Several studies assessed the association of PA with cancer incidence. In a prospective study covering 1.44 million participants, the authors found that high-level PA was associated with a decrease in 13 types of cancer (top-ranking reduction esophageal, liver, lung, and kidney). An adjustment to BMI mildly attenuated the results, but PA remained associated with a marked reduction in 10 cancers regardless of the adjustment. Surprisingly, PA was associated with an increased risk for melanoma and prostate cancer [52]. A recent review cites five prospective studies assessing the effect of PA on the incidence of cancer; four studies found a protective effect of PA against several digestive, gynecological, and pancreatic cancers, while one study did not find a protective effect of PA against prostate cancer [53]. Data from the NHS suggests that PA during adolescence is associated with a marked decrease in the incidence of breast cancer [54].

Data on the anti-HCC effect of exercise has been slowly accumulating from animal studies. Aguiar et al. assessed the effect of training on hepatic carcinogenesis. Male Wistar rats were submitted to a DEN-induced liver carcinogenesis protocol and divided into groups according to a high or low-fat diet and swimming training (5 days a week for eight weeks). There was a marked improvement in the biochemical parameters of the trained rats on the low-fat diet (LFD) and high-fat diet (HFD) groups. However, while there was a marked decrease in pre-neoplastic lesion development in the LFD group, no such effect was noticed in the HFD group. The researchers concluded that exercise attenuated liver carcinogenesis together with dietary manipulations [55]. In another study, serum was collected from a young overweight man without metabolic syndrome before and after 3 weeks of LFD and moderate aerobic exercise. Post-intervention serum was added to the HCC cell line HepG-2 and markedly attenuated cell proliferation, lipid accumulation, and signalling of various stress pathways [56]. Both previous studies combined diet and PA as interventions. Indeed, dietary and metabolic signals are known to markedly increase the incidence of HCC. In 2015, Piguet et al. used genetically engineered mice to assess dietary and exercise effects separately. They used hepatocyte-specific PTEN-deficient mice, which develop steatohepatitis and HCC spontaneously. Mice were fed the same diet and were divided into exercise (treadmill running-1 h, five days a week for 32 weeks) or sedentary groups. The exercise group developed hepatic nodules larger than 15 mm³ in 71% vs. 100% in the sedentary group. There was also a marked decrease in the number of tumors per liver and tumor volume per liver independently of hepatic steatosis. Exercise was associated with increased AMPK and Raptor phosphorylation and decreased Mammalian Target of Rapamycin (mTOR) activity [57]. Later on, Arfianti A et al. assessed the same effect in a chemically induced DEN HCC model in two groups of genetically engineered mice, obese/diabetic Alms1 mutant (foz/foz) mice, and JNK1 deficient foz/foz mice. Mice were divided to exercise or sedentary groups (exercise wheel from week 4 to 12 or 24 weeks). Exercising, foz/foz mice developed obesity by week 24 but still had less dysplastic hepatocytes and significantly fewer tumors (15% vs. 64% compared to sedentary controls). In contrast to previous studies, these diabetic/obese mice failed to activate AMPK and mTOR Complex 1. Instead, exercise improved insulin sensitivity and steatosis and regulated key signalling pathways, which resulted in decreased hepatocyte proliferation [58]. In another study, tumors derived from an HCC cell line (Morris Hepatoma 3924A) were grown sub-cutaneously and were later resected and surgically implanted in the livers of American cancer Institute rats. The rats were assigned to exercise (treadmill running 1 h, five days a week for four weeks), sedentary or sorafenib± exercise or sorafenib+metformin groups. Tumor area, cell proliferation, and vascular density were all decreased by exercise. AMPK phosphorylation was increased in the exercise group together with the expression of PTEN, while STAT3 and S6 ribosomal proteins were decreased. Transcriptomic analysis showed that exercise affected non-tumoral tissue rather than the tumor itself. The anti-tumoral effects of exercise in the sorafenib group were similar to the effect of metformin, suggesting that metformin induces an exercise-like effect [59]. In a similar study, mice on a choline deficient high-fat diet (CD-HFD) were divided to exercise or sedentary groups. CD-HFD mice developed steatohepatitis and hepatic pre-neoplastic lesions. Exercise improved their metabolic parameters and improved steatohepatitis and hepatic inflammation as manifested by decreased aminotransferases. Similar to the previous study, there were fewer hepatic adenomas with increased AMPK activity and mTOR inhibition. The authors concluded that exercise reduced the transition from fatty liver to non-alcoholic steatohepatitis (NASH) and decreased progression to fibrosis and tumorigenesis [60]. In two recent studies, the intensity and method of exercise were assessed. HCC was induced by DEN in C57BL/6 mice, which were subjected to high-intensity interval training (HIIT) or endurance training on a treadmill (from 8–26 weeks). Endurance training resulted in lower cancer incidence and growth and less fibrosis. Furthermore, endurance training resulted in less lipotoxicity and improved body composition, inflammation, and metabolomics compared to the HIIT group. The authors suggested that moderate-intensity endurance training may be superior to HIIT in its anti-tumoral effect [61]. Similarly, Cao et al. assessed moderate endurance training compared to HIIT, arguing that HIIT may induce an acidic micro-environment that may be tumor-promoting. The DEN induced HCC model was utilized in C57BL/6 mice that were subjected to HIIT or moderate endurance training on a treadmill for 18 weeks. Again, endurance training decreased tumor incidence and size as compared to HIIT. There were no significant differences in the mRNA levels of key gluconeogenesis enzymes [62].

The data presented from mouse and cellular models, although preliminary, show a beneficial effect of exercise on HCC initiation and progression in chemically, metabolically, and surgically implanted HCC induced models. While these models do not bear a good resemblance to human HCC, the data is compelling. Furthermore, recent, small publications suggest that exercise type is important, as endurance training showed better anti-tumoral effects than HIIT. Finally, several of the models suggest a synergistic effect to a healthy diet on tumor incidence in these models.

Initial data on the effect of PA on HCC risk in humans came from prospective studies in Korea and Japan. In Korea, a prospective study of 444,963 men older than 40 showed that moderate-high leisure-time physical activity (LPA) (>2 times/week, >30 min) had a significant protective effect against several cancers, including HCC as compared to low LPA (<2 per week, <30 min) [63]. In Japan, 79,771 men and women aged 45–74 responded to a questionnaire and were followed until 2004. LPA was assessed using a metabolic equivalents/day score. Increased PA was associated with a decrease in the risk for various cancers and was more pronounced in women, especially among the elderly and those who exercised regularly. HCC was decreased only in men and reached significance only in the group who were in the highest quartile of PA compared to the lowest [64]. Wen et al. developed prediction models to assess the risk of HCC in a large cohort in Taiwan. Their cohort consisted of 428,584 individuals that were divided into two groups, one with HCV (130,533) and the other without. Data were collected from a standard medical screening program, and the average follow-up was 8.5 years. During this period, 1668 HCC cases were identified. The data included PA, which was categorized as inactive, low-active, and active according to the intensity in a metabolically equivalent task (MET); X represented the duration of exercise in hours per week (METs hours/week). Although individuals with low active and active PA showed significantly reduced risk for HCC in the initial model, this difference became non-significant when other variables (age, sex, health history, HBV and HCV, AFP and transaminases) were added to the model [65]. A study conducted in the USA in 2013 focused on 507,897 participants aged 50–71 in the NIH-AARP diet and health study that were followed for ten years. A total of 628 cases of HCC were identified during the study period. Physical exercise was defined as the performance of 20 min of vigorous activity per week. Groups were classified between “rarely perform” (lowest level) to >5 times per week (highest level). Comparison of the lowest to highest group of PA showed a significant decrease in the risk for HCC (RR = 0.64; 95% CI, 0.49–0.84) independently of BMI [66]. Arem et al. looked at PA in almost 300,000 men and women in the same NIH-AARP study in order to assess PA patterns over the life course and their association with HCC. They used modelling starting from teenage years to middle age and identified seven distinct PA trajectories. They showed that those who maintained PA levels over life had an approximately 30% reduction in HCC risk compared to those with consistently low PA. The specific pattern of PA (increased or decreased PA through life) had different evolving risks of HCC. Their results suggest that maintaining PA from early age onwards had the best protective effect and warn that increasing PA later in life may not yield the same protective effect [67].

Recent years have shown a plethora of studies assessing the role of PA in association with hepato-pancreato-biliary (HPB) cancers. Data from the EPIC cohort looked at cause-specific hazard ratios (HR) among 467,000 participants focusing on HPB cancers. They identified 275 HCCs among 532 all HPB cancers during follow-up. There was a 45% decrease in the risk for HCC comparing active to inactive participants and a 50% decrease for those engaging in vigorous PA (defined a >2 h/week). Markers associated with obesity such as BMI and waist circumference partially explained the reduced risk [68]. In a rigorous prospective study spanning almost 30 years, the authors assessed the effect of various lifestyle parameters, including PA, on liver-related mortality in over 125,000 participants. HCC accounted for a third of the deaths. The risk for overall liver-related mortality declined progressively with increasing PA and increased with higher BMI. The HR for liver-related mortality tripled in obese sedentary compared to lean, active participants. Findings were similar for HCC or cirrhosis-specific mortality. The authors suggested that engaging in average pace walking >3 h/week could have prevented 25% of liver-related deaths [69]. More recently, the emphasis focused on the intensity, mode, and duration of PA. Luo et al. looked at moderate intensity PA in two well-defined cohorts of the NHS and the HPFS over an average of 23 years. Surprisingly, total and vigorous PA was not associated with a reduced risk of HCC, while moderate-intensity PA showed an inverse association with HCC. The reduced risk was especially associated with brisk walking, suggesting that the mode of PA may play a role in HCC prevention [70]. Lee et al. wanted to establish whether there is a minimum PA threshold for the prevention of HCC. They conducted a meta-analysis and divided the PA performed in the selected studies into three groups: high >3 h/week, moderate 2–3 h/week, or low <2 h/week. A total of 10 prospective cohort studies were included. PA was associated with a dose-dependent decrease in HCC risk and mortality. High and moderate PA reduced the risk of HCC by 54% and 45%, respectively. According to their data, the authors state that 2 h/week PA is mandatory to reduce HCC risk [71]. An interesting recent study looked at variability in the association of obesity and PA and HCC at the state level in the US. Trends of HCC incidence from 2001–2017 were calculated using data from the Centers for Disease Control (CDC) and the National Cancer Institute Surveillance. There were striking state-level disparities in HCC incidence ranging from 6.3 to 0.9 in various states and ethnicities. There was a moderate inverse correlation with state-level PA and the incidence of HCC (r = −0.40, p = 0.004) [72].

Two studies looked at the effect of PA on specific, at-risk populations. In the first, Feng et al. looked at the effect of PA on alcohol-related cancer (including liver cancer). Data were collected from British and Scottish population-based surveys spanning the years 1994–2008. Alcohol consumption was categorized from “never drinkers” to “harmful” (>35 units/week for women; >49 units/week for men) and PA to lower (7.5 METs h/week) or upper recommended limits (15 METs h/week). There were 54,686 participants, and hazardous/harmful alcohol consumption was associated with a marked increase in cancer-related mortality. Although cancers were not separated by type, the increased risk was eliminated among participants who exercised more than 7.5 METs hours/week and persisted in the upper recommended limits group [73].

Lastly, a recent study looked at whether PA is associated with HCC risk in patients with chronic HBV infection. The authors looked at 9727 treatment-naïve Korean HBV carriers who started treatment with nucleoside/tide analogous from 2012–2017. During the study period, the cumulative HCC incidence was 8.3%. There was an inverse correlation between carriers engaged in PA measured in METs and those without PA. PA had a protective effect in patients with and without cirrhosis that resulted in an approximately 40% reduction in risk for HCC. The authors conclude that PA was significantly associated with a reduced risk of HCC in HBV carriers, treated with anti-viral medications [74].

The data presented here from multiple studies, cohorts, and meta-analyses, clearly shows a beneficial effect for PA in reducing the risk of HCC. The data comes from diverse populations with different gender and ethnic backgrounds. However, there are multiple questions remaining to be answered, especially the obesity independent effect of PA that was only partially shown. Other issues include the duration, mode, and intensity of PA and whether PA is effective in reducing the risk for HCC across multiple liver diseases.

Following the promising data on HCC prevention prior to tumor development, researchers set out to assess whether PA can improve recovery, prevent a recurrence, and possibly prolong survival in patients with HCC that were treated with or without curative intent.

Kiabori et al. from Japan, looked at 51 patients who underwent hepatectomy for HCC. The patients were randomized to diet alone or diet and PA after the operation. PA was started one week before the operation, resumed one week post-operatively, and continued for six months. The authors show that patients in the PA group had significantly improved metabolic parameters such as whole-body mass, fat mas, insulin resistance and insulin levels and recommend early resumption of PA after HCC resection [75]. The same group assessed whether perioperative PA was associated with long-term survival in a larger group of HCC patients undergoing hepatectomy. One hundred and six patients underwent cardiopulmonary exercise assessment utilizing various methods pre and six months post-hepatectomy. Patients were classified as the maintenance group if they had >90% anaerobic threshold six months post-operatively compared to pre-operatively (n = 78) or a decreased group if the threshold was below 90% (n = 28). 5-year recurrence-free survival and overall survival were significantly improved in the maintenance group compared to the decreased group (39.9% vs. 9.9% p = 0.018, and 81.9% vs. 61.7% p = 0.006 respectively). Thus the maintenance of the anaerobic threshold, which is maintained by PA, was an independent positive prognostic marker [76].

Some concern was raised, whether PA could exacerbate CLD in patients with HCC. It was thought that exercise might increase portal pressures and decrease glomerular filtration rate and thus may expose cirrhotic patients to the risk of variceal bleeding and the development of hepatorenal syndrome.