THURSDAY, June 30, 2016 (HealthDay News) -- A metabolomics signature can predict the transition from gestational diabetes mellitus (GDM) to type 2 diabetes (T2D), according to a study published online June 23 in Diabetes.
Erica Gunderson, Ph.D., of Kaiser Permanente Northern California, Division of Research, in Oakland, Calif., enrolled a prospective cohort of 1,035 women with GDM into the SWIFT study and screened them for type 2 diabetes at six to nine weeks postpartum and then annually for two years. Of 1,010 women without diabetes at enrollment, 113 went on to develop diabetes within two years and another 17 beyond two years after enrollment. Amina Allalou, from the University of Toronto, and colleagues analyzed the specimens from the cohort for metabolic markers.
Metabolomics were conducted with baseline fasting plasma in a nested case-control study involving incident T2D cases matched to non-cases by age, pre-pregnancy body mass index, and race/ethnicity. Twenty-one metabolites that differed significantly by incident T2D status were identified. Machine learning optimization was able to predict T2D incidence with a discriminative power of 83.0 and 76.9 percent in the training set and an independent testing set, respectively.
"Our metabolomics signature predicted T2D incidence from a single fasting sample," the authors write. "This study represents the first metabolomics study of the transition from GDM to T2D validated in an independent testing set, facilitating early intervention."