P47. First trimester urine and serum metabolomics to predict preeclampsia and gestational hypertension

Introduction New methods are required for improved prediction of preeclampsia. Metabolomics is the study of low molecular weight metabolites in tissues or biofluids. Objectives To evaluate early prediction of preeclampsia and gestational hypertension by 1H nuclear magnetic resonance (NMR) metabolomics. Methods 1H NMR spectra of urine and serum samples from 599 women with medium-to high risk of preeclampsia in weeks 11–14 of pregnancy were analyzed by principal component analysis and partial least squares discriminant analysis. Variable selection was applied on the metabolic profiles to find the best markers of prediction. Preeclampsia developed in 26 of the women and gestational hypertension in 21 women. Selected metabolites were combined with maternal characteristics in a logistic regression model. Results Using urine metabolomic profiles, preeclampsia could be predicted at 51.3% sensitivity, gestational hypertension at 40% sensitivity and both combined at 37% sensitivity at 10% false positive rate (FPR). Increased creatinine and decreased hippurate were the main discriminating metabolites in urine. Using serum metabolomic profiles, preeclampsia could be predicted at 15% sensitivity, gestational hypertension at 33% sensitivity and both combined at 30% sensitivity at 10% FPR. Women who later developed preeclampsia or gestational hypertension had increased serum VLDL and decreased HDL. Combining maternal markers (age >35 or Conclusion First trimester metabolomic profiles in urine and serum can independently predict hypertensive disorders of pregnancy. A panel of metabolites measured in urine in the first trimester may improve prediction rates for preeclampsia in combination with maternal biophysical markers.