1828-P: Development of Preliminary Predictive Models for Dorzagliatin Treatment Response and Diabetes Remission in Type 2 Diabetes: A Focus on Personalized Medicine

Introduction and Objective: Dorzagliatin is the first-in-class glucokinase activator (GKA) launched in 2022 for treating T2D. Two Phase III randomized controlled trials (SEED and DAWN) established the efficacy and safety of dorzagliatin, and an observation study (DREAM) suggested diabetes remission for 52 weeks without medication can be achieved after receiving treatment of dorzagliatin. In order to identify those patients who could be benefited from dorzagliatin, predictive models were developed to assist clinicians in quantifying patients' anticipated glycemic responses and diabetes remission probabilities, thereby achieving personalized medicine. Methods: Using data from SEED, DAWN and DREAM, fit a logistic regression model to predict HbA1c response probabilities, and apply a Cox proportional hazards regression model for diabetes remission probabilities. Subsequently, develop visual prediction applications based on these models. Results: HbA1c-response-probability prediction model included the following factors: age, gender, BMI, diabetes duration, metformin use, baseline HbA1c, HOMA- β, and HOMA-IR. Diabetes-remission-probability model included predictors of age at diabetes diagnosis, BMI, diabetes duration, HbA1c before treatment, and HOMA 2-β change value after treatment. Both models incorporated statistically significant predictors of beta cell function, which can be improved after receiving treatment of dorzagliatin demonstrated in SEED and DAWN. The areas under ROC curve were more than 0.7 for both models. Based on the predictive models, web-based visual prediction applications were successfully developed. Conclusion: This study provided a practical tool for personalized medicine with dorzagliatin treatment in T2D. Predictive models and web-based visual prediction applications were developed to assist clinicians in identifying potential patients who can achieve glycemic control and diabetes remission. Disclosure X. Liu: None. X. Gao: None. Y. Tang: None. B. Zhao: None. Y. Zhao: None. Y. Zhang: None.