Development and predictive value of the empirical dietary inflammatory index for pregnant women in the second trimester

Background: We aim to develop and validate the empirical dietary inflammatory index for pregnant women in the second trimester (EDII-PW) and assess its predictive value for maternal and neonatal outcomes. Methods: We utilized data from three sub-cohorts conducted in Hefei, China, including the Maternal and Infant Health Cohort (MIH-I: 2015-2019) with 5,052 pregnant women, MIH-II (2018-2022) with 1,932 pregnant women, and MIH-III (2023-2024) with 1,225 pregnant women, all recruited between gestational ages of 16 and 23 weeks. The development and validation of the EDII-PW index were based on three plasma inflammatory biomarkers (IL-6, hsCRP, and TNF-α) in MIH-I. In MIH-II, three additional biomarkers (E-selectin, VCAM-1, and ICAM-1) were included for independent biomarker validation. In MIH-III, six biomarkers (hsCRP, IL-6, TNF-α, E-selectin, VCAM-1, and ICAM-1) were measured in 1,225 participants and were used to derive inflammatory and endothelial biomarker profiles for the outcome prediction analyses. Results: The EDII-PW was significantly associated with plasma concentrations of inflammatory biomarkers. In MIH-I, participants in the highest quartile of EDII-PW had significantly higher levels of hsCRP, IL-6, and TNF-α compared to those in the lowest quartile. Similar associations were observed for E-selectin, VCAM-1, and ICAM-1 in MIH-II. In MIH-III, ROC curve analyses suggested that multivariable models containing the EDII-PW may discriminate maternal and neonatal outcomes, with AUC values as follows: 0.81 for premature rupture of membranes (PROM), 0.83 for gestational diabetes mellitus (GDM), 0.74 for endothelial dysfunction, 0.81 for inflammatory status, 0.80 for low birth weight (LBW), 0.82 for small for gestational age (SGA), 0.83 for preterm birth, and 0.76 for gestational hypertension. Conclusion: The EDII-PW showed reproducible associations with inflammatory biomarkers across independent sub-cohorts and potential utility as a component of multivariable models for adverse maternal and neonatal outcomes.