Short-term metformin and exercise training effects on strength, aerobic capacity, glycemic control, and mitochondrial function in children with burn injury

Abstract

Severely burned children experience a chronic state of sympathetic nervous system activation that is associated with hypermetabolic/cardiac stress and muscle wasting. Metformin, a diabetes medication, helps control hyperglycemia in obese diabetic populations, and exercise has been shown to improve exercise strength and aerobic exercise capacity after severe burns. However, whether exercise improves glycemic control in burned children and whether combining exercise and metformin improves outcomes to a greater degree than exercise alone are unknown. We tested the hypothesis that a 6-wk exercise program combined with short-term metformin administration (E + M) improves aerobic and strength exercise capacity to a greater degree than exercise and placebo (E), while improving glucose tolerance and muscle metabolic function. We found that, before exercise training, the metformin group compared with the placebo group had attenuated mitochondrial respiration (pmol·s⁻¹·mg⁻¹) for each state: state 2 (−22.5 ± 3), state 3 (−42.4 ± 13), and oxphos (−58.9 ± 19) (P ≤ 0.02, M vs. E + M group for each state). However, in the E + M group, exercise increased mitochondrial respiration in each state (P ≤ 0.05), with respiration being comparable to that in the E group (each P > 0.05). In both groups, exercise induced comparable improvements in strength (change from preexercise, Δ1.6 ± 0.6 N-M·kg^LBM) and V̇o_2peak (Δ9 ± 7 mlO₂·kg^LBM) as well as fasting glucose (Δ19.3 ± 13 mg·dl) and glucose AUC (Δ3402 ± 3674 mg·dl⁻¹·min⁻¹), as measured by a 75-g OGTT (all P ≤ 0.03). Exercise reduced resting energy expenditure in E + M (Δ539 ± 480 kcal/24 h, P < 0.01) but not E subjects (P = 0.68). Both groups exhibited reduced resting heart rate (Δ30 ± 23 beats/min, P ≤ 0.02). These data indicate that short-term metformin combined with exercise provides no further improvement beyond that of exercise alone for strength, exercise capacity, and glycemic control.

INTRODUCTION

Burn trauma causes a state of chronic stress that is associated with hypermetabolism, increased catecholamines and energy expenditure (20–100% above normal), altered lipid metabolism, and leads to muscle wasting (9, 17, 19, 31, 62). This hypermetabolic response is also thought to contribute to a loss of glucose control, resulting in a stress-induced diabetic state (36, 39, 49, 62). These alterations in substrate metabolism are sustained for up to 3 yr after burn injury (15, 23, 31, 57, 61), contributing to long-term morbidity in the severely burned. Major advancements over the last 50 yr have increased survival rates in patients with burn injuries covering 90% of the total body surface area (35). Specifically, resuscitation protocols, early wound closure, support for the hypermetabolic response, infection control, early enteral nutrition, and better glucose control have led to improved survival rates, shorter hospital stays, and decreased rates of morbidity and mortality (43).

Several pharmacological strategies have been investigated to moderate the stress response to severe burns. These include anabolic proteins such as growth hormone, insulin, insulin-like growth factor (IGF-I), and the combination of IGF-I and IGF-binding protein 3; anabolic steroids such as testosterone esters or analogues; and β-blockers such as propranolol or metoprolol (9, 23, 24, 32, 38, 42, 43). Each of these different classes of drugs have been shown to improve postburn outcomes, such as heart rate, resting energy expenditure (REE), and muscle wasting (2, 9, 23, 38). Recently, the common antidiabetic drug metformin has been shown to improve stress-induced diabetes in burned adults (30) and attenuate muscle catabolism in burned children (9, 17, 19).

Skeletal muscle mitochondrial function is related to the hypermetabolic state in severely burned children and adults (46, 47). Mitochondrial production of reactive oxygen species has been reported to affect glucose metabolism by inhibiting glycolysis (5, 12, 28, 29). Moreover, burn trauma-induced hypermetabolism is accompanied by mitochondrial stress and a diabetic state (36, 39, 40, 49, 62). Thus mitochondrial dysfunction may be associated with insulin resistance in burn populations.

Metformin, a biguanide that is commonly prescribed to treat type II diabetes, is thought to alter energy metabolism in tissues such as skeletal muscle and liver by activating AMP-activated protein kinase (AMPK), a key energy-sensing regulatory kinase, ultimately improving glucose control (6, 22). Similar to metformin, single or repeated bouts of exercise can enhance insulin sensitivity and glucose tolerance in healthy, obese, and diabetic populations (4, 27, 33, 41, 54). However, whether exercise or metformin improves glucose control in severely burned children remains unknown. Exercise restores lean body mass (LBM), exercise capacity, and quality of life in severely burned individuals (20, 55, 58, 59). Such improvements may be beneficial to glycemic control in burn populations; however, whether the combination of metformin and exercise improves these outcome measures to a greater degree than exercise alone in burned patients is unknown. Therefore, we tested the hypothesis that, in burned children, administration of metformin would amplify the benefits of 6 wk of exercise training on exercise capacity and glucose tolerance.

METHODS

Ethical approval.

All experiments were approved by the Institutional Review Board of the University of Texas Medical Branch and are in agreement with the Declaration of Helsinki. A combined 34 children participated. All parents or legal guardians provided informed consent before participation, and child assent was obtained, as applicable. The ClinicalTrials.gov identifier is NCT01666665.

Experimental design.

This study had a randomized double-blind design (see Fig. 1). Children received placebo (n = 8) or metformin (n = 9) (500 mg, twice daily) at 72 to 96 h postadmission and at discharge participated in exercise training after discharge. All patients received standard of care involving reconstruction and skin grafting and were discharged once wounds were 95% healed. At discharge and after 6 wk of exercise training, cardiorespiratory function (peak V̇o₂, modified Bruce protocol), strength and power (dynamometer), and body composition [LBM; dual-energy X-ray absorptiometry (DXA)] were determined as previously described (20). Exercise training consisted of resistance exercise 3 days/wk and aerobic exercise (10–45 min) 3–5 days/wk. See Fig. 2 for a schematic of the study timeline.

Fig. 1.

CONSORT-like flow diagram showing subject enrollment. TBSA, percent total body surface area burn.

Fig. 2.

Schematic timeline of study design.

Rehabilitative exercise program.

Within 48 h of admission to Shriners Hospitals for Children (Galveston, TX), pediatric patients underwent standard of care treatment involving total burn excision. Wounds were covered with available autograft and remaining open wounds covered with homograft. Total fluid resuscitation was administered within 24 h of admission and according to the Galveston formula (5,000 ml/m² total body surface area burned plus 2,000 ml/m² total body surface area lactated Ringer solution). All patients received the same nutritional support during the first week. Nutritional support was calculated as 1,500 kcal/m² body surface plus 1,500 kcal/m² area burned and was administered through the enteral duodenal or nasogastric route (21). During the remainder of the acute stay, intake was modified to 1.4 × three weekly measured REE. REE was obtained via indirect calorimetry (Sensor-Medics 2900, Yorba Linda, CA) under fully resting conditions between 12:00 AM and 5:00 AM while the patient was asleep. The composition of inspired and expired gases was sampled and analyzed at 60-s intervals. Values of carbon dioxide production, volume of and oxygen consumption were accepted once they reached a steady state for 5 min and averaged and calories per day obtained utilizing the abbreviated Weir equations; REE = [3.9 (V̇o₂) + 1.1 (V̇co₂)] 1.44. Patients were discharged once wounds were 95% healed. They were then enrolled in our standard of care exercise rehabilitation exercise-training program. The training program entailed supervised aerobic exercise and resistance exercise sessions. The aerobic sessions involved treadmill or cycle ergometer sessions at least 3–5 days/wk at 60–75% of peak V̇o₂. During the first week, an aerobic warmup of 10 min would start the main exercise session and a cool down would end the session. Patients complete as much as they could due to open skin grafts wounds that may have limited mobility and the ability to exercise on the treadmill or cycle ergometer. The strength-training program consisted of at least 3 days/wk of whole body resistance exercise on free weights, such as bench, leg, and shoulder presses; leg extension; biceps, leg, and triceps curls; and toe raises. During the first week of training, the patients were familiarized with equipment and proper technique with minimal weights/loads. Weights/loads were then gradually increased over time from 50 to 60% of the patient’s three-repetition maximum to a goal of 80 to 85% of the three-repetition maximum by week 6. An American College of Sports Medicine-certified exercise specialist supervised all sessions. Both groups exercised for a comparable number of weeks [exercise only (E): 5.7 ± 0.9 wk vs. exercise plus metformin (E + M): 5.7 ± 2.3 wk; P = 0.94].

Peak aerobic exercise capacity test.

Aerobic exercise capacity (peak V̇o₂) was determined by a modified Bruce protocol maximal treadmill exercise test to volitional exhaustion. Respiratory gasses were analyzed using breath-by-breath data using an automated MedGraphics Cardi O₂ metabolic cart (St. Paul, MN) after O₂ and CO₂ gas and air flow were calibrated using known gasses and a 3-liter syringe. Subjects were constantly encouraged to complete 3-min stages, and the test was terminated once peak volitional effort was achieved. Similar criteria that have been used by others in children were used here (42, 47). Verification of maximal exertion was determined once subjects signaled to stop exercise and at least three of the following criteria were met: a respiratory exchange ratio ≥1.05, a leveling off in V̇o₂ with increasing workloads (<2 ml·kg⁻¹·min⁻¹), volitional fatigue, final exercise heart rate ≥190 beats/min, or a final test time between 8 and 12 min. Values were normalized to lean body mass (LBM) to account for differences from pre to postexercise training and from healthy nonburned controls.

Peak strength and mean power test.

Muscle strength was assessed using the Biodex System-3 Dynamometer (Shirley, NY). The isokinetic test was performed at an angular velocity of 150°/s on the dominant leg extensors. Subjects were familiarized with procedures through visual and verbal explanations before testing. Following three submaximal repetitions without load, 10 maximal voluntary muscle contractions (full extension and flexion) were performed consecutively without rest between repetitions. Values of peak torque and average power were calculated by the Biodex software system and normalized to LBM to account for differences from pre- to postexercise training and from healthy nonburned controls.

Oral glucose tolerance test.

All subjects underwent a fast that started at midnight before each oral glucose tolerance test. The 2-h oral glucose tolerance test entailed ingestion of a glucose load determined by body weight: 1.75 ml/kg body wt of glucose up to a maximum of 75 g dissolved in a maximal volume of 296 ml. Venous blood samples were collected in BD Vacutainer serum collection tubes, allowed to clot, and centrifuged at 3,600 rpm for 10 min. An aliquot was tested for glucose using the Siemens Dimension EXL 200 (Malvern, PA) analyzer with the hexokinase glucose-6-phosphate dehydrogenase method. A second aliquot was frozen and sent to LabCorp for measurement of insulin using a chemiluminescence immunoassay. Incremental area under the curve was analyzed by a trapezoidal model. Glucose tolerance testing after exercise rehabilitation training was performed within 1–2 wk after the last bout of exercise (E + M: 7 ± 13 days vs. E: 9 ± 19 days; P = 0.84).

Mitochondrial respiration.

For both groups, skeletal muscle was obtained before exercise training and at 6 wk after exercise for the measurement of mitochondrial respiratory capacity and function. Measurement of mitochondrial respiration has been previously described (48). Skeletal muscle samples were obtained via percutaneous biopsy, placed in ice-cold BIOPS buffer, and immediately transferred to a laboratory preparation of permeabilized myofibers. Myofiber bundles were separated and cleaned of all visible connective tissue before analysis. State 2 mitochondrial respiration supported primarily by complex I of the electron transport chain was determined following titration of substrates (5 mM pyruvate, 1.5 mM octanoyl-l-carnitine, 2 mM malate, and 10 mM glutamate) (state 2_I). The addition of 5 mM ADP was used to record coupled state 3_I respiration. The addition of the complex II substrate succinate (10 mM) was used to determine maximal respiration in the coupled state supported by electron transfer through both complex I and II (state 3_I + 3_II). Mitochondrial quality was determining by calculating the respiratory control for ADP (state 3/state 2) and substrate control ratio for succinate (oxphos/state 3). Muscle samples were obtained at 1.7 ± 0.5 days after the last bout of exercise.

Body morphology.

Standard, calibrated scales were used to determine weight and height. Body composition was determined using DXA (Hologic QDR 4500 densitometer; Hologic, Bedford, MA). DXA measurements were obtained within 7 days of beginning the exercise protocol during the subject’s first visit to the laboratory and after the 6-wk exercise rehabilitation. On the day of each test, the DXA instrument was calibrated using the procedures provided by the manufacturer, and DXA scans were performed and analyzed using pediatric software. The three body mass indexes used were body mass index (BMI), BMI percentile, and body surface area. BMI (kg/m²) was calculated by weight in kilograms divided by the square of height in meters, BMI percentile was computed according to the normative values provided by the Centers for Disease Control and Prevention (7), and body surface area (m²) was calculated according to DuBois and DuBois (11). Total body surface area burned and with third-degree burns was calculated by nursing staff.

Statistical analysis.

All data were blinded into group A and B, and unblinded after the statistical results were obtained. Baseline characteristics were compared between burn groups using unpaired t-test. The nonburned healthy control group was compared with the combined burn groups using unpaired t-test. A two-way factorial ANOVA assessed the interaction and main effects for exercise training (pre to post) and group [exercise only (E) and exercise plus metformin (E + M)]. The changes (Δ) from pre- to postexercise for E and E + M were compared using unpaired Student’s t-test. The nonburned control group was used for reference values where appropriate and did not complete the exercise training. An unpaired t-test was used to compare burned and nonburned control values. For the mitochondrial analysis for each state, a three-way factorial ANOVA was used to examine interactions and main effects for respiratory state × group × training with appropriate post hoc analysis (Holm-Sidak multiple comparisons test). A three-way factorial ANOVA (group × exercise × time) was used to analyze the 2-h oral glucose tolerance test results. For each analysis, where appropriate significant interactions were found, post hoc Bonferroni comparisons were made with appropriate Greenhouse-Geisser corrections for violations of sphericity. Data and figures were analyzed with GraphPad Prism (Version 6.0, La Jolla, CA) with significance set at P < 0.05. Data are reported as means ± SD or ± 95% confidence intervals where indicated.

RESULTS

Subjects.

Subject characteristics are presented in Table 1. Burned children (n = 9) in the exercise-only group (E + placebo) were matched to an additional group of burned children (n = 8) that received exercise plus metformin (E + M). The groups were matched for age (mean: 14.4–14.9 yr; P = 0.62), total body surface area burned (40%; P = 0.99) and with third-degree burns (16–21%; P = 0.46), sex, characteristics of hypermetabolism (resting heart rate, 128–131 beats/min; P = 0.96), and resting metabolic energy expenditure (1,872 kcal/day; P = 0.60) before the exercise rehabilitation program. Nonburned children (n = 17) were likewise matched to each burned child for age and sex. These children did not complete the 6 wk of exercise training and served as a reference only. In the metformin group, 75% of subjects were admitted for flame injury and 25% for electric/flame. In the placebo group, 55% were admitted for flame injury, 33% for electric/flame, and 11% for scald. Height (pre: 157 ± 7; post: 158 ± 7 cm), weight (pre: 51 ± 11; post: 51 ± 11 kg), and BMI (pre: 20 ± 4; post. 20 ± 4 kg/m²) did not change from pre- to postexercise training (each, P > 0.05). However, LBM was increased by ~9% (pre: 35 ± 6; post: 38 ± 6 kg; P = 0.001) in both groups (E: 3.4 ± 3.3 kg vs. E + M: 2.8 ± 2.3 kg; P = 0.67). After exercise training, REE was reduced in the E + M group (Δ539 ± 480 kcal/24 h; P < 0.01) but not the E group (Δ289 ± 687 kcal/24 h; P = 0.68).

Table 1.

Subjects’ physical characteristics

	Burned			Nonburned
	E	E + M	P value	Control	P value
n (male, female)	9 (4/5)	8 (4/4)		17 (8/9)
Age, yr	14.4 [12, 17]	14.9 [13, 17]	0.62	14.8 [13, 16]	0.85
Start of exercise postburn (weeks)	6.6 [4, 9]	5.4 [4, 7]	0.33
Exercise training duration (weeks)	5.7 [5, 6]	6.3 [4, 9]	0.55
Body morphology
Height, cm	155.6 [144, 167]	157.8 [154, 162]	0.72	162.7 [156, 170]	0.15
Weight, kg	47.0 [39, 55]	55.9 [46, 66]	0.14	63.2 [52, 75]	0.06
BSA, m−2	1.4 [1.3, 1.6]	1.6 [1.4, 1.7]	0.18	1.7* [1.5, 1.8]	0.04*
BMI, kg/m2	19.3 [17, 22]	22.4 [18, 26]	0.14	23.7 [21, 27]	0.08
BMI, %tile	41.2 [17, 65]	62.5 [36, 89]	0.18	71.3 [57, 86]	0.06
Lean mass, kg	33.9 [28, 40]	37.5 [34, 41]	0.29	46.4* [38, 55]	0.02*
TBSA burn, %	40 [32, 49]	40 [29, 52]	0.99
TBSA 3rd-degree burn, %	21 [7, 35]	16 [8, 24]	0.46	3
Values are means ± 95% confidence interval
Heart rate, beats/min	131 [123, 138]	128 [115, 140]	0.96	86* [78, 93]	<0.0001*
MEE kcal/day	1,872 [1,506, 2,268]	1,887 [1,343, 2,401]	0.60	208* [37, 379]	<0.0001*

Values are means ± 95% confidence interval. E, exercise; M, metformin; BSA, body surface area; BMI, body mass index; BMI, %tile, body mass index percentile for age; TBSA, percent total body surface area burn; MEE, metabolic energy expenditure.

^*P < 0.05, significantly different from burn group.

Exercise training improves strength and cardiorespiratory fitness in children with burn injury.

Peak strength and cardiorespiratory capacity (peak V̇o₂) test characteristics are reported in Fig. 3. Six weeks of exercise rehabilitation improved strength and cardiorespiratory measures similarly in both groups (change from baseline for each, P > 0.30). Metformin provided no additional benefit for absolute and relative measures (normalized to LBM) strength or cardiorespiratory fitness, as both groups showed a 40–35% increase in peak torque (relative change from preexercise training), 43–39% increase in average power, and 30% increase in cardiorespiratory fitness (main effect for exercise training, P < 0.0001). Relative to the nonburned age- and sex-matched children, both burned groups had significantly decreased strength (normalized to LBM, peak torque: 65% of nonburned control and mean power: 65%) and cardiorespiratory capacities (peak V̇o₂: 59%) before exercise rehabilitation. Exercise training returned exercise strength (peak torque: +17% and mean power: +22%, P < 0.0001) and cardiorespiratory capacity (peak V̇o₂: +14%, P < 0.0001) closer to control levels.

Fig. 3.

Absolute and relative peak torque (A), maximal power (B), and V̇o_2peak (C) before (pre) and after (post) 6 wk of exercise training without (E) and with (E + M) metformin. Relative values are normalized to kilograms of lean body mass (LBM). Values from nonburned healthy children are shown as a reference (control). The change from pre- to posttraining is shown in the vertical histogram at right. Data are expressed as means ± 95% confidence interval (CI). **P < 0.01, ***P < 0.001, ****P < 0.0001, significantly different compared with control nonburned children.

Exercise training counters metabolic and cardiac stress, while enhancing glycemic control in children with burn injury.

Resting values for metabolic energy expenditure and heart rate did not differ between the E and E + M groups before the start of exercise rehabilitation (Fig. 4). Six weeks of exercise reduced resting metabolic energy expenditure for the E + M group only (26% change from preexercise training, group × exercise interaction, P = 0.03). Before exercise rehabilitation, resting heart rate was 156% higher in burned subjects than nonburned controls (P < 0.0001), and exercise training reduced this value by 40% in both groups (main effect for exercise, P < 0.0001). Fasting glucose and insulin values were similar before training. Exercise training reduced both measures of resting glycemic control by 18–21% (main effect for exercise, P = 0.018). The 2-h oral glucose tolerance test showed no effect for metformin post glucose ingestion (Fig. 5). However, glucose tolerance was improved only after exercise rehabilitation (group × exercise × time interaction, P < 0.03). Similarly, the area under the curve for plasma glucose for the 2-h glucose tolerance test was reduced by 16% (main effect for exercise). Insulin values were changed after metformin administration from discharge to the end of exercise training but only at 30 min at the (P < 0.05). Insulin area under the curve did not differ from pre- to postexercise training.

Fig. 4.

Resting metabolic energy expenditure (MEE; A), heart rate (B), fasting glucose (C), and fasting insulin (D) before (pre) and after (post) 6 wk of exercise training without (E) and with (M + E) metformin. MEE and heart rate are compared with values from nonburned healthy children (control). Data are expressed as means ± 95% CI.

Fig. 5.

Plasma insulin (A) and glucose responses (B) to a 2-h 75-g oral glucose tolerance test before (pre) and after (post) 6 wk of exercise training without (E) and with (E + M) metformin. Area under the curve (AUC) is shown at right. Data are expressed as means ± 95% CI. *P < 0.05 vs. preexercise.

Exercise training restores mitochondrial respiratory function in children with burn injury.

Figure 6 shows the maximal capacity for mitochondrial respiration and calculated intrinsic mitochondrial quality. Mitochondrial capacity was lower in the E + M group than the control group at baseline for each state (state 2_I: 37%; state 3_I: 57%; oxphos: 60%; each, P < 0.02); however, exercise rehabilitation restored capacity (Fig. 6A, group × condition interaction, P = 0.004). After exercise training, respiratory capacity (Fig. 6B, state 3_I) did not differ between the E + M or E groups (main effect for exercise, P = 0.014). The respiratory control ratio for ADP, an index of mitochondrial quality (Fig. 6C), did not differ between groups or within groups from pre to postexercise. The substrate control ratio for succinate (Fig. 6D), another index of mitochondrial quality, was improved after exercise training in both groups (main effect for exercise, P = 0.029).

Fig. 6.

Mitochondrial respiration before (pre) and after (post) 6 wk of training without (E) and with (M + E) metformin. Data are expressed as mean ± 95% CI. †Significantly different between preexercise E + M and postexercise E; §significantly different between preexercise E + M and postexercise E + M; *significantly different between pre- to postexercise.

DISCUSSION

This study tested the hypothesis that, in children with burn injury, metformin combined with exercise training would improve exercise capacity and glycemic control to a greater degree than exercise training alone. We show, for the first time, that exercise training improves glycemic control in a burn population. Additionally, we found that short-term metformin combined with exercise provides no further improvement beyond that of exercise alone in terms of strength, exercise capacity, and glycemic control.

Burn injury is associated metabolic and cardiovascular impairments that are sustained for up to 3 yr after injury (15, 23, 31, 57, 61). We have found that, at discharge, children with severe burn injury have reduced exercise capacity. Moreover, our findings are consistent with those of our previous studies, showing that exercise rehabilitation immediately after discharge is an important component of standard of care in hospital settings owing to its ability to restore LBM and exercise capacity (20, 45, 63). Here, severely burned children had strength and cardiorespiratory fitness that were 65 and 60% of age- and sex-matched nonburn values, respectively. According to studies conducted by the Cooper Institute for Aerobics Research in children aged 13–19 yr, most of the burned children in this study fall under the very poor-to-poor category for aerobic fitness (25). We have recently reported that this may be attributable to burn-induced impairments in cardiac function during exercise (53) as well as the significant muscle wasting and orthopedic complications that severe burns incur.

A novel finding in our study was that 6 wk of exercise rehabilitation improved fasting glucose and insulin values as well as glycemic control, as seen by oral glucose tolerance testing. In adults, it is well established that acute or repeated bouts of exercise training can enhance the insulin sensitivity and glucose tolerance in healthy, obese, and diabetic populations (24–28). Mechanistically, exercise (muscle contraction) improves whole body insulin sensitivity and glucose uptake through insulin-independent mechanisms [via the cellular energy sensor AMPK (34), calcium flux, and glycogen depletion (52)]. Notably, we found that metformin given before exercise training, from admission to discharge (5.4 wk), reduced mean insulin values, which were 21% lower than exercise values; however, these changes were insignificant. Glucose tolerance testing, although applicable to real world settings, has limitations because it provides no insight into the source of improvement, which can be associated with either hepatic suppression or skeletal muscle glucose uptake. Gore et al. (18) showed that 8 days of metformin treatment diminished endogenous glucose production and oxidation in adults with severe burn injury. In this study, we showed that exercise training is associated with an improvement in glycemic control in burn populations. Our results are consistent with findings in children that glucose levels are reduced after a single bout (56) or repeated bouts of exercise (14). According to a meta-analysis performed by Fedewa et al. (14) in children, exercise training has a moderate effect in reducing insulin resistance, as calculated with fasting insulin and glucose levels using the homeostasis model assessment. We found a reduction in insulin at 30 min only in the metformin preexercise training condition; however, this response at 30 min changed to similar response after the exercise training and compared with the placebo. Typically, the first 10–15 min of the insulin response to a glucose load represent the first phase of insulin release and reflect β-cell function (57). Metformin may have improved this response during the first phase, although this is speculative and requires further understanding in burn populations. We also found that the combination of metformin with exercise training provided no synergistic effect. Belotto et al. (3) found that exercise and metformin may interact and influence the response of the other, reporting that the glycemic response was lower with the combination of these than with either alone.

We have previously shown that skeletal muscle mitochondrial respiratory capacity and coupling control are reduced in burn patients (46, 47) Specifically, while respiration uncoupled from ATP production appears well maintained in muscle following burn trauma, respiration coupled to ATP production is diminished. In line with this, we have reported that mitochondrial coupling response to ADP and the ATP synthase inhibitor oligomycin (46, 47), much like other extreme human models of hypermetabolism and altered muscle bioenergetics (37). Indeed, in acutely burned patients, we have reported respiratory coupling ratio (RCR) for ADP in the range of ~1.5–2, which is in line with our current findings. While these RCR for ADP appear low, we should note that in our system we are interested in the capacity for uncoupled respiration and altered coupling control after burn trauma. Thus we utilize high fatty acid concentrations (1.5 mM octanoyl-carnitine) to support respiration, which likely drive proton leak, thereby influencing respiratory control in response to ADP. With that said, while RCRs for ADP can range from ~4 to 10 in isolated mitochondrial preparations depending on the substrates used to fuel respiration (50), permeabilized fiber preparations typically produce lower RCRs for ADP, particularly when lipids are used to support respiration (10, 35). Interestingly, while we see low RCRs for ADP in burned patients, respiratory control in response to both ADP and oligomycin is somewhat restored in response to exercise training.

Metformin is thought to act on AMPK, and this is secondary to its effect on mitochondria through inhibition of complex I activity and cell respiration (6). With regard to improving insulin sensitivity, metformin reduces the ATP-to-AMP and ATP-to-ADP ratios, which are associated with decreased cellular respiration and cause insulin sensitization (6, 22). In this context, we found that, before exercise training, mitochondrial respiration at each state was decreased in the E + M group. While this observation is difficult to explain, it may have been driven by the high variable in mitochondrial respiratory function seen in burned patients (particularly in the acute period post injury), the presence/absence of inhalation injury or sepsis when baseline biopsies were collected, and the relatively small sample size. Notably, after exercise training, mitochondrial respiratory capacity and coupling control in both groups were not different, suggesting the metformin treatment per se was not affecting mitochondrial respiratory function in our patients.

Burn injury is associated with a profound increase in energy expenditure and catecholamines (49). Although we did not measure catecholamines, a strong relationship exists between burn injury and elevated resting heart rate (62). Thus the stress response to burn injury is associated with a chronic state of sympathetic hyperactivity. Our exercise rehabilitation program was associated with a 40% reduction in resting heart rate (an indirect marker of adrenergic stress) in both groups. In physically fit individuals, lower resting heart rates are thought to be due to higher parasympathetic activity (vagal tone) or lower sympathetic activity (8). Thus exercise training may have suppressed the hyperadrenergic response to the burn injury and increased vagal activity. Nevertheless, this is speculative and requires further understanding. We have previously reported that exercise rehabilitation has negligible effects on resting metabolic energy expenditure (1); however, the combination of metformin and exercise reduced these values by 24%. Notably, we found that burned children have lower peak heart rate values (metformin: 172 beats/min, placebo: 174 beats/min) than nonburned children (193 beats/min). Interestingly, only the combination of metformin and exercise restored heart rate to nonburn values (metformin: Δ20 beats/min, P < 0.01; placebo Δ5 beats/min, not significant). Burn injury has been reported to impair resting cardiac function in children (51). We recently found that children with severe burn injury exhibit a suppressed cardiovascular response to submaximal exercise that may be attributed to adrenergic responsiveness (53). Metformin with exercise may improve cardiac function; however, this requires further understanding. Notably, others have shown that long-term treatment with metformin substantially reduces cardiovascular events in high-risk populations (26). Work from our institute has shown that children with severe burn injury have long-term (up to 3 yr) hypermetabolic and inflammatory alterations as well as elevated insulin and glucose values (31) and that they may be at risk for long-term cardiovascular disease (13). Thus exercise training may attenuate these exaggerated responses to burn trauma. Exercise also exerts an anti-inflammatory response that can reduce proinflammatory cytokines in obese and diabetic humans and improve metabolic profiles (3, 16, 44, 60). However, whether this also occurs in burned children is unknown.

Placebo-controlled clinical trials in severely injured children come with several caveats that must be acknowledged. First, monitoring adherence to study interventions in any clinical trial can be challenging. While we have the advantage of studying individuals who are outpatient residents in our hospital, study medication compliance largely depends on patients and their guardians. Furthermore, while all exercise testing and training were supervised by a qualified exercise physiologist, testing and training are highly dependent on the effort and engagement of the patients. One must also bear in mind that patients were randomized to placebo and metformin on admission and that several factors may influence whether burn patients develop stress-induced diabetes during their acute hospital stay. Thus the therapeutic value of metformin may vary considerably from patient to patient depending the degree of glucose intolerance. While the randomized design of our study is likely the best way to address this potential limitation, this study has a relatively small group size and may be underpowered to detect any effects of metformin on glucose tolerance test outcomes. Last, while we matched both groups for age, sex, and severity of burn injury and we found a significant suppression of respiratory capacity after metformin treatment before exercise training, we cannot objectively confirm that this was the direct effect of metformin administration because we did not obtain muscle biopsies before the start of exercise rehabilitation training.

In summary, we show for the first time that exercise training improves fasting glucose metabolism and glycemic control in response to a glucose load. Additionally, we provide evidence that short-term metformin provides no additional benefit beyond that of exercise alone in terms of strength, exercise capacity, and glycemic control.

GRANTS

This work was supported by National Institutes of Health Grants 90DP-00430100, P50-GM-060338, R01-GM-056687, R01-HD-049471, and 3R01-HD-049471-12S1 and Shriners Hospital for Children Grants SHC-84080 and SHC-84090.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

D.N.H. and O.E.S. conceived and designed research. E.R., C.P., and W.M. performed experiments; E.R. and O.E.S. analyzed data; E.R., D.N.H., C.P., and O.E.S. interpreted results of experiments; E.R. prepared figures; E.R. drafted manuscript; E.R., D.N.H., C.P., W.M., and O.E.S. edited and revised manuscript; E.R., D.N.H., C.P., W.M., and O.E.S. approved final version of manuscript.