# Maternal amino acid metabolites during pregnancy and preterm birth: results from two prospective cohort studies

## TL;DR

This study identifies maternal amino acid metabolites linked to preterm birth and shows they can improve risk prediction during pregnancy.

## Contribution

The study introduces a new amino acid preterm prediction score (APPS) validated across two cohorts for preterm birth risk assessment.

## Key findings

- Higher Ile-Val in mid-pregnancy is associated with longer gestational duration and lower preterm birth risk.
- Higher Val-Gly is linked to shorter gestational duration and increased preterm birth risk.
- The APPS model outperforms clinical models in predicting preterm birth.

## Abstract

Precise regulation of amino acid concentrations in the maternal–fetal circulation is essential for fetal development, but the association of maternal amino acid metabolites with gestational duration and preterm birth risk remains obscure. Therefore, we aimed to identify biomarkers consistently associated with gestational duration and preterm birth risk across cohorts.

The discovery cohort is a nested case–control study based on the Tongji-Huaxi-Shuangliu Birth Cohort (THSBC), which recruited pregnant women in early pregnancy (6–15 weeks of gestation) between 2017 and 2020 and followed the women during middle and late pregnancy periods. In this discovery cohort we performed trajectory analysis for the amino acid metabolism related metabolites across 3 pregnancy periods. Regression modeling was performed to identify amino acid metabolites that were associated with gestational duration and preterm birth. We also developed an amino acid preterm prediction score (APPS) based on the identified metabolites to stratify patients according to their risk of preterm birth. Validation of the identified metabolites and the APPS was performed in the Westlake Precision Birth Cohort study (WeBirth) consisting of pregnant women with gestational diabetes recruited in mid pregnancy (22–28 weeks’ gestation) between 2019 and 2023.

A total of 723 participants (241 preterm cases) in the THSBC and 1597 participants (96 preterm cases) in the WeBirth were included. In the THSBC, each 1 standard deviation increase in Ile-Val in mid pregnancy was associated with an increment of 0.19 (95% CI, 0.03, 0.35) week for gestational duration, while Val-Gly was inversely associated with gestational duration (− 0.20; 95% CI, − 0.34, − 0.06). In addition, lower Ile-Val and higher Val-Gly were associated with higher risk of preterm birth (OR, 0.72; 95% CI, 0.57, 0.91; OR, 1.25; 95% CI, 1.03, 1.51, respectively). These associations were consistent in the WeBirth cohort. Compared to the clinical model, the incorporation of APPS exhibited better performance in predicting preterm birth.

The findings suggest that circulating amino acid metabolites may serve as biomarkers for preterm birth prediction. Maternal amino acid metabolites may have potential clinical utility in improving prenatal risk assessment.

The online version contains supplementary material available at 10.1186/s12916-026-04710-5.