CGM has Similar Predictive Performance to OGTT for Monitoring Early-Stage Type 1 Diabetes

Abstract Objective Compare continuous glucose monitoring (CGM) and other metabolic markers in progression to stage 3 type 1 diabetes (T1D) in early-stage T1D and evaluate CGM time above 140 mg/dl (TA140) across different stages. Methods In a combined dataset from three prospective studies, 152 early-stage T1D individuals had baseline CGM, oral glucose tolerance test (OGTT) and HbA1c with 54 (36%) progressing to stage 3. ROC curves were generated to compare the areas under the curve (AUC) for stage 3 prediction. Risk of progression was estimated by Accelerated Failure Time (AFT) and Kaplan–Meier analyses. Results TA140 > 10% and 2 hr glucose>140 mg/dl had the exact same specificity (83%), negative predictive value (88%) and sensitivity (55%) for stage 3 prediction at 2 years. In univariable AFT, the HRs for stage 3 at 5 years were significant for all CGM metrics tested (all p<=0.008), while significant OGTT metrics included 2hr-glucose, AUC_glucose, AUC_C-peptide and peak_C-peptide (all p<=0.021). In multivariable AFT, younger age, male sex, HbA1c>=5.7%, TA140 > 15% and 2-hr glucose>140 mg/dl were strong predictors of accelerated progression to stage 3. Cumulative incidence curves for stage 3 were similar for TA140 > 15%, HbA1c>=5.7% and 2hr-glucose>140 mg/dl. Among 406 longitudinal CGMs, participants at stage 3 had higher median TA140 (17.2%, IQR 7.9-33.2) compared to stage 2 (8.6%, IQR 3.8-15.2) and stage 1 (3.2%, IQR 1.3-7.7) (all pairwise p<=0.024). Conclusions In this largest dataset of individuals at early-stage T1D, we define CGM-derived TA140 criteria for the various stages. Furthermore, we show that TA140 has similar performance to a 2-hour OGTT value for predicting stage 3.