259 IN THE PRECEDING ARTICLE ENTITLED, “Pharmacokinetics, Pharmacodynamics, and DoseDependent Relationship of Repaglinide in Type 2 Diabetes” (page 247), Dr. Strange and associates describe some of the characteristics and advantages of the relatively fast acting, oral beta-cell-stimulant, repaglinide. The study illustrates an important principle of insulin/ glucose kinetics that was well recognized in the early 1970s and is just now becoming clinically applicable. This principle emphasizes that the time of insulin provision can be as important for glucose homeostases as the total insulin provided. Insulin and glucose form a closed loop system, in which glucose stimulates insulin secretion and insulin increases glucose uptake and reduces circulating glucose levels, thereby reducing further secretion. However, the time frames for these two phenomena are significantly different. As shown in early studies using the in vitro perfused pancreas,1 insulin secretion increases within 15 seconds of a rise in glucose levels. In contrast, 8–15 minutes are required for insulin to reach its target tissues and activate sufficient glucose uptake and metabolism in muscle and fat (or decrease hepatic output from the liver) to change the circulating glucose level. Predictably, in such a closed loop system, if insulin secretion is regulated only by the concentration of glucose, insulin release increases with a rising glucose before insulin reaches and becomes effective at its target sites. Even when, in time, tissue insulin is adequate to cause a peaking and partial decline in blood glucose, insulin secretion at these elevated glucose levels continues. When glucose levels reach basal, insulin secretion returns to normal, but tissue insulin levels are excessive. Therefore, if the beta cell only recognizes the concentration of glucose, a resulting delayed hypoglycemia is inevitable. Recognition of the inadequacy of a simple relationship between glucose concentration and insulin secretion was shown clearly 25 years ago with the development of an experimental “Artificial pancreas.”2 (This was not a portable sustainable device but a large bedside apparatus to which the subject was briefly connected by intravenous catheters.) Algorithms containing only a glucose concentration dependency, when providing sufficient insulin to produce a normal rise in glucose, overinsulinized the tissues and could not prevent subsequent hypoglycemia. Only when an algorithm was added with a glucose rate-of-change component could near perfect control be achieved. Most interestingly, when insulin was provided rapidly to meet the rising glucose, only one-tenth the amount of insulin was required to achieve control! Although a reliable closed-loop, portable artificial pancreas is not yet available, it is probable that only minor refinements of the algorithms developed decades ago will be adaptable. To avoid overinsulinization, the normal pancreas also utilizes a form of glucose-rate sensitivity. In 1968,1 it was shown that the pancreas Analysis