Mechanisms of Impaired Glucose Tolerance and Insulin Secretion during Isoflurane Anesthesia

This study was approved by the Animal Investigation Committee of Tokushima University, Tokushima, Japan, and was conducted according to the animal use guidelines of the American Physiologic Society, Bethesda, Maryland.

After fasting for approximately 15 h, male Japanese white rabbits (13 weeks old, 2.5–3.0 kg) were anesthetized with intravenous sodium pentobarbital (30 mg/kg). A surgical plane of anesthesia was maintained by continuous infusion of sodium pentobarbital (17 mg · kg⁻¹· h⁻¹) via  an ear vein catheter.21A tracheostomy was performed through a ventral midline incision, and the trachea was cannulated. Rabbits were ventilated (model 122; New England Medical Instruments, Medway, MA) with positive pressure using 100% oxygen. The ventilator was adjusted as needed to maintain blood pH and arterial carbon dioxide tension within normal physiologic ranges. A heating blanket was used to maintain body temperature at 38.5°C. Heparin-filled catheters were inserted into the right carotid artery and the right jugular vein for the measurement of arterial blood pressure and the administration of drug and maintenance fluid, respectively. Maintenance fluid consisted of lactated Ringer’s solution (15 ml · kg⁻¹· h⁻¹), which was continuously infused for the duration of the experiment.

Thirty minutes after surgery, rabbits were randomized to receive an intravenous bolus followed by a 195-min intravenous infusion of either glibenclamide (50 μg/kg + 33.5 μg · kg⁻¹· h⁻¹) or an equivalent volume of vehicle solution (20% hydroxypropyl β cyclodextrin).22Forty-five minutes after the infusion was started, 0 or 1 minimum alveolar concentration (MAC) isoflurane was reached and was maintained throughout the duration of the study (fig. 1). Thirty minutes after 0 or 1 MAC isoflurane was reached, a solution of 0.6 g glucose/kg body weight was administered into the right jugular vein. One milliliter of blood was collected to measure levels of blood glucose and plasma insulin at baseline, and again immediately before administration of isoflurane, immediately before administration of glucose, and at 1, 3, 5, 10, 15, 30, 60, and 120 min after administration of glucose (fig. 1). Blood glucose concentrations were measured using a small electrode-type blood glucose meter (ANTSENSE III; HORIBA Ltd., Kyoto, Japan). Plasma was stored at −80°C until immunoreactive insulin (IRI) levels were measured. Plasma IRI levels were determined using a sandwich enzyme-linked immunosorbent assay (Insulin kit; Morinaga Institute of Biologic Science, Yokohama, Japan). The insulinogenic index was calculated by dividing the area under the insulin–time curve by the corresponding area under the glucose–time curve from 0 to 5 min after administration of glucose.

Pancreata from 25–30 g male Swiss-Webster mice were used for patch clamp experiments. The bile duct was cannulated, and 2 mg/ml collagenase was injected into the pancreas, which was then removed and incubated for 5–10 min at 37°C to isolate individual islets. Single-cell suspensions were obtained by gently shaking islets in a low-calcium medium. Cell suspensions were plated onto glass coverslips in 35-mm Petri dishes containing RPMI-1640 medium supplemented with fetal bovine serum, l-glutamate, and penicillin–streptomycin. Petri dishes were kept in a 95% air–5% CO²incubator at 37°C. Single cells were fed every other day and incubated no more than 1 week after isolation. For patch clamp experiments, individual coverslips were transferred to the test chamber and placed on an inverted microscope.

Cell-attached configurations were used to record currents through single channels using a patch clamp amplifier as previously described.21In the cell-attached configuration, the bath solution was composed of 140 mm KCl, 10 mm HEPES, and 1 mm EGTA. The pipette solution contained 140 mm KCl, 1 mm CaCl², 1 mm MgCl², and 10 mm HEPES. The pH of all solutions was adjusted to 7.3–7.4 with KOH. Recordings were made at 36°± 0.5°C. Patch pipettes were pulled with an electrode puller (PP-830; Narishige, Tokyo, Japan). The resistance of the pipettes filled with internal solution and immersed in Tyrode solution was 5–7 MΩ. Single-channel data were sampled at 5 KHz using a low-pass filter (1 KHz).

Channel currents were recorded with a patch clamp amplifier (CEZ 2200; Nihon Kohden, Tokyo, Japan) and digitized using an analog-to-digital converter (DigiData 1200; Axon Instruments, Foster, CA) and a personal computer (Aptiva; International Business Machines Corporation, Armonk, NY). pClamp version 7 software (Axon Instruments) was used for data acquisition and analysis. The open probability (P^o) was determined from current amplitude histograms and was calculated using the following equation:

where t^jis the time spent at current levels corresponding to j = 0, 1, 2, N channels in the open state; T^dis the duration of the recording; and N is the number of active channels in the patch. Two- to 3-min recordings were analyzed to determine P^o. Channel activity is expressed as NP^o(number of active channels in a patch, N, times open probability, P^o).

Isoflurane was delivered as previously described.21Briefly, a mixture of concentrated isoflurane and bath solution was prepared in graduated syringes. Isoflurane superfusion was achieved using a syringe pump with a constant flow rate of 1.0 ml/min. A clinically relevant concentration of 0.5 mm isoflurane was used, which is equivalent to 2.4 vol% or 1.8 MAC for a mouse. To verify the concentration of isoflurane in solution, 1.5 ml of the superfusate was collected in a metal-capped 3-ml glass vial at the end of each experiment. Gas chromatography (G-3500; Hitachi, Tokyo, Japan) was used to quantify the concentration of isoflurane in the superfusate, which was 0.50 ± 0.05 mm.

Power analysis revealed a group size of n = 6, providing a power of 0.8 with a minimum detectable difference of means of the insulinogenic index of 0.05, an expected SD of 0.03, and an α value of 0.05. Data are expressed as mean ± SD. Repeated-measures analysis of variance was used for within- and between-group comparisons, followed by Student–Newman–Keuls tests. Changes within and between groups were considered statistically significant when the P  value was less than 0.05. Analysis was performed on a Macintosh computer with Statview 5.0 statistical analysis software (SAS Institute Inc., Cary, NC).

There were no statistically significant differences in hemodynamics between groups under baseline conditions or before administration of isoflurane (table 1). Administration of isoflurane significantly decreased heart rate and mean arterial pressure compared with vehicle control. Administration of isoflurane and glibenclamide together resulted in heart rates that were significantly lower than those observed in the vehicle, isoflurane, and glibenclamide groups at 30 min, 1 h, and 2 h after glucose administration.

Mean plasma glucose and IRI concentrations are shown in figures 2 and 3, respectively. Baseline concentrations of glucose and IRI were comparable across groups. In the vehicle and isoflurane groups, concentrations of glucose and IRI measured before administration of isoflurane and before administration of glucose were not significantly different from baseline (figs. 2 and 3). Before glucose administration, glibenclamide decreased blood glucose concentrations and increased IRI levels, whereas isoflurane slightly decreased IRI levels. As a result, blood glucose concentrations measured immediately before glucose administration in the glibenclamide and isoflurane plus glibenclamide groups were significantly lower than those in the vehicle and isoflurane groups. IRI levels measured immediately before glucose administration in the isoflurane group were significantly lower than those in the glibenclamide group.

After intravenous administration of 0.6 g/kg glucose, blood glucose concentrations rapidly increased and then returned to baseline levels within 2 h in all experimental groups (fig. 2). Blood glucose concentrations were highest in the isoflurane group at all time points studied and were significantly higher than those observed in the other groups 5 min after glucose administration. IRI levels rapidly increased and then returned to baseline levels in the vehicle group after administration of glucose (fig. 3). In the isoflurane group, IRI levels were significantly increased after glucose administration, but were lower than those in the vehicle or glibenclamide groups at all time points. One minute after glucose administration, IRI levels in the groups that received 1.0 MAC isoflurane (the isoflurane and isoflurane plus glibenclamide groups) were significantly lower than those in the experimental groups that did not receive isoflurane (the vehicle and glibenclamide groups). In the glibenclamide group, IRI levels increased rapidly and remained elevated for 2 h. In the isoflurane plus glibenclamide group, IRI levels were significantly lower than those in the glibenclamide group at 1, 10, 30, 60, and 120 min after glucose administration. The insulinogenic index in the two groups that received isoflurane was significantly lower than in the vehicle control group (fig. 4).

To investigate whether isoflurane affects K^ATPchannel activity in murine pancreatic β cells, we measured single K^ATPchannel currents using a cell-attached patch clamp technique. Spontaneous single channel activity was observed in the absence of glucose (fig. 5A). Application of 300 μm diazoxide, a selective K^ATPchannel opener, to the bath solution significantly activated K⁺-selective channels (NP^o0.493 ± 0.157; P < 0.05 vs.  baseline; n = 12). The effects of diazoxide were completely blocked by 1 μm glibenclamide, a specific K^ATPchannel blocker (fig. 5A). A single-channel conductance of 58.6 ± 4.8 pS (n = 12) was calculated from the current–voltage relation observed between membrane potentials of −80 and +60 mV, which is consistent with previous studies.23 

We also examined the effects of diazoxide on glucose-induced inhibition of pancreatic K^ATPchannels in a cell-attached configuration. As shown in figure 5B, spontaneous K^ATPchannel activity was significantly decreased when the glucose concentration in the bath (extracellular) solution was increased from 0 to 10 mm. The subsequent application of 300 μm diazoxide to the bath reactivated this channel, resulting in an increase in mean K^ATPchannel activity. Diazoxide-induced K^ATPchannel activity was inhibited by 1 μm glibenclamide. The changes in NP^oare summarized in figure 5C.

Representative traces illustrating the effects of 0.5 mm isoflurane on glucose-induced inhibition of pancreatic K^ATPchannel activity in a cell-attached configuration are shown in figure 5D. Application of isoflurane to the bath reversed the inhibitory effect of glucose, whereas this channel activity was completely inhibited by subsequent application of 1 μm glibenclamide to the bath. The changes in NP^oare summarized in figure 5E. These results demonstrate that isoflurane can mimic the effects of diazoxide by activating pancreatic K^ATPchannels.

Although 1.0 MAC isoflurane had no effect on blood glucose concentration or IRI values when animals were fasted for 15 h, isoflurane significantly decreased insulin secretion during the early phase of the IVGTT and significantly increased blood glucose concentrations 5 min after glucose administration. These results are consistent with previous studies10–15showing that volatile anesthetics impair glucose tolerance and insulin secretion. Tanaka et al.  15reported that isoflurane and sevoflurane anesthesia impair glucose tolerance and insulin secretion to the same extent in humans, an effect that was independent of the dose used. In that clinical study, the authors also showed that the insulinogenic index of individuals in the isoflurane and sevoflurane groups was significantly lower than that observed in a control group.15In the current study, the insulinogenic index of the isoflurane groups was significantly lower than in the vehicle control group. These results suggest that isoflurane inhibits insulin secretion during the early phase of an IVGTT.

The normal response of β cells to a rapid and sustained increase in glucose concentration has been reported to involve a biphasic secretion of insulin.18,24–26The first phase consists of a rapid but transient (4–8 min) increase in the secretory rate.26The second phase consists of an initial decrease followed by sustained or gradually increasing secretion that lasts as long as glucose is present. The first phase involves triggering pathways that initiate the following sequence of events: entry of glucose by facilitated diffusion, metabolism of glucose by oxidative glycolysis, an increase in the ATP–to– adenosine diphosphate ratio, closure of K^ATPchannels, membrane depolarization, opening of voltage-dependent Ca²⁺channels, Ca²⁺influx, an increase in cytoplasmic free Ca²⁺concentration, and activation of the exocytotic machinery (fig. 6).18,24–26The second phase involves continuous stimulation of the triggering signal and amplification of the effects of Ca²⁺on exocytosis through a K^ATPchannel–independent mechanism. The molecular mechanisms underlying this amplification are still unknown.25Our results have shown that isoflurane inhibits insulin secretion during the early phase of an IVGTT, suggesting that isoflurane inhibits the triggering pathways through a K^ATPchannel–dependent mechanism.

In the current study, we demonstrated that isoflurane inhibits insulin secretion induced by administration of glucose and by administration of a K^ATPchannel blocker. Patch clamp experiments further showed that isoflurane significantly increased K^ATPchannel activity in the presence of 10 mm glucose. These results strongly suggest that isoflurane-induced activation of K^ATPchannels in pancreatic β cells mediates the effects of isoflurane on insulin secretion and glucose utilization. It has been known since the 1970s that volatile anesthetics can impair insulin secretion and glucose utilization.10–15However, the precise mechanism underlying these effects is unknown. We are the first to suggest that isoflurane-induced inhibition of insulin secretion is mediated by the isoflurane-induced opening of K^ATPchannels in β cells of the pancreas.

Glibenclamide treatment in the glibenclamide alone and glibenclamide plus isoflurane groups consisted of a 50-μg/kg bolus of glibenclamide followed by 33.5 μg · kg⁻¹· h⁻¹. In the glibenclamide (alone) group, IRI levels were markedly increased after glucose administration and remained elevated for 2 h. Interestingly, isoflurane significantly decreased insulin secretion during the early and late phases of the IVGTT in the glibenclamide plus isoflurane group compared with the glibenclamide group. The continuous infusion of glibenclamide resulted in high IRI levels during the early and late phases of the IVGTT. The inhibitory effect of glibenclamide on K^ATPchannels in pancreatic β cells would be expected to induce and maintain high levels of insulin secretion during the early and late phases of the IVGTT. However, isoflurane inhibited the effects of glibenclamide, suggesting that isoflurane-induced activation of K^ATPchannels inhibits glibenclamide-induced insulin secretion.

Glibenclamide (45 min after administration) increased insulin levels from 5.6 ± 3.3 to 13.7 ± 7.5 μU/ml and from 4.6 ± 1.5 to 12.3 ± 6.9 μU/ml in the glibenclamide and glibenclamide plus isoflurane groups, respectively. Thirty minutes after administration of 1.0 MAC isoflurane, insulin levels were decreased (although not significantly) to 10.0 ± 6.7 μU/ml. In the glibenclamide group, insulin levels remained unchanged (13.7 ± 5.1 μU/ml) when isoflurane was not administered. Likewise, there were no significant differences between the glibenclamide and glibenclamide plus isoflurane groups in blood glucose levels at baseline, before administration of isoflurane, or before glucose administration. These results suggest that the relatively modest effects of isoflurane on insulin secretion in the absence of administration of glucose are unlikely to have clinical relevance.

Hyperglycemia, severe hyperglycemia in particular, is associated with increased morbidity and mortality in a variety of patient populations.27Although volatile anesthetics have been reported to have cardioprotective effects mediated by activation of cardiac K^ATPchannels,5,6the results of the current study indicate that isoflurane impairs glucose tolerance and insulin secretion via  activation of pancreatic K^ATPchannels. How, then, should anesthesiologists respond to these conflicting effects of volatile anesthetics in the operating room? Several randomized clinical trials of volatile anesthetics in patients undergoing coronary artery surgery indicate that volatile anesthetics decrease troponin release, duration of stay in intensive care units, and late cardiac events, and enhance left ventricular function compared with the nonuse of volatile anesthetics.28These trials formed the basis for the suggestion in the American College of Cardiology–American Heart Association 2007 guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery that it can be beneficial to use volatile anesthetic agents during noncardiac surgery for maintenance of general anesthesia in hemodynamically stable patients at risk for myocardial ischemia (class IIa, level of evidence B).28These data suggest that volatile anesthetics have beneficial cardioprotective effects that outweigh the adverse effects of decreased insulin secretion. However, hyperglycemia has been reported to attenuate isoflurane-induced cardioprotection29and to impair isoflurane-induced K^ATPchannel activation in vascular smooth muscle cells.30A recent study has demonstrated that β-cell function was already impaired in subjects with impaired glucose tolerance and in patients with early type 2 diabetes.31In this study, we have shown in healthy rabbits that exposure to isoflurane by itself without glucose injection produced insulin and blood glucose values comparable to the vehicle group. However, isoflurane by itself without glucose overload may decrease insulin secretion in these subjects. Taken together, to induce the beneficial cardiovascular effects of isoflurane, control of blood glucose concentration using insulin may be required during general anesthesia with volatile anesthetics, because isoflurane inhibits insulin secretion, as shown in the current study. Insulin has been reported to protect the myocardium against ischemia and reperfusion injury32and to induce hyperpolarization via  K^ATPchannel opening in vascular smooth muscle cells.33 

Volatile anesthetics are thought to activate K^ATPchannels in cardiac myocytes and vascular smooth muscle cells. Selective K^ATPchannel antagonists such as HMR-1098 (selective for sarcolemmal channels) and 5-hydroxydecanoate (selective for mitochondrial channels) abolish the protective effects of desflurane in response to ischemia and reperfusion injury in dogs,34suggesting that sarcolemmal and mitochondrial K^ATPchannels might play a role in anesthetic preconditioning in cardiac myocytes. Some volatile anesthetics have been shown to enhance sarcolemmal K^ATPchannel currents in cardiac myocytes through facilitation of channel opening after initial activation,20,35to produce coronary vasodilatation through activation of vascular K^ATPchannels,7,8and to hyperpolarize vascular smooth muscle through protein kinase A–induced K^ATPchannel activation.9Data from mutagenesis patch clamp experiments using expressed sulfonylurea (SUR) 2B/Kir6.1 channels (vascular type K^ATPchannels) from our laboratory have shown that protein kinase A–mediated phosphorylation of Kir6.1 and SUR2B subunits plays a pivotal role in isoflurane-induced vascular type K^ATPchannel activation.21In the current study, we extend these findings to show that isoflurane activates K^ATPchannel activity in pancreatic β cells.

There are some limitations of the current study. We did not determine whether other K^ATPchannel agonists, such as diazoxide, have a similar effect to isoflurane on insulin secretion. We could not determine the appropriate concentration of diazoxide to inhibit insulin secretion during IVGTTs in rabbits. We demonstrated in the patch clamp experiments that isoflurane can mimic the effects of diazoxide on activation of pancreatic K^ATPchannels during hyperglycemia (fig. 5). This may mean that diazoxide-induced activation of pancreatic K^ATPchannels also inhibits insulin secretion. We also did not determine whether other volatile anesthetics inhibit insulin secretion. Therefore, it is unclear whether this is an agent-specific phenomenon or whether it is a common property of the volatile anesthetics. Further studies are required to answer this question.

In conclusion, the results of the current study demonstrate that 1.0 MAC isoflurane decreases glucose-induced insulin secretion and increases blood glucose concentrations during the early phase of an IVGTT in rabbits. Isoflurane also attenuates glibenclamide-induced insulin secretion during the late phase of an IVGTT. Patch clamp experiments indicate that 0.5 mm isoflurane increases K^ATPchannel activity in pancreatic β cells in the presence of 10 mm glucose. Taken together, these results suggest that isoflurane impairs insulin secretion and glucose utilization. These results further indicate that the inhibitory effects of isoflurane on glucose utilization might be mediated by isoflurane-induced K^ATPchannel activation in pancreatic β cells.