Brian J Koos1, Jeffrey A Gornbein2. 1. Departments of Obstetrics and Gynecology, David Geffen School of Medicine at University of California Los Angeles, Los Angeles, CA. Electronic address: bkoos_1999@yahoo.com. 2. Department of Medicine, Computational Core, David Geffen School of Medicine at University of California Los Angeles, Los Angeles, CA; Department of Medicine, Statistical Core, David Geffen School of Medicine at University of California Los Angeles, Los Angeles, CA.
Abstract
BACKGROUND: Aberrant fetal programming in gestational diabetes mellitus seems to increase the risk of obesity, type 2 diabetes, and cardiovascular disease. The inability to accurately identify gestational diabetes mellitus in the first trimester of pregnancy has thwarted ascertaining whether early therapeutic interventions reduce the predisposition to these prevalent medical disorders. OBJECTIVE: A metabolomics study was conducted to determine whether advanced analytical methods could identify accurate predictors of gestational diabetes mellitus in early pregnancy. STUDY DESIGN: This nested observational case-control study was composed of 92 gravidas (46 in the gestational diabetes mellitus group and 46 in the control group) in early pregnancy, who were matched by maternal age, body mass index, and gestational age at urine collection. Gestational diabetes mellitus was diagnosed according to community standards. A comprehensive metabolomics platform measured 626 endogenous metabolites in randomly collected urine. Consensus multivariate criteria or the most important by 1 method identified low-molecular weight metabolites independently associated with gestational diabetes mellitus, and a classification tree selected a subset most predictive of gestational diabetes mellitus. RESULTS: Urine for both groups was collected at a mean gestational age of 12 weeks (range, 6-19 weeks' gestation). Consensus multivariate analysis identified 11 metabolites independently linked to gestational diabetes mellitus. Classification tree analysis selected a 7-metabolite subset that predicted gestational diabetes mellitus with an accuracy of 96.7%, independent of maternal age, body mass index, and time of urine collection. CONCLUSION: Validation of this high-accuracy model by a larger study is now needed to support future studies to determine whether therapeutic interventions in the first trimester of pregnancy for gestational diabetes mellitus reduce short- and long-term morbidity.
BACKGROUND: Aberrant fetal programming in gestational diabetes mellitus seems to increase the risk of obesity, type 2 diabetes, and cardiovascular disease. The inability to accurately identify gestational diabetes mellitus in the first trimester of pregnancy has thwarted ascertaining whether early therapeutic interventions reduce the predisposition to these prevalent medical disorders. OBJECTIVE: A metabolomics study was conducted to determine whether advanced analytical methods could identify accurate predictors of gestational diabetes mellitus in early pregnancy. STUDY DESIGN: This nested observational case-control study was composed of 92 gravidas (46 in the gestational diabetes mellitus group and 46 in the control group) in early pregnancy, who were matched by maternal age, body mass index, and gestational age at urine collection. Gestational diabetes mellitus was diagnosed according to community standards. A comprehensive metabolomics platform measured 626 endogenous metabolites in randomly collected urine. Consensus multivariate criteria or the most important by 1 method identified low-molecular weight metabolites independently associated with gestational diabetes mellitus, and a classification tree selected a subset most predictive of gestational diabetes mellitus. RESULTS: Urine for both groups was collected at a mean gestational age of 12 weeks (range, 6-19 weeks' gestation). Consensus multivariate analysis identified 11 metabolites independently linked to gestational diabetes mellitus. Classification tree analysis selected a 7-metabolite subset that predicted gestational diabetes mellitus with an accuracy of 96.7%, independent of maternal age, body mass index, and time of urine collection. CONCLUSION: Validation of this high-accuracy model by a larger study is now needed to support future studies to determine whether therapeutic interventions in the first trimester of pregnancy for gestational diabetes mellitus reduce short- and long-term morbidity.