Early Endotoxemia Increases Peripheral and Hepatic Insulin Sensitivity in Healthy Humans

Abstract

Context: Sepsis-induced hypoglycemia is a well known, but rare, event of unknown origin.

Objective: The aim of the study was to obtain insight into the mechanism of sepsis-induced hypoglycemia, focusing on glucose kinetics and insulin sensitivity measured with stable isotopes by using the model of human endotoxemia.

Design: Glucose metabolism was measured during two hyperinsulinemic [insulin levels of 100 pmol/liter (low-dose clamp) and 400 pmol/liter (medium-dose clamp)] euglycemic (5 mmol/liter) clamps on two occasions: without or with lipopolysaccharide (LPS).

Setting: The study was conducted at the Academic Medical Center, Metabolic and Clinical Research Unit (Amsterdam, The Netherlands).

Participants: Eighteen healthy male volunteers participated in the study.

Intervention: A hyperinsulinemic euglycemic (5 mmol/liter) clamp with LPS (two groups of six subjects; insulin infusion at rates of either 10 or 40 mU · m⁻² · min⁻¹) or without LPS (n = 6; both insulin infusions in same subjects).

Main Outcome Measure: We measured hepatic and peripheral insulin sensitivity.

Results: Hepatic insulin sensitivity, defined as a decrease in endogenous glucose production during hyperinsulinemia (100 pmol/liter), was higher in the LPS group compared to the control group (P = 0.010). Insulin-stimulated peripheral glucose uptake was higher in both clamps after LPS compared to the control setting (P = 0.006 and 0.010), despite a significant increase in the plasma concentrations of norepinephrine and cytokines in the LPS group during both clamps.

Conclusions: These data indicate that shortly (2 h) after administration of LPS, peripheral and hepatic insulin sensitivity increase. This may contribute to the hypoglycemia occurring in some patients with critical illness, especially in the setting of intensive insulin therapy.

Sepsis and critical illness are associated with an acute and reversible state of insulin resistance, characterized by hyperglycemia. Hypoglycemia is also seen in critical illness as an infrequent feature of early sepsis (1–3). In animal studies, it is known that hypoglycemia is the consequence of a decreased glucose production combined with a relative stimulation of glucose disposal by selective macrophage-rich tissues, mostly by an insulin-independent mechanism (4, 5). In humans, the pathophysiology of hypoglycemia during sepsis is unexplored. Because hypoglycemia is a rare and suddenly occurring event, requiring immediate glucose infusion, controlled studies with stable isotopes are not possible. However, iv administration of Gram-negative bacterial lipopolysaccharide (LPS) induces a systemic inflammatory response mimicking many of the clinical features associated with sepsis and can be administered safely in a well-defined manner to produce well-controlled effects (6). Only a few studies have actually measured aspects of whole body glucose metabolism in humans after administration of LPS (6–8). The study by Bloesch et al. (8) showed a significant decrease in plasma glucose concentration 2 h after LPS injection, followed by hyperglycemia. The decrease in glucose concentrations was ascribed to both a decrease in endogenous glucose production rate (rate of appearance of glucose/Ra) as well as an increase in peripheral glucose uptake (rate of disappearance/Rd). The changes in Ra and Rd of glucose, however, did not match in time with the short-lasting drop in plasma glucose concentrations (8). Agwunobi et al. (7) reported a significant increase in glucose infusion rate during a hyperinsulinemic euglycemic clamp 2 h after a LPS bolus, which they ascribed to an increase in peripheral glucose disposal. However, because they did not use isotopes to measure glucose fluxes, this explanation remains an assumption.

The later phases of critical illness are characterized by insulin resistance, influencing the normal balance between glucose production and uptake. An aspect of the pathophysiology of hypoglycemia in early sepsis could be increased insulin sensitivity. This is especially relevant because it is more or less daily practice that severely ill patients, admitted to the intensive care unit, receive intensive insulin therapy to achieve euglycemia and start almost immediately with feeding, both resulting in hyperinsulinemia (9, 10).

Because no data on insulin sensitivity in early sepsis in humans exists, we studied glucose metabolism with the use of stable isotopes during a hyperinsulinemic (100 pmol/liter and 400 pmol/liter) euglycemic (5 mmol/liter) clamp in healthy male volunteers after LPS administration and in a control setting.

Subjects and Methods

Subjects

Eighteen healthy, nonsmoking, male volunteers were included. None of them used medication, had an infection in the preceding 3 months, was obese (defined as a body mass index >25 kg/m²), or had a positive family history of diabetes. All volunteers had normal plasma values of fasting glucose concentration, erythrocyte sedimentation rate, complete blood count, lipid profile, creatinine, and liver enzymes, and all had a normal oral glucose tolerance test according to the American Diabetes Association criteria (11). The study was approved by the Medical Ethical Committee of the Academic Medical Center in Amsterdam, and all subjects gave written informed consent.

Protocol

Volunteers (n = 18) were studied during euglycemia (5 mmol/liter) during either a plasma insulin level of 100 pmol/liter (low-dose hyperinsulinemic clamp) or an insulin level of 400 pmol/liter (medium-dose hyperinsulinemic clamp) to study hepatic and peripheral insulin sensitivity, respectively. In the control group, volunteers (n = 6) were studied twice, i.e. during both clamps. Because it is not feasible to administer LPS twice to the same subjects, six volunteers underwent the low-dose hyperinsulinemic clamp concomitant with LPS, and six other volunteers underwent the medium-dose hyperinsulinemic clamp aimed at an insulin level of 400 pmol/liter plus LPS.

The present study was part of a study on the differential effects of plasma glucose and insulin concentrations on several parameters relevant for critically ill patients (12–15). The data on glucose metabolism, measured with stable isotopes in the present study, have not been published earlier.

For 3 d before the study, all volunteers consumed approximately 250 g of carbohydrates and were asked to refrain from vigorous exercise. Infusion protocols for the control and LPS groups have been published previously (12, 14). In short, at T = 0:00 (0800 h in the LPS group, and 0900 h in the control group), a blood sample was drawn for determination of background enrichment of plasma glucose. (The difference in time was due to a slight difference in study protocol.) Thereafter, a primed-continuous infusion of [6,6-²H₂]glucose (prime, 8.0 μmol/kg; continuous, 0.11 μmol/kg · min; >98% pure and >99% enriched; ARC Laboratories BV, Apeldoorn, The Netherlands) together with somatostatin, glucagon, insulin (10 or 40 mU/m² · min) and glucose (10 or 20%) at a variable rate to maintain euglycemia (5 mmol/liter), was started and continued for 6 h in the control group and for 5 h in the LPS group. [6,6-²H₂]glucose was added to the variable glucose infusion to minimize changes in the achieved plasma isotopic enrichment of approximately 1%.

At T = 3:00 (after 3 h of clamping), LPS (Escherichia coli LPS) was administered to the LPS group (4 ng/kg).

Analytical procedures

Plasma glucose was measured every 5 min at bedside from T = 0:00 until T = 6:00 in the control group and until T = 5:00 in the LPS group. From T = 2:20 until T = 3:00 and from T = 5:20 until T = 6:00 in the control group, and from T = 2:20 until T = 3:00 and T = 4:20 until T = 5:00 in the LPS group, five samples were drawn for measurement of glucose enrichment. The results of the glucose enrichments will be presented as mean of the five samples at T = 3:00 and at T = 5:00 (in the LPS group) or T = 6:00 (in the control group).

Glucagon, insulin, cortisol, (nor)epinephrine, free fatty acid, and total adiponectin were measured at T = 3:00 and T = 6:00 in the control group and at T = 3:00 and T = 5:00 in the LPS group.

IL-6, IL-8, IL-10, and TNF-α were measured at T = 6:00 in the control group and at T = 5:00 in the LPS group.

Measurements of all these parameters and [6,6-²H₂]glucose enrichment were performed as described before (12, 16).

Calculation and statistics

Ra and Rd of glucose were calculated using the modified form of the Steele equations for non-steady-state measurements as described previously (17). Endogenous glucose production (EGP) was calculated as the difference between Ra glucose and glucose infusion rate. The change of EGP, Rd, and glucoregulatory hormones during the low-dose and the medium-dose hyperinsulinemic clamp (T = 6.00 compared with T = 3.00 in the control group, and T = 5:00 compared with T = 3:00 in the LPS group) were calculated and expressed as the relative (%) difference.

Volunteer characteristics were compared using a Kruskal-Wallis test. For cytokines, single values measured during the low-dose and the medium-dose hyperinsulinemic clamp (T = 6.00 in the control group and T = 5:00 in the LPS group) were compared. All results were compared between the LPS and the control group using the Mann-Whitney U test. Data are presented as median [range]. Probability values of less than 0.05 were considered statistically significant. SPSS statistical software version 12.0.1 (SPSS Inc., Chicago, IL) was used to analyze the data.

Results

Subjects characteristics

Subjects in the control group and the LPS group (low-dose and medium-dose hyperinsulinemic clamp, respectively) were similar in all aspects; age (in years), 22 [22–23] vs. 23 [23–25] and 25 [21–27], respectively; body mass index (kg/m²), 21.7 [21.0–23.4] vs. 22.0 [19.9–23.6] and 22.1 [20.9–23.7], respectively.

Glucose kinetics during the low-dose hyperinsulinemic clamp in the control group and the LPS group (Table 1)

Table 1.

Glucoregulatory hormones and glucose kinetics during the low-dose and medium-dose hyperinsulinemic euglycemic (5 mmol/liter) clamps in the control and LPS groups

T	Low-dose hyperinsulinemic clamp				Medium-dose hyperinsulinemic clamp
	Control group		LPS group		Control group		LPS group
	3:00	6:00	3:00	5:00	3:00	6:00	3:00	5:00
Glucose (mmol/liter)	4.8 [4.7–5.0]	5.1 [5.0–5.4]a	4.9 [4.6–5.1]	5.1 [4.6–5.5]	4.9 [4.7–5.2]	5.0 [4.6–5.3]	4.9 [4.6–5.0]	5.1 [4.7–5.5]
Insulin (pmol/liter)	93 [80–123]	93 [80–106]	96 [62–122]	70 [58–84]a	436 [332–463]	431 [332–487]	307 [277–483]	260 [234–360]a
Total adiponectin (μg/ml)	4.8 [4.2–7.7]	4.5 [3.8–6.5]a	7.2 [4.6–11.5]	7.6 [4.2–11.5]	5.6 [2.9–8.0]	5.4[2.8–7.2]a	6.6 [6.1–11.8]	6.5 [6.1–10.6]
FFA (mmol/liter)	0.02 [0.02–0.03]	0.02 [0.02–0.03]	0.02 [0.02–0.09]	0.04 [0.02–0.09]	0.02 [0.02–0.05]	0.02	0.02 [0.02–0.03]	0.03 [0.02–0.03]
Norepinephrine (nmol/liter)	1.14 [0.46–10.2]	0.95 [0.71–3.74]	0.73 [0.49–1.07]	3.01 [0.91–4.17]a	0.77 [0.42–1.76]	0.90 [0.47–1.24]	0.69 [0.41–0.82]	2.20 [1.41–6.18]a
Epinephrine (nmol/liter)	0.06 [0.05–0.21]	0.06 [0.05–0.18]	0.08 [0.05–0.09]	0.19 [0.07–0.59]	0.08 [0.05–0.14]	0.08 [0.05–0.23]	0.07 [0.05–0.14]	0.23 [0.06–0.48]a
Glucagon (ng/liter)	63 [42–102]	58 [49–99]	50 [21–81]	42 [26–55]	59 [52–92]	54 [44–82]	52 [32–71]	48.[27–56]
Cortisol (nmol/liter)	199 [73–355]	211 [183–497]	170 [54–295]	530 [463–674]a	163 [115–287]	206 [124–312]	252 [142–306]	545 [510–646]a
EGP (μmol/kg · min)	10.0 [6.0–16.3]	4.2 [2.9–7.3]a	8.9 [4.9–17.8]	1.5 [0.9–1.7]a	0.3 [0.0–3.4]	0.3 [0.0–3.8]	2.9 [0.0–5.0]	2.0 [0.0–5.0]
Rd (μmol/kg · min)	28.2 [18.1–33.5]	31.0[25.0–34.5]a	21.4 [14.9–36.9]	35.4 [20.2–64.8]a	54.1 [44.0–67.2]	58.4 [51.9–66.5]	55.1 [40.0–71.3]	80.2 [66.9–115.4]a

Data are expressed as median [range], n = 6 per group. T, Time; FFA, free fatty acids.

^a

P < 0.05 when comparing T = 6:00 or T = 5:00 to T = 3:00 within the control or LPS groups.

Euglycemia was maintained during all clamps. At T = 3:00 (just before LPS administration), glucose concentrations, EGP, and Rd in the control and the LPS group were similar.

In the control subjects between T = 3:00 and T = 6:00, EGP decreased significantly by 55% and Rd increased significantly by 5%.

In the LPS-treated subjects, EGP decreased significantly after LPS administration by 82%. Rd increased significantly after LPS infusion by 75%.

In the LPS group, the relative decrease in EGP and the relative increase in peripheral glucose disposal were significantly higher compared with those values in the control group (P = 0.010 and 0.006, respectively).

Glucose kinetics during the medium-dose hyperinsulinemic clamp in the control group and the LPS group (Table 1)

Euglycemia was maintained in all clamps. At T = 3:00, glucose concentration, EGP, and Rd in the control group and the LPS group were similar.

In the control group, EGP and Rd remained similar between T = 3:00 and T = 6:00.

In the LPS group between T = 3:00 and T = 5:00, EGP was unaffected 2 h after LPS. Rd increased significantly by 53% 2 h after LPS.

This relative increase in Rd in the LPS group was significantly higher compared with the relative increase in Rd in the control group (12%) (P = 0.010).

Glucoregulatory hormones during the low-dose and the medium-dose hyperinsulinemic clamp in the control group and the LPS group (Table 1)

At T = 3:00 during the low-dose and the medium-dose hyperinsulinemic clamp, there was no difference in plasma concentration of any of the hormones between the control and the LPS group. During the low-dose clamp in the control group, plasma concentration of total adiponectin decreased significantly by 7%. In the LPS group during the low-dose hyperinsulinemic clamp, norepinephrine and cortisol significantly increased by 296 and 250%, respectively. The increase in plasma concentration of norepinephrine in the LPS group was significantly higher compared with the 11% decrease of plasma norepinephrine concentration in the control group (P = 0.006).

During the medium-dose hyperinsulinemic clamp in the control group, plasma concentration of total adiponectin decreased significantly by 6%. In the LPS group, during the medium-dose clamp, there was a significant decrease in plasma insulin concentration of 19% and a significant increase in plasma norepinephrine, epinephrine, and cortisol concentration by 413, 243, and 153%, respectively. This decrease in the plasma insulin concentration and the increase of the norepinephrine concentration in the LPS group were significantly higher compared with the increase of plasma insulin concentration (3%) and the plasma norepinephrine concentration (11%) in the control group (both P = 0.004).

Cytokines during the low-dose and the medium-dose hyperinsulinemic clamp in the control and the LPS group (Table 2)

Table 2.

Plasma cytokine concentrations during the low-dose and medium-dose hyperinsulinemic euglycemic (5 mmol/liter) clamps in the control (T = 6:00 h) and LPS (T = 5:00 h) groups

	Low-dose hyperinsulinemic clamp		Medium-dose hyperinsulinemic clamp
	Control group	LPS group	Control group	LPS group
TNF-α (pg/ml)	3 [3–302]	10,002 [4,055–38,158]a	3 [3–173]	10,610 [6,266–19,583]a
IL-6 (pg/ml)	3 [3–498]	2,556 [1,962–10,555]a	5 [3–341]	2,948 [1,809–4,868]a
IL-8 (pg/ml)	13 [9–755]	32,78 [2,162–4,378]a	10 [7–460]	3143 [2181–4304]a
IL-10 (pg/ml)	3 [3–28]	53 [37–76]a	3 [3–15]	69 [24–300]a

Data are expressed as median [range], n = 6 per group.

^a

P < 0.05 when comparing the LPS group to the control group.

During the low-dose and the medium-dose hyperinsulinemic clamp, all cytokines were significantly higher in the LPS group compared with the control group.

Discussion

In early sepsis (i.e. during the first hours), hypoglycemia is an infrequent feature, whereas hyperglycemia is a more common phenomenon in later stages (1–3). The shift of hypoglycemia to hyperglycemia is not completely explained, but it is assumed that increased need for substrate availability for repair of damaged tissues or tissues with an increased need (immune system) is the driving force. The pathophysiological mechanism underlying hypoglycemia is unknown and cannot be investigated in clinical sepsis because it has to be corrected instantaneously. Administration of LPS, which initiates a systemic inflammation (fever, headache, tachycardia, and stress hormone responses), can be used as a validated model to mimic the clinical presentation of bacterial sepsis. In the present study, we show that administration of LPS results in an increase in peripheral and hepatic insulin sensitivity. Our data on EGP and Rd during insulin clamps extend prior assumptions about LPS-induced glucose kinetics that the decrease in plasma glucose concentration 2 h after LPS is due to an increased Rd as well as a decrease in EGP (8).

The decrease in EGP is a phenomenon found previously in animal studies (5). Theoretically this decrease could be caused by either a depletion of glycogen or a decrease in gluconeogenesis and/or glycogenolysis. The explanation may be due to a lack of substrate (such as alanine), a change in activity of glucose transporters in hepatocytes, or an acute inhibition of enzymes involved in these processes (18, 19). Although decreased expression of genes encoding for important gluconeogenetic enzymes, such as posphoenolpyruvate carboxykinase and glucose-6-phosphatase, has been reported after long-term exposure to LPS, this seems less likely in our acute experiments (20, 21).

The increased Rd can be the result of an increase in insulin and/or non-insulin-mediated glucose uptake. In rodents, it has been shown that LPS directly stimulates non-insulin-mediated glucose uptake by macrophage-rich tissue. However, indirect stimulation of glucose uptake via an increase in LPS-induced circulating factors with insulin-like properties, such as cytokines, cannot be ruled out (4). As expected, after the administration of LPS, all plasma cytokine concentrations were significantly higher compared with the control group. The role of these inflammatory mediators is still controversial. On the one hand, TNF-α, IL-6, IL-8, and IL-10 are associated with insulin resistance, which hampers peripheral glucose uptake (22–24). On the other hand, infusion of TNF-α and IL-6 is known to enhance Rd glucose, indicating a possible role for these mediators (25–28). However, because the infusion of IL-6 has previously shown to increase EGP, whereas TNF-α did not affect EGP, these factors cannot fully explain the increased insulin sensitivity found in our study (26, 28).

The decrease of total plasma adiponectin concentrations in the control group during hyperinsulinemia is a known phenomenon (15). Although the change in plasma adiponectin levels did not differ between the LPS and the control group, the blunted decrease in adiponectin upon insulin infusion after administration of LPS may contribute to the higher Rd and the lower EGP in this group.

Norepinephrine induces insulin resistance and can explain neither the increased Rd nor the decreased EGP in the LPS group (29, 30).

A potential confounding factor in our study might be the 1-h difference in insulin infusion between the LPS group and the control group. However, this is an unlikely explanation for our results. It is known that the effect of insulin infusion increases in time (i.e. lower EGP and higher Rd) (31). The results of our study, however, show the opposite: in the group with the longest duration of insulin infusion (the control group), suppression of EGP was less and Rd was lower compared with the group with the shorter duration of insulin infusion (the LPS group).

The same holds true for the lower insulin concentrations obtained in the LPS group. For the decreased plasma insulin concentrations, an increased clearance, rather than a decreased production, seems an explanation (32). The decrease in plasma insulin concentration, however, does not contradict our conclusion; as with similar insulin concentrations (i.e. the higher concentrations in the control group), the differences in Rd and EGP between groups probably would have been even more pronounced.

In the control group during the low-dose hyperinsulinemic 5 mmol/liter clamp, plasma glucose concentration at T = 6:00 was significantly, but only slightly (0.3), higher than at T = 3:00. Although statistically significant, we do not consider this small difference to be relevant. In addition, it does not falsify our data because higher plasma concentrations of glucose stimulate Rd via mass effect. Because Rd was lower in the control group, the difference between control group and LPS group would therefore probably have been more significant in the case of equal glucose concentrations.

LPS injection into healthy humans results in a transient rise in body temperature. In this respect, it is important to note that this rise only occurred from 2 h after LPS injection onward and to a similar extent in all LPS groups (data not shown). Hence, changes in body temperature are unlikely to have influenced our results.

The studies were done in relatively young, lean males. Our results are therefore not necessarily applicable to insulin-resistant subjects, like overweight or elderly patients. Furthermore, the model of LPS used in this experiment might induce more transient changes than are probably the case in “real” sepsis. It is unknown whether LPS- and sepsis-induced hypoglycemia share the same mechanism.

In conclusion, as an extension to earlier studies, the results of our study show that 2 h after administration of LPS, peripheral and hepatic insulin sensitivity increase. This is of particular importance because, nowadays, most severely ill patients admitted to the intensive care unit receive intensive insulin therapy to achieve euglycemia and start on feeding very early on, resulting in exogenous and endogenous hyperinsulinemia, respectively. This clinical setting in combination with the increased biological effects of insulin during LPS-induced sepsis may put the patients at risk for developing hypoglycemia and in fact may explain the incidence of hypoglycemia seen in septic patients during intensive insulin treatment (9). Unraveling the mechanism through which LPS establishes this effect seems helpful in treating patients by stabilizing glucose metabolism in sepsis.

Acknowledgments

We thank A. F. C. Ruiter, B. C. E. Voermans, and the staff of the Clinical Research Unit of our hospital for their indispensable help during the experiments.

Disclosure Statement: S.N.v.d.C., R.M.E.B., M.T.A., E.E., M.W.T.T., M.J.S., T.v.d.P., and H.P.S. have nothing to disclose. M.E.S. has received a grant from the Dutch Diabetes Research Foundation (no. 2002.00.008).

For editorial see page 416

Abbreviations

 
• EGP

  Endogenous glucose production

 
• LPS

  lipopolysaccharide

 
• Ra

  rate of appearance

 
• Rd

  rate of disappearance