“Gestational diabetes is typically identified between 24 and 28 weeks of pregnancy by checking blood glucose, but unfavorable metabolic changes in pregnant women and their developing fetuses could occur from early pregnancy,” explains Yeyi Zhu, PhD. “Thus, methods to identify gestational diabetes sooner could benefit patients by giving them more time during their pregnancies to manage their blood sugar and health.”
For a study published in Diabetes, Dr. Zhu and colleagues sought to contribute to a growing area of study known as diabetes metabolomics, which focuses on identifying metabolites, which are end-products of bodily processes such as the use of energy. “People who have certain metabolites in their blood may be at higher risk for developing gestational diabetes,” Dr. Zhu says. “We wanted to contribute to work into the mechanisms and prediction of gestational diabetes by utilizing our unique access to a relatively large, multi-racial and ethnic group of pregnant individuals who participated in previous studies; we also used robust data in metabolomics and clinical factors. Our study team had access to patients’ fasting blood samples and could track their health status over time through Kaiser Permanente EHRs.”
Two Panels of Metabolites Were Developed and Compared
Dr. Zhu and colleagues worked with the NIH West Coast Metabolomics Center at University of California Davis to conduct laboratory analysis of blood specimens. They used patient data from the Pregnancy, Environment and Lifestyle Study (PETALS), a long-term study tracking lifestyle factors and health status of 3,346 pregnant patients in the Kaiser Permanente Northern California system.
“We developed two panels of metabolites using machine learning methods and compared their ability to predict later gestational diabetes with typical risk factors such as obesity, family history of diabetes, history of gestational diabetes, and even fasting glucose levels,” Dr. Zhu says. “Another strength of this study was that we validated it twice, with a sub-group of PETALS participants and a group of participants in the Gestational Weight Gain and Optimal Wellness (GLOW) trial, a previous Division of Research study that provided health coaching to about 200 pregnant individuals on weight gain during pregnancy.”
Evidence from Real-World Pregnant Patients Is Valuable
At gestational weeks 10-13, LASSO regression analysis identified a 17-metabolite panel that showed better predictive performance compared with a model with conventional risk factors (eg, BMI, family history of diabetes, previous gestational diabetes). Similarly, at gestational weeks 16-19, LASSO regression identified a 13- metabolite panel that demonstrated better predictive performance compared with that of a conventional risk factor model (Figure).
Dr. Zhu and colleagues concur that their findings are too preliminary to prompt any change in current medical practice. “However, it is encouraging to know that there is additional evidence from real-world pregnant patients suggesting metabolomics has the potential to make important contributions to preventive care and disease prediction of gestational diabetes in the future. While our study was well-designed and double-validated, the results need verification by researchers with other, larger populations of patients.”
Future research, she adds, is needed to examine the associations of potentially modifiable risk factors, such as diet and physical activity, with these metabolites. “Findings from these studies could further provide important information regarding preventive targets for gestational diabetes,” Dr. Zhu says.
