RESUMO
This study aimed to investigate whether fetal growth trajectories (FGTs) could predict early childhood development, indicate intrauterine metabolic changes, and explore potential optimal and suboptimal FGTs. FGTs were developed by using an unsupervised machine-learning approach. Children's neurodevelopment, anthropometry, and respiratory outcomes in the first 6 years of life were assessed at different ages. In a subgroup of participants, we conducted a metabolomics analysis of cord blood to reveal the metabolic features of FGTs. We identified 6 FGTs: early decelerating, early decelerating with late catch-up growth, early accelerating, early accelerating with late medium growth, late decelerating, and late accelerating. The early accelerating with late medium growth pattern might be the optimal FGT due to its associations with better psychomotor development, mental development, intelligence quotient, and lung function and a lower risk of behaviour and respiratory problems. Compared with the optimal FGT, early decelerating and late decelerating FGTs were associated with poor neurodevelopment and lung function, while early accelerating FGT was associated with more severe autistic symptoms, poor lung function, and increased risks of overweight/obesity. Metabolic alterations were enriched in amino acid metabolism for early decelerating and late decelerating FGTs, whereas altered metabolites were enriched in lipid metabolism for early accelerating FGT. These findings suggest that FGTs are predictors of early life development and may indicate intrauterine adaptive metabolism. The discovery of optimal and suboptimal FGTs provides potential clues for the early identification and intervention of fetal origin dysplasia or disease, but further research on related mechanisms is still needed.