Introduction

Hepatocellular carcinoma (HCC) is the fifth most common cancer and the third most common cause of cancer-related deaths worldwide [1–3]. Various factors, including chronic viral infection, age, sex, alcoholic liver disease (ALD), and diabetes mellitus, are involved in HCC development and progression. Hepatitis C virus (HCV) and/or hepatitis B virus (HBV) infections are the major cause of HCC development in Japan [4–6]. Recently, the incidence of HCC without HBV or HCV infection has increased, accounting for 12–16 % of HCC patients in Japan [7, 8]. Obesity is the leading cause of non-alcoholic fatty liver disease (NAFLD) [9] and represents the second major cause of cancer development in western countries and Japan [10, 11]. Almost 25 % of all HCC cases in western countries, and 8–15 % in Japan are estimated to originate from NAFLD [12, 13]. Because effective predictive biomarkers for non-B non-C (NBNC) HCC development in NAFLD patients have not been clarified, NBNC-HCC is often diagnosed in advanced stages with poor prognosis [14].

There is compelling evidence that host genetic polymorphisms are involved in HCC pathogenesis and may be useful as predictive biomarkers [15]. The associations of functional genetic polymorphisms of interleukin 6 and tumor necrosis factor alpha with HCC development have been reported [16, 17]. In the present study, we focused on the association of patatin-like phospholipase domain-containing protein 3 (PNPLA3) genetic polymorphisms with HCC prognosis. PNPLA3, also called as adiponutrin, is a member of the patatin-like phospholipase family and its activity is regulated by the same hormonal pathways that regulate fat deposition in the liver [18–20]. Non-synonymous single nucleotide polymorphism (SNP) rs738409 C>G transversion (Ile148Met) in exon 3 of the gene encoding for PNPLA3 has been reported to be associated with NAFLD development [21]. Moreover, the PNPLA3 rs738409 SNP is related to the degree of fibrosis observed in ALD and NAFLD [22–24]; however, it is not clear whether this polymorphism affects HCC development and prognosis in NAFLD and ALD. The present study was conducted to determine whether rs738409 C>G non-synonymous PNPLA3 SNP is associated with HCC development and prognosis in HBV, HCV, or NBNC patients.

Methods

Patients

The present study enrolled 638 consecutive Japanese patients newly diagnosed with HCC at the Okayama University Hospital between 2001 and 2010. Liver disease etiologies were as follows: 462 patients with HCV-related disease, 72 with HBV-related disease, and 104 with non-B non-C (NBNC) disease. NBNC etiologies included metabolic disorders such as NAFLD in 57 patients, ALD in 34 patients, and autoimmune liver diseases and cryptogenic liver injury in 13 patients. Heavy drinking behavior was defined as daily alcohol intake of at least 70 g for ALD patients, and non-drinking behavior was defined as daily alcohol intake of less than 20 g for NAFLD patients. Fifty NAFLD patients without HCC were also included in the study. The study was conducted in accordance with the Declaration of Helsinki, and all protocols were approved by the ethics committees of the institutes. All patients provided informed consent before study enrollment.

HCC diagnosis and follow-up

In accordance with the clinical practice manual from the Japan Society of Hepatology [25, 26], HCC diagnosis was accomplished using imaging techniques, including ultrasonography, computed tomography, magnetic resonance imaging, hepatic angiography, and/or tumor biopsy, in combination with the detection of serum levels of alpha-fetoprotein (AFP), AFP-L3, and des-gamma-carboxy prothrombin (DCP). HCC staging conformed to the standards of general rules for clinical and pathological study of primary liver cancer by the Liver Cancer Study Group of Japan [27]. The cancer was treated following suitable measures according to the treatment algorithm for HCC in the clinical practice manual from the Japan Society of Hepatology [18, 19]. After curative treatment of primary HCC, all patients underwent liver function tests every 1–3 months and ultrasonography or dynamic study every 3–4 months. Serum levels of AFP, AFP-L3, and DCP were also determined every 1–3 months. HCC recurrence was diagnosed using the same criteria as those used for primary HCC diagnosis. HCC recurrence was treated following suitable measures.

SNP genotyping

Genomic DNA was extracted from whole blood samples using the QIAamp DNA Mini Kit according to the manufacturer’s protocol (Qiagen, Tokyo, Japan). PNPLA3 SNP was genotyped using the TaqMan predesigned SNP genotyping assays, as recommended by the manufacturer (Applied Biosystems, Tokyo, Japan).

Statistical analysis

Data are expressed as mean ± standard deviation or median (range). Patient laboratory data were compared among the groups using the Kruskal–Wallis test. Survival rates of patients were estimated using the Kaplan–Meier method and compared with the log-rank test. P < 0.05 was considered significant. Statistical analysis was performed using JMP software (SAS Institute, Cary, NC, USA).

Results

Patient characteristic at diagnosis

HCC development occurred at a much younger age in HBV patients than in HCV or NBNC patients (P < 0.001, Kruskal–Wallis test, Table 1). The majority of NBNC and HBV patients were male (P < 0.001). Body mass index (BMI) of NBNC patients was significantly higher than that of HBV or HCV patients (P = 0.001). The NBNC patients also exhibited the highest incidence of diabetes mellitus (38.5 %). The incidence of diabetes mellitus was much lower among HCV (18.8 %) and HBV patients (4.0 %, lowest, P < 0.001). In terms of drinking behavior, 32.7 % of NBNC patients drank heavily, whereas only 5.5 % of HBV patients and 11.0 % of HCV patients (P < 0.001) were heavy drinkers. No significant differences were observed among the experimental groups regarding the ratio of patients with Child–Pugh A (P = 0.44). Significantly higher serum levels of aspartate aminotransferase and alanine aminotransferase as well as lower platelet counts were observed in HCV patients (P < 0.001, 0.003, and <0.001, respectively). Gamma-glutamyl transpeptidase was significantly higher in NBNC patients than in HBV or HCV patients (P < 0.001).

Table 1 Patient characteristics
Full size table

HCC diagnosis and prognosis

At diagnosis, NBNC patients exhibited a higher incidence of large HCC tumors of advanced TNM stages as well as higher serum DCP levels than HBV or HCV patients (P < 0.001, Table 1). After curative treatment of primary HCC, all patients underwent periodical liver function and imaging tests at follow-up. The mean follow-up period was approximately 3.0 years (range 0.1–9.5 years). This period was defined as the time between diagnosis and death or the latest confirmation of survival. The accumulated observation time was 1,914 person-years. The survival rate of NBNC patients was significantly lower than that of HBV or HCV patients (P = 0.011, log-rank test, Fig. 1), with an estimated 68.7 % survival at 3 years and 41.8 % survival at 5 years. The survival of HCV or NBNC patients was not dependent on TNM stage (Fig. 2). The survival of HBV patients on TNM stage II was significantly better than that of HCV or NBNC patients (P = 0.044).

Fig. 1
figure 1

Patient survival rates in the three groups. The patient survival rate was presumed for each group by the Kaplan–Meier method. The survival of NBNC patients was significantly poorer than that of HBV or HCV patients (P = 0.011, log-rank test), and it was estimated as 68.7 % at 3 years and 41.8 % at 5 years

Full size image
Fig. 2
figure 2

Comparison of survival rates among the three groups in different TNM stages. By dividing the patients on the basis of the TNM stage at diagnosis, the patient survival rate was compared among the three groups of HBV, HCV, and NBNC etiology for each TNM stage. No prognostic differences were observed among the three groups for TNM stage I, III, or IV (P = 0.13, 0.36, and 0.64, respectively, log-rank test). The survival of HBV patients on TNM stage II was significantly better than that of HCV or NBNC patients (P = 0.044)

Full size image

Subgroup analysis based on PNPLA3 rs738409 SNP

The ratio of patients with the G/G genotype of PNPLA3 SNP was significantly greater among NBNC patients (46.2 %) than among HBV or HCV patients (16.7 and 24.0 %, respectively, P < 0.001, Table 1). Age and gender did not significantly correlate with PNPLA3 SNP genotype distribution among HBV, HCV, or NBNC patients. Among NBNC patients, patients with the G/G genotype of PNPLA3 SNP had significantly higher BMI than those with the C/C or C/G genotype (P < 0.001, Table 2), whereas no significant correlation was observed between HBV or HCV patients regarding BMI and PNPLA3 SNP genotypes. As for laboratory data, NBNC patients with the G/G genotype exhibited a slight tendency toward smaller tumors (P = 0.047), and higher aspartate aminotransferase to platelet ratio index (APRI) (P = 0.054) than those with C/C or C/G genotype; however, no significant differences were observed in the TNM stage and serum levels of AFP and DCP. We also compared patient survival to the distribution of the different PNPLA3 SNP genotypes (Fig. 3). No significant correlation was observed between survival and the genotypes among HBV, HCV, or NBNC patients. For HBV patients, the viral status including HBe antigen, HBs antigen, and HBV DNA at diagnosis did not differ on the basis of the different PNPLA3 SNP genotypes. We further investigated ALD or NAFLD patients among NBNC patients. The ratio of patients with the G/G genotype of PNPLA3 SNP was significantly greater among NAFLD patients (56.1 %) than the other groups with ALD or other etiologies (35.3 and 30.8 %, respectively, P = 0.046, Table 3). As presented in Fig. 4, ALD patients with the G/G genotype of PNPLA3 SNP who had BMI of less than 25 kg/m2 exhibited significantly lower survival rates than those who had BMI of at least 25 kg/m2 (P = 0.028, log-rank test). A similar association was not observed for ALD patients with the C/C or C/G genotype or NAFLD patients. To investigate the association of PNPLA3 SNP with carcinogenesis, the NAFLD patients with HCC were compared to those without HCC in terms of the prevalence of the G/G genotype of PNPLA3 SNP. The genotypes of PNPLA3 SNP in the NAFLD patients without HCC were C/C in 3 patients (6.0 %), C/G in 22 patients (44.0 %), and G/G in 25 patients (50.0 %), which did not differ significantly from those in the NAFLD patients with HCC (P = 0.61).

Table 2 Characteristics of NBNC patients with PNPLA3 rs738409 polymorphisms
Full size table
Fig. 3
figure 3

Comparison of survival rates based on the different PNPLA3 SNP genotypes in the three groups of HBV, HCV, and NBNC etiology. The PNPLA3 SNP genotype was not significantly related to survival rates among HBV, HCV, or NBNC patients

Full size image
Table 3 Frequency of PNPLA3 rs738409 polymorphisms in NBNC patients
Full size table
Fig. 4
figure 4

Comparison of survival rates based on different BMI for NBNC patients with ALD or NAFLD showing the impact of PNPLA3 SNP on survival. Among ALD patients with the G/G genotype, patients with BMI of less than 25 kg/m2 had significantly poorer survival than those with BMI of at least 25 kg/m2 (P = 0.028, log-rank test). ALD patients with the C/C or C/G genotype of PNPLA3 or NAFLD patients showed no prognostic difference in BMI

Full size image

Discussion

HCC is commonly associated with HBV or HCV infections; however, the incidence of NBNC-HCC with NAFLD or ALD has increased significantly in recent years [12, 13]. The incidence of NBNC-HCC in the present study supports the trend that non-viral-associated HCC is increasing as the ratio of NBNC-HCC has increased from 11.8 % between 2001 and 2005 to 15.1 % between 2006 and 2010. Several studies have attempted to elucidate useful predictors for NBNC-HCC; however, these studies were not completely successful. Recent genome-wide association studies revealed that PNPLA3 rs738409 SNPs are associated with hepatic steatosis and fibrosis [21, 28]. PNPLA3 (rs738409 C>G) SNP can modulate hepatic inflammation and fibrosis progression [29]. The G allele of PNPLA3 SNP is associated with increased steatosis [30–32] and severe fibrosis [22, 27, 33–35]. Liver cirrhosis and steatohepatitis are associated with carcinogenesis and increased susceptibility to HCC development; however, no published studies have focused on the role of PNPLA3 SNP genotypes in HCC development and prognosis. Therefore, we investigated PNPLA3 SNP genotypes in HCC patients to clarify the importance of this polymorphism as a predictor for HCC development and prognosis.

Our results of patient characteristics at HCC diagnosis revealed that NBNC-HCC patients exhibited a significantly higher incidence of the G/G genotype than HCV or HBV patients with a higher incidence of diabetes mellitus, obesity, and alcoholic liver injury, supporting the results of previous studies [36–38]. Furthermore, subgroup analysis revealed that the NBNC-HCC patients with the G/G genotype had higher BMI and APRI, reflecting obesity and severe fibrosis, than those with the other genotypes, and that the patients with NAFLD had a higher frequency of the G/G genotype than those with ALD. In contrast, among HBV or HCV patients with HCC, the presence of obesity or severe hepatic fibrosis showed no significant correlation with the PNPLA3 SNP genotypes. Therefore, these results indicated that among the patients with HCC, the G/G genotype of PNPLA3 SNP is strongly associated with NAFLD with obesity and severe fibrosis, which is a similar situation in those without HCC in the previous studies.

Our results of HCC prognosis clearly showed that no significant association was observed between HCC prognosis and the PNPLA3 SNP genotypes in HCV or NAFLD patients. Interestingly, subgroup analysis of ALD patients indicated prognostic differences according to the genotypes of PNPLA3 SNP; the ALD patients with the G/G genotype of PNPLA3 SNP and high BMI might maintain body weight against HCC progression and malnutrition, and resulted in better prognosis, by comparing to the ALD patients with low BMI, although the number of ALD patients might be relatively small.

In conclusion, our results indicate that differences in the distribution of PNPLA3 SNP genotypes have important and independent roles in the regulation of various liver disease etiologies and HCC prognosis. While the G/G genotype showed greater distribution in NBNC-HCC patients, only among ALD patients was the G/G genotype, but not the C/C and C/G genotypes, associated with HCC prognosis. More research is necessary to further clarify the exact manner in which differences in expression of PNPLA3 SNP genotypes may regulate liver etiology as well as HCC development and prognosis. Longitudinal study of HCC development in patients without HCC is required.