Development and external validation of a metabolic and lipid-based model for predicting subsequent pregnancy in women...

Endometriosis is frequently associated with impaired fertility. Metabolic abnormalities and dyslipidemia may be relevant to reproductive dysfunction among women with endometriosis; however, their associations with subsequent reproductive outcomes and their predictive value remain unclear. This study aimed to examine the associations of metabolic and lipid-related indicators with subsequently recorded delivery and to develop and validate a prediction model among women with endometriosis.

We included 5,604 women of reproductive age with endometriosis from the UK Biobank and an independent cohort from the Affiliated Hospital of Qingdao University. The primary outcome in the UK Biobank was subsequently recorded delivery during follow-up, whereas the corresponding outcome in the external cohort was clinical pregnancy. Associations were evaluated using logistic regression, subgroup analysis, restricted cubic splines, and threshold-effect analysis. The UK Biobank cohort was randomly divided into training and internal validation sets at a 7:3 ratio. Predictors were selected using least absolute shrinkage and selection operator regression with 10-fold cross-validation, followed by multivariable logistic regression and nomogram construction. Model performance was evaluated in the training, internal validation, and external validation cohorts.

During a median follow-up of 13 years, 209 of the 5,604 women in the UK Biobank cohort had a subsequently recorded delivery. Higher body mass index, waist circumference, triglyceride levels, atherogenic index of plasma, triglyceride-glucose index, and non-high-density lipoprotein cholesterol-to-high-density lipoprotein cholesterol ratio were independently associated with a lower likelihood of subsequently recorded delivery. The final prediction model incorporated age, body mass index, Townsend deprivation index, current smoking status, educational attainment, total cholesterol, and triglyceride-glucose index. The model demonstrated areas under the receiver operating characteristic curve of 0.869 in the training set, 0.854 in the internal validation set, and 0.851 in the external validation cohort.

Less favorable metabolic and lipid profiles were associated with a lower likelihood of subsequently recorded delivery among women with endometriosis. The model may provide complementary information for reproductive risk stratification. However, differences in outcome ascertainment between cohorts and the limited number of events warrant cautious interpretation. Further prospective, multicenter validation is required before clinical application.