Hepatocellular carcinoma in non-alcoholic steatohepatitis: Growing evidence of an epidemic?


Dr Etsuko Hashimoto, Department of Internal Medicine and Gastroenterology, Tokyo Women's Medical University, 8-1 Kawada-cho Shinjuku-ku Tokyo 162-8666, Japan. Email: drs-hashimoto@mti.biglobe.ne.jp


The incidence of hepatocellular carcinoma in non-viral-related chronic liver disease has gradually increased in Japan. Obesity and diabetes mellitus type 2 have been established as a significant risk factor for hepatocellular carcinoma (HCC) by epidemiologic observations and experimental studies. The risks of these factors for HCC are likely conferred by two factors: the increased risk for development of non-alcoholic steatohepatitis (NASH) and the carcinogenic potential of themselves. Hepatocellular carcinoma in NASH is difficult to evaluate because histological diagnosis is required for diagnosis of NASH, which can lead selection bias. Furthermore, end-stage NASH is in effect “burned-out” NASH, for which the diagnosis of NASH cannot be made any more. At all events, previous studies on the etiology of Japanese HCC showed that non-alcoholic fatty liver disease accounts for 1–5% of all HCC (male predominant, median age 72 years). They have high prevalences of obesity and/or diabetes mellitus type 2 and 10–75% of the HCC arose from non-cirrhotic livers. HCC in NASH may be of multicentric origin, similar to HCC based on viral hepatitis. Regular screening for HCC is extremely important especially in cirrhotic NASH patients and recurrence should be warned. In western and Asian countries, the prevalence of non-alcoholic fatty liver disease in the general population is increasing dramatically. Therefore, there is an urgent need to elucidate pathogenesis and clinical features of HCC in NASH. In this review we summarize current concepts for HCC in NASH.


PRIMARY LIVER CANCER is the fifth most common cancer worldwide and the third most common cause of cancer mortality.1–8 Hepatocellular carcinoma (HCC) accounts for about 90% of primary liver cancers. HCC has several interesting features, including an association with chronic liver disease, geographic and ethnic variations, and gender differences. It is well known that in any underlying chronic liver diseases, fibrosis is the single most important risk factor for HCC. Recently, obesity and diabetes have been revealed as risk factors for HCC by clinical studies and experimental studies. The risks of obesity and diabetes for HCC are likely conferred by two factors: the increased risk for development of non-alcoholic fatty liver disease (NAFLD) and the carcinogenic potential of themselves.9,10

Non-alcoholic fatty liver disease consists of simple steatosis and non-alcoholic steatohepatitis (NASH), while NASH comprises a wide spectrum of conditions from NASH without fibrosis to cirrhosis. There is increasing evidence that NASH is a risk factor for HCC. In western and Asian countries, the prevalence of NAFLD in the general population is increasing dramatically and ranges from 5% to 40%.11–19 Therefore, there is an urgent need to elucidate pathogenesis, clinical features, and treatments for especially in NASH advanced stages and NASH-related HCC.

There is a growing understanding of the molecular mechanisms responsible for hepatocarcinogenesis. In this review we summarize current concepts for HCC in NASH about epidemiologic factors, characteristic features and pathogenesis.


IN MOST DEVELOPED countries, hepatitis C virus (HCV) infection is the most common cause of chronic liver disease underlying HCC and accounts for 40–80% of all HCC, but the etiology of 5–50% of new HCC cases remains unclear.1–8 The true prevalence of NASH in HCC is difficult to evaluate because histological diagnosis is required for diagnosis of NASH, which can cause selection bias. Furthermore, characteristic features of NASH disappear in the end stage (so-called “burned-out” NASH), making diagnosis impossible. Malik et al.20 reported that among 98 cirrhotic NASH patients who underwent liver transplantation at University of Pittsburgh School of Medicine, 17 patients (17.3%) were diagnosed as having HCC. In this study, NASH-related HCC accounted for 3.8% of all HCC cases.

In a case series of 105 patients with HCC at the University of Michigan Medical Centre, NAFLD accounted for 13% (14 patients) of the patients with HCC. Of the 14 patients, six patients had prior histologic confirmation of NASH and eight patients had cryptogenic cirrhosis and features consistent with NAFLD.21 A recent retrospective case study using liver explants and liver resections from 50 cases at the University of Illinois at Chicago Medical Center22 showed that 10% of HCC cases were caused by NASH.

In Japan, the most recent nationwide Journal of Health and Welfare Statistics reported that primary liver cancer ranks fourth for men and sixth for women as a cause of death from malignant neoplasm.23 HCC accounts for 94% of primary liver cancer.24 Concerning underlying liver disease, the Liver Cancer Study Group of Japan showed that HCV is the most common underlying liver disease in HCC. HCV-related HCC accounts for 68% of all cases of HCC, followed by hepatitis B virus (HBV)-related HCC at 15%. However, the incidence of HCV-related HCC has gradually decreased in recent years, and that of HCC in cases with non-viral-related chronic liver disease has gradually increased.

According to a nationwide retrospective survey on the etiology of liver cirrhosis, NAFLD accounts for 2.7% of all cirrhosis and HCC caused by cirrhotic NAFLD accounts for 1.6% of all HCC.25 In the 2009 Japanese Nationwide Survey of HCC, a total of 2299 patients (15.8% of all HCC patients) were diagnosed as having non-viral-related HCC.26 Among non-viral HCC, alcoholic liver disease-related HCC (7.2% of all HCC patients) was the most common etiology, followed by unknown HCC (5.1%) and NAFLD-HCC (2.0%). Concerning surgical cases, NASH-related HCC was found in eight patients (1% of all HCC patients) at Osaka Medical Center for Cancer and Cardiovascular Diseases from 1990 to 2006.27

In comparison with the United States, the prevalence of NASH-related HCC is lower in Japan, which may be partly due to the higher incidence of HCV related-HCC. In the future, NASH-related HCC will increase due to the marked increase of NASH in Japan.


IN ALMOST ALL populations, males have higher HCC rates than females, with male-to-female ratios between 2:1 and 4:1. This gender difference may be attributable to differences in exposure to risk factors for HCC, such as alcohol consumption and cigarette smoking. However, a similar gender disparity is seen in mice given a chemical carcinogen, diethylnitrosamine (DEN).28 The authors of that study traced HCC gender differences to interleukin-6 (IL-6), which plays a role in the pathogenesis of HCC. DEN administration caused greater increases in serum IL-6 concentrations in males than in females. It is proposed that estrogen-mediated inhibition of IL-6 production by Kupffer cells reduces HCC risk in females. Nakagawa et al.29 conducted a retrospective cohort study on 330 chronic hepatitis C patients to examine whether this hypothesis is applicable to human HCC. Although female patients showed a weak negative correlation between serum IL-6 levels and estradiol levels, the lower risk of HCC in female patients cannot be fully explained by estrogen-mediated inhibition of IL-6 production.

It is well known that NAFLD exhibits age and gender differences in both prevalence and severity.30–34 Among 492 biopsy-proven NASH patients diagnosed at our university hospital, men were significantly more common in the younger ages (24% in woman, 76% in men); however, the number of NASH cases over 50 years in women was higher than that in same-aged men (67% in woman, 33% in men).31 These age and gender differences of NAFLD may just reflect differences in prevalence of obesity and metabolic syndrome in general population. Concerning NASH severity, the prevalence of cirrhotic NASH in women was higher than that in men (57% in women and 43% in men).32 However, in HCC the prevalence was higher in men than in women (38% in women versus 62% in men, the median age 70.5 years in men, and 68.0 years in women) (Fig. 1).33 This might be due to the higher risk of HCC in males. Further studies are needed to clear the pathogenesis of gender differences in HCC.

Figure 1.

(a) Gender distribution in liver cirrhosis (LC) and hepatocellular carcinoma (HCC); among cirrhotic non-alcoholic steatohepatitis (NASH) patients, the prevalence was higher in women than in men. In contrast, for HCC, the prevalence was higher in men than in women. (b) The median ages of male and female HCC patients were almost equal. (inline image): women; (inline image): men.


THE DEVELOPMENT OF NASH is thought to require basal steatosis, followed by a “second hit” that is capable of inducing necroinflammation and eventually fibrosis. This hypothesis is the so-called “two hit theory”.35,36 The second hit can include insulin resistance, inflammatory cytokines, and oxidative stress and so on. Recently, a new concept for the pathogenesis of NASH was reported by Tilg et al. as the “multi-parallel hit”.37 Their new hypothesis suggests that many hits act together to cause liver inflammation. They suggested that among several hits inducing NASH, gut-derived and adipose tissue-derived factors may play a central role in this process. An interesting point is that inflammation might precede steatosis in certain causes. Several studies revealed endoplasmic reticulum stress and autophagy may have a role in the pathogenesis of NASH.38,39

Regarding the pathogenesis of HCC, many basic and clinical studies have assessed it in chronic viral liver diseases and recently it has also been assessed in NASH.40,41 Insulin resistance, oxidative stress, inflammatory cytokines and autophagy contribute to the carcinogenic potential of chronic liver diseases including NASH. Figure 2 shows a hypothesis of pathogenesis in NASH related HCC. It is well known that obesity and insulin resistance induce the release of multipleproinflammatory cytokines including tumor necrosis factor-α (TNF-α), and IL-6 as well as decreased amounts of adiponectin.42,43 To investigate whether inflammatory cytokines and NASH induced by high fat diet (HFD) promote hepatocarcinogenesis, Wang et al. injected rats with a low dose of DEN and then fed either a control diet or HFD. The data clearly demonstrated that NASH induced by HFD promoted DEN-initiated early hepatocarcinogenesis, which was associated with elevated TNF-αl/NF-κB signaling.44 Other studies have confirmed that IL-6-/- mice or TNFreceptor-1 -/- mice abolished obesity-enhanced HCC development in DEN-treated mice.45,46 These data suggest that IL-6 and TNF signaling via TNF receptor 1 are important in NASH-related HCC. Concerning adiponectin, it has been shown that hypoadiponectinemia accelerated hepatic tumor formation in a NASH mouse model.47,48

Figure 2.

Pathogenesis of hepatocellular carcinoma (HCC) in non-alcoholic steatohepatitis (NASH); obesity and insulin resistance induce liver steatosis (NAFLD). The release of multiple proinflammatory cytokines and oxidative stress occurred in some cases with obesity, insulin resistance and non-alcoholic fatty liver disease (NAFLD). Hyperinsulinemia leads to upregulation of the production of insulin-like growth factor-1 (IGF-1) and activation of insulin receptor substrate-1 (IRS-1). These factors then activate several molecules and signal pathways; p53, mitogen-activated protein kinase (MAPK), Jak/STAT3, phosphatidylinositol-3 kinase (PI3K)/AKT. These signal pathways play important roles in both cell proliferation and inhibition of apoptosis. Especially, the JNK-1 is thought to have the most important role in both the development of NASH and carcinogenesis. Finally, ethnic and genetic factors are associated with the development of NASH and carcinogenesis.

Oxidative stress and the release of reactive oxygen species (ROS) likely contribute to the development of both NASH and HCC. The oxidative stress may favor tumorigenesis through inflammation, and cell proliferation, or it may directly induce cancer-promoting gene mutations. Trans-4-hydroxy-2-nonenal (4HNE), a product of lipid peroxidation, may be an important etiological agent for human cancers that have a mutation at codon 249 of the p53 gene.49 Concerning anti-oxidative stress, nuclear respiratory factor-1 (Nrf1) is an essential transcription factor to prevent oxidative stress. Nrf-1 knockout mice demonstrated steatohepatitis with fibrosis and developed to hepatic neoplasia due to increased oxidative stress.50 In a human study, carcinogenesis due to oxidative stress has been reported by Maki et al. They found that patients whose liver tissue showed high levels of 8-OHdG and 4HNE had a higher incidence of recurrent HCC.51 These data strongly suggested that oxidative stress is involved in the pathogenesis of HCC.

Regarding hyperinsulinemia, it upregulates the production of insulin-like growth factor-1 (IGF-1), which plays a role in regulating hepatocyte growth factor-scatter factor and HCC progression.52–54 Insulin also activates the insulin receptor substrate-1 (IRS-1). Tanaka et al. demonstrated that IRS-1-mediated signals may act as survival factors and protect against TGF-β-induced apoptosis in HCC cell lines.55 In addition, they showed that overexpression of IRS-1 led to activation of promotion of hepatocyte proliferation by mitogen-activated protein kinase and phosphatidylinositol-3 kinase (PI3K), which are important pathways in the growth of HCC cells.

PTEN, retinoic acid, c-Jun amino-terminal kinases (JNKs) and several other molecules in animal models are associated with development of NASH and hepatic carcinogenesis.56,57 Among these molecules, the JNKs are thought to have the most important role in both the development of NASH and carcinogenesis.58–63 Hirosumi et al. reported that JNKs activity is abnormally elevated in obese mice induced by HFD.58 Furthermore, an absence of JNK1 results in decreased adiposity, significantly improved insulin sensitivity and enhanced phosphorylates IRS-1. They concluded that JNKs are a crucial mediator of obesity and insulin resistance. Puri et al. analyzed the relationship between JNKs and NAFLD. Human NASH increased the activation of JNKs in comparison to patients with simple steatosis by western blot analysis.59 They also demonstrated that activation of JNKs is associated with the level of histologic activity. Concerning the relationship between HCC and JKN-1, several groups have reported an important role of JNK1 in the pathogenesis of human HCC and its close association with the expression of HCC signature genes.60–63 Enhanced JNK1 activation was noted in 17 out of 31 HCC samples (55%), compared to in non-cancerous tissues. JNK2 activation was roughly equal in HCC and non-cancerous tissues.61 JNK1 plays important roles in the development of human HCC partially through epigenetic mechanisms. Fujishita et al. reported that JNK signaling promoted intestinal tumorigenesis through activation of mammalian target of rapamycin complex 1 (mTORC1), which is implicated in various human cancers.64,65 The PI3K/PTEN/AKT/mTOR pathway is also reported to be activated in human HCC subjects.66,67

Caldwell et al.68 reported that HCC rates between US and Japan were clearly different, suggesting that cultural and genetic differences might influence carcinogenesis. Kato and coworkers reported that several genomic DNA polymorphisms were associated with Japanese HCV-related HCC.69,70 Valentin et al. reported that the rs738409 patatin-like phospholipase domain-containing 3 (PNPLA3) genotype, which influences hepatic fat content and the development of NASH,71,72 was independently associated with cirrhosis and HCC in chronic hepatitis C.73 Liver iron overload and C282Y mutation of the HFE gene are associated with a high risk of HCC in patients with alcoholic liver cirrhosis.74,75 The roles of genetic background in hepatic carcinogenesis of NASH patients should be clarified.


OBESITY, AS DEFINED by a body mass index (BMI), has been established as a risk factor for HCC by epidemiologic studies in the general population and also in cirrhosis patients (the relative risk is increased 1.6- to 4.5-fold in the general population and 2- to 3-fold in cirrhosis patients).76–82 The relative risk of HCC in cirrhosis patients varies with etiology and it is higher for non-viral cirrhosis.

Utsunomiya et al.83 evaluated the impact of obesity on the outcome of HCC surgery in 328 consecutive patients with primary HCC and 60 patients with recurrent HCC. They found that among patients with recurrent HCC, the obese patients showed a significantly poorer long-term prognosis than those in the non-obese group (5-year survival rate; 51.9% and 92.0%, respectively, P < 0.05). They suggested that the worsened prognosis in obese patients might be associated with the synergistic impacts of the obesity itself and perioperative blood transfusions.

Concerning visceral fat accumulation and recurrence of HCC, Ohki et al. evaluated 62 HCC patients with suspected NASH after curative treatment.84 The analysis identified older age and the accumulation of visceral fat as independent risk factors for the recurrence of HCC. They concluded that the majority of HCC “recurrence” patients may in fact be the development of de novo HCC and the visceral fat accumulation may be associated not only with tumor recurrence but also with initiation.


DIABETES AND IMPAIRED glucose tolerance are among the complications of cirrhosis, introducing a potential bias in case-control studies. Only prospective cohort studies can shed light on the association between diabetes and development of HCC.

El Serag et al.85 conducted a study of HCC incidence in a large cohort of the Department of Veterans Affairs patients (173 643 patients with and 650 620 patients without diabetes). The incidence of HCC was significantly higher among patients with diabetes (incidence rate: 2.39 vs. 0.87 per 10 000 person-years, respectively, P < 0.0001). Diabetes was associated with a hazard ratio of 2.16 of HCC. Other large population-based cohort studies from Europe and the United States have demonstrated a 1.86-fold to 4-fold increase in the risk of HCC among patients with diabetes.86–89 According to a large-scale population-based Japanese cohort study,90 patients with diabetes may have an increased overall risk of cancer and the risk is especially high for cancer of the liver (hazard ratio: 2.24).

Recently, important findings about the association between HCC and diabetes treatments have been reported. Donadon et al.91 performed a retrospective case-control study on 465 HCC patients, 618 patients with cirrhosis and 490 control subjects. They confirmed that diabetes mellitus type 2 is an independent risk factor for HCC. Moreover, in male patients with diabetes mellitus type 2, the data show a direct association of HCC with insulin and sulphonylureas treatment and an inverse relationship with metformin therapy. Recently, Hassen et al.92 also found a similar association between the risk of HCC and the treatment for diabetes.


CHEN ET AL.93 investigated 23 820 residents of Taiwan over a follow-up period of 14 years. Obesity (BMI > 30 kg/m2) was independently associated with a risk of HCC in persons with anti-HCV positive and in persons without HBV and anti-HCV, but not in persons with HBV (relative risk; 4.13 in persons with anti-HCV, relative risk; 2.36 in persons without HBV and anti-HCV). Diabetes was associated with HCC in all three groups. They found more than 100-fold increased risk in HBV or HCV carriers with both obesity and diabetes, indicating the synergistic effects of metabolic factors and viral hepatitis. Polesel et al.94 conducted a hospital-based case-control study of 185 HCC patients and 404 hospital controls. Among subjects without HBV or HCV infection, obesity and diabetes were confirmed to increase the risk of HCC (odds ratio; 3.5 for obesity, and 3.5 for diabetes), and these two conditions combined led to a 10-fold increase in risk (odds ratio; 11.8). A synergistic effect of diabetes and obesity is evident in relation to HCC.


LUDWIG ET AL.95 studied tissue iron in 447 explant livers histologically and chemically. In this study, positive iron staining was found in 32.4% and increased chemical hepatic iron concentration was found in 20.3%, including 38 cases (8.5%) with hepatic iron overload in the hemochromatosis range. However, among these 38 cases, only five were suspected of hemochromatosis. Prevalence of iron overload was very common in cryptogenic cirrhosis (65.1%), chronic hepatitis B (64.7%), and alcoholic cirrhosis (62.8%) but rare in biliary cirrhosis (6.7% in PSC, 10% in PBC), and NASH was 42.9%. In this study, they found variation of iron overload depending on the etiology of cirrhosis. Nahon et al.74 studied histological liver iron overload and C282Y/H63D HFE gene mutations in 162 alcoholics and 139 patients with HCV-related cirrhosis (LC-C) to assess iron overload on risk of HCC. Liver iron overload and C282Y mutation are associated with a higher risk of HCC in patients with alcoholic but not LC-C.

Concerning NASH, excess hepatic iron may increase the risk for NASH and its progression to cirrhosis.75,96–99 Recently, Sorrentino et al.100 conducted a retrospective study investigating hepatic iron measured with a semiquantitative method in liver biopsies of 153 cirrhotic NASH patients: 51 with HCC and 102 without HCC, matched for age, sex and stage of liver disease. They showed that iron deposits were more prominent in HCC patients than in cirrhotic NASH patients. They found that excess sinusoidal iron deposition may play a role in liver injury, as well as possible carcinogenesis. Hepatic iron overload is frequently seen in NASH patients, but its role is not well defined. Further studies are needed to clarify this important issue.


THE LONG-TERM PROGNOSES of NAFLD, simple steatosis, NASH and cirrhotic NASH have been reported in population-based studies as well as hospital-based cohort studies (Table 1).32,101–111 According to these studies, different prognoses among these conditions have been revealed.

Table 1.  Studies on long-term mortality in non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH)
AuthorDiagnosisnAverage F/U (Years)Cirrhosis Prevalence n (%)HCCn (at baseline)Death
Overall n (%)Liver-related /Overall (%)
  • At the end of the follow-up period.

  • ‡The diagnosis was made by imaging or liver biopsy.

  • §

    §The diagnosis was made by liver biopsy. NR, no record.

Adams et al.101NAFLD4207.621 (5%)2 (0)53 (12.6%)13.2%
Ekstedt et al.102NAFLD§12913.710 (7.8%)3 (0)26 (20.2%)7.7%
Rafiq et al.103NAFLD§13118.5NR1 (0)78 (59.5%)15.4%
Soderberg et al.104NAFLD§118219 (7.6%)5 (0)47 (39.8%)19.1%
Sørensen HT et al.105NAFLD18006.204 (0)NRNR
Teli et al.106Simple§ Steatosis409.600 (0)14 (35%)0.0%
Dam-Larsen et al.107Simple§ Steatosis17020.42 (1.2%)0 (0)48 (28.2%)2.1%
Evans et al.108NAFLD§268.71 (4%)0 (0)4 (15%)0.0%
Hui et al.109Cirrhotic-NAFLD§237.0100%0 (0)6 (26%)83.3%
Sanyal et al.110Cirrhotic-NAFLD§15210100%13 (3)29 (19.1%)69.0%
Yatsuji et al.32Cirrhotic-NAFLD§683.4100%21 (14)19 (27.9%)78.9%
Soderberg et al.104Cirrhotic-NAFLD§921100%3 (0)8 (88.9%)50.0%
Ascha MS et al.111Cirrhotic-NAFLD1953.2100%25 (0)NRNR

Compared with individuals of the general population of the same age and gender, those with NAFLD had lower-than-expected survival at a standardized mortality ratio from 1.34 to 1.69, due to increases in cardiovascular disease and liver-related death.101,102,104,107 The most common causes of death in NAFLD and NASH were cardiovascular disease and malignancy, followed by liver-related death. It is also important to be aware that the patients with NAFLD who underwent liver biopsy had a significantly shorter survival compared with those who did not, reflecting an increased likelihood of performing biopsy in patients with more advanced disease.101

Sørensen et al.105 examined the incidence of cancer in 7326 fatty liver patients discharged from Danish hospitals during 1977–1993. Patients with a prior diagnosis of liver cirrhosis were excluded. Overall, 523 cancers were diagnosed over a 16-year period, yielding a 1.7-fold increased risk compared with the Danish general population. The risk of primary liver cancer was markedly elevated in patients with alcoholic as well as NAFLD with a standardized incidence ratio of 9.5 and 4.4, respectively.

To elucidate the clinical features of cirrhotic NASH, we analyzed the natural history of 68 patients of biopsy-proven cirrhotic NASH and compared them with 69 cases of age- and sex-matched cirrhosis based on LC-C.32 The mean age of cirrhotic NASH patients was 62.7 years. Although the outcome of cirrhotic NASH was better than that for LC-C, cirrhotic NASH followed a similar course to that of LC-C (5-year HCC development rate 11.3% for cirrhotic NASH and 30.5% for LC-C; 5-year survival rates of 75.2% and 73.8%, respectively). In patients with cirrhotic NASH, HCC was the leading cause of death, followed by liver failure. Development of HCC and Child–Turcotte–Pugh scores were significant risk factors for mortality in cirrhotic NASH patients on multivariate analysis.

Sanyal et al.110 compared 152 cirrhotic NASH patients with a median age of 55 years with 150 matched LC-C. They reported that there were no significant across-group differences in mortality in patients with Child class B or C cirrhosis, except that cirrhotic NASH had a significantly lower risk of development of HCC (10/149 vs. 25/147 patients P < 0.01). Recently, Ascha et al.111 compared 195 cirrhotic NASH patients with 315 LC-C patients. The yearly cumulative incidence of HCC was found to be 2.6% in cirrhotic NASH, compared with 4.0% in LC-C (P = 0.09). The incidence of HCC in their study was similar to our results. Multivariate regression analysis revealed that older age and alcohol consumption were independent variables associated with the development of HCC in cirrhotic NASH. However, if the definition of “no alcohol intake” was made very strict, even NASH patients with minimal alcohol consumption would be excluded and alcohol would not be selected as a risk factor for HCC. In the future, risk factors for HCC in patients with alcoholic steatohepatitis and NASH stratified by several levels of alcohol intake should be studied to clarify the influence of a small amount of alcohol on HCC.

We compared 34 patients with NASH-related HCC and 348 NASH patients without HCC. Advanced fibrosis, older age, histological low-grade inflammation, and low AST levels were selected as risk factors for the development of HCC by multivariate regression analysis.33 It is well known that when NASH progresses to the end stage, necroinflammatory changes and transaminase levels gradually decline, so low grade inflammation and a low AST level may be present at the very end stage. Accordingly, risk factors for the development of HCC in NASH patients consist of the features of end-stage NASH and older age.

All previous studies32,104,109–111 confirmed that cirrhotic NASH has a similar course to LC-C. In the natural history of cirrhotic NASH, liver failure mainly due to HCC is the main cause of morbidity and mortality. Regular screening for complications of liver cirrhosis and development of HCC is thus extremely important for cirrhotic NASH patients.


THERE HAVE BEEN many reports describing HCC arising in patients with NASH.20–22,25–27,33,112–135 Interestingly, most of them are from Japan. Yasui et al. performed a cross-sectional Japanese multicenter study. They examined histologically proven 87 NASH patients (male; 54, female; 33, median age, 72 years) with HCC. Obesity, diabetes, dyslipidemia, and hypertension were present in 62%, 59%, 28%, and 55% of the patients, respectively. In non-cancerous liver, the degree of fibrosis was stage 1 in 11% of the patients, stage 2 in 17%, stage 3 in 21%, and stage 4 (i.e., liver cirrhosis) in 51%. The prevalence of cirrhosis was significantly lower among male patients (39%) compared with female patients (70%) (P = 0.008). This gender difference is very interesting and important. Further studies are needed to clarify the pathogenesis of this issue. According to previous studies, 10–75% of NASH-related HCCs occur in non-cirrhotic NASH.22,27,33,124,127–133 The high incidence of HCC arising from non-cirrhotic NASH may be partly due to the fact that the diagnosis of NASH is based on histology and liver tissue can only be obtained in patients with preserved liver function by liver biopsy or surgery. Moreover, end-stage cirrhotic NASH cannot be diagnosed with any confidence due to its “burned out” histology. Further studies are required to clarify the true incidence of HCC arising from non-cirrhotic NASH.

We compared the clinical features of 34 NASH-related HCC patients with 56 age- and sex-matched patients who had HCV-related HCC and similar treatment.134 As expected, there was a significantly higher prevalence of obesity, diabetes, and dyslipidemia in the NASH group. Transaminases were significantly higher in the HCV-related HCC patients, while the GGTP level was significantly higher in the patients with NASH-related HCC (AST (IU/L) 42 in NASH-HCC/62 in HCV-related HCC, ALT (IU/L) 36/62, GGTP (IU/L) 102/55, respectively). Concerning tumor markers, 35.3% of patients with NASH-related HCC showed elevation of α-fetoprotein (AFP) and 52.9% showed elevation of des-γ-carboxy prothrombin (DCP) (Fig. 3). In contrast, 69.6% of the patients with HCV-related HCC showed elevation of AFP and 41.3% showed elevation of DCP. These differences of tumor markers between NASH-related HCC and HCV-related HCC are interesting and should be studied further. Among our NASH-related HCC patients, 24 (71%) had a single tumor nodule with a median diameter of 24 mm (8–83 mm). Treatment for NASH-related HCC consisted of surgery in 15 patients (48%), radiofrequency ablation in three (10%), and transcatheter arterial chemoembolization in 13 (42%). Histologically, moderately differentiated HCC with trabecular type was the most common.

Figure 3.

Prevalence of elevation of tumor markers in patients with non-alcoholic steatohepatitis (NASH)-related hepatocellular carcinoma (HCC) and hepatitis C virus (HCV)-related HCC: In NASH related-HCC, des-γ-carboxy prothrombin was more commonly elevated. In contrast, patients with HCV-related HCC frequently showed elevation of α-fetoprotein (AFP). (inline image): NASH-HCC; (inline image): HCV-HCC.

During a median follow-up period of 35.4 months, 13 NASH-related HCC patients died (12 due to HCC or hepatic failure and one due to gastric cancer). The 5-year survival rate was 55.2% in NASH-related HCC versus 50.6% in HCV-related HCC (Fig. 4). The 5-year recurrence rate of 16 NASH-related HCC patients after curative treatment was 69.8% versus 83.1% in the 32 HCV-related HCC patients. According to the Takuma et al. study,128 the recurrence-free survival rates at 1, 3 and 5 years were 72%, 60%, and 60%. Hashizume et al.124 reported nine NASH-related HCC patients and after an average follow-up of 32 months, there was no recurrence except one patient. These results were different from our findings. Recently, Wakai et al.132 reported that postoperative morbidity and 30-day mortality rates in HCC patients with NAFLD were significantly higher than those in HCC patients with viral hepatitis; however, patients with NAFLD showed better disease-free survival on univariate (P = 0.048) and multivariate (P = 0.020) analyses. Further studies are required to elucidate the recurrence of NASH-related HCC.

Figure 4.

Survival curve of 34 non-alcoholic steatohepatitis (NASH)-hepatocellular carcinoma (HCC)patients and 56 hepatitis C virus (HCV)-HCC patients. The survival rates at 1, 3, and 5 years were 93.5%, 82.5%, 55.2% in NASH-HCC and 83.0%, 60.0%, 50.6% in HCV-HCC, respectively. (inline image): HCV-HCC; (inline image): NASH-HCC.

Zen et al.113 have reported a NASH patient with HCC for whom liver histology suggested a multicentric origin. Their patient developed four nodules at different times after the diagnosis of NASH. Liver biopsies were performed from each nodule. The first nodule waspathologically diagnosed as a pseudolymphoma, the second moderately differentiated HCC, the third well-differentiated HCC, and the fourth a dysplastic nodule.

Hepatocellular carcinomain NASH may be of multicentric origin, similar to HCC based on viral hepatitis.133 HCC recurrence was very high in patients with NASH in our study. Regular screening for HCC is extremely important for NASH patients with HCC even if they have received curative treatment.


PATIENTS WITH END-STAGE cirrhotic NASH have “burned-out NASH”, leading to a diagnosis of cryptogenic cirrhosis.135 It has been acknowledged that a substantial proportion of patients with cryptogenic cirrhosis may actually have previously unrecognized NASH, since cryptogenic cirrhosis patients have a high prevalence of obesity and/or type 2 diabetes82,136–138 and because these patients often develop NASH after liver transplantation (recurrence of NASH?).138

Ratziu et al.139 analyzed retrospectively 27 obesity-related cryptogenic cirrhosis (suspected of burned-out NASH), 10 lean patients with cryptogenic cirrhosis and 391 patients with LC-C. Patients with obesity-related cryptogenic cirrhosis were older than patients with LC-C at the time of development of cirrhosis (62.1 versus 53.7, P < 0.001) and also older at the time when HCC was detected (66.8 versus 63.4). The survival of patients with obesity-related cryptogenic cirrhosis was lower than that of age- and sex-matched patients with untreated LC-C and 27% of them developed HCC. Ratzu's results are worse than those for patients with cirrhotic NASH, suggesting that burned-out NASH is probably the very end-stage of cirrhotic NASH.

Bugianesi et al.140 conducted a case-control study of HCC in patients with cryptogenic cirrhosis, viral cirrhosis, and alcoholic cirrhosis. They demonstrated that the prevalence of obesity, diabetes and other features suggestive of metabolic syndrome was significantly higher in patients with HCC in cryptogenic cirrhosis. Other retrospective studies have yielded similar results.141,142 It seems that a substantial number of HCCs arising in cryptogenic cirrhosis may have occurred in patients with “burned-out NASH”. Prospective cohort studies of biopsy-proven NASH are needed to reveal the clinical features of end-stage cirrhotic NASH (burned-out NASH) and its association with cryptogenic cirrhosis.

Paradis et al.143 analyzed HCC patients with features of metabolic syndrome as the only risk factor for liver diseases (MS group, n = 31), HCC patients with overt causes of chronic liver disease (n = 81) and cryptogenic group (n = 16). The patients with MS group were older (mean age: 67.7), and the non-cancerous liver was significantly more often free of significant fibrosis (F0–F2: 65%). In addition, the HCC associated with MS group were more often well differentiated (65%), and some of them arise through malignant transformation of a pre-existing liver cell adenoma. Regimbeaua et al.140 also found that well-differentiated tumors were significantly more common in patients with cryptogenic liver disease (89%). These interesting features also remain to be clarified.

In conclusion, the exact mechanisms leading to development of HCC in patients with NASH remain unclear. In addition to older age and fibrosis, the pathophysiological factors related to development of NASH itself, such as insulin resistance, oxidative stress, and inflammatory cytokines, are likely to contribute to hepatocarcinogenesis. In the natural history of cirrhotic NASH, liver failure (mainly due to HCC) is the main cause of morbidity and mortality. In patients with HCC, tumor recurrence and de novo HCC should be considered. Regular screening for HCC is extremely important especially in patients with cirrhotic NASH and NASH patients with HCC.