Yunbei Rao, Xuan Wang, Qingling Li, Lingling Yang, Xiaoyun Xiong, Dongshan Han, Chuanzhong Yang, Xueyu Chen
High human milk feeding was associated with lower BPD severity and a less activated inflammatory profile at 36 weeks PMA. IL-1Ra- and CXCL10-related inflammatory pathways may partially account for this association, although the observational design precludes definitive causal inference.
OBJECTIVE: This study investigates the mediating role of inflammatory cytokines in the protective effect of human milk against bronchopulmonary dysplasia (BPD) in preterm infants.
METHODS: This prospective cohort study included 73 infants born at < 28 weeks' gestation.Infants were classified into high-dose human milk and high-dose formula groups according to feeding patterns at 36 weeks postmenstrual age (PMA). Cross-sectional group comparisons of inflammatory factors at postnatal day 7, day 28, and PMA 36 weeks were examined by Mann-Whitney U test, while longitudinal changes in plasma inflammatory cytokines were evaluated using generalized estimating equations (GEE), and mediation analysis was performed using PROCESS version 4 to assess the potential indirect effects of inflammatory cytokines on the association between human milk exposure and BPD severity.
RESULTS: Infants in the high-dose human milk group had a significantly lower incidence of grade II/III BPD and lower plasma concentrations of CXCL10, IL-1Ra, and IL-13 at 36 weeks PMA than those in the high-dose formula group. Longitudinal GEE analysis showed significant temporal changes in these cytokines but no significant group × time interactions. Mediation analysis demonstrated a significant joint indirect effect through IL-1Ra and CXCL10 (effect = -0.245, 95% CI -0.487 to -0.046), accounting for 34.9% of the total association between human milk exposure and BPD severity.
CONCLUSION: High human milk feeding was associated with lower BPD severity and a less activated inflammatory profile at 36 weeks PMA. IL-1Ra- and CXCL10-related inflammatory pathways may partially account for this association, although the observational design precludes definitive causal inference.