Insulin Dose-Adjusted Hemoglobin A1c has Suboptimal Performance for Identifying C-peptide-defined Partial Remission: The SEARCH for Diabetes in Youth Study

We analyzed data from 422 SEARCH for Diabetes in Youth study participants with mixed-meal tolerance test (MMTT)-stimulated C-peptide within 12 months of diagnosis. Partial remission was defined using MMTT-stimulated area under the curve (AUC) mean C-peptide. Participant characteristics were compared by BMI category (<85th percentile [normal/underweight] and ≥85th percentile [overweight/obese]) using Wilcoxon rank-sum and Fisher’s exact tests.

Sensitivity, specificity and receiver operating characteristic (ROC) AUC of IDAA1c ≤9 for identifying partial remission were evaluated. Results: Normal/underweight participants, versus overweight/obese participants, had lower IDAA1c (8.95 vs 9.28, p=0.043) despite lower stimulated C-peptide (1.25 vs 1.88 ng/mL, p<0.001). Overall, IDAA1c ≤9 had 61% sensitivity and 79% specificity for predicting C-peptide ≥0.91 ng/mL with ROC AUC=0.77.

Among overweight/obese participants, sensitivity was 45% and specificity 81%. A Youden index-derived cutoff of IDAA1c ≤11.6 improved sensitivity to 93%, with specificity of 54% in overweight/obese participants. Conclusions: IDAA1c ≤9 demonstrated suboptimal accuracy as a screening test for partial remission, especially underestimating β-cell function in youth with overweight/obesity.

BMI-specific cutoffs improved sensitivity but reduced specificity.</p>