Effect of Acute Hyperglycemia on Insulin Secretion in Humans

Thirteen nondiabetic subjects with wide-ranging BMIs were studied (Table 1). None had a family history of diabetes or was taking drugs known to affect glucose metabolism. The potential risks of the study were carefully explained to all subjects, who gave their informed consent before the study. The study protocol was approved by the Institutional Ethics Committee.

Subjects were admitted to the Metabolism Unit between 8:00 and 9:00 a.m. after an overnight (11- to 12-h) fast. The study was performed with the subject resting supine in a comfortable bed in a quiet air-conditioned room. A 20-gauge catheter was inserted into an antecubital vein (for the infusion of test substances); another catheter was threaded retrogradely into a wrist vein of the same arm and used for blood sampling. The hand was kept in a heated box for the entire duration of the study to achieve and maintain arterialization of venous blood. The study protocol consisted of three consecutive 30-min square-wave steps of hyperglycemia (2.8, 2.8, and 5.6 mmol/l above baseline) followed by an intravenous bolus of 5 g arginine. The priming doses of glucose for each step were calculated by multiplying the desired increments in plasma glucose concentration by the body glucose space (estimated to be 150 ml per kilogram of body weight). The 20% dextrose volumes thus calculated were administered over 1 min. After the glucose bolus, plasma glucose concentrations were maintained at the desired plateau by means of a variable 20% glucose infusion according to the glucose clamp technique (11). Blood samples for the determination of plasma glucose, insulin, and C-peptide concentrations were obtained every 2 min for the first 10 min and every 5 min for the other 20 min of each step. After the arginine bolus, blood was sampled every 2 min for 10 min.

Plasma glucose was assayed by the glucose oxidase method (Glucose Analyzer; Beckman Instruments, Fullerton, CA). Specific insulin and C-peptide were assayed in plasma by radioimmunoassay (Human Insulin-specific RIA kit from Linco Research, St. Charles, MI; C-peptide Myria from Techno Genetics, Milan, Italy).

Insulin secretion rate (ISR) was reconstructed from C-peptide levels by deconvolution analysis according to the approach developed by Van Cauter et al. (12) using the sex- and age-specific version. For each glycemic step, first-phase insulin response was calculated as the area under the insulin secretion peak (6 min), whereas second-phase insulin response was defined as the average insulin secretory rate during the final 10 min of each step. The arginine response was the integral of insulin secretion over the 10 min after arginine injection. All insulin secretion values were expressed per square meter of body surface area.

Results are given as means ± SE. Mean values during the various protocol phases were compared by analysis of variance for repeated measures. Linear regression analysis was carried out by standard methods.

The time course of the main variables is shown in Fig. 1. In each of the three hyperglycemic steps, fasting plasma glucose concentrations (which averaged 5.2 ± 0.2 mmol/l) were rapidly raised to the desired plateau and clamped there for the following 30 min. Plasma insulin and C-peptide concentrations followed a parallel pattern during the protocol. In particular, each hyperglycemic step was associated with a distinct peak followed by a gentle rise in hormone levels. Arginine caused a much greater increase in both hormones, also in the form of a peak lasting 10 min. Both the insulin and C-peptide peaks were progressively smaller at each successive glucose plateau.

When compared with plasma insulin or C-peptide concentrations, insulin secretion showed more rapid time dynamics (Fig. 1). On average, peak values were reached within 2–4 min after each glucose bolus and rapidly decreased thereafter to the respective prebolus values; only after 8–10 min of hyperglycemia did insulin secretion start to increase steadily. The acute response to the first hyperglycemic peak was greater than the response to the second identical glucose gradient (Table 2); a further glucose step twofold higher than the preceding ones failed to stimulate a proportionate acute insulin response. In contrast, second-phase insulin secretion increased progressively throughout the protocol. Thus, the dose-response curves for first- and second-phase insulin secretion ran opposite to one another(Fig. 2). With arginine, a large amount of insulin was rapidly released: the peak secretion was fivefold greater than that elicited by the first hyperglycemic step (4.79 ± 2.33 vs. 0.97 ± 0.30 nmol · min⁻¹ · m⁻²). As shown in Table 3, all phases of insulin secretion were interrelated, with the exception of the first-phase response to the second and third hyperglycemic step.

It could be argued that the β-cell is programmed to sense changes in glucose rather than absolute glucose levels by mounting proportionate increments in insulin release. To test this possibility, for each step, we calculated the increments in insulin secretion and glucose concentration above the respective preceding levels, and expressed the former in relation to the latter (Table 2). Whether incremental insulin secretion was divided by the absolute or the percentage plasma glucose gradient, first-phase response still showed a pronounced decline with consecutive hyperglycemic steps (P < 0.0001 for both). In contrast, second-phase insulin secretion, as estimated by either index, did not change significantly across hyperglycemic plateaus.

In this study, insulin secretion was measured by applying a two-compartment model for the deconvolution of peripheral C-peptide concentrations (12). The use of this approach, which has been extensively validated in humans, allows not only quantification of pancreatic insulin secretion, but also discrimination of rapid secretory events (e.g., first vs. second phase) (13). For example, in a study of subjects with impaired glucose tolerance, second-phase insulin secretion was judged to be impaired when using plasma insulin concentrations, whereas the corresponding reconstructed secretion rates were similar to those of control subjects (14). Similarly, in our experiments, plasma insulin and C-peptide concentrations remained above the respective baseline in the transition between the first and second phase, whereas the corresponding secretory rates were clearly separated by a return to baseline (Fig. 1).

The main finding of the present study is that short-term moderate hyperglycemia exerts an inhibitory influence on the first-phase insulin secretory response to a subsequent glucose stimulus. This refractoriness is not overcome when the glucose stimulus is further enhanced and does not involve the response to arginine. This inhibitory effect of hyperglycemia on first-phase insulin secretion is evident both when considering absolute secretion rates and when relating changes in secretion to the corresponding changes in glucose levels. Thus, whether the β-cell is sensing the full glucose excursion or only its increment above the preexisting glucose level, the secretory response is depressed when preceded by another glucose stimulus. In contrast, second-phase insulin secretion responds to steady-state glucose concentrations in a quasilinear manner.

The effects of a prestimulus glucose concentration on first- and second-phase insulin secretion in response to intravenous glucose in humans have been controversial. In vitro studies have demonstrated that the increased insulin release that occurs with continued presence of a high glucose concentration tends to wane over a few hours (15). A similar phenomenon was demonstrated in vivo in the rat, in which 48 h of hyperglycemia (maintained by continuous glucose infusions) led to a selective loss of glucose-stimulated insulin secretion, with the response to arginine being preserved (16). In nondiabetic humans, while some studies have demonstrated an apparent increase in insulin secretion after short-term hyperglycemia (7), other studies have shown decreased insulin concentrations after short-term hyperglycemia (10) and increased insulin secretion after prolonged hyperglycemia (17). In light of the current findings, previous conflicting results may be reconciled. In the early studies by Cerasi et al. (9), periods of euglycemia of variable duration were interposed between the hyperglycemic challenges. With this format of stimulation, the emergence of inhibitory or potentiating influences on insulin secretion depended on the duration of the intermittent euglycemia. Thus, when sufficient time was allowed for the β-cell to recover from a first glucose challenge, the response to a subsequent glucose stimulus was enhanced (potentiation). If, on the other hand, the glucose infusions were in quick sequence, inhibition of first-phase insulin release was evident. In the study by Ferner et al. (10), which used square waves of hyperglycemia similar to those applied in our studies, the analysis relied on plasma insulin concentrations rather than secretion rates, and the interpretation depended on whether one assumed that islets respond to absolute or incremental glucose levels. In our experiments, inhibition of first-phase insulin secretion was obvious regardless of the way in which data were calculated.

The physiological relevance of this finding needs to be appraised. From a wealth of in vitro and in vivo evidence, it is clear that β-cells respond phasically to constant stimulation and that both inhibitory and enhancing influences are necessary to explain this behavior adequately (18). The cellular basis for these dynamic adaptations to glucose stimulation are incompletely described. A two-pool model, in which mature secretory granules rapidly dock on the plasma membrane to deliver their insulin content to the extracellular space at the same time as the processing machinery is set into motion, still offers the most plausible explanation for the biphasic insulin response to acute hyperglycemia. Our results indicate that moderate preexisting hyperglycemia limits the ability of only the “fast pool” to respond to an incoming challenge; this limitation is not due to exhaustion because an arginine stimulus still elicits a quick release of large amounts of insulin. Such selective functional refractoriness is closely reminiscent of the effects of chronic hyperglycemia, i.e., glucose toxicity (16). On the other hand, the fact that second-phase insulin secretion was closely related to the actual glucose concentration at each hyperglycemic plateau suggests that any toxic effect of glucose exposure on second-phase insulin release must have a longer time constant than the 2 h of our experiment.

To what extent the secretory dynamics evoked by acute intravenous glucose administration relate to early- and late-phase insulin secretion during oral glucose or mixed meal administration remains, in our view, uncertain. It is possible that the slow rise in glucose levels that follows glucose ingestion triggers inhibition of “fast pool” insulin delivery and potentiation of “slow pool” insulin availability simultaneously. This, however, remains a purely descriptive conjecture. More plausible is that the chronic fasting hyperglycemia of diabetes may institute refractoriness of “fast pool” insulin release, which both nonglucose secretagogues and lowering of fasting hyperglycemia (by antidiabetic treatment) can circumvent only in part.