MEASUREMENT OF INSULIN SENSITIVITY BY THE MINIMAL MODEL METHOD USING A SIMPLIFIED INTRAVENOUS GLUCOSE-TOLERANCE TEST - VALIDITY AND REPRODUCIBILITY

This study aimed at testing whether 12 rather than 26 plasma glucose and insulin determinations can be used to calculate the indices of insulin sensitivity and of glucose effectiveness using Bergman's minimal model during a simple intravenous glucose tolerance test performed without tolbutamide injection. Two intravenous glucose tolerance tests (separated by 1 week) were performed in 7 lean normal subjects and a single test was performed in 9 severely obese nondiabetic subjects. Intra-subject reproducibility of insulin sensitivity was not significantly different when 26 or 12 time-points were analyzed (CV = 16.8 +/- 3.4 versus 18.9 +/- 3.8 % respectively). Compared with the insulin sensitivity of the lean subjects, that of obese subjects was significantly (P < 0.001) and similarly reduced when using 12 (2.14 +/- 0.34 versus 7.97 +/- 1.29 10(-4) min(-1)/mU.1(-)1) rather than 26 determinations (2.13 +/- 0.42 versus 6.95 +/- 1.12 10(4) min(-1)/mU.1(-1)) respectively. Glucose effectiveness was less reproducible than insulin sensitivity and was slightly diminished by the reduction of blood samples (relative error : -9.7 +/- 4.4 %; P < 0.05). Finally, glucose effectiveness tended to be slightly lower in the morbidly obese subjects than in the lean controls with both modes of calculation. In conclusion, in non-diabetic subjects, the insulin sensitivity index can be accurately measured during a simple intravenous glucose tolerance test, without tolbutamide injection and with only 12 blood samples. The possibility of performing a simplified test should contribute to increase the use of the minimal model method for estimating insulin sensitivity in clinical practice.