Bayesian Two-Compartment and Classic Single-Compartment Minimal Models: Comparison on Insulin Modified IVGTT and Effect of Experiment Reduction

Models describing plasma glucose and insulin concentration of an intravenous glucose tolerance test (IVGTT) allow a noninvasive cost-effective approach to estimate important indexes characterizing the efficiency of glucose-insulin control system, i.e., glucose effectiveness (S_G) and insulin sensitivity (S_I). To overcome some limitations of the classic single compartment minimal model (1CMM) of glucose kinetics, a two-compartment Bayesian minimal model (2CBMM) has been recently proposed for the standard IVGTT. This study aims to assess 2CBMM ability to describe the insulin-modified IVGTT (IM-IVGTT) which is the protocol of choice since it allows to study insulinopenic states. Both a full-length IM-IVGTT (240 min) as well as a reduced version (90 min) of it are studied. Results of the maximum a posteriori identification of IM-IVGTT (240 min) in 13 normals agree with those of standard IVGTT, i.e., a 42% decrease (P <0.002) of S_G and a 13% increase (P <0.006) of S_I with respect to 1CMM. When identified from IM-IVGTT (90 min), 2CBMM not only provides S_G and S_I estimates 46% lower (P <0.002) and 41% higher (P <0.002) than 1CMM ones respectively, but also seems to overcome some limitations of the 240 min-based identification that probably arise because the minimal model is unable to properly account for the hyperglycemic hormonal response taking place in the second half of IM-IVGTT.