Fitness versus fatness: Moving beyond weight loss in nonalcoholic fatty liver disease


  • Nathan A. Johnson,

    1. Discipline of Exercise and Sport Science, University of Sydney, Sydney, Australia
    2. Institute of Obesity, Nutrition and Exercise, University of Sydney, Sydney, Australia
    Search for more papers by this author
  • Jacob George

    Corresponding author
    1. Storr Liver Unit, Westmead Millennium Institute and Westmead Hospital, University of Sydney, Sydney, Australia
    • Department of Medicine, Westmead Hospital, Westmead NSW 2145, Australia
    Search for more papers by this author
    • fax: 9635 7582

  • Potential conflict of interest: Nothing to report.

  • Jacob George is supported by the Robert W. Storr bequest to the University of Sydney and a program grant from the Australian National Health and Medical Research Council (358398)


The rapid emergence of nonalcoholic fatty liver disease (NAFLD) as a cause of both liver-related morbidity and mortality and cardiometabolic risk has led to the search for effective lifestyle strategies to reduce liver fat. Lifestyle intervention comprising dietary restriction in conjunction with increased physical activity has shown clear hepatic benefits when weight loss approximating 3%-10% of body weight is achieved. Yet, the poor sustainability of weight loss challenges the current therapeutic focus on body weight and highlights the need for alternative strategies for NAFLD management. Epidemiologic data show an independent relationship between liver fat, physical activity, and fitness, and a growing body of longitudinal research demonstrates that increased physical activity participation per se significantly reduces hepatic steatosis and serum aminotransferases in individuals with NAFLD, independent of weight loss. Mechanistic insights to explain this interaction are outlined, and recommendations for the implementation of lifestyle intervention involving physical activity are discussed. In light of the often poor sustainability of weight loss strategies, and the viability of physical activity therapy, clinicians should assess physical fitness and physical activity habits, educate patients on the benefits of fitness outside of weight loss, and focus on behavior change which promotes physical activity adoption. (HEPATOLOGY 2010)

Nonalcoholic fatty liver disease (NAFLD) affects ∼30% of adults and a majority of obese individuals.1 In addition to increasing morbidity and mortality from liver disease and cancer, excess liver fat is an independent risk factor for cardiovascular disease, insulin resistance, prediabetes and type 2 diabetes.1 Thus, strategies to modulate the burden of NAFLD are likely to have benefits beyond attenuating liver disease to the broader realm of obesity-related cardiometabolic risk reduction.

Despite its prevalence, treatment options for NAFLD are limited. This is important both for the amelioration of liver disease, as well as for the reduction in morbidity from insulin resistance and diabetes that is often signified by the presence of liver fat. In nondiabetic cohorts, metformin improves aminotransferase levels and reduces steatosis, whereas thiazolidinediones show promise in some studies.1 Concomitant with pharmacotherapy trials, there is increased interest in the efficacy of lifestyle interventions to reduce liver fat and steatohepatitis.2-5 In this context, weight reduction and behavior therapy–based interventions have been reviewed in HEPATOLOGY,6 but there is little information on the role and importance of physical activity in NAFLD.

Physical activity (PA) encompasses structured “exercise” involving aerobic activities at moderate to vigorous intensity (e.g., jogging, brisk walking, bicycling, swimming, skiing, and ball games) and resistance training which comply with current exercise recommendations,7 as well as other leisure-time tasks performed at low intensity below current guidelines for improving cardiorespiratory fitness7 (e.g., casual walking, bicycling, dancing, and nonstructured lifestyle activities such as gardening, house-work, hobbies, and yoga).

This review will trace the history of PA in fatty liver disease management, focusing on studies reporting on the independent effects of PA and the mechanism(s) by which PA may ameliorate hepatic steatosis. The review will conclude with a discussion on practical issues concerning PA prescription in the management of NAFLD.


ALT, alanine aminotransferase; AMPK, adenosine monophosphate–activated protein kinase; AST, aspartate aminotransferase; BMI, body mass index; FFA, free fatty acid; 1H-MRS, proton magnetic resonance spectroscopy; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; PA, physical activity; SREBP-1c, sterol regulatory element binding protein 1c; VLDL, very low density lipoprotein; VO2max, maximal aerobic power.

Effect of Weight Loss Via Lifestyle Modification in NAFLD

When compared with conditions such as type 2 diabetes for which there have been several major randomized trials to examine the efficacy of lifestyle intervention (e.g., Knowler et al.8 and Laaksonen et al.9), there is paucity of such research in NALFD. This, in part, reflects the invasive nature of grading hepatic steatosis by needle biopsy and histology, which limits the capacity for repeated measures of liver fatness. The available data clearly show that lifestyle modification involving combined diet restriction and PA promotion improves liver tests and ameliorates steatosis when reduction in body weight/body mass index (BMI) of ∼6.5%-10% is achieved.10-14 In children, this benefit is comparable to metformin treatment15 (Table 1). The effectiveness of weight loss on hepatic steatosis has been confirmed and quantified by use of proton magnetic resonance spectroscopy (1H-MRS). A loss of 10% of body weight via diet or combined diet and exercise results in a 44%-58% reduction in hepatic triglyceride concentration in overweight adults.14

Table 1. Effect of Lifestyle Modification (Diet and Exercise) on Liver Fat and Liver Tests
Author (Year)nPopulationAge, in Years (± SD)Mean BMI (kg/m2)Sex (M/F,%)Lifestyle InterventionPrimary Outcome MeasureFinding
  1. ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; HCV, hepatitis C virus; 1H-MRS, proton magnetic resonance spectroscopy; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; NR, not reported. Values are means ± standard deviation (SD). ↓ indicates significantly lower (P < 0.05).

Palmer et al. (1990)39Overweight45 (7)32 (4)59/4115.9 ± 12.0 months dietary energy restriction plus unstructured PA therapySerum aminotransferasesALT normalized in individuals achieving ≥10% body weight loss
Ueno et al. (1997)25Overweight39 (13)31 (5)52/4812 weeks diet energy restriction plus progressive increase in daily walking (3000 to 10,000 steps per day) culminating in jogging for 20 minutes twice per day (n = 15) versus Control (n = 10)Histological grading; serum aminotransferasesMean 10% reduction in BMI associated with ↓ALT, ↓AST, and ↓ hepatic steatosis
Hickman et al. (2002)23HCV43 (8)30 (5)58/4212 weeks weekly counseling on diet energy restriction plus physical activity. Diaries used to quantify self-reported exercise habits suggested that most patients had increased their activity to at least 30 minutes of aerobic exercise per day after the intervention phase, with a mean exercise volume at 3 months of ∼215 minutes per weekHistological grading in subset of individuals (n = 10); serum aminotransferasesMean 6.5% reduction in BMI associated with ↓ALT and ↓ waist circumference plus reduction in hepatic steatosis (median reduction of 1 grade)
Hickman et al. (2004)35HCV and non-HCV44 (10)31 (6)60/40Encouraged to maintain an appropriate diet and exercise routine for 12 months maintenance period following 3 months intervention (total 15 months) (Hickman et al. [2002]). Reported physical activity approximately 120 minutes per weekSerum aminotransferasesMean weight loss of 9% body weight and ↓ waist circumference. Individuals who maintained weight loss achieved ALT normalization
Huang et al. (2005)23NASH48 (12)34 (6)48/52Diet counseling targeting weight loss (weekly for 8 weeks; biweekly from 3-6 months; monthly to 12 months). Individualized advice on increasing physical activity with recommendations targeting 70% of calculated “target heart rate”Histological grading in subset of individuals (n = 15); serum aminotransferasesMean 3% weight reduction associated with improvement in serum aminotransferase levels, visceral adiposity, and the NASH score
Suzuki et al. (2005)348serum ALT (>35 IU/L)42 (7)26 (3)100/0Provided information of diet therapy (total energy [especially fat] restriction) plus exercise instruction (increase daily activity and fitness exercise for 20-30 minutes/day, at least 2-3 times per week)Serum aminotransferasesAll subjects who achieved ≥5% weight reduction had improvement in serum ALT levels. The factors associated with ALT normalization were weight loss and commencing or maintaining “regular exercise” (≥ exercise sessions per week, comprising 20-30 minutes of aerobic exercise per session)
Tamura et al. (2005)14Type 2 diabetes diet plus exercise versus diet51 (10)27 (8)57/43Two weeks of energy-restricted diet provision with or without instruction to perform two or three sessions of exercise (30 minutes each) by walking on 5-6 days per week. Exercise intensity was targeted at 50%-60% of maximum oxygen uptake1H-MRSMean 2.6% reduction in body weight associated with ∼20% reduction in hepatic triglyceride concentration
Larson-Meyer et al. (2006)48Overweight diet plus exercise versus diet37 (23)28 (6)42/586 months: 12 weeks of energy-restricted diet provision (then self-selected) with or without instruction to increase energy expenditure by 12.5% via structured exercise (i.e., walking, running, or stationary cycling) 5 days per week. Participants were required to undertake three sessions per week under supervision1H-MRSMean 10% reduction in body weight associated with ∼44% reduction in hepatic triglyceride concentration and ↓ visceral adipose tissue
Sreenivasa Baba et al. (2006)59NASH39 (9)27 (4)78/223 months: advised an exercise program involving brisk walking, jogging or rhythmic aerobic exercises for ≥ 45 minutes/day for ≥ 5 days/week. Counseling to achieve 60%-70% of maximal heart rate and to maintain it for a period of 20 minutes. In addition, those with a high BMI (>25 kg/m2) were advised a moderately energy-restricted dietSerum aminotransferasesMean 2.6% reduction in BMI associated with ↓ALT, ↓AST and ↓ waist circumference. No change in noncompliant subjects (who exercised on fewer than 4 days a week). Nonsignificant weight loss in group prescribed exercise only associated with ↓ALT, ↓AST, and ↓ waist circumference
Osland et al. (2007)65Hospital outpatients with NAFLD and chronic hepatitis C50 (12)33 (5)49/5112 weeks intensive intervention followed by a weight maintenance program involving monthly review for 12 months. Counseling on food choice, nutrition label interpretation, physical activity with the goal of achieving weight loss ≤ 0.5 kg weight loss per weekSerum aminotransferasesMean weight loss of 4.6% associated with ↓ALT and ↓ waist circumference
Schafer et al. (2007)181Impaired glucose tolerance vs. normal46 (11)29 (5)41/599 month intervention: dietary counseling targeting weight loss plus encouraged to perform at least 3 hours of moderate sports per week1H-MRSMean 3.0% reduction in body weight associated with ∼28% reduction in hepatic triglyceride concentration
Nobili et al. (2008)57Overweight and obese children12 (NR)26 (NR)70/3024 months diet energy restriction plus recommendation to engage in moderate aerobic physical exercise for 45 minutes/day (Control, n = 28) versus metformin with lifestyle intervention (n = 29). Patients and responsible guardians also received 1-hour counseling sessions on nutrition and physical activity on a monthly basisHistological grading in subset of individuals (n = 27); serum aminotransferases; ultrasoundMean 10% reduction in BMI associated with ↓ALT, ↓AST, and ↓ steatosis and liver fatness
Kantartzis et al. (2009)170Overweight men and women45.6 (10.4)28.9 (3.9)41/59Up to 10 sessions of dietary counseling plus encouraged to partake in 3 hours of moderate-intensity sports per week1H-MRSMean 3.5% reduction in body weight associated with ∼35% reduction in hepatic triglyceride concentration associated with ↓ body fatness, ↓ visceral adiposity, and ↓ insulin resistance

If improvement in body weight is the goal, other reports show that more modest reductions (<5% of body weight) also confer benefits.3, 16-18 Suzuki et al. hinted at the critical role of PA, showing that both weight loss and commencing or maintaining “regular exercise” were associated with alanine aminotransferase (ALT) reduction.18 For every 5% weight loss, a 3.6 greater likelihood of ALT normalization was observed.18 Studies employing 1H-MRS confirm these reports. Two weeks of combined diet and exercise therapy evoking a 2.6% reduction in weight led to an ∼20% reduction in hepatic triglycerides in patients with type 2 diabetes.19 Likewise, a mean 28% reduction was seen in individuals with prediabetes following a 3% loss of body weight20 (Table 1). This extent of change is outside the coefficient of variation for this technique.1 The largest study to quantify changes in hepatic triglyceride concentration with lifestyle intervention was undertaken by Kantartzis et al.2 In overweight and obese men and women who achieved a 3.5% reduction in body weight with diet and exercise therapy, hepatic triglycerides decreased by 35% after 9 months in those with liver fat >5.56% at baseline. This benefit was associated with significant improvement in cardiorespiratory fitness.2 The results suggest that the synergy of dietary energy restriction and PA therapy positively influences hepatic steatosis when weight loss approximating 3%-10% of body weight is achieved (Table 1).

Physical Activity in NAFLD

Weight loss remains fundamental to the management of NAFLD, but is mistakenly perceived as the primary rationale for promoting PA participation. However, obesity management is not simply a function of weight loss. Outside the context of liver disease, it is well established that exercise enhances insulin sensitivity, reduces progression to type 2 diabetes, and favorably modifies serum lipids independent of weight loss.8, 9 When combined with the observation that high fitness and habitual physical activity are associated with improved functional capacity, quality-of-life measures, well-being, and reduced all-cause mortality,7 the importance of incorporating PA therapy, beyond assisting weight loss, becomes apparent.

This argument of “fitness versus fatness” is relevant given that results of randomized clinical trials suggest that weight loss via diet and/or PA therapy is typically modest (1-8 kg) and returns to baseline within 1-3 years.21 Thus, although weight loss should be the goal, there is a practical challenge to achieving sustainable weight loss with lifestyle therapy. A beneficial independent effect of PA would provide a second practical intervention target.

Cross-Sectional Data.

Epidemiologic studies show a negative relationship between NAFLD and self-reported habitual PA levels,22-26 although this may not persist when adjusted for body weight23, 24 (Table 2). These observations need to be viewed with caution because subjective measures of PA level can be inaccurate and prone to reporter bias.27 Yet, the argument that there is a benefit in exercise participation against NAFLD is also supported by data, which demonstrate an inverse correlation between measures of NAFLD and cardiorespiratory fitness.2, 28-30 This is commonly described as maximal aerobic power (VO2max) and typically increases with exercise training (Table 3). However, the association is not apparent in all studies31 including those with lean individuals32; it appears to be independent of BMI but not visceral adiposity2, 28-30 (Table 3).

Table 2. Cross-Sectional Studies: Reported Habitual Physical Activity, Liver Fat and Liver Tests
Author (Year)nPopulationAge, in Years (± SD)Mean BMI ISO (kg/m2)Sex (M/F%)ModeOutcome MeasureFinding
  1. ALT, alanine aminotransferase; GGT, gamma-glutamyltransferase; 1H-MRS, proton magnetic resonance spectroscopy; LF, liver fatness; NAFLD, nonalcoholic fatty liver disease; NR, not reported; PA, physical activity level inferred by questionnaire. Values are means ± standard deviation (SD). ∝ indicates correlation; ↓ indicates significantly lower; ↔ indicates no significant relationship (P > 0.05).

Hsieh et al. (1998)3331Japanese male48.3 (8.4)23.1 (2.5)100/0aerobicultrasound↓ prevalence of NAFLD in reported exercise
↓ prevalence of NAFLD with 2 or ≥3 days exercise per week vs. sedentary
↓ prevalence of NAFLD with ≥3 days exercise per week vs. 1 day per week
Tiikainen et al. (2002)27Previous gestational diabetes20-5028-350/100aerobic1H-MRS↓ reported exercise in NAFLD vs. normal (3 days vs. 5 days/week)
Lawlor et al. (2005)3789Older British female69.9 (NR)27.6 (NR)0/100aerobicALT, GGT↓ GGT with increasing PA
↓ ALT with increasing PA but not with BMI adjustment
Perseghin et al. (2007)191NAFLD vs. controls34.3 (8.8)27.4 (3.8) vs. 23.7 (3.6)60/40aerobic1H-MRS↓ reported habitual exercise in NAFLD vs. controls
↓ LF in most physically active individuals
LF ∝ reported habitual physical activity level
Zelber-Sagi et al. (2009)349Mixed Israeli cohort50.7 (10.4)27.2 (4.5)53/47aerobic + resistanceultrasound↓ LF in more physically active but not with BMI adjustment for aerobic exercise
↓ reported habitual exercise in NAFLD vs. normal
Table 3. Cross-Sectional Studies: Cardiorespiratory Fitness, Liver Fat, and Liver Tests
Author (Year)nPopulationAge, in Years (± SD)Mean BMI (kg/m2)Sex (M/F%)ModeOutcome MeasureFinding
  1. BMI, body mass index; CRF, cardiorespiratory fitness; CT, computed tomography; 1H-MRS, proton magnetic resonance spectroscopy; LF, liver fatness; NAFLD, nonalcoholic fatty liver disease. Values are means ± standard deviation (SD). ∝ indicates correlation; ↓ indicates significantly lower; ↔ indicates no significant relationship (P > 0.05).

Seppala-Lindroos et al. (2002)30NAFLD vs. controls43.0 (7.7)25.5 (3.9)100/0cycle1H-MRS↔ LF ∝ CRF
Nguyen-Duy et al. (2003)161Caucasian men52.4 (6.1)27.4 (3.7)100/0treadmillCTSignificant inverse ∝ LF and CRF but not independent of visceral adiposity
Kuk et al. (2005)86Premenopausal females40.8 (3.9)21.4 (1.7)0/100treadmillCT↔ LF ∝ CRF
Church et al. (2006)218NAFLD vs. normal52.0 (6.7)32.4 (4.3) vs. 27.8 (3.9)100/0treadmillCTSignificant ∝ ↓ NAFLD prevalence high CRF but not independent of visceral adiposity
McMillan et al. (2007)293High vs. low LF50.5 (8.2)28.6 (4.2)100/0treadmillCTSignificant inverse ∝ LF and CRF but not independent of visceral adiposity
Kantartzis et al. (2009)170Overweight men and women45.6 (10.4)28.9 (3.9)41/59cycle1H-MRSSignificant inverse ∝ LF and CRF but not independent of visceral adiposity

The Impact of PA Intervention in NAFLD (Delineating PA from Diet Effects).

From their 1-year longitudinal cohort study, Suzuki and colleagues noted that normalization of ALT was 2.5 times more likely in those participating in regular exercise (≥once a week) than those who did not.18 Likewise, Huang and coworkers noted improvements in liver histology in patients who reported increasing their PA habits.16 Finally, on the basis of data from their lifestyle intervention, Kantartzis and colleagues also showed that although reduction in hepatic fat was a function of weight loss, cardiorespiratory fitness was an independent and best predictor of change in hepatic triglyceride, independent of total and visceral adiposity.2 It was suggested that cardiorespiratory fitness modulates liver fat via effects on hepatic mitochondrial biogenesis, leading to increased lipid oxidation.2 Interestingly, reported habitual PA level and increase in PA levels did not predict the hepatic benefit.2

The Impact of PA per se in NAFLD.

Research that has examined the efficacy of PA in isolation provides generally favorable evidence for an independent effect in modulating liver tests and liver fat (Tables 1 and 4). In a prospective study in which a subgroup with NASH received exercise prescription without dietary intervention, Sreenivasa Baba et al. found an ∼50% reduction in aminotransferase levels after 3 months. This was associated with a reduction in waist circumference, yet no change in BMI. No changes in ALT or aspartate aminotransferase (AST) were observed in patients who were noncompliant.3 The largest study to examine for possible independent effects of PA in NAFLD showed that 3 months of a behavior change–based intervention significantly improved liver enzymes independent of weight loss.5 In this study, participants who achieved low (60-119 minutes/week) and moderate (120-239 minutes/week) volumes of incidental PA experienced a hepatic benefit, with the greatest improvement in those who increased their PA levels to >150 minutes/week and had improved VO2max.5

Table 4. Intervention Studies: Exercise (Without Diet Modification) and Liver Fat
Author (Year)nAge, in Years (± SD)Mean BMI (kg/m2)Sex (M/F%)ModeProgram Length (weeks)Session Frequency (per week)Exercise Session Duration (minutes)Exercise IntensityOutcome MeasureFinding
  1. BMI, body mass index; CT, computed tomography; 1H-MRS, proton magnetic resonance spectroscopy; LF, liver fatness; VO2peak, peak aerobic capacity. Values are means ± standard deviation (SD). ↓ indicates significantly lower (P < 0.05); ↔ indicates no significant relationship (P > 0.05).

Shojaee-Moradie et al. (2007)1750 (10)27.6 (2.1)100/0variable63≥20 min60%-85% VO2peak1H-MRS↔ LF
Devries et al. (2008)3440 (8)28.6 (4.8)44/56cycling122-3week 1:30 min; increased to 60 min by final weekweek 1:50% VO2peak; increased to 65%-70% VO2peak by final weekCT↔ LF
Johnson et al. (2009)1948 (9)31.8 (2.8)65/35cycling43week 1:30 min; week 2:45 min; week 3,4:45 minweek 1:50% VO2peak; week 2:60% VO2peak; week 3,4:70% VO2peak1H-MRS↓ LF
Van der Heijden et al. (2010)2915 (1)33.7 (4.3) vs. 20.6 (3.0)59/41variable122 + 2 (unsupervised)30 min∼70% VO2peak1H-MRS↓ LF

Four studies have examined the effect of exercise therapy in isolation on hepatic steatosis (Table 4). Twelve weeks of structured, supervised cycling had no measurable effect on liver fatness as inferred by computed tomography in a mixed cohort of obese and lean individuals.33 This outcome was corroborated by Shojaee-Moradie and colleagues.34 Subjects in this report were overweight but did not have NAFLD (mean hepatic triglyceride = 3.95%). In contrast, using a short-term (4-week) aerobic exercise intervention, Johnson et al. demonstrated that both steatosis and visceral adiposity were reduced without any change in body weight in previously sedentary obese individuals with NAFLD. Subjects allocated to a progressive aerobic exercise program over 4 weeks experienced a mean 21% reduction in hepatic triglycerides. This occurred despite no loss of subcutaneous adiposity or change in dietary macronutrient content and composition.35 An independent benefit of aerobic exercise training has recently been confirmed by van der Heijden et al., who observed a reduction in hepatic triglyceride concentration (∼37%) and visceral adiposity, despite body weight maintenance in previously sedentary obese adolescents, but not in previously sedentary lean adolescents.36 That no hepatic benefit was detectable in the previous report that examined exercise training effects via liver density estimates (computed tomography)33 is not unexpected given the qualitative nature of the technique and its poor sensitivity.1 The reason for the conflicting findings from studies employing 1H-MRS is unclear and may reflect differences in subject population, baseline liver fatness, or the exercise training intensities and modalities employed (Table 4).

NAFLD Pathophysiology and the Mechanism of an Exercise Effect

Hepatic triglyceride concentration is a function of (1) the delivery of free fatty acids (FFAs) to the liver from dietary sources and adipose tissue; (2) de novo lipogenesis; (3) hepatic β-oxidation; and (4) very low density (VLDL) lipoprotein synthesis, export, and clearance (Fig. 1A). Donnelly et al. demonstrated that in obese individuals with NAFLD, adipose-derived plasma FFAs are the dominant contributor to hepatic steatosis, with de novo lipogenesis and dietary fatty acids accounting for approximately 25% and 15% of hepatic triglyceride formation, respectively37 (Fig. 1A). Based on this data,37 it could be argued that strategies which ameliorate the delivery of FFAs to the liver from adipose tissue should impart the most significant benefit in reducing liver fat.

Figure 1.

(A) Processes involved in the development of hepatic steatosis. Thick lines indicate dominant effect on triglyceride concentration. FFAs from adipose tissue are the dominant source of fatty acids for hepatic triglyceride, followed by de novo lipogenesis and chylomicron-derived fatty acids from dietary fat. Hepatic triglyceride concentration is also a function of hepatic β-oxidation and the synthesis, removal and clearance of very low density lipoproteins. (B) Mechanisms for a benefit of exercise on hepatic triglyceride concentration. Unproven effects are indicated by dashed lines. Regular exercise is known to a) reduce subcutaneous and visceral adipose tissue mass; b) reduce availability of adipose-derived FFAs to the liver; c) enhance adipose insulin sensitivity leading to lower FFAs and possibly reduced de novo lipogenesis. Rodent studies demonstrate that exercise increases d) β-oxidation; e) VLDL synthesis, removal and clearance and f) decreases de novo lipogenesis. β-ox, hepatic β-oxidation; DNL, de nova lipogenesis; FABPs, fatty acid binding proteins; FFAs, free fatty acids; TGs, triglycerides; VLDL, very low density lipoprotein.

PA and Fatty Acid Delivery to the Liver.

Exercise substantially increases whole-body fatty acid oxidation, reflecting the augmented respiration rate within working skeletal muscle. Fat oxidation increases as a function of exercise duration and intensity, with the absolute rate highest at ∼50%-70% of VO2max. It declines during vigorous exercise, and often remains elevated for hours into the postexercise period.38 Whether this acute redistribution of fatty acids to muscle positively affects the hepatic triglyceride pool is unknown, but would seem unlikely given that hepatic FFA uptake is a function of FFA delivery, which increases with blood flow and the elevated plasma FFA concentration during acute exercise.39

The adaptive response to regular exercise (training) involves a number of putative candidates, which possibly contribute to hepatic benefits. Regular exercise increases mitochondrial biogenesis and capillarization which augments the capacity for fatty acid uptake, β-oxidation, and triglyceride storage within skeletal muscle,38 potentially serving to partition fatty acids away from the liver (Fig. 1B). Although there are few studies on hepatic FFA delivery, a report by Hannukainen et al. has shown in monozygotic twins that those with higher aerobic capacity had a lower uptake of FFAs into the liver.40 Because portal vein blood flow is unaffected by training,41 this suggests an effect of FFA concentration and/or hepatic FFA extraction. The exercise benefit may therefore reflect the cumulative effect of regular exercise training contributing to net fat loss and/or visceral adiposity reduction over time,42, 43 thereby reducing the FFA load on the liver (Fig. 1B). The clear effectiveness of moderate lifestyle-induced weight loss, which is almost invariably associated with visceral adipose tissue reduction,42 is consistent with this notion.

The apparent independent hepatic benefit of PA therapy also suggests that alterations in adipose function beyond actual fat loss, including alterations in adipose insulin sensitivity and adipokine secretion, may be of importance. Regular aerobic exercise enhances insulin sensitivity, and in obese individuals, the benefit of exercise without weight loss is similar to that observed following weight loss.44 Although exercise-induced insulin sensitization is commonly discussed with reference to amelioration of skeletal muscle insulin resistance,45 in adipose tissue, insulin resistance manifests as a reduced ability to suppress lipolysis with insulin,45 leading to increased FFA release. The degree of adipose insulin resistance correlates with hepatic triglyceride concentration in individuals with type 2 diabetes and NAFLD.46 By improving insulin sensitivity, aerobic exercise training thus results in a lower FFA concentration under both basal and insulin-stimulated conditions.44

Whether the hepatic benefit of exercise reflects lower adipose lipolysis and FFA availability and/or a direct effect on hepatic FFA uptake independent of FFA concentration is unclear. In the context of diet-induced weight loss, reductions in liver fat parallel the attenuation of hepatic FFA uptake despite similar basal FFA concentrations,47 which suggests that the liver may not be a passive bystander with adipose tissue acting as the locus of control. Therefore, it has been suggested that hepatic fatty acid uptake may be regulated,48 perhaps through altered expression and activity of fatty acid translocase or cluster of differentiation 36 (CD36).

PA and Intrinsic Hepatic Effects: VLDL Metabolism, β-Oxidation, and Lipogenesis.

Despite evidence in rodents which suggests that aerobic exercise can increase VLDL secretion and clearance,49 in exercising humans, fatty acids from adipose-derived and intramyocellular triglyceride-derived lipolysis account for almost all whole-body fat oxidation. The contribution of VLDL triglyceride-derived fatty acids is believed to be negligible38 (Fig. 1B).

Emerging evidence indicates that exercise may also modulate hepatic fat by directly altering hepatic β-oxidation and/or lipogenesis. The propensity to store triglyceride within hepatocytes is related to low mitochondrial content and associated low rates of fatty acid β-oxidation, which is exceeded by hepatic FFA uptake (Fig. 1B). Similarly, lower fasting and glucose-stimulated insulin concentrations after exercise training44 may reduce insulin-mediated hepatic conversion of FFAs to triglycerides (Fig. 1B). Unfortunately, human studies examining direct hepatic effects of exercise therapy on hepatocellular biochemistry are restricted by the limitation of obtaining liver tissue, and no human data are available.

Sedentary rats genetically bred for low aerobic capacity have higher sterol regulatory element binding protein 1c (SREBP-1c), a transcription factor that regulates genes which promote triglyceride synthesis, with associated reductions in hepatic mitochondrial volume density and capacity for fatty acid oxidation.50 However, it is difficult to dissociate these adaptations from factors external to the liver. For instance, when compared with that of high-fitness rats, those with low aerobic capacity had increased adiposity, including visceral adiposity and insulin resistance, which is known to increase hepatic fatty acid synthesis via SREBP-1c.51 More recently, Rector et al. have shown that hepatic fatty acid oxidation increases and de novo lipogenesis declines with exercise training in rodent models of obesity and type 2 diabetes, but initiation of sedentary behavior elevates hepatic triglyceride (Fig. 1B). The latter was accompanied by enzyme alterations which initiate hepatic fat accumulation.52, 53

In human NAFLD, variability in the expression of peroxisome proliferator-activated receptor-delta, which is involved in the regulation of hepatic mitochondrial biogenesis, has been shown to affect liver fatness. Namely, homozygous and heterozygous carriers of the rs1053049, rs6902123, and rs2267668 single-nucleotide polymorphisms experienced less pronounced reductions in visceral and hepatic fat in response to lifestyle intervention.54 The signal for these adaptations may be adenosine monophosphate-activated protein kinase (AMPK), whose activity is increased during and after exercise in rodents.55 Although direct studies of exercise training are absent, AMPK activation is known to attenuate malonyl-coenzyme A and subsequently to increase fatty acid entry and oxidation within mitochondria (perhaps due to hepatic acetyl-coenzyme A carboxylase inhibition),55 and reduce lipid synthesis and insulin resistance.55 These effects are modulated by adipokines, particularly adiponectin, which up-regulates AMPK in both skeletal muscle and liver and also reduces hepatic glucose production. Although adiponectin concentration has been shown to increase following significant weight loss (∼10% body weight), an independent effect of exercise is yet to be established.56

Clearly, a number of plausible candidates exist to explain the hepatic benefit of PA therapy, and synergistic actions likely exist. The pertinent question concerns the primary locus of an exercise-mediated benefit in NAFLD, because this has direct implications for exercise prescription. For example, if exercise exerts the bulk of its benefit via lowering visceral adiposity, therapies known to effect visceral adipose tissue reduction (including weight loss) would be best advocated in NAFLD. Yet, if enhancement of cardiorespiratory fitness or insulin sensitivity confers substantial hepatic improvements, there are methods of achieving this which are not contingent upon high energy expenditure and/or weight loss. For instance, progressive resistance training is a stimulus for whole-body insulin sensitization43 and carries less time cost than current aerobic exercise guidelines. In this regard, Zelber-Sagi et al. recently noted an inverse relationship between resistance training and NAFLD, which persisted after adjustment for BMI.24 Data from experimental studies involving young, lean cohorts clearly show that exercise training involving repeated (5-8 times) short bursts of cycling exercise (10-30 seconds) increases maximal aerobic power and muscle oxidative enzymes and lowers plasma triglycerides to an equivalent level to that seen with traditional aerobic exercise training regimes, despite a 70%-90% reduction in energy expenditure and weekly time commitment.57 Such studies are clearly warranted, because lack of time is the principal reason for drop-out from structured exercise programs and the most commonly cited barrier to initiating exercise.27

Implementing PA Therapy in NAFLD

At present, there is an overall paucity of evidence concerning the benefits of PA as treatment for NAFLD. What is available shows a conclusive benefit of PA when coupled with energy restriction when weight loss is achieved, and it is encouraging for an independent benefit in the absence of weight loss. Although weight loss remains fundamental, patients should be counseled on the spectrum of benefits conferred by regular PA. Management should include assessment of cardiorespiratory fitness and PA levels, and the setting of lifestyle goals based on adoption of regular exercise, with a focus on the attainment of sustainable PA habits.

The dose (intensity and volume) of PA required to reduce liver fat remains unclear. Furthermore, from the present evidence, it is difficult to discern the relative importance of structured exercise and fitness versus less structured PA. This conundrum is borne out in data from cross-sectional research, which shows that both high PA and cardiorespiratory fitness correlate negatively with fatty liver (Tables 2 and 3). Although several examples of a hepatic benefit from low-dose PA therapy have been cited,5, 18, 25, 35 in the absence of robust data and knowledge of the long-term sustainability of such outcomes, it would seem reasonable to promote the current public health recommendations for health promotion, disease prevention, and weight management (Table 5). This recommends that individuals accumulate 20-60 minutes or more of moderate intensity (∼45%-70% of VO2max) exercise on most days of the week.7 If weight loss is the goal, exercise, even when prescribed without associated restriction of energy intake, confers a reduction in body weight in an apparent dose-response fashion with exercise volume.43 Greater amounts of exercise may be needed for most individuals to induce significant weight loss or prevent weight being regained in the long term. The consensus suggests that little weight loss is achieved with <150 minutes of exercise per week, modest (∼2-3 kg) losses are attainable with >150 minutes/week (with an energy equivalent of ∼1200-2000 kcal/week), and moderate weight loss (∼5-7.5 kg) often results from 225-420 minutes/week (∼1800-3300 kcal) of aerobic activity.43

Table 5. Recommendations for Physical Activity in NAFLD
Patients should be appropriately screened for contraindications prior to initiating exercise testing or therapy
Physical fitness assessment via exercise testing. Physical activity level assessment by subjective (questionnaire/diary) or objective (e.g., accelerometer) means
Accumulate 20-60 minutes or more of moderate intensity rhythmic exercise using large muscle groups on at least 5 days/week
Moderate intensity PA between 150 and 250 min/week for preventing weight gain
PA >250 min/week for clinically significant weight loss
Moderate-to-high intensity resistance training 3 days/week for enhancing insulin sensitivity

These targets can be achieved using a variety of exercise modalities, with the outcome of cardiorespiratory fitness being a reliable and easily quantifiable endpoint measure of structured aerobic exercise. Although there is currently no longitudinal evidence available concerning its benefit in NAFLD, progressive resistance training may be useful for the management of obesity-related comorbidities, particularly insulin resistance.43 The benefits of nonstructured leisure-time PA, including reduced sedentary time, are becoming increasingly recognized and have, in some studies, shown efficacy in improving cardiometabolic risk and promoting weight loss.43 Clear guidelines for such “lifestyle PA” are lacking, and reliable measurement, particularly of intensity, is more difficult. PA habits and adherence can be estimated by questionnaires, pedometers, and accelerometers (reviews of which can be found elsewhere27), and the latter may further promote adherence to PA.27

Sustainable Lifestyle Change Involving PA.

A major consideration for lifestyle therapy is that adherence to diet and PA regimens can be poor in a clinical setting, for example.8, 9 The diabetes prevention studies provide important insights regarding behavior therapy to target PA adherence. Although different approaches were used, the intervention arms in all studies included behavioral strategies for reinforcing prescribed changes in PA, dietary intake, or a combination of the two, and included initial lifestyle counseling sessions and ongoing regular contact, self-selection of goals and PA strategies, and recording of participation, which is known to enhance adherence.8, 9

The success of these interventions and their relatively low drop-out rate (<10%) is partly attributable to the way in which lifestyle modification was reinforced. Common to all interventions was individual counseling, goal setting, regular assessment (every 3-12 months), and multiple contacts (∼6-20 times per year) with staff, an approach mirrored in some of the intervention studies in NAFLD. The absence of behavior change or risk reduction in the control groups highlights the importance of intensive support but also the difficulties of implementing lifestyle therapy in clinical practice. In a recent study, 152 subjects (141 with NAFLD and 11 with chronic hepatitis C) with elevated liver enzymes, abdominal obesity, and a range of metabolic risk factors were randomized to either a moderate or low-frequency lifestyle counseling intervention or to a control group.4 Improvements in all metabolic risk factors and liver enzymes were observed in the moderate-frequency counseling group, versus a smaller number of changes in the low-frequency intervention group and no change in any risk factors in control subjects.4 Such intensive approaches imply the need to reevaluate the traditional role of the clinician and suggests the need for multidisciplinary teams.6

Summary and Recommendations

Diet and/or PA intervention is important in the management of NAFLD, and there is increasing evidence that exercise per se beneficially modulates liver fat independent of weight loss. The latter effects should be emphasized and ideally delivered using a multidisciplinary approach. There is an obvious need for further research to understand the effectors of exercise-mediated benefits in NAFLD, including PA dose, modality, and the relative importance of structured exercise and cardiorespiratory fitness versus less-structured lifestyle PA levels. Clarification of these will enable the formulation of effective and time-efficient PA programs which may ultimately enhance patient benefit, participation, and adherence.


The authors are grateful to Don Chisholm and Michael Baker for their assistance with the manuscript.