The predictive value of cord blood 25-hydroxyvitamin D and vitamin A levels for bronchopulmonary dysplasia in preterm infants

Background Despite substantial progress in the treatment of bronchopulmonary dysplasia (BPD), its overall incidence remains high, with increasing rates of moderate-to-severe cases. Reliable early biomarkers are urgently needed to facilitate timely intervention and optimize long-term prognosis. Methods This prospective cohort study enrolled 144 preterm infants with gestational age ≤32 weeks admitted to the neonatal intensive care unit from March 2021 to December 2024. Umbilical arterial blood was collected to detect 25(OH)D, vitamin A (VA), IL-6, PCT, and CRP levels. Infants were classified into BPD ( n = 63) and non-BPD ( n = 81) groups, and the BPD group was further stratified into Grade I, II, and III subgroups according to oxygen requirement and respiratory support. Group differences in biomarker levels and clinical variables were analyzed. Univariate and multivariate logistic regression models were constructed to identify independent BPD predictors. ROC curves, DeLong test, NRI, and IDI were applied to evaluate incremental predictive performance. Results Cord blood 25(OH)D and VA were significantly decreased, while PCT was elevated in BPD infants (all P < 0.05). Within BPD subgroups, 25(OH)D gradually decreased and IL-6 increased with disease severity. Multivariate regression confirmed that cord blood 25(OH)D remained an independent protective factor for BPD after adjustment for gestational age, birth weight, and caffeine exposure (OR = 0.854, P < 0.001), whereas VA showed no independent predictive significance. The baseline clinical model yielded an AUC of 0.784. The addition of 25(OH)D significantly improved predictive efficacy to AUC = 0.857 (ΔAUC = 0.071, P = 0.001), with robust NRI and IDI improvements (both P < 0.001). Further inclusion of VA produced no additional predictive benefit. Conclusion Umbilical arterial 25(OH)D is a stable independent predictive biomarker for BPD and provides valuable incremental predictive value beyond conventional clinical parameters, supporting early risk stratification in preterm infants, whereas VA exhibits limited predictive utility.