Maternal Plasma Proteins Associated with Birth Weight: A Longitudinal, Large Scale Proteomic Study

Small infants for gestational age (SGA) and large infants for gestational age (LGA) have increased risk of complications during delivery and later in life. Prediction of the fetal weight is currently limited to biometric parameters obtained by ultrasound scans that can be imprecise. Biomarkers of fetal growth would be crucial for tailoring clinical management and optimizing outcomes for the mother and child. Seventy pregnant women participated in the current study, including 58, 7, and 5 giving birth to adequate for gestational age (AGA), SGA, and LGA infants, respectively. Maternal venous blood was drawn at gestational weeks 12-19, 21-27, and 28-34 and quantified for nearly 5000 proteins on the SomaLogic platform. We used machine learning algorithms with leave-one-out cross-validation to construct multiprotein models for prediction of birth weight groups. Random forest models using only 20 predefined proteins (selected by moderated <i>t</i> tests) were able to predict LGA with good discrimination (AUC > 0.8) at all three visits, while prediction of SGA was less successful. Protein differential abundance analysis revealed 148 proteins with higher abundance in LGA compared to AGA pregnancies, while only four proteins were differentially abundant between the SGA and AGA. The principal findings indicate that the maternal plasma proteome may hold potential biomarkers of LGA.