Exercise and physical activity in cirrhosis: opportunities or perils

Abstract

Reduced exercise capacity and impaired physical performance are observed in nearly all patients with liver cirrhosis. Physical activity and exercise are physiological anabolic stimuli that can reverse dysregulated protein homeostasis or proteostasis and potentially increase muscle mass and contractile function in healthy subjects. Cirrhosis is a state of anabolic resistance, and unlike the beneficial responses to exercise reported in physiological states, there are few systematic studies evaluating the response to exercise in cirrhosis. Hyperammonemia is a mediator of the liver-muscle axis with net skeletal muscle ammonia uptake in cirrhosis causing signaling perturbations, mitochondrial dysfunction with decreased ATP content, modifications of contractile proteins, and impaired ribosomal function, all of which contribute to anabolic resistance in cirrhosis and have the potential to impair the beneficial responses to exercise. English language-publications in peer-reviewed journals that specifically evaluated the impact of exercise in cirrhosis were reviewed. Most studies evaluated responses to endurance exercise, and readouts included peak or maximum oxygen utilization, grip strength, and functional capacity. Endurance exercise for up to 12 wk is clinically tolerated in well-compensated cirrhosis. Data on the safety of resistance exercise are conflicting. Nutritional supplements enhance the benefits of exercise in healthy subjects but have not been evaluated in cirrhosis. Whether the beneficial physiological responses with endurance exercise and increase in muscle mass with resistance exercise that occur in healthy subjects also occur in cirrhotics is not known. Specific organ-system responses, changes in body composition, or improved long-term clinical outcomes with exercise in cirrhosis need evaluation.

INTRODUCTION

Sarcopenia is a progressive loss of muscle mass and strength that contributes to adverse outcomes, including increased mortality and decreased quality of life (43, 233). Primary sarcopenia, or sarcopenia without underlying disease, is due to aging, whereas secondary sarcopenia is that which occurs in patients with chronic disease including advanced liver, heart, lung, and renal disease (43). Both primary and secondary sarcopenia have been shown to be states of anabolic resistance or attenuated muscle response to nutrients and physical activity (100, 167). Of the various disorders associated with secondary sarcopenia, there is considerable literature on the clinical consequences of sarcopenia in cirrhosis, or end-stage liver disease, but there are limited data on the underlying mechanisms of sarcopenia (46, 48, 50). We critically evaluated the current literature on the responses to exercise in cirrhosis and identify areas of unmet need.

Cirrhosis is a pathological consequence of injury to the liver due to a number of causal factors and is characterized by hepatocyte loss, fibrosis of the hepatic lobule with loss of hepatic lobular architecture and vascular perturbations that result in elevated pressure in the portal venous system and shunting of blood across the liver, and regenerative activity that may progress to neoplasm or hepatocellular carcinoma. The most common causes of cirrhosis in the United States include nonalcoholic fatty liver disease (NAFLD), hepatitis C or B virus infection, alcohol abuse, cholestasis, and metabolic/inherited disease (206). Both NAFLD and alcohol-related liver disease are rapidly becoming the most frequent causes of cirrhosis (99, 200). The etiology of liver disease (especially alcohol related) affects the severity of muscle loss (237), but whether exercise capacity or responses vary by the cause of cirrhosis is not known. Of the complications of cirrhosis, sarcopenia, or skeletal muscle loss, is the most frequent.

Dysregulated protein homeostasis (proteostasis) in cirrhosis contributes to loss of muscle mass and may also impair contractile function and limit exercise capacity (46, 47, 50, 155). A number of metabolic and molecular perturbations in cirrhosis result in dysregulated signaling pathways and an increased rate of gluconeogenesis, lipolysis, and proteolysis, which induce anabolic resistance and consequent loss of muscle mass (46–48, 50). Impaired skeletal muscle mitochondrial oxygen consumption and response of components of the electron transport chain substrates and inhibitors, as well as reduced ATP content, modifications of contractile proteins, and altered substrate metabolism, have been reported and may contribute to decreased muscle contractility in cirrhosis (47, 48, 50, 51, 155). However, whether impaired skeletal muscle contractile function and decreased ATP production affect the response to exercise training in cirrhosis is not known. Additionally, there is a lack of uniformity in interpretation of patient-reported symptoms of weakness and fatigue that contribute to their limited functional capacity, mobility, and response to exercise. This heterogeneity limits the interpretation of studies on muscle contractile strength and the response to exercise in cirrhosis (40).

Systematic reviews of nutritional supplementation have shown modest to no benefit in patients with liver disease, consistent with the interpretation that cirrhosis is a state of anabolic resistance (74, 176). Whether anabolic resistance extends to physical activity or exercise in cirrhosis has not been evaluated. In contrast to the extensive literature evaluating nutritional supplementation in cirrhotics, there have been few studies evaluating the impact of physical activity and exercise on muscle mass or contractile function in this cohort (46, 251). The present work will provide an overview of the physiological responses to exercise and the potential maladaptive responses that may occur in cirrhosis. We critically reviewed published studies that evaluate the impact of exercise on skeletal muscle in patients with liver disease and potential strategies to optimize the response to anabolic stimuli in cirrhosis.

Response to Exercise

Much of our understanding of the physiological, metabolic, molecular, and functional responses to exercise is based on studies in healthy subjects, athletes, and the aged population. Resistance and endurance exercise are the two best studied forms of structured physical activity (8, 191, 214). The terms “exercise” and “training” have been used interchangeably in the literature (and will be used interchangeably throughout this review based on the terms used in the original studies) even though exercise is an episode of physical activity that aims to enhance health, whereas training is a longer-term program of physical activity with a goal of improved physical performance (36). Resistance training is low-repetition exercise against heavy resistance (such as weight lifting) that results in stimulation of muscle protein synthesis and an increased muscle mass and strength with little or no increase in maximum oxygen uptake capacity. In contrast, endurance training is low-resistance, frequent repetition exercise (including swimming and cycling) that results in an increase in muscle mitochondrial mass, maximum oxygen uptake, and the ability to perform prolonged work with lesser impact on body composition than resistance exercise. Concurrent training refers to a combination of varying duration and intensity of both resistance and endurance exercise performed multiple times as part of a longer-term regimen (8). Unlike the robust literature on resistance and endurance training, there are emerging data on the effects of concurrent training on skeletal muscle outcomes (8, 41, 81, 87, 172, 196). Published studies suggest that during concurrent training, the benefits of endurance training may be enhanced but the increases in muscle mass and strength due to resistance training are diminished (called interference or concurrence training effect) (8, 41, 108, 161). The mechanistic basis of responses to either endurance or resistance training have been well studied in healthy subjects, but mechanistic understanding of the response to concurrent training is only recently being evaluated systematically (8, 41, 81, 161, 171, 172). An overview of the metabolic, molecular, subcellular, physiological, functional, and phenotypic responses to exercise and training in healthy subjects is shown in Table 1 (21, 60, 71, 109, 138, 157, 164, 182, 185, 201, 202, 204, 211, 224).

Table 1.

Physiological responses to exercise in healthy subjects

Effect	Endurance Training	Resistance Training
Metabolic response
Lactate production	No change in lactate; detraining caused a decrease in lactate dehydrogenase	Increased lactate (breaks between sessions minimize lactate production)
Physiological response
V̇o2max	Increased V̇o2max, anerobic threshold, workload, and endurance	Small to no increase in V̇o2max (may be related to increased muscle mass), increased anerobic threshold
Blood vessels	Increased vascularization after training (26–29%); increased capillary density and contacts, with myofibers independent of muscle hypertrophy	No change in capillary contacts; however, capillaries around the fibers increased (33%) independent of hypertrophy
Blood flow	Lowered blood pressure, decreased splanchnic blood flow	Increased portal pressure
Molecular response
Signaling/regulatory proteins	AMPKα2 is activated, resulting in the alterations of downstream pathways/targets	Increased mTORC1 response resulting in increased muscle protein synthesis
Genetic factors	Proteolytic genes increased, myostatin decreased	Proteolytic enzymes had a varied response during exercise but returned to normal after exercise, myostatin decreased, and cytokines including TNFα, IL-6, Il-8, and IL-15 increased
Functional and phenotypic response
Mitochondria	Increased number of mitochondria and oxidative function	Increased mitochondria content and quality, increased muscle size, unchanged or decreased mitochondrial volume
Body mass	Significantly increased lean body mass and decreased fat mass	Increased lean mass and deceased fat mass, although contradictory data have also been reported
Muscle strength	Modest increase in muscle strength and power	Robust increase in muscle strength and power
Fiber type	Increased fast-twitch muscle fibers, glycogen content, and oxidative capacity; potential fiber-type switch to oxidative fibers	Decreased type IIb fibers, muscle fiber area increased

AMPK, AMP-activated kinase; mTORC1, mammalian target of rapamycin complex 1.

Clinical Relevance

In cirrhosis, there is a reduction in muscle mass, impaired contractile strength, and decreased oxygen uptake, all of which are consistently associated with adverse outcomes including decreased survival, reduced quality of life, and worsened post-liver transplant outcomes (46, 50). Limited published data suggest that an increase in muscle mass in cirrhosis is associated with improved survival (228). However, whether improvement in contractile strength and maximum oxygen uptake following endurance exercise impacts survival or long-term clinical outcomes in cirrhosis is currently not known. Therefore, identifying the potential muscle-specific beneficial responses to exercise in cirrhosis (e.g., improved muscle mass, contractile strength, oxygen consumption) is of high clinical relevance.

Data Extraction Strategy

A literature search on PubMed was conducted for papers in the English language with the search words including “cirrhosis,” “exercise,” and “sarcopenia” (Fig. 1). In an initial PubMed search, there were 3,500 publications relating to cirrhosis and sarcopenia (n = 188) or cirrhosis and exercise (n = 3,305). Publications that included only patients with noncirrhotic liver disease were not included because our goal was to evaluate the impact of exercise in cirrhosis and even though exercise may be beneficial in patients with liver disease without cirrhosis (nonalcoholic fatty liver disease, alcoholic hepatitis, viral hepatitis), the heterogeneity of the clinical conditions precludes robust conclusions from being drawn. There were nine reviews that were evaluated to ensure that all relevant original publications were included in this review. The remaining publications were categorized by articles that addressed safety of exercise in cirrhosis and response to exercise in cirrhosis as their primary or secondary outcomes. The majority of the studies included a small number of patients, so no exclusion criteria were defined for subject counts. Additionally, we did not require that healthy controls be used as the active comparator. Not all of the investigators (110) used the same measurements to quantify response to exercise or comparable definitions of impaired exercise capacity. Of the 36 studies evaluated, 21 publications evaluated the responses to exercise (Table 2) (5, 58, 73, 83, 84, 110, 118, 128, 135, 144, 145, 165, 181, 193, 194, 208, 221, 222, 239, 245, 246), whereas 15 publications also evaluated the safety metrics (Table 3) (12, 13, 15, 16, 56, 65, 72, 82, 88, 114, 134, 149, 199, 209, 251). Because of the heterogeneity of study design, patient populations, and outcome measures, a meta-analysis of the data could not be done; instead, a critical evaluation of the published data is presented.

Fig. 1.

Consort statement of search output from PUBMED on exercise and cirrhosis in humans. HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MESH, medical subject headings; NASH, nonalcoholic steatohepatitis.

Table 2.

Physiological responses to exercise as the primary outcome in patients with cirrhosis

Author and Year	Patient Number	Study Protocol	Primary Outcomes	Secondary Outcomes	Results	Comments
Kennedy et al., 1977 (135)	1 AC	Case report	Effect of exercise in pulmonary function	None	Hypoxia and hypocapnia exist in AC at rest; hypoxia becomes more severe with endurance exercise, suggesting a shunt-like mechanism	Measurements obtained at rest in supine, sitting, and standing positions and during 20 min of exercise
Almenoff et al., 1989 (5)	7 LC, 7 HC	Case control	Compare lactate metabolism in LC patients vs. HC	None	LC had a more rapid increase in arterial lactate vs. HC (34.8 LC to 14.1 min in HC), 3× lower lactate clearance in LC vs. HC	Cycle V̇o2max test, starting at 25 W with increments increasing by 25 W every 2 min until AT; lactate levels were measured during and after exercise
Petrides et al., 1997 (181)	9 LC, 9 HC	Case control	Examine the effects of a single exercise session on glucose metabolism	None	Glucose utilization (mg·kg−1·min−1) changed from 2.42 to 2.25 in HC and 2.41 to 2.55 in LC after exercise; during clamp, HC changed from 11.92 to 11.49 in HC and 4.93 to 4.73 in LC; glucose oxidation (mg·kg−1·min−1) changed from 1.35 to 1.22 in HC and 1.28 to 0.66 in LC after exercise; and during glucose clamp changed from 3.34 to 3.28 in HC vs. 3.03 to 3.06 in LC; lipid oxidation (mg·kg−1·min−1) in HC changed from 0.93 to 0.91 and 1.23 to 1.41 in LC during exercise; and during clamp, HC changed from 0.31 to 0.28 and 0.42 to 0.40 in LC patients	Both groups underwent 30 min of treadmill exercise with increasing work intervals to 60%; glucose clamp study performed
Tarter et al., 1997 (221)	49 AC, 42 NAC, 50 HC	Case control	Compare isokinetic muscle strength in patients with AC vs. NAC and HC	None	AC had muscle weakness in concentric and eccentric movements, suggesting a need for physical rehabilitation in this population	Testing performed to determine isokinetic muscle strength
Wiesinger et al., 2001 (239)	19 LC	Cohort	Determine physical fitness by isokinetic muscle strength and AT	None	54% (Child-Pugh A), 36% (Child-Pugh B), and 31% (Child-Pugh C) reached predicted AT; isokinetic muscle strength of the quadriceps femoris was 149 Nm (Child-Pugh A), 108 Nm (Child-Pugh B), and 89 Nm (Child-Pugh C)	Cycle exercise until AT was reached; thigh muscle and grip strength were measured
Jacobsen et al., 2001 (118)	14 LC, 6 HC	Case control	Determine if maximal rate of ATP synthesis in skeletal muscle is decreased in LC	None	Decreased exercise capacity shown by decreased intracellular Mg2+ (0.733 mM in HC, 0.594 mM in Child Class A, and 0.385 mM in Child Class B-C) and decreased maximal ATP synthesis rate (HC: 0.425 mmol·L−1·s−1, Child Class A: 0.402 mmol·L−1·s−1, and Child Class B-C: 0.189 mmol·L−1·s−1)	Exercise protocol with pNMR testing during the 2 min of rest, 3 min of exercise, and 14 min of recovery
Wong et al., 2001 (245)	39 LC, 12 HC	Case control	Investigate the cardiac link between low V̇o2 and AT in patients with LC	None	Predicted values of work performance were achieved as follows: 88% in HC, 71% in pre-ascites, and 46% in LC + ascites; cardiac index was 7.43 L·min−1·m−2 in HC, 4.81 L·min−1·m−2 in pre-ascites, and 4.30 L·min−1·m−2 in LC + ascites; LC ± ascites have a lower V̇o2 and AT than HC	Cardiac testing and cycle exercise with increments equivalent to 50 kilopond meters per 30 s until exhaustion
Shiraishi et al.,2005 (208)	5 AC, 9 HC	Case control	Investigate abnormal energy metabolism of skeletal muscle in ALD	None	LC had decreased AT compared with HC and required a much lower amount of exercise to achieve AT	Patients exercised, then AT and respiration compensation point were measured
Mori et al., 2007 (165)	6 LC, 15 HC	Case control	Evaluate level of myocardial blood flow	None	In the LC group, MIBI uptake and % uptake (mean: 52.5 ± 5.8%) were not lower than those of the HC group	Exercise consisted of a single exercise treadmill test
Terziyski et al., 2008 (222)	19 LC, 19 HC	Case control	Investigate exercise performance and ventilatory efficiency in patients with LC as compared with HC	Compare exercise performance, ventilatory efficiency, and Child-Pugh score	Exercise capacity (23.8 vs. 30.6 mL·min−1·kg−1 in LC vs. HC), ventilatory efficiency (30.4 vs. 26.3 in LC vs. HC), and oxygen uptake efficiency [2,187 vs. 2,745 mL/min per log (10) in LC vs. HC] were lower in LC vs. HC; decreased ventilatory efficiency and exercise capacity correlate with increasing Child-Pugh score	Patients were compared in a maximal exercise test on a treadmill; LC patients performed worse than HC, and those with worse disease also had worse outcomes
Dharancy et al., 2008 (58)	135 LC	Prospective	Reported V̇o2max in LC	Prognostic value of V̇o2max, influence of peak V̇o2 on hospitalization	LC had significantly lower V̇o2max vs. predicted levels (61.2%), lower V̇o2max correlated with lower AT (0.74 vs. 0.94 mL/min in patients with decreased V̇o2max), patients with a higher MELD score and lower V̇o2max (low V̇o2max, mean MELD score 16 vs. normal V̇o2max, mean MELD score 13) had lower survival rates and increased hospitalizations	Pre-OLT subjects cycled in increasing increments to fatigue
Lemyze et al., 2010 (144)	20 LC	Prospective	Effect of OLT on exercise capacity in LC	None	Peak V̇o2 increased by a mean of only 7.7% (from 63.4 to 71.1% of predicted value) and decreased in one-quarter of the patients following OLT	Aerobic capacity and respiratory indicators measured before and after OLT
Galant et al., 2011 (84)	26 LC	Cross-sectional study	Evaluate quality of life and compare it to aerobic capacity and RMS	None	V̇o2max correlates with MIP and MELD	LC measured for V̇o2max and RMS in a treadmill exercise test (5 3-min stages)
Lemyze et al., 2011 (145)	30 LC +widened P(A-a)O2 vs. 30 LC +normal P(A-a)O2	Cohort	Comparison of response to exercise	None	Patients with decreased pulmonary gas exchange by widened P(A-a)O2 (mean 44.9 mmHg.) vs. those with normal P(A-a)O2 (20.8 mmHg) had lower mean V̇o2max (53 vs. 61.9% of predicted) and lower maximal workload (87.3 vs. 100.4 W), respectively	Cycle exercise
Wu et al., 2012 (246)	40 LC	Correlation design	Explore the correlation between fatigue and physical activity in patients with LC	None	Correlation between liver disease and fatigue but no correlation between disease severity or laboratory results and fatigue	No exercise, questionnaires only
Galant et al., 2013 (83)	27 AC	Prospective study	Establish correlation with V̇o2max, MELD, and survival over 3 yr	None	Inverse correlation between V̇o2max and MELD (r = −0.91); individuals who had a V̇o2max ≤ 14 mL/kg showed 60% mortality vs. V̇o2max ≥ 14 mL/kg	V̇o2max test performed and patients were followed up for 3 yr
Kaibori et al., 2013 (128)	51 HCC	Randomized nonblinded	Exercise therapy in patients with HCC who underwent hepatectomy	None	Whole body mass (95% of baseline in exercise vs. 100% in diet) and fat mass decreased in exercise and diet group (86% of baseline in exercise vs. 97% in diet); patients who exercised often had improved HOMA-IR (10.7 μU/mL and 5.8 μU/mL high frequency) and peak V̇o2 (103 standard exercise to 118% of baseline high frequency exercise) as well as AT V̇o2 (102% standard exercise to 115% of baseline to high frequency)	2 groups of patients: diet and exercise vs. diet only; exercise included stretching and walking for 1 h 3 days/wk; patients exercised 1 mo before surgery and then for 6 mo after surgery
Roman et al., 2014 (194)	17 LC patients	Randomized nonblinded pilot	Exercise capacity, muscle mass, HRQoL	None	Patients in exercise group showed improvement in 6-MWT (365 to 445 m), moderate increase in BMI (26.7 to 27.0), low thigh circumference (41 to 46 cm), and improved patient perceived HRQoL; unaltered values in no exercise group	Exercise vs. no exercise, both groups given BCAA; exercise group performed moderate exercise 3 day/wk for 1 h for 12 wk, exercise was treadmill or cycling
Faustini-Pereira et al., 2015 (73)	178 LC (92 with HPS and 86 without)	Cross-sectional study	Determine the impact of HPS on exercise capacity, functional condition, and RMS	None	Patients with HPS had low V̇o2peak (14.2 vs. 17.6 mL/kg), lower 6-MWT (340.8 vs. 416.5 m), lower MIP (−49.1 vs. −74.2 cmH20), and lower maximum expiratory pressure (60.1 vs. 76.8 cmH20)	V̇o2max measured during a 15-min exercise test
Roman et al., 2016 (193)	23 LC patients	Randomized nonblinded	Functional capacity by CPET, BC, anthropometry, and DEXA	Risk of falls	Exercise group showed an increase in effort time (2 min), AT (0.5 min), and thigh circumference (4.25 cm)	Group 1: exercised either by treadmill or cycle for 1 h, 3 days/wk for 12 wk; group 2: no exercise
Hiraoka et al., 2017 (110)	33 LC	Randomized nonblinded	Determine the effects of BCAA and walking exercise on BC and muscle strength	None	Serum ammonia and HbA1c were not altered; BCAA/tyrosine ratio improved (4.3 ± 1.4 to 5.2 ± 2 interquartile range); increases in daily steps, muscle leg strength ratio (1.0 to 1.013) and leg strength ratio (1.0 to 1.1), and hand-grip strength ratio (1.00 to 1.06)	LC patients were prescribed 2,000 additional steps and BCAA as a late evening snack for 3 mo

6-MWT, 6-min walk test; AC, alcoholic cirrhosis; ALD, alcoholic liver disease; AT, aerobic threshold; BC, body composition; BCAAs, branched-chain amino acids; BMI, body mass index (kg/m²); HbA1c, hemoglobin A1c; CPET, cardiopulmonary exercise test; DEXA, dual-energy X-ray absorptiometry; HBF, hepatic blood flow; HC, healthy control; HOMA-IR, homeostasis model of assessment for insulin resistance; HPS, hepatopulmonary syndrome; HRQoL, health-related quality of life; LC, liver cirrhosis; MELD, model for end-stage liver disease; MIBI, myocardial perfusion imaging; MIP, maximal inspiratory pressure; NAC, N-acetyl cysteine; Nm, Newton meter; OLT, orthotopic liver transplant; P(A-a)O₂, alveolar arterial oxygen pressure gradient; pNMR, p nuclear magnetic resonance spectroscopy; RMS, respiratory muscle strength; V̇o₂, oxygen consumption; V̇o_2max, maximum oxygen consumption.

Table 3.

Physiological responses and safety of exercise in patients with cirrhosis

Author and Year	Patient No.	Study Protocol	Primary Outcomes	Secondary Outcomes	Results	Comments
Sinniah et al., 1970 (209)	15 HC, 15 LC	Case control	Determine effects of 20, 40, and 80% power forearm exercise on ammonia concentrations	None	LC patients had higher levels of ammonia compared with HC at 20% (139.8 vs. 63.2 μg/100 mL), at 40% (101.2 to 75.6 μg/100 mL), and at 80% power (104.6 to 93.2 μg/100 mL). After exercise, ammonia concentration increased earlier and stayed elevated for longer (after 30 min, LC had ammonia concentration of 110 vs. 76 μg/100 mL)	Patients were asked to do forearm exercise with a hand grip at 20, 40, and 80% power with rest for 30 min between each exercise
Kelbaek et al., 1987 (134)	29 AC, 29 HC	Case control	Elucidate alterations in hemodynamics during exercise	None	AC subjects had lower cardiac ejection fraction; workload was decreased (122 vs. 186 in men and 60 vs. 119 W in women), RHR increased (91 vs. 78 bpm), and MHR decreased (159 vs. 170 bpm)	Cycle exercise until exhaustion with workload increase of 30–50 W per 3 min; workload was defined by highest level completed during the last 3 min of exercise
DeLissio et al., 1991 (56)	4 HC, 4 LC	Case control	Determine effects of exercise on glucose metabolism in HC vs. LC	Tolerance of exercise in HC vs. LC	After exercise, endogenous glucose production increased by 2.5-fold in HC but not in LC; forearm muscle glucose extraction was 3.6% in HC LC vs. 0.55% in LC, suggesting that LC patients have increased fat oxidation	Exercise for 90 min on a treadmill at 50% V̇o2max
Garcia-Pagan et al., 1996 (88)	8 LC	Randomized nonblinded	Effects of hemodynamic and humoral changes caused by exercise influence, portal, and systemic hemodynamics in LC	None	Portal hypertension increased at 30% (HVPG increased from 16.7 to 19.2 mmHg, HBF decreased from 1,291 to 1,034 mL/min) and at 50% of peak performance (HVPG from 16.7 to 19.9 mmHg, HBF from 1,291 to 900 mL/min); concluded that moderate exercise is not safe for patients with LC and portal hypertension	8- to 10-min exercise capacity tests (cycling), performed at 30% and 50% of peak performance
Salo et al., 1997 (199)	21 nonazotemic patients with LC and ascites	Cohort study	Determine effects of moderate exercise on renal function	None	Moderate exercise impairs renal function, GFR (77 vs. 67 mL/min), renal plasma flow (555 vs. 442 mL/min), and free water clearance (5.8 vs 4.5 mL/min); all values at rest vs. after exercise	30 min of moderate cycling at a workload equal to 3 METs
Bandi et al., 1998 (12)	23 LC patients	Double-blinded (propranolol or placebo administration)	Determine effects of moderate physical exercise on hepatic hemodynamics and collateral blood flow in LC patients with portal hypertension	Whether propranolol therapy may prevent worsening of portal hypertension during exercise	Exercise increased HVPG by 14%, HBF decreased by 18% in the placebo group, patients on propranolol had decreased HVPG by 23% and HBF by 26%, and AZBF did not change with exercise; CO (17.6 to 11.1 L/min), HR (119 vs. 100 bpm), and mean arterial pressure (126 vs. 113 mmHg) were lower in the propranolol group in response to exercise; AEs not experienced in either group	8–12 min of monitored cycle exercise in a single session
Epstein et al., 2004 (65)	156 LC patients (59 of whom underwent OLT)	Cohort study	Investigate the link between low V̇o2 and AT on post-OLT outcomes	None	Patients with a pretransplant V̇o2max <60% of predicted V̇o2max who underwent OLT had a higher 100-day posttransplant mortality, if AT <50%, patients with V̇o2max < 60% of predicted: 25% died (4 of 16), patients with V̇o2max< 60% predicted and V̇o2-AT < 50%: 36% died (4 of 11)	Determined V̇o2max with a cycle exercise working at increasing rate of 10–15 W/min; subject monitored for 100 days post-OLT
Bay et al., 2005 (13)	8 LC	Cross-sectional study	Does the ability to reduce HS blood flow during exercise affect CO	None	With exercise HR increased from 68 bpm to 142 bpm, CO increased from 5.1 to 12.9 L/min, mean arterial pressure from 89 to 115 mmHg, HS blood flow declined from 0.97 to 0.62 L/min, cerebral oxygenation by NIRS from 61% to 72%	Patients performed cycle exercise at 60 rpm to achieve a HR of 90–95 bpm in the first round and 115–120 bpm in the second round, the final round HR was 140 bpm
Hollingsworth et al., 2010 (114)	16 PBC, 16 HC; 4 PBC patients in a separate study	Case control	Compare bioenergetics effects of exercise on LC vs. HC	Explore feasibility, acceptability and effects of exercise on PBC patients	pH recovery after initial exercise was prolonged in PBC compared with HC (160 vs. 25 s) with the longest recovery time in fatigued patients (210 s); all patients tolerated and completed 4 wk of exercise training	Subjects performed 3 × 3-min exercise at 35% maximum voluntary capacity
Galant et al., 2011 (84)	27 LC patients	Case report	Influence of V̇o2 on pre-OLT patients for 3 yr	None	An inverse relationship between V̇o2max and MELD score, individuals with a V̇o2max ≤14 mL/kg showed 60% mortality when compared with those who had a V̇o2max >14 mL/kg	V̇o2max and MELD score were assessed and patients were followed for 3 yr
Bernal et al. 2014 (15)	399 pre-OLT LC	Retrospective observational	Characterize risk factors for decreased AT	Review risk factors associated with decreased AT	Patients with decreased AT had prolonged hospitalization after OLT, and nonsurvivors had lower AT values than survivors 1 yr after transplantation; 176 patients did not undergo OLT and had 1-yr mortality rate of 34.6%; AT and peak V̇o2 were lower in patients who died	Pre-OLT LC patients performed cycling exercise; V̇o2max was reached
Zenith et al., 2014 (251)	36 LC screened, 19 completed study	Randomized prospective pilot	Change in V̇o2max from baseline	Quadriceps muscle thickness, thigh circumference, 6-MWT, QoL, and safety	8 wk of exercise increased V̇o2 by 5.3 mL·kg−1·min−1 in exercise group, thigh circumference (52.4 to 53.6 cm) and quadriceps thickness and QoL improved, and 6-MWT improved (529.1 to 570.5 meters) in exercise group	Investigated 8 wk of supervised treadmill exercise program for 3 days/wk for 8 wk at 60–80% of baseline peak V̇o2 vs. no exercise
Faustini Pereira et al., 2016 (72)	86 LC patients	Prospective cohort	Identify predictors for outcomes in LC	None	6-MWT distance < 410 m, 55% survival vs. 6-MWT distance > 410 m, 97% survival, MIPs < 70 cmH2O, 62% survival vs. >70 cmH2O, 93% survival	Patients were followed for 3 yr
Macias-Rodriguez et al., 2016 (149)	22 LC patients	Open-label pilot	Effect of PEP in HVPG and QoL	Changes in physical fitness, nutritional status, ammonia levels, and safety	HVPG decreased with exercise (−2.5 mmHg) and increased in the no-exercise controls (4 mmHg), with a significant between-groups difference (P = 0.009), there was improvement in ventilatory efficiency in exercise (−1.9) vs. no-exercise controls (−0.4)	Group 1: no exercise controls who received nutritional supplements. Group 2: exercise with nutritional supplements + exercise, 14-wk follow-up in both groups; PEP included both types of exercise (40 min of cycling exercise and 30 min of kinesiotherapy)
Berzigotti et al., 2017 (16)	50 obese LC patients	Cohort	Explore the effects of diet and exercise on obese LC patients	None	Hepatic pressure (13.9 to 12.3 mmHg), leptin (29.7 to 20.3 ng/mL), and insulin (912 to 670 pg/mL) were decreased; fat mass decreased (35.1 to 30.2 kg) with no change to lean mass (56.7 to 55.6 kg) after 4 mo of exercise	A weekly 60-min moderate exercise program with 40 min of aerobic and strength training for 4 mo

6-MWT, 6-min walk test; AC, alcoholic cirrhosis; AEs, adverse events; AT, anerobic threshold; AZBF, azygous blood flow; BC, body composition; BCAAs, branched-chain amino acids; bpm, beats per minute; CO, cardiac output; GFR, glomerular filtration rate; HC, healthy controls; HCC, hepatocellular carcinoma; HR, heart rate; HRQoL, health-related quality of life; HS, hepatosplanchnic; HVPG, hepatic-venous pressure gradient; LC, liver cirrhosis; MELD, model for end-stage liver disease; MET, metabolic equivalent; MHR, maximal heart rate; MIP, mean inspiratory pressure; NIRS, near-infrared spectroscopy; OLT, orthotopic liver transplant; PBC, primary biliary cholangitis; PEP, physical exercise program; RHR, resting heart rate; rpm, rotations per minute.

ANALYSIS AND INTERPRETATION OF PUBLISHED LITERATURE

Exercise Capacity and Functional Responses in Cirrhosis

Exercise capacity is reduced in cirrhosis and may be due to factors directly related to liver disease that include reduced ventilatory capacity (125, 222), decreased inspiratory pressure (84), an impaired V̇o_2max, or indirect factors including reduction in oxygen delivery to the muscle due to anemia and/or alterations in regional blood flow. Skeletal muscle blood flow, measured by forearm blood flow, is not altered at rest but increases in response to exercise in cirrhosis (13, 148). Comorbidities or cardiopulmonary dysfunction in cirrhosis can also contribute to impaired exercise capacity and response. V̇o_2max correlates with inspiratory pressure (84) and 6-min walk test (73) and inversely correlates with the model for end-stage liver disease (MELD) score, a measure of the severity of the underlying liver disease (84). Oxygen consumption above which aerobic energy production is supplemented by anaerobic pathways with increased lactate generation and metabolic acidosis is termed “anaerobic threshold” (AT) (235). In addition to decreased V̇o_2max, AT was also reduced in cirrhotics compared with healthy controls or predicted values (31, 58, 73, 144–146, 208, 245). Reduced AT results in sustained elevation in lactate generation and decreased hepatic lactate clearance with delayed return to baseline in cirrhosis during exercise. This combination of increased muscle lactate generation, possibly due to impaired mitochondrial oxidative function, in conjunction with impaired lactate clearance due to hepatocyte dysfunction lowers plasma pH during exercise (56). The increase in lactate and decrease in pH have adverse metabolic and functional consequences that reduce exercise capacity in cirrhosis (173).

Impaired exercise capacity has been attributed to greater fatigue in patients with cirrhosis. Notably, recent data show that caregivers of patients with cirrhosis notice that, despite recommendations to exercise, these patients display reduced physical activity and complain of increased fatigue (40). Fatigue is considered to be due to either central effects from a feeling of tiredness that is generally perceived only at rest or peripheral effects usually due to impaired contractile function and occurs either during or shortly after physical activity (155, 177). A reduction in AT with increased lactate and lowered pH has been suggested to contribute to central fatigue and the consequent impaired exercise capacity in cirrhosis (5, 246). Another potential mechanism of fatigue in cirrhosis is the lower tissue and circulating magnesium content with a consequent decrease in ATP synthesis (118). This is consistent with recent data that hyperammonemia of cirrhosis lowers muscle phosphocreatine content and mitochondrial oxidative function (51, 155). Hyperammonemia causes posttranslational modifications, including nitration of muscle proteins that can impair their function and contribute to decreased exercise capacity and fatigue (155). Even though the reduced muscle and coronary blood flow in cirrhosis may contribute to lower exercise capacity and altered biochemical responses, decreased blood flow has not been consistently observed (135, 165). In addition to fatigue, muscle strength in cirrhotics is also decreased (221). These data suggest that skeletal muscle contractile dysfunction is important because reduced muscle strength in cirrhosis and the resultant perception of fatigue can also impair exercise capacity.

Supervised endurance exercise ranging from 8 wk to 6 mo has been reported to improve metabolic and functional parameters in cirrhotics. AT, work effort (193), 6-min walk test (194), and V̇o_2max increase in cirrhotics after endurance exercise (199, 251). Measures of functional capacity (higher AT, V̇o_2max, and 6-min walk test) are associated with better survival and quality of life and reduced hospital stays, but whether improvement in these measures following exercise training in cirrhosis results in better clinical outcomes is not known (15, 58, 65, 72, 73, 216). Multiple studies report that V̇o_2max predicts survival with an accuracy similar to MELD scores in cirrhosis (58, 64, 65, 83, 84, 146). Even though cirrhotics have low V̇o_2max, those with a V̇o_2max < 14 mL/kg had a 60% greater mortality rate than those with a V̇o_2max ≥ 14 mL/kg. Higher V̇o_2max before transplantation in cirrhotics is associated with better survival while waiting for a liver transplant and also after transplantation (15). Patients with a V̇o_2max lower than 60% of predicted rate had decreased survival after liver transplantation compared with those who had a V̇o_2max greater than 60% of predicted (65). Because higher V̇o_2max is associated with greater survival and better quality of life, an increase in V̇o_2max is expected to translate into improved clinical outcomes, but a direct link between a structured exercise program and long-term survival in pretransplant cirrhotics has not been reported. Several groups have shown that improvement in V̇o_2max posttransplant is associated with improved survival (240). In healthy patients, V̇o_2max begins to decline when endurance exercise is stopped (169, 174). Whether this rate of decline is accelerated, and what the consequences of a reduction in V̇o_2max upon discontinuation of an exercise program in cirrhosis are, also remain unknown.

Both endurance and resistance exercise (to a lesser extent) result in an increase in V̇o_2max and a decrease in fat mass in healthy subjects. Similarly, in addition to an increase in V̇o_2max, changes in measures of body composition [increase in body mass index (BMI), lean muscle mass, and thigh circumference and a decrease in fat mass] have been reported in cirrhotic patients following 8–12 wk of endurance exercise (193, 194, 251). DeLissio et al. (56) reported an increase in fat oxidation during and after endurance exercise in cirrhotics. Loss of fat mass can potentially aggravate the accelerated starvation in cirrhosis, and the proteolysis-dependent gluconeogenesis may worsen muscle loss and aggravate hyperammonemia. Studies on metabolic and substrate utilization in cirrhosis during and after exercise are necessary to determine if the beneficial responses to exercise are translated into improved muscle mass, contractile function, and overall survival.

Pulmonary vascular complications, including hepatopulmonary and portopulmonary syndromes that occur in 4–30% of cirrhotics depending on the stage and severity of disease, can alter the pulmonary gas exchange and lower exercise capacity and V̇o_2max even further (73). Interpretation of exercise capacity necessarily requires both screening for hepatopulmonary syndrome and recognizing a lower V̇o_2max in these patients. Anemia due to blood loss and/or due to chronic disease also contributes to impaired cardiopulmonary responses and exercise capacity in cirrhosis (64, 215).

Physiological and Hemodynamic Responses to Exercise in Cirrhosis

Exercise increases blood flow to the active muscle and reduces blood flow to splanchnic organs, including the liver and kidneys (104, 127, 132, 136, 187, 197). Splanchnic blood flow and pressure decrease during acute endurance exercise, but resistance exercise increases portal pressure (104, 139, 180, 188). Neither endurance nor resistance exercise alter renal function in healthy subjects (187), but whether the hemodynamic responses are similar in patients with cirrhosis who frequently have impaired renal and hepatic function is not known (231).

Cardiac output and heart rate increase with both endurance (105) and resistance exercise (130) in healthy controls and with endurance exercise in patients with cirrhosis (13). Although cardiac responses are similar to those in controls, cirrhotics have lower hepatic blood flow and increased hepatic venous pressure gradient (HVPG) accompanying the increased heart rate in response to an exercise test (12, 88, 165, 219). The increase in HVPG and decrease in hepatic blood flow responses in cirrhosis are reported at 30% of peak workload and are more pronounced at 50% of peak workload (88). Pre-exercise administration of propranolol, a nonselective β-blocker used in cirrhotic patients to lower portal pressure and reduce the risk of variceal bleeding, decreased HVPG from 19.3 to 12.9 mmHg as compared with an increase in HVPG (16.9 to 19.0 mmHg) with placebo in response to endurance exercise in cirrhosis (12). Others have also reported lower HVPG during endurance exercise while on propranolol (149). Thus, sustained endurance exercise without propranolol causes an adverse splanchnic hemodynamic response (increase in HVPG and a decrease in hepatic blood flow) in cirrhotic patients, suggesting the need for supervised exercise, especially in those with portal hypertension. Sustained increases in HVPG can result in rupture of varices and worsen ascites and hepatic encephalopathy in cirrhotics with deleterious consequences (217). However, nonselective β-blockers can reduce exercise tolerance and capacity, with increased subjective fatigue (ratings of perceived exertion) that can limit exercise training in cirrhosis (69, 92, 163). Thus, propranolol may help lower the adverse splanchnic hemodynamics during endurance exercise in cirrhosis at the cost of reduced exercise tolerance. Splanchnic vasoconstriction, altered gut motility in cirrhosis, protein losing enteropathy due to portal hypertension, and anemia of chronic disease and that due to splenomegaly (79) can significantly contribute to impaired nutrient availability and adaptive responses to exercise that may contribute to deleterious effects during and after exercise.

Cirrhotic patients are also at risk for renal failure due to splanchnic and renal vasoconstriction (23) and peripheral vasodilation (249). An increase in hepatic sinusoidal pressure results in further reduction in renal arterial flow, aggravating renal dysfunction in cirrhotics (119). Although renal function does not change in healthy subjects following endurance exercise (40), impaired renal hemodynamics in cirrhotics may be aggravated by the hemodynamic alterations during and after endurance exercise. Renal plasma flow, glomerular filtration rate (GFR), and the activity of the renin-aldosterone and sympathetic nervous systems, or neurohumoral activity at rest and after moderate endurance exercise for 30 min in 21 cirrhotic patients were evaluated (199). In 10 patients, mean GFR (76 to 49 mL/min) and mean renal plasma flow (535 to 354 ml/min) decreased and plasma norepinephrine (576 to 925 pg./mL) and mean plasma renin (4.7 ng/mL to 7.1 ng/mL) concentrations increased, before and after exercise, respectively. In the remaining 11 patients, no significant change in GFR, a modest reduction in mean renal perfusion (573 to 522 mL/min), and increases in mean plasma renin (1.4 ng/mL to 2.6 ng/mL) and plasma norepinephrine (288 to 630 pg/mL) were noted before and after endurance exercise (199). After endurance exercise, cirrhotic patients with ascites and activation of the renin-angiotensin system (RAS) had further worsening of renal function (199). Because neurohumoral activity is not measured routinely, impaired response to diuretics may be used as a surrogate to determine the renal safety of exercise in cirrhosis. Moderate endurance exercise may therefore be safe in cirrhotics who have ascites but respond to diuretic treatment or in those without ascites. Because marked activation of the RAS is associated with poor response to diuretics, renal dysfunction may worsen in cirrhotics with ascites who are nonresponders to diuretics. Safety of hemodynamic responses in different organs to exercise in cirrhosis needs to be assessed before an exercise program is recommended.

Muscle Fiber Responses to Exercise

In healthy subjects, the effects of endurance training are believed to be primarily on the slow-twitch (type 1) fibers with an increase in number, diameter, and, consequently, overall muscle mass (76, 203, 241). Whether there is hypertrophy of these mitochondria-rich, oxidative fibers or if there is a fiber-type switch from the fast, glycolytic type II fibers to type I fibers is not yet certain (241, 248). Even though there are limited skeletal muscle histological data in human cirrhosis, in preclinical models, loss of type IIb fibers and reduction in muscle fiber diameter have been reported (89, 141). Therefore, resistance exercise may benefit cirrhotics, but anabolic resistance may limit the anticipated responses.

Metabolic Responses to Exercise in Cirrhosis

Adaptive responses to glucose, protein, and fat metabolism as well as changes in body composition occur during exercise training in healthy subjects (191, 214), but whether these responses are altered in cirrhosis is currently not well known, given the critical contribution of hepatic metabolism to exercise (225). Although endurance exercise is beneficial in patients with metabolic syndrome and obesity, whether similar benefits occur in cirrhosis, a state of accelerated starvation with an increased fat metabolism, is not certain even though these benefits have been reported in cirrhotics with nonalcoholic fatty liver disease (NAFLD) (17, 91). Muscle responses to endurance exercise in nonobese cirrhotics need to be evaluated to ensure that further increases in fat metabolism do not increase muscle amino acid catabolism. Resistance exercise also results in reduction of fat mass, but to a lesser extent compared with endurance exercise (210). However, increase in muscle mass is greater with resistance exercise compared with endurance exercise (8, 39, 98, 214).

Few studies have evaluated metabolic responses and consequences of exercise in cirrhosis (12, 54, 55, 128, 140, 175, 193, 194, 220, 251). During exercise, the rapid demand for ATP results in anaerobic glycolysis and depletion of glucose stores, which increases utilization of amino acids (derived from skeletal muscle protein stores) as a substrate for oxidation (20, 166, 191, 223). After recovery from exercise, muscle protein mass is restored in healthy subjects (13). However, impaired protein synthesis in cirrhosis and anabolic resistance (46, 50, 227, 229) may contribute to suboptimal recovery of muscle protein mass in cirrhotic patients. Both endurance and resistance exercise improve peripheral glucose uptake and insulin sensitivity in healthy subjects (85, 107, 181, 234). Patients with cirrhosis have underlying insulin resistance (37, 131) that may also improve with exercise (endurance). An improvement in insulin resistance occurs with exercise in patients with NAFLD, but there is limited data in cirrhotic patients (14, 101, 178, 181). Exercise induces an increase in muscle lactate generation, and lactate disposal is impaired in cirrhosis (93, 121). The consequent accumulation of lactate in cirrhotics (34) may contribute to exaggerated fatigue. Additionally, lactate causes perturbations in signaling pathways that may impair the beneficial responses to exercise (173, 183, 218, 252). Unlike glucose and amino acid oxidation in healthy subjects, data on fatty acid oxidation are less consistent. An increase in non-plasma-derived fatty acid oxidation and differences between trained and untrained subjects have been reported (115, 120, 122, 243). The increasing prevalence of NAFLD has resulted in increasing interest in fatty acid metabolism in the liver, but once cirrhosis develops, whether these alterations are modified is not known (56, 91, 208). During exercise in healthy subjects, the increased energy demand promotes lipolysis, and fatty acid concentrations increase in circulation with a decline during the recovery period (78, 94, 120, 243). Because cirrhosis is an insulin-resistant state with accelerated fatty acid oxidation at rest, it is not known whether a response similar to that of healthy patients occurs. A greater increase in amino acid oxidation rather than fatty acid oxidation, especially during exercise, may contribute to greater loss of muscle mass in cirrhotics, but this is not known.

Net ammonia generation by the contracting skeletal muscle occurs due to a combination of purine breakdown and transamination of amino acids as a source of energy (9, 10, 26, 27, 33, 35, 57, 66, 67, 93, 95–97, 116, 143, 150–152, 154, 155, 212, 213, 242, 247, 253). Ammonia concentrations increase following exercise (209) and as such, may be exaggerated in cirrhotics due to impaired ureagenesis (86, 207). A study in patients with well-compensated chronic liver disease reported elevated plasma ammonia with exercise (59), but there are no reports on the development of overt or minimal encephalopathy in response to acute or long-term exercise in cirrhosis. In addition to hyperammonemia during exercise, muscle anaerobic glycolysis results in elevated serum lactate concentrations. Even though hepatic lactate clearance (Cori cycle) is impaired in cirrhosis (121) and lactate concentrations increase following exercise (34), it is not clear whether elevated lactate after exercise affects clinical outcomes. Elevated exercise/postexercise lactate concentrations may contribute to early onset fatigue in these patients, limiting exercise capacity (28, 173, 198). Exercise-induced lactate stimulates mitochondrial biogenesis (102), but this may be impaired in cirrhosis due to the competing adverse effects of hyperammonemia on skeletal muscle mitochondrial oxidative function (51). Substrate utilization during exercise has not been evaluated in cirrhosis, but since cirrhosis is a state of accelerated starvation (91), anaerobic glycolysis with increased demands on substrate utilization may impair the physiological benefits of exercise on protein homeostasis. Studies on metabolic responses to exercise in cirrhosis will help optimize training regimes.

Exercise and the Gut Microbiome

There is increasing evidence of alterations in the gut microbiome in response to exercise (153). Cross-sectional studies on gut microbiome and metabolites showed distinct differences with a higher α-diversity and a lower abundance of Bacteroides and Lactobacillus species in athletes than in lean counterparts (153). An increase in butyrate-producing bacteria has been reported in women who exercised (4). In longitudinal studies, even though changes in gut microbiome have been reported by some but not others, there is heterogeneity in the alterations in the bacterial species (4, 42, 170). There seems to be agreement that butyrate-producing bacteria increase in response to exercise and may benefit gut homeostasis with improved gut barrier function and gut immune function (4). Short-term exercise-induced changes in gut microbiome, however, revert to the initial diversity after a relatively short washout period, suggesting that either longer-term or different approaches are needed for sustained beneficial microbiome responses (153). In cirrhosis, gut ammoniagenesis has been shown to contribute to hepatic encephalopathy, but skeletal muscle responses have not been reported (2). Bacteria-producing short-chain fatty acids, including butyrate, have been reported to be beneficial in cirrhosis (232). Also, lowering gut ammoniagenic bacteria may be beneficial in improving muscle responses to exercise. The effects of exercise on gut microbial composition last only for a short term with variable responses. Despite the potential of targeting the gut microbiome to enhance exercise benefits in cirrhosis, these data are currently not available.

Molecular Responses to Exercise

Skeletal muscle molecular responses to exercise in healthy human subjects have been extensively reviewed by others (8, 29, 77, 81, 103, 156, 172, 214, 226). Briefly, endurance exercise stimulates molecular pathways that promote mitochondrial protein synthesis and metabolic adaptations via peroxisome proliferator-activated receptor (PPAR) γ coactivator 1-α (PGC1α), calcium/calmodulin-dependent kinases, calcineurin, AMP-activated kinase (AMPK), and mitogen-activated kinases (MAPK), whereas resistance exercise promotes myofibrillar and global muscle protein synthesis and inhibits muscle protein breakdown via the Akt-mammalian target of rapamycin complex 1 (Akt-mTORC1) pathways (Fig. 2) (29, 61, 63, 133, 172). However, the molecular and functional responses to concurrent exercise are more complex than the interference/concurrent training effects reported in the past and depend on a number of factors (Table 4) (41, 126, 161). The order, in addition to FITTVP (250) [frequency, intensity (defined by %V̇o_2max or target heart rate), time (duration), type, volume (how much exercise, e.g., repetitions), and progression (adjustment of exercise to adaptive responses, e.g., change in V̇o_2max)] (which we call OFITTVP), and the goals determine the response to concurrent exercise (Table 4) (8, 41, 87, 161, 171, 172, 186). In brief, when concurrent exercise is recommended, the primary desired outcome, whether metabolic adaptation or muscle mass, determines the recommendations of OFITTVP(Table 4).

Fig. 2.

Simplified schematic of major cellular responses to exercise. A: endurance exercise initiates a number of metabolic, molecular, and cellular responses that contribute to increased glucose uptake, fiber-type switch, and mitochondrial biogenesis. Competing influences (stimulatory and inhibitory) on the mammalian target of rapamycin complex 1 (mTORC1) determine the outcome on myofibrillar protein synthesis and autophagy. B: resistance exercise responses include consistent activation of mTORC1 and ribosomal biogenesis with increased myofibrillar protein synthesis and increased muscle mass. AMPK, AMP-activated kinase; ATF, activating transcription factor; GLUT, glucose transporter; HIF, hypoxia-inducible factor; IGF, insulin like growth factor; MAPK, mitogen-activated protein kinase; NRF, nuclear respiratory factor; PI3K, phosphoinositide 3 kinase; PGC, peroxisome proliferator-activated receptor (PPAR) γ coactivator; SIRT1/3, sirtuin 1 and 3; SLC, solute linked carrier; TFAM, transcription factor A, mitochondrial.

Table 4.

Concurrent training effects and strategies for optimizing concurrent exercise

Concurrent Training Effects
Response	Endurance response improves, and resistance response is diminished; conflicting data but broad consensus and guidance as below
Training background	Responses are different; in untrained subjects, the concurrent training effect is less/absent and becomes more evident with training
Duration of the exercise (volume of exercise and intensity/severity)	High intensity (≥85% of V̇o2max or at 100% V̇o2max), low volume (1–2 km per session) short bouts (4–10 min) (HITT), interval training have less negative effects on resistance training-induced adaptations during a concurrent program, cycling has fewer negative effects than running
Intrasession exercise sequence	Endurance mitigates resistance response especially when performed first, maybe due to glycogen depletion and/or molecular pathways. Order of exercise depends on primary goal: exercise performed first has greatest response effect. Endurance first if goal is metabolic adaptation and resistance first if muscle mass increase is primary goal
Proportion ratio	Ratio of time spent (1:2 or 1:3) on endurance or resistance endurance: greater proportion on exercise goal (metabolic adaptation, more time on endurance, and vice versa)
Single bout or long-term responses	Long-term exercise responses different from single bout studies due to metabolic/molecular adaptations and training effect
Gap between exercise	3–4 h gap limits the negative effects, >24 h almost completely mitigates negative effects
Gender effects	Women fatigued by endurance exercise faster and therefore have less response to resistance; for women, perform resistance exercise before endurance, but determine ratio based on goal.

	Goal-Dependent Strategies to Optimize Concurrent Training
	Endurance	Resistance
Order	Endurance before resistance if desired outcome is V̇o2max, mitochondrial, fiber-type related	Resistance before endurance if goal is hypertrophy, insulin signaling, strength, or power; endurance first may interfere with molecular responses of resistance; particularly true for females whose muscles tend to fatigue faster from endurance
Frequency	Less frequent: 3–4 days/wk	Daily with focus on different muscle groups each day
Intensity	High intensity ≥85% V̇o2max (SIT, HIIT) rather than low intensity <85% V̇o2max	Dependent on goal: strength, power, or endurance
Time	<20 min if before resistance; one-third or one-fourth of session should be spent on endurance	Minimum of 3 h postendurance with preference for 24 h; two-thirds to three-fourths of session should be spent on resistance
Type	Cycle rather than treadmill	Dependent on goal: strength, power, or endurance
Volume	Low volume rather than high volume (1–2 km)	RM, number of reps, number of exercises dependent on goals and training status to avoid injury
Progression	Increase to 85% V̇o2max as adaptation occurs	Increase weight lifted, repetitions, or time held as adaption occurs

HIITT, high-intensity interval training; RM, repetition maximum; SIT, short-intensity training; V̇o_2max, maximum oxygen consumption.

The evaluation of skeletal muscle molecular responses in cirrhosis is critical because the causal factors of cirrhosis (e.g., ethanol, fatty acids) and consequences [reduced food intake, hyperammonemia, increased bile salts, decreased branched-chain amino acids (BCAAs), decreased testosterone and growth hormone] can perturb skeletal muscle signaling responses (46, 48, 129, 189, 190, 236). There are limited data on muscle metabolic and molecular changes in the resting state and no published data on muscle molecular responses to exercise in liver disease (45–48, 53, 117, 129, 142, 158, 179, 189). Attenuated muscle response to physiological anabolic stimuli, known as anabolic resistance, that has been reported in aging-related sarcopenia has also been reported in cirrhosis (7, 50, 100, 167). It is known that patients with liver disease, including cirrhosis, do not respond to nutrient supplementation with reversal of muscle loss (45, 46, 50, 229), but the anabolic response to exercise or training has not been as well studied. A number of molecular perturbations have been suggested to contribute to the anabolic resistance in cirrhosis, including increased skeletal muscle expression of myostatin, impaired mTORC1 signaling, and increased autophagy. Hyperammonemia in liver disease, due to a combination of impaired hepatocyte function and portosystemic shunting, results in a unique hyperammonemic cellular stress response in the skeletal muscle with increased phosphorylation of eukaryotic initiation factor 2α (eIF2α) with reduced protein synthesis (47–49, 51–53, 155, 189, 190). Consistently, high doses of leucine were able to overcome the hyperammonemic stress response, suggesting a potential mechanism of anabolic resistance in cirrhosis (47, 53, 227). Whether exercise-induced muscle hypertrophy is blunted or recovery following immobilization-induced muscle loss is delayed in cirrhosis is currently not known.

There are also emerging data on epigenetic changes with alterations in DNA methylation, histone modifications, and regulation of noncoding RNA-associated genes in response to exercise (156). Gene-specific alterations can mediate the expression of genes involved in metabolic regulation and mitochondrial biogenesis, including PGC1α, following exercise (238). Similarly, metabolic alterations, including changes in TCA cycle intermediates (α-ketoglutarate, succinate, fumarate) that occur in the skeletal muscle during exercise can change DNA methylation regulatory enzymes (156). Alterations in AMPK phosphorylation and NAD⁺ that occur during exercise are potential mediators of these alterations. In cirrhosis, hyperammonemia increases AMPK phosphorylation, decreases NAD⁺, and can result in histone and DNA modifications with changes in chromosome conformation and transcription factor binding. Whether such perturbations alter exercise-induced responses in cirrhosis and how long these changes persist is currently unknown, even though there are data to suggest persistence of “memory” to hypertrophy and inflammation (205, 230).

Organelle Responses to Exercise

Ribosome biogenesis, content, and translational capacity increase after resistance, but not concurrent, training (75, 80) (192). Ribosome responses in human subjects with endurance training have not been reported even though preclinical data suggest no changes (195). Whether the mediators of the liver-muscle axis, including hyperammonemia, negatively affect ribosome biogenesis and therefore muscle hypertrophy responses in cirrhosis is not known (52). Both endurance and resistance exercise have been reported to increase mitochondrial efficiency and substrate utilization to meet the bioenergetic demand of active contraction as well as the energy dependence of protein synthesis and myofiber hypertrophy (90, 248). Skeletal muscle mitochondrial dysfunction in cirrhosis and hyperammonemia include impaired oxygen consumption, defects in substrate oxidation, reduced ATP synthesis/content, defects in electron transport chain function, and increased generation of reactive oxygen species (51, 118), but whether mitochondrial mass, efficiency, and response to exercise are altered in these patients is currently not known. Endurance exercise increases lysosomal autophagy that may exacerbate hyperammonemia-induced increased muscle autophagy with loss of muscle mass in liver disease (24, 48, 190). Whether exercise-induced autophagy will be additive to the dysregulated muscle autophagy in cirrhosis with accelerated muscle loss is a concern.

Safety of Exercise in Cirrhosis

Hemodynamic and metabolic responses to exercise have the potential for adverse clinical outcomes in patients with cirrhosis due to elevated portal pressure. During exercise of any type, cardiac output increases and there is redistribution of blood flow, which enhances muscle oxygen delivery to accommodate the increased workload (104, 168). Redistribution of blood volume is accompanied by increased vascular resistance and a dysfunctional response to vasodilators and vasoconstrictors, both of which may be responsible for the increase in portal pressure in cirrhotics during exercise (12, 16, 88). These physiological responses to exercise may have negative outcomes for patients with cirrhosis, such as increased portal hypertension and an increased risk of variceal bleeding (88).

Another potential adverse consequence of exercise is related to net muscle ammonia release (94, 96). In cirrhosis, hepatocellular dysfunction and portosystemic shunting impair ureagenesis with net muscle uptake of cytotoxic ammonia (86, 147, 189, 207). Skeletal muscle uptake of ammonia may be protective against hepatic encephalopathy, one of the best described neurological consequences of hyperammonemia in liver disease (3). Consistently, it has been reported that sarcopenic patients with cirrhosis have more severe encephalopathy (38, 123, 159). One group suggested that if exercise increases muscle mass in cirrhotic patients, hepatic encephalopathy may improve (1). However, ammonia is released during exercise by muscle deamination of AMP and by amino acid catabolism (95). During exercise, skeletal muscle ATP utilization exceeds ATP regeneration with accumulation of ADP and AMP. To avoid accumulation of excess AMP, which alters cellular metabolism and signaling responses, inosine monophosphate and ammonia are generated by the enzyme AMP deaminase, resulting in net muscle ammonia release (106). Deamination of BCAA, which serves as a source of energy and intermediary metabolites for the tricarboxylic acid cycle in the skeletal muscle during exercise, may also result in net ammoniagenesis (111–113). Net release of ammonia due to increased muscle ammoniagenesis during exercise could contribute to the hyperammonemia of cirrhosis with the potential to aggravate muscle loss and encephalopathy. Even if overt encephalopathy is not detected, minimal hepatic encephalopathy may develop, affecting exercise performance and coordination. Interestingly, despite compelling data on the health benefits of exercise, excessive exercise has been reported to be associated with adverse consequences that include lower plasma HDL and elevated systolic blood pressure, plasma triglycerides, and plasma fasting insulin (22). Whether there are adverse metabolic consequences of intense exercise and what constitutes intense exercise in patients with liver disease is yet to be defined.

Published data suggest that moderate intensity, relatively short duration endurance exercise is safe in patients with cirrhosis (220). The type, intensity, and duration of exercise in published studies are shown in Table 3. Resistance exercise increases muscle mass in healthy controls and may increase muscle mass in cirrhotics, but there are no published data except one study in abstract form. However, resistance exercise in cirrhotics has the potential to increase portal pressure and the consequent risk of variceal bleeding (88). Concurrent exercise may be another option in cirrhosis, mitigating the potential negative consequences of either type of exercise alone (8). Notably, concurrent exercise has been reported to improve the quality of life, blood pressure, and body composition in cirrhotics (16, 149). A combination of aerobic exercise and strength training decreased hepatic pressure and fat mass with no change in lean body mass in cirrhosis (16). Others have reported that concurrent exercise improved ventilatory efficiency (percentage of carbon dioxide in exhaled volume per breath) (149). Thus, concurrent training may be superior to endurance or resistance exercise alone in cirrhosis, but more systematic studies on safety and responses are needed. Additionally, the optimal time of the day, order, intensity, and duration of the components of concurrent training for patients with cirrhosis are still not known. It is important to note that due to potential balance and coordination problems given the physical and mental impairments resulting from end-stage liver disease, such as ascites, edema, and hepatic encephalopathy, most of the current literature recommends supervised exercise, at least in the initial phase of a new exercise regimen (140, 251).

Nutrient Supplementation with Exercise

The European Society Parenteral and Enteral Nutrition and the European Association for the Study of the Liver recommend that cirrhotics need 25–40 kcal·kg⁻¹·day⁻¹ based on their body weight and 1.0–1.5 g/kg protein daily to prevent muscle loss (62, 124, 162, 184). Nutrient supplementation, especially that which contains protein, enhances resistance exercise-induced increases in muscle mass in healthy subjects (137, 244), but whether such benefits occur in cirrhotics with anabolic resistance is not known. In obese cirrhotics, increased fat mass may mask the low lean body mass (sarcopenic obesity), and modifications of nutritional supplements (decreased fat calories and increased protein/amino acid content) may enhance exercise benefits (68, 70). A late evening snack increases lean body mass in cirrhosis (229), but contractile function and exercise capacity have not been evaluated.

Exercise increases muscle protein synthesis but also increases muscle protein breakdown (244). For muscle mass to increase, the rate of protein synthesis needs to exceed the rate of muscle protein breakdown with exercise (244). Even though cirrhosis is a state of anabolic resistance with impaired muscle protein synthesis, specific nutrient supplements (l-leucine, β-hydroxy β-methylbutyrate) may enhance muscle protein synthesis. In healthy subjects, nutrient intake induces insulin release and decreases protein breakdown (7, 18, 244), making nutrient intake beneficial to the exercise-induced muscle protein synthesis. Amino acids are a key driver for protein synthesis, suggesting that amino acid or protein supplements may be a source of calories in addition to being a substrate for protein synthesis compared with carbohydrates/fats in cirrhotics (244). Therefore, a snack consisting of protein before, during, or after exercise may increase amino acid availability. The International Society of Sports Nutrition suggests that protein supplementation before, after, and during the time of exercise will increase amino acid levels and muscle protein synthesis (30). Additionally, the optimum time of supplementation may also depend on the desired results (muscle hypertrophy, reduced time to fatigue, improvements to muscle protein breakdown/synthesis) and type of exercise (244).

BCAAs improve the response to both endurance and resistance exercise by decreasing the muscle protein breakdown associated with exercise, increasing muscle protein synthesis, and improving time to exhaustion of skeletal muscle and endurance time in healthy patients (30). Of the three BCAAs, leucine is a known stimulator of mTOR and thus can aid in muscle protein synthesis (6, 11, 53). Comparing the response of leucine supplementation in cirrhotic patients with and without endurance exercise showed an improvement in 6-min walk test, thigh circumference, and BMI in the exercise group only (193, 194). An increase of 2,000 steps per day over baseline and a BCAA supplement improved grip and leg strength (110). Decreased muscle protein breakdown in response to nutrient-induced insulin release may lead to gains in muscle mass in healthy controls with a normal response to insulin, but cirrhotics are insulin resistant and therefore may not respond similarly. Whether improvement in insulin resistance with endurance exercise in cirrhotic patients (16, 128, 181) restores the benefits of supplemental nutrition with exercise is also unknown. Hyperammonemia, impaired lactate clearance, and increased tryptophan entry into the brain may all contribute to accelerated fatigue in cirrhosis (19). BCAAs decrease cerebral tryptophan entry and may potentially improve muscle ammonia metabolism, thereby decreasing fatigue in these patients, but this has not been evaluated. The anticipated benefits of BCAAs or BCAA-enriched protein supplements need to be balanced with the elevated ammonia levels and renal dysfunction in cirrhosis (25, 30, 44, 160, 231), given the inconsistent data on the benefits of BCAA in ammonia detoxification and the risk of worsening renal dysfunction with nitrogenous loads (111–113).

Summary and Conclusions

The broad terms “frailty” and “malnutrition” have been used to refer primarily to loss of muscle mass and contractile function in chronic diseases including cirrhosis (32, 70, 179). We suggest that the term “sarcopenia of cirrhosis,” a type of secondary sarcopenia (43), be used to refer to the loss of muscle mass and contractile and metabolic dysfunction in cirrhosis. We also suggest that the general term “malnutrition” in cirrhosis be used to refer to a combination of disordered nutrient intake, absorption, assimilation, and consequent metabolic and physiological perturbations. Of the components of sarcopenia, skeletal muscle loss is well recognized in liver disease, and its mechanisms are now being identified. In contrast, despite the recognition of the adverse clinical consequences of contractile dysfunction, there has been a very limited evaluation of the underlying mechanisms. Impaired contractile function contributes to decreased physical activity (measured by hand grip strength, 6-min walk test, sit-to-stand test) and also increases mortality (220). There are observational studies on responses to short-term endurance exercise in cirrhosis with very limited knowledge on the changes in body composition or molecular or tissue metabolic responses. Human data suggest an increase in V̇o_2max with endurance exercise, and concurrent exercise may be beneficial in cirrhosis (16, 149). However, the details of the strength/resistance exercise component in these two studies are not clear. The safety and benefits of resistance exercise alone in cirrhosis are currently not known. Unlike in healthy subjects, there are no data on whether supplementary nutrition provides any additional benefit to an exercise training program in cirrhosis.

Fatigue is a major barrier to exercise, and contributory factors include encephalopathy, decreased ATP synthesis/content, and modifications of contractile proteins (51, 155). Ascites and pedal edema of varying severity also contribute to the impaired ability to exercise. Diuretics cause hypokalemia with muscle weakness and β-blockers in cirrhosis aggravate fatigue. Strategies to reverse skeletal muscle bioenergetic dysfunction and perturbations in contractile proteins are likely necessary to overcome fatigue in cirrhosis. Additionally, published studies on exercise responses in cirrhosis have been restricted to well-compensated patients, but whether similar benefits will occur in decompensated cirrhosis, and if the improvements noted are sustained over time, is not known.

Current publications also do not provide guidance as to the type (resistance, endurance, or concurrent), frequency, or duration of exercise required to improve muscle mass or functional capacity. Studies on the response to exercise have focused on V̇o_2max, a measure of cardiorespiratory fitness, and have found that V̇o_2max improves with endurance exercise in cirrhosis but to a lesser extent than in healthy controls. However, whether these benefits contribute to increased survival is not known. Endurance exercise does not produce a substantial increase in muscle mass, an outcome that is directly related to survival in cirrhosis (32). Resistance exercise improves muscle mass in healthy subjects. However, given the very limited data on resistance exercise response in cirrhosis, potential risks of variceal bleeding and the anabolic resistance that can impair expected gains in muscle mass, current data do not provide evidence to support resistance exercise in cirrhosis. Observational studies suggest that concurrent exercise benefits cirrhotics, especially those who are obese, with improvement in body composition (increase in lean mass and decrease in fat mass) and V̇o_2max and decreased heart rate and portal pressure, but mechanistic and physiological studies are still needed (16, 17, 149). Despite the recommendations of multiple societies to increase caloric intake in cirrhosis, modifications need to take into consideration the increasing prevalence of sarcopenic obesity in cirrhosis. For nonobese cirrhotics, the source of calories to be added (e.g., protein alone, a combination of protein and carbohydrates) and the frequency and the timing of calories in relation to exercise need to be determined. A circadian dependency of muscle mass and strength responses to exercise or nutrient supplementation is now recognized, and cirrhosis is associated with circadian perturbations. Whether and how these will alter nutrient and exercise responses in cirrhosis is unknown.

In summary, despite nascent data on exercise in cirrhosis, physiological and mechanistic studies are required to optimize and maximize the beneficial responses.

GRANTS

This work was funded in part by National Institutes of Health (NIH) Grants R21 AA022742; RO1 DK 113196; RO1 GM119174; P50 AA024333; UO1 AA021890; UO1 AA026975; and UO1 DK061732 and the Mikati Foundation Grant support to S. Dasarathy. N. Welch was partially supported by the American College of Gastroenterology Clinical Research Award and NIH Grant T32 DK083251.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

A.B., N.W., and S.D. conceived and designed research; A.B., N.W., and S.D. analyzed data; A.B., N.W., and S.D. interpreted results of experiments; A.B., N.W., and S.D. prepared figures; A.B., N.W., and S.D. drafted manuscript; A.B., N.W., and S.D. edited and revised manuscript; A.B., N.W., and S.D. approved final version of manuscript.