Introduction
Hepatocellular carcinoma (HCC) or liver cancer accounts for the cancer with the fourth highest mortality rate in the world. In 2018, there are estimated to be 841,080 of HCC new cases worldwide [
1]. Patients with HCC usually have poor prognosis and high mortality rates, even in developed countries [
2]. Until now, the pathogenesis of HCC has not been fully understood, but it is known so far that it is influenced by hepatitis B and C virus infections and also influenced by environmental factors (smoking, alcohol, aflatoxin B1) [
3]. It is known that there are differences in the risk of HCC within each person, in which case the host factor has an important role [
4].
Tumor Necrosis Factor-α (TNF-α) is an important inflammatory cytokine in the development of liver disease. This cytokine can cause hepatic injury, cirrhosis and eventually promote hepatocellular carcinoma [
5,
6]. Several previous studies have identified some Single Nucleotide Polymorphism (SNP) s in TNF-α gene, especially in the promoter region. SNPs of TNF-α -1031 T/C (rs1799964), −863C/A (rs1800630), −857C/T (rs1799724), −308G/A (rs1800629), and -238G/A (rs361525) are SNPs in the TNF-α promoter site that have often been investigated regarding their association with HCC in several previous studies [
7,
8]. It was also said that those SNPs could affect TNF-α production at the transcription level [
9,
10]. This plays important role because cytotoxic T lymphocytes (CTLs) in the liver secrete TNF-α [
11]. Increased TNF-α level due to those SNPs can cause persistent inflammatory condition in liver tissue which is the most important risk factor for HCC [
7].
High production of TNF-is related to the increase of pro-inflammatory cytokine secretion, the activation of proto oncogenes and several genes associated with cell growth, invasion, and cancer cells metastasis [
12,
13]. Excessive production of TNFα can also induce the generation of free radicals in the form of Reactive Oxygen Species which can cause further liver damage and genomic instability [
14]. It is also said that high TNF-α expression is an independent predictor of poor survival in HCC patients [
15].
The results of various previous studies regarding the relationship between TNF-α polymorphism and HCC across various ethnicities and populations show mixed results, related [
16‐
18] or there is no relationship [
19,
20]. Research on TNF-α gene SNPs in patients with Hepatitis B Virus (HBV) and Hepatitis C Virus (HCV) infection is specific for each ethnicity and shows different results in each population [
21,
22]. As the results regarding the role of these five SNPs of TNF-α genes against HCC are still controversial, we conducted this meta-analysis to determine the relationship between those TNF-α gene polymorphisms and HCC.
Methods
Database searching
Our meta-analysis was reported based on the items outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement to ensure adequate reporting of this meta-analysis of observational studies in epidemiology [
23]. We conducted an electronic database searching to identify all previously published cohort or case-control studies investigating the association between five types of TNF-α gene SNPs (− 1031 T/C, −863C/A, −857C/T, −308G/A, and -238G/A) and risk of HCC. We searched data from PubMed, ProQuest, EBSCO, Science Direct, and Springer by using MeSH terms: “Tumor Necrosis Factor-alpha” or “TNF-alpha” and “polymorphism” or “SNP” or “single nucleotide polymorphism” or “variant” and “hepatocellular carcinoma” or “HCC” or “liver cancer”. The search was conducted in September 2019–January 2020. We also performed a manual search was also performed to obtain potential sources cited in other meta-analysis.
Criteria for inclusion and exclusion
All included studies should meet the following criteria: (1) investigate the association between any of the five SNPs (− 1031 T/C, −863C/A, −857C/T, −308G/A, and -238G/A) in TNF-α gene and risk of HCC; (2) have cohort or case-control study design; (3) risk of HCC was reported as RR (relative risk) and/or OR (odds ratio) with 95% Confidence Interval (CI) or provide sufficient data to extract RR and/or OR with 95% CI data; (4) include human subjects; (6) in English language. We excluded studies with the following criteria: (1) studies with design other than case-control or cohort; (2) duplicated studies; (3) studies with unmeasurable population and not qualified data; (4) using non-English language; (5) Studies in which the full text or main data could not be obtained.
Study selection
Two investigators independently performed the electronic search and retrieved the articles that matched with our searched terms. Any disagreement was settled by discussion and consensus with all the authors. Final decision was merely based on the agreements of all authors.
A standardized reporting form was used to extract the data from each article which included the first author’s name, year of publication, population country, TNF-α SNP type, study design, etiology of HCC, SNP genotyping method, controls, and frequencies of SNP. Hardy–Weinberg equilibrium test was performed and its significance of the control groups was calculated when the original information was not provided.
Quality assessment
Newcastle - Ottawa quality assessment scale (NOQS) was used for measuring the quality of the included studies. This scale was designed through collaboration between the Universities of Newcastle, Australia and Ottawa, Canada. The purpose of this scale was to assess the quality of observational studies for producing good meta-analysis. The studies were qualified as high quality (9 stars), medium quality (7–8 stars), and low quality (less than 7 stars) [
24].
Data analysis
Analysis were conducted using Review Manager 5.3 software (The Cochrane Collaboration, UK). Hardy-Weinberg Equilibrium was examined by Chi Square test when the original information was not provided. We evaluate 5 genetic models (allele, dominant, recessive, codominant major vs minor homozygote, and codominant heterozygote vs major homozygote) for each SNP type separately. Major alleles for each SNP were: -1031TT, −863CC, −857CC, −308GG, and -238GG, while minor alleles for each SNP were: -1031CC, −863AA, −857TT, −308AA, and -238AA. Heterogeneity assumption was assessed with Cochrane Q statistic and I2 statistic. The pool estimated ORs was calculated with either fixed or random effects model assumptions. If Q test showed significant result (p < 0.05), we used a random effects model. Otherwise, if Q test showed insignificant result (p > 0.05), we used a fixed effect model. We also calculated the 95% confidence interval (CI) of pool estimated OR. Inverted funnel plots were conducted to find any presence of publication bias.
Discussion
The development of HCC depends on some factors such as viral infection, environmental, behavioral (smoking, alcoholism), metabolism, and genetics [
44,
45]. The contribution of SNP as a genetic factor is widely studied related to its role in the development of HCC, in which the prevalence of SNP is different in each population [
5,
46,
47]. From the host factor, TNF-α is thought to play an important role in hepatocarcinogenesis through necroinflammation and the induction of fibrogenic factors [
48]. Five biallelic SNPs in the promoter region Tumor Necrosis Factor-α gene were known: -1031 T/C, −863C/A, −857C/T, −308G/A, and -238G/A [
7]. However, to date, only few meta-analyses have analysed all these five SNPs with HCC risk. Hereby, we performed this updated meta-analysis to clarify the independent role of each of these five SNPs on HCC risk.
As research on SNP TNF-α − 1031 T/C and HCC risk is still limited, we only found five studies investigating this SNP, even though many studies have shown significant association between this SNP with other diseases such as polycystic ovary syndrome and endometriosis [
49,
50]. In a study conducted by Shin et al. survival of HCC cases with TNF-α − 1031 wild type (TT) genotype or SNP TC genotype was significantly better than those with the SNP CC genotype [
8]. However, in this meta-analysis, no significant relationship was found between the SNP and HCC risk in all genetic models. Meta-analysis conducted by Wei et al. also shows that there is no significant relationship between this SNP and HCC risk [
51].
This study shows a significant relationship between SNP TNF-α − 863 C/A with HCC risk in allele models and dominant model analysis. A meta-analysis conducted by Wei et al. also showed a significant relationship betweehis SNP and HCC in dominant and codominant (CA vs CC) model analysis [
51]. Polymorphism of TNF-α − 863 C/A in the promoter region can influence TNF-α expression, however, the result is still conflicting. Some research suggest that it may increase TNF-α expression [
52,
53], while other research show the opposite result in which it may decrease TNF-α expression [
30,
54]. A study by Skoog et al. proposed that TNF-α − 863 C/A polymorphism affects the binding of nuclear protein(s) to the promoter region of the TNF-α gene, with accompanying changes in TNF-α expression, thus leading to variation in TNF-α levels [
55]. The role of ethnicity may also play a role as well, carriers of the rare ‘A’ allele have a significantly lower TNF-α levels in Swedish and Indian population [
55,
56], while they have a significantly higher levels in Japanese population [
52,
53].
In the present study, we also found a significant relationship between SNP TNF-α − 857 C/T with dominant model analysis. Limited studies were present regarding the relationship between this polymorphism with HCC risk, thus there were only 7 studies that came within our inclusion criteria. This is different from the meta-analysis conducted by Wei et al. which showed no significant relationship between this SNP and HCC risk, however in that meta-analysis there were only 3 included studies [
51].
There have been many previous studies investigating SNP TNF-α − 308 G/A with HCC risk, thus we got 19 included studies. All five genetic analysis models showed a significant relationship between the SNP and HCC risk. This is in line with several previous meta-analysis studies, including those conducted by Hu et al. and Tavakolpour and Sali on allele models and dominant model analyses [
17,
57]. Wei et al. on codominant and dominant model analyses [
51], and Xiao et al. in all except recessive model analysis [
18]. Of all other TNF-α SNPs, − 308 G/A is the most studied SNP. This SNP is also correlated with the risk of other cancer, such as breast cancer and gastric cancer [
56,
58]. This SNP seems to be variable within ethnicities, as in meta-analysis study conducted by Wei et al., SNP − 308 AA was associated with an increased risk of HCC in Asian ethnicities, but not for Caucasian [
51]. A study conducted by Shin et al. on South Korean population showed that TNF-α − 308 SNP alone was not significantly associated with HCC, but when several genotypes were combined (e.g. -1031 / − 308 / -238 TT / GG / GA), there was a significant association with the incidence of HCC [
8].
Studies on SNP TNF-α − 238 G/A have also been extensively performed. Nevertheless, meta-analyses of these SNPs are still limited and yield conflicting results. The present study showed a significant relationship between this SNP and HCC risk. This is in line with a meta-analysis conducted by Xiao et al. showing that this SNP is associated with HCC risk, although in that study only HBV-related HCC was studied [
18]. Another meta-analysis conducted by Hu et al., however, shows no significant relationship between this SNP with HCC risk in Asian population [
57].
Those SNPs on TNF-α promoter can influence TNF-α level. They resulted in higher and constitutive TNF-α expression and were associated with an increased risk of HCC [
10,
17]. TNF-α itself is a potent pro-inflammatory cytokine [
17,
59]. Necroinflammation in hepatocytes triggers mutagenesis and activation of oncogenes from proto-oncogenes in host cells, causing HCC [
48]. In addition, through the chronic inflammatory pathway, TNF-α is also known to induce HCC through activation and differentiation of hepatic progenitor cells [
60].
This meta-analysis still had several limitations. To date, there has rarely been any meta-analysis discussing the relationship between all each SNPs of TNF-α − 1031 T/C, − 863 C/A, − 857 C/T, and − 238 G/A with HCC. The second limitation of this study was that we only included studies in English language so that it does not rule out the exclusion of any good research in non-English languages. Third, the role of other factors, such as variations in genes adjacent to the TNF-α gene or epigenetic factors are said to also be able to regulate TNF-α expression [
61,
62]. Fourth, our study is a meta-analysis so it is not possible to generate a per-patient haplotype analysis. Further research is required to determine the effects of these various TNF-α SNPs in each different population.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.