Clinical Risk Score is Superior to Fasting C‐peptide–Based β‐Cell Function Markers in Predicting Insulin Requirement in Individuals With Type 2 Diabetes

AIMS: This study aims to evaluate the predictive performance of clinical characteristics and beta-cell function markers for future insulin requirements in Chinese individuals with type 2 diabetes who were not insulin deficient. MATERIALS AND METHODS: A prospective cohort of 2421 individuals with type 2 diabetes and fasting plasma C-peptide ≥ 0.25 nmol/L (1996-2012) was followed until 2019 for insulin requirement, defined as insulin initiation or two consecutive HbA1c ≥ 8.5% while on at least two non-insulin glucose-lowering drugs. Cox regression with stepwise selection was used to identify predictors and derive a clinical risk score. Model calibration was assessed using the Hosmer-Lemeshow test, and discrimination using the Harrell's C-index and area under the receiver operating characteristic curve (AUC) at one, five and 10 years. RESULTS: Over a median follow-up of 10.2 (IQR 4.0-14.1) years, 1143 individuals progressed to insulin requirements. Younger age at diabetes diagnosis, male sex, poorer glycaemic control, kidney complications and greater use of non-insulin glucose-lowering drugs were identified as independent predictors. The clinical risk score demonstrated better discrimination (C-index: 0.729, AUCs: 0.829 at 1 year, 0.792 at 5 years, 0.779 at 10 years) than fasting plasma C-peptide (C-index: 0.532, AUCs: 0.502, 0.553, 0.537), C-peptide-to-glucose ratio (C-index: 0.560, AUCs: 0.660, 0.572, 0.571) and HOMA2-%B (C-index: 0.618, AUCs: 0.734, 0.651, 0.639) (all p < 0.001). Adding fasting plasma C-peptide or HOMA2-%B to the clinical model did not significantly improve its predictive performance. CONCLUSIONS: Clinical characteristics outperformed fasting beta-cell function markers in predicting insulin requirements in non-insulin-deficient type 2 diabetes.