Amniotic fluid metabolic biomarkers of fetal physiology and pregnancy success

Abstract

Amniotic fluid (AF) profiling provides a minimally invasive window into early fetal physiology. We characterized the AF metabolome from first trimester (Day 68) Holstein dairy heifers (n = 45), considering fetal sex, conception method [in vitro fertilization vs. artificial insemination (AI)], and eventual pregnancy outcome as key variables. Multivariate statistics uncovered differentially abundant metabolites for each comparison—including markers that preceded spontaneous abortion—independently of recipient age, weight, gestation length, or fetal genetics. Thereafter, a machine learning algorithm using panels of six metabolites accurately predicted fetal sex (AUROC = 0.76; P = 0.023) and pregnancy viability (AUROC = 0.81; P = 0.018), while corroborating conception method (AUROC = 0.91; P = 0.001). External validation using AF (Day 42) from an independent cohort of beef heifers (n = 22) reproduced the fetal sex classifier with similarly high sensitivity and specificity (AUROC = 0.85, P = 0.029). These findings reveal metabolic signatures that forecast fetal sex and pregnancy viability, while confirming distinct metabolic imprints of assisted-conception modalities. These data lay the groundwork for next-generation AF prenatal diagnostics in veterinary and human obstetrics.