Addition of New Therapeutic Agents to an Established Type 2 Diabetes Simulation Platform for Therapy Optimization: A Bayesian Model-Based Approach

Abstract Patients with type 2 diabetes mellitus (T2DM) typically take blood glucose level lowering oral or injectable therapeutic agents to treat their condition. Titration and timing of administration of these agents can be difficult under optimal conditions. Largely because of these challenging tasks, less than half of patients with T2DM under therapy are reaching desired glycemic targets. Computer simulations have been shown in both types of diabetes to be powerful tools to design and test optimal therapies. However, the diversity of available therapeutic agents makes the construction of such a platform challenging. In this manuscript, we present a methodology to integrate pharmacokinetics (PK) and pharmacodynamics (PD) of anti-diabetic drugs into an existing T2DM population simulation platform to optimize therapy dosage and timing, and inform clinical trial designs; the mixture of insulin glargine and a glucagon-like peptide 1 receptor agonist (GLP1-RA) was used as an example. The platform was augmented with several drug-specific new/modified sub-models and the associated parameter distributions were derived from various blood measurements collected during clinical studies. The joint model parameter distribution of the augmented platform was obtained by fitting simulated glucose profiles on 2000 days of glucose sensor data in a novel Bayesian framework. The resulting platform was then validated by reproducing glucose distributions from a large clinical study, originally excluded from the training data. Finally, simulation experiments of optimal administration timing of the studied mixture were run.