Insulin Sensitivity and Mitochondrial Function Are Improved in Children With Burn Injury During a Randomized Controlled Trial of Fenofibrate

Abstract

Objective:

To determine some of the mechanisms involved in insulin resistance immediately following burn trauma, and to determine the efficacy of PPAR-α agonism for alleviating insulin resistance in this population.

Summary Background Data:

Hyperglycemia following trauma, especially burns, is well documented. However, the underlying insulin resistance is not well understood, and there are limited treatment options.

Methods:

Twenty-one children 4 to 16 years of age with >40% total body surface area burns were enrolled in a double-blind, prospective, placebo-controlled randomized trial. Whole body and liver insulin sensitivity were assessed with a hyperinsulinemic-euglycemic clamp, and insulin signaling and mitochondrial function were measured in muscle biopsies taken before and after ∼2 weeks of either placebo (PLA) or 5 mg/kg of PPAR-α agonist fenofibrate (FEN) treatment, within 3 weeks of injury.

Results:

The change in average daily glucose concentrations was significant between groups after treatment (146 ± 9 vs. 161 ± 9 mg/dL PLA and 158 ± 7 vs. 145 ± 4 FEN; pretreatment vs. posttreatment; P = 0.004). Insulin-stimulated glucose uptake increased significantly in FEN (4.3 ± 0.6 vs. 4.5 ± 0.7 PLA and 5.2 ± 0.5 vs. 7.6 ± 0.6 mg/kg per minute FEN; pretreatment vs. posttreatment; P = 0.003). Insulin trended to suppress hepatic glucose release following fenofibrate treatment (P = 0.06). Maximal mitochondrial ATP production from pyruvate increased significantly after fenofibrate (P = 0.001) and was accompanied by maintained levels of cytochrome C oxidase and citrate synthase activity levels. Tyrosine phosphorylation of the insulin receptor and insulin receptor substrate-1 in response to insulin increased significantly following fenofibrate treatment (P = 0.04 for both).

Conclusions:

Fenofibrate treatment started within 1 week postburn and continued for 2 weeks significantly decreased plasma glucose concentrations by improving insulin sensitivity, insulin signaling, and mitochondrial glucose oxidation. Fenofibrate may be a potential new therapeutic option for treating insulin resistance following severe burn injury.

Insulin resistance and hyperglycemia are common after burn injury, but the mechanisms were unknown, and treatment options limited. A randomized controlled trial was performed during the acute burn period in children with >40% TBSA burns examining glucose metabolism before and after 2 weeks PPAR-α treatment. Treatment decreased mean plasma glucose concentrations and increased whole body insulin sensitivity, muscle insulin signaling, and mitochondrial glucose oxidation compared with placebo-treated controls.

Multiple studies have documented hyperglycemia and insulin resistance following trauma, myocardial infarction, stroke, or surgery.^1–4 Both are of serious clinical concern, as hyperglycemia is associated with increased morbidity and mortality of critically ill, surgical, and burned patients.^3–5 These outcome studies emphasize the importance of better understanding insulin resistance in this large patient population. Pediatric burns provide a unique model for studying acute onset insulin resistance, since the nature of the injury is severe and quantifiable and the timing of onset is exactly known. Further, lean children are much less likely to have preexisting comorbid conditions such as diabetes, or to have been taking medications prior to injury.

We have recently demonstrated that mitochondrial oxidative function is impaired in burned children by more than 70% at 1 week postburn.⁶ A limited number of animal studies have found that mitochondrial genes, protein, and function are also decreased in skeletal muscle, liver, and cardiac muscle following burn injury.^7–9 Similar findings were reported from the liver of patients who died after prolonged stays in intensive care for nonsurgical illness.¹⁰ In patients with type II diabetes, correlations between mitochondrial function and insulin sensitivity have been documented.^11,12 These studies indicate that improving mitochondrial function may be a potential treatment target for improving insulin sensitivity.

There are limited proven therapies for treating acute injury-induced insulin resistance. Insulin therapy to prevent hyperglycemia has been shown to reduce mortality and/or morbidity in intensive care patients, patients undergoing surgical procedures and in burns.^4,10 However, intensive insulin therapy requires constant vigilance and does not alter the underlying pathology. One study found that metformin decreased muscle loss after burn,¹³ but other medications that are typically prescribed for treatment of insulin resistance have not been tested in these populations.

PPAR-γ agonists, or thioglitazones, have been shown to improve insulin sensitivity in patients with diabetes.¹⁴ The mechanism of action apparently involves suppression of peripheral lipolysis and redistribution of triglyceride stores to subcutaneous fat cells.¹⁴ However, burned children often have a large amount of their subcutaneous triglyceride stores destroyed or removed, decreasing the availability for redistribution sites. Further, burn patients may be treated with the beta-blocking agent propranolol to reduce myocardial stress, which also suppresses lipolysis.¹⁵ Therefore, a PPAR-γ agonist would be anticipated to be redundant in children given propranolol.

PPAR-α agonists have been shown in numerous animal studies and a few human studies to improve insulin sensitivity by enhancing mitochondrial function.^16–18 We therefore hypothesized that PPAR-α agonism would stimulate mitochondrial function and improve insulin sensitivity in pediatric burn patients.

METHODS

Protocol

This was a prospective, randomized, double-blind, placebo-controlled clinical trial. The protocol was approved by the IRB at the University of Texas Medical Branch. A legal guardian provided permission for participation of the child. Medical assent was obtained in children 7 to 17 years of age when possible.

Children 4 to 17 years of age and >20 kg with >40% total body surface area burns that would require skin grafting, who arrived at our hospital within 96 hours of injury, were eligible. Children with major electrical burns, renal or hepatic failure, iodine allergies, severe sepsis, or who had required resuscitation from cardiac arrest were excluded.

Two 8-hour isotopic tracer studies were conducted 2 weeks apart, and the patients received treatment with fenofibrate (Tricor; Abbott Laboratories, Abbott Park, IL) a PPAR-α agonist, or placebo (Fig. 1A). The study duration chosen was 2 weeks, since many patients with 40% to 60% burns only require stays in the intensive care unit 3 weeks postinjury.

FIGURE 1. A, Overall study design. OR, operation. B, Infusion study design.

The details of the tracer studies protocol are shown in Figure 1B. The tracer studies were performed approximately 4 days after the child's first and third excision and skin grafting operations. This timing avoided acute surgery-induced metabolic changes, yet made use of the existing arterial and venous access. All initial metabolic studies were conducted within 10 days postburn and the second studies were ∼18 to 21 days postburn. Each study was preceded by a 4-hour fast with intravenous fluids of only saline, and 8 hours without blood or albumin transfusions. The tracer protocol started with background samples taken before a 4-hour basal period that was followed by a 4-hour hyperinsulinemic-euglycemic clamp. During the clamp, insulin was infused at a rate of 1.5 mU · kg⁻¹ · min⁻¹ into a central vein and 20% dextrose was simultaneously infused to maintain a plasma glucose concentration between 80 and 90 mg/dL. A primed constant infusion of 6.6 d₂ glucose was started at the beginning of the study at a rate of 0.444 μmol · kg⁻¹ · min⁻¹ and maintained throughout the entire study period. Arterial samples for glucose enrichment were taken 10 minutes apart in triplicate at the end of each period. Muscle biopsies (∼50 mg) were taken from the vastus lateralis at 1 and 8 hours of tracer infusion using a Bergstrom needle for analysis of insulin signaling and mitochondrial function.

Medical care was determined by faculty surgeons, fellows, and residents according to clinical protocols that have been previously described.¹⁹ Plasma glucose was kept below 180 mg/dL by insulin infusion if necessary. Tight glycemic control below 120 mg/dL in this population is currently under study but has not yet proven to be more efficacious than maintaining glucose values below 180 mg/dL. Liver enzymes were measured every other day to monitor possible hepatotoxic effects of fenofibrate.

Patients were fed with Vivonex T.E.N. (Novartis, Minneapolis, MN; 82% carbohydrate, 15% protein, 3% fat) at 1.4 times their measured resting energy expenditure. The resting metabolic rate was determined once a week in the early morning at 30°C with a V-max 29 metabolic cart (Sensormedics, Yorba Linda, CA). The patients received nutritional supplements including a multivitamin, folic acid, zinc, and vitamin C.

Prior to the start of the study, it was determined that 6 subjects would be needed to see a significant 2 mg · kg⁻¹ · min⁻¹ change in glucose uptake, the primary end point of the study. All data are presented as the mean ± standard error of the mean. The baseline data between groups was compared with an unpaired t test with an alpha value of 0.05, to ensure that the groups were similar prior to treatment. The response to insulin before and after treatment was the primary endpoint for glucose metabolism and insulin signaling. Thus, the change from the basal period to the clamp period for each study was calculated and compared from pre to posttreatment using a paired t test. For the mitochondrial data, the difference between pretreatment and posttreatment was compared. Values were compared by means of a paired t test; and since change was predicted a priori, the alpha value is 0.10. All statistics were preformed with Sigma Stat software package, version 2.03.

Assignment

Each individual was assigned to a drug allocation upon enrollment in the trial. Drug allocations were predetermined prior to initiation of the protocol by means of a randomization schedule created by M.G.C. on Sigma Stat version 2.03.

Blinding Procedures

M.G.C. was unblinded to treatment allocation but had no clinical responsibilities, and A.A. was unblinded to monitor safety parameters in the fenofibrate treated children. All others were blinded at the time of the protocol, especially the treating physicians, D.N.H., J.O.L., and A.P.S. All clinical records reflected fenofibrate treatment, and the true allocation was secured within the hospital pharmacy; 5 mg/kg of fenofibrate, the European pediatric dose for hyperlipidemia, was ground in 1 to 2 mL of 200 proof ethyl alcohol and suspended in 4 mL of drug suspension agent with flavoring. The placebo consisted of the suspension and flavoring agents; thus, the drugs appeared identical to administering nursing staff and to patients. The medications were administered daily at 13:00 through a nasogastric tube or orally.

Sample Analysis

Plasma glucose concentration was measured on an YSI 2300 Stat glucose/lactate analyzer (YSI, Inc., Yellow Springs, OH). Serum insulin concentrations were measured using radioactive immuno assay (Diagnostic Laboratories, Los Angeles, CA). Plasma glucose was purified for measurement of enrichment, and the isotopic enrichment of the penta-acetate derivative was determined by gas chromatography mass spectrometry as described previously.²⁰ Plasma resistin and adiponectin were measured using standard ELISA kits and TNF-α using a Lincoplex bead assay (Linco, St. Charles, MO).

Mitochondrial enzyme activities of cytochrome C oxidase and citrate synthase were measured from homogenates of vastus lateralis biopsies in a sucrose/EDTA/Tris buffer as previously described.²¹ Maximal mitochondrial oxidation of pyruvate and mitochondrial respiratory rates were measured in fresh muscle tissue as previously described.^22,23 Following these measurements, the samples were weighed, and the citrate synthase activity and protein contents were measured.^24,25

The insulin receptor (IR) tyrosine phosphorylation and insulin receptor-substrate-1 (IRS-1) tyrosine phosphorylation were measured from frozen muscle samples using Western-blot separation and antibody probing methods.²⁶

Calculations

Glucose infusion rate in mg · kg⁻¹ · min⁻¹ was calculated by measuring the amount of 20% dextrose infused to maintain plasma glucose between 80 and 90 mg/dL.

The rate of endogenous release of glucose by the liver, representing both gluconeogenesis and glycogenolysis, was calculated by measuring the dilution of the infused tracer by unlabeled glucose. Thus during the basal period the calculation is:

Equation 1: Rate of appearance = rate tracer infusion/arterial enrichment.

where arterial enrichment is expressed as tracer/tracee ratio. However, during the hyperinsulinemic-euglycemic clamp, the rate of appearance of glucose was corrected by subtracting the amount of unlabeled glucose infused to maintain euglycemia. The resultant units for both of these equations are glucose release in mg · kg⁻¹ · min⁻¹.

RESULTS

The demographics and sustained injuries of the PLA and FEN groups were comparable at randomization. Nine subjects completed the study in each group. An additional 3 children, 1 in PLA and 2 in FEN, did not complete the study and are not included in the data. Failure to complete the study was due to an early discharge from smaller third degree injuries (PLA), missed doses of medication (FEN), and a death due to sepsis (FEN). The total percentage of surface area burned was similar between the groups. The mean time from burn to admission was 2 days, and the average length of stay between the groups was similar. The sample size would have had to be substantially larger to analyze any changes in the rates of sepsis or complications in clinical outcomes. The majority of subjects were of Hispanic descent, which reflects the racial distribution of patients admitted to our hospital. Table 1 shows some common laboratory values taken from the patients on the days of their studies. There were no differences in the laboratory values between the groups prior to the study.

TABLE 1. Patient Laboratory Values

There were no significant changes in any of the laboratory measurements following fenofibrate treatment (Table 1). Of note was the lack of increase in liver enzymes, compared with a significant increase in alkaline phosphatase and GGT in the PLA group. There were no significant changes after treatment in hormones commonly associated with insulin resistance including cortisol, adiponectin, resistin, or TNF-α. Fasted plasma glucose in the PLA trended to increase, whereas the FEN trended to decrease, but there was not adequate power to show a significant difference. 4 PLA and 3 FEN children required at least one insulin administration for hyperglycemia above 180 mg/dL, and the average number of days receiving insulin was 3 ± 1 in PLA and 2 ± 1 in FEN.

There was no difference in the basal plasma glucose concentrations between groups prior to initiation of drug treatment. PLA experienced a significant increase (146 ± 9 vs. 161 ± 9 mg/dL; pretreatment vs. posttreatment; P = 0.004) in the mean daily plasma glucose concentration after treatment, whereas there was no change in FEN (Fig. 2A) (158 ± 7 vs. 45 ± 4 mg/dL; pretreatment vs. posttreatment). The change between the groups was also significant (P = 0.004).

FIGURE 2. A, Mean daily plasma glucose concentrations in mg/dL before and after treatment. The change between FEN and PLA was significant (*P = 0.004). B, Glucose infusion rate in mg/kg per minute during a hyperinsulinemic-euglycemic clamp is shown before and after treatment. The infusion rate, and thus insulin sensitivity, was significantly (*P = 0.003) increased after fenofibrate treatment. C, Suppression of hepatic glucose output during hyperinsulinemia as compared with basal. The percent suppression of endogenous glucose output was almost greater after fenofibrate treatment (P = 0.06). D, Total glucose uptake (endogenous and exogenous) during the hyperinsulinemia in mg/kg per minute. Uptake was significantly greater after fenofibrate (*P = 0.002).

There was no change in the glucose infusion rate required to maintain euglycemia during hyper-insulinemia in the PLA after treatment (4.28 ± 0.57 vs. 4.47 ± 0.70 mg · kg⁻¹ · min⁻¹, pretreatment vs. posttreatment). However, there was a significant increase in FEN after treatment 5.23 ± 0.52 vs. 7.56 ± 0.66 mg · kg⁻¹ · min⁻¹; pretreatment vs. posttreatment; P = 0.003). Glucose uptake increased in every individual in FEN.

The suppression of hepatic glucose production during the clamp in PLA was 47% ± 8% and 42% ± 8% of basal, before and after treatment (Fig. 2C). After treatment suppression trended to be greater in FEN (47% ± 7% vs. 61% ± 7%, pretreatment vs. posttreatment; P = 0.06)

To ascertain if the greater rate of glucose infusion during the clamp in the FEN group was merely a manifestation of enhanced suppression of endogenous glucose production, the glucose release by the liver was added to the amount of glucose infused to calculate the total amount of glucose uptake during the clamp. Whole body glucose uptake was still significantly greater in FEN after treatment (P = 0.002, Fig. 2D).

There was a significant decrease over time in citrate synthase activity in the PLA group, whereas the levels were maintained in FEN (P = 0.01) (Fig. 3A). The activity level of cytochrome C oxidase had a similar pattern to that of citrate synthase and the difference between PLA and FEN was significant (P = 0.02) (Fig. 3B).

FIGURE 3. A, Change from pretreatment to posttreatment in the ability of muscle citrate synthase to metabolize acetyl-CoA (μmol acetylCoA/μg protein per minute) (*P = 0.01). B, Change in the capacity of cytochrome C oxidase to metabolize cytochrome C (μmol cytochrome C/μg protein per minute) (*P = 0.01). C, State 3 coupled muscle mitochondria maximal pyruvate oxidative capacity (μmol O₂ /mg protein per minute) There was a significant increases in ATP production after fenofibrate treatment (*P = 0.001), whereas there was a decrease in the PLA group (*P = 0.006).

The maximal ability of mitochondria to oxidize pyruvate through coupled oxidation was similar prior to treatment between the groups. There was a significant increase in the FEN group after treatment (P = 0.001), whereas the PLA group decreased significantly (P = 0.006) (Fig. 3C).

There was no measurable IR tyrosine phosphorylation after hyper-insulinemia in PLA at either study (0.06 ± 0.21 vs. 0.19 ± 0.30 absolute units, pretreatment vs. posttreatment; P = 0.37) In FEN, IR tyrosine phosphorylation increased significantly after treatment (0.62 ± 0.17 vs. 1.98 ± 0.95, pretreatment vs. posttreatment; P = 0.04) The tyrosine phosphorylation of IRS-1 was similar (−0.06 ± 0.32 vs. 0.56 ± 0.65, pretreatment vs. posttreatment; P = 0.25; PLA and −0.79 ± 0.58 vs. 2.25 ± 1.09 absolute units, pretreatment vs. posttreatment; P = 0.04; FEN) (Fig. 4).

FIGURE 4. A, Change in phosphorylation of muscle insulin receptor from basal to clamp on the study days before and after either placebo of fenofibrate treatment, expressed as arbitrary band absorption measurements. Insulin stimulated phosphorylation was significantly increased following fenofibrate (*P = 0.04). B, Change in phosphorylation of muscle insulin receptor substrate: 1 from basal to clamp on the study days before and after either placebo off fenofibrate treatment expresses as arbitrary band absorption measurements. Insulin-stimulated phosphorylation was significantly increased following fenofibrate (*P = 0.04).

DISCUSSION

This is the first study to show that 2 weeks of fenofibrate treatment can ameliorate insulin resistance in pediatric burn patients, a model of uncomplicated acute insulin resistance. The improvement in insulin sensitivity occurred in both the liver and muscle, and was associated with improved insulin signaling in the muscle. Further, the maximal mitochondrial oxidative capacity of pyruvate was increased by fenofibrate treatment. There were no observed adverse affects from the medication, and it was much easier and safer to administer than current suggested protocols of hyperinsulinemia to maintain euglycemia.⁴

The improvement in insulin sensitivity observed in the pediatric burn patients is similar to those seen after fenofibrate treatment in obese patients with metabolic syndrome, yet it was likely not restored to preinjury levels.^18,27 The rate of insulin stimulated glucose uptake increased by 50% following fenofibrate treatment, however it was still below 10 mg · kg⁻¹ · min⁻¹, whereas healthy and even obese children studied with similar clamps had insulin stimulated glucose uptake rates of greater than 10 mg · kg⁻¹ · min⁻¹.^28,29 It is not surprising that the insulin sensitivity in the placebo group did not change, since insulin resistance has been demonstrated in burn patients that are still within the ICU setting 2 months following injury.³⁰ Further, it is not likely that the prevalence of insulin resistance is related to the larger Hispanic proportion of children, since these children were all lean, and insulin resistance is very common following burn injury, regardless of heritage.² While insulin sensitivity has been shown to vary with age, every child in the fenofibrate group responded, indicating that insulin age does not affect overall response.³¹ Studies are currently being conducted to see if tight glucose control benefits the pediatric burn population; and based on preliminary results from these studies, it is likely that fenofibrate would still have an effect on insulin sensitivity in the face of tight glucose control.

Hyperinsulinemia should prevent hepatic glucose release in healthy and insulin-resistant children.³² A decreased response to the suppressive effect of insulin on liver gluconeogenesis has been noted in trauma and burns patients previously.³³ Whereas the suppressive effect of insulin on hepatic glucose release was greater after fenofibrate, it remained less than would have been expected in normal children.

PPAR-α agonists have been shown to increase mitochondrial function and glucose oxidation in multiple animal models,¹⁷ although they have not been used in burn animal models. However, it is likely that this is the mechanism of action of fenofibrate in these patients (Fig. 5). Citrate synthase activity increased in skeletal muscle of overfed rats that were also given the PPAR-α agonists Wy-14643 as compared with control and overfed only rats.¹⁶ Further, PPAR-α agonists have been shown to up-regulate the expression of many genes for proteins involved in oxidation, including those involved in the tricarboxylic acid (TCA) pathway and respiratory chain.³⁴ Since studies in burned animals have found an association between injury and protein transcription of TCA cycle proteins, including cytochrome C-oxidase,⁸ it is likely that this was also a mechanism of the fenofibrate-treated children, perhaps contributing to the increased maximal pyruvate oxidative capacity. It could be that the fenofibrate increased the transcription of new proteins and thus maintained the levels of activity of these enzymes, whereas the placebo group experienced further decrement of the same proteins.

FIGURE 5. Glucose enters the mitochondria as pyruvate, and typically becomes acetyl CoA before entering the TCA cycle, where carbons are released for entry into the electron transport chain. Here either ATP or H free radicals are made. In burns, it may be that more free radicals are made relative to ATP primarily to generate more heat through uncoupling protein 1 (UCP-1), but that also creates reactive oxygen species (ROS), that in turn cause TNF-α production, and activation of protein kinase C (PKC), which inhibits insulin signaling. Further, if the TCA cycle is decreased, acetyl CoA can leave the mitochondrial and become malonyl CoA, which also activates PKC. Fenofibrate may work by increasing the ratio of glucose turned into ATP, thereby decreasing PKC activity though decreasing TNF-α and malonyl CoA.

The changes we observed in the insulin signaling pathway were similar to those found in the few burn animal studies that have been conducted.^35,36 For example, insulin injections did not stimulate IR tyrosine phosphorylation or IRS-1 tyrosine phosphorylation in rats 3 days postburn compared with controls rats.³⁵ The lack of response of the insulin signaling pathway in week 1 in our patients was very similar to that reported in type II patients with diabetes.³⁷ Further, recent studies have found that patients with diabetes have fewer smaller mitochondria, indicating that a parallel may exist between acute and chronic insulin resistance.¹²

The fact that the improved in vivo insulin sensitivity occurred concomitantly with increased IR tyrosine phosphorylation and mitochondrial pyruvate oxidative capacity adds further evidence to support the concept that these factors are linked in the origin of insulin resistance.³⁸ However, it is not clear if the maximal mitochondrial oxidative capacity of pyruvate reflects the central mechanism, since under physiologic conditions in burn patients the rate of pyruvate oxidation is not maximized, and there is little evidence of a deficiency.³⁹ Rather, it may be that the disruption of normal oxidative phosphorylation that links mitochondrial function and insulin sensitivity may stem from the release of reactive oxygen species (ROS) due to the greater rate of uncoupled oxidation.⁴⁰ ROS can in turn indirectly inhibit insulin action by activating TNF-α, and then protein kinase C, which can prevent tyrosine phosphorylation of the insulin receptor and IRS-1.^41,42 Fibrates have been shown to decrease plasma levels of TNF-α, although on the days of our study, the plasma TNF-α concentrations were often undetectable. It may be that plasma concentrations did not accurately reflect tissue concentrations of TNF-α or that fenofibrate possibly affected TNF-α release during stressful events such as surgery, staple removal, or bathing with debridement. It may be that the production of ROS links the oxidative efficiency to the insulin signaling pathway in burns. Regardless of the mechanism, this is the first demonstration of the pharmacologic stimulation of mitochondrial function in human burn patients.

CONCLUSION

Whole body insulin sensitivity, muscle insulin signaling, mitochondrial enzyme activity, and mitochondrial pyruvate oxidation are improved after approximately 2 weeks of fenofibrate treatment in severely burned children, compared with placebo. Decreased mitochondrial oxidative capacity may play a role in burn trauma-induced insulin resistance. This is the first study to show that mitochondrial function, insulin signaling, and insulin sensitivity can be improved in pediatric burn patients through pharmacologic means. Further studies are needed to determine if similar results are found in adults with burn trauma, and the optimal treatment regimen.

ACKNOWLEDGMENTS

The authors thank the families and volunteers; Dr. Jong O. Lee for his clinical care of the patients; Dr. Gerald Lynis Dohm for his assistance with the insulin signaling data; and Paulette Rousette as well as the Shriners Hospital of Galveston research, clinical, respiratory and pharmacy staff; and the UTMB Metabolism Unit staff.