Independent effects of physical activity in patients with nonalcoholic fatty liver disease

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Nonalcoholic fatty liver disease, characterized by elevated liver enzymes, central obesity, and insulin resistance, is becoming increasingly prevalent. The effects of changes in physical activity on the metabolic profile of this group have not been reported. We assessed at 3 months the impact of a behavior change-based lifestyle intervention on physical activity and the effects of this change on the metabolic profile of people with fatty liver disease. In all, 141 participants with nonalcoholic fatty liver disease were prospectively enrolled into either a low- or moderate-intensity lifestyle intervention or to a control group. Physical activity was assessed using a validated reporting tool and physical fitness was measured using the YMCA protocol on a cycle ergometer. Individualized counseling to increase physical activity was provided. Overall, 96% of participants attended the 3-month follow-up assessment. Participants in the moderate- and low-intensity intervention groups were 9 times more likely to increase physical activity by an hour or more per week compared to controls. Patients increasing or maintaining their reported physical activity to ≥150 minutes/week, and those who increased their objective levels of fitness, had the greatest improvements in liver enzymes and other metabolic indices compared to those who were least active. This effect was independent of weight loss and was corroborated by an objective measure of fitness. There was no dose-response effect on liver enzymes with incremental increases in physical activity above 60 minutes/week. Conclusion: Lifestyle counseling interventions are effective in improving physical activity behavior. Maintaining or increasing physical activity provides health benefits for patients with fatty liver, independent of changes in weight. (HEPATOLOGY 2009.)

Nonalcoholic fatty liver disease (NAFLD) is the commonest form of chronic liver disease in developed countries.1–3 It is characterized by elevated liver enzymes and components of metabolic risk, including obesity, insulin resistance, and type 2 diabetes mellitus (T2DM).4–6 Both elevated body mass index (BMI) and T2DM are associated with higher rates of fibrosis progression and are predictors of advanced NAFLD.7–9 Although lifestyle change is currently the main recommendation for patients with NAFLD,10 the effect of increasing physical activity (PA) has not been thoroughly investigated.11–16

PA has been shown to reduce the risk of T2DM, insulin resistance, hypertension, dyslipidemia, impaired fasting glucose (IFG), and the metabolic syndrome.17–20 This indicates that PA might play a role in the treatment of patients with NAFLD, yet previous interventions have predominantly targeted weight loss through dietary change.11, 15, 16, 21 Studies that have assessed changes in PA have not reported on its effect per se on physiologic and metabolic outcomes.12, 14–16, 22

Given that PA may directly assist in reducing hepatic steatosis and hepatic insulin resistance,23–25 studies that assess the impact of interventions aimed at increasing PA in patients with liver disease are warranted. This is particularly important as weight loss and weight loss maintenance are often difficult to achieve and not all patients presenting with hepatic steatosis are overweight.10

This study reports on the effects of a lifestyle intervention on PA levels in patients with a metabolic component to their liver disease, and specifically focuses on the effects of changes in PA on health outcomes at 3 months. The aims were to report at 3 months (1) the impact of a counseling-based lifestyle intervention on PA behavior; (2) the relationship between changes in PA and fitness on liver enzymes and metabolic parameters; (3) the extent of change in PA that is required to influence metabolic targets; and (4) whether the relationship between change and health outcomes was similar for both PA and objective cardiorespiratory fitness changes.

Abbreviations

ACSM, American College of Sports Medicine; ALT, alanine aminotransferase; ANCOVA, analysis of covariance; ANOVA, analysis of variance; AST, aspartate aminotransferase; BMI, body mass index; GGT, gamma glutamyl transpeptidase; HDL, high-density lipoprotein; HOMA-IR, Homeostasis Model Score for Insulin Resistance; IFG, impaired fasting glucose; LDL, low-density lipoprotein; NAFLD, non-alcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; PA, physical activity; PAQ, Physical Activity Questionnaire; SD, standard deviation; T2DM, type 2 diabetes mellitus; VO2, maximum volume of oxygen.

Patients and Methods

Study Group.

Patients were enrolled from the Sydney West Area Health Service. The study protocol was approved by the Human Research Ethics Committees of the Sydney West Area Health Service and the University of Sydney. Written informed consent was obtained from all participants and NAFLD was diagnosed on the basis of abnormal liver enzymes, a detailed history, and exclusion of other causes of liver disease by appropriate serological and biochemical tests as reported.26 The present dataset only comprises subjects with NAFLD and excludes the 11 patients with chronic hepatitis C included in the earlier analysis.26 Of 179 patients that underwent baseline screening, 141 were eligible for the study. Participants were deemed ineligible if they had liver cancer, severe depression, thyroid disease, Cushing's syndrome, or chronic pancreatitis. Other liver diseases were excluded by appropriate tests including hepatitis B (negative for hepatitis B surface antigen), hepatitis C (negative for hepatitis C antibody), autoimmune liver disease (antinuclear antibody, antismooth muscle antibody, anti-Liver Kidney microsomal antibody and antimitochondrial antibody), Wilson's disease (serum copper and ceruloplasmin levels), alpha-1 antitrypsin deficiency (alpha-1 antitrypsin levels and alpha-1 antitrypsin genotype), and hemochromatosis (iron studies and genotyping for C268Y and H63D mutations). Participants with alcohol consumption >20 g/day in males and 10 g/day in females were excluded. Alcohol consumption was excluded by two treating physicians and close family members of the patient as reported.4 In addition, detailed interview-based diet histories, 3-day food records, and questionnaires on alcohol consumption were completed by patients at the baseline assessment. During the course of the study alcohol consumption history was confirmed at each clinic visit.

Study Design.

At baseline, 141 eligible patients were randomized to one of three intervention arms or a control group. Group A (n = 38) was a low-intensity intervention in which patients received three counseling sessions (Baseline, Weeks 2 and 4). Groups B (n = 33) and C (n = 36) received an identical moderate-intensity intervention in the first 3 months, comprising six fortnightly counseling sessions. Group C went on to receive additional long-term telephone support postintervention (3 to 12 months) to examine the effects of a maintenance program versus none. We have reported baseline data for Groups B and C separately in Table 1 to demonstrate the randomization process and trial design, and then combined Groups B and C in reporting the results, as they had identical interventions to 3 months. Subjects in the control group (n = 34) received one consultation at baseline and the results of their assessment were discussed along with the PA goals of the intervention. However, they were not provided with support, advice, or assistance in making changes to PA and similarly not advised to not make changes if they wanted to do so. The control phase was limited to 3 months as it was not considered ethical to withhold the intervention from patients for a longer period of time. At 3 months, 24 of the 34 subjects in the control group elected to continue in the study and were randomized to one of the three intervention Groups A (n = 11), B (n = 7), or C (n = 6). The results reported for the “intervention” groups here includes data from the original participants in Groups A, B, and C, as well as those from the control group who were subsequently allocated to one of the intervention groups.

Table 1. Baseline Data for Demographic Factors, Metabolic Indices, Liver Enzymes, Physical Activity, and Diet for Patients Originally Randomized to Each of the 3 Intervention Groups and the Control Group (N = 141)
Characteristics (Reference Range)Intervention GroupsControls
ABCD
N38333634
  • The baseline data presented here are similar to that published elsewhere, but this study group is different as it is confined to patients with NAFLD only.

  • Data are mean ± SD. There were no significant differences between groups in any of the study variables.

  • Continuous variables were analyzed by ANOVA with Scheffe post-hoc contrast.

  • Log transformations were undertaken for all variables with nonnormal distributions.

  • Chi-squared analysis was used to compare categorical data.

  • Variable reference range

  • Based on results of 102 subjects completing a 2-hour oral glucose tolerance test (OGTT) at baseline.

 Male (N; %)20 (53%)25 (76%)20 (56%)22 (65%)
 Age (years)47.5 ± 12.449.8 ± 10.248.6 ± 1.446.8 ± 12.2
Metabolic Indices    
 BMI (kg/m2)31.9 ± 6.031.7 ± 5.331.8 ± 5.031.5 ± 5.1
 Weight (kg)88.2 ± 16.692.0 ± 15.690.2 ± 20.790.1 ± 18.2
 Waist circumference (cm)106.9 ± 12.9106.4 ± 11.5106.5 ± 14.2105.4 ± 13.4
 Systolic pressure (110-130 mmHg)129.1 ± 13.7126.6 ± 11.6129.0 ± 12.7129.0 ± 13.2
 Diastolic pressure (70-80 mmHg)86.0 ± 9.785.7 ± 5.284.7 ± 6.986.5 ± 9.0
 Fasting glucose (<5.5 mmol/L)5.9 ± 1.95.6 ± 0.65.7 ± 1.65.7 ± 1.7
 Fasting insulin (< 26 mU/L)21.9 ± 13.817.6 ± 11.218.9 ± 9.922.3 ± 12.0
 120m insulin (<60 mU/L)88.7 ± 50.1113.8 ± 73.8126.9 ± 76.9125.3 ± 67.3
 HOMA-IR6.2 ± 5.44.3 ± 2.95.2 ± 3.56.0 ± 4.4
 Triglycerides (<2.0 mmol/L)2.2 ± 1.81.8 ± 1.01.9 ± 0.91.8 ± 1.0
 Total cholesterol (<4.0 mmol/L)5.1 ± 1.04.8 ± 1.05.2 ± 0.85.1 ± 1.1
 LDL-C (<3.5 mmol/L)2.80 ± 0.942.74 ± 0.822.95 ± 0.752.92 ± 0.96
Liver Enzymes    
 Ferritin (30-300 μg/L)221.7 ± 186.3237.1 ± 151.0215.6 ± 181.8257.6 ± 180.0
 ALT (< 35 U/L)65.6 ± 57.565.6 ± 34.667.0 ± 31.268.8 ± 41.1
 GGT (M: < 43; F: <30 U/L)92.6 ± 87.470.6 ± 95.298.0 ± 97.889.8 ± 88.4
 AST (<40 U/L)45.0 ± 25.343.9 ± 20.041.9 ± 18.042.2 ± 16.7
Physical Activity and Diet    
 Physical activity (≥150 min/wk)200 ± 288236 ± 314206 ± 214233 ± 230
 Kilojoule intake (per day)8451 ± 27699179 ± 29878287 ± 21258718 ± 2169

Intervention Design.

Exercise scientists provided individually tailored counseling to assist people in making changes to their PA habits. Participants were encouraged to increase both planned and incidental moderate-intensity PA to achieve at least 150 minutes/week for general health and to target more than 200 minutes/week for weight loss.27 Walking was the primary modality encouraged; however, the type, amount, frequency, and intensity of increase varied depending on baseline level of activity, medical history, and the personal preferences of each patient.

Measurement.

Measurement of weight, height, waist, and hip circumference and blood pressure were undertaken using standard protocols.28 The Active Australia Survey, an established and reliable self-report measure,29 was used to assess PA. To corroborate the data obtained from the PA questionnaire, half the study group (n = 69) undertook a submaximal aerobic assessment at baseline and 3 months. The assessment used a cycle ergometer and was administered by exercise physiologists using the YMCA protocol.30 Energy expenditure (VO2) was calculated from the last two workloads using the American College of Sports Medicine (ACSM) metabolic equations.30 Low and high fitness categories were defined using the median value at baseline for cardiorespiratory fitness (estimated VO2) and the median change in fitness level (VO2) was used to define a change in VO2.

Venous blood samples were drawn after a 12-hour overnight fast to determine the levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma glutamyl transpeptidase (GGT), total cholesterol, triglycerides, high- and low-density lipoproteins (HDL, LDL), ferritin, plasma glucose, and serum insulin. All samples were obtained before 10 AM. All biochemical tests were performed using a conventional automated analyzer at the Department of Pathology, Westmead Hospital.

To describe the effect of changes in PA, four categories were created: (1) those who increased PA (≥60 minutes/week); (2) those who maintained PA (above 150 minutes/week at baseline and 3 months but did not increase); (3) low-active (decreased to or remained between 60-150 minutes/week; and (4) sedentary (decreased to or remained at <60 minutes/week). To examine the dose-response relationship of PA those that increased their PA (≥60 minutes/week) were further divided into low increasers (60-119 minutes/week), moderate increasers (120-239 minutes/week), and high increasers (≥240 minutes/week). A categorical measure of “sufficient physical activity” was defined as five sessions and 150 minutes/week. This is the PA threshold to achieve a health benefit.31 Weight change was categorized as a reduction of ≥3% of initial body weight.32

Statistical Analysis.

Continuous variables were summarized as mean ± standard deviation (SD) and categorical data as percentages. SPSS software (version 16.0, Chicago, IL) was used for data analysis. Log-transformations were performed for variables not normally distributed. Analysis of variance (ANOVA) with Scheffé post-hoc contrasts were used to identify between-group differences. Comparison of the changes occurring within each group was undertaken using paired t-tests for all continuous variables. For analysis between groups, analysis of covariance (ANCOVA) was used. Logistic regression analysis was used to compare the prevalence of risk factors between groups at 3 months, with baseline prevalence as the covariate. To compare changes in categorical data within each group, McNemars test was used and Pearson's chi-square test was used to compare categorical data between groups.

Results

Baseline Characteristics

There were no significant differences at baseline in any anthropometric and metabolic variables or in PA (total minutes per week) between the groups (Table 1). Those who were “sufficiently” active at baseline, that is, participating in ≥150 minutes/week and ≥5 sessions/week of PA (36%), did have a significantly lower waist circumference, blood glucose, insulin, and HOMA scores than those who were not (data not shown). The retention rate for the study was high, with 96% (158/165) of participants attending the 3-month assessment.

Impact of the Lifestyle Intervention on PA Behavior

Figure 1 shows the changes in PA in the moderate-intensity and low-intensity intervention groups compared to controls. Sixty-four percent (P < 0.001) of those in the moderate-intensity and 63% (P < 0.001) in the low-intensity intervention group increased PA compared to 16% of those in the control group. Patients in the moderate and low-intensity intervention groups were more than nine times more likely to increase PA by ≥60 minutes/week compared to those in the control group (Moderate-Intensity: odds ratio [OR]: 9.4, 95% confidence interval [CI]: 3.30-22.34, P < 0.001; Low-Intensity: OR: 9.2, 95% CI: 2.99-28.42, P < 0.001). Conversely, the proportion of people undertaking less than 150 minutes/week of PA at 3 months was highest in the control group (50%), compared to 21% (P = 0.03) of those in the moderate-intensity intervention group and 24% (P = 0.016) of the low-intensity intervention group.

Figure 1.

Proportion of patients increasing, maintaining and decreasing physical activity in the intervention groups versus controls at 3 months. *P < 0.05 compared to control.

There was a significant increase in the number of patients participating in “sufficient” PA (≥150 minutes and five sessions per week) in both the moderate (+25%; P < 0.001) and low-intensity (+35%, P < 0.001) intervention groups compared to no change in the control group (−13%, P > 0.05). In addition, at 3 months the proportion of patients undertaking a sufficient level of PA was higher in both the moderate-intensity (63%; χ2 = 5.8, 1 df, P = 0.016) and the low-intensity (61%; χ2 = 5.1, 1 df, P = 0.024) intervention groups compared to the control group (38%).

From baseline to 3 months the proportion of participants undertaking less than 60 minutes/week of PA decreased significantly in both the moderate (30.0% to 12.5%, P = 0.004) and the low-intensity (37.0% to 13.0%, P = 0.007) intervention groups compared to no change (25.0% to 28.1%) in the control group.

Effects of Changes in PA and Fitness (VO2) on Liver Enzymes and Metabolic Parameters After Controlling for Weight Loss

Physical Activity.

We next examined the associations between PA and health outcomes, irrespective of intervention group allocation (Table 2). Changes in liver enzymes, insulin resistance, and metabolic risk factors are shown for those who increased PA by ≥60 minutes/week compared to those who remained sedentary, low-active, or maintained their levels of PA (but did not increase PA). People in the sedentary category had no improvement in liver enzymes or any other parameter, despite a significant reduction in weight over the 3 months. In contrast, those who were active at even a low level (60-149 minutes/week) showed significant improvements in liver enzymes and serum ferritin despite the absence of significant weight loss. Most important, the improvement in liver enzymes was greater in all PA categories compared to the sedentary group even after controlling for change in weight.

Table 2. Changes in Liver Enzymes and Metabolic Indices by Physical Activity Change Category at 3 Months (N = 158)
NPhysical Activity Change Category
No Increase in PAIncreased PA
SedentaryLow ActiveMaintainedIncreased
<60 min/wk60 -149 min/wk≥150 min/wk≥60 min/wk
25192985
  • Table reports on the results from subjects initially randomized to the intervention and control groups (n = 141), as well as results from those subjects who after 3 months in the control group were then randomized to an intervention group (n = 24); 7 subjects did not attend the 3-month assessment, resulting in a total of n = 158 for Table 2.

  • Data are means ± SD.

  • *

    P < 0.05 for the change from baseline to 3 months within each group using paired sample t-test.

  • **

    P < 0.001 for the change from baseline to 3 months within each group using paired sample t-test.

  • P < 0.05 versus Sedentary Category using ANCOVA adjusted for age, gender, and change in weight.

  • P< 0.05 versus both Sedentary and Low Active Categories using ANCOVA adjusted for age, gender, and change in weight.

  • Log transformations used for all variables with nonnormal distributions:

  • a

    subjects on injected insulin excluded from analysis;

  • b

    results from a 2-hour oral glucose tolerance test completed by 86 subjects at baseline and 3 months; Sedentary (n = 13), Low-Active (n = 9), Maintained (n = 18), Increased (n = 46).

 Weight (kg)−1.5 ± 4.4*−0.8 ± 2.8−2.1 ± 3.8*−2.4 ± 3.6**
 BMI (kg/m2)−0.6 ± 1.4*−0.2 ± 1.0−0.7 ± 1.3*−0.8 ± 1.3**
 Waist (cm)+0.4 ± 4.5−0.9 ± 4.0−1.0 ± 4.6−2.2 ± 4.2**
Liver Enzymes    
 Ferritin (μg/L)−8.1 ± 63.7−8.1 ± 44.6*−34.1 ± 67.0*−25.9 ± 81.5**
 ALT (U/L)−2.4 ± 29.2−14.5 ± 21.1*−19.6 ± 26.3**−14.8 ± 26.0**
 GGT (U/L)+1.0 ± 27.5−18.8 ± 25.1*−12.9 ± 15.2**−17.1 ± 36.3**
 AST (U/L)+2.2 ± 17.4−10.1 ± 18.9*−6.9 ± 12.3*−7.6 ± 17.8**
Metabolic Variables    
 FBGa (mmol/L)0.0 ± 0.8+0.2 ± 0.9−0.2 ± 0.4−0.3 ± 1.0*
 Fasting insulina(mmol/L)+0.7 ± 7.7−1.8 ± 7.5−1.5 ± 5.6−3.6 ± 7.9**
 2 hour glucoseab (mmol/L)+0.1 ± 2.6−0.2 ± 2.1−0.5 ± 1.8−0.4 ± 1.9
 2 hour insulina,b (mmol/L)+38.8 ± 142.6−8.8 ± 84.4−37.2 ± 164.0−46.9 ± 87.0**
 HOMA+0.2 ± 2.7−0.1 ± 2.5−0.3 ± 1.5−1.0 ± 3.3*
 Total cholesterol (mmol/L)+0.06 ± 0.42−0.09 ± 0.44−0.14 ± 0.67−0.30 ± 0.76**
 HDL (mmol/L)−0.03 ± 0.15+0.02 ± 0.130.00 ± 0.17−0.04 ± 0.17*
 LDL (mmol/L)+0.11 ± 0.46−0.12 ± 0.67−0.03 ± 0.54−0.18 ± 0.60*
 Triglyceride (mmol/L)−0.06 ± 0.79−0.12 ± 0.33−0.10 ± 0.60−0.23 ± 0.94*
Physical Activity and Fitness    
 Change in VO2 (ml/kg/min)−1.3 ± 2.0−0.2 ± 1.0+1.1 ± 2.3+2.0 ± 3.2**
 Total PA Baseline (min/wk)101.1 ± 1331.1163.4 ± 138.1549.0 ± 317.6136.3 ± 167.1
 Total PA 3 Months (min/wk)17.8 ± 18.287.9 ± 30.1340.7 ± 182.6393.3 ± 282.4
 ΔTotal PA (min/wk)−83.4 ± 131.1*−75.5 ± 143.8−208.3 ± 219.1**+257.0 ± 226.1**

The greatest improvement in liver enzymes occurred in those who either maintained PA at 150 minutes/week (and no increase) or those who increased PA by ≥60 minutes/week. In addition, those who increased their total PA by ≥60 minutes/week had a significant improvement in all metabolic variables at 3 months, except for 2-hour fasting glucose on a glucose tolerance test. Those who maintained PA (and did not increase) had improvements in weight, liver enzymes, and markers of insulin resistance. However, in contrast to those who increased their PA, in patients in the maintenance group there was no improvement in waist circumference, 2-hour fasting insulin, triglycerides, total, LDL, or HDL cholesterol levels.

Consistent with the self-reported PA data, fitness level (VO2) increased only in the group where PA was increased. The improvements in other metabolic variables in those who increased PA by ≥60 minutes/week remained significantly better than changes observed in those who remained sedentary, even after controlling for change in weight. Although there was significant mean weight loss in those who did not increase PA, the proportion of people losing ≥3% of initial body weight was highest in those who increased PA and least in those who did not increase PA and were undertaking less than 150 minutes/week at 3 months (P = 0.013) (Fig. 2).

Figure 2.

Proportion of patients losing ≥3% of initial body weight at 3 months by physical activity change category. *P < 0.05 compared to those with no increase and <150 minutes/week.

Fitness.

At baseline, subjects with a low fitness level had significantly higher weight, BMI, waist circumference, fasting blood glucose, fasting insulin, and HOMA-IR scores compared to subjects with a higher fitness level (Table 3). Irrespective of baseline fitness level, those who improved their fitness level (VO2) from baseline to 3 months had significantly greater reduction in weight than those who did not (Table 3). There was also a clear trend toward greater improvement in waist circumference, ferritin, HOMA-IR, 2-hour fasting insulin, and LDL cholesterol in those who increased fitness versus those who did not. There were significant increases in both total PA and walking minutes per week in those who increased fitness over 3 months compared to no change in those who did not have increases in fitness. Subjects with a low fitness level at baseline who improved their fitness level from baseline to 3 months showed the greatest improvement in liver enzymes and in other metabolic variables (Table 3).

Table 3. Change in Metabolic Variables by Change in Fitness (VO2) for Subjects Completing a Fitness Test at Baseline and 3 Months (N = 69)
 Low Fitness Baseline (n = 35)High Fitness Baseline (n = 34)
Baseline ValueIncreased VO2No Increase VO2Baseline ValueIncreased VO2No Increase VO2
N = 19N = 16N = 16N = 18
  • Data are means ± SD.

  • *

    P < 0.05 for the change from baseline to 3 months in each group using paired sample t-test.

  • **

    P < 0.001 for the change from baseline to 3 months in each group using paired sample t-test.

  • #

    P < 0.05 at baseline for low versus high fitness.

  • P < 0.05 for the change in the Increased VO2 versus No Increase in VO2 controlled for age and gender.

  • P < 0.05 for the change in the Increased VO2 group versus No Increase in VO2 controlled for age, gender, and change in weight.

  • Log transformations undertaken for all variables with nonnormal distributions:

  • a

    subjects on injected insulin excluded from analysis;

  • b

    results from a 2-hour oral glucose tolerance test completed by 43 subjects at baseline and 3 months.

Weight (kg)(90.6 ± 17.0)#−4.6 ± 4.4**−1.1 ± 1.9(85.2 ± 11.2)−3.9 ± 4.4*−1.1 ± 2.2*
BMI (kg/m2)(31.7 ± 4.6)#−1.6 ± 1.4*−0.3 ± 0.8(29.2 ± 2.5)−1.2 ± 1.5*−0.4 ± 0.6*
Waist (cm)(107.4 ± 12.6)#−3.8 ± 5.4*−1.1 ± 2.9(99.3 ± 8.6)−2.7 ± 3.5*−0.9 ± 2.8
Ferritin (μg/L)(211.2 ± 157.9)−47.9 ± 53.4**−30.1 ± 47.6*(242.7 ± 173.6)−49.2 ± 54.0**−5.4 ± 74.5
ALT (U/L)(62.6 ± 26.3)−22.5 ± 26.3**−12.4 ± 17.4*(70.34.0)−16.8 ± 28.2*−17.2 ± 26.8*
GGT (U/L)(62.6 ± 36.9)−21.6 ± 31.9**−1.9 ± 20.7*(62.4 ± 38.1)−5.7 ± 15.3*−10.8 ± 18.7*
AST (U/L)(41.4 ± 15.4)−8.4 ± 12.7*−4.3 ± 12.9(40.2 ± 14.6)−2.9 ± 8.4−7.0 ± 12.8*
FBGa (mmol/L)(6.2 ± 2.2)#−0.2 ± 0.6−0.2 ± 0.3*(5.2 ± 0.5)−0.2 ± 0.50.0 ± 0.3
Fasting Insulina (mmol/L)(20.9 ± 10.7)#−5.2 ± 7.8*−4.4 ± 4.8(15.3 ± 6.1)−2.2 ± 5.3+0.6 ± 4.7
2-hour glucosea,b (mmol/L)(7.6 ± 3.4)−1.0 ± 1.5+0.6 ± 2.1(6.6 ± 2.8)−0.8 ± 2.3−0.2 ± 1.1
2-hour insulina,b (mmol/L)(127.6 ± 73.7)−46.7 ± 51.1−2.4 ± 73.0(110.0 ± 60.5)−43.4 ± 55.6*−17.1 ± 47.6
HOMA(6.7 ± 5.4)#−2.4 ± 4.0*−1.3 ± 1.4*(3.6 ± 1.6)−0.6 ± 1.4+0.2 ± 1.1
Total cholesterol (mmol/L)(4.8 ± 0.8)−0.4 ± 0.5*+0.1 ± 0.6(4.9 ± 1.0)−0.3 ± 0.9−0.2 ± 0.5
LDL (mmol/L)(2.7 ± 0.8)−0.2 ± 0.60.0 ± 0.6(2.8 ± 0.8)−0.2 ± 0.6−0.0 ± 0.6
Triglyceride (mmol/L)(1.8 ± 0.9)−0.2 ± 0.6*+0.4 ± 0.9(1.8 ± 1.1)−0.4 ± 0.7*−0.1 ± 0.5
Total PA (min/wk)(183 ± 225)+200 ± 315*+75 ± 261(250 ± 327)+237 ± 292*+9.0 ± 263
Total walking (min/wk)(110 ± 125)+103 ± 106*+41 ± 126(77 ± 97)+107 ± 183*+13.0 ± 110
VO2 (ml/kg/min)      
 Baseline 15.8 ± 2.215.2 ± 2.5 20.9 ± 2.021.6 ± 2.3
 3 Months 19.7 ± 3.1**15.1 ± 2.8 23.2 ± 3.0**20.6 ± 2.4*

Dose-Response Relationships Between Increases in PA and Health Outcomes

We next examined the relationship between low, moderate, and high increases in reported PA (≥60 minutes/week) and metabolic parameters (Table 4). There was a positive relationship between increasing PA and weight loss but no clear relationship between PA increments and reductions in liver enzyme levels (Table 4). Thus, those who increased their levels of PA even a small amount (60-119 minutes/week) showed similar reductions in liver enzymes compared to those who increased PA by ≥4 hours. With respect to the other metabolic parameters, increasing PA by at least 2 hours/week resulted in decreases in waist circumference and improvements in the level of fasting insulin. Changes in fitness (VO2) reflected the changes in PA for each category, corroborating the self-report PA change data.

Table 4. Effect of Low Versus Moderate and High Increases in Physical Activity
NIncrease in Physical Activity
Low IncreaseModerate IncreaseHigh Increase
60-119 min/week120-239 min/week≥240min/week
252634
  • Data are means ± SD.

  • *

    P < 0.05 for the change from baseline to 3 months in each group using paired sample t-test.

  • **

    P < 0.001 for the change from baseline to 3 months within each group using paired sample t-test.

  • Log transformations undertaken for all variables with nonnormal distributions:

  • a

    subjects on injected insulin excluded from analysis;

  • b

    results from a 2-hour oral glucose tolerance test completed by 43 subjects at baseline and 3 months.

Weight (kg)−1.5 ± 3.9*−2.5 ± 2.6**−2.9 ± 4.1**
BMI (kg/m2)−0.5 ± 1.3*−0.9 ± 0.9**−1.0 ± 1.4**
Waist (cm)−1.1 ± 4.0−2.7 ± 3.5*−2.5 ± 4.8*
Ferritin (μg/L)−20.4 ± 55.4*−24.8 ± 57.1*−39.6 ± 56.1**
ALT (U/L)−20.8 ± 31.1*−11.6 ± 28.1*−12.9 ± 19.3**
GGT (U/L)−16.7 ± 30.2*−20.0 ± 45.2*−15.3 ± 33.9**
AST (U/L)−9.5 ± 11.4**−5.5 ± 21.3−7.68 ± 19.0*
FBGa (mmol/L)−0.2 ± 0.3*−0.1 ± 0.6−0.3 ± 0.6*
Fasting insulina (mmol/L)−0.8 ± 5.3−3.5 ± 7.2*−4.5 ± 6.8*
2-hour glucosea,b (mmol/L)−0.6 ± 1.7−0.6 ± 1.8−0.1 ± 2.0
2-hour insulina,b (mmol/L)−39.7 ± 41.5*−48.3 ± 36.9*−24.2 ± 44.2
HOMA−0.8 ± 2.7−0.8 ± 1.8−1.4 ± 2.3*
Total cholesterol (mmol/L)−0.40 ± 0.79*−0.18 ± 0.89−0.32 ± 0.64*
LDL (mmol/L)−0.30 ± 0.6*5−0.11 ± 0.56−0.15 ± 0.61
Triglyceride (mmol/L)−0.31 ± 0.72−0.14 ± 0.58−0.24 ± 1.27
Change in VO2 (ml/kg/min)+1.5 ± 3.9+1.9 ± 3.0*+2.4 ± 2.9*

Discussion

In this study, both moderate and low-intensity lifestyle interventions resulted in changes in PA behavior. Compliance with the exercise recommendations was determined using a Physical Activity Questionnaire (PAQ) and corroborated by a submaximal fitness test that indicated good correlation between the PAQ and changes in fitness. Patients who increased PA by an hour or more per week had the greatest improvements in metabolic and obesity-related variables, whereas those who remained or became sedentary had no improvement, and a clear trend toward deterioration in several metabolic parameters. Interestingly, improvement in liver enzymes was similar in both those who increased PA and those who maintained the recommended amount for health benefits, despite the latter group not achieving improvement in other metabolic indicators. Although “maintainers” were static with respect to total amount of PA, it is possible that variations in the type and frequency of PA may have contributed to the liver test improvements observed in this group. Consistent with this, we observed that of the group who maintained PA (and did not increase), 28% (8/29) changed the type of activity they were doing from moderate to vigorous without their total PA per week increasing.

Although greater increases in PA (>2 hours/week) appeared to be related to greater weight loss there was no additional benefit in terms of liver enzymes or glucose homeostasis. The reason for this is unclear, but we hypothesize that the threshold for change in liver enzymes may be low so that even a slight increase in PA is sufficient to improve liver tests and additional increases in PA may not translate to “dose-response” reductions in liver enzymes. Alternatively, the present study may not have been powered to detect a “dose-response” effect between PA and liver tests, given the known variability between individuals. The data presented here do clearly suggest that among patients with NAFLD even small increments in regular PA, even in the absence of large amounts of weight loss, can contribute to improvements in liver enzymes.

Patients with low fitness levels at baseline achieved significant metabolic health benefits with improved levels of fitness, despite their overall fitness level not even reaching the baseline level of the high fitness group (Table 3). Some of the changes (fasting insulin and HOMA) were greater than those achieved by the high fitness group. These findings support the hypothesis that even small gains in fitness and PA may have significant health benefits for patients with NAFLD.

At baseline, participation in sufficient levels of PA and levels of cardiorespiratory fitness in subjects in this study were lower than in the general population.31 The finding that patients with higher levels of fitness at baseline had lower levels of insulin resistance and waist circumference compared to those with lower fitness levels (Table 3) highlights the importance of PA for patients with NAFLD.

Despite being a relatively low-intensity level lifestyle intervention (compared to studies such as the Diabetes Prevention Project33), the proportion of subjects increasing their PA was typical of behaviorally based PA intervention studies.34 We found that increasing PA by an hour or more a week resulted in a significant reduction in waist circumference compared to remaining sedentary, even after controlling for weight loss (Table 2). In patients with NAFLD, waist circumference has been shown to correlate with insulin sensitivity and ALT levels.20 This may partly explain why sedentary patients in this study did not achieve reductions in insulin resistance or ALT as a group, despite significant weight loss (Table 2). An additional impact was the positive effect of increasing PA on reducing the levels of serum ferritin. A number of studies have demonstrated a close relationship between elevated levels of serum ferritin and insulin resistance,35–36 the metabolic syndrome,37 and T2DM.38 In the present study, those who remained or became sedentary over the 3 months had no change in ferritin levels. In all other groups, serum ferritin decreased significantly (Table 2).

The metabolic pathway by which PA improves insulin sensitivity may be different from that of weight loss39 and this is particularly relevant for patients with NAFLD, where insulin resistance is a pathological component of the condition.4 PA improves insulin resistance through positive changes in fatty acid metabolism in muscle, which cannot be achieved through energy restriction.40 Some small studies have indicated that PA may also have a direct effect on improving hepatic lipid metabolism and insulin sensitivity, independent of weight loss.23, 24, 39, 41 In addition, several other studies have indicated that there may be a relationship between PA levels, fitness, and NAFLD severity.42, 43

Study Strengths and Limitations.

The lack of objective measures of change in PA is a limitation of many PA interventions.13, 19, 33–34 This study used both a validated self-report measure of PA and an objective measure of fitness and the results were similar for either measure. Of note is the reduction in total PA from baseline to 3 months for those in the “maintained PA” category, which may be due to regression to the mean (Table 2).

The subjects in this study are representative of patients seen in hepatology clinics, suggesting that these findings might be generalizable to similar clinic settings. Another strength that supports the practical dissemination of this kind of intervention was the use of PA counseling, rather than structured exercise sessions. While providing structured, supervised exercise sessions is an effective way to assist patients in increasing PA, they are more expensive and often difficult to adopt in routine clinical practice. Ueno et al.11 demonstrated that exercise therapy and diet resulted in improvements in liver histology and biochemical indices of NAFLD; however, this involved structured exercise sessions, including hospital admission for the initial month of the intervention. We have shown here that PA counseling can result in significant increases in PA and fitness and subsequent improvement in health, without the need for supervised exercise sessions. This study expands on the findings of Ueno et al.11 in that it was conducted in free-living individuals and we have been able to importantly discern the independent effects of PA from those due to weight loss.

In conclusion, PA counseling as a component of a lifestyle intervention is an effective method of increasing levels of PA in patients with fatty liver disease and elevated metabolic risk factors. Improvements in PA can have positive effects on liver enzymes, insulin resistance, and other metabolic parameters in people with NAFLD. This is a particularly important outcome, as liver fat has been shown to correlate independently with all risk factors of the metabolic syndrome.44 Intensive and longer duration interventions may be efficacious, but are not practical at a population level. Our results also demonstrate that low-intensity, short-duration interventions can bring about significant metabolic improvements in people with this common form of chronic (liver) disease. Lastly, this study highlights the risk of remaining or becoming physically inactive (irrespective of whether weight loss is achieved) on metabolic parameters in patients with liver disease, and demonstrates the importance of PA for patients with NAFLD. Longer-term maintenance of PA change is essential for metabolic health, but will require further follow-up.

Acknowledgements

We thank Ms. Jennifer Green, Ms. Lauren McGrath, Ms. Lee Russell, and Ms. Rosemary Carney for assistance with this study.

Ancillary