A minimal-model-based glucose clamp yielding insulin sensitivity independent of glycemia

A new technique is introduced for automatic control of the blood sugar ("glucose clamp") at basal (euglycemic) or elevated (hyperglycemic) levels during variable insulin infusion. A minimal mathematical model of glucose kinetics is implemented on a laboratory minicomputer during clamp experiments. From the measured time course of plasma glucose concentration, the computer estimates the fractional disappearance rate of glucose (X) and calculates the rate of exogenous infusion required to match the desired concentration (M). Eight clamp experiments were performed on three conscious dogs. Euglycemic (N = 4): insulin was infused at 8, 40, and 150 microU/min for sequential 2.5-h periods )I, II, and III); hyperglycemic (N = 4): following the establishment of 140 mg/dl hyperglycemia (period I), insulin was infused at 30 and 120 mU/min for 2.5 h each (periods II and III). Plasma insulin levels [range: 21 +/- 9 (basal) to 1729 +/- 209 microunits/ml] were matched in comparable periods in the two types of experiments [P greater than 0.8 (I); P greater than 0.5 (II), P greater than 0.2 (III)]. Glucose was successfully clamped for all periods at euglycemia (99.6 +/- 0.7% of desired value) and hyperglycemia (100.0 +/- 1.1% desired 140-mg/dl value). Glucose infusion rates necessary to achieve stable glycemia were greater at elevated than basal glucose (P less than 0.001) despite comparable insulinemia during the three insulin infusion periods. Thus, the glucose infusion rate versus insulin level (M/I) was glycemia-dependent, and therefore not a direct measure of insulin sensitivity. In contrast, the insulin-induced increases in fractional glucose disappearance (delta X), also provided by our clamp technique during experiments, were highly correlated within insulin (r = 0.08 to r = 0.99, P less than 0.01) and independent of glycemia (P greater than 0.7). The ratio delta X/delta I (increase in glucose fractional disappearance rate/incremental insulin level) was a measure of insulin sensitivity independent of glycemia, and could successfully differentiate insulin-resistant from normal animals (P less than 0.001; delta X/delta I in three animals: 6.9 X 10(-4), 5.5 X 10(-4), and 1.4 X 10(-4) min-1 microunits/ml). Thus, this model-dependent glucose clamp technique provides a direct measurement of insulin sensitivity independent of glycemia.