Physical deconditioning is the common denominator in both obese and overweight subjects with nonalcoholic steatohepatitis

Summary

Background:

Obesity and decreased physical activity mirror increasing prevalence of nonalcoholic fatty liver disease (NAFLD).

Aim:

We aimed to investigate associations between aerobic fitness, anthropometrics and disease parameters in patients with nonalcoholic steatohepatitis (NASH). We hypothesised that NASH subjects have lower aerobic power and capacity than untrained, sedentary, normal subjects.

Methods:

Forty subjects (60% obese, 40% overweight) with biopsy-confirmed NASH and NAFLD activity score (NAS) ≥4 were enrolled in a clinical trial where anthropometrics, laboratories, liver fat content by MRI, activity, and aerobic fitness by cycle ergometry data were obtained.

Results:

NASH subjects were significantly deconditioned compared to 148 untrained, sedentary, healthy subjects from our laboratory in aerobic power (VO2peak) (NASH 16.8 ± 6.6 vs control 28.4 ± 10.6 mL/kg/min, P < 0.0001) and capacity (VO2 at lactate threshold [LT]) (NASH 8.3 ± 2.5 vs control 14.1 ± 5.9 mL/kg/min, P < 0.0001). NASH subjects’ fitness was comparable to the “least fit” tertile of controls: VO2peak [NASH 16.8 ± 6.6 vs “least fit” 17.3 ± 3.3, P = 0.64]) and VO2 at LT (NASH 8.3 ± 2.5 vs “least fit” 9.3 ± 2.1, P = 0.31). Fitness was similar in obese compared to overweight subjects (adjusted for gender) and was not correlated with visceral adiposity or NAS. Engaging in dedicated cardiovascular activity correlated with higher VO2peak and VO2peak at LT.

Conclusions:

Aerobic deconditioning was universally present in NASH subjects. NASH subjects’ fitness was similar to our laboratory’s “least fit” untrained, sedentary control subjects. Further research investigating NASH patients’ ability to improve low baseline aerobic fitness is warranted.

1 |. INTRODUCTION

As obesity has grown increasingly problematic in Westernised societies, nonalcoholic fatty liver disease (NAFLD) has become a condition that is increasing in prevalence,¹ with estimates that NAFLD affects up to 20% of the general population and up to 75% of obese patients^2–5 and is closely connected to insulin resistance. In the US, NAFLD is the most common explanation for abnormal liver enzymes among adults.⁶ Nonalcoholic steatohepatitis (NASH) represents a more severe form of NAFLD that leads to advanced fibrosis in up to 20% of patients.⁷

Because of the clear association between insulin resistance and physical inactivity, exercise has been repeatedly cited as a therapeutic option for NAFLD and NASH.^8,9 While weight loss has been shown to improve steatosis and aminotransferase abnormalities, studies to evaluate exercise as a function of fitness in its role in obesity and NAFLD have had mixed results, mainly limited by relatively small numbers of participants.^10–20 The experience thus far has been nicely summarised in a recent systematic review/meta-analysis, which also demonstrated a benefit for physical activity in reducing liver fat in NAFLD.²¹

In insulin-resistant conditions such as type II diabetes, mitochondrial dysfunction in skeletal muscle in the setting of hyperinsulinaemia or high plasma lipids plays a central role in mediating insulin resistance. Excess plasma free fatty acids leads to high levels of intramyocellular lipids which ultimately results in downregulation of genes for ATP and mitochondrial production as well as mitochondrial oxidative function, leading to decreased oxidative capacity. In addition, increased electron flux through the respiratory chain, particularly in the setting of lipid peroxidation and other mechanisms of lipotoxicity, may result in increased reactive oxygen and nitrogen species generation. These downstream effects on mitochondrial gene expression and respiration products lead to less efficient insulin signalling and worsened insulin sensitivity.²²

VO2peak and lactate threshold exercise testing are standard methodologies to measure aerobic fitness. VO2 (volume of oxygen consumed) is a long-accepted measure of fitness and endurance. VO2peak and the VO2 at lactate threshold are both measures of aerobic fitness. VO2peak is a strong indicator of fitness and provides the “ceiling” on fitness because it measures maximum oxygen consumption at a subject’s highest tolerated levels of exertion. The lactate threshold is an objective measure of lactate handling and tolerance during intense but submaximal exertion where aerobic capacity is reached and some anaerobic metabolism occurs during exercise. Lactate measurements are taken from the blood at preset time points during incremental exercise as VO2 is simultaneously measured. Lactate threshold testing helps to establish the exercise intensity a subject is likely to be able to maintain for a prolonged period without significant fatigue.²³

We evaluated the associations specifically between NASH, body-mass index (BMI) and parameters of aerobic fitness, including VO2peak and VO2 at the lactate threshold (LT), as standard measures of aerobic power and capacity. We hypothesised that these aerobic fitness measures would be lower in NASH subjects compared to untrained normal populations of similar age and gender. We also hypothesised that aerobic fitness would be poorer in obese NASH patients compared to overweight NASH patients despite comparable leisure time physical activity levels.

2 |. PATIENTS AND METHODS

This is a cross-sectional clinical study of baseline data collected from subjects who were patients ≥18 years with biopsy-proven NASH recruited between 2008 and 2011 to participate in a single-centre (University of Virginia) randomised, controlled, double-blind clinical trial evaluating the efficacy of n-3 fish oil to resolve NASH (ClinicalTrials.gov identifier: NCT00681408). Inclusion criteria included liver histology collected within 6 months of enrolment confirming presence of NASH (steatosis greater than or equal to 5%, hepatocyte ballooning, lobular inflammation), NASH Clinical Research Network (CRN) NASH activity score (NAS) of ≥4,²⁴ and a history of alcohol use of ≤30 g/d (men) and ≤20 g/d (women). Standard evaluation including laboratory and histological testing was completed to exclude patients with alternate aetiologies of chronic liver disease including viral hepatitis, autoimmune liver disease and metabolic liver disease such as haemochromatosis, Wilson disease and alpha-1 antitrypsin deficiency. Patients with decompensated cirrhosis or those with known risk factors for “secondary NASH” including jejunoileal bypass, total parenteral nutrition, previous bariatric surgery and medication or environmental toxin exposure were also excluded. Any patient with contraindications to safely undergoing intensive graded exercise bicycle testing (GXBT) including advanced cardiopulmonary disease or mobility limitations were not enrolled. While prescription of a beta-blocker was not a contraindication, no patients were receiving a beta-blocker medication at study entry. Our control group included a cohort of 148 otherwise healthy but untrained, sedentary subjects who performed <90 minutes per week of dedicated exercise. Subjects’ medical histories were queried specifically related to meeting no metabolic syndrome conditions. All testing was completed on-site at the University of Virginia. Anthropometric measurements were all completed in the General Clinical Research Center. Exercise testing was completed in the Exercise Physiology Laboratory to provide more complete metabolic/fitness profile of subjects. Hepatic fat content was obtained by MRI using volumetric assessments and Dixon technique and in- and out-of-phase imaging techniques performed at the Bioengineering Research Center. The University of Virginia IRB-HSR approved all parts of this study protocol, and all participants gave written informed consent.

2.1 |. Histological assessment of NASH

All patients underwent their initial liver biopsy as a part of their general hepatology care, and slides from all biopsies were reviewed in blinded fashion in random order by both a hepatologist and pathologist for confirmation of the diagnosis of NASH. The NASH activity score (NAS) was assessed as the sum of semiquantitative scores for steatosis, lobular inflammation and hepatocellular ballooning and fibrosis was also assessed per the NASH-CRN guidelines.²⁴ Presence of histological features of NASH and a composite NAS of 4 or greater were required for inclusion in this study. Biopsy cores were required to have length of ≥1.5 cm, width of ≥1.5 mm and at least 11 portal tracts to be deemed adequate for diagnostic evaluation. Mean biopsy length was 1.9 cm (±0.6 cm). All anthropometric, exercise testing/GXBT and MRI testing was completed within 6 months of the initial liver biopsy at their index clinical trial visit. Data regarding weight at the time of entry liver biopsy was incomplete-many patients were referred for clinical trial enrolment from community providers. In patients where this clinical data were available, there were no significant weight changes observed in the interval between entry biopsy and trial enrolment.

2.2 |. Demographic and laboratory data

Patient interviews were utilised to obtain all medical history. The National Cholesterol Education Program-Adult Treatment Panel III (NCEP-ATP III) criteria²⁵ were used to classify patients with metabolic syndrome. Typical anthropometric, biochemical and clinical assessments were obtained. Laboratory parameters included alanine aminotransferase (ALT), aspartate aminotransferase (AST), fasting blood glucose, fasting morning insulin and standard fasting lipid measurements including total cholesterol, LDL, HDL and triglycerides. Estimations of insulin resistance were calculated according to the original HOMA-IR model²⁶ and QUICKI-FFA model.²⁷ In addition, serial blood lactate levels were collected during maximal intensity exercise for estimation of LT and VO2peak at LT.

2.3 |. Physical conditioning and fitness measures

A symptom-limited continuous incremental cycle ergometer protocol was used to assess VO2 using standard open circuit spirometry and serum lactate and glucose drawn from an indwelling venous cannula (YSI analyser) at the end of each exercise stage in all patients. This modality of exercise testing was selected specifically because it is weight-independent. This was the first exercise testing event for each subject-no previous testing sessions had been administered. The test was terminated when the subject reached volitional exhaustion or was unable to maintain the desired pedal cadence. During the incremental protocols VO2peak was assessed as the highest oxygen uptake with maximal HR within 10 beats per minute of age-predicted HR max, a respiratory exchange ratio (RER) value >1.05, and/ or an Rate Perceived Exertion (RPE) score of >18 (6–20 Borg scale), which is routine in our exercise physiology laboratory.²⁷ In addition to testing of subjects in our laboratory, age-predicted VO2peak levels were estimated using formulas for untrained normal weight individuals per Bruce et al (VO2peak (men) = 57.8 – (0.445 × age); VO2peak (women) = 42.3 – (0.356 × age)).²⁸ VO2peak values are provided in both typical, standard units of mL O2/kg/min as well as L/min. Expressing VO2peak in L/min is more appropriate in bicycle ergometry given that it is a body weight-independent testing modality.²⁷ The peak power output and power output in watts at the LT were also measured. The LT was determined from the blood lactate-power output relationship function by standard methodology. The highest power output prior to the curvilinear increase in blood lactate was selected as the power output associated with the LT. A lactate elevation of at least 0.2 mM (the degree of error associated with the lactate analyser) was required for LT determination.²³

2.4 |. Abdominal/Visceral/Liver fat calculations

Abdominal, visceral and subcutaneous fat areas were measured in each subject using magnetic resonance imaging.^29–32 Liver fat was quantified by the 3-point Dixon and modified Dixon methods, both of which use in- and out-of-phase MRI imaging to detect hepatocyte lipid content as a result of chemical shift artefact. These methods were applied to both the whole liver as well as to an approximate liver biopsy site in the right lobe by drawing regions of interest for each set of baseline and study exit images.^33,34

2.5 |. Physical activity participation assessment

The time spent in physical activity at different intensities was assessed using the Aerobic Center Longitudinal Study’s Physical Activity Questionnaire administered using a standardised interview technique to increase accuracy.^35,36 Total physical activity was calculated as MET•H/Week (1 MET = 3.5 mL/kg/min), as recommended in the guidelines of the Compendium of Physical Activities.³⁷ METs or metabolic equivalents are best considered as an index of intensity of physical activity. 1 MET equals the amount of oxygen consumed per unit body over 1 minute while at rest, or, alternatively as energy expenditure of 1 kcal/kg * h. “Dedicated exercise” was activity completed specifically for improved health (ie excludes physical activity performed as an occupational or household duty). “Active” signified dedicated exercise of intensity of at least 3 METs for at least 150 minutes per week, and “sedentary” characterised patients who participated in no dedicated exercise beyond usual occupational or household duties.

2.6 |. Data analysis

Descriptive statistics including means and standard deviations were calculated for continuous variables, and frequencies and fractions were calculated for noncontinuous variables. Chi-square analysis and Wilcoxon’s test were used to test for differences among groups for nominal categorical and other nonparametric measures. Spearman’s correlation coefficient was used to estimate the association of ordinal categorical variables. Primary analysis used independent t tests to assess differences in NASH vs control subjects as well as in obese- and overweight-BMI patients. A significance level of P ≤ 0.05 was set for all statistical analyses.

3 |. RESULTS

3.1 |. Subjects

Forty patients were included in the original clinical trial study cohort (Table 1). 24 patients were obese by BMI ≥30, and the remainder were overweight (BMI ≥25). Four obese patients did not complete the LT exercise protocol (2 lost intravenous accesses during exercise testing and 2 of who did not achieve a true VO2peak due to arthritis-related joint-based symptoms during exercise). Four patients did not undergo MRI with fat quantitation due to severe claustrophobia (one obese patient), habitus too large to allow scanning (one obese patient), metal in orbit on scout plain films (one overweight patient) and inability to travel in a timely manner (one obese patient). One male patient was East Asian Indian ethnicity, and all other patients in the cohort were Caucasian. Nine patients were on stable doses of oral agents for diabetes for at least 3 months prior to enrolment (no patients were receiving insulin therapy), 22 patients were prescribed at least 1 agent for hypertension and 9 patients were prescribed at least 1 agent for hyperlipidaemia. 53% of our cohort met the National Cholesterol Education Panel-ATP-III criteria for metabolic syndrome.

TABLE 1.

Demographic/baseline characteristics by body mass index

	All (N = 40)	Obese (N = 24)	Overweight (N = 16)	P value
Age (y)	46.6 ± 12.2	45.9 ± 11.6	48.3 ± 13.4	0.69
Sex (M/F)	16/24	6/19	11/5	0.004
BMI (kg/m2)	32.5 ± 7.1	36.1 ± 7.1	27.1 ± 1.6	<0.0001
Waist/Hip Ratio	0.998 ± 0.076	1.011 ± 0.085	0.977 0.053	0.16
Comorbidities, N (%)
Hyperlipidaemia	27 (68)	18 (75)	9 (56)	0.30
Hypercholesterolaemia	14 (35)	11 (46)	3 (19)	0.10
Hypertriglyceridaemia	15 (38)	11 (46)	4 (25)	0.32
Type 2 diabetes	9 (23)	7 (29)	2 (13)	0.27
Metabolic syndrome	26 (65)	17 (65)	9 (56)	0.50
Laboratory test
Fasting glucose (mg/dL)	107.7 ± 18.0	107.3 ± 18.9	109.5 ± 16.7	0.88
Abnormal glucose, N (%)	24 (60)	15 (63)	9 (56)	0.74
Fasting insulin (mU/L)	22.8 ± 13.5	25.5 ± 12.6	18.9 ± 14.3	0.17
HOMA-IR	6.2 ± 4.3	6.8 ± 3.6	5.4 ± 5.1	0.36
AST (U/L)	56.8 ± 42.8	64.9 ± 51.8	44.6 ± 19.7	0.09
ALT (U/L)	74.2 ± 45.4	81.8 ± 48.1	62.9 ± 39.7	0.18
Steatosis (fat morphometry, %)	19.2 ± 8.5	19.6 ± 8.0	18.7 ± 9.3	0.76
Histology
Steatosis	2.0 ± 0.6	2.0 ± 0.5	1.9 ± 0.7	0.83
Lobular inflammation	2.0 ± 0.5	2.1 ± 0.5	1.8 ± 0.5	0.14
Ballooning	1.2 ± 0.6	1.2 ± 0.6	1.1 ± 0.5	0.79
NAFLD activity score	5.2 ± 0.9	5.3 ± 0.8	4.9 ± 0.9	0.10
Fibrosis	1.7 ± 1.0	1.8 ± 0.9	1.7 ± 1.2	0.54

Within the NASH subject group, obese and overweight subjects were similar regarding medical comorbidities, laboratory testing, insulin resistance and histological parameters associated with NASH. The gender composition between groups was not balanced with more females in the obese group (76%), and more males in the overweight group (69%).

All values are expressed as means ± standard deviation except where noted above.

P values quantify the differences in overweight versus obese subjects.

Comorbidities were defined by NCEP ATP-III guidelines.

Histology was scored according to the NASH-CRN scoring system for NASH activity score.

Steatosis (fat morphometry, %) refers to steatosis measured using computer-based fat measurements from photomicrographs of liver biopsies.

Group names: All: all NASH subjects. Obese: NASH subjects with BMI ≥30 kg/m². Overweight: NASH subjects with BMI 25–29.9 kg/m².

Demographic data regarding the subjects by BMI are also shown in Table 1. There were significant differences in the proportion of females between groups, but otherwise, the obese and overweight groups were comparable in terms of age, waist-hip ratio and incidence of hyperlipidaemia, hypercholesterolaemia, hypertriglyceridaemia, type II diabetes mellitus and metabolic syndrome. Aminotransferase levels, fasting glucose levels and histological parameters also were not statistically different between groups.

3.2 |. Physical conditioning and fitness: NASH vs sedentary, untrained controls and age- and gender-matched norms

Compared to our laboratory’s exercise testing of untrained, sedentary individuals of similar age range (Table 2), NASH patients had significantly lower fitness by VO2peak (16.8 T± 6.6 vs 28.4 ± 10.6 mL/kg/min, P < 0.0001) and VO2 at LT (8.3 ± 2.5 vs 14.1 ± 5.9 mL/kg/min, P < 0.0001). When using the weight-independent unit of L/min for VO2-related comparisons of NASH subjects to controls, VO2peak (1.54 ± 0.54 vs 2.35 ± 0.22 L/min, P < 0.0001) and VO2 and LT (0.76 ± 0.24 vs 1.17 ± 0.12, P < 0.0001) both continued to be significantly different. Results for NASH patients were comparable to but even lower than the “least fit” tertile of untrained control subjects for both VO2peak [16.8 ± 6.6 vs 17.3 ± 3.3, P = 0.64]) and VO2 at LT (8.3 ± 2.5 vs 9.3 ± 2.1, P = 0.31). Mean BMI of the least fit controls was 34.3 which is actually higher than the mean BMI of the NASH cohort (32.5).

TABLE 2.

Physical conditioning and fitness by body mass index and gender in NASH subjects

	All NASH N = 40	All obese N = 24	All overweight N = 16	All women N = 24	All men N = 16	Obese women N = 19	Overweight women N = 5	Obese men N = 5	Overweight men N = 11
VO2 peak (mL/kg/min)	16.8 ± 6.6	14.2 ± 4.7	20.3 ± 7.3a	13.1 ± 3.6	21.8 ± 6.3	12.6 ± 3.3d	14.8 ± 3.7	19.4 ± 4.4	22.8 ± 6.7
VO2 at LT (mL/kg/min)	8.3 ± 2.5	7.5 ± 2.3	9.3 ± 2.4b	7.1 ± 1.7	9.9 ± 2.4	6.8 ± 1.6e	8.2 ± 1.6	9.9 ± 2.3	9.8 ± 2.4
VO2 at LT/VO2 peak	0.51 ± 0.12	0.54 ± 0.13	0.49 ± 0.11	0.55 ± 0.12	0.47 ± 0.11	0.54 ± 0.12	0.56 ± 0.07	0.52 ± 0.11	0.45 ± 0.10
Blood lactate (mmol/L)	4.9 ± 1.8	4.7 ± 1.7	5.3 ± 1.9	4.1 ± 1.4	6.1 ± 1.6	4.3 ± 1.5	3.5 ± 0.7	6.1 ± 1.4	6.1 ± 1.6
Peak power (watts)	93.8 ± 38.1	80 ± 30.7	113.8 ± 39.7c	70.9 ± 20.6	126.9 ± 31.3	69.2 ± 18.6	77 ± 25.8	119 ± 30.9	130.5 ± 30.9

Taken as a whole, there were differences between obese and overweight groups; however, these differences appear to be mainly related to differences in gender makeups of the groups, as few differences persisted when obese and overweight subjects were compared according within gender-based sub-groups.

All values are expressed as means ±standard deviation. P values quantify the differences in overweight versus obese subjects.

^aBetween group P-value obese vs overweight = 0.007

^bBetween group P-value obese vs overweight = 0.03

^cBetween group P-value obese vs overweight = 0.008

^dBetween group P-value obese vs overweight = 0.007

^eBetween group P-value obese vs overweight <0.0001.

Group names: All NASH: all NASH subjects. All Obese: all NASH subjects with BMI ≥30 kg/m². All Overweight: all NASH subjects with BMI 25–29.9 kg/ m². All Women: all female NASH subjects. All Men: all male NASH subjects. Obese Women: all female NASH subjects with BMI ≥30 kg/m². Overweight Women: all female NASH subjects with BMI 25–29.9 kg/m². Obese Men: all male NASH subjects with BMI ≥30 kg/m². Overweight Men: all male NASH subjects with BMI 25–29.9 kg/m².

Exercise data was missing in 2 obese subjects (missing VO2 peak data with both tests severely limited by joint-related symptoms and thus did not reach a true VO2 peak) and 2 additional obese patients did not have lactate threshold data due to loss of intravenous access during exercise testing.

When compared with age- and gender-matched norms,²⁸ nearly all (91%) overweight and obese NASH patients demonstrated well below-average physical conditioning (≤10th %ile) as measured by VO2peak when compared with the percentile values for VO2peak in untrained men and women per age-norms after adjustment for mode of exercise. All female subjects had fitness <10th %ile (“well below average”) by this standard, and all but 3 male subjects (81%) had fitness <10th %ile, with the other 3 male subjects all testing “below average,” with CRF ≤30th %ile.

3.3 |. Physical conditioning and fitness: Overweight vs obese NASH

Fitness testing was generally well-tolerated with no adverse events. The median rating of perceived exertion at LT and at VO2peak for all subjects was 10 and 17 respectively (Borg scale of perceived exertion: 6 = no exertion, 20 = maximal exertion to exhaustion).³⁸ Conditioning and fitness data by BMI are shown in Table 3. All tests were stopped secondary to “maximal” or “near-maximal” perceived exertion by the participant or due to inability to maintain the predetermined pedalling cadence per standard exercise testing protocol. There were statistically significant differences in means of VO2peak (14.2 ± 4.7 vs 20.3 ± 7.3 mL/kg/min, P = 0.007), VO2 at LT (7.5 ± 2.3 vs 9.3 ± 2.4 mL/kg/min, P = 0.03) and peak power (80 ± 30.7 vs 113.8 ± 39.7 watts, P = 0.008) between obese and overweight NASH subjects, respectively, with higher mean values found in all measures in the overweight subjects.

TABLE 3.

Physical conditioning and fitness in NASH vs untrained, control subjects

	All NASH N = 40	All controls N = 148	“Least fit” controls N = 50	NASH women N = 24	Control women N = 74	NASH men N = 16	Control men N = 74
VO2peak (mL/kg/min)	16.8 ± 6.6	28.4 ± 10.6a	17.3 ± 3.3	13.1 ± 3.6	23.9 ± 9.4h	21.8 ± 6.3	32.9 ± 9.8l
VO2peak (L/min)	1538 ± 535	2345 ± 217b	1716 ± 76e	1178 ± 232	1943 ± 226i	1999 ± 459	2760 ± 149m
VO2 at LT (mL/kg/min)	8.3 ± 2.5	14.1 ± 5.9c	9.3 ± 2.1f	7.1 ± 1.7	12.7 ± 5.8j	9.9 ± 2.4	15.4 ± 5.7n
VO2 at LT (L/min)	758 ± 240	1165 ± 121d	923 ± 48g	630 ± 124	1032 ± 139k	919 ± 254	1292 ± 94o
RPE at LT	9.8 ± 2.2	10.4 ± 2.0	10.2 ± 2.0	9.7 ± 2.2	10.1 ± 2.1	9.9 ± 2.3	10.7 ± 1.9

Physical conditioning NASH subjects were comparable to the “least fit” (lowest tertile) subgroup of the controls, although NASH subjects demonstrated slightly lower oxygen utilisation once the measure was no longer expressed relative to weight (in L/min). When subjects’ genders were stratified, differences between NASH and control subjects were maintained. Perceived effort was similar across all groups.

All values are expressed as means ± standard deviation. Statistically significant differences are highlighted in bold.

VO2, volume of oxygen consumed; LT, lactate threshold; RPE, rating of perceived exertion, measured by Borg scale (6–20; 6: none, 20: maximal).

Group names: All NASH: all NASH subjects. All Controls: all control subjects. “Least fit” Controls: control subjects with VO2peak in lowest tertile of control subjects. NASH Women: all female NASH subjects. NASH Men: all male NASH subjects. Control Women: all female control subjects. Control Men: all male control subjects.

^aBetween group P-value all NASH vs all controls <0.0001.

^bBetween group P-value all NASH vs all controls <0.0001.

^cBetween group P-value all NASH vs all controls = 0.03.

^dBetween group P-value all NASH vs all controls = 0.04.

^eBetween group P-value all NASH vs “least fit” controls = 0.02.

^fBetween group P-value all NASH vs “least fit” controls = 0.04.

^gBetween group P-value all NASH vs “least fit” controls <0.0001.

^hBetween group P-value NASH vs control women <0.0001.

ⁱBetween group P-value NASH vs control women <0.0001.

^jBetween group P-value NASH vs control men = 0.0003.

^kBetween group P-value NASH vs control women <0.0001.

^lBetween group P-value NASH vs control men <0.0001.

^mBetween group P-value NASH vs control men <0.0001.

ⁿBetween group P-value NASH vs control men = 0.0003.

^oBetween group P-value NASH vs control men <0.0001.

When subjects were stratified by gender and BMI, there were no differences in VO2peak, VO2 at LT and peak power between obese and overweight NASH subjects in either gender, although the comparisons were hampered by uneven and small numbers per group. When subjects were stratified by activity level (active versus inactive/sedentary), or by level of non-HDL-cholesterol (≤130 mg/dL versus >130 mg/dL), there were no differences in aerobic conditioning between groups.

3.4 |. Subcutaneous and visceral fat by MRI: Overweight vs obese

Measurements of fat areas are presented in Table 3. There were significant differences in subcutaneous (P = 0.0002) and total fat area (P = 0.001) between the obese and overweight groups, with higher mean values for both parameters in the obese group as displayed in Table 4. Interestingly, visceral fat area was not significantly different between groups despite the differences in BMI (P = 0.31).

TABLE 4.

Fat quantitation from MRI by Dixon method and subcutaneous, visceral and total fat areas by body mass index

	All	Obese	Overweight	P value
% steatosis by Dixon method	14.8 ± 8.8	16.9 ± 9.6	12.3 ± 7.5	0.26
Subcutaneous fat area (cm2)	385.0 ± 157.2	457.2 ± 160.8	283.8 ± 79.5	0.0002
Visceral fat area (cm2)	221.3 ± 87.0	233.4 ± 94.2	204.4 ± 75.6	0.31
Total fat area (cm2)	636.4 ± 225.9	727.0 ± 243.6	509.5 ± 115.0	0.001
Total abdominal area (cm2)	872.6 ± 209.1	960.7 ± 201.7	749.2 ± 152.4	0.001

Differences in abdominal and total fat between obese and overweight subjects appear to be attributable to variability in subcutaneous fat, as visceral fat volumes and liver fat percentages were not significantly different between the obese and overweight groups.

All values are expressed as means ± standard deviation.

P values reflect the differences in overweight versus obese subjects.

Group names: All: all NASH subjects. Obese: NASH subjects with BMI ≥30 kg/m². Overweight: NASH subjects with BMI 25–29.9 kg/m^2.

Four patients were unable to undergo fat quantitation by magnetic resonance imaging with Dixon scoring (1 overweight patient had metal in orbit, 1 obese patient was limited by habitus, 1 obese patient was unable to reasonably arrange travel and 1 obese patient had claustrophobia).

3.5 |. Physical activity participation: Overweight vs obese

Current activity participation was comparable between overweight and obese groups, with no differences in proportion of sedentary subjects (P = 0.53) or means for total activity per week (P = 0.97), activity per week at any intensity (low (<3 METs), P = 0.37; moderate (3–6 METs), P = 0.51; vigorous (> 6 METs), P = 0.23) or amount of dedicated exercise per week (P = 0.2) as shown in Table 5. There was a difference in overall MET-minutes per week of dedicated exercise that approached significance (P = 0.06), favouring higher amounts of more intense dedicated exercise in the overweight subjects.

TABLE 5.

Physical activity participation by BMI

Activity	All	Obese	Overweight	P value
Active, N (%)	15 (38)	8 (33)	7 (44)	0.53
Inactive, N (%)	18 (45)	12 (50)	6 (38)	0.53
Sedentary, N (%)	7 (17)	4 (17)	3 (18)	0.54
Total physical activity (in min/wk [±SD])	3506 ± 1785	3514 ± 1901	3494 ± 1656	0.97
Low: activity of < 3 METs (in min/wk [±SD])	2763 ± 1648	2939 ± 1902	2498 ± 1181	0.37
Moderate: activity of 3–6 METs (in min/wk [±SD])	597 ± 660	535 ± 566	689 ± 791	0.51
Vigorous: activity of > 6 METs (in min/wk [±SD])	146 ± 549	39 ± 103	307 ± 850	0.23

Physical activity levels did not vary significantly between obese and overweight NASH subjects.

All values are expressed as means ± standard deviation.

SD, standard deviation.

P values quantify the differences in overweight versus obese subjects.

MET, metabolic unit, standard measure of exercise intensity. 1 MET is approximately equal to 3.5 mL O2/kg/min.

Activity assessed using the Aerobic Center Longitudinal Study’s Physical Activity Questionnaire.

Group names: Active: dedicated exercise of intensity of at least 3 METs for at least 150 min/wk. Inactive: some dedicated exercise of intensity of at least 3 METs but less than 150 min/wk.

Sedentary: no dedicated exercise beyond usual occupational or household duties.

3.6 |. Correlations of visceral fat, anthropometrics, physical activity, NAS score, fibrosis, VO2peak and LT: Overweight vs obese

Pearson correlations of fat amounts and fitness parameters were calculated and showed no correlation between VO2peak and subcutaneous or visceral fat areas. There was, however, a significant correlation between LT and subcutaneous fat area (r = −.35, P = 0.04), but there was no correlation between LT and visceral fat area (r = −.06, P = 0.72). As expected, VO2peak was correlated with the amount of dedicated exercise (r = .35, P = 0.03), whereas LT was only modestly correlated (r = .26, P = 0.12). Visceral fat area and waist circumference was significantly correlated (r = .37, P = 0.02).

4 |. DISCUSSION

Our cross-sectional study characterises physical activity levels, cardiorespiratory/aerobic fitness, body fat measurements by MRI imaging and histological data in a group of NASH subjects of variable BMI at baseline. Despite overweight subjects showing marginally better cardiorespiratory fitness as measured by VO2peak than obese subjects, we found that all subjects had severe deficits in baseline physical fitness compared to our laboratory’s untrained, sedentary control groups fitness as well as gender- and age-matched norms using VO2peak and VO2 at LT. Our NASH subjects had performance that was comparable to but lower than the “least fit” tertile subjects from our laboratory’s control group. Our NASH subjects also had significantly lower fitness compared by gender with our laboratory’s untrained, sedentary controls. There were statistically significant differences in VO2peak and LT between the overweight and obese NASH subjects despite similar activity levels and severity of NASH and fibrosis, although the groups had disproportionate membership by gender (many more women in the obese group, many more men in the overweight group). The level of deconditioning in NASH patients may negatively impact their ability to increase physical activity and lose weight, the 2 mainstays of recommendations for NASH patients, although we found that patients with higher amounts of weekly dedicated exercise tended to have higher VO2peak values. Whether this deficit is intrinsic in NASH patients and contributes to their risk for developing the condition or is a result of development of the NASH phenotype is not known. Our study was not designed to investigate the trainability of NASH subjects, although this would be an interesting investigation.

VO2peak has been demonstrated to inversely vary with existence of the metabolic syndrome³⁹ and NAFLD¹⁸ in both men and women. In our cohort of NASH patients, mean VO2peak and LT by bicycle ergometer testing for all patients (16.8 mL/kg/min ± 6.6 and 8.3 mL/kg/min ± 2.5), overweight patients (20.3 mL/kg/min ± 7.3 and 9.3 mL/kg/min ± 2.4) and obese patients (14.2 mL/kg/ min ± 4.7 and 7.5 mL/kg/min T 2.3) were far below untrained, sedentary controls⁴⁰ from our laboratory as well as age- and gender-predicted norms. Our results for VO2peak and muscle power are consistent with but lower than either Krasnoff et al²⁰ or Church et al⁴¹ Our results also are consistent with previous exercise studies from our laboratory in subjects with metabolic syndrome,⁴² so we feel confident that these results reflect accurate assessments of performance for this cohort. We did not collect information regarding effects of training or weight loss on fitness as this survey was preliminary and derived from data collected from a primary, randomised, controlled, medication trial.

Visceral fat was markedly elevated in all subjects, with normative values for Caucasians falling under 100 cm².⁴³ Increased visceral fat was present relatively uniformly across both groups and may have led to the lack of correlation of VO2peak or LT with visceral fat. 100% of our cohort is overweight with 60% being obese. Visceral fat has a clear bearing on insulin resistance and has been implicated in increased cardiovascular risk.⁴⁴ The lack of significant difference in visceral fat area in the overweight versus obese NASH patients fits the theory that increased visceral fat is a hallmark of NASH regardless of BMI. We also show that the amount of visceral fat is fairly linearly related to insulin resistance, regardless of BMI (Figure S1).

Although there were measurable differences in physical activity testing, the self-reported physical activity habits of our NASH cohort demonstrated no significant differences between obese and overweight patients. Nearly 40% of our cohort engaged in dedicated cardiovascular exercise of moderate intensity (at least 3 METs) for at least 150 minutes per week, which implies that, the majority of our NASH patients’ activity levels fall below the CDC’s Guidelines for fitness.⁴⁵ The proportion of patients who engaged in no dedicated cardiovascular activity was higher than the proportion who were active in the overall cohort (45% vs 38%) as well as in the obese group (50% vs 33%). The overweight NASH patients surpassed the obese NASH patients in moderate (3–6 METs) and vigorous activity (>6 METs), although the findings were not statistically significant, most likely due to either wide variability in activity levels in both groups, recall bias or self-reporting bias.

Our study has several limitations. It is cross-sectional and does not reflect the dynamic state of energy homeostasis. There were no normal weight individuals in our NASH cohort, thus reducing generalisability of our results to the low proportion of NASH patients with normal BMI. Despite including only patients with histologically diagnosed NASH, there were no trends or relationships between visceral fat, fitness and histological parameters. The extremely poor physical conditioning in both groups of NASH subjects is unlikely to be related to differences in comorbidities, as no significant differences in occurrence of different medical illnesses were observed. Our cohort size is comparable to those included in most investigator-initiated NASH treatment trials. There was a difference in gender makeup in the overweight versus the obese groups, so it is possible that the differences in groups may reflect gender differences rather than those related to BMI, although the differences were not different in within-gender comparisons. The differences between groups were quite small, and adjusting the measurements for gender did not alter the relationship of BMI to fitness level. In addition, our cohort was not ethnically diverse, thus it does not adequately characterise outcomes for ethnicities aside from Caucasian given that 39 of 40 subjects were Caucasian. There are inherent limitations to the physical activity questionnaire due to the self-reporting nature of the instrument. It is possible that inability to blind patients to their diagnosis of NASH may have affected their effort in exercise testing, but we identified no literature to support or refute this phenomenon.

Despite VO2peak being a longtime standard of reporting fitness testing results, the use of a weight-based measure such as VO2peak may be perceived as a limitation, and several studies regarding weight- and body fat-related effects on exercise capacity support this assertion,^46–48 with the most popular assertion being that if baseline respiratory capacity is limited even at BMIs near 30 kg/m², then this population’s ability to compensate with moderate-to-high intensity exercise will be even further below expected levels.⁴⁸ In contrast, there have been several studies that demonstrate the trainability of high-BMI and even high-body fat individuals to VO2peak values that match average fitness levels for age- and gender-based standards.^49,50 Clearly there remains a great deal of controversy on the degree to which VO2peak represents actual cardiorespiratory exercise capacity in overweight and obese patients. Our results in using either unit of measure of VO2-mL/kg/min or L/min-had similar results. Further work in exercise science to evaluate measures of physical conditioning in overweight and obese subjects and possibly clarify an optimal measure would help to address this important question.

Our study shows that NASH patients have substantially decreased cardiorespiratory fitness, well-below average compared to our laboratory’s historical control group and published age- and gender-adjusted norms. Obese NASH patients have mildly lower levels of fitness compared to overweight NASH patients, but there was no correlation of fitness levels to fibrosis or severity of NASH as assessed by NAS score. NASH patients, regardless of BMI, also have lower than recommended participation in physical activity, and elevated amounts of visceral fat compared to adjusted norms. Because of the severity of deconditioning, fitness parameters such as VO2peak and LT may be useful predictors or evaluation tools for patients’ response to an exercise intervention, both in research and in typical clinical scenarios, and they appear feasible in chronic liver disease patients.⁵¹ Future research regarding the effects of exercise in NASH, trainability of NASH patients with exercise programs, and the applicability of VO2peak and other classic measures of exercise capacity in this population should include objective evaluation of cardiorespiratory fitness, fat/body composition,⁵² and histological criteria to ensure that interventions and assessment design are standardised to give clinicians and patients useful targets in attempting to alter the course of NASH through exercise.