Association of TLR4 gene 2026A/G (rs1927914), 896A/G (rs4986790), and 1196C/T (rs4986791) polymorphisms and cancer susceptibility: Meta-analysis and trial sequential analysis.

With the development of society, the prevalence of cancer is increasing. Cancer is a kind of disease caused by malignant proliferation of cells. It is invasive and has become one of the important diseases threatening people’s life quality and life span. Cancer is also a multi-factorial disease. In addition to lifestyle, occupation and biological, physical and chemical factors in the environment, the influence of genetic factors has emerged as another crucial factor. [ 1 ] Single nucleotide polymorphism (SNP) refers to the change of genetic information at the nucleotide locus of a gene. [ 2 ] It was previously believed that missense mutations altering the amino acid composition of translated proteins would lead to changes in genetic traits, susceptibility to disease and other outcomes. [ 2 , 3 ] However, a recent study published in Nature showed that many synonymous mutations are also harmful to organisms, rather than neutral or near neutral. [ 4 ] Both nonsense mutations and missense mutations may lead to changes in individual genetic traits to varying degrees, thus changing factors such as external performance or susceptibility to disease. [ 4 ] Therefore, the study of the relationship between gene polymorphisms and disease susceptibility can provide strong evidence for the genetic diagnosis and prevention of diseases and has important clinical significance. Toll-like receptor 4 (TLR4) is one of the receptors that closely related to immunity. [ 5 ] SNPs of genes encoding TLR4 often lead to a series of changes in the body’s immune system, leading to certain diseases, such as immune related diseases like asthma, [ 6 ] rheumatoid arthritis, [ 7 ] systemic lupus erythematosus, [ 8 ] and even intracranial aneurysm, [ 9 ] hypertension, [ 10 ] and endometriosis. [ 11 ] More and more case-control studies have confirmed that the SNPs of TLR4 gene 2026A/G (rs1927914), 896A/G (rs4986790), and 1196C/T (rs4986791) are related to the susceptibility and prognosis of prostate cancer, [ 12 ] lung cancer, [ 13 ] gastric cancer, [ 14 ] cervical cancer, [ 15 ] and many other cancers. However, most of the studies only focused on the same cancer type. Moreover, most of the patients included in a single study were confined to the same hospital, lowering the representativeness of the results. Therefore, focusing on case-control studies on the relationship between SNPs of TLR4 gene 2026A/G (rs1927914), 896A/G (rs4986790), and 1196C/T (rs4986791) and cancer susceptibility, meta-analysis and trial sequential analysis (TSA) were carried out in this paper. In this way, the secondary research results with data of multi centers and large samples were obtained, to give hints for clinical practice and basic research.

Conceptualization: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Data curation: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Formal analysis: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Funding acquisition: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Investigation: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Methodology: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Project administration: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Resources: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Software: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Supervision: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Validation: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Visualization: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Writing – original draft: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu. Writing – review & editing: Fengzhen Wang, Xianming Wen, Ting Wen, Ziyou Liu.

The study design of the included articles should be a publicly published case-control study. It should assess the association between TLR4 gene 2026A/G, 896A/G, and 1196C/T polymorphisms and cancer susceptibility. The case group consisted of patients with clinically and pathologically diagnosed cancer, and the control group consisted of healthy people. All the people included in the study were not restricted by race, gender or age. Moreover, the articles should contain the numbers of people in the case and control groups with each genotype. The data quality should be reliable, and the results should be clearly expressed. Articles with incomplete analytical data or unavailable after contacting the original author, were not included in this meta-analysis. Articles that were repeatedly published and retrieved, as well as articles whose original research object was not human, were also excluded. Web of Science, PubMed, Embase, CBM, WanFang Data, CNKI, and VIP database were used for retrieving. The retrieval time interval was from the date of database establishment to May 31, 2022. Articles in the reference list were also reviewed to find eligible case-control studies. The retrieval was carried out by combining free words with subject words. The key words used include TLR4, TLR4, cancer, carcinoma, malignant tumor, neoplasm, lymphoma, SNP, polymorphism, variant, SNP, etc. Taking PubMed as an example, its retrieval strategy is shown in Table 1 . Retrieval strategy of PubMed. The 2 researchers independently screened the articles, extracted the data, and performed independent cross examination on them. Disputes were settled through discussion with a third party. When screening articles, the title was read firstly to exclude apparently unrelated articles. Then, the abstract and full text was further read to determine whether it could be included. If necessary, the author will be contacted by email to obtain the key information that has not been mentioned in the original study. The content of data extraction includes the basic information of each included study: the first author, publication year, country, number of included people in case group and control group, number of cases and controls corresponding to each genotype, type of cancer studied, etc. According to the principle of Newcastle-Ottawa Scale, the general qualities of included case-control studies were evaluated independently by 2 researchers. [ 16 ] Disputes were settled through discussion with a third party. Meta-analysis was performed by RevMan 5.4 (RevMan, College Station, TX) and Stata MP-17 software (Stata, London, UK). TSA was performed by TSA 0.9.5.10 Beta software. The test level of heterogeneity test is α = 0.10. If the result of the heterogeneity test shows P   .10, it means that there is no heterogeneity, and the fixed effect model is used. The heterogeneity judgment method of sensitivity analysis is the same as the above. Odds ratio (OR), 95% confidence interval (95% CI) and P value were recorded. The test level of meta-analysis and TSA analysis is α = 0.05. The publication bias was tested by Begg’s test, and the test level was α = 0.05. Ethical approval is not applicable as data were derived from published articles.

According to the meta-analysis article retrieving and screening process recommended by Preferred Reporting Items for Systematic Reviews and Meta-Analyses, 87 case-control studies from more than 30 countries were included in this meta-analysis (Fig. 1 ). [ 12 – 15 , 17 – 76 ] Among them, there were 25,969 people in the case group and 32,119 in the control group. The diseases involved in case groups include prostate cancer, lung cancer, gastric cancer, hepatocellular carcinoma, colorectal cancer, etc. Study quality evaluation showed that all original studies had clear case groups and control groups with reliable inclusion and diagnostic criteria. The basic characteristics of the original studies included in the meta-analysis are shown in Tables 2 – 4 . Article screening process. Study characteristics included in meta-analysis for rs1927914. The lower case letters after the first author are used to distinguish different types of cancer in the same article. AA = adenine/adenine, AG = adenine/guanine, GG = guanine/guanine. Study characteristics included in meta-analysis for rs4986790. The lower case letters after the first author are used to distinguish different types of cancer in the same article. AA = adenine/adenine, AG = adenine/guanine, GG = guanine/guanine. Study characteristics included in meta-analysis for rs4986791. The lower case letters after the first author are used to distinguish different types of cancer in the same article. CC = cytosine/cytosine, CT = cytosine/thymine, TT = thymine/thymine. Summary of meta-analysis results. For rs1927914, a represents A, b represents G; For rs4986790, a represents A, b represents G; For rs4986791, a represents C, b represents T. CI = confidence interval; OR = odds ratio. Forest plot of rs1927914 A versus G model was shown as Figure 2 . Heterogeneity test showed that χ 2  = 24.04, P  = .06  1, indicating individuals carrying allele A have a higher risk of developing cancer than individuals carrying allele G. Besides, 95% CI = [1.01, 1.15] which did not pass through 1, and P  = .02 < .05, indicating the results were statistically significant. Cancer susceptibility characteristics of rs1927914 A versus G model. Forest plot of rs4986790 A versus G model was shown as Figure 3 . Heterogeneity test showed that χ 2  = 79.95, P  < .0001 < .10, indicating heterogeneity exists. In this way, statistical analysis was performed using random effects model. The results of meta-analysis showed that OR = 0.85 < 1, indicating individuals carrying allele A have a lower risk of developing cancer than individuals carrying allele G. Besides, 95% CI = [0.75, 0.96] which did not pass through 1, and P  = .007 < .05, indicating the results were statistically significant. Cancer susceptibility characteristics of rs4986790 A versus G model. Forest plot of rs4986791 C versus T model was shown as Figure 4 . Heterogeneity test showed that χ 2  = 58.37, P  = .004 < .10, indicating heterogeneity exists. In this way, statistical analysis was performed using random effects model. The results of meta-analysis showed that OR = 0.74 < 1, indicating individuals carrying allele C have a lower risk of developing cancer than individuals carrying allele T. Besides, 95% CI = [0.63, 0.86] which did not pass through 1, and P  = .0001 < .05, indicating the results were statistically significant. Cancer susceptibility characteristics of rs4986791 C versus T model. The results of meta-analysis were shown as Table 5 . For AA versus GG, AA+AG versus GG and A versus G models of rs1927914, all the OR value was higher than 1, and the results were statistically significant. This indicates that considering TLR4 gene 2026A/G (rs1927914) polymorphism, individuals with the A allele have a higher risk of cancer. For AA versus AG+GG, AA versus AG and A versus G models of rs4986790, all the OR value was lower than 1, and the results were statistically significant. This indicates that considering TLR4 gene 896A/G (rs4986790) polymorphism, individuals with the G allele have a higher risk of cancer. For all the models of rs4986790, all the OR value was lower than 1, and the results were statistically significant. This indicates that considering TLR4 gene 1196C/T (rs4986791) polymorphism, individuals with the T allele have a higher risk of cancer. Taking the allele model as examples, the sensitivity analysis was carried out by one-by-one exclude method. The results of sensitivity analysis showed that for the A versus G model of rs1927914, the OR value was 1.09 at the highest and 1.07 at the lowest after excluding a single study, and the results were statistically significant. For the A versus G model of rs4986790, the OR value was 0.87 at the highest and 0.83 at the lowest after excluding a single study, and the results were statistically significant. For the C versus T model of rs4986791, the OR value was 0.76 at the highest and 0.71 at the lowest after excluding a single study, and the results were statistically significant. Sensitivity analysis showed that the results of meta-analysis were relatively stable, and the results were less affected by changes in a single original study. Taking the allele model as examples, Begg’s test was used to analyze publication bias. For A versus G model of rs1927914, publication bias analysis results showed that Z  = 1.13, P  = .260 > .05, indicating no publication bias analysis exists. The funnel plot showed that the included original studies were generally distributed in a symmetrical funnel shape along the symmetry axis, indicating that there was no obvious publication bias (Fig. 5 A). These conclusions were the same for A versus G model of rs4986790 ( P  = .460 > .05) and C versus T model of rs4986791 ( P  = .477 > .05), and the funnel plots were shown as Figure 5 B and C. This shows that the reliability of the meta-analysis conclusion is less affected by publication bias. Funnel plots of publication bias analysis. Taking the allele model as examples, the results of TSA were shown in Figure 6 . The results showed that although the accumulated information did not reach the required information size, the Z curve had intersected with the boundary value, indicating that the association between the current gene and the high risk of cancer has been confirmed. Therefore, more tests are generally not required for further verification, and the possibility of false positives is small. Trial sequential analysis results.

Cancer is a serious threat to human life quality and longevity. In the past 2 decades, a large number of original clinical studies and related secondary studies have been published, revealing the association of TLR4s with cancer susceptibility and prognosis. Our meta-analysis showed that the individual carrying A allele in TLR4 Gene 2026A/G (rs1927914) polymorphism, G allele in 896A/G (rs4986790) polymorphism, or T allele in 1196C/T (rs4986791) polymorphism had an increased risk for cancer. Sensitivity analysis showed that the results of the meta-analysis were slightly affected by a single study, and the results were stable. Publication bias analysis and TSA showed that no significant publication bias was found in the results of the meta-analysis, and the probability of false positives was small. In conclusion, the A allele of rs1927914, G allele of rs4986790, and T allele of rs4986791 are the susceptibility genes to cancer, and the results are reliable. TLRs are a class of pattern recognition receptors that can interact with other pattern recognition receptor families and activate a variety of pathogen-associated molecular patterns to initiate a sequence of signal transduction. [ 77 ] It is closely related to inflammatory responses, and the variation of related genes will affect several pathways of the body, thus resulting in a series of changes in health status or the occurrence of diseases. The effects of an inflammatory immune response have 2 aspects: on the 1 hand, they improve an organism’s ability to fight against infection; on the other hand, a persistently inflammatory environment may make it easier for tumor cells to escape the immune system. [ 78 ] That is to say, TLRs appear to represent a potential link between infections, persistent inflammation, and the emergence of tumors in the context of cancer. In this way, TLRs are one of the prime choices to determine how inflammation plays a part in cancer. Two key pathways are used to transmit the signals that TLR4 mediates: one uses the adapter protein myeloid differentiation factor 88, and the other uses the adaptor-inducing interferon protein, which contains a toll/interleukin-1 receptor domain. [ 79 ] TLR4 can mediate reactions to host molecules like oxidized low-density lipoprotein, amyloid peptide, heat shock proteins, and those made in response to tissue injury. Recognition of their ligands causes a series of signaling events, the first of which is represented by the activation of the interleukin-1 receptor family. This is followed by the activation of the transcription factor nuclear factor kappa-B (NF-κB) with the transcription of pro-inflammatory genes. [ 80 ] Besides, a specific collection of genes implicated in proinflammatory, antiviral, and antibacterial responses begin to be transcribed when TLR4 activates myeloid differentiation factor 88, in this way NF-κB activation is promoted which leads to the production of inflammatory cytokines. [ 5 , 79 , 81 ] It has been hypothesized that activation of NF-κB is a key mediator of inflammation-induced tumor growth and progression. Numerous studies have shown that NF-κB is an important regulator of Snail expression and that it is particularly important for the spread of carcinoma. [ 82 , 83 ] The development and spread of cancer may be aided by the production of inflammatory mediators via the NF-κB pathway that is activated by the TLRs. [ 80 , 84 , 85 ] Besides, it has been suggested that innate immune activation brought on by TLR-mediated identification of pathogens or endogenous chemicals, such as those created by cell and DNA damage, can foster the growth of cancer in an inflammatory environment. [ 86 ] Therefore, the polymorphisms of TLR4 gene may affect the above pathways, thus leading to changes in cancer susceptibility, or affecting the prognosis of cancer by affecting metastasis and invasion ability. However, our study also has some limitations. First of all, most of the meta-analysis of each group has high heterogeneity. In view of this situation, we used random effects model for statistical analysis. Then, for the meta-analysis of some loci, the countries where the original research were conducted were mostly confined to Asia, especially China. At the same time, the original research involved a large number of cancer types, resulting in a small number of original studies for each cancer type. Therefore, subgroup analyses were not performed. Finally, if personal data containing additional factors, such as age, sex, and smoking status, becomes available, a more accurate analysis should be carried out. In summary, people with A allele of rs1927914, G allele of rs4986790, or T allele of rs4986791 were more susceptibility to cancer. The results are basically stable and there is less chance of false positives. To support our findings, additional sizable, thoughtful, comprehensive research across a range of groups is required.