Methods are presented for assessing insulin therapies using a physiologic pharmacokinetic model of glucose homeostasis in man. The model is composed of simultaneous differential equations that represent physiologic compartments and spaces in which glucose and insulin are distributed and undergo metabolic reactions. The model is used to simulate clinical experiments in which blood glucose concentration is controlled by artificial device therapies. Predictions of the theoretical model for responses of normal and diabetic individuals to standard intravenous and oral glucose tolerance tests are compared to clinical data. Reasonable agreement is obtained between predictions of the computer simulations and clinical data for normal individuals. The responses of a diabetic person to oral glucose tolerance tests are simulated by removal of the pancreas from the glucose homeostasis model and introduction of insulin into the model by a prescribed therapy. Model simulations reaffirm expectations concerning the poor blood glucose control attainable by intramuscular insulin injection. Simulations of blood glucose regulation by an artificial pancreas using closed-loop feedback control for controlling insulin delivery rate reveal hyperinsulinemia that results in a net shift in the deposition of a glucose load from liver to peripheral tissues. Simulations of this system in which the time delay for glucose measurement is varied from 1.5 to 30 min show that increases in sensor delay result in progressive loss in glucose regulation, exacerbation of hyperinsulinemia, and increased insulin requirements.

An implantable glucose sensor is being developed that is based on the use of a high-area platinum electrode. The sensor is operated in a controlled potential mode, in which the potential of the platinum working electrode versus an unpolarized reference electrode is periodically varied according to a preselected voltage-time regimen. The change in potential is accompanied by a flow of current between the platinum working electrode and a counter-electrode. This current serves to periodically rejuvenate the electrode surface and to provide a signal that is proportional to the glucose concentration. A method for analyzing this signal has been developed, the compensated net charge (CNC) method, that involves integration of the current over one complete potential cycle. This method significantly improves both the sensitivity and selectivity for determining glucose in the presence of the normal physiologic coreactants. For a solution containing glucose, amino acids, and urea, a change in glucose concentration from 50 to 150 mg/dl gives about a 50% change in the net charge. A change in the urea concentration from 20 to 40 mg/dl has no effect on the net charge. A change in the amino acids concentration from 35 to 65 mg/dl has little effect on the net charge above a glucose concentration of about 100 mg/dl. Negligible effects on the net charge have also been found for creatinine (0-1.5 mg/dl) and uric acid (0-4 mg/dl).

No abstract