Single Nucleotide Polymorphisms in PLCE1 for Cancer Risk of Different Types: A Meta-Analysis

Background: Recent studies have investigated the relationships between PLCE1 polymorphisms and cancer susceptibility. However, some findings lack consistency. Objectives: In the current study, we conducted a meta-analysis to more accurately evaluate the relationships between PLCE1 (rs2274223, rs3765524, rs753724, rs11187842, and rs7922612) single nucleotide polymorphisms (SNPs) and risk for different types of cancer. Methods: We performed a comprehensive search strategy in PubMed, Web of Science, Medline, EMbase, and Scopus for articles available until 19 March 2018. A total of 54 case-control studies comprising 17,955 cases and 20,400 controls were included in the current meta-analysis, which together comprised a total of 32 publications. The pooled odds ratios (ORs) with 95% confidence intervals (CIs) were used to evaluate relationships between the PLCE1 polymorphisms and cancer susceptibility. All statistical analyses were performed using Stata 11 software. Results: Results of the meta-analysis demonstrated that the rs2274223 polymorphism showed a significant correlation with increased overall cancer susceptibility (AG vs. AA: OR 1.168, 95% CI 1.084–1.259; GG vs. AA: OR 1.351, 95% CI 1.163–1.570; AG+GG vs. AA: OR 1.193, 95% CI 1.103–1.290; GG vs. AA+AG: OR 1.262, 95% CI 1.102–1.446; G vs. A: OR 1.163, 95% CI 1.089–1.242). Results of subgroup analysis showed that the rs2274223 polymorphism was associated with higher risk for esophageal cancer and gastric cancer relative to colorectal cancer and head and neck cancer. In addition, the rs2274223 polymorphism was found to be associated with increased cancer risk, especially among the subgroups comprising Asians, studies with population-based controls, studies employing the TaqMan genotyping method, and studies consistent with Hardy-Weinberg equilibrium (HWE). The association between the rs3765524 polymorphism and reduced overall cancer risk was detected in one specific genetic model (CT vs. CC: OR 0.681, 95% CI 0.523–0.886). Results of subgroup analysis showed that the rs3765524 polymorphism was associated with cancer risk in a specific genetic model among the subgroups of colorectal cancer, esophageal cancer, Asians, studies with population-based controls, and studies consistent with HWE. However, relationships among the PLCE1 rs753724, rs11187842, and rs7922612 polymorphisms and tumor risk were not identified. Conclusions: Results of the current meta-analysis suggested that PLCE1 (rs2274223, rs3765524) polymorphisms are associated with cancer susceptibility.


INTRODUCTION
Cancer has become a major threat to public health worldwide (1). In 2018, there is a predicted 1,735,350 new cancer cases, which are equivalent to over 4,700 new cancer diagnoses each day in the United States, which correspond to an expected 609,640 cancer deaths (2). In addition, cancer has become the leading cause of death in China (3). Therefore, there is an urgent need to investigate cancer, identify relevant biomarkers, and develop strategies for active prevention and early diagnosis and treatment. Cancer is well-established to be the result of a combination of genetic and environmental factors. In the last few decades, extensive experimental and epidemiological findings demonstrated the close association between genetic alterations and tumor risk (4). Single nucleotide polymorphisms (SNPs), the most common form of gene alteration in the human genome, refers to single-nucleotide variations with distribution frequencies that are >1% in the population.
Phospholipase C epsilon1 (PLCE1), which is located on chromosome 10q23, is a member of the phospholipase C protein family (5). In 2010, the results of genome-wide association studies indicated that PLCE1 is associated with cancer risk (6,7). Since then, multiple researchers investigated the relationship between PLCE1 polymorphisms and cancer risk. Cui et al. (8) explored the association between PLCE1 polymorphisms and risk for esophageal squamous cell carcinoma. Li (9), Zhang (10) and other authors investigated the relationship between PLCE1 polymorphisms and colorectal cancer risk. Yuan (11), Malik (12) and other authors investigated the association between PLCE1 polymorphisms and gastric carcinoma risk. Sharma (13) and other authors showed that PLCE1 polymorphisms were associated with susceptibility to developing gall bladder cancer. Among all studies that investigated PLCE1 polymorphisms and cancer susceptibility, the SNPs rs2274223, rs3765524, rs753724, rs11187842, and rs7922612 were five of the most extensively studied polymorphic loci. However, we noted significant differences in the results, sample size, race, or selection of controls among the different studies. In addition, the latest meta-analysis on the relationship between the rs2274223 polymorphism and the overall cancer risk was published in 2015 (14). Furthermore, to the best of our knowledge, no studies conducted meta-analysis of the association of rs3765524, rs753724, rs11187842, and rs7922612 polymorphisms with overall cancer risk. Therefore, in the present study, we summarized all currently qualified casecontrol studies to obtain a more accurate understanding of the relationship between the PLCE1 polymorphism rs2274223 and overall cancer risk [(15) studies were added to the current meta-analysis from the meta-analysis published in 2015 (14)]. And we firstly performed a meta-analysis of the association between the rs3765524, rs753724, rs11187842, and rs7922612 polymorphisms and cancer risk in the overall population.

Literature Search
We carried a comprehensive search strategy to retrieve qualified publications from PubMed and Web of Science until 19 March 2018. The search queries comprised a combination of the Medical Subject Headings (MeSH) and the following keywords: (rs2274223 OR rs3765524 OR rs753724 OR rs11187842 OR rs7922612) OR (PLCE1 OR PLCE OR PPLC OR NPHS3) and (cancer OR tumor OR carcinoma OR neoplasm OR malignancy). In addition, we searched literatures from Medline, EMbase, and Scopus, as complementary data. The references of qualified articles or other reviews were additionally searched. For publications with no available original data, we contacted the authors to ensure that data from all qualified literatures were included in the current meta-analysis. The authors of three out of six publications responded.

Inclusion Criteria and Exclusion Criteria
The inclusion criteria for qualified literatures were as follows: (a) The studies evaluated the associations between PLCE1 polymorphisms (rs2274223 or rs3765524 or rs753724 or rs11187842 or rs7922612) and cancer risk. (b) The study had available genotyping data required for the calculation of the odds ratios (ORs) with 95% confidence intervals (95% CIs). (c) The studies were case-control studies. (d) Studies were complete original articles. Exclusion criteria of qualified literatures were as follows: (a) Articles did not estimate the relationships between the PLCE1 (rs2274223, rs3765524, rs753724, rs11187842, or rs7922612) polymorphisms and cancer susceptibility. (b) The article was a repeated publication. (d) Primary data were missing and were not obtained after contacting the authors. (e) The subjects were not human. Two researchers independently retrieved the literature. In the case of different views in the selected literature, the two researchers discussed to reach an agreement or the decision was made by an independent researcher (Xuelian Li). of cases and controls, and the P-values of the HWE test of the controls. In the case of different views, the two researchers reached an agreement through discussion.

Quality Score Assessment
All qualified literatures were individually assessed by two researchers based on the Newcastle-Ottawa scale (NOS) (16). The assessment results indicated that all selected literatures were of relatively high quality (all NOS scores were ≥6). In addition, the two researchers assessed the quality of the studies using the STREGA (strengthening report of genetic association studies) quality score system (15). All STREGA scores were >12, which indicated that the quality of the studies was moderate-high or high.

Statistical Analysis
Hardy-Weinberg equilibrium (HWE) was examined by performing a Chi-square test in the controls. Heterogeneity was evaluated by conducting Q-test and I 2 -test. In addition, the pooled ORs with 95% CIs were calculated based on the random effects model when heterogeneity was significant (I 2 > 50%) (17). Otherwise, pooled ORs with 95% CIs were calculated according to the fixed-effects model (18). The pooled ORs with 95% CIs were used to evaluate relationships between the PLCE1 polymorphisms (rs2274223, rs3765524, rs753724, rs11187842, and rs7922612) and cancer susceptibility. To investigate the potential sources of heterogeneity across different studies, stratification and meta-regression analyses were conducted. Sensitivity analyses were carried out to evaluate the stability of the results. The effect of publication bias was evaluated using Begg's funnel plot (19) and Egger's test (20). All the above analyses were performed using Stata 11 software. P < 0.05 was considered statistically significant.

Study Characteristics
A total of 32 literatures were eventually included based on the above described comprehensive search strategy. The workflow of the enrollment in the meta-analysis is presented in Figure 1.

Meta-Analysis of the Association Between PLCE1 rs3765524 Polymorphism and Cancer Susceptibility
There were eight qualified case-control studies in this metaanalysis, which assessed the relationship between the PLCE1 rs3765524 polymorphism and cancer susceptibility. The results of meta-analysis on the relationship between the PLCE1 rs3765524 polymorphism and cancer risk are summarized in Table 3 and    Meta-Analysis of the Relationship Between the PLCE1 rs753724, rs11187842, and rs7922612 Polymorphisms and Cancer Risk The rs753724, rs11187842, and rs7922612 polymorphisms were involved in four, four, and three case-control studies, respectively. The results of meta-analysis on the association between PLCE1 (rs753724, rs11187842, and rs7922612) polymorphisms and cancer risk are summarized in Table 4 and Figure S1. Results indicated no significant relationship between the PLCE1 rs753724, rs11187842, and rs7922612 polymorphisms and cancer risk. In addition, further subgroup analysis did not identify statistically significant relationships in any genetic model.

Heterogeneity, Sensitivity Analysis, and Publication Bias
Substantial heterogeneities were identified in our metaanalysis. For example, we observed significant heterogeneity in the overall analysis for rs2274223 (I 2 > 50%). Therefore, we conducted meta-regression analyses to investigate the source of heterogeneity for rs2274223. Results suggested that ethnicity is the likely source of heterogeneity for rs2274223 in the three genetic models (GG vs. AA: P = 0.009; GG vs. AA+AG: P = 0.009; G vs. A: P = 0.048). The genotyping method is a possible source of heterogeneity for rs2274223 in one genetic model (AG vs. AA: P = 0.020) ( Table S1). The results of stratified analyses for rs2274223 were basically consistent with results of meta-regression. However, identifying the source of heterogeneity for rs3765524, rs753724, rs11187842, and rs7922612 was difficult based on stratified analyses. Further, results of sensitivity analysis suggested that the results of meta-analysis were not influenced by any single study in all genetic models for all five polymorphisms, which indicated that our analysis was robust and stable (Figure 4). Next, publication bias was evaluated by Egger's test and funnel plot. (Table S2 and Figure S2).
The results of Egger's test showed that all P-values were >0.05 and that the funnel plots were relatively symmetrical, indicating no publication bias was detected in the current analysis.

DISCUSSION
Tumor pathogenesis involves both genetic and environmental factors. As the effects of genetic mutations on cancer continued to be revealed, many authors have focused on the associations between SNPs and cancer susceptibility. PLCE1 is one of the members of the phospholipase C protein family, which can interact with Ras and participate in cellular signal transduction, produce secondary messengers by hydrolyzing phosphatidylinositol-4, 5-bisphosphate, and regulate cell growth, differentiation, apoptosis, and angiogenesis. (49)(50)(51)(52)(53) The role of PLCE1 in cancer remains controversial. The studies of Wang et al. (54,55) demonstrated that PLCE1 plays a tumor suppressor role in colorectal carcinoma. However, some studies indicated that PLCE1 acts as an oncogene in numerous cancers, such as non-small cell lung cancer (56) and head and neck cancer (57). Recent years have witnessed an increasing number of studies that investigate PLCE1 polymorphisms and cancer susceptibility. Likewise, several meta-analyses assessed the association between PLCE1 polymorphisms and cancer risk. However, most of these studies focused on the relationship between PLCE1 polymorphisms and digestive tract cancer rather than the overall tumor risk.
Our current findings showed that the rs2274223 polymorphism was associated with overall tumor susceptibility in five genetic models, consistent with the results reported by Xue et al. (14). However, the current results were slightly different from those reported by Umar (58), in which the rs2274223 polymorphism showed no significant association with overall cancer susceptibility in one specific genetic model (GG vs. AG+AA). Further stratified analysis revealed that the rs2274223 polymorphism was associated with gastric cancer and esophageal cancer susceptibility, but not with other types of cancer. The above findings were consistent with those reported by Umar (58), but slightly different from the findings of Xue et al. (14), which suggested that rs2274223 polymorphism was not associated with susceptibility to gastric cancer. The results based on the esophageal cancer subgroup were consistent with the results of Wang et al. (59) and Guo et al. (60). Moreover, the results of the stratified analysis indicated that the rs2274223 polymorphism was associated with cancer susceptibility in Asians but not in Caucasians, consistent with the findings of Umar et al. (58). Results of subgroup analysis according to the source of controls identified a relationship between rs2274223 polymorphism and tumor risk regardless of whether controls were obtained from a hospital or a population and were also consistent with the findings of Umar et al. (58). For rs3765524, the results of the present meta-analysis showed that the association between the rs3765524 polymorphism and overall cancer risk was identified in only one genetic model (CT vs. CC). The results of stratified analysis indicated that the rs3765524 polymorphism was associated with colorectal cancer and esophageal cancer susceptibility but not with the other types of cancer. The above findings were distinct from those of Mocellin et al. (61), which identified an association between the rs3765524 polymorphism and gastric cancer susceptibility. Finally, our results of both the total cancer analysis or subgroup analysis indicated that the rs753724, rs11187842, and rs7922612 polymorphisms were not related to tumor risk. The consistencies between the current and previous meta-analyses might be because some of the literatures included in meta-analyses were the same. Meanwhile, the inconsistencies between the current and previous meta-analyses could be attributed to differences in inclusion criteria. For example, the present meta-analysis specifically required that the qualified studies were case-control studies, which was different from the meta-analysis of Mocellin et al. (61). Some limitations still existed in the present analysis, though the analysis was performed carefully. First, relatively few qualified studies were included for investigating rs753724, rs11187842, and rs7922612, and some subgroups included in the stratified analysis had low sample sizes, which might have affected statistical results. Second, unified adjustment about confounders could not be carried out in our analysis because the original data were not obtained. Third, ICD-O codes of cancers from qualified studies were not obtained, and differences in cancers included in the studies might lead to biases. Finally, unpublished materials were not obtained, which might have caused publication bias, although publication bias was not detected based on Begg's funnel plots and Egger's test in this meta-analysis.

CONCLUSIONS
Our findings indicated that the PLCE1 rs2274223 polymorphism is significantly associated with cancer susceptibility in the overall population. On the other hand, the PLCE1 rs753724, rs11187842, and rs7922612 polymorphisms showed no significant associations with cancer risk. In addition, the results suggested that the PLCE1 rs3765524 polymorphism is associated with overall cancer risk under the heterozygote model (CT vs. CC).