Calculation of the systemic delivery rate of insulin in normal man.

The accuracy with which the systemic delivery rate of insulin can be estimated from peripheral insulin concentrations was investigated in eight normal men. Endogenous insulin secretion was suppressed by somatostatin, and insulin was infused exogenously via a peripheral vein. The infusion rate was progressively increased and then decreased to stimulate the changes in insulin secretion that occur after a secretory stimulus. The infusion rate of insulin was then estimated by analyzing peripheral insulin concentrations using both single and two-compartment mathematical models and was compared with the actual infusion rates. Model parameters were measured in each subject by analysis of the insulin decay curve after a bolus injection. Estimated infusion rates based on the single compartment model were 84.3 +/- 3.0% (mean +/- SE) of the actual infusion rate as it was increasing and 110.3 +/- 3.7% of the infusion rate as it was decreasing. The mean overall absolute percent error associated with this approach was 14.1 +/- 1.0% (range, 10.2-18.8%), and model estimates of the total amount of insulin infused over the duration of the experiment were 97.1 +/- 3.2% of the quantity actually infused (range, 85.4-110.7%). Estimates of the infusion rate based on a two-compartment model represented 90.3 +/- 3.1% of the actual infusion rate as it was increasing and 98.2 +/- 3.2% of the actual rate as it was falling. The area under the derived infusion rate curve was 94.8 +/- 2.8% of the area under the actual curve. The overall error associated with the use of the two-compartment model was 12.9 +/- 1.1% (range, 8.8-18.6%). Differences between the single and two-compartment models were minor and not statistically significant. These data demonstrate that under euglycemic conditions, changes in the plasma insulin concentration within the physiological range are not associated with changes in the clearance kinetic rate constants of the hormone. Furthermore, the systemic delivery rate of insulin can be accurately derived from peripheral insulin concentrations using either a single or two-compartmental model if model parameters are accurately measured in individual subjects by analysis of insulin decay curves.