The predictive value of homocysteine levels for pregnancy outcomes in PCOS women: A retrospective study

Abstract

Background Polycystic ovary syndrome (PCOS) poses a challenge in reproductive medicine, particularly in understanding the role of hyperhomocysteinemia in its pathogenesis, especially among women experiencing recurrent pregnancy loss (RPL). Objective We aimed to assess the predictive value of homocysteine (HCY) levels for pregnancy outcomes in women with PCOS. Materials and Methods A consecutive series of PCOS-diagnosed patients participated in this prospective study. Pregnancy outcomes were collected prospectively. Women with PCOS were divided into two groups: Group I (n = 67) experienced RPL, while Group II (n = 162) had successful live births. Group I was further categorized into subgroups A and B based on insulin resistance status. Basic participant information and serum HCY levels were recorded. A logistic regression model was constructed, with model selection based on Akaike information criteria (AIC) using backward stepwise selection. Receiver operating characteristic (ROC) curves were plotted to evaluate model sensitivity and specificity. Optimal cut points for predicted probabilities were determined using the Youden index. Model performance was assessed using five-fold cross-validation, repeated ten times. Results Patients without live births were more likely to have a history of adverse pregnancy outcomes (78% vs. 22%, p < 0.05). They also exhibited higher baseline HCY levels (7.7 vs. 6.85, p < 0.05) compared to those with live births. Univariate logistic regression revealed significant associations between live birth and adverse pregnancy history and HCY levels. Backward stepwise selection identified adverse pregnancy history (odds ratio [OR] = 12.92, 95% CI: 6.46–27.40, p = 3.89×10− 12), insulin resistance (OR = 2.08, 95% CI: 1.04–4.26, p = 0.041), and HCY (OR = 1.34, 95% CI: 1.09–1.68, p = 0.008) as important predictors of live birth. The area under the ROC curve (AUC) was 0.839, indicating strong discrimination ability. The optimal cut point for predicted probabilities, determined by the Youden index, was 0.258, resulting in a sensitivity of 0.772 and specificity of 0.806. Conclusion This study underscores the predictive value of homocysteine levels in assessing pregnancy outcomes among women with PCOS. Combining homocysteine levels with medical history and insulin resistance status serves as a robust predictor of live birth outcomes.