The effect of LTA gene polymorphisms on cancer risk: an updated systematic review and meta- analysis

Increasing studies have demonstrated that a number of proinflammatory cytokines could be associated with the development of cancer [1,2]. Lymphotoxin-α (LTA) is the predominant member of the tumor necrosis factor (TNF) ligand family, which responds to immune and inflammatory reaction and plays an important role in the pathogenesis of cancer [3,4]. The human LTA gene is located on the short arm of chromosome 6 (6p21.3) [5]. The presence of single nucleic polymorphism (SNP) may affect cytokine expression level, which might be an important mediator of cancer [6,7]. SNP rs1041981 is a mutation of LTA gene at the 804 (C/A) position of exon 3 in codon 26, causing the amino acid threonine to be asparagine, which may be related to the transcriptional regulation of LTA, then activate the lymphocytes and induce apoptosis [8]. While, SNP rs909253 is a mutation of LTA gene at 252 (A/G) position in intron 1, which may lead to increase in the transcriptional activities of LTA [1]. In addition, SNP rs2239704, rs746868 and rs2229094 are associated with the LTA expression, which may affect subsequent inflammatory responses and immunomodulatory diseases, including cancers [9,10].

There are ample evidences that have demonstrated the association between LTA polymorphisms and cancer [11–34]. However, these results are inconsistent and even contradictory, which might be due to the heterogeneity within cancer types, ethnicities, source of control, Hardy–Weinberg equilibrium (HWE), small sample sizes and so on. Huang et al. reported a meta-analysis about this topic, and they found that the LTA rs1041981, rs2239704 and rs2229094 polymorphisms were associated with the increased risk of cancers [35]. However, based on the current studies, we found that more studies were negatively correlated with LTA polymorphisms and cancer [11,13–15,18–24,27,28,31,32]. Therefore, we conducted the current updated systematic review and meta-analysis to accurately determine the association between genetic variation of LTA gene and cancer susceptibility.

We conducted a systematic literature search on PubMed, Medline, Embase, Google Scholar and Web of Science to retrieve all eligible publications on the association between LTA polymorphisms and the risk of cancer (up to 28 February 2020) with the following keywords: (LTA OR Lymphotoxin alpha OR TNF-β OR tumor necrosis factor-beta) AND (polymorphism OR mutation OR variation OR SNP OR genotype) AND (cancer OR tumor OR neoplasm OR malignancy OR carcinoma OR adenocarcinoma). The language of enrolled studies was restricted to English. After carefully screening, five polymorphisms were left for further investigation.

Articles enrolled in our meta-analysis satisfied the following inclusion criteria: (1) case–control studies that evaluated the association between LTA polymorphisms and cancer risk; (2) publications focusing on population genetic polymorphisms; (3) articles with sufficient genotype data to assess odds ratios (ORs) and the corresponding 95% CIs; (4) blood sample only for SNP analysis; (5) the control subjects satisfied HWE. The major exclusion criteria were: (1) case-only studies, case reports or reviews; (2) studies without raw data for the LTA genotype.

Two investigators (Jingdong Li and Yaxuan Wang) independently extracted the data from each study. All the case–control studies satisfied the inclusion criteria and consensus for any controversy was achieved. The data from the eligible articles comprise the first author’s name, year of publication, ethnicity, source of control, cancer type and numbers of cases and controls in LTA genotypes. Ethnicity was categorized as ‘Asian’, ‘Caucasian’, and ‘Mixed’.

The risk between the LTA polymorphisms and cancer was evaluated using summary ORs and the corresponding 95% CIs in allelic (B vs. A), dominant (BA + BB vs. AA), and recessive (BB vs. BA + AA) models (A: wild allele; B: mutated allele). The Cochrane’s Q-statistic test was used to assess the heterogeneity between studies, and the inconsistency was quantified with the I² statistic. The substantial heterogeneity was considered significant when I² > 50% or PQ ≤ 0.1, then, a random-effects model was used; otherwise, the fixed-effects model was applied. Subgroup meta-analysis were performed by cancer type, ethnicity, genotyping, HWE and the source of control. We also conducted sensitivity analysis to assess stability of the results by omitting one study each time to exclude studies. HWE was estimated by the asymptotic test, and deviation was considered when P<0.05. The potential publication bias of the eligible studies was evaluated by Begg’s and Egger’s regression test quantitatively. Trial sequential analysis (TSA) was performed as described by Xie et al. [36]. The required information size was calculated after adopting a level of significance of 5% for type I error and of 30% for type II error. The data was analyzed using the Stata 14.0 software (version 14.0; State Corporation, College Station, Texas, U.S.A.). A two-tailed P<0.05 was considered statistically significant.

The study selection processes were presented in Figure 1. For polymorphisms of LTA gene (rs1041981, rs2229094, rs2239704, rs746868, rs909253), a total of 24 articles (including 43 case–control studies) with 24577 cases and 33351 controls met the inclusion criteria [11–34]. Sixteen of these studies were performed in Asians, 17 studies were performed in Caucasians, 10 studies in Africans and the others were in mixed ethnic groups (including at least one race). Controls of 30 studies were population-based controls and 13 studies were hospital-based controls. All studies were in compliance with HWE except for two studies [30,32]. Table 1 shows the characteristics of all the eligible studies and genotype frequency distributions of the five LTA polymorphisms included in our meta-analysis. Newcastle–Ottawa scale (NOS) was used to evaluate the quality of the enrolled studies, as shown in Table 2.

rs1041981

The pooled results based on six included studies [11–16] (including 6427 cases and 9714 controls) indicated that no significant association between rs1041981 polymorphism and cancer risk was found. However, in the stratification analysis by ethnicity, we observed that Asian group was significantly related to a reduced risk of cancer in allelic contrast (B vs A: OR = 0.79, 95% confidence interval (CI) = 0.64–0.97, P=0.027, Figure 2) and dominant model (BB+AB vs AA: OR = 0.67, 95% CI = 0.52–0.87, P=0.002). Moreover, when the subgroup analysis was performed based on source of controls, hospital-based control group was significantly related to a decreased risk of cancer in allelic contrast (B vs A: OR = 0.69, 95% CI = 0.55–0.87, P=0.002) and dominant model (BB+AB vs AA: OR = 0.58, 95% CI = 0.42–0.78, P=0.000) (Table 3).

The pooled results based on four included studies [11,17–19] (including 6227 cases and 8562 controls) indicated that no significant association between rs2229094 polymorphism and cancer risk was found. Further subgroup analysis by ethnicity also indicated that no significant result was uncovered (Supplementary Table S1).

The pooled results based on eight included studies [18–24] (including 3550 cases and 3962 controls) suggested that rs2239704 reduced the risk of cancer in allelic contrast (B vs A: OR = 0.90, 95% CI = 0.85–0.97, P=0.003), dominant model (BB+AB vs AA: OR = 0.88, 95% CI = 0.80–0.96, P=0.006) and recessive model (BB vs AA+AB: OR = 0.88, 95% CI = 0.77–0.99, P=0.040). Furthermore, in the stratification analysis by cancer type, we observed that rs2239704 reduced the risk of NHL in allelic contrast (B vs A: OR = 0.89, 95% CI = 0.83–0.96, P=0.001, Figure 3), dominant model (BB+AB vs AA: OR = 0.88, 95% CI = 0.80–0.97, P=0.011) and recessive model (BB vs AA+AB: OR = 0.83, 95% CI = 0.72–0.95, P=0.006). Moreover, when the subgroup analysis was performed based on ethnicity, source of control and genotyping, we found mixed ethnicity was significantly related to a reduced risk of cancer in allelic contrast (B vs A: OR = 0.89, 95% CI = 0.83–0.97, P=0.006), dominant model (BB+AB vs AA: OR = 0.89, 95% CI = 0.79–1.00, P=0.042) and recessive model (BB vs AA+AB: OR = 0.82, 95% CI = 0.71–0.96, P=0.013) (Table 4).

The pooled results based on four included studies [18,25–27] (including 1351 cases and 2145 controls) indicated that no significant association between rs746868 polymorphism and risk of cancer was uncovered. Moreover, in the subgroup analysis by cancer type, ethnicity and source of control, similar results were found. (Supplementary Table S2).

The pooled results based on 21 included studies [13–15,18–34] (including 7022 cases and 8968 controls) indicated that no significant association between rs909253 polymorphism and cancer risk was found. However, in the stratification analysis by cancer type, we observed that rs909253 polymorphism was significantly related to an increased risk of HCC in allelic contrast (B vs A: OR = 3.52, 95% CI = 2.73–4.54, P=0.000, Figure 4), dominant model (BB+AB vs AA: OR = 4.33, 95% CI = 3.07–6.09, P=0.000) and recessive model (BB vs AA+AB: OR = 4.92, 95% CI = 2.92–8.29, P=0.000). In addition, in the stratification analysis by ethnicity, we observed that rs909253 polymorphism was significantly related to an increased risk of Caucasian ethnicity in allelic contrast (B vs A: OR = 1.10, 95% CI = 1.02–1.20, P=0.019), and mixed ethnicity in allelic contrast (B vs A: OR = 1.09, 95% CI = 1.01–1.17, P=0.024), dominant model (BB+AB vs AA: OR = 1.11, 95% CI = 1.01–1.23, P=0.039). Moreover, in the stratification analysis by source of control, genotyping and HWE, null result was found (Table 5).

Sensitivity analysis were performed to evaluate the influence of each separate case–control study. The results showed that there was no material alteration in corresponding pooled ORs for rs1041981, rs2229094, rs2239704, rs746868, rs909253 (Supplementary Figures S1–S5). In addition, Begg’s test and Egger’s regression test were performed to evaluate the publication bias. As for rs1041981, rs2229094, rs2239704, rs746868 and rs909253, no evidence of publication bias was identified (Supplementary Table S3).

To evaluate random errors, we performed TSA (Figure 5). This analysis showed that the cumulative z-curve did not cross the trial sequential monitoring boundary and the required information size, suggesting that more evidences are needed to verify the conclusions.

In the present study, a total of 24 articles including 43 case–control studies were enrolled to validate the association between five LTA gene polymorphisms (rs1041981, rs2229094, rs2239704, rs746868, rs909253) and the risk of cancer. We identified that rs2239704 was inversely associated with the risk of cancer under different genetic models. However, for LTA rs1041981, rs2229094, rs746868, rs909253 polymorphisms, no significant association with cancer risk was uncovered.

In subgroup meta-analysis stratified by cancer type, we found that rs2239704 was significantly reduced NHL susceptibility. Huang et al. reported rs2239704 polymorphism was correlated with cancer and positive association in North Americans [35]. However, they included studies that contained buccal samples for SNP analysis or insufficient data studies [37–39]. We strictly follow the inclusion and exclusion criteria to include the literature. And our results indicated that rs2239704 was significantly reduced cancer susceptibility in mixed ethnicity, hospital-based control and polymerase chain reaction (PCR) genotyping subgroups. Despite of several possible bias, we still could conclude that rs2239704 could reduce cancer susceptibility.

In the stratified analysis of rs1041981, we found that Asians might have less susceptibility to cancer. Unlike the study by Huang et al. [35], we excluded two studies, one of which was autopsy specimen for SNP analysis [10] and the other was a study of HIV-infected patients [40]. The literature thus incorporated has a better baseline consistency and is more reflective of the real situation. Our results were consistent with the results of Huang et al.[35]. Due to the small sample size, we were unable to evaluate the role of rs1041981 in Caucasians. Larger sample size studies are needed for further evaluation. However, based on the current studies, we might conclude that rs1041981 could reduce cancer susceptibility in Asians.

For LTA rs2229094 and rs746848, only four studies reported their relationship with cancer in each group. No significant results were found. Huang et al. reported positive association between rs2229094 and cancer risk [35], which could be the bias from report by Takei et al. [10]. Because of the small sample size, we could not draw any conclusions based on current literature.

Although the overall analysis of rs909253 indicated a null result for cancer risk, the risk of cancer for Caucasians and HCC susceptibility were significantly increased in the stratified analysis by ethnicity and cancer types. In addition, some of the control groups did not match HWE, we can not exclude the possibility that may cause the bias. Then, subgroup analysis by HWE showed that HWE status did not cause the bias of results. Huang et al. did not report the relationship of rs909253 and cancer risk, because it might be present in high linkage disequilibrium with other four SNPs [8,9]. However, our results identified that the function of rs909253 was opposite to rs2239704 and rs1041981. So, further studies with larger sample size are required to identify the role of LTA rs909253 and the linkage disequilibrium with other SNPs.

In the present study, we have put great effort on carefully searching for eligible studies. In order to obtain more accurate and reliable results, we conducted a comprehensive search to verify more eligible studies. Then, we used NOS to evaluate the quality of the included studies, eliminate low-quality studies and improve overall research quality. In order to provide the sources of heterogeneity, subgroup analysis was performed by ethnicity, cancer type, source of controls, genotyping and so on. In addition, sensitivity analysis was used to confirm the stability of the studies. Egger’s and Begg’s tests were used to assess publication bias. However, several limitations in our study should be noted. First, small sample size limits the reliability of the results for some polymorphisms. Second, we just included the studies published in English, which may influence the effects of the polymorphisms. Third, we mainly evaluated the relationship between LTA polymorphisms with various cancers, and we could not get enough data for some cancer types. Fourth, we did not assess the linkage disequilibrium, which might not reflect the real function correctly. In future, more well-designed case–control studies are needed to investigate the functions of LTA polymorphisms.

Our meta-analysis suggests that LTA rs2239704 polymorphism is inversely associated with the risk of cancer, as is LTA rs1041981 polymorphism in Asia. While, LTA rs909253 polymorphism is a risk factor for HCC in Caucasians. Further studies with larger sample size are needed to confirm these findings.

The authors declare that there are no competing interests associated with the manuscript.

The authors declare that there are no sources of funding to be acknowledged.

Zhenwei Han had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Zhenwei Han and Jingdong Li. Acquisition of data: Jingdong Li and Yaxuan Wang. Analysis and interpretation of data: Jingdong Li and Yaxuan Wang. Drafting of the manuscript: Jingdong Li and Zhenwei Han. Critical revision of the manuscript for important intellectual content: Jingdong Li, Yaxuan Wang and Xueliang Chang. Statistical analysis: Yaxuan Wang and Xueliang Chang.

We thank Dr. Chawnshang Chang at University of Rochester Medical Center for helping with the preparation of the manuscript.

CI

confidence interval

HWE

Hardy–Weinberg equilibrium

LTA

lymphotoxin-α

NHL

non-Hodgkin lymphoma

NOS

Newcastle–Ottawa scale

OR

odds ratio

SNP

single nucleic polymorphism

TNF

tumor necrosis factor

TSA

trial sequential analysis