Development of a ketone-aware automated insulin delivery system to enable safe use of sodium-glucose cotransporter inhibitors in people with type 1 diabetes - Project Summary Despite recent advances in diabetes treatment and technology (e.g., automated insulin delivery [AID] systems), achieving recommended glycemic targets is still difficult for people with type 1 diabetes (T1D) and, worrisomely, T1D confers substantial cardiorenal risk even when conventional glycemic targets are achieved. The sodium- glucose cotransporter inhibitors (SGLTi) are antihyperglycemic medications that have demonstrated significant cardiorenal benefits in people with and without type 2 diabetes, including reductions in kidney disease progression and hospitalization for heart failure. SGLTi medications have been tested in people with T1D and shown beneficial effects on glycemic control; however, they also contribute to an increased risk of ketoacidosis that has drastically limited their clinical utility in this population. AID systems pair continuous glucose monitors (CGM) and insulin pumps with control algorithms that automatically modulate insulin delivery in real-time. Continuous ketone monitoring (CKM) is a rapidly evolving technology with the potential to prevent ketoacidosis by tracking ketone levels, transmitting results to a dedicated smartphone or device, and providing early warnings of ketone elevations. A combined CKM/CGM biowearable sensor (Abbott Diabetes Care; Alameda, CA) is currently in clinical testing for commercial use and has the potential to be integrated into closed-loop AID systems. The current proposal will specifically address this by developing a “ketone-aware” AID system that receives dual input from the CKM/CGM sensor and subsequently optimizes glycemic control while minimizing ketosis. Our study aims are as follows: 1. We will refine and implement the Ketone-Aware Predictive Algorithm (kAPA), a ketone-aware algorithm with its roots in our recent SGLTi + AID crossover trial. Using archival data and CKM characteristics, we will model ketone elevation and measurement in our well-known simulation environment that is FDA-accepted for preclinical testing of AID systems. In parallel, we will create a kAPA model predictive AID controller and refine it in silico before implementing kAPA into our established DiAS prototyping platform. 2. We will demonstrate the feasibility and safety of kAPA in people with T1D by conducting a 36-hour supervised pilot study to demonstrate acceptable real-time ketosis control while taking sotagliflozin 200 mg; 3. We will demonstrate that kAPA improves glycemic control and minimizes ketosis by conducting a 12- week randomized, single-blind, placebo-controlled crossover clinical trial comparing kAPA + sotagliflozin to AID + placebo in people with T1D. Successful completion of this project will develop a proactive, ketone-aware AID system that can optimize glycemic control, minimize ketosis, and allow for safe administration of SGLTi medications in people with T1D.
