Simultaneous fitting of the Minimal Model to IVGTT and Glucose Clamp Data
Mills, Richard (1,2), Kayode Ogungbenro (1) & Leon Aarons (1)
(1) School of Pharmacy and Pharmaceutical Sciences, University of Manchester, Manchester, UK. (2) Medeval Ltd. Manchester, UK.
The minimal model of glucose kinetics has been extensively used to provide quantitative estimates of insulin sensitivity (SI) and glucose effectiveness (SG) in both healthy and diabetic subjects. Using NONMEM the minimal model was fitted to glucose and insulin concentration data from an Intravenous Glucose Tolerance Test (IVGTT) where 0.3 g/kg glucose was administered to 28 healthy volunteers. Blood samples were taken pre- and post-administration of glucose to measure the insulin and glucose concentrations. The IVGTT was performed prior to a two-period glucose clamp experiment. During each clamp period 0.2 IU/kg of Actrapid‚ was administered and blood glucose concentrations were monitoring using HemoCue‚ measurements every 5 minutes. These were used to determine the glucose infusion rate (GIR) which maintained the blood glucose concentration at the clamp level. Blood samples were taken at numerous intervals of the 8-hour clamp for determination of insulin concentrations.
It has previously been shown that the minimal model parameter SI and the insulin action parameter (p2) show high correlation and in the fitting of the minimal model to the IVGTT data both parameters were found to be unidentifiable.
Consequently the minimal model was simultaneously fitted to glucose and insulin concentration data from an IVGTT and a glucose clamp. The addition of the clamp data in the simultaneous fit stabilised the fitting enabling reliable and reproducible estimates for all minimal model parameters to be obtained. Population estimates for the parameter values were SI = 0.00134 min-1 per mU.L-1, SG = 0.0183 min-1, VG = 0.164 L/kg and p2 = 0.0138 min-1.
The Fisher Information Matrix (FIM) showed that the associated CV% for the SI and p2 parameters was greatly reduced for the simultaneous fit data when compared to IVGTT data alone. For the IVGTT data alone the SI and p2 parameters had CV% values of 43.0% and 46.5% respectively, which were reduced to 1.7% and 1.0 % respectively, for the simultaneous fit.