Obesity and diabetes as a risk factor for hepatocellular carcinoma

Abstract

Ten percent of patients who undergo resection for hepatocellular carcinoma (HCC) associated with chronic liver disease have no detectable cause for this underlying liver disease. Recent studies have shown that patients with cryptogenic chronic liver disease frequently have risk factors for nonalcoholic fatty liver disease (NAFLD). This study examines the incidence of risk factors for NAFLD in patients with chronic liver disease who underwent resection for HCC. Among 210 patients with chronic liver disease who underwent resection for HCC, 18 (8.6%) had no identifiable cause for the underlying liver disease. These patients were assessed for obesity, diabetes mellitus, and histological features of the tumor and the adjacent liver parenchyma. Comparisons were made with matched patients with alcohol- and chronic-viral-hepatitis-related HCC. The prevalence of obesity (50% vs. 17% vs. 14%), diabetes (56% vs. 17% vs. 11%), aspartate aminotransferase / alanine aminotransferase ratio < 1 (50% vs. 19% vs. 17%), and steatosis > 20% (61% vs. 17% vs. 19%) was significantly higher in patients with cryptogenic liver disease than in patients with alcohol abuse and chronic viral hepatitis (P < 0.0001 for each). Well-differentiated tumors were significantly more common in patients with cryptogenic liver disease (89% vs. 64% in patients with alcohol-related HCC vs. 55% in patients with chronic viral hepatitis-related HCC, P < 0.0001).

In conclusion, the hypothesis that obesity and diabetes mellitus may be important risk factors for cryptogenic chronic liver disease in patients with HCC is supported by the analysis of surgically treated patients. Whether HCC is primarily related to obesity and diabetes mellitus or secondarily to a NAFLD-like parenchymal lesions remains to be clarified. (Liver Transpl 2004;10:S69–S73.)

In Western countries, 75 to 90% of hepatocellular carcinomas (HCC) are associated with chronic liver diseases.1 The most common causes of chronic liver disease-underlying HCC are hepatitis B or C infection and alcohol abuse. Cholestatic, autoimmune, and inherited metabolic disorders are other potential but much more uncommon causes.2 Patients in whom none of these disorders can be identified are considered to have cryptogenic underlying liver disease or may often be characterized as occult alcohol abusers. It must be noted that these patients account for at least 10% of those who undergo resection for with chronic liver disease-associated HCC in Western series.3

Nonalcoholic fatty liver disease (NAFLD) has recently been reported to be a potential cause of severe liver disease in patients without HCC.4 NAFLD refers to a wide spectrum of liver disorders, including steatosis, steatohepatitis (pathological features of alcohol-related liver injury but in patients who do not have excessive alcohol intakes), extensive fibrosis, and occasionally cirrhosis.5 NAFLD is associated with a high prevalence of obesity and diabetes.4 Four recent studies have shown that HCC may represent a late complication of NAFLD-related cirrhosis6–9 and that HCC must therefore be considered as part of the clinical spectrum of liver disease in NAFLD. These studies included patients with widely variable degrees of severity of liver disease (patients with end-stage cirrhosis who could not undergo surgery, cirrhotic patients who underwent resection, and others who underwent liver transplantation). In those who did not have resection or transplantation, the analysis of the nontumoral liver parenchyma has frequently been incomplete.

In this study, we analyzed a homogeneous surgical series of patients with cryptogenic chronic liver disease who underwent surgical resection. In all patients, detailed pathological analysis of the tumor as well as the non-tumor-bearing liver was performed, and associations with risk factors for NAFLD could be examined precisely.

Abbreviations

HCC, hepatocellular carcinoma; NAFLD, nonalcoholic fatty liver disease; HBV, hepatitis B virus; HCV, hepatitis C virus; BMI, body mass index; CVH, chronic viral hepatitis.

Patients and Methods

Between January 1990 and December 1999, 301 patients underwent liver resection for HCC at our institution. The diagnosis of HCC was established on the resected specimen by a senior pathologist (MC) in all studied patients. Liver sections sampled at a distant site of the tumor (as well as those originating from the tumor itself) were reanalyzed in a retrospect for the presence of fibrosis and steatosis and other features of NAFLD. Fibrosis was graded with a score analogous to the METAVIR score.10 Fifty-four (18%) patients did not have evidence of fibrosis, and 37 (12%) had only mild periportal fibrosis. These 91 patients who did not have severe chronic liver disease were therefore excluded from further analysis. The remaining 210 patients had either moderate-to-extensive fibrosis (n = 58) or cirrhosis (n = 152, all Child-Turcotte-Pugh class A) and were included in the study and stratified according to the cause of their chronic liver disease using a comprehensive panel of tests.

Of the 210 patients with chronic liver disease, 192 were found to have at least one recognized cause of chronic liver disease. In most of the patients (107) in this group, more than one cause could be identified. A total of 81 patients had chronic hepatitis B virus (HBV) infection. Eighty-four patients had chronic hepatitis C virus (HCV) infection, including 21 who had concomitant HBV infection. Genetic hemochromatosis (serum iron transferrin saturation level > 45%; massive iron deposition on liver biopsy and/or presence of homozygote C282Y gene mutation) was present in 11 patients, and autoimmune cirrhosis (presence of elevated titers of antinuclear and anti-smooth muscle antibodies) was present in 2 patients. Primary biliary cirrhosis (defined by presence of antimitochondrial antibodies and consistent histological changes), repeated carbon tetrachloride exposure, and Budd-Chiari syndrome were each identified in one patient. No patient had Wilson's disease.

Alcohol consumption was thoroughly evaluated by specialist interview. Alcohol abuse was defined by an average daily consumption of more than 20 g alcohol for women and 30 g for men. If they had stopped using alcohol at the time of interview, patients were considered to have a past history of alcohol abuse if they had had the same level of consumption for a period of more than 2 years within the previous 20 years.11 For the purpose of the present study, all surviving patients were reinterviewed by an additional specialist, a psychologist, and a social worker for ongoing or previous use of alcohol (informed consent was obtain from each patient). Based on these criteria, 59 patients were determined to have alcohol-related liver disease. Several of the patients with alcohol-related liver disease also had other identified risk factors. The remaining 18 (8.6%) patients had no recognizable cause for their chronic liver disease and HCC, and were considered to have cryptogenic chronic liver disease (Table 1).

Table 1. Demographics and Histological Data of the 18 Male Patients With Hepatocellular Carcinoma Developed on Cryptogenic Liver Disease

Patient	Age (years)	BMI (kg/m^2)	Diabetes	Fibrosis*	Steatosis (%)	Ballooning Degeneration	Lobular Infiltration
Abbreviations: Pt., patient; BMI, body mass index. * Fibrosis grade (METAVIR score) 1-4; see Patients and Methods.
1	63	22.5	No	F3	0	—	—
2	75	23.9	Yes	F2	5	—	—
3	76	24.5	Yes	F2	0	—	—
4	54	25.5	Yes	F3	20	—	—
5	73	28.1	Yes	F4	5	—	—
6	75	28.2	No	F2	5	—	—
7	67	28.4	No	F4	30	Moderate	—
8	69	28.4	No	F4	50	Mild	—
9	67	28.7	Yes	F2	50	—	—
10	70	30.0	Yes	F2	20	—	—
11	71	30.4	Yes	F4	20	Moderate	—
12	62	30.7	Yes	F2	50	—	—
13	47	31.4	Yes	F2	70	Moderate	—
14	72	33.1	No	F2	10	—	—
15	72	34.6	No	F2	20	—	—
16	70	35.2	No	F2	30	—	—
17	62	36.3	Yes	F2	10	—	—
18	52	36.9	No	F2	20	—	—

Preoperative body mass index (BMI) was recorded in all patients, and obesity was defined by BMI > 30 kg/m².12 Prevalence of type II diabetes mellitus was based on fasting glucose level.13 Preoperative fasting serum cholesterol and glucose levels as well as liver function tests and alpha-fetoprotein levels were recorded in all patients. These variables recorded in the group of patients with cryptogenic liver disease were compared with those of matched patients in two other groups: patients with well-documented alcohol-related liver disease (and no other cause of liver injury), and patients with HBV- or HCV-related liver disease. Patients were matched for age (5 year strata), gender, and the extent of fibrosis.

Resected specimens were assessed to determine the extent of fibrosis, steatosis grade (expressed as the percentage of hepatocytes containing fat droplets) in all study patients. Due to the pathologic similarities in NAFLD and alcohol-related liver disease, retrospective pathological reanalysis of the non-tumor-bearing liver was performed for patients with cryptogenic and alcohol-related HCC. In these groups, the necro-inflammatory activity (hepatocyte necrosis and ballooning, lobular and portal inflammatory infiltration), the presence of Mallory bodies, or iron deposits were assessed.14 The tumor characteristics (including differentiation) were also determined for all patients.15 Data are expressed as mean ± standard deviation (SD). Comparisons between groups were performed by univariate analysis (using Chi-square test, Fisher exact test, or Student's t test) where appropriate.

Results

The 18 patients with cryptogenic chronic liver disease were males with a mean age of 66 ± 8 years (range, 47 to 75 years). The mean BMI was 29.8 ± 4.2 kg/m² (22.5 to 36.9 kg/m²), and 50% of patients were obese. Type II diabetes mellitus was present in 10 (56%) and had been recognized with a mean interval of 10 years prior to the diagnosis of HCC. Five patients (28%) had both diabetes and obesity, while 14 (78%) had either diabetes or obesity. Cholesterol level was above the normal upper range in (38%). Eight (44%) had an AST/ALT ratio less than 1. The prevalence of these risk factors was more common in the cryptogenic chronic liver disease group than in the matched groups of patients with alcohol-related HCC or chronic viral-hepatitis-related HCC (Table 2). These differences (except for the cholesterol levels) were statistically significant. Steatosis of the non-tumor-bearing liver was present in 16 patients (89%) with cryptogenic chronic liver disease. Steatosis was over 20% in 11 patients (61%). Steatosis was significantly more frequent and more intense in this group than in those with alcohol-related and with chronic viral-hepatitis-related HCC (Table 2).

Table 2. Clinical, Biological, and Pathological Characteristics of Patients With Hepatocellular Carcinoma Developed on Cryptogenic, Alcohol-Related Liver Disease, and Chronic Viral Hepatitis

Characteristic	Cryptogenic n = 18	Alcohol n = 36	CVH n = 36	Cryptogenic vs.
				Alcohol (P)	CVH* (P)
NOTE. Table values are expressed as number of patients (%), or mean ± standard deviation. Patients were matched for age, gender, and degree of fibrosis. Abbreviations: n, number; BMI, body mass index; AST, aspartate aminotransferase; ALT, alanine aminotransferase; NS, not statistically significant; CVH, chronic viral hepatitis. * No statistical difference was found when alcohol-related liver disease was compared with CVH patients.
Obesity (BMI > 30 kg/m^2)	9 (50%)	6 (17%)	5 (14%)	0.01	0.006
Diabetes mellitus type II	10 (56%)	6 (17%)	4 (11%)	0.004	0.009
AST/ALT < 1	9 (50%)	7 (19%)	6 (17%)	0.02	0.05
Alfa-fetoprotein ≥ 20 ng/mL	6 (33%)	20 (55%)	16 (44%)	NS	NS
Cholesterol > 200 ng/dL	7 (38%)	5 (14%)	6 (17%)	0.04	NS
Steatosis present	16 (89%)	14 (39%)	15 (42%)	0.004	0.004
Steatosis > 20%	11 (61%)	6 (17%)	7 (19%)	0.015	0.003
Tumor size (cm)	8.1 ± 5	6.0 ± 4	6.5 ± 2	NS	NS
Multiple tumors	4 (22%)	2 (5%)	4 (11%)	NS	NS
Well-differentiated tumor	16 (89%)	23 (64%)	20 (55%)	0.04	0.01

No patients were found to have all the features consistent with nonalcoholic steatohepatitis, and only 4 patients were found to have mild or moderate hepatocyte ballooning degeneration (Table 1). No patients had Mallory bodies or iron overload on pathologic analysis. Mean alpha-fetoprotein level (890 ng/mL), and the percentage of patients with an alpha-fetoprotein level > 20 ng/mL were comparable in patients with cryptogenic-liver-disease-related HCC (33%), than in those with alcohol-related (55%) and chronic viral-hepatitis-related (44%) HCC. Patients with cryptogenic liver disease were more likely to have multiple tumors (22%) or large tumors (8.1 cm) than those with alcohol-related and viral hepatitis-related HCC (5%, 6 cm and 11%, 6.2 cm, respectively) though these differences were not statistically significant. Well-differentiated tumors were significantly more common in patients with cryptogenic liver disease (89%) than in those with alcohol-related (64%) and viral hepatitis related HCC (55%) (Table 2).

Discussion

The results of this study confirm that patients with HCC and an underlying liver disease of unknown origin frequently have some of the characteristic features of NAFLD as well as significant risk factors for this condition.6–9

In our series, the low proportion of patients who are categorized as having cryptogenic liver disease among those who undergo resection for HCC (9%) was comparable to or lower than that reported by others in patients who did3 or did not undergo surgical resection.2 This low proportion is likely to reflect an intensive preoperative viral screening and a thorough evaluation of alcohol use or abuse (including oriented reevaluation of all living patients) in this analysis. Thus, it is unlikely that patients categorized in the cryptogenic liver disease group in the present series had in fact an unrecognized cause, given currents information regarding risk factors for liver diseases and HCC.

Further strengthening the hypothesis that NAFLD is a risk factor for HCC is the high incidence of diabetes and obesity in the cryptogenic liver disease group as compared with the alcohol-related liver disease patients, patients who may otherwise share most of the features of NAFLD. To avoid the anticipated difficulties for differentiating the diagnosis of NAFLD from alcohol-related disorders in the present series, we included reevaluation of clinical and pathologic features in these specific groups. When possible, reinterview concerning present and past alcohol consumption was also performed.9, 11

Interestingly, patients in the cryptogenic liver disease group had a high prevalence of risk factors for NAFLD. In this group, 89% had steatosis, 56% had type II diabetes, 50% were obese, and 50% had an AST/ALT ratio < 1, suggesting that even if they did not meet all the pathological criteria for NAFLD, at the time they had end-stage liver disease, it seems likely that most of them could have a past history of NAFLD. The impact of obesity and diabetes, present in 50% and 56%, respectively, of the patients in the cryptogenic disease group, might be underestimated, since these conditions could also have been significant risk factors for the development of HCC in patients with other causes of cirrhosis.9, 16 For example, 14% of our patients with chronic viral hepatitis were obese, and 11% had type II diabetes, proportions that are twice as high as those in the general French population (8% and 5%, respectively).17, 18 The incidence of diabetes and obesity in alcohol-related HCC patients this study is even more striking:17% and 17%, respectively. Moreover, it must be kept in mind that only BMI at the time of resection was considered and that a past history of overweight was not taken into account, which could also contribute to an underestimation of the true impact of obesity on chronic liver disease and HCC. The finding that HCCs associated with cryptogenic liver diseases were larger and more frequently multiple raises the concern that patients in this group could have been referred later to specialized centers.

Some of the characteristic features of NAFLD, such as the presence of parenchymal inflammation, hepatocyte necrosis or ballooning degeneration, were absent in the majority of the patients with cryptogenic liver disease,19, 20 This discrepancy could be anticipated in the present series in which all patients had extensive fibrosis or cirrhosis. Indeed, the development of fibrosis tends to make these pathological manifestation in patients with NAFLD disappear.4

Whether the development of HCC is related to obesity and diabetes or to underlying NAFLD- related changes remains unclear.21 Obesity and diabetes have been shown to increase the risk of uterus, breast, colon, and pancreas malignancies.22–26 Until recently, only three cases of HCC have been reported in NAFLD patients.20, 27, 28 Since then, four papers have specifically analyzed the occurrence of HCC in NAFLD.6–9 In these series of patients with well-documented NAFLD-related cirrhosis, the risk of HCC was estimated to be 7% to 27%.6, 7, 9 A predominance of males and well-differentiated HCC were common findings in these series.6, 9

In conclusion, this series of patients with surgical resection for cryptogenic-liver-disease-related HCC shows that these patients frequently exhibit the same risk factors as those for NAFLD. These results support the hypothesis that NAFLD is a significant risk factor for HCC. Therefore, the diagnosis of HCC might be considered in obese and/or diabetic patients with liver nodules, even if they do not have overt manifestations of chronic liver disease. Finally, features of past or present NAFLD should always be accurately assessed before labeling as occult alcohol abusers those patients who have HCC and who deny alcohol consumption.