Author: Bruce Ling, PhD, Assistant Professor, Department of Surgery, School of Medicine, Stanford University
Pregnancy triggers longitudinal metabolic alterations in women to allow precisely programmed fetal growth. Comprehensive characterization of such a “metabolic clock” of pregnancy may provide a molecular reference in relation to studies of adverse pregnancy outcomes. However, a high-resolution temporal profile of metabolites along a healthy pregnancy remains to be defined.
Two independent, normal pregnancy cohorts with high-density weekly urine sampling (discovery: 478 samples from 19 subjects in California; validation: 171 samples from 10 subjects in Alabama) were studied. Urine samples were profiled by liquid chromatography-mass spectrometry (LC-MS) for untargeted metabolomics, which was applied for gestational age dating and prediction of time to delivery.
5,473 urinary metabolic features were identified. Partial least-squares discriminant analysis on features with robust signals (n = 1,716) revealed that the samples were distributed based on the first two principal components according to their gestational age. Pathways of bile secretion, steroid hormone biosynthesis, pantothenate and CoA biosynthesis, benzoate degradation, and phenylpropanoid biosynthesis were significantly regulated, which was collectively applied to discover and validate a predictive model that accurately captures the chronology of pregnancy. With six urine metabolites (acetylcholine, estriol-3-glucuronide, dehydroepiandrosterone sulfate, α-lactose, hydroxyexanoy-carnitine, and L-carnitine), models were constructed based on gradient-boosting decision trees to date gestational age in high accordance with ultrasound results, and to accurately predict time to delivery.
This study characterizes the weekly baseline profile of the human pregnancy metabolome, which provides a high-resolution molecular reference for future studies of adverse pregnancy outcomes.
该项研究基于两个独立的正常妊娠组，每周进行高密度尿检 (研究组: 来自加州19名抽样对象的478份样本; 验证组: 来自阿拉巴马州10名抽样对象的171份样本)。尿液样本经过液相色谱-质谱(LC-MS)进行非靶向代谢组学分析，用于检测孕龄和预测分娩时间。
在研究中，5,473例尿液代谢组特征得到了鉴定。对具有稳定信号的特征(n = 1716)进行偏最小二乘判别分析的结果表明，样本在前两个主成分的维度上根据胎龄分布。孕期的胆汁分泌、类固醇激素生物合成、泛酸和辅酶A生物合成、苯甲酸降解、以及苯丙素生物合成的途径受到明显调节，这些发现共同揭开并验证了一个能够准确捕捉妊娠时间的预测模型。利用六种尿液代谢物(乙酰胆碱、雌三醇-3-葡萄糖醛酸苷、硫酸脱氢表雄酮、α-乳糖、羟色胺-肉碱、以及L-肉碱)建立的模型基于梯度提升决策树来测算孕龄，其结果与超声影像结果高度一致，并能够准确预测分娩时间。
Gestational Dating by Urine Metabolic Profile at High Resolution Weekly Sampling Timepoints: Discovery and Validation. K. G. Sylvester, S. Hao, Z. Li, Z. Han, L. Tian, S. Ladella, R. J. Wong, G. M. Shaw, D. K. Stevenson, H. J. Cohen, J. C. Whitin, D. B. McElhinney, and X. B. Ling. Front. Mol. Med, 27 April 2022 https://doi.org/10.3389/fmmed.2022.844280`