Abstract
Background/Aim: Fibroblast growth factor (FGF), vascular endothelial growth factor, and hepatocyte growth factor play a critical role in the pathogenesis of hepatocellular carcinoma (HCC). Materials and Methods: We assessed nine single nucleotide polymorphisms (SNPs) in the FGF1, FGF2, FGF receptor (FGFR)-2, Flt-1, and c-MET genes in 245 HCC patients and 483 chronic hepatitis B virus (HBV) carriers without HCC. Results: Kaplan–Meier analysis showed that patients with the FGF2 rs308447 TT genotype had shorter overall survival than patients with the CC or CT genotype (p=0.016) and that FGF2 rs308379 A allele carriers had shorter overall survival than patients with the TT genotype (p=0.020). Conclusion: Multivariate Cox proportional analysis revealed that the FGF2 rs308379 A allele (hazard ratio(HR)=1.663, p=0.004) and advanced tumor stage (HR=3.430, p<0.001) were independent prognostic factors for overall survival in patients with HCC.
- Fibroblast growth factor
- fibroblast growth factor receptor
- single nucleotide polymorphism
- hepatocellular carcinoma
- hepatitis B virus
Hepatocellular carcinoma (HCC) is a devastating disease that is common worldwide and has the third highest mortality rate among cancers. Hepatitis B virus (HBV), hepatitis C virus, and heavy alcohol drinking are well known risk factors of HCC (1). Chronic HBV infection, in particular, accounts for 70-80% of HCC cases in HBV endemic areas such as Korea (2, 3). The pathogenesis of HCC involves a complex multistep process, including the activation of growth factor signaling pathways (4).
HCC is a highly vascularized tumor. Among the widely recognized angiogenic factors, such as vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF), angiopoietin, and hepatocyte growth factor (HGF) (5), VEGF is considered as the most potent angiogenic factor in HCC (6). VEGF-A and its receptor (FMS-like tyrosine kinase [FLT1]) have been identified as major mediators in VEGF signaling pathways (7). VEGF binds to its receptors, VEGFR1/FLT1 and VEGFR2, and subsequently activates several signaling pathways involved in the proliferation, migration, and invasion of endothelial cells (8).
FGF2 is also a potent angiogenic molecule that is involved in tumor progression (9, 10). Although VEGF is the key driver of tumor angiogenesis, there is a crosstalk between VEGF and FGF signaling in angiogenesis (11, 12), and FGF can act synergistically on VEGF signaling pathways to induce angiogenesis. HGF is a potent mitogen for hepatocytes. FGF and HGF control the proliferation and invasion of liver cancer cells (13, 14). HGF is the ligand for the MET receptor (15), the MET signaling pathway has been found to be activated in approximately 50% of advanced HCC cases (16). Overexpression of c-met defines a subgroup of HCC with poor prognosis and an aggressive phenotype (17).
Although the understanding of HCC pathogenesis has increased, the exact mechanism underlying tumor development and the host genetic factors determining prognosis are not yet defined. Genome-wide association studies have indicated that several polymorphic variants are associated with HCC (18). Recently, a single nucleotide polymorphism (SNP) 61*G (rs4444903) in the epidermal growth factor gene was reported to be associated with increased epidermal growth factor expression and elevated risk of HCC development in cirrhotic Caucasian patients (19, 20). Another report suggested that the determination of VEGF and VEGFR genotypes might play a role in the prediction of clinical outcome of HCC patients receiving sorafenib (21).
There is evidence that SNPs in germ-line DNA may contribute to an individual's susceptibility to HCC or to the prognosis of the disease. The aim of this study was to determine whether SNPs in genes that encode for proteins involved in growth factor signaling pathways can influence the development or progression of HCC and the survival of patients with HBV-associated HCC.
Materials and Methods
Study subjects. The case-control population included a total of 728 Korean patients who were admitted at the outpatient clinic of the Gastroenterology Department of Ajou University Hospital (Suwon, South Korea) between June 2000 and February 2006. They were divided into three groups according to their HBV infection status, clinical data, and serological profile: the chronic hepatitis B group, the HBV-associated liver cirrhosis group, and the HBV-associated HCC group. The chronic hepatitis B group included 293 patients with chronic hepatitis B (male patients=214; female patients=79; mean age, 38.0 years). One hundred and ninety patients were diagnosed as having liver cirrhosis (male patients=144; female patients=46; mean age, 45.8 years), based on the typical morphological findings on computed tomography or ultrasonography, and on evidence of portal hypertension or corresponding laboratory features. The HCC group included 245 HCC patients (male patients=185; female patients=60; mean age, 53.9 years). Patients were defined as having HCC if they had a tumor with a maximum diameter greater than 2 cm, serum α -fetoprotein (AFP) level greater than 200 ng/ml, and typical features of HCC observed using a dynamic imaging technique. If the findings were not characteristic or if the vascular profile was not typical, a biopsy was performed (22). All the subjects were of a single ethnic group, that is, Korean. Informed consent was obtained from each subject, and the Institutional Review Board of Human Research of Ajou University Hospital approved the study protocol (GN3-08-030).
Genotyping. SNPs at nine polymorphic sites were assessed: the FGF1 gene at position rs152524 (C to T); the FGF2 gene at positions rs3088447 (C to T) and rs308379 (T to A); the FGFR2 gene at positions rs1047100 (G to A), rs2981578 (G to A), rs1219648 (T to C), and rs2981582 (C to T); the FLT1 gene at position rs4771249 (G to C); and the MET gene at position rs41736 (C to T) (Table I). Genotyping was performed using a Golden gate genotyping assay kit according to a standard protocol (Illumina Inc., San Diego, CA, USA) as previously described (23). Briefly, 250 ng of genomic DNA were mixed with oligomers, and allele-specific extension was carried out by ramping the temperature from 70°C to 30°C over 16 h. Specific extension products were then used in a polymerase chain reaction (PCR) consisting of 34 cycles for 35 sec at 95°C, 35 sec at 56°C and 2 min at 72°C. PCR products were purified using 96-well filter plates (Millipore, Billerica, MA, USA). For hybridization, all samples were transferred to a 384-well microplate. The Sentrix array matrix chip and purified PCR products were hybridized at 60°C for 30 min and then at 45°C for 16 h. The Sentrix array matrix chip was then washed and imaged at a resolution of 0.8 mm using a BeadArray Reader (Illumina Inc.). Genotyping analysis was performed using Illumina's BEADSTUDIO software (Version 3.0.22). All other reagents were purchased from Sigma-Aldrich (St. Louis, MO, USA).
Statistical analysis. Statistical testing was performed using SPSS version 22.0 (SPSS, Chicago, IL, USA). To check for deviations from the Hardy–Weinberg equilibrium, χ2 tests were performed. Genetic models for the association tests were divided into co-dominant, dominant, and recessive models. Associations between genotypes and HCC development were analyzed using a multiple logistic regression model after controlling for age and gender as covariates. Chronic HBV carriers (chronic hepatitis B and HBV-associated liver cirrhosis groups) were regarded as a control group compared to an HBV-associated HCC group for analyzing association between genotypes and HCC development. A linkage disequilibrium (LD) block of SNPs was confirmed using the HAPLOVIEW software (version 4.0; http://www.broad.mit.edu/mpg/haploview). Individual haplotypes were inferred by the EM algorithm using the SAS haplotype procedure. Haplotype analysis was performed by multiple logistic regressions. Within the HBV-associated HCC group, associations between tumor characteristics and genotypes were also analyzed using a multiple logistic regression model. Association between liver transplantation-free overall survival of patients with HCC and genotypes were analyzed using a multivariate Cox regression analysis. For all statistical tests, significance was set at p<0.05. For more precise estimates, statistical computation was used to determine the p-values of Fisher's exact test by Monte Carlo simulation (number of iterations=100,000).
Results
Patient characteristics. Genotypic and clinical data obtained from 728 subjects (483 chronic HBV carriers and 245 HBV-associated HCC) were analyzed. Proportion of male gender was similar between patients who were chronic HBV carriers without HCC and those with HBV-associated HCC (74.3% vs. 75.9%, p=0.640). The age of patients with HCC was greater than that of patients who were chronic HBV carriers without HCC (53.83±10.34 vs. 41.13±10.31 years; mean±standard deviation, p<0.001). The baseline patient characteristics and clinicopathological features of the tumors in the enrolled HCC patients are summarized in Table II. Approximately one-third of the patients had serum AFP levels greater than 400 ng/ml. One hundred and nine (44.5%) of the 245 tumors had maximum diameters, greater than 5 cm in size. Seventy-eight (3.18%) patients had vascular invasion. Advanced tumor stage (modified union for international cancer control stages III+IV) was observed in 132 (53.8%) patients.
Genotype distribution and haplotype construction. The observed genotype frequencies of all nine SNPs in the patients with HCC and chronic HBV carriers without HCC were in Hardy-Weinberg equilibrium. The SNPs in the FGF2 and FGFR2 genes were analyzed for LD, and haplotypes were constructed. LD coefficients (|D’| and r2) among SNPs were calculated based on the genotypes of the study subjects. Two polymorphic sites in the FGF2 gene had no linkage. Three (rs2981578, rs1219648, and rs2981582) of four polymorphic sites in the FGFR2 gene were closely linked together (Figure 1). Using the EM algorithm, four common haplotypes [ht1 (A-T-C), ht2 (G-C-T), ht3 (G-T-C), and ht4 (G-C-C)] of rs2981578, rs1219648 and rs2981582 were inferred from those SNPs in the FGFR2 gene and used for haplotype association analysis.
Association between the genotype/haplotype frequencies and the development of HCC. When we compared the genotype frequencies between the patients who were chronic HBV carriers (chronic hepatitis B and HBV-associated liver cirrhosis) and the patients with HCC, none of the SNPs showed a significant association with HCC development. Similarly, haplotype frequencies in the FGFR2 gene were not associated with HCC development (data not shown).
Association between the genotype/haplotype frequencies and the severity of HCC. Table III summarizes the association between tumor characteristics and nine SNPs of the analyzed genes. The FGF2 rs308379 A allele was associated with small tumor size (p=0.042 in a co-dominant model and p=0.028 in a dominant model) and early tumor stage (p=0.010 in a co-dominant model and p=0.003 in a recessive model). Furthermore, the FGF2 rs308379 A allele was found to be significantly associated with a lower likelihood of vascular invasion (p=0.034 in a recessive model) in patients with HCC (Table III). With haplotype analyses, FGFR2 ht3 (G-T-C) was found to be in marginal association with small tumor size in a co-dominant model (p=0.041).
The Flt-1 rs4771249 SNP C allele was associated with low AFP levels (AFP<400 ng/ml) in a dominant model (p=0.036; Table III).
Association between the genotype frequencies and the overall survival in patients with HCC. The genotype frequencies and the median survival time of patients with HCC are depicted in Table IV. Among the nine SNPs evaluated, two SNPs in the FGF2 gene and one SNP in the FGFR2 gene were significantly associated with patient overall survival. Kaplan–Meier analysis showed that patients with the FGF2 rs308447 TT genotype had shorter survival rates than patients with the CC genotype (p=0.021) or the CT genotype (p=0.008). Similarly, patients with the FGF2 rs308447 TT genotype showed lower survival rates thanpatients with the CC or CT genotypes (p=0.016). The FGF2 rs308379 A allele carriers had a shorter survival than those with the TT genotype (p=0.020; Figure 2), which was contradictory considering the association of the A allele with early tumor stage. The FGFR2 rs1219648 CC genotype was significantly associated with increased overall survival rates (p=0.047; Figure 3).
No significant association between the SNPs of the FGF1, FLT1, and MET genes and overall survival were observed in patients with HCC.
Multivariate analysis for overall survival using the Cox proportional hazard model. We performed multivariate analysis of the effects of genotype on survival, with the Cox proportional hazard models, which included all the SNPs evaluated and other clinically significant covariates such as age, sex, AFP levels, and tumor stage (stages I and II and stages III and IV). In multivariate analyses using a forward stepwise selection method, FGF2 rs308379 and tumor stage remained significant. The FGF2 rs308379 A allele (HR=1.663, 95% confidence interval=1.171–2.361, p=0.004) and advanced tumor stage (hazard ratio=3.430, 95% CI=2.414-4.875, p<0.001) were independent poor prognostic factors for overall survival in patients with HBV-associated HCC (Table V).
Discussion
We demonstrated that the SNPs in the FGF2 and FGFR2 genes were significantly associated with overall survival in patients with HBV-associated HCC. Furthermore, multivariate analysis showed that the FGF2 rs308379 SNP A allele was independently a poor prognostic factor for overall survival. To our knowledge, this is the first study to show the association between SNPs in FGF-related genes and survival of HCC patients.
Tumor progression might depend on the close network of signaling pathways rather than just a single pathway. Previous studies have reported a crosstalk between the FGF and VEGF signaling pathways, indicating a possible synergistic effect on angiogenesis (11, 12). Moreover, FGF signaling is also implicated in the development of resistance to drugs targeting VEGF (24). Therefore, we attempted to investigate the effect of SNPs in the key genes related to various growth factor signaling pathways implicated in the pathogenesis of HCC.
The FGF signaling pathway has an established role in the molecular pathogenesis of HCC (4). In HCCs, the expression of FGF1 and FGF2 has been found to increase sinusoidal capillarization, suggesting a role for FGF signaling in tumor angiogenesis (25). FGF2 stimulates HCC proliferation through an autocrine mechanism and plays a key role in HCC invasion and induction of angiogenesis (9). In clinical samples, high preoperative serum FGF2 levels predicted an invasive HCC phenotype with shorter overall survival after surgical resection (10). FGFR2 expression was also reported to be a potential prognostic indicator of postsurgical survival in Korean patients with HCC (26). Taken together, it can be assumed that the expression levels of FGF2 and FGFR2 could be used as prognostic markers in HCC patients.
Interestingly, Kaplan–Meier and Cox multivariate analyses showed that the FGF2 rs308379 SNP A allele was a prognostic factor for poor overall survival, even though the FGF2 rs308379 A allele was associated with small tumor size, less vascular invasion, and early tumor stage in patients with HCC. If we consider that both tumor status and liver dysfunction contribute to prognosis in patients with HCC, these discrepant results can be acceptable. Another explanation is that the cross-sectional study design of the present study might be a cause of the false association between tumor characteristics and SNPs in the FGF2 gene.
Considering its location within the intron, the functional effect of the FGF2 rs308379 SNP on the transcript is unpredictable. Given the multiple splicing variants of FGFs and FGFRs, dysregulation of mRNA splicing can result in altered FGF signaling (27). The F-SNP database used to predict the functional effects of the analyzed polymorphisms showed that the intron-located FGF2 rs308379 polymorphism may predict changes in transcriptional activity (http://compbio.cs.queensu.ca/F-SNP/). However, it is unclear how the FGF2 rs308379 SNP has biological effects on the clinical outcomes of patients with HCC.
Overexpression of VEGF, as well as VEGFR1 and VEGFR2, frequently correlates with increased microvascularity and metastasis in HCC (7). Increased serum VEGF levels are correlated with disease progression and tumor recurrence in patients with HCC undergoing surgical resection (6, 28). Previous results established that p53 can differentially stimulate transcription in the case of a polymorphic variant of the flt-1 promoter (29). The FLT1 rs4771249 C allele is associated with colon cancer with a TP53 mutation, which is one of the most frequent mutations in HCC (30). In our study, the Flt-1 rs4771249 SNP C allele was associated with low AFP levels, but showed no association with overall survival in patients with HCC.
In conclusion, FGF2 and FGFR2 SNPs were significantly associated with overall survival in patients with HBV-associated HCCs. Furthermore, the FGF2 rs308379 SNP was an independent prognostic factor for overall survival in multivariate analysis. Functional analysis of this polymorphism is required to clarify the usefulness of FGF2 genotyping in clinical practice.
Acknowledgements
This study was supported by a grant from the Korean Health R & D Project, Ministry of Health and Welfare, Republic of Korea (HI16C2011). The biospecimens from Ajou University Hospital were provided by the Ajou Human Bio-Resource Bank (AHBB), a member of the National Biobank of Korea, which is supported by the Ministry of Health and Welfare.
Footnotes
Authors' Contributions
Sung Won Cho and Jae Youn Cheong designed the study. Soon Sun Kim and Jae Youn Cheong wrote the draft of the manuscript. Jung Woo Eun, Hyo Jung Cho, Hyun-Young Lee and Chul Won Seo conducted and analyzed the experimental and clinical data. Choong Kyun Noh, Sung Jae Shin and Kee Myung Lee supervised the overall study and revised the manuscript.
Conflicts of Interest
There are no conflicts of interest from any of the Author regarding the data and contents of this manuscript.
- Received January 13, 2019.
- Revision received March 17, 2019.
- Accepted March 18, 2019.
- Copyright© 2019, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved