Bidirectional prospective relationship of endometriosis with major depressive disorder: the mediating role of metabolic biomarkers

BACKGROUND: Endometriosis and major depressive disorder (MDD) pose significant global health burdens. Despite pathophysiological overlaps, existing epidemiological studies predominantly examine their unidirectional associations, often constrained by cross-sectional designs and small sample sizes. We investigated their bidirectional relationship and the mediating effect of circulating metabolites. METHODS: This prospective study followed 244,725 women from the UK Biobank cohort who were free of MDD at the baseline survey. Cox proportional hazard models estimated the hazard ratio (HR) of endometriosis status for incident MDD. Parallelly, a unified statistical analysis model was used to assess the risk of incident endometriosis according to MDD status among 62,255 premenopausal women without baseline endometriosis. Mediation analyses were performed to evaluate the effects of 22 circulating biomarkers. RESULTS: With a 13.6-year median follow-up, 13,292 incident MDD cases were documented. Compared with participants without endometriosis, those with endometriosis had HRs (95% confidence interval [CI]) of 1.39 (1.27-1.51) for MDD after adjusting for multiple confounders. In the parallel analysis, 947 incident endometriosis cases were identified. Compared with those without MDD, the HRs (95% CI) for endometriosis were 1.30 (1.08-1.57) in premenopausal patients with MDD. A cluster of circulating biomarkers, including triglycerides, high-density lipoprotein cholesterol, C-reactive protein, insulin-like growth factor-1, total bilirubin, alanine aminotransferase, glycated hemoglobin, and apolipoprotein B, was identified to explain the bidirectional prospective association between endometriosis and MDD, with the mediation proportion being 0.4% ∼ 4.1%. CONCLUSION: Endometriosis significantly increased the risk of MDD and vice versa in premenopausal women. Recognizing this bidirectional link may enhance prediction and treatment approaches for both conditions.