Researchers at the University of Toronto have developed a test to determine which patients who experience gestational diabetes are most at risk for post-gestational T2D.
The C.D.C. estimates that 2% to 10% of women will develop gestational diabetes (G.D.M.) during pregnancy. Of the population that develops G.D.M., up to 50% will develop type 2 diabetes (T2D). Current testing includes fasting plasma glucose (F.P.G.) and a burdensome 2-hour oral glucose tolerance test (OGTT). The ability of the 2-hour OGTT to predict future T2D ranges from 65% to 77%. In women who develop G.D.M., the A.D.A. recommends postpartum testing, but research shows that fewer than 30% follow up and complete the testing. As we know, prevention is critical in T2D. Testing is paramount in ensuring clinicians identify patients so that prevention techniques such as weight loss and dietary changes can be employed.
Research by Lai et al. at the University of Toronto has used metabolomics to shed light on the pathophysiology of type 2 diabetes mellitus (T2D) as it pertains to its transition from gestational diabetes mellitus (G.D.M.). The researchers obtained plasma metabolite samples collected during the SWIFT trial, an 8-year prospective observational cohort of 1035 women with G.D.M., to conduct a cross-sectional and longitudinal analysis of individual study subjects’ metabolites as they changed over time. The researchers were able to develop a model using machine learning (random forest classification) that predicted the development of T2D in women with G.D.M. with a higher probability than either F.P.G. or 2-hour OGTT.
The researchers conducted a cross-sectional analysis of the study group at baseline with plasma metabolite samples from the SWIFT trial. These were obtained from G.D.M. patients six to nine weeks postpartum. The cross-sectional analysis showed dysregulation of the metabolites, with particular notice of amino acid dysregulation, suggesting that dysfunction in amino acid metabolism is a preclinical indicator of T2D. Patients who progressed to T2D showed continued dysregulation of metabolites in the longitudinal analysis conducted from baseline to two years. Metabolites with continued dysregulation in the longitudinal analysis were narrowed to ten, showing a marked dysregulation of amino acids. A panel of 20 metabolites (hexose, six amino acids, six glycerophospholipids, two acylcarnitines, two sphingolipids, and three biogenic amines) was selected to create the predictive model for progression from G.D.M. to T2D. The metabolites were selected based on previous studies showing their dysregulation and link to increased risk of T2D. The model predicted progression to T2D with a predictive power area under the curve receiver operating curve (AUC-ROC) of 0.883 (95% CI 0.820-0.945, p <0.001).
A limitation of the study is that the predictive model was created used a single cohort. To best test the model and determine whether or not it is generalizable to a broader population, it would have been ideal to have a separate cohort of patients with G.D.M. to test the model’s T2D predictive capability. To negate overfitting or to create a model that is only capable of predicting an outcome with accuracy within the study cohort, the researchers split the cohort into 70% for training the model and 30% for testing. Despite this limitation, the results of this research present a step forward in the prevention of T2D. Now, instead of an F.P.G. or a 2-hour OGTT, which are burdensome and lack the predictive power of the metabolite panel presented in this research, an additional analysis of labs already taken will provide clinicians with a powerful predictive tool. Hannes Röst, one of the researchers, was quoted as saying, “this is the holy grail of personalized medicine, to find molecular differences in seemingly healthy people and predict which ones will develop a disease.” Indeed, the ability to determine with a substantial degree of certainty which patients will develop T2D perhaps years before their diagnosis would be a landmark shift from the usual reactive approach to diabetes to a proactive one.
- Up to 50% of women with G.D.M. progress to T2D, but only 30% of those will have the appropriate testing done.
- A new predictive model using metabolites from patient plasma samples can predict progression to T2D from G.D.M. with 88% predictive power.
- When implemented, the new screening test would allow providers to closely monitor high-risk patients and make interventions to prevent progression to T2D.
References for “A New Test Capable of Predicting Post-Gestational T2D”:
“Gestational Diabetes.” Centers for Disease Control and Prevention, Centers for Disease Control and Prevention, May 30, 2019, www.cdc.gov/diabetes/basics/gestational.html.
Lai, Mi et al. “Amino acid and lipid metabolism in post-gestational diabetes and progression to type 2 diabetes: A metabolic profiling study.” PLoS medicine, vol. 17,5 e1003112. May 20. 2020, doi:10.1371/journal.pmed.1003112
Gunderson, Erica P, et al. “Study of Women, Infant Feeding, and Type 2 diabetes mellitus after G.D.M. pregnancy (SWIFT), a prospective cohort study: methodology and design.” B.M.C. public health vol. 11 952. 23 Dec. 2011, doi:10.1186/1471-2458-11-952
Henderson, Emily. “Blood Test Could Help Doctors Identify Patients at Risk of Developing Type 2 Diabetes.” News Medical Lifesciences, AZoNetwork Ltd, May 23, 2020, www.news-medical.net/news/20200522/Blood-test-could-help-doctors-identify-patients-at-risk-of-developing-type-2-diabetes.aspx.
David Clarke, PharmD Candidate, University of Colorado, Skaggs School of Pharmacy and Pharmaceutical Sciences