Does Depression Increase the Risk of Endometriosis? A Combined Population‑Based and Laboratory Study Authors: Yuru Wang, Zihan Zhu, Xu Jin, Zhaoyan Ding, Wen Zhong, Yueming Zhang, Songwei Feng, Ting Wu Published: 2026 ABSTRACT BackgroundEndometriosis is a chronic, estrogen-dependent inflammatory disorder that causes substantial pain and infertility. Observational studies have reported higher rates of multiple psychiatric disorders among individuals with endometriosis; however, confounding and reverse causation limit causal interpretation.MethodsWe performed bidirectional two-sample Mendelian randomization (MR) analyses using genome-wide association study (GWAS) summary statistics from European-ancestry populations to test causal relationships between genetic liability to depression, anxiety, insomnia, bipolar disorder, schizophrenia, and endometriosis. Instrumental variables were selected using genome-wide significance thresholds and linkage disequilibrium clumping, and robustness was evaluated using MR Egger regression, the weighted median method, and mode-based estimators, together with heterogeneity and pleiotropy tests. To assess biological coherence, SH-SY5Y cells were differentiated into sympathetic neuron-like cells, norepinephrine was quantified by enzyme-linked immunosorbent (ELISA) assay, and paracrine effects on 12Z endometriotic epithelial cells were examined using CCK-8 assays. Transcriptomic profiling was followed by Gene Ontology (GO) enrichment, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses, and protein–protein interaction (PPI) network analysis.ResultsForward-direction MR supported an association between genetic liability to depression and increased endometriosis risk, whereas anxiety, insomnia, bipolar disorder, and schizophrenia showed no robust causal evidence. Reverse-direction MR did not support endometriosis genetic liability as a causal driver of psychiatric disorders. In vitro, sympathetic neuron-like differentiation increased norepinephrine secretion and enhanced 12Z cell proliferation, with stronger effects under conditioned-medium exposure. Enrichment analyses identified immune-related pathways, with prominent enrichment in Ras signaling, phosphatidylinositol 3 kinase and protein kinase B signaling, mitogen-activated protein kinase signaling, and cyclic adenosine monophosphate signaling.ConclusionThese findings suggest that depression may contribute to endometriosis susceptibility and are supported by complementary in vitro and transcriptomic evidence. Introduction Endometriosis is a chronic, estrogen-dependent inflammatory disorder characterized by ectopic endometrial-like tissue, chronic pelvic pain, infertility, and impaired quality of life.1,2 Diagnostic delays and limited disease-modifying therapies contribute to substantial clinical and socioeconomic burden, highlighting the need to clarify modifiable upstream determinants and biologically actionable pathways.3–5 Psychiatric disorders are frequently reported in individuals with endometriosis, including depression, anxiety, insomnia, bipolar disorder, and schizophrenia.6,7 This comorbidity has direct clinical relevance because psychiatric symptoms can exacerbate pain experience, worsen functional impairment, and complicate long-term management. However, observational associations do not establish directionality. Pain severity, infertility-related stress, healthcare utilization, medication exposure, and socioeconomic factors can confound the relationship, and endometriosis itself can contribute to psychiatric symptoms through chronic pain and inflammation. In addition, clinical depression as a diagnosed phenotype should be distinguished from genetic liability to depression, which represents inherited susceptibility captured by genome-wide association study summary statistics. Clarifying whether psychiatric liability contributes to endometriosis risk, or whether endometriosis itself increases susceptibility to psychiatric disorders, is essential for understanding the biological basis of this comorbidity. Genetic studies support endometriosis as a polygenic disease and implicate pathways related to hormone signaling, inflammation, tissue remodeling, and neuroimmune interactions.8 Similarly, depression and other psychiatric disorders have been linked to immune dysregulation and altered stress responses.9,10 This genetic architecture provides an opportunity to evaluate whether psychiatric disorder liability is causally related to endometriosis risk, rather than merely correlated through clinical consequences or shared confounders. Mendelian randomization uses genetic variants as instrumental variables to test causal hypotheses under explicit assumptions, thereby reducing bias from confounding and reverse causation.11 Bidirectional two-sample Mendelian randomization based on genome-wide association study summary statistics enables efficient evaluation of directionality between psychiatric disorders and endometriosis at population scale, while sensitivity estimators and pleiotropy diagnostics improve robustness to violations of instrument validity. Therefore, we conducted a bidirectional two-sample Mendelian randomization analysis to evaluate causal relationships between genetic liability to depression, anxiety, insomnia, bipolar disorder, schizophrenia, and endometriosis. Guided by the Mendelian randomization findings that prioritized depression as the psychiatric trait most strongly associated with endometriosis risk, we next modeled sympathetic neuron–derived signaling in vitro using a sympathetic neuron-like differentiation system and assessed its paracrine effects on endometriotic epithelial 12Z cells. Finally, we performed ribonucleic acid sequencing and downstream bioinformatics analyses to characterize transcriptional programs and pathways responsive to sympathetic neuron-conditioned signals. Materials and Methods Mendelian Randomization Analysis Using large-scale genome-wide association study (GWAS) summary statistics, we performed a bidirectional two-sample Mendelian randomization (MR)12 analysis to investigate the potential causal relationships between endometriosis and common psychiatric disorders, including depression, anxiety, insomnia, bipolar disorder, and schizophrenia. The overall study workflow is presented in Figure 1. To support valid causal inference, genetic variants used as instrumental variables (IVs) were selected to satisfy the three core MR assumptions: (i) relevance, genetic instruments are robustly associated with the exposure; (ii) independence, instruments are not associated with confounders of the exposure–outcome relationship; and (iii) exclusion restriction, instruments influence the outcome only through the exposure and not via alternative pathways (horizontal pleiotropy).13 Figure 1Outline of the study design. (1) relevance; (2) Independence; (3) exclusion restriction.The diagram illustrates the relationships between genetic variants, endometriosis, mental disorders and confounders. On the left, a box labeled 'IVs: SNPs associated with EMs' points to 'EMs' with an arrow marked (1). 'EMs' then points to 'Mental disorders' with another arrow marked (1). On the right, a box labeled 'IVs: SNPs associated with mental disorders' points to 'Mental disorders' with an arrow marked (1). Above, 'Confounders' are shown with arrows marked (2) pointing to both boxes labeled 'IVs'. Below, arrows marked (3) connect the two 'IVs' boxes, indicating a bidirectional relationship. Cross marks are placed near the arrows marked (2) and (3), suggesting exclusion or restriction in these pathways.Diagram showing relationships between genetic variants, endometriosis, mental disorders and confounders. Outline of the study design. (1) relevance; (2) Independence; (3) exclusion restriction. Data Sources Genetic association data for endometriosis were obtained from the GWAS Catalog (GWAS ID: ebi-a-GCST90018839).14 Summary statistics for psychiatric disorders were retrieved from the FinnGen study (Release 12), reported as the most up-to-date version (Data accessed on: 2025-07-01) available at the time of data extraction (Table 1). All datasets were restricted to individuals of European ancestry to reduce population stratification and between-study heterogeneity. All data used in this study were derived from publicly available GWAS summary statistics; therefore, no additional ethical approval or informed consent was required.Table 1Data Sources and Characteristics of GWAS Summary Statistics Used in MR AnalysesTraitNo. of Variants (SNPs) in Summary DataSample Size (Cases/Controls)GWAS IDEndometriosis24,089,7524,511 / 227,260ebi-a-GCST90018839Depression21,326,97359,333 / 434,831finngen_R12_F5_DEPRESSIOAnxiety21,325,74256,552 / 562,304finngen_R12_KRA_PSY_ANXIETY_EXMOREInsomnia21,327,0176,776 / 490,763finngen_R12_F5_INSOMNIASchizophrenia21,326,8747,234 / 484,776finngen_R12_F5_SCHZPHRBipolar disorder21,326,2218,946/ 4,34,831finngen_R12_F5_BIPO Table 1Data Sources and Characteristics of GWAS Summary Statistics Used in MR AnalysesTraitNo. of Variants (SNPs) in Summary DataSample Size (Cases/Controls)GWAS IDEndometriosis24,089,7524,511 / 227,260ebi-a-GCST90018839Depression21,326,97359,333 / 434,831finngen_R12_F5_DEPRESSIOAnxiety21,325,74256,552 / 562,304finngen_R12_KRA_PSY_ANXIETY_EXMOREInsomnia21,327,0176,776 / 490,763finngen_R12_F5_INSOMNIASchizophrenia21,326,8747,234 / 484,776finngen_R12_F5_SCHZPHRBipolar disorder21,326,2218,946/ 4,34,831finngen_R12_F5_BIPO Data Sources and Characteristics of GWAS Summary Statistics Used in MR Analyses Selection of Instrumental Variables Single-nucleotide polymorphisms (SNPs) associated with each exposure were selected as genetic instruments using a genome-wide significance threshold of P ≤ 5 × 10−8.15 For endometriosis, which had a limited number of genome-wide significant SNPs, the significance threshold was relaxed to P <1 × 10−5 to ensure an adequate number of instrumental variables. To obtain independent instruments, we performed linkage disequilibrium (LD) clumping with thresholds of r2 < 0.001 within a 10,000 kb window,16 thereby reducing bias from correlated variants. Instrument strength was evaluated using the F-statistic, and SNPs with F ≤ 10 were considered weak instruments and excluded to minimize weak-instrument bias.17 Exposure and outcome datasets were harmonized to align the effect alleles. Ambiguous and palindromic SNPs were removed to avoid strand inconsistencies. The detailed significance threshold, SNP counts, and F-statistic for each exposure are provided in Table S1. Statistical Analyses All MR analyses were conducted using R (version 4.5.1) with the TwoSampleMR package (version 0.6.19). The inverse-variance weighted (IVW) method18 was used as the primary approach to estimate causal effects. For binary traits, causal estimates were reported as odds ratios (ORs) with 95% confidence intervals (CIs). A two-sided P < 0.05 was considered statistically significant. Sensitivity analyses were performed using MR-Egger regression,19 the weighted median method,20 and mode-based estimators (simple mode and weighted mode)21 to evaluate the robustness of IVW results under different assumptions regarding horizontal pleiotropy. Between-instrument heterogeneity was assessed using Cochran’s Q test. When heterogeneity was suggested (P < 0.05), IVW estimates were obtained under a random-effects model; otherwise, a fixed-effects IVW model was applied. Directional horizontal pleiotropy was evaluated using the MR-Egger intercept test. In addition, leave-one-out analyses were conducted to assess whether any single SNP disproportionately influenced the overall causal estimate. Establishment of a Sympathetic Neuron–Endometriotic Epithelial Cell Co-Culture System Cell Culture and Differentiation of SH-SY5Y Cells Into Sympathetic Neuron-Like Cells Human endometriotic epithelial cells (12Z) and human neuroblastoma cells (SH-SY5Y) were purchased from Wuhan Pricella (Wuhan, China).22,23 Since the experiments involved established cell lines and in vitro procedures only, no additional ethical approval was required. SH-SY5Y cells were maintained in a dedicated culture medium at 37°C in a humidified incubator with 5% CO2 (Heracell 150iGP, Thermo Fisher). Culture medium was replaced every 3 days. When cells reached the logarithmic growth phase, they were seeded into 6-well plates (51110, LABSELECT) at a density of 2×105 cells/well and allowed to adhere for 24 h. To induce differentiation toward a sympathetic neuron-like phenotype, SH-SY5Y cells in the experimental group were cultured in differentiation medium consisting of DMEM supplemented with 2% FBS, 1% penicillin–streptomycin, and 10 μM all-trans retinoic acid (ATRA) for 7 days.24 Control cells received the same medium containing 0.1% DMSO instead of ATRA. After differentiation, both groups were switched to standard growth medium (DMEM with 10% FBS and 1% penicillin–streptomycin) and maintained for an additional 48 h before sample collection. Quantification of Norepinephrine Secretion by ELISA Norepinephrine (NE) secretion was quantified using an ELISA assay to validate functional sympathetic neuron-like differentiation. After 7 days of induction, cell supernatants were collected and centrifuged at 12,000 × g for 10 min at 4°C to remove debris. NE concentration in the supernatants was measured using a Human Norepinephrine (NE) ELISA Kit (ml024646, Mlbio) according to the manufacturer’s instructions. Absorbance was read at 450 nm, and NE concentrations (ng/mL) were calculated based on a standard curve. All experiments were performed with at least three independent biological replicates. Statistical analysis was conducted using GraphPad Prism (version 10.4.1). Comparisons between two groups were performed using an unpaired two-tailed t-test, and P < 0.05 was considered statistically significant. Immunofluorescence Analysis For immunofluorescence, cells were fixed with 4% paraformaldehyde, permeabilized in 0.3% Triton X-100, and blocked with 3% bovine serum albumin (BSA). Cells were then incubated overnight at 4°C with primary antibodies against tyrosine hydroxylase (TH) or β-hydroxylase (DBH), followed by fluorophore-conjugated secondary antibodies. Nuclei were counterstained with DAPI. Images were acquired using a fluorescence slide scanner, and fluorescence intensities were quantified via Fiji software across at least three independent experiments. Preparation of Conditioned Medium and Co-Culture with 12Z Cells 12Z cells were maintained in standard growth medium (DMEM supplemented with 10% FBS and 1% penicillin–streptomycin) at 37°C in a humidified incubator with 5% CO2. Conditioned medium (CM) was collected from differentiated SH-SY5Y cells after replacing the differentiation medium with standard growth medium and culturing for an additional 48 h. Supernatants were collected and centrifuged at 1,500 × g for 10 min at 4°C to remove cellular debris, followed by filtration through a 0.22 μm membrane filter. The filtered supernatant was designated as conditioned medium (CM) and stored at −20°C until further use. To evaluate the paracrine effects of sympathetic neuron-like cells on endometriotic epithelial cells, a direct co-culture system was established using Transwell inserts. 12Z cells were seeded in the lower chamber of a Transwell system (5% Corning, 0.4 μm pore size) and allowed to reach approximately 80% confluence. Differentiated SH-SY5Y cells were seeded into the upper chamber inserts at a 1:1 cell ratio relative to 12Z cells. The Transwell inserts had a pore size of 0.4 μm, which allows soluble factors to diffuse between chambers while preventing direct cell–cell contact. Co-cultures were maintained in standard growth medium for 48h before subsequent analyses. Control groups consisted of 12Z cells cultured alone under identical conditions without Transwell inserts. Cell Viability Assay (CCK-8) Cell viability and metabolic activity was assessed using the Cell Counting Kit-8 (CCK-8) assay under both direct co-culture and conditioned medium treatment conditions. 12Z cells were seeded in 96-well plates at 1 × 104 cells/well in 100 μL medium, with three technical replicates per condition. After cell attachment, CCK-8 assays were performed at 0 h, 24 h, and 48 h. Briefly, 10 μL of CCK-8 reagent was added to each well, followed by incubation for 2 h at 37°C. Absorbance at 450 nm was measured using a microplate reader (Multiskan FC, Thermo Fisher). All experiments were repeated at least three independent times. Statistical analyses were conducted using GraphPad Prism (version 10.4.1). To evaluate the independent and interactive effects of treatment group and time, a two-way analysis of variance (ANOVA) was applied. RNA Sequencing and Bioinformatics Analyses Total RNA from 12Z cells treated with CM or control medium (n = 3 per group) was extracted using TRIzol reagent. RNA integrity was assessed prior to library construction (RIN > 7.0). Poly(A)+ RNA was enriched using oligo(dT) beads, followed by RNA fragmentation and reverse transcription to generate strand-specific cDNA libraries. After end repair and adaptor ligation, libraries were amplified by PCR and sequenced on an Illumina NovaSeq 6000 platform with paired-end 150 bp reads. Raw sequencing reads were quality-filtered to obtain clean reads by removing adaptor sequences and low-quality reads. Clean reads were aligned to the human reference genome (Homo sapiens) using HISAT2. Transcript assembly and quantification were performed with StringTie, and gene expression levels were normalized and reported as fragments per kilobase of transcript per million mapped reads (FPKM).25,26 Differential expression analysis was conducted using DESeq2 based on gene-level read counts.27 Genes were considered differentially expressed if they met the thresholds of |log2(fold change)|≥1 and a Benjamini–Hochberg adjusted P value (q values/FDR) < 0.05. Functional enrichment analyses of differentially expressed genes were performed for Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways.28 Enrichment was evaluated using a hypergeometric test by comparing the number of differentially expressed genes annotated to each term/pathway against the annotated background gene set; only genes with valid GO/KEGG annotations were included in the calculations. Terms/pathways with nominal p < 0.05 were considered significantly enriched. To investigate potential functional associations among differentially expressed genes, protein–protein interaction (PPI) analysis was performed using the STRING database. Differentially expressed genes were mapped to STRING,29 and interactions with a confidence score > 400 were retained. The resulting interaction network was visualized and further analyzed in Cytoscape.30 Results Selection of Genetic Instruments In the forward-direction MR analyses, we evaluated the causal effects of genetic liability to five psychiatric disorders on the risk of endometriosis. After instrument selection and LD clumping, we included 30, 23, 28, 28, and 41 independent SNPs as instrumental variables for depression, anxiety, insomnia, schizophrenia, and bipolar disorder, respectively. In the reverse-direction MR analyses, endometriosis was treated as the exposure. After harmonization and removal of ambiguous/palindromic variants, 7 SNPs were retained as genetic instruments for endometriosis across all reverse MR models. Forward MR Analyses Using the IVW method, genetically predicted liability to depression was nominally associated with a higher risk of endometriosis (OR = 1.20, 95% CI 1.00–1.43, P = 0.04) (Table 2). Sensitivity analyses using the weighted median, simple mode, and weighted mode methods yielded directionally consistent but statistically non-significant estimates. MR-Egger regression produced an imprecise estimate with a wide confidence interval (OR = 0.84, 95% CI 0.36–1.94, P = 0.69). No evidence of heterogeneity was observed (Cochran’s Q = 28.11, P = 0.51) and the MR-Egger intercept test did not indicate directional horizontal pleiotropy (intercept = 0.02, P = 0.40) (Table 3), supporting the robustness of the primary IVW finding. The nominal association between depression genetic liability and endometriosis risk should be interpreted with caution, as it does not survive correction for multiple testing across the five psychiatric traits evaluated.Table 2Forward-Direction Two-Sample MR Estimates for the Effects of Genetic Liability to Psychiatric Disorders on Endometriosis RiskExposure OutcomeN.SNPsMethodsOR (95% CI)PvalDepressionEndometriosis30Inverse variance weighted1.20 (1.00 to 1.43)0.04MR Egger0.84 (0.36 to 1.94)0.69Weighted median1.17 (0.91 to 1.50)0.216Simple mode1.15 (0.68 to 1.95)0.613Weighted mode1.12 (0.70 to 1.79)0.641AnxietyEndometriosis23Inverse variance weighted1.20 (0.97 to 1.47)0.09MR Egger0.21 (0.06 to 0.77)0.03Weighted median1.12 (0.84 to 1.49)0.449Simple mode1.57 (0.89 to 2.78)0.134Weighted mode1.31 (0.77 to 2.21)0.332InsomniaEndometriosis28Inverse variance weighted1.01 (0.93 to 1.10)0.80MR Egger1.04 (0.90 to 1.22)0.58Weighted median1.02 (0.90 to 1.14)0.8Simple mode0.93 (0.75 to 1.15)0.507Weighted mode0.95 (0.79 to 1.15)0.618SZEndometriosis28Inverse variance weighted1.00 (0.93 to 1.07)0.99MR Egger0.85 (0.69 to 1.06)0.16Weighted median1.01 (0.91 to 1.11)0.882Simple mode1.06 (0.87 to 1.29)0.565Weighted mode1.02 (0.87 to 1.21)0.799BPDEndometriosis41Inverse variance weighted0.98 (0.91 to 1.06)0.61MR Egger0.85 (0.72 to 1.02)0.09Weighted median0.96 (0.86 to 1.07)0.457Simple mode0.96 (0.75 to 1.22)0.732Weighted mode0.97 (0.78 to 1.22)0.818Abbreviations: IVW, inverse-variance weighted; OR, odds ratio; CI, confidence interval; SZ, Schizophrenia; BPD, Bipolar disorder. Table 3Heterogeneity and Horizontal Pleiotropy Assessments for Forward-Direction MR AnalysesExposureOutcomeHeterogeneityPleiotropyQ statisticp valueMR–Egger Interceptp valueDepressionEndometriosis28.110.510.020.40AnxietyEndometriosis22.440.430.080.01InsomniaEndometriosis16.970.93−0.010.62SZEndometriosis24.360.440.020.14BPDEndometriosis38.500.540.020.09Abbreviations: SZ, Schizophrenia; BPD, Bipolar disorder. Table 2Forward-Direction Two-Sample MR Estimates for the Effects of Genetic Liability to Psychiatric Disorders on Endometriosis RiskExposure OutcomeN.SNPsMethodsOR (95% CI)PvalDepressionEndometriosis30Inverse variance weighted1.20 (1.00 to 1.43)0.04MR Egger0.84 (0.36 to 1.94)0.69Weighted median1.17 (0.91 to 1.50)0.216Simple mode1.15 (0.68 to 1.95)0.613Weighted mode1.12 (0.70 to 1.79)0.641AnxietyEndometriosis23Inverse variance weighted1.20 (0.97 to 1.47)0.09MR Egger0.21 (0.06 to 0.77)0.03Weighted median1.12 (0.84 to 1.49)0.449Simple mode1.57 (0.89 to 2.78)0.134Weighted mode1.31 (0.77 to 2.21)0.332InsomniaEndometriosis28Inverse variance weighted1.01 (0.93 to 1.10)0.80MR Egger1.04 (0.90 to 1.22)0.58Weighted median1.02 (0.90 to 1.14)0.8Simple mode0.93 (0.75 to 1.15)0.507Weighted mode0.95 (0.79 to 1.15)0.618SZEndometriosis28Inverse variance weighted1.00 (0.93 to 1.07)0.99MR Egger0.85 (0.69 to 1.06)0.16Weighted median1.01 (0.91 to 1.11)0.882Simple mode1.06 (0.87 to 1.29)0.565Weighted mode1.02 (0.87 to 1.21)0.799BPDEndometriosis41Inverse variance weighted0.98 (0.91 to 1.06)0.61MR Egger0.85 (0.72 to 1.02)0.09Weighted median0.96 (0.86 to 1.07)0.457Simple mode0.96 (0.75 to 1.22)0.732Weighted mode0.97 (0.78 to 1.22)0.818Abbreviations: IVW, inverse-variance weighted; OR, odds ratio; CI, confidence interval; SZ, Schizophrenia; BPD, Bipolar disorder. Forward-Direction Two-Sample MR Estimates for the Effects of Genetic Liability to Psychiatric Disorders on Endometriosis Risk Abbreviations: IVW, inverse-variance weighted; OR, odds ratio; CI, confidence interval; SZ, Schizophrenia; BPD, Bipolar disorder. Table 3Heterogeneity and Horizontal Pleiotropy Assessments for Forward-Direction MR AnalysesExposureOutcomeHeterogeneityPleiotropyQ statisticp valueMR–Egger Interceptp valueDepressionEndometriosis28.110.510.020.40AnxietyEndometriosis22.440.430.080.01InsomniaEndometriosis16.970.93−0.010.62SZEndometriosis24.360.440.020.14BPDEndometriosis38.500.540.020.09Abbreviations: SZ, Schizophrenia; BPD, Bipolar disorder. Heterogeneity and Horizontal Pleiotropy Assessments for Forward-Direction MR Analyses Abbreviations: SZ, Schizophrenia; BPD, Bipolar disorder. For anxiety, the IVW estimate did not support a statistically significant causal association with endometriosis (OR = 1.20, 95% CI 0.97–1.47, P = 0.09) (Table 2). MR-Egger regression suggested an inverse association (OR = 0.21, 95% CI 0.06–0.77, P = 0.03); however, the MR-Egger intercept test indicated evidence of directional pleiotropy (intercept = 0.08, P = 0.01) (Table 3). Therefore, the MR-Egger estimate for anxiety was considered unreliable. Other robust methods (weighted median, simple mode, weighted mode) did not show significant associations. No significant heterogeneity was detected (Q = 22.44, P = 0.43). No causal effects on endometriosis risk were observed for genetically predicted liability to insomnia (IVW OR = 1.01, 95% CI 0.93–1.10, P = 0.80), schizophrenia (IVW OR = 1.00, 95% CI 0.93–1.07, P = 0.99), or bipolar disorder (IVW OR = 0.98, 95% CI 0.91–1.06, P = 0.61) (Table 2). Across these analyses, sensitivity methods yielded consistent null findings, and neither heterogeneity nor directional pleiotropy was evident based on Cochran’s Q tests and MR-Egger intercept tests (Table 3). We applied a Bonferroni correction to account for five independent tests (one per psychiatric trait). The adjusted significance threshold was set at P < 0.01 (0.05/5). Consequently, the nominal association observed for depression (IVW P = 0.04) did not pass multiple testing correction. The primary findings of all forward MR analyses should therefore be regarded as exploratory and suggestive rather than confirmatory. Reverse MR Analyses Using IVW as the primary analysis, there was no evidence that genetic liability to endometriosis causally influenced the risk of depression (OR = 1.04, 95% CI 0.99–1.08, P = 0.09), anxiety (OR = 1.02, 95% CI 0.95–1.08, P = 0.62), insomnia (OR = 0.94, 95% CI 0.84–1.05, P = 0.26), schizophrenia (OR = 0.91, 95% CI 0.78–1.07, P = 0.27), or bipolar disorder (OR = 1.09, 95% CI 0.99–1.21, P = 0.09) (Table 4). Estimates derived from MR-Egger, weighted median, simple mode, and weighted mode methods were broadly consistent with the IVW results (Table 4). Cochran’s Q tests provided no strong evidence of heterogeneity for most reverse-direction models; however, the endometriosis to anxiety analysis showed borderline heterogeneity (Q = 12.61, P = 0.05) (Table 5). MR-Egger intercept tests did not indicate directional horizontal pleiotropy across reverse-direction analyses (all P > 0.05).Table 4Reverse-Direction Two-Sample MR Estimates for the Effects of Genetic Liability to Endometriosis on Psychiatric DisordersExposure OutcomeN.SNPsMethodsOR (95% CI)PvalEndometriosisDepression7Inverse variance weighted1.04 (0.99 to 1.08)0.09MR Egger1.08 (0.88 to 1.31)0.50Weighted median1.03 (0.98 to 1.08)0.28Simple mode1.01 (0.93 to 1.10)0.84Weighted mode1.01 (0.92 to 1.11)0.82EndometriosisAnxiety7Inverse variance weighted1.02 (0.95 to 1.08)0.62MR Egger0.96 (0.70 to 1.31)0.80Weighted median1.03 (0.97 to 1.09)0.34Simple mode1.00 (0.91 to 1.11)0.94Weighted mode1.01 (0.93 to 1.10)0.78EndometriosisInsomnia7Inverse variance weighted0.94 (0.84 to 1.05)0.26MR Egger1.05 (0.63 to 1.75)0.86Weighted median0.95 (0.82 to 1.09)0.47Simple mode0.94 (0.78 to 1.14)0.55Weighted mode0.95 (0.78 to 1.16)0.63EndometriosisSZ7Inverse variance weighted0.91 (0.78 to 1.07)0.27MR Egger0.49 (0.25 to 0.99)0.10Weighted median0.96 (0.78 to 1.18)0.69Simple mode0.99 (0.73 to 1.35)0.96Weighted mode0.98 (0.73 to 1.31)0.90EndometriosisBPD7Inverse variance weighted1.09 (0.99 to 1.21)0.09MR Egger1.35 (0.84 to 2.16)0.27Weighted median1.09 (0.96 to 1.24)0.18Simple mode1.09 (0.89 to 1.34)0.42Weighted mode1.09 (0.88 to 1.35)0.45Abbreviations: IVW, inverse-variance weighted; OR, odds ratio; CI, confidence interval;SZ, Schizophrenia; BPD, Bipolar disorder. Table 5Heterogeneity and Horizontal Pleiotropy Assessments for Reverse-Direction MR AnalysesExposureOutcomeHeterogeneityPleiotropyQ statisticp valueMR–Egger Interceptp valueEndometriosisDepression5.750.45−0.010.72EndometriosisAnxiety12.610.050.010.72EndometriosisInsomnia2.870.83−0.020.68EndometriosisSZ6.730.350.090.14EndometriosisBPD4.410.62−0.030.41Abbreviations: SZ, Schizophrenia; BPD, Bipolar disorder. Table 4Reverse-Direction Two-Sample MR Estimates for the Effects of Genetic Liability to Endometriosis on Psychiatric DisordersExposure OutcomeN.SNPsMethodsOR (95% CI)PvalEndometriosisDepression7Inverse variance weighted1.04 (0.99 to 1.08)0.09MR Egger1.08 (0.88 to 1.31)0.50Weighted median1.03 (0.98 to 1.08)0.28Simple mode1.01 (0.93 to 1.10)0.84Weighted mode1.01 (0.92 to 1.11)0.82EndometriosisAnxiety7Inverse variance weighted1.02 (0.95 to 1.08)0.62MR Egger0.96 (0.70 to 1.31)0.80Weighted median1.03 (0.97 to 1.09)0.34Simple mode1.00 (0.91 to 1.11)0.94Weighted mode1.01 (0.93 to 1.10)0.78EndometriosisInsomnia7Inverse variance weighted0.94 (0.84 to 1.05)0.26MR Egger1.05 (0.63 to 1.75)0.86Weighted median0.95 (0.82 to 1.09)0.47Simple mode0.94 (0.78 to 1.14)0.55Weighted mode0.95 (0.78 to 1.16)0.63EndometriosisSZ7Inverse variance weighted0.91 (0.78 to 1.07)0.27MR Egger0.49 (0.25 to 0.99)0.10Weighted median0.96 (0.78 to 1.18)0.69Simple mode0.99 (0.73 to 1.35)0.96Weighted mode0.98 (0.73 to 1.31)0.90EndometriosisBPD7Inverse variance weighted1.09 (0.99 to 1.21)0.09MR Egger1.35 (0.84 to 2.16)0.27Weighted median1.09 (0.96 to 1.24)0.18Simple mode1.09 (0.89 to 1.34)0.42Weighted mode1.09 (0.88 to 1.35)0.45Abbreviations: IVW, inverse-variance weighted; OR, odds ratio; CI, confidence interval;SZ, Schizophrenia; BPD, Bipolar disorder. Reverse-Direction Two-Sample MR Estimates for the Effects of Genetic Liability to Endometriosis on Psychiatric Disorders Abbreviations: IVW, inverse-variance weighted; OR, odds ratio; CI, confidence interval;SZ, Schizophrenia; BPD, Bipolar disorder. Table 5Heterogeneity and Horizontal Pleiotropy Assessments for Reverse-Direction MR AnalysesExposureOutcomeHeterogeneityPleiotropyQ statisticp valueMR–Egger Interceptp valueEndometriosisDepression5.750.45−0.010.72EndometriosisAnxiety12.610.050.010.72EndometriosisInsomnia2.870.83−0.020.68EndometriosisSZ6.730.350.090.14EndometriosisBPD4.410.62−0.030.41Abbreviations: SZ, Schizophrenia; BPD, Bipolar disorder. Heterogeneity and Horizontal Pleiotropy Assessments for Reverse-Direction MR Analyses Abbreviations: SZ, Schizophrenia; BPD, Bipolar disorder. Establishment and Validation of Cell Co-Culture Model Based on the MR findings suggesting depression as a potential risk factor for endometriosis, and given prior evidence linking depression to chronic stress responses and sustained sympathetic activation, we further examined the effects of sympathetic neuronal activity on endometriotic epithelial cells. SH-SY5Y cells were differentiated with 10 μM ATRA for 7 days to establish a sympathetic neuron-like model. ELISA assays showed that NE concentrations in supernatants from the differentiated group (588.72 ± 18.60 ng/mL) were significantly higher than those of the undifferentiated control group (443.45 ± 28.98 ng/mL), indicating successful induction of NE-secreting sympathetic neuron-like properties (Figure 2A). Consistent with a sympathetic neuron-like phenotype, immunofluorescence co-staining revealed the expression of key catecholaminergic enzymes, TH and DBH, following ATRA-induced differentiation (Figure S1). TH, the rate-limiting enzyme in catecholamine biosynthesis, and DBH, which catalyzes the conversion of dopamine to norepinephrine, serve as canonical markers for sympathetic neurons. Immunofluorescence analysis revealed markedly enhanced TH and DBH signals in ATRA-treated cells relative to untreated controls. In the induction group, TH immunoreactivity localized predominantly to the soma and neurites, with DBH exhibiting a congruent distribution, collectively indicating the acquisition of a catecholaminergic phenotype. Quantitative fluorescence intensity analysis further confirmed significant upregulation of both markers following ATRA treatment. The mean fluorescence intensity of TH increased to 82.65 ± 11.40 in the ATRA group compared with 56.65 ± 9.60 in controls (*P < 0.05). Likewise, DBH expression was significantly elevated (83.99 ± 6.06 vs. 66.25 ± 7.15, *P < 0.05; Figure S1B–C). Next, direct co-culture and conditioned medium systems were established. CCK-8 assays demonstrated that, in the conditioned medium system, 12Z cells exposed to sympathetic neuron CM exhibited significantly increased cell viability at 24 h and 48 h compared with 12Z monoculture controls (Figure 2B). In the direct co-culture system, an increased metabolic activity trend was observed at 48 h (Figure 2C). Figure 2Sympathetic neuron–like differentiation enhances norepinephrine secretion and promotes proliferation of endometriotic epithelial cells. (A) Quantification of norepinephrine (NE) levels in the supernatant of SH-SY5Y cells after 7 days of differentiation with 10 μM all-trans retinoic acid (ATRA). (B) Proliferation of 12Z cells assessed by CCK-8 assay following treatment with conditioned medium derived from differentiated sympathetic neuron–like cells (Co-culture) or control medium. Relative cell viability (OD450 nm) was measured at 0, 24, and 48 h. (C) Proliferation of 12Z cells in a Transwell direct co-culture system with differentiated sympathetic neuron–like cells. Relative cell viability was measured at 0, 24, and 48 h. A significant increase in proliferation was observed at 48 h in the co-culture group compared with the control group. Data are presented as mean ± SD from at least three independent biological experiments. Statistical analysis was performed using an unpaired two-tailed t-test for panel (A) and two-way ANOVA for panels (B) and (C). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.The image A shows a bar graph comparing norepinephrine concentration in nanograms per milliliter between ATRA-treated and control groups. The ATRA group shows a higher concentration, approximately 588.72 nanograms per milliliter, compared to the control group at 443.45 nanograms per milliliter, with a significance level of double asterisk. The image B shows a line graph of a conditioned medium assay with time in hours on the x-axis and CCK-8 relative ratio on the y-axis. The co-culture group shows increased cell viability over time, reaching about 3.5 at 48 hours, while the control group reaches about 2.5, with significance levels of triple and quadruple asterisks. The image C shows a line graph of a direct co-culture with time in hours on the x-axis and CCK-8 relative ratio on the y-axis. Both groups show increased viability, with the co-culture group reaching about 4 at 48 hours, slightly higher than the control group, with a single asterisk significance level.Three graphs showing norepinephrine concentration and cell viability in ATRA-treated and control groups over time. Sympathetic neuron–like differentiation enhances norepinephrine secretion and promotes proliferation of endometriotic epithelial cells. (A) Quantification of norepinephrine (NE) levels in the supernatant of SH-SY5Y cells after 7 days of differentiation with 10 μM all-trans retinoic acid (ATRA). (B) Proliferation of 12Z cells assessed by CCK-8 assay following treatment with conditioned medium derived from differentiated sympathetic neuron–like cells (Co-culture) or control medium. Relative cell viability (OD450 nm) was measured at 0, 24, and 48 h. (C) Proliferation of 12Z cells in a Transwell direct co-culture system with differentiated sympathetic neuron–like cells. Relative cell viability was measured at 0, 24, and 48 h. A significant increase in proliferation was observed at 48 h in the co-culture group compared with the control group. Data are presented as mean ± SD from at least three independent biological experiments. Statistical analysis was performed using an unpaired two-tailed t-test for panel (A) and two-way ANOVA for panels (B) and (C). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Transcriptomic Profiling Suggests Signaling Pathways Involved in Sympathetic Neuron–Driven Proliferation To explore the molecular programs associated with the pro-proliferative effect of sympathetic neuron–conditioned medium on endometriotic epithelial cells, RNA sequencing was performed on 12Z cells treated with conditioned medium (CM; n = 3, CM1–CM3) and control conditions (P; n = 3, P1–P3). Principal component analysis (PCA) showed clear separation between CM-treated and control samples along PC1 (25.46%) and PC2 (20.75%), indicating a reproducible transcriptomic shift induced by conditioned medium (Figure 3A). Differential expression analysis identified 327 differentially expressed genes (DEGs), including 124 upregulated and 203 downregulated genes. The volcano plot highlighted robust changes in representative transcripts, including upregulation of RPL34, RARRES1, PEPD, and OAS3, along with downregulation of MFN2 and NFKB2 (Figure 3B). Figure 3Transcriptomic profiling and functional enrichment analyses of 12Z cells treated with sympathetic neuron–derived conditioned medium. (A) Principal component analysis (PCA) of RNA-seq data showing clear separation between conditioned medium (CM)–treated 12Z cells and control (P) cells. (B) Volcano plot illustrating differentially expressed genes (DEGs) between CM-treated and control 12Z cells. Red dots represent significantly upregulated genes, blue dots represent significantly downregulated genes, and gray dots indicate non-significant genes. Differential expression thresholds were defined as |log2 fold change (FC)| ≥ 1 and P < 0.05. (C) Gene Ontology (GO) enrichment analysis of DEGs categorized into Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). The bar plots show the number of enriched genes in representative GO terms. DEGs were prominently enriched in processes related to signal transduction, immune response, cell adhesion, membrane components, and protein binding. (D) KEGG pathway enrichment analysis of DEGs. Bubble size represents the number of genes enriched in each pathway, and color indicates the significance level (P value).Image A shows a PCA graph with PCA1 (25.46%) and PCA2 (20.75%), separating CM-treated 12Z cells from control cells. CM1, CM2 and CM3 are distinct from P1, P2 and P3. Image B presents a volcano plot with log2 fold change and log10 P value, highlighting DEGs between CM-treated and control 12Z cells. Upregulated genes include RPL34, RARRES1, PEPD and OAS3, while MFN2 and NFKB2 are downregulated. Image C displays Gene Ontology enrichment analysis, categorizing genes into Biological Process, Cellular Component and Molecular Function, with enrichment in signal transduction, immune response, cell adhesion, membrane components and protein binding. Image D shows KEGG pathway enrichment analysis, with pathways like arachidonic acid metabolism, Ras signaling and PI3K-Akt signaling. Bubble size indicates gene count and color shows significance (P value).Four graphs showing PCA, volcano plot, Gene Ontology enrichment and KEGG pathway analysis of 12Z cells. Transcriptomic profiling and functional enrichment analyses of 12Z cells treated with sympathetic neuron–derived conditioned medium. (A) Principal component analysis (PCA) of RNA-seq data showing clear separation between conditioned medium (CM)–treated 12Z cells and control (P) cells. (B) Volcano plot illustrating differentially expressed genes (DEGs) between CM-treated and control 12Z cells. Red dots represent significantly upregulated genes, blue dots represent significantly downregulated genes, and gray dots indicate non-significant genes. Differential expression thresholds were defined as |log2 fold change (FC)| ≥ 1 and P < 0.05. (C) Gene Ontology (GO) enrichment analysis of DEGs categorized into Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). The bar plots show the number of enriched genes in representative GO terms. DEGs were prominently enriched in processes related to signal transduction, immune response, cell adhesion, membrane components, and protein binding. (D) KEGG pathway enrichment analysis of DEGs. Bubble size represents the number of genes enriched in each pathway, and color indicates the significance level (P value). GO functional enrichment analyses indicated that DEGs were over-represented in biological processes related to signal transduction, phosphorylation, cell adhesion, and immune-associated programs (eg., innate immune response, immune response, and cellular response to lipopolysaccharide) (Figure 3C). KEGG pathway enrichment further prioritized several canonical signaling cascades implicated in proliferation and survival, including the Ras, PI3K–Akt, and MAPK signaling pathways, as well as the cAMP signaling pathway (Figure 3D). Protein–Protein Interaction Network Identifies Key Hub Proteins To further explore potential interaction modules within the CM-responsive transcriptional program, a protein–protein interaction (PPI) network was constructed based on the identified DEGs using the STRING resource and visualized in Cytoscape. Several highly connected nodes were observed in the network, including CD244, PLEK, and GNGT2, suggesting potential hub-like positions within the DEG interaction landscape. Additional connected components included immune- and signaling-associated proteins such as CD70, HAVCR2, FLT3, and EGF (Figure 4), indicating a network architecture that may integrate immune modulation with growth factor–related signaling in conditioned medium–treated 12Z cells. Figure 4Protein–protein interaction (PPI) network analysis of differentially expressed genes in 12Z cells treated with sympathetic neuron–derived conditioned medium.The image shows a protein–protein interaction network analysis with various nodes connected by lines. Key nodes include CD244, PLEK and GNGT2, which are highlighted with larger circles indicating a higher degree of interaction. Other nodes such as CD70, HAVCR2, FLT3 and VEGFD are connected, forming a network architecture. The score threshold is represented by different line styles, ranging from 0.5 to 1.0. The degree of interaction is indicated by the size of the nodes, with larger nodes representing a degree of 4, medium nodes a degree of 2 and smaller nodes a degree of 1. The network suggests potential hub-like positions within the interaction landscape, integrating immune modulation with growth factor-related signaling in conditioned medium-treated 12Z cells.Network of protein interactions: nodes CD244, PLEK, GNGT2, etc., with different degrees and scores. Protein–protein interaction (PPI) network analysis of differentially expressed genes in 12Z cells treated with sympathetic neuron–derived conditioned medium. Discussion In this study, we performed bidirectional two-sample MR analyses to systematically evaluate the bidirectional associations between genetic liability to depression, anxiety disorders, insomnia, bipolar disorder, schizophrenia, and endometriosis. We found that genetic liability to depression was nominally associated with increased endometriosis risk in the primary inverse-variance weighted MR analysis (OR = 1.20, 95% CI 1.00–1.43, P = 0.04), but this signal was not supported after stringent Bonferroni correction (P threshold < 0.01), and sensitivity analyses yielded directionally consistent but non-significant estimates. No robust causal evidence was observed for other psychiatric traits, and reverse-direction MR analyses found no significant effect of endometriosis genetic liability on the risk of any of the included psychiatric disorders. These findings provide preliminary genetic evidence to clarify the directionality underlying the clinical comorbidity between these conditions. Globally, endometriosis has a prevalence of approximately 10% among women of reproductive age,31 and nearly half of affected patients present with comorbid psychiatric disorders including depression and anxiety in clinical settings. A meta-analysis included 9 studies confirmed that patients with endometriosis had a significantly increased risk of comorbid depression and anxiety (pooled relative risk RR = 2.93 for depression, pooled RR = 2.82 for anxiety). However, the authors also noted that limited by constraints including heterogeneity in study design and publication bias, previous observational studies have been unable to clarify the causal temporality and underlying mechanisms linking these conditions.32 Previous genetic studies have identified a significant genetic correlation between endometriosis and depression (rg = 0.27–0.36),6,33 as well as positive genetic correlations with anxiety disorders (rg = 0.33),6 eating disorders (rg = 0.616, and post-traumatic stress disorder (PTSD) (rg = 0.31).34 A study further demonstrated that genetic susceptibility loci for endometriosis are significantly enriched in pathways associated with pain and emotional regulation,8 providing a theoretical genetic basis for the comorbidity between endometriosis and psychiatric disorders. Notably, most previous MR studies on this topic have focused exclusively on depression and anxiety phenotypes, with no subsequent mechanistic validation performed.35 We verified the robustness of our findings through multidimensional sensitivity analyses, and identified that depression is an upstream causal risk factor for endometriosis. Building on the core risk factor identified via MR analyses, we proposed the scientific hypothesis that abnormal sympathetic activation serves as the key intermediate link through which depression mediates the elevated risk of endometriosis onset, and we performed preliminary validation of this hypothesis using in vitro experiments. In this study, we treated SH-SY5Y cells with 10 μM all-trans retinoic acid (ATRA) for 7 days to successfully establish a sympathetic neuron-like cell model with norepinephrine (NE) secretory function. This induction protocol has been validated in multiple studies,24,36 which enables stable differentiation of SH-SY5Y cells toward a sympathetic neuron-like phenotype, and is a well-established in vitro model for investigating sympathetic neuron function. Our cellular experimental results showed that conditioned medium from differentiated sympathetic neuron-like cells significantly promoted the increased the cell viability of 12Z endometriotic epithelial cells, which is highly consistent with the previous findings. Consistent with our results, multiple studies have confirmed that β2-adrenergic receptor (β2-AR), the primary endogenous receptor of norepinephrine, is highly expressed in endometriotic lesions, and that sympathetic activation via norepinephrine promotes the malignant biological behavior of endometriotic cells through β2-AR signaling.37 Notably, some studies have reported opposite findings, showing that β2-AR agonists can inhibit the viability and invasion of ectopic lesions via β2-AR signaling.38 This bidirectional effect of β2-AR activation may reflect context-dependent signaling outcomes under different experimental settings. Further transcriptome sequencing analysis revealed that differentially expressed genes (DEGs) were significantly enriched in the Ras, phosphatidylinositol 3 kinase and protein kinase B (PI3K-Akt), mitogen-activated protein kinase (MAPK), and cyclic adenosine monophosphate (cAMP) signaling pathways. Among these, the PI3K-Akt and MAPK pathways have been well documented as core pathways regulating the proliferation, survival, and invasion of endometriotic cells, with sustained activation observed in ectopic endometrial tissues.39,40 Meanwhile, cAMP, a key second messenger downstream of sympathetic neurotransmitters, has also been confirmed to participate in the pathological process of endometriosis. Collectively, our results provide a molecular-level explanation for sympathetic activation-mediated endometriosis progression at the transcriptomic level. Current clinical management of endometriosis remains dominated by hormonal therapy and surgical resection, with key unmet clinical challenges including high postoperative recurrence rates and a lack of curative treatment options. In addition to the sympathetic activation pathway validated in this study, previous research has uncovered other potential mechanisms underlying the comorbidity between these conditions, including dysregulation of the stress response mediated by hypothalamic-pituitary-adrenal (HPA) axis dysfunction,41 structural and functional abnormalities of the periaqueductal gray (PAG),42 and the vicious cycle of pain-depression central sensitization.43 These pathways may exert synergistic effects with sympathetic activation, collectively participating in the comorbid progression of endometriosis and mood disorders. Future studies should also explore the potential roles of burn-out and sexual dysfunction, which have been linked to both depression and endometriosis, in the comorbidity and progression of these conditions.44 Further in-depth mechanistic exploration of the crosstalk between these multiple pathways is warranted in future studies, which may identify novel targets for combined clinical intervention. Several limitations of this study should be acknowledged. First, all GWAS data included in this study were derived from individuals of European ancestry, which not only limits the generalizability of our findings to non-European populations including Asian populations, but also precludes assessment of the potential impact of ethnic differences on the genetic association between endometriosis and psychiatric disorders. Second, the in vitro functional validation in this study was performed using only the 12Z endometriotic epithelial cell line, lacking multidimensional validation using primary cells, other cell lines, and in vivo animal models. Furthermore, we did not conduct in-depth exploration of the core receptor and key regulatory genes underlying the pro-proliferative effect of sympathetic neurons on endometriotic cells. Third, our genetic analysis only covered common genetic variants, while rare variants and more complex genetic structures are increasingly recognized as contributors to psychiatric and reproductive traits.45,46 Meanwhile, the substantial clinical underdiagnosis of endometriosis may compromise the accuracy of phenotypic definition in the GWAS datasets, and we were unable to evaluate the potential roles of rare variants and complex genetic structures in subgroups with comorbid endometriosis and psychiatric disorders. Conclusion In bidirectional Mendelian randomization analyses, genetic liability to depression showed a nominal association with increased endometriosis risk that did not survive multiple testing correction; whereas reverse-direction analyses did not support endometriosis as a causal determinant of psychiatric disorders. In vitro sympathetic neuron-like models and transcriptomic analyses showed that sympathetic neuron–derived factors enhanced 12Z cell proliferation and were accompanied by enrichment of immune-related and signaling pathways. In exploratory bidirectional Mendelian randomization analyses, genetic liability to depression showed a nominal association with increased endometriosis risk that did not survive multiple testing correction.