Journal of Gastroenterology

, 46:1230

Hepatocellular carcinoma in Japanese patients with nonalcoholic fatty liver disease, alcoholic liver disease, and chronic liver disease of unknown etiology: report of the nationwide survey

Authors

  • Katsutoshi Tokushige
    • Department of Internal Medicine and GastroenterologyTokyo Women’s Medical University
    • Department of Internal Medicine and GastroenterologyTokyo Women’s Medical University
  • Yoshinori Horie
    • Department of Internal MedicineInternational University of Health and Welfare, Sanno Hospital
  • Makiko Taniai
    • Department of Internal Medicine and GastroenterologyTokyo Women’s Medical University
  • Susumu Higuchi
    • National Hospital Organization Kurihama Alcoholism Center
Original Article—Liver, Pancreas, and Biliary Tract

DOI: 10.1007/s00535-011-0431-9

Cite this article as:
Tokushige, K., Hashimoto, E., Horie, Y. et al. J Gastroenterol (2011) 46: 1230. doi:10.1007/s00535-011-0431-9

Abstract

Purpose

To clarify the etiology of hepatocellular carcinoma (HCC) in Japanese patients with non-viral liver disease, we performed a nationwide survey. The influence of obesity, lifestyle-related diseases, and alcohol consumption was focused on.

Methods

A nationwide survey of 14,530 HCC patients was conducted in 2009. Clinical features were studied for HCC patients with nonalcoholic fatty liver disease (NAFLD-HCC; n = 292), alcoholic liver disease (ALC-HCC; n = 991), and chronic liver disease of unknown etiology (unknown HCC; n = 614). The unknown HCC was divided into two subgroups, a no alcohol intake group and a modest alcohol intake group.

Results

ALC-HCC accounted for 7.2% of all HCC, followed by unknown HCC (5.1%) and NAFLD-HCC (2.0%). The characteristics of these three groups were clearly different (median age was 72 years for NAFLD-HCC, 68 years for ALC-HCC, and 73 years for unknown HCC, p < 0.01; female gender was 38, 4, and 37%, respectively, p < 0.01). Obesity and lifestyle-related diseases were significantly more frequent in NAFLD-HCC than in ALC-HCC and unknown HCC. The no alcohol intake subgroup of unknown HCC showed female predominance (58%) and was older, without a high prevalence of obesity and lifestyle-related diseases. In contrast, the modest alcohol intake subgroup showed the same trends regarding gender, body mass index, prevalence of lifestyle-related diseases, and liver function as the ALC-HCC group.

Conclusions

The clinical features of ALC-HCC, NAFLD-HCC, and unknown HCC were clearly different. Modest intake of alcohol might have a more significant role in hepatic carcinogenesis than is presently thought.

Keywords

Hepatocellular carcinomaAlcoholic liver diseaseNonalcoholic fatty liver diseaseQuestionnaire survey

Introduction

Primary liver cancer is the fifth most common cancer worldwide and the third most common cause of cancer mortality [13]. According to the most recent nationwide Japanese registry data, primary liver cancer ranked third for men and fifth for women as a cause of death from malignancy [4]. Hepatocellular carcinoma (HCC) accounts for about 90% of primary liver cancer. With respect to the underlying liver disease, the latest nationwide report of the Liver Cancer Study Group of Japan showed that hepatitis C virus (HCV)-related liver disease is the most common underlying cause of HCC [5]. HCV-related HCC accounts for 67% of all HCC, followed by 15% for hepatitis B virus (HBV)-related HCC. The incidence of HCV-related HCC has been gradually decreasing in recent years, whereas the incidence of HCC associated with non-viral chronic liver disease has gradually been increasing.

Recently, obesity and diabetes have been shown to be risk factors for HCC in both large cohort studies and experimental studies [68]. The increased risk of HCC associated with obesity and diabetes is probably due to two factors, which are an increased prevalence of nonalcoholic fatty liver disease (NAFLD) and the carcinogenic potential of the two conditions. Because of the rapid increase in NAFLD in Japan due to a dramatic increase of obesity and metabolic syndrome, HCC associated with NAFLD may be the main reason for the increase of HCC among patients with non-viral chronic liver disease. Also, a synergistic effect of NAFLD and alcoholic liver disease may promote the development of HCC.

Historically, steatosis and steatohepatitis have been divided into two categories: alcoholic and nonalcoholic. Nonalcoholic liver disease is usually defined by a daily alcohol consumption of less than 20 g in women and less than 30 g in men, because alcohol-induced steatosis and steatohepatitis can occur above these thresholds [9]. However, there is no clear consensus regarding the threshold alcohol consumption for defining nonalcoholic steatosis and steatohepatitis. Moreover, the level of alcohol intake associated with the development of alcoholic liver disease differs among individuals and ethnic groups due to differences of alcohol metabolism. The alcohol consumption criterion for defining alcoholic liver disease proposed by the Japanese Study Group on Alcoholic Liver Disease is the intake of more than 70 g/day for more than 5 years [10]. Thus, the cutoff levels for nonalcoholic and alcoholic steatosis/steatohepatitis in Japan are defined as intake of less than 20 g/day and more than 70 g/day, respectively. One problem with these definitions is the lack of a term for patients with steatosis and/or steatohepatitis whose alcohol consumption lies between the above two levels. This suggests the need for a change in nomenclature to terms such as metabolic fatty liver disease or metabolic steatohepatitis, which avoid the need for a definition of “nonalcoholic”.

Against this background, we performed a nationwide survey to study the etiology of HCC and analyze the clinical features of HCC in patients with NAFLD (NAFLD-HCC), alcoholic liver disease (ALC-HCC), or liver disease of unknown etiology (unknown HCC). Alcohol may be a potent carcinogen, even when its intake is low [11]. To investigate the pathogenesis of unknown HCC, the influence of obesity was investigated, as well as the effect of lifestyle diseases and no to modest alcohol intake (below the Japanese criterion for alcoholic liver disease of 70 g/day).

Patients and methods

The Nationwide Survey of HCC was conducted in 2009. We sent questionnaires on the etiology of HCC and case cards for NAFLD-HCC, ALC-HCC, and unknown HCC to hospitals throughout Japan that were approved by the Japanese Society of Gastroenterology. We asked for data on all patients with a diagnosis of HCC between April 2006 and March 2009. A total of 115 hospitals throughout Japan responded to the questionnaire and provided case cards. These institutions are listed in the Appendix. This study was conducted according to the Declaration of Helsinki (2000 version). By studying 14,530 patients with HCC from 115 hospitals nationwide, we accurately determined the liver diseases underlying HCC in Japan.

Etiology of the underlying liver diseases

We enrolled patients with a clinical and/or histological diagnosis of HCC. Table 1 shows a list of the causes of HCC and their definitions. All patients were negative for hepatitis B surface antigen and for anti-HCV antibody and/or HCV RNA by polymerase chain reaction analysis.
Table 1

Clinical diagnosis of liver disease in the patients with HCC and classification of etiology

I. Criteria for diagnosis of HCC

 HCC was diagnosed by abdominal imaging and/or histological findings

II. Criteria for classification of etiology

 1. Hepatitis B virus (HBV): positive for HBs antigen

 2. Heptatitis C virus (HCV): positive for anti-HCV and HCV-RNA

 3. HBV + HCV

 4. Alcohol: modified criteria of the Japanese Study Group of Alcoholic Liver Disease

 5. Nonalcoholic fatty liver disease (NAFLD)—fulfills the following criteria:

  A. Alcohol consumption less than 20–30 g/day

  B. Confirmation of hepatic steatosis (or steatohepatitis) by liver biopsy or abdominal imaging

  C. No other etiology for liver disease

 6. Congestive disease

 7. Metabolic disease (Wilson’s disease, hemochromatosis, etc.)

 8. Autoimmune liver disease [primary biliary cirrhosis (PBC), autoimmune hepatitis (AIH), primary sclerosing cholangitis (PSC)]

 9. Unknown etiology

Diagnosis of NAFLD was based on the following criteria: (1) detection of hepatic steatosis (or steatohepatitis) by liver biopsy or imaging; (2) intake of less than 20–30 g of ethanol per day (as confirmed by the attending physician and family members who were in close contact with the patient); and (3) appropriate exclusion of other liver diseases (such as alcoholic liver disease, viral hepatitis, autoimmune hepatitis, drug-induced liver disease, primary biliary cirrhosis, primary sclerosing cholangitis, biliary obstruction, and metabolic liver diseases such as Wilson’s disease and hemochromatosis) [12, 13]. Alcoholic liver disease was diagnosed according to the modified criteria proposed by Takada et al. [10], and was defined as habitual alcohol consumption over 70 g daily. Patients with alcoholic liver disease were also negative for HBV and HCV. The final diagnosis of HCC and assessment of etiology were done at each participating institution.

Comparison of patients with NAFLD-HCC, ALC-HCC, and unknown HCC

We also collected detailed clinical data (age, gender, body mass index (BMI), obesity, lifestyle-related diseases, and laboratory data) about the patients with NAFLD-HCC, ALC-HCC, or unknown HCC.

All patients were negative for hepatitis B surface antigen and for anti-HCV antibody and/or HCV RNA by polymerase chain reaction analysis. Obesity was defined as a BMI > 25 kg/m2 according to the Japanese Obesity Association criteria [14]. Diagnosis of type II diabetes mellitus was based on the World Health Organization (WHO) criteria. Dyslipidemia was diagnosed if the patient was currently on treatment with lipid-lowering medications or had elevated serum levels of total cholesterol (>220 mg/dl) and/or triglycerides (>150 mg/dl) on at least three occasions. Hypertension was diagnosed if the patient was receiving antihypertensive therapy or had a blood pressure >140/90 mmHg on at least three occasions. Liver cirrhosis was diagnosed on the basis of histological biopsy findings, laparoscopy, or abdominal imaging (left lobe hypertrophy with splenomegaly, nodular changes in liver surface) and laboratory findings (lower platelet count, albumin level, and/or prolonged prothrombin time) compatible with liver cirrhosis. Also taken into account were the clinical findings of esophageal varices, ascites, or hepatic encephalopathy. Metabolic syndrome was diagnosed by applying the criteria of the Japanese Society of Internal Medicine [15].

The following laboratory parameters were measured: albumin, total bilirubin, aspartate aminotransferase (AST), alanine aminotransferase (ALT), gamma-glutamyltranspeptidase (γ-GTP), fasting blood sugar (FBS), hemoglobin A1c (HbA1c), platelet count, and α-fetoprotein (AFP).

Unknown liver disease classified by alcohol consumption (subgroup analysis)

To determine whether modest alcohol intake could influence carcinogenesis in patients with unknown HCC, we divided the patients into a no alcohol group (alcohol consumption <20 g/day) and a modest alcohol intake group (alcohol consumption of 20–70 g/day).

Statistical analysis

Analyses were performed with SPSS version 13.0J software (SPSS Inc., Tokyo, Japan). Data are shown as the median ± standard deviation (SD) or as percentages. The Kruskal–Wallis test or the chi-square test was used to compare data among the NAFLD-HCC, ALC-HCC, and unknown HCC groups. The Mann–Whitney test or the chi-square test was used to compare data between the no alcohol and modest alcohol subgroups of the unknown HCC group. In addition, multilogistic analysis was performed between the no alcohol intake and the modest-alcohol intake subgroups of unknown HCC. A p value <0.05 was considered significant.

Results

Etiology of the underlying liver disease

Among the 14,530 patients with HCC, 2,299 (15.8%) patients were diagnosed as having non-HBV, non-HCV HCC (Table 2). Of the 12,231 patients with viral-related HCC, 2,054 (14.1%) were positive for HBs antigen, 9,640 (66.3%) were positive for HCV-RNA, and 537 (3.7%) were positive for both HBs antigen and HCV-RNA. Among the HCC patients with non-viral liver diseases, ALC-HCC (7.2%) was the most common diagnosis, followed by unknown HCC (5.1%). NAFLD-HCC (2.0%) was the third most common etiology, while HCC associated with autoimmune liver disease (AIH + PBC + PSC) (1.2%) came fourth.
Table 2

Etiology of HCC

Total, n = 14,530

%

HBV

14.1

HCV

66.3

HBV + HCV

3.7

Alcohol

7.2

NAFLD

2.0

Congestive disease

0.1

Metabolic diseasesa

0.1

PBC + AIH + PSC

1.2

Unknown etiology

5.1

NAFLD nonalcoholic fatty liver disease, PBC primary biliary cirrhosis, AIH autoimmune hepatitis, PSC primary sclerosing cholangitis

aWilson’s disease, hemochromatosis, etc.

Comparison among NAFLD-HCC, ALC-HCC, and unknown HCC

Among patients with non-HBV or non-HCV HCC, 292 had NAFLD-HCC, 991 had ALC-HCC, and 614 had unknown HCC. Table 3 shows the characteristic features among the three groups. The median age at the diagnosis of HCC was youngest in the ALC-HCC group at 68 ± 9.1 years (NAFLD-HCC, 72 ± 8.4 years; unknown HCC, 73 ± 10.1 years). The percentage of women was lowest in the ALC-HCC group at 4% (NAFLD-HCC, 38%; unknown HCC, 37%). Regarding lifestyle-related diseases, the prevalence of obesity, diabetes, dyslipidemia, hypertension, and metabolic syndrome were highest in the NAFLD-HCC group. The prevalence of cirrhosis was highest in the ALC-HCC group at 78% (NAFLD-HCC, 62%; unknown HCC, 52%). Serum levels of total bilirubin, AST, ALT, and γ-GTP were significantly higher in the ALC-HCC group compared with the other groups, whereas the platelet count and serum albumin level were lowest. The HbA1C and fasting blood glucose levels were highest in the NAFLD-HCC group.
Table 3

Comparison among NAFLD-HCC, ALC-HCC, and unknown HCC

 

NAFLD-HCC (n = 292)

ALC-HCC (n = 991)

Unknown HCC (n = 614)

p value

Age at diagnosis

72 ± 8.4

68 ± 9.1

73 ± 10.1

<0.001

Gender (female)

38%

4%

37%

<0.001

Obesity (BMI > 25)

66%

37%

35%

<0.001

BMI (kg/m2)

27.0 ± 4.0

23.8 ± 3.7

23.5 ± 4.1

<0.001

Diabetes

70%

49%

43%

<0.001

Hypertension

60%

43%

46%

<0.001

Dyslipidemia

35%

14%

15%

<0.001

Metabolic syndrome

56%

18%

9%

<0.001

Liver cirrhosis

62%

78%

52%

<0.001

Albumin (g/dl)

3.8 ± 0.6

3.6 ± 0.6

3.6 ± 0.6

<0.001

Total bilirubin (mg/dl)

0.9 ± 1.3

1.1 ± 1.9

0.9 ± 1.7

<0.001

AST (IU/l)

40 ± 36

80 ± 301

43 ± 71

<0.001

ALT (IU/l)

35 ± 35

45 ± 176

30 ± 44

0.03

γ-GTP (IU/l)

91 ± 202

147 ± 271

88 ± 198

<0.001

FBS (mg/dl)

119 ± 57

111 ± 63

107 ± 53

<0.001

HbA1c (%)

6.3 ± 1.4

5.9 ± 1.6

5.7 ± 1.4

<0.001

Platelet count (×104/mm3)

14.1 ± 7.4

12.6 ± 8.0

15.2 ± 9.1

<0.001

AFP (ng/ml)

12 ± 427,557

11 ± 368,512

13.0 ± 94,155

0.284

FBS fasting blood sugar

Unknown liver disease classified by alcohol consumption (subgroup analysis)

Eighty-four patients were excluded because we could not obtain detailed information about their alcohol intake. As a result, we compared 316 patients with no alcohol intake and 214 patients with modest alcohol intake. Table 4 shows the characteristic features in these two subgroups. Among the no alcohol subgroup, the prevalence of women was markedly higher (p < 0.001) at 58% versus only 8% in the modest alcohol subgroup. The mean age at the diagnosis of HCC was older in the no alcohol subgroup than the patients with modest alcohol intake subgroup (75.5 vs. 72 years, p < 0.001). The prevalence of obesity and liver cirrhosis were both higher in the no alcohol subgroup, but the prevalence of diabetes, hypertension, dyslipidemia, and metabolic syndrome did not differ significantly between the two subgroups. Liver function was relatively well preserved in the modest alcohol intake subgroup, but their serum γ-GTP level was significantly higher than that of the no alcohol subgroup. Multivariate analysis of six main factors (age, gender, obesity, cirrhosis, serum γ-GTP level, serum albumin level) was performed and the results compared between the two groups. Significant differences were observed in age (p < 0.001), gender (p < 0.001), and serum albumin level (p = 0.008).
Table 4

Comparison between the no alcohol and modest alcohol subgroups of unknown HCC

 

No alcohol intake n = 316

Modest alcohol intake n = 214

p value

Age at diagnosis

75.5 ± 10.2

72 ± 9.0

<0.001

Gender (female)

58%

8%

<0.001

Obesity (BMI > 25)

39%

30%

0.044

BMI (kg/m2)

23.8 ± 4.5

23.5 ± 3.4

0.396

Diabetes

41%

46%

0.214

Hypertension

45%

49%

0.424

Dyslipidemia

15%

15%

0.989

Metabolic syndrome

8%

13%

0.119

Liver cirrhosis

57%

42%

0.001

Albumin (g/dl)

3.6 ± 0.7

3.8 ± 0.6

0.030

Total bilirubin (mg/dl)

0.9 ± 1.5

0.8 ± 1.2

0.266

AST (IU/l)

44 ± 63

39 ± 73

0.081

ALT (IU/l)

29 ± 45

29 ± 42

0.455

γ-GTP (IU/l)

75 ± 184

103.5 ± 213

0.003

FBS (mg/dl)

106 ± 51

110 ± 56

0.050

HbA1c (%)

5.7 ± 1.3

5.7 ± 1.5

0.307

Platelet count (×104/mm3)

14.6 ± 9.0

16.8 ± 8.7

0.001

AFP (ng/ml)

13.3 ± 77,396

10 ± 31,196

0.378

FBS fasting blood sugar

Discussion

In this study, we investigated the etiology of HCC in Japan. Among all HCC patients, ALC-HCC accounted for 7.2%, followed by unknown HCC (5.1%) and NAFLD-HCC (2.0%). A 2008 nationwide retrospective survey of the etiology of cirrhosis [16] found that HCC associated with cirrhotic NAFLD accounts for 1.6% of all HCC with cirrhosis, whereas alcoholic cirrhosis and unknown cirrhosis account for 6.3 and 3.8%, respectively. These findings about the etiology of HCC with cirrhosis in Japan are similar to those of our national survey. Regarding the etiology of HCC in western countries, NAFLD-HCC has been reported to account for 3.8–13% of all HCC [17, 18], whereas ALC-HCC accounts for 9–38% and unknown HCC for 7% [1921]. In comparison with western countries, the prevalence of NAFLD-HCC and ALC-HCC was lower in Japan. This is not only due to the low incidence of NAFLD-HCC and ALC-HCC per se, but also to the high incidence of hepatitis virus-related HCC in Japan. In fact, the incidence of NAFLD-HCC in Japan is expected to increase in the future due to the rising prevalence of NAFLD associated with obesity and/or diabetes.

Comparison of NAFLD-HCC, ALC-HCC, and unknown HCC revealed several findings. (1) ALC-HCC patients were predominantly men and were younger than other groups at the onset of HCC. (2) Obesity and lifestyle-related diseases were significantly more common in the NAFLD-HCC group. (3) Cirrhosis showed a significantly higher prevalence in the ALC-HCC group, with higher transaminase levels and worse liver function.

Concerning the gender difference in ALC-HCC, the low number of women was attributed to very low alcohol consumption among older Japanese women. Nowadays, Japanese women drink alcohol and alcoholic consumption is similar between men and women in younger generations. In the future, the meaning of gender difference in the development of HCC will become clear. As expected, the prevalence of obesity and lifestyle-related disease in the NAFLD-HCC group was higher than in the ALC-HCC and unknown HCC groups. In these last two groups, the prevalence of obesity, hypertension, and dyslipidemia did not differ compared with the findings for the age- and sex-matched general Japanese population [4]. However, the prevalence of diabetes was significantly higher in the ALC-HCC and unknown HCC groups than in the general population (about 11% in women and 21% in men) or in patients with HCV (23%) [22]. El-serag et al. [23] demonstrated that diabetes increased the risk of HCC in a large cohort of Department of Veterans Affairs patients. This increase in risk was independent of alcoholic liver diseases, viral hepatitis, or demographic features. Together with our data, this suggested that diabetes might be a risk factor for not only virus-related HCC, but also non-viral HCC.

It has been acknowledged that a substantial percentage of patients with cryptogenic cirrhosis have previously unrecognized NAFLD, because cryptogenic patients have a high prevalence of obesity and/or type 2 diabetes [24], and because steatohepatitis occurs frequently after liver transplantation for cryptogenic cirrhosis [25].

To clarify the characteristics of the unknown HCC groups, firstly we divided the unknown HCC group according to the presence of obesity, presence of diabetes, and alcohol consumption and compared the difference of median age and gender (Fig. 1). The comparison between no alcohol and modest alcohol subgroups only showed significant differences. Then, we focused on the subgroup analysis of unknown liver diseases classified by alcohol consumption. The no alcohol subgroup showed distinct female predominance, a very unusual feature, as men generally have higher rates of HCC even in NAFLD-HCC. Moreover, the no alcohol subgroup, which seemed to have pure cryptogenic chronic liver disease, did not show a high prevalence of obesity and lifestyle-related diseases in comparison with NAFLD-HCC. Naugler et al. [26] and Nakagawa et al. [27] reported that the reduction of estrogen-mediated IL-6 was associated with lower liver cancer risk in females. The no alcohol intake subgroup showed female predominance and higher age, suggesting that carcinogenesis in this subgroup might be associated with decreased estrogen due to older age, which may induce greater production of IL-6.

The etiology of unknown HCC could include “burnt-out” nonalcoholic steatohepatitis (NASH) [28], occult HBV infection, HBV carriers with previous seroconversion to HBs antigen [29], and “burnt-out” autoimmune hepatitis [30]. In our hospital, we measured anti-HBc antibody in NAFLD-HCC patients, ALC-HCC patients, and HCV-related HCC patients. The prevalence of anti-HBc antibody was 28% in NAFLD-HCC, 44% in ALC-HCC, and 39% in HCV-HCC patients [31, 32], with the differences among them not being significant. In addition, none of the NAFLD-HCC patients or ALC-HCC patients had high HBc antibody titer. Therefore, we think that, even if HBV did influence carcinogenesis in non-viral HCC, the influence would be minimal. Unfortunately in this survey, we did not assess the prevalence of HBc antibody and auto-antibodies. To evaluate the etiology of unknown HCC, further studies will be needed.
https://static-content.springer.com/image/art%3A10.1007%2Fs00535-011-0431-9/MediaObjects/535_2011_431_Fig1_HTML.gif
Fig. 1

Median age (a) and the prevalence of women (b) in the unknown HCC group stratified by alcohol intake (no alcohol intake vs. modest alcohol intake), obesity (BMI < 25 vs. BMI ≥ 25), or diabetes (diabetes vs. non-diabetes). The comparison between no alcohol and modest alcohol subgroups only showed significant difference of age and gender

The modest alcohol intake subgroup showed the same trends in regard to gender, BMI, lifestyle-related diseases, and γ-GTP levels as the ALC-HCC group. Median age was higher and prevalence of cirrhosis was lower in the modest alcohol intake group than those of the ALC-HCC group. In the modest alcohol intake subgroup, liver damage may progress slowly, so that HCC develops at an old age.

Alcohol itself may be a potent carcinogen [11]. Asians are often slow at converting acetaldehyde to acetic acid. Thus, a relatively low alcohol intake may lead to the development of non-viral HCC in Asian populations. Ascha et al. [33] reported that NASH patients with cirrhosis had a greatly increased risk of liver cancer, and even social alcohol consumption appeared to be the most significant factor associated with the risk of HCC. However, Suzuki et al. [34] reported that light-to-moderate alcohol intake (10–40 g/day) could protect against hypertransaminasemia. The alcohol consumption criteria for diagnosis of alcoholic liver disease vary around the world [35, 36], and the alcohol consumption criterion for alcoholic liver disease proposed by the Japanese Study Group of Alcoholic Liver Disease is more than 70 g/day. Our data suggest that social or modest intake of alcohol might have a more significant role in hepatic carcinogenesis than is presently thought. In the future, more detailed studies need to be performed, including assessment of alcohol metabolism genotypes.

Acknowledgments

This work was supported in part by a grant-in aid from the Ministry of Health, Labour and Welfare of Japan.

Copyright information

© Springer 2011