Improved Hepatic Lipid Composition Following Short-Term Exercise in Nonalcoholic Fatty Liver Disease

Abstract

Context:

Hepatic steatosis, insulin resistance, inflammation, low levels of polyunsaturated lipids, and adiponectin are implicated in the development and progression of nonalcoholic fatty liver disease (NAFLD).

Objective:

We examined the effects of short-term aerobic exercise on these metabolic risk factors.

Design and Participants:

Obese individuals (N = 17, 34.3 ± 1.0 kg/m²) with clinically confirmed NAFLD were enrolled in a short-term aerobic exercise program that consisted of 7 consecutive days of treadmill walking at ∼85% of maximal heart rate for 60 minutes per day. Preintervention and postintervention measures included hepatic triglyceride content, and a lipid saturation index and polyunsaturated lipid index (PUI) of the liver, obtained by ¹H magnetic resonance spectroscopy (N = 14). Insulin sensitivity was estimated from an oral glucose tolerance test (OGTT), and mononuclear cells were isolated to assess reactive oxygen species production during the OGTT. Circulating glucose, insulin, and high molecular weight (HMW) adiponectin were determined from plasma.

Main Outcome:

Short-term aerobic exercise training improved hepatic lipid composition in patients with NAFLD.

Results:

Exercise training resulted in an increase in liver PUI (P < .05), increased insulin sensitivity (Matsuda Index: P < .05), HMW adiponectin (P < .05), and maximal oxygen consumption (P < .05). Reactive oxygen species production during the OGTT was reduced following exercise training (P < .05). HMW adiponectin was increased after the exercise program and the increase was positively correlated with the increase in liver PUI (r = 0.52, P = .05). Body weight remained stable during the program (P > .05).

Conclusion:

Short-term exercise can target hepatic lipid composition, which may reduce the risk of NAFLD progression. The improvement in hepatic lipid composition may be driven by adiponectin.

Nonalcoholic fatty liver disease (NAFLD) represents a spectrum of liver diseases ranging from simple hepatic steatosis to nonalcoholic steatohepatitis (NASH), which may progress to hepatic fibrosis, cirrhosis, and subacute liver failure. NAFLD is strongly associated with obesity, insulin resistance, type 2 diabetes, and the metabolic syndrome. Similar to recent trends for obesity and type 2 diabetes, the prevalence rates for NAFLD have escalated, and it is now estimated that upward of 30% of the population have some degree of hepatic steatosis (1). Indeed, NAFLD is now regarded as the most common form of chronic liver disease in adults in the US (2). Although simple hepatic steatosis is relatively benign with a cirrhotic risk of <4% over a 10- to 20-year period (3), up to 25% of those with NASH develop cirrhosis, of which 30% to 40% succumb to liver-related death within 10 years (4). Thus, NAFLD and its subsequent complications create a significant health burden and are problematic for treating clinicians due to the lack of specific and successful therapies.

Progression of NAFLD is characterized by insulin resistance and elevated levels of tissue inflammation and oxidative stress. Current treatment strategies include lifestyle interventions, such as diet and exercise, and medications targeting parameters of metabolic syndrome, such as insulin resistance and hypertriglyceridemia (5). In addition, more recent therapies, such as vitamin E, and anti-TNF-α agents, including pentoxifylline, have shown promise (6–8). Increased physical activity in combination with a hypocaloric diet is known to alter metabolism in line with improved hepatic and peripheral insulin resistance in obesity and type 2 diabetes (9). Several longer term lifestyle intervention studies that have specifically targeted histology-proven NAFLD have demonstrated significant weight loss along with improved liver histology and steatosis, and reduced liver enzymes (10–12). However, 2 recent studies report reductions in hepatic lipid content in NAFLD patients after exercise training intervention programs that did not induce weight loss (13, 14). Although these data suggest that exercise per se can reverse hepatic steatosis, they do not identify the physiological or cellular adaptations that elicit this improvement. However, mechanistic studies that directly measure liver dysfunction in humans are challenging because of limitations in obtaining repeated liver biopsy samples. Imaging tools provide a noninvasive measure of the liver and this approach is amenable to repeated measures. Recent advances in magnetic resonance imaging (MRI) and spectroscopy (MRS) offer dramatically increased sensitivity for measuring liver tissue composition, and its separation into saturated, unsaturated, and polyunsaturated lipid subtypes (15). The work of Johnson et al demonstrated that in addition to quantification of hepatic steatosis, using ¹H-MRS spectra it is also possible to determine an index of lipid saturation and polyunsaturation (15). We applied this approach to examine the effect of exercise on hepatic lipid composition in our patient population.

Short-term (7 d) aerobic exercise interventions provide an ideal stimulus to investigate primary disease mechanisms in the absence of changes in body composition. Because of the relatively short duration of the intervention, there is less concern regarding increased caloric intake and the effect of diet on the study outcomes. Short-term exercise training improves peripheral tissue insulin-stimulated glucose uptake, reduces hepatic glucose production, improves whole body glucose tolerance, and elevates adiponectin levels and basal fat metabolism in overweight and obese men and women across all age groups and levels of insulin resistance (16–18). However, the effect of short-term exercise on the underlying pathophysiology in NAFLD patients is unknown. The purpose of this study was to identify the effects of exercise training on indices of hepatic lipid saturation, insulin resistance, and markers of inflammation and oxidative stress in adults with clinically diagnosed NAFLD. We hypothesized that in patients with NAFLD, short-term aerobic exercise training would significantly decrease insulin resistance, inflammation, and oxidative stress, and this would facilitate improvements in hepatic lipid composition.

Materials and Methods

Participants

This research involved the prospective study of 17 obese (body mass index = 34.4 ± 1.0 kg/m²) men and women with hepatic steatosis, which was confirmed by MRS and was based on a hepatic triglyceride content greater than 5%. Patients were recruited from the liver clinic in the Department of Gastroenterology/Hepatology at the Cleveland Clinic. Exclusion criteria included body mass index <26 or >40 kg/m², a history of alcohol use (>20 g/d for men and >10 g/d for women), or other causes of liver disease or evidence of overt type 1 and 2 diabetes currently being treated by insulin or thiazolidinediones, evidence of cardiovascular, cerebrovascular, renal, hematological, thyroid disease, or cancer. Participants were weight stable (<2 kg weight change in the past 6 mo), sedentary (<20 min of exercise, 2 times per week), and free of any contraindications to participation in an exercise-training program. Written informed consent was obtained according to the guidelines of the Institutional Review Board for Human Research at the Cleveland Clinic (Project approved 5/16/2008).

Control period

Prior diet and physical activity were carefully controlled and monitored by a dietitian. All subjects were instructed to maintain their normal dietary eating habits and completed 3-day diet records. Subjects were also given detailed instructions to maintain their habitual physical activity levels during the course of the study. On the day preceding metabolic testing, subjects were counseled to consume ∼55% of calories as carbohydrates to meet a goal of 250 g.

Measurement of aerobic exercise capacity

Maximal oxygen consumption (VO_2max) during an incremental treadmill test was used as the criterion measure of physical fitness. The VO_2max test was performed using a Modified Bruce protocol as previously described (18). Speed was set between 2 and 4 miles per hour, and the incline of the treadmill increased 2% to 3% every 2 minutes until fatigue. Inspired air volumes were measured from pressure changes detected by a bidirectional digital volume sensor (Triple V) pneumotach, and concentrations of O₂ (electrochemical detection) and CO₂ (thermal conductivity detection) were measured using a Jaeger OxyCon Pro/Delta System (Version 4.6; Hoechberg, Germany). At least 2 of the following criteria were attained to assure a maximum test: plateau in VO₂, heart rate within 10 beats per minute of age-predicted maximum, and/or a respiratory exchange ratio >1.0. Measurements of VO_2max and maximal heart rate were used to determine the appropriate exercise training intensity. Postintervention VO_2max testing was performed the morning after day 7 of exercise training.

Exercise training

All exercise sessions were supervised by an exercise physiologist and were conducted at the Cleveland Clinic Fitness Center. Exercise consisted of walking and jogging on a treadmill. Subjects trained for 7 consecutive days, 50 to 60 minutes per day, at 80% to 85% of maximum heart rate. The participants wore heart rate monitors (Polar Electro Inc, Woodbury, New York) during each training session to provide visual feedback of their individualized target heart rate. Each training session incorporated a brief standardized warm-up and cool-down and included a series of stretching exercises. A typical exercise session began with a 5- to 10-minute warm-up, 40- to 50-minute walking/jogging on a treadmill with an appropriate grade, and a 5- to 10-minute cool-down. Compliance with the training was documented through regular attendance at the training sessions, recording exercise heart rate and blood pressure, and the use of training logs.

Body composition assessment

Height without shoes was measured to the nearest 1.0 cm. Body weight was measured to the nearest 0.1 kg with the subject wearing their underclothing and a hospital gown. Waist circumference was measured midway between the lower rib margin and the iliac crest to the nearest 1.0 cm to estimate abdominal adiposity. Hip circumference was measured at the widest girth near the buttocks. In addition, all subjects underwent dual energy x-ray absorptiometry to determine percentage of total body fat using a Lunar iDXA model scanner (GE Healthcare, Madison, Wisconsin). Computerized tomography scans (Somotom Sensation 16 scanner; Siemens Medical Solutions, Malvern, Pennsylvania) were also used to quantify cross-sectional abdominal subcutaneous and visceral adiposity. Subjects were positioned in the supine position and cross-sectional 5-mm slices were obtained at the fourth lumbar vertebral body (L4) using a sampling time of 2 seconds at 120 kV and 100 mÅ. The location for the postintervention scans was standardized using distances from bony landmarks identified on the preintervention scan.

Resting metabolic rate

After an overnight fast and a mandatory supine rest period of 30 minutes, exhaled air was collected using the ventilated hood technique. Resting energy expenditure and basal substrate oxidation rates were determined using indirect calorimetry (Vmax Encore; Viasys, Yorba Linda, California) as previously described (19).

MRS hepatic lipid subspecies quantification

Three subjects could not complete the MRS studies for safety reasons (metallic tattoos, metal implants, and so on). Subjects arrived at the laboratory at 06:00 am following an overnight (8 h) fast. Briefly, a body array MRI coil was affixed to each subject's back with Velcro straps. The center of the body array coil was aligned with the subject's spine and shoulders for accurate repositioning during longitudinal studies. Each subject was positioned face down and head first in the Siemens Verio 3T MRI scanner on a memory foam mattress to further minimize respiratory motion. This configuration has been used by our group and others to limit respiratory motion in liver scans (19). After localizer scans for positioning, an 8-cm³ voxel was positioned within the right posterior lobe of the subject's liver with guidance from the high-resolution localization images. The localizer images from the initial scanning session were used to ensure that the same voxel position was being selected for the subsequent longitudinal scans using anatomical landmarks of the liver. Care was taken to avoid large blood vessels and large concentrations of hepatic ducts. Alternative voxel positioning sites were confirmed to have similar lipid profiles. Manual shimming was performed to a line width of ∼40 Hz to generate the spectra required to delineate water and the various lipid species. Multiple MR spectra (with and without water suppression) were acquired with a single-voxel PRESS acquisition with a long repetition time (5000 ms), and a short echo time (30 ms) to limit the effects of magnetic relaxation (20). The acquisition was acquired with 32 averages to obtain sufficient signal to accurately assess lipid components with low concentrations such as the diallylic lipids. The total scan time for each subject was less than 30 minutes total. The data was Fourier-transformed, filtered, baseline corrected, and phased. A representative spectrum identifying the detected lipid species is presented in Figure 1. All NAFLD patients were confirmed to have greater than 5% hepatic triglyceride content, which is the diagnostic criterion for hepatic steatosis (21). MRS scans of the liver were repeated on the morning after the final exercise session to assess the effects of exercise training on lipid content. Indices of liver lipid saturation, unsaturation, and polyunsaturation were calculated as described by Johnson et al (15).

Figure 1.

Representative water-suppressed MRS spectrum of hepatic lipid subspecies obtained before and after the 7-day intervention. Data were Fourier-transformed, filtered, baseline corrected, phased, and fitted with Gaussian lineshapes to integrate the individual lipid peaks shown. Saturated lipid species are methylene, methyl, and α-methylene; unsaturated lipid species are allylic; polyunsaturated lipid species are diallylic.

Measurement of insulin resistance

Within 30 minutes of the MRS scans, a 75 g oral glucose tolerance test (OGTT) was performed after an overnight fast, preintervention, and postintervention. Following baseline blood draws, the glucose drink was ingested and blood samples were drawn in EDTA tubes containing aprotinin at 30, 60, 90, 120, and 180 minutes after ingestion. Plasma glucose was determined using a YSI 2300 STAT Plus analyzer (Yellow Springs, Ohio), and plasma insulin and high molecular weight (HMW) adiponectin were determined via RIA and ELISA (Millipore, Billerica, Massachusetts), respectively. Insulin sensitivity during the OGTT (ISI_OGTT) was calculated using the Matsuda Index (22) and homeostasis model of assessment-insulin resistance (HOMA-IR) (23).

Measurement of reactive oxygen species (ROS) production

Mononuclear cells (MNCs) were isolated from peripheral blood by density gradient centrifugation using polymorphonuclear cell isolation medium from fasting and glucose stimulated (2-h OGTT) whole blood samples as previously described (24). Isolated MNCs were washed twice with HEPES buffered saline (HBS) and reconstituted to a concentration of 4 × 10⁵ cells/mL in HBS. Respiratory burst activity of MNCs was measured by detection of superoxide radical via chemiluminescence. Duplicate cuvettes containing 500 μL of MNC (400 cells/μL) were placed into a 2-channel lumi-aggregometer (Chrono-Log, Havertown, Pennsylvania). Fifteen microliters of 10 mM luminol followed by 1 μL of 10 mM formylmethionyl leucine phenylalanine was added to each cuvette. Chemiluminescence was recorded in millivolts. This method was developed and validated by Thusu et al (25) and has been used successfully in our previous work (26).

Statistics

Values were tested for normality using the D'Agostino and Pearson omnibus normality test on GraphPad Prism 4.0 (Graphpad Software Inc, San Diego, California). Preintervention to postintervention changes were assessed using a repeated measures ANOVA for normally distributed samples and Wilcoxon signed rank test for nonnormally distributed data (ROS, HMW Adiponectin). Bivariate correlation analyses were used to identify relationships between changes (Δ) in variables and multiple regression analyses were used to further examine significant relationships and various potential predictors for changes in dependent variables. Data transformations were applied where appropriate to Δ values to improve normality. Statistical significance was accepted when P < .05. These analyses were carried out using StatView for Windows 5.0.1 (SAS Institute, Cary, North Carolina), and all data are expressed as mean ± SEM.

Results

Baseline subject characteristics and anthropometric variables are presented in Table 1 for both the preintervention and the postintervention periods. Markers of liver function at baseline, AST and ALT, were above normal limits (47.1 ± 9.2 and 51.4 ± 9.3 U/L, respectively). The exercise program did not affect body weight or adiposity. However, exercise training did improve VO_2max, reduced resting respiratory quotient, and increased resting fat oxidation (Table 2).

Table 1.

Subject Characteristics and Anthropometrics

Variable (unit)	Pre	Post
Age, y	54 ± 2	–
Weight, kg	100.2 ± 3.3	100.4 ± 3.3
BMI, kg/m2	34.4 ± 1.0	34.5 ± 1.0
WC, cm	111.9 ± 2.1	111.3 ± 2.3
Fat mass, kg	43.9 ± 2.1	43.6 ± 2.1
Fat-free mass, kg	56.4 ± 3.2	56.9 ± 3.2
TAT, cm2	566.5 ± 32.3	562.3 ± 33.9
SAT, cm2	460.8 ± 31.6	456.0 ± 32.8
VAT, cm2	105.6 ± 12.9	106.4 ± 11.5
HTGC, %	19.4 ± 3.3	18.7 ± 2.5
SI, %	96.6 ± 0.4	96.4 ± 0.4
UI, %	3.4 ± 0.4	3.6 ± 0.5
PUI, %	0.47 ± 0.07	0.60 ± 0.06a

Abbreviations: BMI, body mass index; HTGC, hepatic triglyceride content; SAT, subcutaneous abdominal adipose tissue; SI, hepatic saturated lipid index; TAT, total abdominal adipose tissue; VAT, visceral abdominal adipose tissue; UI, hepatic unsaturated lipid index; WC, waist circumference. Data are presented as mean ± SE.

^aP < .05.

Table 2.

Metabolic Variables

Variable (unit)	Pre	Post	P Value
FPG, mg · dL−1	113.6 ± 5.1	108.7 ± 3.3	ns
FPI, μU · mL−1	25.7 ± 2.4	22.0 ± 2.2	.03
HOMA-IR, AU	7.2 ± 0.9	6.0 ± 0.7	.03
Glucose iAUC, mg · dL−1 · 2 h	9420 ± 1002	8940 ± 938	.001
Insulin iAUC, μU · mL−1 · 2 h	12325 ± 2176	11767 ± 3889	ns
Matsuda Index, AU	1.5 ± 0.2	1.8 ± 0.2	.0004
Resting RER, AU	0.85 ± 0.02	0.81 ± 0.01	.008
Resting FOX, mg · min−1	60 ± 1	77 ± 1	.02
VO2max, mL · kg−1 · min−1	24.3 ± 1.5	25.8 ± 1.8	.05
HMW adiponectin, ng · mL−1	3041 ± 499	3480 ± 635	.04

Abbreviations: FOX, fat oxidation rate; FPG, fasting plasma glucose; FPI, fasting plasma insulin; ns, nonsignificant; RER, respiratory exchange ratio. Data are presented as mean ± SE.

Seven consecutive days of aerobic exercise training increased the lipid polyunsaturation index (PUI) (P < .05) (Figure 2). However, hepatic triglyceride content, the lipid saturation index, and lipid unsaturation index were similar to baseline following the exercise program (Table 1).

Figure 2.

Short-term aerobic exercise training increases hepatic polyunsaturated lipids. Hepatic lipid subspecies were determined via proton MRS at baseline and the morning following the last exercise session. P < .05; Mean ± SEM

Fasting plasma glucose and insulin values are presented in Table 2 for both preintervention and postintervention. Both HOMA-IR (reduced) and the Matsuda Index of insulin sensitivity (increased) were improved following exercise training (Table 2). Incremental area under the curve (iAUC) values for glucose and insulin during the OGTT are also presented in Table 2. Whereas, iAUC glucose was decreased, and iAUC insulin was unchanged following exercise.

ROS was measured at basal and after glucose stimulation using the OGTT from peripheral blood mononuclear cells both before and following the intervention period. Basal ROS production was unchanged (mV/min, Pre: 0.21 ± 0.05; Post: 0.22 ± 0.05, P > .05); however, glucose-stimulated ROS production was markedly reduced following the exercise program (mV/min, Pre: 0.46 ± 0.15; Post: 0.20 ± 0.04, P < .05). The change in ROS production from basal to glucose stimulation was also significantly attenuated by the intervention (Figure 3). Furthermore, the decrease in ROS production was inversely correlated with increases in fat oxidation (r = −0.61, P < .01), suggesting that adaptations in whole body lipid metabolism are linked to ROS production.

Figure 3.

Hyperglycemia-induced ROS production is attenuated following short-term aerobic exercise training. ROS production was determined in isolated MNCs during basal and under glucose stimulated conditions during the OGTT. Measures were obtained before and after the exercise intervention. P < .05; Mean ± SEM.

The short-term aerobic exercise program increased plasma concentrations of HMW adiponectin by 26 ± 16% (P < .05) (Table 2). We then explored the relationship between the increase in HMW adiponectin and the increase in hepatic lipid indices and fat oxidation. The change in HMW adiponectin was positively correlated with the increase in PUI (Figure 4, r = 0.52, P = .05). To examine potential predictors of the change in PUI, we used a multiple regression model with 3 independent predictors (ΔHMW adiponectin, baseline Matsuda Index, postintervention Matsuda Index). This model produced R² = 0.58, F(3, 10) = 4.51, P = .03, where ΔHMW adiponectin (t = 2.41, P = .04) and baseline Matsuda Index (t = 2.36, P = .04) had significant positive regression weights, indicating the underlying insulin sensitivity may influence PUI responsiveness following short-term exercise training. Postintervention Matsuda Index did not contribute to the multiple regression model (t = −0.73, P = .48). When postintervention Matsuda Index was removed from the model, the contribution of baseline Matsuda Index and ΔHMW adiponectin predicted 53% and 47% of the variability, respectively. A separate model with 2 independent predictors of ΔPUI (ΔHMW adiponectin and ΔMatsuda Index) produced R² = 0.38, F(2, 11) = 3.41, P = .07. Only the ΔHMW adiponectin (t = 2.41, P = .04) had significant positive weight in this model.

Figure 4.

Changes in plasma HMW adiponectin are positively correlated with the change in hepatic polyunsaturated lipids following short-term aerobic exercise. R = 0.52, P = .05.

Discussion

The primary findings of this investigation are that short-term aerobic exercise favorably alters hepatic lipid composition by increasing polyunsaturated lipids and these changes in lipid polyunsaturation are positively associated with increased circulating HMW adiponectin. These data demonstrate the potential benefit of exercise training on hepatic lipid partitioning, insulin resistance, and oxidative stress, risk factors known to influence NAFLD severity and progression to NASH, or more serious forms of liver disease. Our data point to a plausible mechanism for improved liver lipid polyunsaturation that may be regulated by HMW adiponectin.

Obesity is associated with greater lipid saturation and decreased lipid polyunsaturation. The degree of polyunsaturation is further decreased in obese individuals exhibiting hepatic steatosis (15). This observation is consistent with evidence of hepatic polyunsaturated fatty acid depletion in NAFLD (27). The data presented herein are in agreement with relatively low levels of PUI in NAFLD and indicate that exercise training can increase the PUI. However, these MRS data should be interpreted with caution as reproducibility studies and biochemical comparisons to human NAFLD liver biopsies, indicating sufficient reliability of this approach, have not yet been reported. It is also unclear if measures of PUI are influenced by the total hepatic fat fraction. We acknowledge that this MRS approach has limitations in detecting and quantifying allylic and diallylic lipid signatures given their low abundance, the variability in line width of up to 40 Hz, and limited signal-to-noise ratio, especially in cases of low liver fat fraction. Despite these limitations, some studies indicate that polyunsaturated lipids create a beneficial metabolic profile compared to saturated lipids (28). Furthermore, recent studies indicate a beneficial effect of polyunsaturated lipid supplementation on markers of liver function and steatosis in NAFLD patients (29). These data suggest that targeting PUI by either exercise or diet, or a combination of the 2, may produce a significant beneficial outcome in NAFLD patients.

Adiponectin is a known regulator of lipid oxidation with anti-inflammatory properties, and circulating levels are low in patients with NAFLD (30). Recent data from the Dallas Heart Study extended these observations to NASH, where it was reported that adiponectin is an independent predictor of hepatic steatosis (31). Additional data are also emerging to show the positive benefits of adiponectin on NAFLD and fibrosis (32). Recently, adiponectin treatment of primary hepatocytes in culture showed reduced oxidative stress and synthesis of extracellular matrix proteins (33). Our data indicate that HMW adiponectin was increased after exercise training. Furthermore, we show that changes in adiponectin are correlated with changes in hepatic polyunsaturated lipid. We recognize that this relationship does not imply causality; however, the work of Sofi and colleagues (29) suggests that polyunsaturated fatty acid supplementation increases adiponectin levels in NAFLD patients. This plausible mechanism linking PUI and adiponectin may involve the recently described signaling pathway initiated at the adiponectin receptor (34). Iwabu et al have shown that the adiponectin receptor signals calcium-calmodulin kinase and AMP-activated protein kinase, which both converge on sirtuin (silent mating type information regulation 2 homolog) 1 to activate peroxisome proliferator-activated receptor γ, coactivator 1 α and subsequent gene transcription for lipid oxidation pathways and mitochondrial biogenesis (34). Additional mechanisms such as improved adipocyte insulin sensitivity (16) or reduced circulating insulin may also be responsible for the observed improvements in HMW adiponectin and future studies using human liver biopsies following exercise training are required to address this potential mechanism.

It is well established that exercise training improves resting substrate oxidation and creates a metabolic milieu that promotes lipid utilization in skeletal muscle and liver (35, 36). This improvement is regulated by increased mitochondrial function and enhanced gene/protein expression (37). We found a significant increase in basal fat oxidation in our NAFLD patients after the 7-day exercise program. These data reflect a shift in whole body substrate use to favor oxidizing fat as a fuel and may serve to facilitate a shift in lipid partitioning in the liver. However, whether this adaptation directly promotes an increase in polyunsaturated lipids in the liver remains an open question. We also observed an inverse correlation between fat oxidation and ROS production, which is interesting because in NAFLD increased hepatic fat oxidation is associated with increased ROS production and mitochondrial dysfunction (38). Although we did not measure fat oxidation and ROS generation directly in the liver, it seems reasonable to suggest that exercise did indeed enhance hepatic respiratory efficiency. Exercise increased VO_2max in our participants, and this increased oxidative capacity has been shown to manifest in multiple organs, including the liver. Our conclusion is further supported by recent data showing an increase in hepatic fat oxidation in exercised rats with fatty liver (35). This increased efficiency may translate to less oxidative stress and hence reduced activation of inflammatory pathways and insulin resistance. Our observation that HOMA-IR was attenuated after exercise further supports this supposition.

With regard to the observed decrease in ROS production, it should be noted that a large body of evidence supports a role for oxidative stress and proinflammatory cytokine production in the pathogenesis of NAFLD. Metabolic stress from hyperglycemia and increased oxidation of free fatty acids can result in excessive production of ROS and cytokines from circulating MNCs, hepatocytes, and Kupffer cells, all of which contribute to raising intrahepatic levels of proinflammatory cytokines (39). The model of oxidative stress used in the current investigation capitalizes on the peripheral blood MNCs' ability to produce ROS during hyperglycemia. We have previously shown that TNF-α secretion and ROS generation in response to hyperglycemia is associated with obesity and insulin resistance (24, 26). This seems to be mediated by increased phosphorylation of cytosolic p47^phox and activation of NADPH oxidase (26). Following exercise, we observed that ROS generation from MNCs during the hyperglycemic challenge was normalized to baseline levels. These data are consistent with the postintervention increase in oxidative capacity and insulin sensitivity, which would be expected to attenuate the glucose excursion from the oral glucose load. This method is not without limitations and does not provide a direct indicator of hepatic ROS production; however, it is reflective of decreased ROS production in the context of improved whole body glucose tolerance and insulin sensitivity. Thus, exercise seems to counter 2 of the main protagonists (insulin resistance and oxidative stress) in the 2-hit hypothesis of NASH and this may partially explain how exercise reverses hepatic steatosis (4).

The approach used in this study provides a model to perturb the established pathophysiological state and probe for mechanisms that contribute to the development of NAFLD/NASH. We found that short-term exercise alone generates a favorable change in hepatic lipid composition in patients with well-characterized NAFLD, and this may help prevent disease progression. We conclude that the metabolic response to exercise includes the triggering of cytoprotective mechanisms, possibly involving an increase in HMW adiponectin and a decrease in oxidative stress. Our data suggest that these events are part of the physiological response that contributes to the exercise-induced increase in polyunsaturated lipid in these patients.