O-258 Nuclear magnetic resonance profiling of the vaginal metabolome in adverse early pregnancy outcomes

Abstract

Abstract Study question Is the vaginal metabolome predictive of early pregnancy outcomes? Summary answer The vaginal metabolome is moderately predictive of live normally-sited pregnancies when compared to a pregnancy with an adverse outcome. What is known already Approximately 15% of all pregnancies are lost in the first trimester and a further 1-2% are ectopic. To date, no non-invasive and clinically useful tests have been developed to differentiate live normally-sited pregnancies from non-viable and ectopic pregnancies. Metabolomics is the comprehensive analysis of small molecules in biological samples used to identify disease biomarkers. The vaginal metabolome has previously been shown to be predictive of later adverse pregnancy outcomes including preterm birth. In this study, we aimed to identify biomarkers of live normally-sited pregnancies when compared to non-viable intrauterine and ectopic pregnancies using NMR metabolomic profiling of vaginal swabs (LVS). Study design, size, duration This was a prospective cohort study (EXPEDITE). 222 pregnant women <10 weeks’ gestation with pain and/or bleeding were recruited over a period of 36 months at the Liverpool Women’s Hospital, a tertiary referral centre. Pregnancy outcomes were categorised according to ESHRE guidelines as live normally-sited pregnancy, miscarriage, tubal ectopic pregnancy or pregnancy of unknown location (PUL). Participants/materials, setting, methods Participants consented to self-collecting a LVS, which was extracted in acetonitrile and analysis by NMR spectroscopy; 1D 1H and 2D 1H-13C. Spectra were acquired at 25 °C on a 700 MHz spectrometer. Relative metabolite abundances underwent univariate and multivariate statistical analysis using MetaboAnalyst 5.0 and custom-built R scripts (p-value false discovery rate adjusted). Area under receiver operating characteristic curve (AUC) values were used to evaluate predictive models. Main results and the role of chance Our cohort of 222 women were categorised in to 100 (45 %) live normally-sited pregnancies, 29 (13 %) tubal ectopic pregnancies, 52 (23 %) miscarriages and 31 (14%) pregnancies of unknown location (PUL). Ten spectra from rare or uncertain pregnancy outcomes (e.g. molar pregnancy) were excluded. Univariate statistical analysis of all four groups identified isoleucine levels as significantly higher (p < 0.05) in live normally-sited pregnancies compared with all groups with adverse pregnancy outcome groups. Multivariate partial least squares discriminant analysis (PLS-DA) models of all groups failed to accurately predict pregnancy outcomes. To enhance the predictive models, miscarriage, tubal ectopic pregnancy and PUL groups were combined and compared to live normally-sited pregnancies. Univariate analysis revealed levels of several metabolites (including isoleucine, tyrosine and maltose) as significantly altered in live normally-sited pregnancies compared with all other outcomes. PLS-DA models demonstrated modest discrimination between live normally-sited pregnancies versus other outcomes (AUC = 0.75). In summary, this study has identified a vaginal metabolite profile and potential biomarkers associated with live normally-sited pregnancy in the first trimester. Limitations, reasons for caution The number of tubal ectopic pregnancies and PULs were relatively low compared to other outcomes. Additionally, several factors that may influence the vaginal metabolome, including infection and the vaginal microbiome, were not characterised in this study. Wider implications of the findings These findings may be useful in the development of diagnostic tests to confirm an intrauterine pregnancy and exclude potential life-threatening pregnancies. It is well established that the microbiome contributes to the vaginal metabolome. As such, our findings will inform future correlative studies of vaginal microbiome and metabolome in early pregnancy. Trial registration number NA