Prediction of gestational diabetes mellitus based on an analysis of amniotic fluid by capillary electrophoresis

Abstract

Aim: To detect gestational diabetes mellitus biomarkers in human amniotic fluid collected for age-related genetic testing using capillary electrophoresis and a sophisticated data analysis methodology. Materials & methods: Amniotic fluid samples were separated by capillary electrophoresis. Samples were classified using a genetic algorithm with Bayesian benefit function. The best model maximized the sensitivity and specificity and employed a leave-one-out cross-validation strategy. Results: Gestational diabetes mellitus (GDM; n = 14) was distinguished from non-GDM (n = 95) with 86% sensitivity and 99% specificity using two wavelets. These wavelets were located in the unresolved protein region and on the edge of the maternally derived albumin peak. Conclusion: GDM is a maternal pathology; however, it was shown that it alters the biochemical profile of amniotic fluid. Testing for GDM is normally carried out at 24–28 weeks, but changes can be detected at 15 weeks gestation, suggesting that GDM onset occurs early in gestation.