CYP1A2 rs762551 polymorphism contributes to cancer susceptibility: a meta-analysis from 19 case-control studies

Background Genetic polymorphism (rs762551A>C) in gene encoding cytochrome P450 1A2 (CYP1A2) has been shown to influence the inducibility of CYP1A2 expression and thus might be associated with risk of several types of human cancer. However, the results of previous studies on the associations of this polymorphism with risk of cancer are not all consistent. To clarify the potential contribution of CYP1A2 rs762551 to cancer risk, we performed a meta-analysis of the published case–control studies. Methods We used PubMed, Embase, OVID, ScienceDirect, and Chinese National Knowledge Infrastructure databases to identify the related publications for this meta-analysis. The pooled odds ratio (OR) and 95% confidence interval (CI) were calculated using random effect model to evaluate the association of rs762551 with cancer risk. A χ2-based Q-test was used to examine the heterogeneity assumption and the funnel plot and Egger’s test were used to examine the potential publication bias. The leave-one-out sensitivity analysis was conducted to determine whether our assumptions or decisions have a major effect on the results of the review. Results Our analysis of 19 eligible case–control studies showed a significant association between rs762551C variant with risk of cancer in the genetic model of CC versus AA (OR = 1.30, 95% CI = 1.02-1.64) and the dominant model (OR = 1.19, 95% CI = 1.04-1.36). In subgroup analysis based on ethnicity, the rs762551CC genotype was associated with increased cancer risk (OR = 1.29, 95% CI = 1.27-1.63 in co-dominate model and OR = 1.17, 95% CI = 1.02-1.34 in dominant model in Caucasians, but not in Asians and the mixed population. Conclusion These results suggested that CYP1A2 rs762551 polymorphism is likely to be associated with susceptibility to cancer in Caucasians.


Background
The cytochromes P450 (CYPs) played an important role in the development of various cancers, since they involved in the metabolic transformation of numerous endogenous and exogenous compounds including carcinogens and anti-cancer drugs [1]. Cytochrome P450 1A2 (CYP1A2), as one of important CYP enzymes, is responsible for the metabolic activation of pro-carcinogens such as heterocyclic aromatic amines (HAA), polycyclic aromatic hydrocarbons (PAHs) and 4-methylnitrosamino-1-(3-pyridyl)-1-butanone (NNK) and further contributes to the risk of cancer [2,3].
CYP1A2 gene has been mapped on chromosome 15q24.1 and head to head shares a bidirectional promoter with CYP1A1 gene [4]. It has been demonstrated that the expression of CYP1A2 existed large interindividual variability in the liver [5] and it is believed that the expression of CYP1A2 is regulated by constitutive expression and inducible expression from environmental chemicals [3,6].
Recently, many studies have investigated the association of CYP1A2 polymorphisms and the risk of various cancers, including lung cancer, breast cancer, colorectal cancer, stomach cancer and others in various populations [15][16][17][18]. However, the results of these studies are not consistent and inconclusive. Considering the role of CYP1A2 in defending against environmental carcinogens and in the development of cancers, we performed a systematic meta-analysis from all eligible studies to address the overall risk of CYP1A2 variants in the development of all cancers involved.

Identification and eligibility of relevant studies
Remote PubMed, Embase, OVID, ScienceDirect, and Chinese National Knowledge Infrastructure database (between January 2003 and December 2011) was searched using the search terms: CYP1A2/P4501A2/phase II enzymes, polymorphism/polymorphisms/genotype and cancer/ carcinoma/adenomas to identify all publications, which investigated the association of the CYP1A2 polymorphism with cancer risk in all ethnic populations. We evaluated the titles and abstracts of all relevant papers, but excluded case reports, editorials and reviews. All publications in English language with available full text matching the eligible criteria were retrieved. For inclusion in this metaanalysis, the identified articles had to provide information on the following: (1) using a case-control design, (2) sufficient data for examining an odds ratio (OR) with 95% confidence interval (CI) and (3) was a study of the CYP1A2 rs762551 polymorphism and cancer risk, (4) genotype distributions of polymorphism are consistent with hardy-Weinberg Equilibrium (HWE). In addition, we checked the references of relevant reviews and eligible articles that our search retrieved by two investigators independently.

Methods for quantitative analysis
We examined the association between CYP1A2 rs762551 A>C polymorphism and the risk of cancer by calculating pooled odds ratio (ORs) and 95% confidence intervals (CI) in genetic model of CC versus AA, dominant model (CC+CA versus AA) and recessive model (CC versus CA+AA). The significance of pooled OR was tested by Z test. The χ2-based Q-test was also used to examine the heterogeneity assumption [19]. If studies' findings only differ by the sampling (P≥0.05), a fixed-effects model could be used to calculate the combined OR. By contrast, if the P value of the Q tests is below 0.05, which showed that the study results statistically differ by heterogeneous case and sampling, a random-effects model could be more suitable. Since we used accumulating data from a series of studies, which had been conducted by researchers operating independently, the random model was more easily justified than the fixed model [20,21]. The summary OR and 95% CI were calculated under the random effect model.
The leave-one-out sensitivity analysis was conducted to determine whether our assumptions or decisions have a major effect on the results of the review by omitting each study (one at a time) [22]. Furthermore, subgroup analyses were performed to test whether the effect size varied by the ethnicity and the source of control population. To evaluate the published bias, we used funnel plot analysis, which is graphical display of sample size plotted against effect size for the studies included in a metaanalysis [23]. To test for funnel plot asymmetry, Egger's test was performed [24]. All of calculations were performed using R program.
The characteristics of populations and cancer types of final 19 publications were listed in Table 1. This metaanalysis involved in 5 breast cancer studies, 5 lung cancer studies, 4 colorectal cancer studies and 5 studies with other cancer types. Of these, there were 10 studies conducted in Caucasian, 7 studies in Asian, 1 study in Tunisian, and 1 study included multiple ethnicities. There were 6 population-based studies, 12 hospital-based studies, and 1 study with mixed controls study. Variant genotyping methods were used, which included polymerase chain reaction restriction fragment length polymorphism assay (PCR-RFLP) in 14 studies, TaqMan assay, MassARRAY, microarray, arrayed primer extension (APEX) and unknown method in one study each. Overall, the genotyping frequencies of CYP1A2 rs762551 polymorphism were in agreement with the Hardy-Weinberg equilibrium in both cases and controls.

Quantitative synthesis
Regarding CYP1A2 rs762551 polymorphism, the eligible studies involved in 8218 cases and 11165 controls. For each study, we investigated the association based on the assumption of different inheritance models of CYP1A2 rs762551 A>C polymorphism. In all inheritance models of rs762551 polymorphism, due to the between-study heterogeneity in the individual studies (all P for Q test < 0.01 and I 2 >25%), the random-effect model was used to analyze the data [42]. We identified that rs762551 polymorphism had a weak correlation with the risk of cancer (CC versus AA, OR = 1.30, 95% CI = 1.02-1.64; dominant model, OR = 1.19, 95% CI = 1.04-1.36), but not in recessive model (OR = 1.19, 95% CI = 0.99-1.44) ( Figure 2 and Table 2).
Several subgroup analyses were also performed according to the population ethnicity, cancer type, or source of control population (   mixed population, respectively ( Figure 2 and Table 2). We didn't observe any significant association among other subgroups (cancer type and source of control) subgroup in any genetic model using random effect model (Table 2).

Sensitivity analysis
In order to compare the sensitivity of the meta-analysis, we conducted a leave-one-out sensitivity analysis (Additional file 1: Table S1). A single study involved in this meta-analysis was evaluated each time to reflect the influence of the individual data set to pooled ORs. The results pattern was not impacted by single study in all genetic models. The P for Q test and the I 2 value also showed that none of single study affected the heterogeneity of this meta-analysis.

Statistical uncertainty of finding
For risk assessment, statistical uncertainty is associated with the model selected [43]. In this meta-analysis, we evaluated the association of CYP1A2 rs762551 polymorphism with cancer risk in different genetic models (CC versus AA model, dominant model and recessive model). The results showed the difference among these genetic models. We found the rs672551 polymorphism was associated with cancer risk under CC versus AA model and dominant model, but not under recessive model.

Bias diagnostics
To evaluate the publication bias of rs762551 variant in the overall meta-analysis, the funnel plot and Egger's test were used. In this analysis, the funnel plot showed a relatively symmetric distribution (Figure 3), but the point cloud didn't have a distinctive form. No publication bias was detected by the Egger's test (t = 1.4870, P = 0.1553). However, the deficient funnel form of the funnel plot could be due to the relatively high heterogeneity with respect to the different ethnicity and the source of control population. Furthermore, the number of studies was relatively small and the publication bias may still exist.

Discussion
Various studies provided evidences that genetics play an important role in determining cancer risk and association studies have been identified to evaluate cancer susceptibility [44]. However, many association studies failed to provide convincing evidence of linkage and have resulted in contradicting findings, especially in small sample sizes [45]. Meta-analysis provided a popular method for combining world literatures across studies to resolve the statistical power and discrepancy problem in associate studies [46]. Based on 19 studies providing data on CYP1A2 rs762551 polymorphism and cancer risk, we conduced a meta-analysis involving in 8218 cancer cases and 11165 controls to indicate if the rs762551 polymorphism was significantly associated with risk of cancer. We evaluated the publication bias. The CYP1A2 rs762551 A>C genotypes funnel plot was approximately symmetrical and the Eggle's test showed that there is no publication bias in the study of CYP1A2 rs762551 (P = 0.1553). We found that the carriers of CYP1A2 rs762551 C allele had a weak effect on the overall cancer risk in allele genetic model and in the dominant genetic model. These results suggested that the CYP1A2 rs762551 polymorphism might be useful for assessing cancer risk. CYP1A2 is a critical enzyme involved in drug metabolism and carcinogen bioactivation. The expression and activity of CYP1A2 has been demonstrated to relate to the risk of various cancers [2,6,41,47]. CYP1A2 gene is genetically polymorphic in human. To date, 177 SNPs have been deposited in the NCBI database and the frequency of these SNPs varies by ethnicity. Many of these SNPs are in linkage disequilibrium and a few SNPs have been reported to be functional [14,48]. For example, CYP1A2*1F (rs762551) polymorphism can result in 2-3 fold increase in activity/protein and has been associated with inducibility [49]. Our current analysis showed that the CC genotype of rs762551 elevated the individual susceptibility to the cancer risk. This is consistent to the function of CYP1A2 rs762551 polymorphism. However, we didn't observe any correction of CC genotype with the cancer risk in subgroup analysis by cancer type. This biochemical mechanism was still unclear. In addition, due to the low OR in our study, our results should be interpreted cautiously. The CYP1A2 polymorphism may prove to be useful for assessing cancer risk.
It has been well known that cancer occurrence and mortality varied by ethnicity and geographic location [50]. In this meta-analysis, all subjects were subgrouped into three groups (Caucasian, Asian and other populations). No association of rs762551 polymorphism with cancer risk was detected in Asian and mixed population, while increased cancer risk was demonstrated in Caucasians. This finding reflected the difference of cancer susceptibility  in different ethnicity, due to different genetic background and environmental exposure. However due to the low OR in this meta-analysis, further investigation still need to be conducted in a large scale Asian population.
Any meta-analysis has it limitations. To better interpreting the finding, several limitations need to be considered in current analysis. Firstly, potential publication biases may exist in this meta-analysis because studies excluded the non-English-language publications. Secondly, the total study size was still too small to perform subgroup analysis. Thirdly, this meta-analysis was based on unadjusted data due to a lack of detailed genotype information stratified by main confounding variables, such as gender, age, smoking status in original articles.

Conclusions
In summary, our meta-analysis demonstrated a weak association of CYP1A2 rs762551 polymorphism with cancer risk, mainly in Caucasian population. However, as a potentially powerful tool for assessing population effects of genetic variants, meta-analysis cannot replace for adequate genetic association studies. Also, to reach a more definitive conclusion, further gene-gene interaction and gene-environment interaction studies, which based on large sample size, are still needed in different population.