Fotteler B., Weber C. and Steimer JL.
Pharma Development, F.Hoffmann-La Roche, Basel, Switzerland.
To get an early indication on the efficacy of a new insulin sensitizer, “Minimal Model” analysis was applied to glucose and insulin concentrations obtained after an insulin modified frequently sampled intravenous glucose tolerance test (FSIVGT) using stable labeled glucose, before and after six weeks of treatment with placebo or one of four doses in 40 type 2 diabetes patients (N=8 per treatment group). The “Hot Two-Compartment Minimal Model” was implemented into the software WinNonlinTM to fit the stable labeled glucose profiles. The following features were included:
- Linear interpolation for insulin and total glucose to provide defined (continuous) input to the model.
- Logarithmic transformation of all structural parameters for the fitting.
- Use of predose glucose and insulin concentrations for glucose and insulin steady state-values.
- Weighting of data points before tmax set to 0, i.e ascending phase of tracer glucose neglected.
- Incorporation of assay performance for both, total and tracer glucose in the variance model for the fitting.
- Blinding of individual profiles with respect to dose, patient number and study day to avoid “Modeller Bias” (use of profile numbers).
The stable-labeled FSIVGT and the use of the hot-2 compartment model permitted estimation of in vivo parameters describing glucose disposal in all type 2 diabetes patients on each occasion. The hot minimal model revealed a reliable and accurate “measurement device” for estimation of the insulin sensitivity index (SI) and standard errors of the estimates. This allowed assessment of the treatment effect in an early study exposing a small number of patients only.
Reference: PAGE 10 (2001) Abstr 179 [www.page-meeting.org/?abstract=179]
Poster: poster