Introduction

Cancer is the leading cause of death worldwide, both in developed and developed countries. GLOBOCAN estimates of cancer incidence and mortality indicate that 14.1 million new cancer cases and there were 8.2 million deaths in 2012 [1]. Cancer is a complex disease that can be affected by genetic factors [2], which can influence the susceptibility and pathophysiology of cancer [3, 4]. Single nucleotide polymorphisms (SNPs) are the most common genetic variation and have become a focus in cancer research. An increasing number of studies have found that SNPs can predict the risk, prognosis, and effect of drugs on individuals with cancer [5, 6].

Caveolin-1 (CAV-1) is an 18-24 kDa protein located at 7q31.1. The protein has three exons and belongs to the caveolin protein family [7]. It serves as a scaffolding protein in charge of recruiting related signaling molecules to the caveolae and regulating their activity [8]. Studies showed that CAV-1 plays roles in cancer, diabetes, cardiovascular disease, and pulmonary fibrosis [8-11]. In cancer, CAV-1 appears to act as both a tumor suppressor and oncogene. It is downregulated and it inhibits malignant potential of tumor cells in ovarian, colon, and breast cancers [12-14]. However, it is upregulated and promotes malignant potential of tumor cells in bladder and prostate cancer [15, 16].

The potential effect of cancer risk due to the CAV-1 gene, a considerable number of recent studies have indicated the link between CAV-1 polymorphism and cancer susceptibility. More than 20 SNPs in the CAV-1 promoter were detected in different types of cancers, some of which were thought to be genetic risk factors [17, 18]. Six SNPs (rs1997623, rs3807987, rs12672038, rs3757733, rs7804372, and rs3807992) were most frequently studied among Asians. However, individual studies have inconsistent or conflicting findings because of heterogeneity of data collection and inadequate sample sizes. To eliminate this inconsistency, we conducted a meta-analysis of all eligible case-control studies that have been published to date and estimated the cancer risk of six common CAV-1 polymorphisms.

Material and Methods

Literature Search

To identify potentially all eligible studies, we performed a systematic computerized search of PubMed, ISI Web of Knowledge, and Chinese National Knowledge Infrastructure (CNKI) Data using the following keywords: “CAV-1,” “Caveolin-1,” “polymorphism,” “variant,” and “cancer.” The last retrieval date was April 30, 2016. In addition, studies were identified through a manual search of reviews and retrieved studies. The language was restricted to English and Chinese. Studies were included in the analysis if they satisfied the following inclusion criteria: 1) evaluation of the association between CAV-1 polymorphism and cancer susceptibility in Asian; 2) case-control study or cohort studies; and 3) the genotype distribution of the polymorphism in cases and controls was sufficient to estimate the odds ratios (ORs) with a 95% confidence interval (CI) and a P value. The main exclusion criteria were listed as follows: 1) case reports, review articles, and editorials; 2) only case population; 3) duplication of a previous publication; and 4) no available genotype frequency.

Data extraction from published studies were extracted independently by two investigators. Discrepancies were resolved by discussing their findings on every item. For each included study, the following information was collected: first author, year of publication, region, study design, sample size, source of control, genotyping method, allele or genotype frequencies, and P value for Hardy–Weinberg equilibrium (HWE) of controls.

Statistical Analysis

The HWE for each study that describes control subjects was evaluated by a chi-square test. A P value < 0.05 was considered significant disequilibrium. OR with 95% CI was performed to assess the strength of the CAV-1 SNPs and cancer susceptibility. The pooled ORs were performed for allele comparison, dominant and recessive models, homozygote comparison, and heterozygote comparison. The Z test was performed to estimate the significance of the pooled ORs, and P < 0.05 was considered statistically significant. The heterogeneity among the studies was verified by the chi-square-based I² test and the P value of the Q test. P > 0.05 indicated a lack of heterogeneity, I²< 25% indicated low heterogeneity, I² of 25–75% indicated moderate heterogeneity, and I² > 75% indicated high heterogeneity. If P > 0.05, then pooled ORs were calculated by using a fixed effects model. Otherwise, a random effects model was used. Furthermore, sensitivity analyses were performed by sequentially removing each eligible study. Publication bias of literature was assessed with Begg’s funnel plot test. P < 0.05 was considered representative of statistically significant publication bias. All P values were two-sided, and statistical analyses were conducted using the STATA software version 12.0 (StataCorp, College Station, TX, USA).

Results

Characteristics of Studies

Following the search strategy, 51 articles in PubMed, 49 articles in ISI Web of Knowledge, and 8 articles in CNKI were identified. After removing duplicates, 69 potential relevant studies were retrieved. Then, 44 articles were excluded after title and abstract screening. Next, 12 articles were excluded after full text reading. In the end, 13 articles were included in this study. Figure 1 presents the detailed process of selecting and excluding studies. In the 13 articles, all studies were case-control studies with the following SNP distribution: 11 articles were about polymorphism rs1997623 [19-29], 11 articles were about rs12672038 [19-29], 10 articles were about rs3757733 [19, 21-29], 12 articles were about rs3807987 [18-29], 9 articles were about rs3807992 [18, 19, 21-29], and 13 articles about rs7804372 [19, 21-30]. The characteristics of the included studies are summarized in Table 1.

×

Figure 1:

Close

Figure 1. Flow chart of Study Indetification.

×

Table 1:

Close

Table 1. Main Characteristics of Included Studies in this Meta - Analysis.

Quantitative Synthesis

Results of CAV-1 rs1997623, rs12672038, rs3757733, rs3807987, rs3807992, and rs7804372 polymorphisms and cancer risk are presented in Table 2. For rs1997623, no significant associations were found using allele comparison (C vs. A: OR=0.921, 95% CI: 0.730–1.160, P=0.484) and recessive models (CC vs. AC+AA: OR=0.922, 95% CI: 0.728–1.169, P=0.504). For rs12672038, rs3757733, and rs3807992, no significant statistic associations were found in allele comparison, dominant and recessive models, homozygote comparison, and heterozygote comparison. The particular data for rs12672038 were as follows: A vs. G: OR=1.034, 95% CI: 0.967–1.104, P=0.329; AA vs. GG: OR=1.086, 95% CI0.932–1.266, P=0.292; AA vs. AG+GG: OR=1.081, 95% CI: 0.930–1.256, P=0.310; AA+AG vs. GG: OR=1.028, 95% CI: 0.947–1.115, P=0.514; AG vs. GG: OR=1.014, 95% CI: 0.930– 1.107, P=0.751. For rs3757733, A vs. T: OR=0.964, 95% CI: 0.990–1.032, P=0.289; AA vs. TT: OR=0.921, 95% CI: 0.791– 1.073, P=0.292; AA vs. AT+TT: OR=0.928, 95% CI: 0.799– 1.077, P=0.326; AA+AT vs. TT: OR=0.968, 95% CI: 0.888– 1.054, P=0.449; AT vs.. TT: OR=0.980, 95% CI=0.894–1.075, P=0.670. For rs3807992, A vs. G: OR=1.767, 95% CI: 1.486– 2.101, P=0.825; AA vs. GG: OR=0.983, 95% CI: 0.843–1.146, P=0.828; AG vs. GG: OR=1.039, 95% CI: 0.943–1.145, P=0.435; AA vs. AG+GG: OR=0.966, 95% CI: 0.834–1.118, P=0.641; AA+AG vs. GG: OR=1.027, 95% CI: 0.938–1.125, P=0.563. rs3807987 it was significantly associated with an increased risk of cancer in allele comparison (A vs. G: OR=1.767, 95% CI; 1.486– 2.101, P=0.000), dominant models (AA+AG vs. GG: OR=2.034, 95% CI=1.639–2.526, P=0.000), recessive models (AA vs. AG+GG: OR=1.724, 95% CI: 1.402–2.119, P=0.000), homozygote comparison (AA vs. GG: OR=2.243, 95% CI: 1.686–2.984, P=0.000), and heterozygote comparison (AG vs. GG: OR=1.976, 95% CI: 1.621–3.410, P=0.000). rs7804372 was significantly associated with a decreased risk of cancer in allele comparison (A vs. T: OR=0.704, 95% CI; 0.645–0.768, P=0.000), dominant models (AA+AT vs. TT: OR=0.706, 95% CI=0.653–0.762, P=0.000), recessive models (AA vs. AT+TT: OR=0.580, 95% CI: 0.508–0.662, P=0.000), homozygote comparison (AA vs. TT: OR=0.528, 95% CI: 0.461–0.604, P=0.000), and heterozygote comparison (AT vs. TT: OR=0.766, 96% CI: 0.705–0.832, P=0.000)(Figure 2).

×

Figure 2:

Close

Figure 2. Forest Plot for the Caveolin-1 Polymorphism and Cancer Susceptibility in the Allele Comparison.

Heterogeneity and Sensitivity Analysis

Heterogeneity was evaluated using the chi-square-based I2 test and Q test. Results showed that heterogeneity existed in the five comparisons of rs3807987 and allele comparison of rs7804372. In the remaining SNPs, no significant heterogeneity was observed (Table 2). Sensitivity analysis was conducted to verify the effect of each study on the overall OR. Each study was omitted time by time. Results showed that the pooled ORs of these six polymorphisms were not materially altered by the contribution of any individual study (Figure3). Thus, the meta-analysis was statistically robust.

×

Table 2:

Close

Table 2. Meta - Analysis of the Caveolin-1 Polymorphism and Cancer Risk in Asian.

×

Figure 3:

Close

Figure 3. Sensitivity Analysis of the Influence of Allele Comparison in Cancer Risk.

Publication bias

Publication bias in all the studies was assessed by Begg’s funnel plot and Egger’s test. All the graphical funnel plots were symmetrical for the comparison of the genetic models of every SNP (Figure 4). Egger’s test results did not indicate any evidence of publication bias in this study.

×

Figure 4:

Close

Figure 4. Funnel Plot for Publication Bias Test.

Discussion

The overall goal of a meta-analysis is to combine the results of previous studies to obtain an overall conclusion. In this metaanalysis, we detected the associations between CAV-1 rs1997623, rs3807987, rs12672038, rs3757733, rs7804372, and rs3807992 polymorphisms and cancer risk. To our knowledge, this study is the first to use meta-analysis to obtain comprehensive insights into the CAV-1 polymorphism and risk associated with all types of cancer in an Asian population. This study showed that rs3807987 A>G polymorphism increased the risk of cancer in all five comparison models and rs7804372 A>T polymorphism decreased the risk of cancer. However, no significant association was found in rs1997623, rs12672038, rs3757733, and rs3807992 polymorphisms and cancer.

The reasons CAV-1 polymorphisms affect cancer risk were not clearly understood. The polymorphism in the promoter may affect the expression of CAV-1. In non-small cell lung carcinoma, the expression of CAV-1 was statistically correlated with pathologic TNM stage and lymph node metastasis [31] because polymorphism affects the binding site of the promoter or changes the methylation site.

Heterogeneities between studies were found in the rs3807987 and rs7804372 polymorphisms in this meta-analysis. This meta-analysis includes only Asian people. However, the heterogeneities due to ethnicity cannot be eliminated. Heterogeneities due to different cancer types must be considered in this study.

This meta-analysis has some limitations. First, the small sample size in the studies may have resulted in low statistical power. Hence, additional detailed and large-scale studies are necessary. Second, the analysis did not consider gene environment interactions because of the lack of sufficient data. Finally, this study mainly focused on an Asian population. Whether the results can be generalized and applied to other populations remains unclear.

Conclusion

Our meta-analysis results showed that CAV-1 rs3807987 A>G polymorphism increased the risk of cancer, and rs7804372 A>T polymorphism decreased the risk of cancer. No significant association was found in rs1997623, rs12672038, rs3757733, and rs3807992 polymorphisms and cancer. Further studies that include different ethnicities and have a large population size should be conducted to obtain a comprehensive conclusion.