Postnatal growth and metabolic blood biomarkers in preterm infants developing reversible retinopathy of prematurity

Purpose To investigate predictive potential of growth and metabolic blood biomarkers in the development of milder, reversible retinopathy of prematurity (ROP) stages. Methods Biomarkers were obtained from blood samples collected every second postnatal week in a prospective, longitudinal cohort study including 108 infants born with a gestational age (GA) <32 weeks in four hospitals in the Capital Region of Denmark, 2018-2019. ROP diagnoses were obtained from the electronic medical record system together with demographic, clinical and laboratory data. Measurement of glucose was summarised as mean and SD for every postnatal week and growth was summarised as increment in weight, head circumference (biparietal diameter) and length every postnatal week. The predictive potential of each biomarker and each marker of growth in turn were evaluated in univariate receiver operating characteristics curve analyses and in multivariate analyses including GA and small for gestational age (SGA) as known predictors. Results The strongest isolated postnatal predictor of ROP was weight gain at the second postnatal week with an area under the curve (AUC) of 0.80 (95% CI: 0.70 to 0.89). However, it only added insignificantly to the AUC (0.85; 95% CI: 0.76 to 0.93, adj. p=0.89) compared with GA and SGA alone (AUC=0.80, 95% CI: 0.70 to 0.90). Mean glucose in PNA weeks 1-4, glycaemic variability as measured by glucose SD weeks 1-3 PNA, and concentrations of adiponectin/glucose (mean) ratio were also associated with ROP diagnosis (AUCs ranging from 0.679 to 0.77) but did also not contribute significantly to the AUC compared with GA and SGA alone. Conclusions Postnatal growth and metabolic blood biomarkers were significantly associated with milder, reversible ROP, but none of these gave prediction over and above GA and SGA. Due to the small sample sizes, potential predictors could only be investigated in univariate analyses. Larger studies are needed to fully explore the predictive potential of all the biomarkers.