The neuroendocrine stress response compensates for suppression of insulin secretion by volatile anesthetic agents: An observational study

Abstract

Alterations in perioperative metabolic function, particularly hyperglycemia, are associated with increased post-operative complications, even in patients without preexisting metabolic abnormalities. Anesthetic medications and the neuroendocrine stress response to surgery may both contribute to altered energy metabolism through impaired glucose and insulin homeostasis but the discrete pathways involved are unclear. Prior human studies, though informative, have been limited by analytic sensitivity or technique, preventing resolution of underlying mechanisms. We hypothesized that general anesthesia with a volatile agent would suppress basal insulin secretion without altering hepatic insulin extraction, and that surgical stress would promote hyperglycemia through gluconeogenesis, lipid oxidation, and insulin resistance. In order to address these hypotheses, we conducted an observational study of subjects undergoing multi-level lumbar surgery with an inhaled anesthetic agent. We measured circulating glucose, insulin, c-peptide, and cortisol frequently throughout the perioperative period and analyzed the circulating metabolome in a subset of these samples. We found volatile anesthetic agents suppress basal insulin secretion and uncouple glucose-stimulated insulin secretion. Following surgical stimulus, this inhibition disappeared and there was gluconeogenesis with selective amino acid metabolism. No robust evidence of lipid metabolism or insulin resistance was observed. These results show that volatile anesthetic agents suppress basal insulin secretion, which results in reduced glucose metabolism. The neuroendocrine stress response to surgery ameliorates the inhibitory effect of the volatile agent on insulin secretion and glucose metabolism, promoting catabolic gluconeogenesis. A better understanding of the complex metabolic interaction between anesthetic medications and surgical stress is needed to inform design of clinical pathways aimed at improving perioperative metabolic function.

FIGURE 1

Perioperative events and blood collection schematic. Triangles represent blood sampling. Black triangles denote samples for metabolomic analyses. A sample was drawn in pre‐operative hold (Pre‐Op) and another upon arrival to the operating room (OR). A sample was drawn immediately following induction, every 15 min for 2 h, then every 30 min until extubation. Finally, a sample was drawn on arrival to the post‐anesthesia care unit (PACU) and again 60 min later. Average time (minutes ± SD) until the next event is noted on the schematic.

FIGURE 2

Perioperative glucose, insulin, and cortisol dynamics. Glucose concentrations (a) and fold change from baseline (b) over the perioperative period. Insulin concentrations (c) and fold change from baseline (d). Circulating c‐peptide levels (e) and fold change from baseline (f). Perioperative cortisol concentrations (g) and fold change from baseline (h). Data are presented as mean ± SEM and are censored between 270 min and arrival in the post‐anesthesia care unit (PACU) due to low subject number. Dashed line represents baseline. *p < 0.05 compared to first sample or baseline by linear mixed effects modeling.

FIGURE 3

Alterations in perioperative insulin homeostasis. Insulin secretion rates (a) calculated from c‐peptide concentrations. Fold change insulin secretion rate from baseline (b; dashed line represents baseline). Hepatic extraction ratio derived from difference in circulating insulin and c‐peptide levels (c). Insulin‐to‐glucose ratios over the perioperative period (d). Data are presented as mean ± SEM. Insulin secretion rates were not calculated from post‐anesthesia care unit (PACU) data. Hepatic extraction and insulin‐to‐glucose ratio data are censored between 270 min and arrival in the PACU due to low subject number. Dashed line represents baseline. *p < 0.05 compared to first sample or baseline by linear mixed effects modeling.

FIGURE 4

Perioperative alterations in the circulating metabolome. Score plots for the first two (a) and first three (b) components of the metabolite data from partial least squares discriminant analysis (PLS‐DA) show clear evolution over the perioperative period. Red triangles (1) are preoperative data, green pluses (2) are 15 min data, blue crosses (3) are 120 min data, and teal diamonds (4) are 60 min after arrival post‐anesthesia care unit (PACU) data. The shaded areas represent the 95% confidence regions for each time point. Percent variance explained by each component is noted in parentheses. Data labels are randomly assigned.

FIGURE 5

Perioperative carbohydrate and amino acid dynamics. Glucose (a), pyruvate (b), glycerol‐3‐phosphate (c), glyceraldehyde (d), and taurine (e) concentrations increase over the course of the perioperative period. Circulating glutamate concentrations are decreased in the post‐anesthesia care unit (f), while tryptophan (g) and xanthurenic acid (h) decreased steadily over the perioperative course. Time points are: 1—pre‐operative, 2–15 min after induction, 3–120 min after induction, and 4–60 min after arrival in post‐anesthesia care unit.

FIGURE 6

Perioperative metabolite pathway dynamics. Tryptophan and linolenic acid metabolism increase following induction (a, white). Carbohydrate and nucleotide metabolite networks are prominent after sustained surgical stimulation (a, light gray). In the post‐anesthesia care unit (PACU), amino acid catabolism and gluconeogenesis dominate the metabolome (b). Analyses of the metabolomic network over the whole perioperative period show broad amino acid and carbohydrate catabolism (c). Pathway impact derived from relative betweenness centrality of the metabolic networks and significance of each pathway determined by enrichment analyses (presented as negative log of the raw p‐value). Data presented have pathway impact >0.05 and false detection rate corrected p‐values <0.05.