677: Integration of multiple ‘omic datasets from a nested prospective observational study reveals linkage between the gut microbiome and metabolites in association with spontaneous preterm birth (sPTB)

Abstract

Reliable prediction of sPTB remains inadequate, in part due to its likely heterogeneous etiology. Previous efforts to identify a genetic link or biomarker that is highly predictive of PTB have typically relied upon a single ‘omic dataset (genome, transcriptome, proteome, metabolome) and have had limited success. However, integration of multiple ‘omic datasets can increase the discriminatory power, enabling more robust predictions of PTB that may better reflect and inform the underlying biology. Here, we sought to conduct the first integrated multi’omic analysis of metagenomics-transcriptomics-metabolomics data from a nested cohort derived from a larger prospectively enrolled cohort of n=526 pregnant women enrolled in the 1st trimester and at-risk for sPTB (the BaBs trial). 20 subjects (11 preterm, 9 term) were selected for a nested integrated multi’omics analysis after matching for baseline demographics. Multiple ‘omic datasets were generated from stool samples collected prospectively at multiple time points. High fidelity data included complex microbiome sequencing (16S rRNA gene & whole genome shotgun metagenomics), metatranscriptomics (host & microbial RNA), and metabolomics (host & microbial small molecules). Custom in-house pipelines were used to identify significant biomarkers and identify significant correlations with microbiota and their associated functional pathways that can inform its possible origin, function and role in PTB In total, >1.5 Tb of host and microbial data was generated and analyzed. Non-targeted mass spectroscopy identified 2297 unique features, with 8 putative biomarkers significantly differing between preterm and term pregnancies (Fig 1A, p<0.01, fold-change>2). These putative biomarkers demonstrated significant correlations with numerous taxa within the maternal gut, most notably Ruminococcus and Lachnospiraceae species (Fig 1B, p<0.05). Here, we demonstrate an integrative multi’omic approach for prediction of preterm birth in an initial nested case-cohort study. Expansion of this analysis into the larger cohort will provide a useful framework for identifying putative biomarkers of PTB and for better understanding the molecular mechanisms underlying premature labor.