Background and Objectives: Insulin treatment may be initially required to stabilize patients presenting with metabolic crisis at type 1 and 2 diabetes mellitus (DM) onset. Some patients with type 2 DM may need persistent insulin treatment. This study aimed to examine the predictive performance of non-stimulated C-peptide level at the time of diagnosis for future insulin use in pediatric diabetic patients. Materials and Methods: We reviewed the medical charts of diabetic patients aged 18 years or younger in a medical center in southern Taiwan from January 2000 to December 2019. Clinical and individual data were collected at the time of DM diagnosis. Outcomes were persistent insulin use at the time of diagnosis, as well as at one and two years after diagnosis. Results: The final analysis included a total of 250 patients. The best cut-off point of non-stimulated C-peptide was 0.95 ng/mL, and the predictive indices for the insulin use were 0.84 for sensitivity and 0.94 for specificity at two years after DM diagnosis. Incorporating age at onset and presence of GAD antibodies can further increase the predictive power of non-stimulated C-peptide. Conclusions: The value of non-stimulated C-peptide at diabetic onset was feasible and effective for predicting future insulin treatment up to the time point of two years after diagnosis.
Keywords: diabetes mellitus (DM); glucagon test; glutamic acid decarboxylase (GAD) antibodies; non-stimulated C-peptide.