Construction and validation of a nomogram for predicting intrahepatic cholestasis of pregnancy (ICP) risk: a prospective cohort study

Abstract

Intrahepatic cholestasis of pregnancy (ICP) is a pregnancy-specific liver disorder. Serum total bile acids (TBA) are the primary diagnostic marker for ICP, but their concentrations are affected by environmental, dietary, and genetic factors, resulting in significant interindividual variation. This study aimed to develop and validate a clinically applicable nomogram for the early prediction of ICP risk. This prospective cohort study enrolled pregnant women attending prenatal care at Shanghai Public Health Clinical Center from June 2022 to June 2024. A total of 391 participants (184 ICP cases and 207 controls) were included. Clinical characteristics and serum biomarker concentrations were collected. Univariate analysis, LASSO regression, and multivariate logistic regression were used to identify independent predictors for ICP risk, which were then incorporated into a nomogram. Multicollinearity was assessed using the variance inflation factor (VIF), with VIF < 2 indicating its absence. Internal validation was performed using bootstrap resampling with 1,000 replicates. Model performance was evaluated for discrimination (AUC), calibration (Hosmer–Lemeshow test with calibration curve), and clinical utility (decision curve analysis and clinical impact curve). Seven independent predictors for ICP risk were identified via multivariate logistic regression (all P  < 0.05): age ≥ 35 years (OR = 5.276, 95% CI 1.155–24.727), ALT (OR = 1.082, 95% CI 1.023–1.163), TG (OR = 4.538, 95% CI 2.236–10.723), Apo A (OR = 0.001, 95% CI 0.000–0.089), FIB (OR = 3.460, 95% CI 1.288–10.726), sICAM-1 (OR = 1.067, 95% CI 1.036–1.112), and PLTP (OR = 1.001, 95% CI 1.000–1.001). Apo A was negatively associated with ICP, while the other six factors showed positive associations. No multicollinearity was detected (all VIF < 2). The nomogram exhibited excellent discrimination (AUC = 0.991) and optimal calibration (Hosmer-Lemeshow test: χ² = 0.991, P  = 0.998). Decision curve analysis confirmed its clinical utility. The nomogram provides a reliable tool for the individualized risk assessment of ICP. It enables early identification of high-risk pregnant women, facilitates the optimization of perinatal monitoring and intervention strategies and may thereby improve maternal and neonatal outcomes.