1232-OR: Dynamic CGM Models to Predict GDM: A GO MOMs Analysis

Introduction and Objective: In GO MOMS, a multicenter observational study, % time above range (TAR) 134mg/dL, as determined from early pregnancy CGM, added statistically significant but not clinically meaningful predictive value for GDM compared to clinical factors alone. We sought to assess if prediction models incorporating CGM metrics derived from multiple time points improved GDM prediction. Methods: Gravidas with no pre-existing diabetes and a singleton gestation enrolled at 10-14 wks gestation. Blinded CGM data were collected at 10-14 (V1), 16-20 (V2), 24-28 (V3) wks gestation, with a minimum of 5 days of data used to calculate CGM metrics. GDM was diagnosed using Carpenter-Coustan criteria at V3. Using lasso logistic regression, models were developed separately for each visit along with dynamic models combining data from multiple visits. Candidate clinical predictors included maternal age, BMI, and past GDM. Results: Of 2178 participants, 1539 had complete GDM outcomes (14.3% with GDM) and CGM data at all 3 visits. Models including % TAR 150mg/dL, coefficient of variation (CV, a measure of glycemic variability), BMI, and past GDM predicted GDM (AUCs: V1=0.81, V2=0.83, V1+V2=0.84). Inclusion of V3 CGM did not improve performance Conclusion: Inclusion of multiple CGM metrics, including CV and TAR, improved model performance compared to clinical data alone when predicting GDM. Future studies should assess the clinical relevance of these findings for GDM prediction in early pregnancy. Disclosure J. Siddique: None. J. Sherr: Other - research support, consultant, advisory board member; Current; Abbott Diabetes. Other - advisory board member, consultant; Current; Vertex Pharmaceuticals Incorporated. Consultant; Current; Ypsomed AG. Research Support; Current; Dexcom, Inc., JDRF, Provention Bio, Inc., National Institutes of Health. Other - research support, consultant, advisory board member; Current; Insulet Corporation, Medtronic. Research Support; Current; Sanofi. Advisory Panel; Current; sequel med tech. P. Catalano: None. F.L. Facco: None. M. Feghali: None. W. Grobman: None. E. LeBlanc: None. J. Lonier: Advisory Panel; Ended; Vertex Pharmaceuticals Incorporated. Research Support; Current; Sanofi. W. Lowe: None. A. Merriam: None. M. Mourad: None. C. Oshiro: None. C. Powe: Research Support; Current; Dexcom, Inc. Other - Associate Editor of Diabetes Care, Honoraria for Educational Materials; Current; American Diabetes Association. Other - Royalties for Up To Date chapters; Current; Wolters Kluwer (Up To Date). Other - Speaker; Ended; Medscape. U. Reddy: None. D. Rouse: None. D. Scholtens: None. A.C. Spadola: None. K. Vesco: None. E. Werner: None. L.M. Yee: None. N. Zork: None. N. Lancki: None. E. Szmuilowicz: None. Funding The GO MOMs study is supported by grants from the National Institute of Diabetes and Digestive and Kidney Diseases. U01DK123795 to Massachusetts General Hospital; U01DK123791 to Kaiser Permanente; U01DK123759 and U01DK123745 to Northwestern University; U01DK123799 to Yale University; U01DK123783 to Women & Infants Hospital of Rhode Island. Dexcom provided the CGM systems used in the study free of charge.