Single‐nucleotide polymorphisms in the coding region of a disintegrin and metalloproteinase with thrombospondin motifs 4 and hepatocellular carcinoma: A retrospective case‐control study

Abstract Previous studies have shown that single‐nucleotide polymorphisms (SNPs) of a disintegrin and metalloproteinase with thrombospondin type 1 motif 4 (ADAMTS4) may involve in the pathogenesis of some diseases. However, it is not clear whether they are associated with hepatocellular carcinoma (HCC). A hospital‐based case‐control study, including 862 cases with HCC and 1120 controls, was conducted to assess the effects of 258 SNPs in the coding regions of ADAMTS4 on HCC risk and prognosis. We found that six SNPs in ADAMTS4 were differential distribution between cases and controls via the primary screening analyses; however, only rs538321148 and rs1014509103 polymorphisms were further identified to modify the risk of HCC (odds ratio: 2.73 and 2.95; 95% confidence interval, 2.28‐3.29 and 2.43‐3.58; P‐value, 5.73 × 10−27 and 1.36 × 10−27, respectively). Significant interaction between these two SNPs and two known causes of hepatitis B virus and aflatoxin B1 were also observed. Furthermore, rs538321148 and rs1014509103 polymorphisms were associated not only with clinicopathological features of tumor such as tumor stage and grade, microvessel density, and vessel metastasis, but with poor overall survival. Additionally, these SNPs significantly downregulated ADATMS4 expression in tumor tissues. These data suggest that SNPs in the coding region of ADAMTS4, such as rs538321148 and rs1014509103, may be potential biomarkers for predicting HCC risk and prognosis.


| INTRODUCTION
Hepatocellular carcinoma (HCC), a major histological type of liver cancer, is the fifth most common malignant tumor worldwide and the third cause of cancer-related mortality. 1 More than half a million cases are diagnosed every year, there being major geographic differences in incidence. The annual incidence rates in eastern Asia, especially the People's Republic of China, exceed 15 per 100 000 inhabitants; whereas the figures are 5-15 per 100 000 individuals in the southern Europe and Mediterranean basin and very low (less than 5/100 000) in northern Europe and America. 1 It's worth noting that approximately 50% new-diagnosed HCC cases locate in China. Evidence from clinical epidemiological studies has proved that chronic hepatitis B virus (HBV) infection and aflatoxin B1 (AFB1) is responsible for HCC carcinogenesis. 1,2 Additionally, growing evidence has also shown that genetic factors may play some kind of role in the control of hepatic tumorigenesis. [3][4][5][6] Results from high incident area of HCC display that genetic single nucleotide polymorphisms (SNPs) some genes (such as XRCC1, ERCC2, and so on) increase HCC risk and modify HCC prognosis 7,8 ; however, despite considerable advances in carcinogenetic fields, HCCrelated genes have been still elaborated less well.
A disintegrin and metalloproteinase with thrombospondin type 1 motif 4 (ADAMTS4) is an important member of a disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS) family genes. This gene is located on chromosome 1q23.3 and consists of 9 exons (PubMed). 9 Its encoding protein contains 837 amino acids and structurally includes following several domains: a signal peptide in N-terminal (directing ADAMTS4 to the secretory pathway), a pro-domain (maintaining ADAMTS4's enzyme latency), a metalloproteinase domain (binding zincs and having aggrecanase activity), a disintegrin-like domain, a thrombospondin type 1 repeat domain, a cysteine-rich domain, and a spacer region in C-terminal. [10][11][12] Functionally, ADAMTS4 acts as an aggrecanase and involves in the degradation of aggrecan and brevican, which is responsible for collagen maturation, matrix degradation, organogenesis, and blood coagulation. 10,[13][14][15] Recently, increasing evidence has indicated that the dysregulation of ADAMTS4 may be associated with the pathogenesis of some tumors such as lung cancer, 16 head and neck cancer, 17 choriocarcinoma, 18 Ewing's sarcoma, 19 melanoma, 20 colorectal cancer, 21 breast cancer, 22 HCC, 23 and glioblastoma. 24 Sever studies have reported that it also involves in the procession of angiogenesis and vascular endothelial growth factor receptor 2 (VEGFR2) phosphorylation. 25 However, the relevance of the SNPs in ADAMTS4 gene and HCC risk and prognosis remains to be elucidated. Here, we investigated whether 258 SNPs in the coding region of this gene are correlated with the risk and prognosis of HCC. Additionally, we also analyzed the potential effects of positive SNPs on ADAMTS4 expression.

| Study population
A hospital-based case-control study was conducted in Guangxi Region, a high-incidence area of HCC in China, according to our previously described criteria. [26][27][28] Briefly, a total of 862 cases with histologically diagnosed HCC (representing 98% eligible cases) and 1120 age-, gender-, and race-matched controls without any evidence of liver diseases (representing 97% eligible cases) were recruited in the affiliated hospitals of Youjiang Medical University for Nationalities and Guangxi Medical University during January 2009 and December 2013 and included in the present final analyses ( Figure S1). For these individuals, 200 matched cases with HCC and controls were randomly selected for screening positive SNPs in ADAMTS4 and the remainder was used for validation analyses. After signed consent was obtained, research samples (including 4 mL peripheral blood samples from every subjects and surgical removed tumor samples from every case with HCC) and corresponding clinicopathological data (including survival data) were collected. In this study, individuals with positive hepatitis B surface antigen or anti-hepatitis C virus (HCV) were considered as HBV or HCV infectors respectively. The exposure levels of AFB1 were evaluated using serum AFB1-albumin adducts (se-AAA) described in our previously published methods. 28 Tumor stage and grade was defined according to WHO tumor, node, and metastasis system and Edmondson differentiation grading system; whereas liver cirrhosis and microvessel density (MVD) was evaluated using histopathological method. 29 For survival analyses, all patients with HCC were followed and the follow-up endpoint was set on 31 July 2018. Overall survival (OS) status was defined as the duration from the date of initial treatment to the date of death or last known date alive. The present study was approved by the ethic committees of the participating hospitals and was carried out in accordance with the approved guidelines (No. AYJM20090112). More detailed method information was available in the Supporting Information.

| SNPs selection and genotypic analyses
Because epidemiological evidence has shown that the dysregulation of ADAMTS4 may involve in HCC tumorigenesis 26 and the SNPs in the coding region often result in the structural and functional alteration of the corresponding proteins, the SNPs resulting in missense mutations were obtained. According to the prediction value of Sorting Intolerant from Tolerant (SIFT, http://sift.bii.a-star.edu.sg) for SNPs in the coding region of ADAMTS4 and the criteria of SIFT score less than 0.05, a total of 258 SNPs were selected for initial screening analyses in the first stage and six SNPs (rs150616368, rs538321148, rs17855812, rs1014509103, rs1485965919, and rs773407656) were further validated in the second stage. More detailed information was available in the Supporting Information.

| ADAMTS4 expression assays
The levels of ADAMTS4 mRNA and protein were tested by TaqMan-PCR and immunohistochemistry (IHC) techniques (described in the Supporting Information) respectively.

| Statistical analysis
All statistical analyses were performed using the SPSS version 18 (SPSS Institute). The distribute difference between the groups was tested by chi-square test, t test, or ANOVA test. To decrease false-positive report probability in the multiple test, 30 positive SNPs in ADAMTS4 were screened and validated using non-conditional regression based on the additive model according to the correlation matrix-based method (CMBM). 31

| ADAMTS4 SNPs and HCC risk
Genotypic distribution of all selected ADAMTS4 SNPs among control individuals was the Hardy-Weinberg equilibrium. Results in the screening stage showed that six SNPs (rs150616368, rs538321148, rs17855812, rs1014509103, rs1485965919, and rs773407656) were significantly associated with HCC risk (Table S3); however, only two of them (rs538321148 and rs1014509103) were further identified as risk factors for HCC via the validation analyses (Table 1).

| Joint effects of ADAMTS4 SNPs and causes on HCC risk
Given that aforementioned HBV infection and AFB1 exposure are two major causes for Guangxiese HCC, 2 we questioned whether there were interactive effects between causes and ADAMTS4 SNPs and accomplished a series of joint analyses based on the combination of ADAMTS4 genotypes and causes (

OS of patients with HCC
To explore the effects of ADAMTS4 SNPs on OS of patients with HCC, we first analyzed the association between ADAMTS4 rs538321148 and rs1014509103 polymorphisms and clinicopathological features of HCC using non-conditional regression models (  (Figure 1). Taken together, these results are indicative of ADAMTS4 SNPs as potential biomarkers for HCC prognosis.

| The effects of ADAMTS4 SNPs on the expression of ADAMTS4
In view of previous studies showing that the dysregulation of ADAMTS4 expression may involve in carcinogenesis, 16 and rs1014509103 polymorphisms were correlated with increasing ADAMTS4 mRNA expression (Figure 2A,B). Immunohistochemistry data further proved that these two SNPs were positively associated with the levels of ADAMTS4 protein (r = .34 for rs538321148 and 0.40 for rs1014509103, respectively) ( Figure 2C,D). Represent IHC plots showed their relationship ( Figure 2E,F).

| DISCUSSION
Through a hospital-based case-control study, we explored the linkage between ADAMTS4 SNPs and the risk of HCC in Guangxi Zhuang Autonomous Region, a high incident area of liver cancer in China. It was found that ADAMTS4 genotypes with rs538321148 T alleles and rs1014509103 A alleles and their haplotypes had a substantial correlation with the growing risk of HCC (ORs, 2.73 for rs538321148, 2.95 for rs1014509103, and 6.11 for their haplotypes respectively). Interestingly, the interactions between ADAMTS4 genotypes and environmental factors HBV and AFB1 were also found for HCC risk. These results may indicate that the potential of ADAMTS4 SNPs as biomarkers for the risk of HCC. In Guangxi area, HCC is frequent and one of the most common malignant types; the most important causes of which are AFB1 exposure and HBV infection. 2 Our previous reports and prevent study also show that patients with HCC had a higher frequency of positive AFB1 exposure and HBV infection 3,33 ; furthermore, we observed the significant evidence of multiplicatively interactive effects of AFB1 and HBV on HCC risk. However, the fact that among these individuals exposing to the same risk factors, only small proportion develop cancer is indicative of individual's genetic susceptibility to HCC carcinogenesis. [3][4][5][6] Therefore, the new approach to understand genetic susceptible factors during triggering the tumorigenesis of hepatic cells is of high priority in current perspective. Human ADAMTS4 (also known as ADMP-1, ADAMTS-2, or ADAMTS-4), a major member of ADAMTS family genes, is an important aggrecanase involving in collagen maturation, matrix degradation, organogenesis, and blood coagulation. 10,14 In the past several decades, the epidemiological and clinicopathological studies have shown that the dysregulation of ADAMTS4 expression and function may involve in the process of some tumors, including head and neck cancer, melanoma, choriocarcinoma, and Ewing's sarcoma. [17][18][19][20] Demircan et al 17 investigated the expression of ADAMTS4 mRNA in primary head and neck squamous cell carcinoma with and without metastasis by quantitative real time polymerase chain reaction technique and found that higher expression was in the metastatic tumor tissues. Lee 18 and Minobe 19 observed that the different expression of ADAMTS4 involves in the process of choriocarcinoma and Ewing's sarcoma. Rao et al 20 further analyzed differential functions of different catalytic fragments of ADAMTS4 in the melanoma growth and angiogenesis. They observed that both ADAMTS4 full-length and N-terminal autocatalytic fragment (NTAF) can promote angiogenesis and tumor growth in mice; whereas their mutant types or C-terminal fragment displayed converse role. Of particular concerns are the coexistence of proteolytic fragments which contain only C-terminal ancillary domains), the full-length fragments,

T A B L E 4 (Continued)
F I G U R E 1 ADAMTS4 rs538321148 and rs1014509103 polymorphisms significantly correlating with hepatocellular carcinoma prognosis.
Cumulative hazard function was plotted by Kaplan-Meier's methodology, and P value was calculated with two-sided log-rank tests. The relative hazard ratio values for genotypes were calculated using multivariable cox regression models (with all significant variables) based on forward-step method with likelihood ratio test. Abbreviations: H-1, ADAMTS4 haplotype with 1 risk allele of rs538321148 and rs1014509103; H-2, ADAMTS4 haplotype with 2 risk alleles of rs538321148 and rs1014509103; H-3, ADAMTS4 haplotype with 3 risk alleles of rs538321148 and rs1014509103; H-4, ADAMTS4 haplotype with 4 risk alleles of rs538321148 and rs1014509103; H-N, ADAMTS4 haplotype without risk alleles of rs538321148 and rs1014509103; HR, hazard ratio and NTAFs in the culture cells and tissues samples with melanoma. 20 Our previous report 26 and the present study shows that the dysregulation of ADAMTS4 by genetic factors such as mutant microRNA-1268a increase the risk of HCC. Together, the collected data imply that ADAMTS4 may play some vital roles in the procession of carcinogenesis, and thus deserve to be further evaluated. More than 3200 SNPs in the ADAMTS4 have been identified and the part of them may involve in human diseases such as lumbar disc degeneratio. [34][35][36] For example, Liu et al 34 investigated the association between five SNPs (rs11585858, rs4656291, rs41270041, rs10908826 and rs4233367) in and around region of ADAMTS4 via a case-control study (including 482 cases with lumber intervertebral disc degeneration and 496 controls without any evidence of diseases) and observed that only rs4233367 altered the risk of lumber intervertebral disc degeneration (OR 0.69; 95% CI 0.50-0.94; P = 1.66 × 10 −2 ). However, Canbek et al 36 reported that two SNPs (rs226794 and rs2830585) in the ADAMTS4 did not change knee osteoarthritis risk. Until now, there have been any meta-analyses focused on this topic because of only two known studies. 35 In this study, 258 known SNPs in the coding regions of ADAMTS4 gene were selected initially for the first-stage screening analyses, mainly because of their implied effects on the structure and function of ADAMTS4 protein by SIFT method (http://sift.bii.a-star.edu.sg). Of these SNPs analyzed, only rs538321148 (Val 104 to Iso 104 ) and rs1014509103 (Thr 493 to Iso 493 ) polymorphism were found to link with increasing HCC risk through the screening-and validation-stage analyses. Moreover, they were also observed to significantly interact with HBV infection and AFB1 exposure during HCC carcinogenesis. These data indicate that rs538321148 and rs1014509103 polymorphisms may act as potential susceptibility factors for cancer risk, and the possibility of the genetic SNPs in ADAMTS4 influencing HCC risk should therefore not be ignored. F I G U R E 2 ADAMTS4 rs538321148 and rs1014509103 polymorphisms significantly associating with ADAMTS4 expression. A and B, The ADAMTS4 mRNA levels in the fresh tissues with hepatocellular carcinoma (n = 25) were tested using TaqMan real-time quantitative PCR method and its relative expression was calculated by the comparative CT method. Data were shown as means ± SD and tested by ANOVA test. C-F, The protein amount of ADAMTS4 in the paraffin-embedded tissues with HCC (n = 862) were tested using immunohistochemistry method. To analyze, the levels of ADAMTS4 protein were divided into three groups: low expression group (LE, immunoreaction score ≤ 4), medium expression group (ME, 4 < IRS < 8), and high expression group (HE, IRS ≥ 8). The associations between the levels of ADAMTS4 protein and either rs538321148 (C) or rs1014509103 (D) were analyzed using Spearman's correlation model. Represent immunohistochemistry plots shows their associations (E, F) To explore possible pathogenesis of ADAMTS4 SNPs modifying HCC risk, we examined the expression of ADAMTS4 mRNA and protein and the pathological characteristics of HCC. Our results showed that ADAMTS4 rs538321148 and rs1014509103 polymorphisms were linked with the upregulation of ADAMTS4 expression, increasing tumor MVD and metastasis risk, and tumor dedifferentiation and progression. Several recent in vitro and in vivo studies 20,23,25 and clinicopathological investigations 18,19,23,25 have exhibited that the dysregulation of ADAMTS4 expression promotes tumor progression and angiogenesis by modifying pVEGFR2. The OS analyses further proved that these two SNPs were significantly relevant to poor survival of patients with HCC, emphasizing the potential values of ADAMTS4 rs538321148 and rs1014509103 polymorphisms in predicting HCC prognosis. Collectively, these findings imply that ADAMTS4 SNPs could affect the expression and function of ADAMTS4 and ultimately increase HCC risk and shorten the prognosis of patients with this cancer.
The major strongpoints of this study were the retrospective two-stage design, relatively large-size samples, and SNPs screening strategy on the basis of SIFT method. The potential effects of known confounders (including gender, age, and race) for Guangxiese HCC were controlled using an individually matched case-control design. In addition, the most two causes of HCC among Guangxi population, also HBV infection and AFB1, were included in the current study and their interaction with SNPs in the ADAMTS4 were also elucidated for HCC risk.
However, there were several disadvantages to limit our study. First, the hospital-based design may result in the selective bias of study subjects. Second, the gene-environmental interactions may be underestimated because of the relatively small sample sizes of some subgroups and low frequency of risk alleles and genotypes, especially some ADAMTS4 haplotypes. Third, although SNPs in the coding region of ADAMTS4 were analyzed in this study, SNPs in the 5′ near region, 5′-UTR, 3′-UTR, and 3′ near region of this gene may change the expression of ADAMTS4 and affect HCC carcinogenesis. Fourth, some SNPs in the coding region of ADAMTS4 (such as rs377253620, rs1428696549, rs1467819272, rs17855812, and so on) may play potential role in the procession of HCC carcinogenesis although they were excluded in the current study according to screening power. Fifth, although the effects of ADAMTS4 SNPs on OS status were investigated among these patients with HCC, their effects on tumor occurrence-free survival were not estimated. Additionally, more analyses for SNPs modifying ADAMTS4 functions are not done, except for ADAMTS4 expression based on tissular samples with HCC. Therefore, more functional and mechanical analyses deserve further elucidation on the basis of a larger sample size, more SNPs, and the combination of SNPs and liver carcinogenesis.

| CONCLUSIONS
To the best of our knowledge, this is the first study to explore correlation between ADAMTS4 SNPs and HCC risk and prognosis among Guangxi Population from a high incidence area of HCC. Some evidence was found to suggest that ADAMTS4 rs538321148 and rs1014509103 polymorphisms may increase HCC risk and modify the prognosis of patients with HCC. Following confirmation in a large sample size (including meta-analyses focused on this topic), these findings may be used to identify susceptible individuals at high risk of HCC, especially those with positive HBV infection and AFB1 exposure.