Upwards of 30% of the world’s population has nonalcoholic fatty liver disease (NAFLD) [1]. At this point in time, there is no regulatory-agency-approved effective drug therapy or cure, and lifestyle modification with dietary change and increased physical activity remains crucial in the clinical management of all types of NAFLD, including nonalcoholic steatohepatitis (NASH). Even when a regulatory-agency-approved drug therapy becomes widely available, lifestyle modification will always play a key role in the prevention and treatment of NAFLD and lessen the burden of associated extrahepatic disease from cardiovascular disease events and cancer. Following years of research, it is widely accepted that regular physical activity and, in particular, exercise training, which is a type of physical activity that is planned, structured, repetitive and with a specific goal in mind ^[2][5], leads to many benefits within and outside the liver in patients with NAFLD ^[3][4][6,7]. These benefits include an improvement in liver fat, histologic NASH activity, change in body composition, gain in physical fitness, reduction in markers of cardiovascular risk, improvement in health-related quality of life and possibly a reduction in oncologic risk ^[4][7]. Importantly, many of these improvements, including the reduction in magnetic resonance imaging (MRI)-measured liver fat, may occur without clinically significant body weight loss ^[5][8].

AMPK is a fuel-sensing enzyme that is activated by energy stress ^[6][18] and is composed of a trimeric complex with a catalytic subunit (α) and two regulatory subunits (β and γ). AMPK also has a liver-specific role in hepatic de novo lipogenesis, fatty acid oxidation, glycogenolysis and gluconeogenesis. AMPK activity is abnormally low in patients with NAFLD, leading to excessive accumulation of liver fat ^[7][19]. For this reason, AMPK remains a drug target of interest, and in fact, the recent Phase 2a STAMP-NAFLD study enrolled 120 patients who were randomized to treatment with PXL770, a direct AMPK activator, or a placebo ^[8][16]. Unfortunately, this study did not meet its primary endpoint of statistically significant MRI-determined liver fat reduction; however, subgroup analysis limited only to patients with type 2 diabetes demonstrated a significant reduction in liver fat and corresponding metabolic parameters, including glycemic control.

Animal models of NAFLD demonstrate exercise changes the AMPK pathway, leading to less liver fat accumulation by reducing lipogenesis and increasing fatty acid oxidation ^{[9][10][11][12][13]}[20,21,22,23,24]. Importantly, exercise-induced AMPK activation appears to be dose-related. As exercise intensity and duration increase, ATP usage increases to the point where it cannot be regenerated quickly enough, increasing the AMP/ADP:ATP ratio and activating AMPK ^[14][25]. Further, in order to generate additional ATP during exercise, glycogen, which is the main energy substrate used during exercise at higher intensities (typically greater than 70% of VO₂max), dissociates from AMPK, leading to AMPK activation (glycogen-bound AMPK is inactive) ^[6][15][16][18,26,27]. Sustained moderate–vigorous intensity exercise seems to be required to deplete glycogen enough to activate AMPK ^{[16][17][18][19]}[27,28,29,30]. While several small studies have demonstrated that exercise can favorably impact targets downstream of AMPK, including FGF-21 ^[20][31] and also ribosomal protein s6 ^[21][32], in patients with NASH, researchers await a definitive study showing that exercise can directly activate AMPK in this patient population.

There is a clear and consistent body of evidence to support the role of the AMPK pathway as a key pathway modulated by exercise training that appears to be related to both intensity and exercise volume across animal and human studies.

FGF is a complex family of peptide hormones that has a crucial implication on regulating energy homeostasis and metabolism ^[22][33]. Multiple isoforms of FGF are involved in the cascade, but FGF-19 and FGF-21 are closely related to fat metabolism ^[23][34] and are the signaling pathways for both hormones involved in NAFLD and NASH development and the focus of drug discovery ^[24][25][12,35]. Both the FALCON and BALANCED trials investigated the efficacy of FGF-21 analogues pegbelfermin and efruxifermin in patients with NASH ^[25][26][27][13,35,36].

FGF-19 is expressed in the ileum and is released in response to bile acid stimuli. FGF-19 is activated postprandially, regulating the transcription of hepatic protein and glycogen synthesis and inhibiting hepatic gluconeogenesis. FGF-19 appears to largely act locally on the hepatocytes, whereas FGF-21 tends to be expressed systemically and, of particular interest, in the skeletal muscle, adipose tissue and in the liver tissue ^[28][29][37,38]. Importantly, FGF-21 relies heavily on the coreceptor β-klotho ^[30][39]. If there is a lower expression of β-klotho, resistance to FGF-21 has been observed, resulting in impaired fatty acid oxidation ^[31][40]. In fact, NAFLD is felt to be an FGF-21-resistant state ^[32][33][41,42].

Because FGF-21 is widely and variably expressed in the human body, it has been challenging to identify the relationship between exercise and FGF-21 expression. Despite this, there is a robust and consistent body of evidence describing this relationship in pre-clinical animal models of NAFLD and in clinical studies involving patients who are inactive, overweight, obese or diabetic ^[34][35][36][43,44,45]. Importantly, the relationship between exercise and FGF-21 expression appears to exist across differing intensities, where both moderate and vigorous intensity exercise can change FGF-21 expression ^[37][46]. In terms of patients with NAFLD, a recent study by Takahashi et al. ^[38][47] found that after 12 weeks of resistance training in which participants performed push-ups and squats three times a week on non-consecutive days, serum FGF-21 level was significantly reduced, confirming the results seen with aerobic exercise training in animal models of NAFLD. Furthermore, a study of 24 patients with biopsy-proven NASH reported that 20 weeks of aerobic exercise training significantly reduced serum FGF-21 in parallel with gains in cardiorespiratory fitness ^[20][31]. Despite these differences in methodology, it is generally agreed upon that while acute exercise tends to increase plasma levels of FGF-21, perhaps due to increased production by skeletal muscle, chronic exercise training programs of four weeks or more duration lead to a reduction in serum FGF-21 level while simultaneously increasing expression of FGF receptors and β-klotho in not only liver tissue but also in adipose tissue and skeletal muscle ^[39][48].

Although closely related to FGF-21, exercise may play a different role in FGF-19 expression. There have been observed correlations between resistance training and upregulation of FGF-19, but further studies are needed to ensure no other external factors are contributing, such as fasting states. These potentially confounding factors may explain the conflicting results of FGF-19 downregulation following an acute one-hour bout of aerobic exercise ^[40][49] as well as multiple negative studies, which are limited by small sample sizes. 

To date, the scientific literature suggests that exercise training across different exercise intensities and volumes can activate FGF-21. 

The liver plays a central role in insulin metabolism and is impacted by multiple gut hormones. One such hormone is GLP-1, an incretin, which helps to regulate satiety and lipid metabolism in both the fasting and glucose-stimulated states ^[41][52]. GLP-1 has recently emerged at the forefront of drug development in NASH given the recent promising results of early-phase studies using both semaglutide ^[42][11] as well as liraglutide ^[43][53] and the fact that several GLP-1 receptor agonists are regulatory-agency-approved for the medical treatment of overweight and obesity as well as type two diabetes ^[44][45][54,55].

Exercise can impact serum levels of GLP-1, and in fact, exercise can increase GLP-1 levels in healthy individuals and in persons with obesity and suppress appetite ^[46][47][56,57]. In patients with NAFLD, Kullman et al. ^[48][58] measured the impact of a short-term, one-week high-intensity exercise program and found that while GLP-1 level remained unchanged, exercise instead reversed the GLP-1 resistant state of NAFLD by restoring the normal physiologic response of GLP-1 to glucose stimulation. This raises the question of the importance of increasing serum GLP-1 level versus ameliorating GLP-1 resistance, building on previous work showing short-term exercise programs to improve hepatic insulin extraction in patients with NAFLD ^[49][59]. To our knowledge, the effects of a long-term exercise training program on GLP-1 have not yet been investigated in patients with NAFLD or NASH.

While GLP-1 and other gut hormones present intriguing avenues of research, the existing scientific data prevent strong conclusions regarding the impact of exercise training programs on this therapeutic target.

The liver plays a principal role in lipid metabolism as the primary site of de novo lipogenesis and fatty acid oxidation ^[50][60]. In fact, lipid-derived energy production in the liver occurs through the β-oxidation of fatty acids ^[50][60]. However, mitochondrial defects, which are related to both physical inactivity and obesity ^[51][61], reduce the oxidative capacity of the mitochondria, which results in incomplete β-oxidation and the accumulation of metabolic by-products, such as ceramides and diacylglycerides ^[52][62]. While intrahepatic triglycerides themselves do not cause hepatic insulin resistance, it is thought that the accumulation of the aforementioned by-products impairs insulin receptor signaling through various mechanisms and has been identified as critical in the pathogenesis of hepatic insulin resistance ^[53][63].

Regular exercise has been shown to improve mitochondrial oxidative capacity and increase mitochondrial content, which are related to increases in cardiorespiratory fitness ^[54][55][64,65]. In fact, cardiorespiratory fitness is inversely related to hepatic steatosis ^[56][66], and improvements in cardiorespiratory fitness are independently associated with improvements in steatosis ^[57][58][67,68]. As cardiorespiratory fitness has been shown to improve to a similar degree with both high-intensity interval training (HIIT) and more traditional moderate-intensity continuous training ^[59][69], this may, in part, explain why HIIT leads to similar improvements in hepatic steatosis to more moderate-intensity continuous exercise despite expending less energy ^[60][70].

Although most work in this space has been conducted in animal models, the available data from human studies support the role of improving cardiorespiratory fitness as a key therapeutic target in the management of NAFLD.

Mitochondria are vital organelles that are at the forefront of cellular metabolism, especially in the liver, which is the primary metabolic organ in the human body. UCPs are a key component of mitochondrial metabolism and are mitochondrial inner-membrane proteins which mediate proton leak across the inner membrane through anion transport and uncouple substrate oxidation from ATP synthesis ^[61][71]. Five key mitochondrial UCPs have been discovered. UCP-1 is found largely in brown adipose tissue and plays a role in thermogenesis and energy expenditure; UCP-2, while fairly ubiquitous, is found in high concentrations in the liver and regulates insulin secretion from pancreatic β-cells as well as fatty acid metabolism; UCP-3 is expressed largely in brown adipose tissue and also skeletal muscle and influences fatty acid metabolism and insulin sensitivity ^{[62][63][64][65]}[72,73,74,75]. UCP-4 and UCP-5 are found in the brain. The dysfunction of these transporters has been correlated with various metabolic disorders, such as obesity and diabetes ^[66][67][76,77], and also NAFLD ^[68][69][78,79]. Genetic variation in UCP polymorphisms is also important. In patients with type 2 diabetes, the CC genotype of the UCP-1 rs3811791 polymorphism blunts the response to regular physical activity in terms of insulin resistance and also lipid control, even at guideline-based amounts of 150 min/wk. of moderate-intensity activity ^[70][80]. Moreover, the INDOGENIC cohort study demonstrated that even in healthy individuals, the GG genotype for the UCP-2 G-866A polymorphism changed the physiologic response to energy intake, making these individuals more prone to weight gain and overeating over a two year follow-up period ^[71][81]. Consequently, targeting UCPs to enhance energy utilization remains an attractive treatment option in NASH. However, when medications with this mechanism of action have been used to induce body weight loss, they have been significantly limited by a strong side effect profile, and this target must be approached with caution ^[72][82]. A recent phase 2a trial for a novel mitochondrial uncoupler, HU6, appears to have addressed some of these concerns by demonstrating a favorable side effect profile while inducing significant body weight loss and MRI-measured liver fat reduction in patients with obesity and NASH ^[73][83].

Exercise is well known to upregulate the expression of various mitochondrial UCPs. Animal models of aerobic exercise have demonstrated that UCP-1 can be upregulated in brown adipose tissue, white adipose tissue as it browns in response to exercise and skeletal muscle ^{[74][75][76][77]}[84,85,86,87]. Animal models have also demonstrated that UCP-2 is modulated by exercise in the vascular endothelium, myocardium, adipose tissue and skeletal muscle ^[74][78][84,88], as is UCP-3 in skeletal muscle ^[79][89]. When limited to animal models of NASH, aerobic training reverses dysfunction in UCP-2 in the liver ^[80][90]; however, researchers are unaware of any studies to date in patients with NAFLD or NASH confirming these findings.

In summary, animal models suggest a role for exercise training in upregulating UCPs.

PPARα is a nuclear receptor that plays a key role in regulating lipid metabolism. It is specifically activated by fatty acids and their derivatives. The receptor can be found in many key organs, including the liver, and has three subtypes, PPAR-α, PPAR-γ and PPAR-β/δ ^[81][91]. When expressed in the liver, PPAR-α is responsible for fatty acid catabolism and energy homeostasis ^[82][83][92,93]. PPAR-γ is heavily involved in glucagon signaling and insulin sensitivity and, thus, is closely related to adiposity-related disorders ^[84][85][94,95]. The PPAR pathway remains one of many targets for drug discovery of anti-steatogenic medications, where late-phase studies have demonstrated PPAR-γ agonists, such as pioglitazone ^[86][96], and dual PPAR-α/γ agonists, such as saroglitazar ^[87][15], improve insulin resistance and lipid metabolism through regulation of fatty acid metabolism and modulation of inflammatory adipocytokines and adiponectin secretion. Moreover, the PPAR pathway is intricately linked to several other signaling pathways involved in NAFLD and NASH pathogenesis, including FGF-21, AMPK, and uncoupling proteins (UCP) ^[88][89][90][97,98,99].

The PPAR-α pathway is known to be strongly influenced by exercise training in non-NAFLD populations ^[91][92][93][100,101,102], including patients who are physically inactive and are overweight or obese ^[94][103]. In animal models of NAFLD, multiple studies have shown exercise-induced PPAR-α activation to mediate liver fat reduction ^{[12][85][90][95]}[23,95,99,104]. Importantly, the impact of exercise training on the PPAR-α pathway appears to be independent of exercise type, in that PPAR activation has been observed in animals who performed either moderate-intensity aerobic exercise training with either swimming or running ^[85][90][95,99], HIIT ^[95][96][104,105] or resistance training ^[97][106]. While exercise per se can influence the regulation of PPAR activity, epigenetic factors may also play a role. For example, maternal exercise during pregnancy may confer protection against the development of NAFLD early in the life of offspring exposed to a high-fat diet. Bae-Gartz et al. ^[98][107] demonstrated that male offspring of exercised mouse dams were protected from adult-onset NAFLD mediated through greater activation of PPAR-α. The PPAR-γ receptor may also be affected by exercise training. Batatinha et al. ^[99][108] demonstrated that after eight weeks of treadmill training, PPAR-α knockout mice that were fed a high-fat diet to induce NAFLD still experienced a decrease in fat accumulation, perhaps owing to changes in PPAR-γ activity and fatty acid oxidation in the skeletal muscle.

In summary, the current available evidence indicates that exercise-mediated activation of PPAR appears to be mediated by exercise intensity and volume.

The thyroid gland and NAFLD are intricately linked in that thyroid hormones, including triiodothyronine (T3) and thyroxine (T4), are not only involved in lipid metabolism regulation, glucose uptake and increased size and number of mitochondria ^[100][101][111,112] but also that hypothyroidism is associated with increased NAFLD risk ^[102][103][113,114]. As a result, thyroid-hormone-based treatments, including resmetirom ^[104][14], are an attractive therapeutic target in patients with NAFLD and NASH. Two THR subtypes are found in the human body: THR-α, which is predominantly expressed in cardiac tissue, and THR-β, which is most commonly found in the liver and is responsible for the intrahepatic response to T3 ^[105][115]. THR-β is a critical receptor in the regulation of cholesterol metabolism and fatty acid oxidation, as exhibited in mouse models ^[69][106][79,116]. Importantly, THR-β is intricately linked to other pathways of interest in NAFLD, including PPAR-α, FGF-21 and UCP1 ^{[69][107][108]}[79,117,118]. Since THR-β is interrelated to many metabolic pathways, it has been the focus of both pharmacologic and non-pharmacologic therapies. Resmetirom is an emerging therapy in NASH clinical trials that has the potential to significantly reduce hepatic fat, making it a potential therapy for patients with NASH ^[104][14].

Regular physical activity, including exercise training, is also well known to significantly impact circulating levels of thyroid hormone at the population level ^[109][119], in smaller groups of healthy adults who perform regular resistance training ^[110][120] and also in post-menopausal women with metabolic syndrome ^[111][121]. Specifically, exercise training leads to increased turnover of T3 and T4 at the same work rate, effectively lowering resting concentrations of thyroid hormone. The intensity of exercise is an important consideration as it appears to cause differential effects on thyroid hormones, with vigorous intensity leading to the greatest change ^[112][122]. When limiting solely to NAFLD and NASH, the data are more sparse.