Effects of exercise training and metformin on body composition and cardiovascular indices in HIV-infected patients

Objective: To determine whether exercise training in combination with metformin improves cardiovascular risk indices and insulin in comparison to metformin alone among HIV-infected patients.

Methods and design: We conducted a prospective, randomized, 3-month study of HIV patients on stable antiretroviral therapy with hyperinsulinemia and fat redistribution. Subjects received metformin alone or metformin and exercise training consisting of 1 h of aerobic and resistance training three times a week. Cardiovascular parameters, including blood pressure and endurance during sub-maximal stress testing, body composition, strength, insulin and other biochemical parameters were determined.

Results: Thirty-seven patients were randomized and 25 subjects completed the study. Subjects receiving exercise training and metformin demonstrated significant decreases in median waist-to-hip ratio [-0.02 (-0.06, -0.01) (median (interquartile range) versus -0.01 (0.03, 0.02), P = 0.026], resting systolic [-12 (-20, -4) versus 0 (-11, 11), P = 0.012] and diastolic blood pressures [-10 (-14, -8) versus 0 (-7, 8), P = 0.001], increased thigh muscle cross-sectional area [3 (-3, 12) versus -7 (-11, 0), P = 0.015], and improved exercise time [3 (0, 4) versus 0 (-1, 1), P = 0.045] compared with subjects receiving metformin alone. Fasting insulin and insulin area under the curve decreased significantly more in the exercise and metformin group (P < 0.05). Lipids and resting lactate did not change significantly between treatment groups.

Conclusions: These data demonstrate that exercise training in combination with metformin significantly improves cardiovascular and biochemical parameters more than metformin alone in HIV-infected patients with fat redistribution and hyperinsulinemia. Combined treatment was safe, well tolerated and may be a useful strategy to decrease cardiovascular risk in this population.

Introduction

Use of highly active antiretroviral therapy (HAART) is associated with changes in fat distribution and metabolic abnormalities, including insulin resistance, dyslipidemia, and increased diastolic blood pressure [^1-9]. Among non-HIV-infected individuals, hyperinsulinemia and truncal obesity are strong independent risk factors contributing to coronary artery disease (CAD) [^10-12] and such factors may similarly increase CAD risk in HIV-infected patients. Although a retrospective study of HIV-infected patients did not show increased risk of CAD with HAART, a recent observational cohort study does suggest increased CAD risk with antiretroviral treatment [^13,14]. Premature coronary artery disease, coronary and carotid plaques have also been reported among HIV-infected patients treated with protease inhibitors [^15-17]. Modification of CAD risk remains an important goal in this population.

Metformin has been shown to improve insulin sensitivity and selected CAD risk indices in non-HIV-infected patients with truncal obesity, impaired glucose tolerance, and insulin insensitivity [^18,19]. Recently, metformin was shown to improve insulin sensitivity and cardiovascular risk indices in HIV-infected subjects with insulin resistance and truncal adiposity and has been suggested as a potential therapeutic strategy in this population [^20,21].

Exercise training may be another potentially useful strategy to reduce cardiovascular risk in HIV-infected men and women with fat redistribution. Exercise has been shown to modify cardiovascular risk factors, including blood pressure, weight, waist-to-hip ratio (WHR), and lipid levels in non-HIV-infected patients [^22,23] and to increase insulin sensitivity [²⁴]. In addition, preliminary studies suggest an effect of aerobic, resistance, or combined exercise training on fat redistribution and aerobic endurance in HIV-infected patients [^25-28].

We conducted a randomized, prospective study investigating the effects of metformin in combination with exercise training in comparison to metformin alone in HIV-infected subjects with fat redistribution and hyperinsulinemia. Our data demonstrate that combined exercise training and metformin was more effective in reducing WHR, blood pressure and insulin levels in this population. Furthermore, exercise training improved aerobic endurance and muscle strength, not seen with metformin alone.

Methods

Subjects

Between October 2000 and August 2002, 84 subjects were recruited through community advertisements and primary care provider referral. Eligible subjects were seen at the Clinical Research Centers of the Massachusetts General Hospital and the Massachusetts Institute of Technology. Eligible subjects were between 18 and 60 years of age, HIV positive, on a stable anti-retroviral regimen for more than 3 months, with fasting insulin ≥ 104 pmol/l or 120 min insulin ≥ 521 pmol/l, and evidence of fat redistribution (truncal obesity with a WHR ratio > 0.90 in men and 0.85 in women and a lipodystrophy score > 2). Fasting and 120 min hyperinsulinemia were defined in advance based on the 90th percentile for healthy subjects aged 26-50 years in the Framingham Offspring Study (personal communication, James Meigs, MD). Lipodystrophy scores were calculated based on evaluation of face, neck/shoulders, arms, abdomen, and hips/legs with graded values between 0 and 2 for fat loss or accumulation.

Subjects were excluded from the study if they had any new opportunistic infection that required hospitalization within 6 weeks of enrollment. Subjects with a history of unstable angina, aortic stenosis, uncontrolled hypertension, severe neuropathy, arthritis, prior history of diabetes mellitus or fasting plasma glucose of ≥ 6.99 mmol/l (126 mg/dl), current substance abuse or other contraindication to exercise were also excluded. In addition, subjects requiring parenteral nutrition, parenteral or oral gluccocorticoid therapy (> 7.5 mg daily), estrogen, progesterone derivatives, supraphysiologic testosterone, or ketoconazole within 3 months were also excluded. All subjects gave written informed consent and the study was approved by the Human Research Committee at the Massachusetts General Hospital and the Committee on Use of Human Subjects at Massachusetts Institute of Technology.

Of the 84 subjects screened, 42 were ineligible, five declined to participate, 12 withdrew and 25 completed the protocol (Fig. 1).

Fig. 1

Protocol

Baseline visit

After eligibility was determined, subjects returned for an 8-h outpatient baseline visit. Subjects arrived in the morning after a 12-h overnight fast and underwent an oral glucose tolerance test for insulin and glucose responses, lipid profile, HIV viral load, hematocrit, and resting lactate. Anthropometric measurements were determined, including WHR, height, and weight. Single slice cross-sectional computed tomography scans at the mid-thigh and abdomen and a total body dual energy X-ray absorptiometry (DEXA) were performed. A sub-maximal exercise stress test was conducted on a cycle ergometer through increased stages of power output (watts) every 5 min until subjects reached fatigue or their sub-maximal heart rate (220 - age × 0.85), as a measure of aerobic endurance.

Randomization

Eligible subjects were randomly assigned to one of two treatment groups at baseline. Randomization was performed by the Biostatistics Center of the Massachusetts General Hospital Clinical Research Center, using a permuted block algorithm. The metformin-only treatment group received metformin 500 mg twice a day, with a dose increase to 850 mg twice a day after 2 weeks, if resting lactic acid levels were within normal limits and no serious side effects were reported.

Subjects randomized to exercise training received metformin, as described above, and also began an aerobic and resistance exercise training program. Exercise training was three times per week for the 12-week intervention, a total of 36 sessions. Sessions were conducted at the Charles River Park Health Club affiliated with the Massachusetts General Hospital.

Subsequent visits

Subjects returned 2 and 6 weeks after the baseline visit for a physical examination, interval history and resting lactate level. Upon completion of the 3-month intervention, subjects returned for a study visit identical to that at baseline. Patients continued on their prescribed medical therapies and those not assigned to exercise were instructed not to change their normal pattern of physical activity.

Predetermined safety parameters were also monitored. Subjects were ineligible or discontinued from the study if serum creatinine was > 114 μmol/l, aspartate aminotransferase > 2.5 times upper limit of normal, resting lactate > twice the upper limit of normal, Hgb < 1.4 μmol/l, or positive urine pregnancy test (in women).

Training protocol

Each training session began with a 5-min warm-up on a stationary bicycle, followed by a standard flexibility routine to minimize the risk of injury [²⁹]. Subsequently, an aerobic training protocol was performed followed by a strength training routine and cool-down period.

The aerobic training program followed the general guidelines established by the American College of Sports Medicine [³⁰]. The duration of the aerobic component was 20 min during the first 2 weeks and 30 min thereafter. The intensity of the exercise was set at 60% of maximal heart rate during the first 2 weeks and 75% thereafter. Resistance training was based on the progressive resistance exercise concept originally proposed by DeLorme et al. [³¹] and was performed using constant external resistance Life Circuit equipment. Selected muscle groups were trained alternating upper and lower body exercises in the following order: (1) hip extension; (2) lateral pull down; (3) knee extension; (4) elbow flexion; (5) knee flexion; and (6) chest press. Each repetition included concentric and eccentric phases, with the total duration of each contraction approximately 6-10 s. Combining concentric and eccentric muscle actions maximizes strength gains and muscle hypertrophy resulting from a strength training program [³²]. Subjects performed three sets of 10 repetitions each for every muscle group, resting 2-3 s between repetitions, 2 min between sets, and 4 min between muscle groups. The initial intensity of exercise was set at 60% of the one-repetition maximum (1 RM). After 2 weeks, the relative intensity was increased to 70% of the 1 RM and after an additional 2 weeks to 80% of the 1 RM. The 1 RM was measured every other week and the absolute load was adjusted accordingly to maintain the relative intensity at 80% of the 1 RM. Each training session was supervised by a physical therapist from the Physical Therapy Services of the Massachusetts General Hospital. Those subjects not randomized to resistance training had a determination of 1 RM made at baseline, 6- and 12-week visits.

Body composition

Whole body DEXA was performed to assess fat mass, with a precision error of 3% [³³]. Cross-sectional abdominal computed tomography scanning at L4 pedicle was performed to assess subcutaneous and visceral abdominal fat areas as previously described by Borkan et al. [³⁴]. In addition a cross-sectional computed tomography scan was performed at the mid-thigh to assess total muscle cross-sectional area.

Laboratory methods

Insulin levels were measured in serum samples using radioimmunoassay (Diagnostic Product Corp., Los Angeles, California, USA). Intra-assay and inter-assay coefficients of variation ranged from 3.1 to 9.3% and 4.9 to 10%, respectively. Glucose was measured with a hexokinase reagent kit (Dade Dimension, Wilmington, Delaware, USA). Plasma lactate levels were analyzed by spectrophotometry (SmithKline Laboratories, King of Prussia, Pennsylvania, USA).

Total cholesterol was measured by enzymatic hydrolysis (Dade Dimension); serum triglycerides were measured using a lipase enzymatic method (Dade Dimension); high-density lipoprotein cholesterol (HDL-C) was measured after precipitation of low-density lipoprotein cholesterol (LDL-C), and LDL-C was calculated indirectly. Patients with triglycerides greater than 4.52 mmol/l (400 mg/dl) had direct LDL measurements using the colorimetric method with the Olympus AU640 (Olympus Inc., Irving, Texas, USA).

CD4 cell counts were measured by flow cytometry (FACS scan analyzer; Becton Dickinson and Co., San Jose, California, USA). HIV viral load was determined by ultrasensitive assay (Amplicor HIV-1 Monitor Assay; Roche Molecular Systems, Branchburg, New Jersey, USA) with limits of detection of 50 to 75 000 copies/ml.

Biostatistical analysis

Baseline characteristics were compared between randomization groups using the Wilcoxon Rank-Sum Test and Fisher's Exact Test for non-continuous variables. Median change from baseline between treatment groups was compared using the Wilcoxon Rank-Sum Test. Statistical analysis was performed using JMP Statistical Database Software (Version 4; SAS Institute, Cary, North Carolina, USA). Statistical significance was defined as P < 0.05. All data is presented as median values with interquartile ranges, except where otherwise indicated. The study was powered at 80% (two-sided 5 % significance level), to detect a difference between treatment groups of 3500 units in insulin area under the curve (AUC), assuming a standard deviation of 3000 units and an evaluable group of 25 subjects.

Results

Subjects were matched at baseline for age, duration of HIV, CD4 count, HIV viral load, and medication use (Tables 1 and 2). Twenty-one percent of the metformin only group and 25% of the metformin and exercise training group had a history of hypertension. Thirty-six percent of the metformin only and 42% of the metformin and exercise training group had a history of hyperlipidemia. The differences between groups at baseline for history of hypertension and hyperlipidemia were not statistically significant (P > 0.05). There was no reported history of myocardial infarction, stroke, or vascular disease in either group.

Table 1

Table 2

In the metformin only group, 21% were current smokers with a mean ± SEM pack year history of 11 ± 7. In the metformin and exercise training group 17% were current smokers with a mean pack year history of 12 ± 5. In addition, 36% of the metformin only group and 42% of the metformin and exercise training group reported current alcohol use with a mean ± SEM of 6 ± 2 and 5 ± 3 ounces of alcohol per week, respectively. Again, these baseline characteristics were not statistically different between treatment groups.

Among subjects completing the protocol, the mean percentage compliance ± SEM with exercise sessions was 93 ± 2%, time on the stationary bicycle was 97 ± 2% of expected, and completion of weight repetitions was 97 ± 1%. Mean ± SEM metformin compliance, determined by pill count, was 99.1 ± 1% in the metformin and exercise training group and 93 ± 5% in the metformin only group. Eight subjects in the metformin and exercise training group versus four subjects in the metformin only group withdrew from the study. Two subjects, one in each group, experienced minor increases in resting lactic acid greater than twice the upper limits of normal without signs or symptoms of lactic acidosis and were withdrawn from the protocol based on pre-specified safety parameters. In addition, two subjects, one in each treatment group, demonstrated minor elevations in aspartate aminotransferase and were withdrawn from the study based on pre-specified safety parameters. One subject in the metformin only group reported gastrointestinal side effects with metformin, necessitating a temporary hold of study drug and reduction of dose to 500 mg twice daily. Six subjects in the metformin and exercise group were withdrawn from the study for various reasons; three due to family/personal issues, including one death in the family, one subject lived too far away, and two subjects were non-compliant with exercise. Two subjects, one from each treatment group, were lost to follow-up.

There was a significant reduction from baseline in median systolic [-12 (-20, -4) versus 0 (-11, 11) mmHg, P = 0.012] and diastolic blood pressure [-10 (-14, -8) versus 0 (-7, 8) mmHg, P = 0.001] in the combined metformin and exercise group in comparison with metformin only (Table 2). Statistically significant (P < 0.05) decreases in median change from baseline in WHR and metabolic indices, such as fasting insulin and insulin AUC, were seen in the metformin and exercise training group versus the metformin only group (Table 2). There were also statistically significant increases in median change from baseline in thigh muscle cross-sectional area, five out of six strength indices measured by 1 RM, and endurance (time on cycle ergometer during sub-maximal stress test) (Table 2, Fig. 2). Both subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT), tended to decrease more in the metformin and exercise training group than the metformin only group (Table 2). There were no statistically significant changes from baseline for either treatment group in weight or body mass index. Lactic acid, aspartate aminotransferase, CD4 cell count and HIV viral load did not change significantly between the groups (Table 2).

Fig. 2

Discussion

Modification of CAD risk factors has become an increasingly important aspect of HIV management, as recent studies suggest increased CAD risk among HIV-infected patients receiving combination antiretroviral therapy [^14,35,36]. We have previously shown that metformin improves hyperinsulinemia and selected cardiovascular risk indices in this population [²⁰]. Limited data are available demonstrating that exercise training improves cardiovascular risk indices, including WHR, lipid levels, and aerobic capacity in HIV-infected patients with fat redistribution [^25-28], but prior studies have not examined combined treatment strategies to improve CAD risk parameters in this population.

In this study, we investigated whether combined metformin and exercise training was safe and more efficacious than metformin alone for HIV-infected patients with fat redistribution and hyperinsulinemia. All subjects received metformin and half were randomized to also receive exercise training. Metformin was used as a minimum treatment for subjects in this study based on prior data showing its efficacy for insulin and CAD risk parameters in this population [²⁰]. A substantial percentage of subjects, 44%, demonstrated impaired glucose tolerance, which may also increase CAD risk [^37,38].

Our data demonstrate a significant effect of combined exercise training and metformin on multiple CAD risk parameters, including WHR and resting diastolic and systolic blood pressure known to predict mortality in non-HIV-infected patients [^10,12]. In addition, the significant decrease in insulin levels in response to this treatment strategy may further reduce CAD risk in this population [¹¹].

The exercise training program used in this study contained both a resistance and aerobic component. Exercise training was effective based on the demonstrated increases in muscle strength and thigh muscle cross-sectional area. In addition, the aerobic exercise component significantly improved aerobic endurance as measured by increased time on standard sub-maximal stress testing. As expected, metformin alone did not increase aerobic fitness, muscle mass or strength.

The benefits of exercise in HIV-infected patients have been investigated in a limited number of studies. Studies of resistance training alone [²⁷] or resistance training combined with aerobic exercise [^25,28] demonstrated increased strength and increased muscle mass or decreased body fat. Only one prior study selected patients for evidence of increased abdominal girth [²⁵]. In addition, improved aerobic endurance was seen in a study of 19 subjects with HIV randomized to aerobic exercise [²⁶] and a small study of six HIV-infected subjects completing an aerobic and resistance training regimen [²⁸].

Our study extends the data from these prior studies and suggests that a thrice weekly aerobic and resistance training program in combination with metformin significantly improves cardiovascular parameters and aerobic endurance in hyperinsulinemic, lipodystrophic HIV patients. Hyperinsulinemia is common and seen in 61% of HIV patients with fat redistribution [⁹] in association with increased WHR [³⁹]. Our data suggest that use of simple measures such as WHR and insulin can be used to identify patients who will benefit from combined metformin and exercise training. The results of this study are relevant for this large sub-population of HIV-infected patients receiving antiretroviral therapy.

Thrice weekly gym sessions may not be practical for all such patients and further studies are needed to determine the minimum effective regimen to improve cardiovascular parameters in this population. Furthermore, this study does not address the question as to whether exercise training alone might be beneficial in HIV-infected patients with fat redistribution and normal insulin levels, in whom the use of metformin may be less appropriate. Further investigation of this question is warranted.

Of importance, subjects in this study were selected for relative truncal obesity as defined by an increased WHR. Use of metformin in low-weight patients with primary lipoatrophy and insulin resistance would not be appropriate, because of the potential for further weight loss. In contrast, the use of a thiazolidinedione, associated with weight and fat gain, might be a more appropriate insulin-sensitizing agent in patients with predominant lipoatrophy [^40,41].

A standard dose of metformin, 850 mg twice daily was used in this study and was generally well tolerated. Using strictly defined pre-specified safety criteria, two subjects, one in each treatment group, demonstrated resting lactic acid levels greater than two times the upper limits of normal, without signs or symptoms of lactic acidosis. Variability in resting lactate levels has been shown among HIV-infected patients [^42,43], and the clinical significance of minimal elevations in lactate without symptomatology remains unknown. Nonetheless, subjects receiving metformin with even modest increases in resting lactate should be monitored carefully for lactic acidosis with consideration of discontinuation of metformin in the presence of a continued rise in lactate levels. In this study no differences in resting lactate levels were seen between treatment groups, suggesting that a rigorous exercise training program can be accomplished among HIV-infected patients receiving metformin, without risk of precipitating lactic acidosis.

In contrast to prior studies of exercise training, significant effects on lipid levels were not seen, although cholesterol, LDL and triglyceride all tended to decrease more in the exercise training group. The absence of an effect on HDL and other lipid parameters may relate to the unique pathophysiology of lipid abnormalities in HIV-infected patients [^44,45] or an ongoing effect of antiretroviral treatment that may blunt the known effects of exercise training on lipid metabolism in this population.

Treatment with metformin and exercise training improves traditional CAD risk markers, including WHR and blood pressure, more than metformin alone among HIV-infected patients with fat redistribution and hyperinsulinemia. Exercise training is well tolerated in this group of patients and improves muscle strength and size as well as aerobic fitness. Combined therapy using exercise training and metformin in selected groups of HIV-infected patients may substantially alter CAD risk.

Acknowledgements

The authors would like to thank both the Massachusetts Institute of Technology and Massachusetts General Hospital Clinical Research Center nursing and bionutrition staff and members of the Massachusetts General Hospital Physical Therapy Services for their dedicated patient care. We would also like to thank the staff at the Charles River Park Health Club for the use of their exercise facilities.

Sponsorship: Funded in part by NIH Grants DK49302, RR300088, and RR01066.