCG inhibition on immune cells; β2, represents the effect of immune cells on EM. Forest plot depicting the effects of inhibiting UGCG activity on immune cells and the influence of immune cells on EM. Mediation diagrams. Mediation effects of inhibiting UGCG activity on EM exerted via various immune cell populations: ( a ) TD CD4 + %T cell, ( b ) IgD + CD24 + %B cell, ( c ) CD127- CD8 br %T cell, ( d ) CD24 on IgD + CD38-, ( e ) IgD on IgD + CD38 dim , ( f ) BAFF-R on CD20 ˗ , ( g ) CD45 on B cell, and ( h ) HLA DR on CD33 br HLA DR + CD14 dim .

We examined the causal relationship between UGCG inhibition and EM. Our findings suggested that UGCG inhibition may help prevent or reduce EM. Additionally, we investigated the role of immune cells in mediating this relationship. Our findings indicated that immune cells regulate the effects of UGCG inhibition on EM. The eight mediator immune cell phenotypes identified included four B cell-related phenotypes, one TBNK-related phenotype, one phenotype associated with mature T cells, one Treg-related phenotype, and one myeloid cell-related phenotype, revealing the biological basis of EM, which may potentially aid in the development of new therapeutic targets and personalized treatments. UGCG expression is elevated in patients with EM. 9 , 41 This upregulation can cause the accumulation of glycosphingolipids, including glucosylceramide, in the endometrial tissues of individuals with EM. Glucosylceramide is essential for immune regulation, and its overexpression may result in immune dysfunction, influencing the EM pathogenesis and progression. 41 The immune microenvironment plays a central role in EM pathogenesis. The peritoneal immune microenvironment, including innate and adaptive immunity, has a significant effect on the vascularization and fibrosis of EM lesions. Immune cells, such as macrophages, natural killer cells, dendritic cells, neutrophils, T cells, and B cells, as well as related cytokines and inflammatory mediators, may accelerate the implantation and development of ectopic lesions by promoting the angiogenesis and fibrosis of lesions. 42 Immune cells play pivotal roles in EM pathogenesis. For instance, Chen et al 42 demonstrated that dysfunctional macrophages can disrupt the immune microenvironment and promote the development and exacerbation of EM. A study by Mobarak et al 43 indicated that UGCG inhibition reduced the glycosphingolipid content in cell membranes, weakening the binding of Toll-like receptor 4 to the intracellular signaling protein Mal, thus inhibiting the immune response to pathogens. This suggests that UGCG inhibition holds significant potential for improving immune cell function, regulating the immune microenvironment, and controlling disease progression in patients with EM. B cell receptor engagement is reportedly involved in the conversion of ceramide into glucosylceramide by UGCG. 44 Therefore, UGCG plays a key role in B cell glycolipid metabolism. UGCG activity is essential for the maintenance of glycolipid expression on the B cell membrane, and UGCG inhibitors may weaken the proinflammatory activity of B cells by reducing the glycolipid content on the B cell membrane, thereby interfering with the immune evasion and chronic inflammatory processes of EM lesions. UGCG expression in plasmacytoid dendritic cells is crucial for their cytokine production and antiviral immune response. The regulation of plasmacytoid dendritic cell glucose and lipid metabolism by UGCG inhibitors may reduce the release of proinflammatory factors and the inflammatory state of the local immune microenvironment, thereby inhibiting the angiogenesis and fibrosis of EM lesions, thus hindering the implantation and development of EM lesions. 45 Ma et al 10 demonstrated that decreased T cell activation and impaired natural killer cell function lead to reduced cell clearance and increased EM lesion growth. TD CD4 + T cells, products of T cell activation and differentiation, have essential immunomodulatory functions. Our study revealed that 26.727% of the effect of UGCG inhibition on the risk of developing EM is mediated by TD CD4 + T cells. These findings align with previous research findings, confirming a causal association between immune cells and UGCG . Furthermore, UGCG inhibition was causally associated with immune cells. Previous research has suggested a causal relationship between immune cells and EM, with a high frequency of CD33 br HLA-DR + CD14 – cells being positively correlated with EM risk. 46 Our study revealed that the negative regulatory effect of UGCG inhibitors on CD33 br HLA-DR + CD14 dim cells may lower EM risk. UGCG inhibitors may reduce EM risk by regulating different immune subsets in the CD33 high and HLA-DR + compartments. Glycosphingolipids are limiting metabolites required for cancer immune escape, and UGCG plays an important role in this process as a key enzyme, catalyzing ceramide to glycosphingolipid conversion. 47 Inhibition of glycosphingolipid synthesis using the UGCG inhibitor eliglustat increases the exposure of major histocompatibility complex and tumor antigens, thereby enhancing the antitumor response of CD8 + T cells. 48 Furthermore, UGCG inhibition reduces the glycolipid content on the immune cell membrane, affecting antigen presentation and cell–cell interactions, thereby weakening the immune escape ability of tumor cells and enhancing the recognition by and killing effects of the immune system. UGCG inhibitors may work through similar mechanisms in EM treatment. Specifically, UGCG inhibitors may inhibit the immune evasion of EM lesion cells, boost immune recognition, and decrease inflammatory responses by regulating glycolipid expression on the surfaces of immune cells. Subsequently, this may enhance the functionality of antitumor immune cells, inhibit immunosuppressive cells, reduce the release of proinflammatory factors, and promote immune cell infiltration, thereby exerting a therapeutic effect. Combination therapy studies have demonstrated the potential synergistic effects of UGCG inhibitors. The FDA-approved UGCG inhibitor eliglustat, when used in combination with a lysosomal autophagy inhibitor, significantly inhibited tumor growth and improved the survival rate in drug-resistant patients. 13 Similarly, eliglustat in combination with checkpoint blockade therapy significantly reduces tumor burden in mice. 47 This combination therapy is not only effective in tumor treatment but also has broad metabolic and immune-regulatory potential, providing a multilevel treatment option for various immune-related diseases such as EM. Mediation analysis separates the effects of an exposure on an outcome into direct effects and those exerted through a mediating variable by estimating the causal effects among the three categories of variables, including exposure, mediating variables, and outcome. This method retains the advantages of using genetic tools for causal inference, avoiding bias due to confounding, while estimating the different effects required for mediation analysis. 19 The present study used MR analysis methods to reduce the influence of confounding factors such as basic research and observational studies in EM research, as well as the possibility of reverse causality. In addition, MR studies reduce the bias due to subjective measurement errors such as self-reporting and improve the credibility of causal inference. Compared with non-instrumental variable mediation methods, instrumental variable mediation methods can improve causal inference in mediation analysis. 49 We verified the potential therapeutic targets for EM through MR research, which can guide drug development. 50–52 Despite the promising results, our study has a few limitations. First, genetic variations mimicking the effects of UGCG inhibitors may reflect the lifelong effects of UGCG inhibitors, but not accurately represent their short-term effects. Second, MR analysis focuses on determining potential causal directions rather than quantifying their magnitude. Therefore, further research is required to determine the generalizability of the study findings to populations of European descent. Future research should explore the potential mechanisms of UGCG inhibitors in treating EM, as well as their effects on immune regulation and apoptosis. Additionally, investigating novel UGCG inhibitors or developing therapeutic strategies that target immune cells may help establish more effective EM treatments.

The present study confirmed the association among genetically predicted UGCG inhibition, immune cells, and EM and demonstrated that eight immune cell types mediate the protective effects of UGCG inhibition on EM. The findings provide genetic insights into how UGCG inhibition reduces the risk of EM.