CYP1A1 Ile462Val Polymorphism Is Associated with Cervical Cancer Risk in Caucasians Not Asians: A Meta-Analysis

Objective: Previous studies have reported that Ile462Val polymorphism in the gene Cytochrome P450 1A1 (CYP1A1) is associated with the risk of cervical cancer, but inconsistent results have emerged. Hence, we performed this updated and cumulative meta-analysis to ascertain a more accurate association between CYP1A1 Ile462Val polymorphism and risk of cervical cancer. Methods: Studies involving the CYP1A1 Ile462Val polymorphism associated with cervical cancer risk were searched from the databases of PubMed, Scopus, ScienceDirect, and Chinese National Knowledge Infrastructure (CNKI). The strength of correlation was evaluated through calculating summary odds ratios (ORs) with the corresponding 95% confidence intervals (95% CIs). Subgroup analyses according to ethnicity, source of control and HWE were completed to further explore specific association between the polymorphism and the cancer risk. Results: Altogether, 11 eligible case-control studies were ultimately encompassed into the current meta-analysis, with 1,932 patients and 2,039 healthy controls. The total analysis revealed a borderline relationship between CYP1A1 Ile462Val polymorphism and cervical cancer risk in general population. Interestingly, after subgroup analyses based on ethnicity and source of control, the polymorphism increased the susceptibility of cervical cancer in Caucasian (G vs. A: OR = 1.97, 95% CI = 1.24–3.13; GG vs. AA: OR = 3 .24, 95% CI = 1.24–8.46; GA vs. AA: OR = 1.62, 95% CI = 1.25–2.10; GA+GG vs. AA: OR = 1.68, 95% CI = 1.16–2.43; GG vs. AA+GA: OR = 2.73, 95% CI = 1.05–7.10) and population-based (G vs. A: OR = 1.49, 95% CI = 1.10–2.02; GA vs. AA: OR = 1.41, 95% CI = 1.20–1.67; GA+GG vs. AA: OR = 1.40, 95% CI = 1.19–1.64) groups. Conclusion: The CYP1A1 Ile462Val polymorphism may enhance the susceptibility to cervical cancer in Caucasian females.


INTRODUCTION
Cervical cancer is followed by breast cancer, which is the first common cancer among women all over the world (Denslow et al., 2012). Besides the onset of cervical cancer was aged 15-49 years in developing countries (Forouzanfar et al., 2011). Cervical cancer is a major threat to woman's health and quality of life, and is also the focal point and the difficulties for medical workers. Hence, to seek the risk factor and people who may be at high risk of cervical cancer for prevention is a significant and important work. As we know, smoking (Sood, 1991;Zeng et al., 2012) and human papillomavirus (HPV) infection (Patel et al., 2016) are the classical risk factor for cervical cancer. However, some women without smoking and HPV infection also got cervical cancer, why? That's might be other factors play a role in the onset of cervical cancer, such as genetic background.
Cytochrome P450 (CYP) gene family and susceptibility to many cancers has been the most widely studied (Rodriguez-Antona et al., 2010). CYP1A1 is belonged to CYP gene family1, subfamily A, polypeptide 1 and has two major functional nonsynonymous polymorphisms: MspI polymorphism m1 and Ile462Val polymorphism m2 (Sugawara et al., 2003). Ile462Val polymorphism is a heme-binding site due to the replacement of isoleucine (Ile) by valine (Val), which caused by G to A transition (A4889G) in exon 7 at codon 462, also called rs1048943 (https://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref. cgi?rs=1048943). There are five published meta-analyses directly or indirectly investigated the association between CYP1A1 Ile462Val polymorphism and risk of cervical cancer (Sergentanis et al., 2012;Yang et al., 2012;Wu et al., 2013;Qin et al., 2014;Wang et al., 2015), however, they obtained inconsistent results and with some deficiency (Table 1). For example, the meta-analysis performed by Yang et al. was conducted in 2012 involving ten case-control studies, and found that CYP1A1 Ile462Val polymorphism was associated with increased risk of cervical cancer in general populations. Besides, ethnic subgroup analyses showed a significant association was found in Caucasians but not in Asians (Yang et al., 2012). However, two publications (Geng et al., 2010;Shi et al., 2011) involved the same subjects were both included in above meta-analysis. Moreover, several new original studies (Abbas et al., 2014;Roszak et al., 2014;Li et al., 2016) on this topic were published since then. The recent meta-analysis by Wang et al. pooled the data of 8 case-control studies and indicated the CYP1A1 Ile462Val polymorphism might be a risk factor for cervical cancer (Wang et al., 2015). Since data extracted from two articles (Sugawara et al., 2003;Joseph et al., 2006) were repeatedly included in the meta-analysis, six case-control studies were actually included. Also, due to different inclusion criteria and uneven sample sizes, several meta-analyses (Wu et al., 2013;Qin et al., 2014) presented opposite conclusions. Additionally, subgroup analyses by ethnicity could not be performed due to limited number of studies.
Thus, we identified relevant published reports through a systematic search strategy, and performed this updated and cumulative meta-analysis to reappraise between CYP1A1 Ile462Val polymorphism and risk of cervical cancer. What's more, subgroup and sensitivity analyses were conducted to further ascertain such relationship.

Ligibility Criteria
Any study was considered eligible if it met all of the following criteria: (1) contain information investigated the association between CYP1A1 Ile462Val polymorphism and risk of cervical cancer; (2) as a cohort or case-control design; (3) both cases and controls were clearly diagnosed and all controls were healthy subjects or tissues; (4) included information that allowed for calculation of the odds ratios (ORs) and 95% confidence intervals (CIs). The study investigated the cervical dysplasia, cervical intraepithelial neoplasia (CIN), or cervical squamous intraepithelial lesions (CSIL) was excluded. If any publication involved the same subjects, we compared them according to the population, exposure, control, and study period and chosen the more comprehensive one.

Literature Search
The PubMed, Scopus, ScienceDirect, and Chinese National Knowledge Infrastructure (CNKI) databases were searched for relevant studies that were published up to January 10, 2017. The search strategy included usage of the following key words: cervical, cervix, cancer, carcinoma, CYP1A1, cytochrome P-450, cytochrome P450, polymorphism, Ile462Val, A2455G, and rs1048943. Moreover, we also searched through the references which were cited in the included studies to obtain additional relevant studies. Two independent authors searched and evaluated the eligibility of all studies and any disagreements were resolved by discussion.

Data Extraction
Two authors extracted the following data independently from all included studies: the last name of first author, year of publication, country of origin, ethnicity, sources of controls, sample sizes of cases and controls, genotype distribution in cases and controls, Hardy-Weinberg Equilibrium (HWE) for controls, and genotyping methods. If the HWE was not reported, we assessed it by the chi-square test and significance was set at P< 0.05 according to the genotype distribution in control. Disagreements were resolved by discussion.

Data Analysis
All analyses were conducted using the Comprehensive Meta-Analysis v2.2 software (Zeng et al., 2016a,b). First, the heterogeneity was evaluated using the I 2 test and Cochran Q test, both I 2 < 50% and P > 0.1 was considered acceptance heterogeneity and the fixed-effect model was chosen to pool single study; otherwise, the random-effect model was used. Subgroup analyses according to ethnicity, source of control, and HWE were conducted. Five genetic models were used for overall and subgroup analyses: G vs. A, GG vs. AA, GA vs. AA, (GA+GG) vs. AA, and GG vs. (AA+GA). The odds ratio (OR) and its 95% confidence interval (95%CI) were used to evaluate the pooled effect size. Sensitivity analysis was performed  Sugawara et al., 2003;Huang et al., 2006;Joseph et al., 2006;Taskiran et al., 2006;Gutman et al., 2009;Zhang, 2009Zhang, , 2011Geng et al., 2010;Ding et al., 2011;Shi et al., 2011 10 case-control studies to investigate the influence of a single study on overall analysis, and to test the robustness of overall results. Cumulative metaanalysis was carried out to observe the change when with sample sizes were enlarged (Pabalan, 2010;Rotondi and Bull, 2012). Publication bias was assessed by visual inspection of funnel plots and Egger's test. Figure 1 shows the study selection process. The primary search yielded potentially related publications and finally 11 casecontrol studies involving 1,932 patients and 2,039 healthy controls were included (Sugawara et al., 2003;Huang et al., 2006;Joseph et al., 2006;Gutman et al., 2009;Zhang, 2009Zhang, , 2011Ding et al., 2011;Shi et al., 2011;Abbas et al., 2014;Roszak et al., 2014;Li et al., 2016). Of these 11 case-control studies, four dealt with probands of Caucasian origin (Joseph et al., 2006;Gutman et al., 2009;Abbas et al., 2014;Roszak et al., 2014) and seven referred to Asian origin (Sugawara et al., 2003;Huang et al., 2006;Zhang, 2009Zhang, , 2011Ding et al., 2011;Shi et al., 2011;Li et al., 2016); five studies were out of Hardy Weinberg Equilibrium (HWE) (Huang et al., 2006;Zhang, 2009;Ding et al., 2011;Shi et al., 2011;Abbas et al., 2014). Table 2 lists the main characteristics of identified studies. analysis was influenced only by Huang's study (Figure 3), which indicated that the results were robust. Moreover, cumulative meta-analysis was conducted by adding one study at a time in the order of publication year. The results showed that evidence of the effect of CYP1A1 Ile462Val polymorphism on cervical cancer incidence was stable during the cumulative meta-analysis, but the CIs became increasing narrower (Figure 4). In total analysis, the association between CYP1A1 Ile462Val polymorphism and cervical cancer incidence is borderline and may indicate a role of this variant.

Meta-Analysis Results
In the subgroup analysis for ethnicity, no significant association was found in Asians under any genetic models, but significantly increased risk was observed in Caucasians under all contrasts (G vs. A: OR = 1.97, 95% CI = 1.24-3.13; GG vs. AA:

Publication Bias
As shown in Figure 5, no obvious publication bias was found. The Egger's test also showed no evidence of publication bias (G

DISCUSSION
Cervical cancer is the second most common cancer among women worldwide, frequently occurring in developing countries in particular (Forouzanfar et al., 2011;Denslow et al., 2012). To date, the pathogenesis of cervical cancer has not been identified utterly, but a few risk factors have been confirmed, including human papilloma virus (HPV) infection, smoking, multiparity and long duration of oral contraceptive use (Zeng   Patel et al., 2016). Currently, the complex interaction of environmental and genetic factors is also proposed to possibly conduce to the occurrence of cervical cancer. Cytochrome P450 1A1 (CYP1A1) is a member of the CYP1 family and participates in the metabolic activation of structurally diverse xenobiotics and endobiotics (Rodriguez-Antona et al., 2010). So far, numerous studies have been conducted on the association between CYP1A1 Ile462Val polymorphism and the susceptibility to cervical cancer, but the conclusions are not unanimous (Sugawara et al., 2003;Huang et al., 2006;Abbas et al., 2014). A few meta-analyses (Sergentanis et al., 2012;Yang et al., 2012;Wu et al., 2013;Qin et al., 2014;Wang et al., 2015) were also conducted to figure out the influence of CYP1A1 Ile462Val polymorphism on cervical cancer susceptibility. However, due to different inclusive criteria and uneven sample sizes, these reports presented different conclusions. Although most of them indicated that CYP1A1 Ile462Val polymorphism might be a risk factor for cervical cancer, the effects of the polymorphism on different ethnic groups were not fully clarified. Therefore, we conducted this cumulative meta-analysis to obtain accurate and up-to-date estimates of the association between the CYP1A1 Ile462Val polymorphism and cervical cancer susceptibility.
In the current meta-analysis, we incorporated 11 casecontrol studies involving 1,932 patients and 2,039 healthy  controls to statistically discuss this issue. According to our inclusion criteria, the controls clearly diagnosed as healthy subjects or tissues were included and those investigated the cervical dysplasia, cervical intraepithelial neoplasia (CIN), or cervical squamous intraepithelial lesions (CSIL) were excluded. Thus, the study by Taskiran et al. was not included in this meta-analysis (Taskiran et al., 2006). After data syntheses, we found a borderline correlation of CYP1A1 Ile462Val polymorphism with the susceptibility to cervical cancer in total analysis. And the polymorphism significantly elevated the risk in Caucasian and population-based groups after subgroup analyses by ethnicity and source of control. Sensitivity analysis was conducted to evaluate the influence of individual study on overall analysis, which indicated that the overall results were robust. Furthermore, assessment of publication bias was confirmed by visual inspection of funnel plots and Egger's test, supporting that publication bias across the studies was negligible. Cumulative meta-analysis was also carried out to observe the change when with sample sizes were enlarged, and the CIs became increasing narrower. Recently, the relationships between some other polymorphisms and cervical cancer development were investigated, such as CYP1A1 MspI (rs4646903), COMT (rs4680), and CYP2E1 (rs3813867) polymorphisms. Importantly, a significant association between the MspI polymorphism and elevated cervical cancer risk was observed (Von et al., 2011;Matos et al., 2016). Studies suggest that Ile462Val polymorphism is in tight linkage disequilibrium with the MspI polymorphism, and these variant genotypes are associated with greater CYP1A1 activity or inducibility (Crofts et al., 1994;Ng et al., 2005). Thus, the borderline association between Ile462Val polymorphism and cervical cancer risk cannot be excluded and may indicate a role of this variant, which warrants further functional studies.
Despite certain advantages, there still were some limitations to be addressed. To begin with, the sample sizes of subgroup analyses were relatively small, which might affect the final results. Second, since meta-analysis is a secondary analysis (Zeng et al., 2015), the possible impacts of gene-gene and geneenvironment interactions with this polymorphism on cervical cancer susceptibility were not evaluated due to insufficient information from original papers. Third, only studies published in English or Chinese language were included, some unpublished due to negative findings or other reasons and in other languages might be missed.
In summary, our meta-analysis outcomes revealed a borderline relationship between CYP1A1 Ile462Val polymorphism and cervical cancer risk in general population, and this polymorphism significantly increase the risk in Caucasian females. It provides an important theoretical basis to reveal the CYP1A1 Ile462Val polymorphism and the biological mechanism of developing cervical cancer, which may be helpful to predict the occurrence of cervical cancer in Caucasian females. Importantly, special attention should be paid upon the design of future studies, such as selection of subjects and adjustment of interfering factors so as to uncover any underlying mechanisms and distinct patterns. In view of the above mentioned restrictions, these results should be applied with prudence and further verified by large-scale and well-designed studies with various ethnic groups.

AUTHOR CONTRIBUTIONS
L-NW and X-QR designed this study; FW and Y-HJ searched databases and collected full-text papers; JL and CF extracted and analyzed data; L-NW, FW, and JL wrote the manuscript, X-QR reviewed the manuscript.