Association between CYP17 T-34C rs743572 and breast cancer risk

Association between CYP17 T-34C (rs743572) polymorphism and breast cancer (BC) risk was controversial. In order to derive a more definitive conclusion, we performed this meta-analysis. We searched in the databases of PubMed, EMBASE and Cochrane for eligible publications. Pooled odds ratios (ORs) with 95% confidence intervals (95% CIs) were used to assess the strength of association between CYP17 T-34C polymorphism and breast cancer risk. Forty-nine studies involving 2,7104 cases and 3,4218 control subjects were included in this meta-analysis. In overall, no significant association between CYP17 T-34C polymorphism and breast cancer susceptibility was found among general populations. In the stratified analysis by ethnicity and source, significant associations were still not detected in all genetic models; besides, limiting the analysis to studies with controls in agreement with HWE, we also observed no association between CYP17 T-34C polymorphism and breast cancer risk. For premenopausal women, we didn't detect an association between rs743572 and breast cancer risk; however, among postmenopausal women, we observed that the association was statistically significant under the allele contrast genetic model (OR = 1.10, 95% CI = 1.03–1.17, P = 0.003), but not in other four models. In conclusion, rs743572 may increase breast cancer risk in postmenopausal individuals, but not in premenopausal folks and general populations.


INTRODUCTION
Breast cancer (BC), the most frequent malignant neoplasm among female worldwide, accounts for approximate 25% of women malignant tumor. It is reported that 1.67 million people were diagnosed as BC ever year, therefore it has become a serious health issue, especially in the developing countries [1]. It is well known that the lifetime presence of the estrogen in the blood is an important pathogenic factor of BC, and this is in consistence with the low incidence of the breast cancer in males that is due to the lower estrogen levels and lower breast tissue volume. By now, researches on the status of hormone receptors and/or menopause associated with Research Paper genetic alterations in BC risk have attracted an increasing number of attention, and lots of genes, including BRIPI, CHEK2, MDM, TGFB, TP53, BRCA1, BRCA2, and PTEN, and also several gene polymorphisms. Among genes of this family, CYP17, CYP19 and CYP1A1 have important functions in synthesis, metabolism and maintaining the levels of the androgen and estrogen hormones [2]. Previous published reasearches have demonstrated that estrogen act as a crucial role in the formation of BC; in addition, evidences have also been found about the positive role of cell surface receptors of estrogen in tumorigenesis [3]. Nevertheless, the precise mechanism behind estrogen in the formation of BC remains unknown. Previous studies have indicated that cytochrome P450c17α, which is a key enzyme in the synthesis of estrogen, and could increase the breast neoplasm risk [4]. The cytochrome P450c17α enzyme, predominantly catalyzes the formation of the precursor dehydroepiandrosterone (DHEA). Meanwhile, precursor DHEA could further be converted into estrogen through a succession of tissue-specific pathways [5,6]. Estrogen, plays a vital part in the etiology of BC and identified the risk between estrogen and BC could well elucidate the biosynthesis and metabolism mechanisms. So far, more and more researches have demonstrated the correlation of estrogen-related genes genetic variations with BC risk. The CYP17T-34C (rs743572) polymorphism which is located on the human chromosome 10, in the 50-untranslated region has been most commonly reported [7].
Many studies about the genetic mutations or SNP occurring in CYP17 gene could enhance CYP17's transcription rate and increase the enzyme cytochrome P450c17 level, resuling in an increasing number of bioavailable estrogen, which is likely to affect the risk and aggressiveness of BC [8]. But many previous article results between rs743572 mutations and BC risk remain conflicting: Han's research [9] revealed that no statistically meaningful correlation of rs743572 with risk of BC. However, significant correlation was found between rs743572 and BC risk in another research on the same theme [10]. Since few new high-quality investigations were published, we performed this study to take a more precise evaluation of rs743572 with the risk of BC.

The main feature of included studies
As showed in Figure 1, 331 references were retrieved at first based on our selection strategy. 186 papers were remained after removing the duplicate reports. After reading titles and abstracts, we excluded 104 studies which were clearly unrelated. In the end, the whole of the rest of the papers were checked based on the inclusion and exclusion criteria. Finally, forty-nine studies on rs743572 and the risk of BC were eventually included in our study.
Thirteen articles showed the number of three genotypes (TT, TC, and CC) among premenopausal women, and thirteen studies report TT, TC, and CC number in postmenopausal women. Main information of included studies were shown in Table 1. Among these qualified researches, seventeen were performed in Asians, twentyfive in Caucasians, one in Africans, one in both Asians and Caucasians, one in both Africans and Caucasians, and four in mixed ethnicity. Moreover, twenty-two studies were considered as moderate-quality studies (NOS scores of these researches were [4][5][6], and other twenty-seven studies were considered as high-quality studies (NOS scores of these studies were seven or above). Except for four included researches were not in agreement with Hardy-Weinberg equilibrium (HWE), genotype distributions in the control groups of other 45 researches were all satisfied with HWE.

Meta-analysis results
Meta-analysis results among overall populations, distribution of this polymorphism in case groups and control groups are presented in Table 2. For premenopausal women and postmenopausal women, distribution of this polymorphism in case groups and control groups are presented in Table 3, and the main outcome of our study are shown in Tables 4 and 5.

Sensitivity analysis
Even though four researches included in our studies were not conformed to the HWE balance (P < 0.05), final consequences were not changed when we excluded the abovementioned four studies. Besides, after performing the sensitivity analysis, the pooled OR values were not statistically significant changed when we delete each of the researches, indicating that this study has good stability and reliability.

Heterogeneity analysis
Heterogeneity was obtained by Q statistic. When the P value more than 0.1 in the Q test, then the fixedeffect models were selected to conduct relevant statistical analysis; otherwise, random-effect models were selected.

Publication bias
No statistical evidence of publication bias was found in the Begg's test and Egger's test. What's more, funnel plot also did not show any evidence of obvious asymmetry (Table 4) ( Figure 4).
Estrogen is mainly produced in the ovaries and mammary glands among premenopausal women. However, in postmenopausal individuals, adipose tissue mainly acts as an important part in estrogen biosynthesis [15,16]. Several studies have reported conflicting results of menopausal and CYP17 polymorphism: the study    by Dunning et al. [17] showed the association between increased A2 genotype and premenopausal breast cancer; while Feigelson et al. [18] reported increasing frequency of A2 genotype associated with postmenopausal BC patients. We observed that rs743572 was correlated with an increasing BC risk among postmenopausal women under the allele contrast genetic model, but not in other models; however, no association was found in premenopausal women. Previous published meta-analysis reported that no association existed both in postmenopausal women and among premenopausal women [12,13,14]. Compared with them, our study used five genetic models to reduce the probability of class I errors, so our result was more reliable.
Unavoidable, there are some limitations in metaanalysis. First, breast cancer is a multifactorial disease involving genetic and environmental interactions; however, it was still not addressed the impact of geneenvironmental interactions in this meta-analysis [19]. Second, the detailed individual information in some studies was unknown; thus, we could not assess the susceptibility of breast cancer according to other risk factors including obesity, family history, radiation therapy in young age, history of pregnancy, breast-feeding, hormone therapy and so on [20]. Last, there are only two studies about Africans, more well designed studies with different population should be performed to make more persuasive conclusions. In summary, our results indicate that rs743572 could increase risk of BC in postmenopausal individuals, but not in premenopausal women and the general population. Further multicenter research with complete risk factors are required to validate the potential role of rs743572 polymorphism in BC. More multicenter studies and complete risk factors are needed to further confirm the possible role of rs743572 polymorphism in the occurrence and development of breast cancer.

Literature and search strategy
We searched the PubMed, EMBASE and Cochrane databases for studies performed prior to March 7, 2017 that reported an association between rs743572 SNP and breast cancer risk. There were no language restrictions in our searching process. The searching strategy was as follow: (breast cancer OR breast carcinoma) AND (polymorphism OR variant OR genotype OR SNP) AND (CYP17 OR CYP17A1 OR P450c17). Besides, the references of the retrieved studies were also reviewed to identify additional eligible studies.

Inclusion criteria
The included studies must meet the following criteria: (1) case-control design; (2) investigating the association between CYP17 T-34C polymorphism and breast cancer risk; (3) sufficient genotyping data that could be used to calculate odds ratios (ORs) and 95% confidence intervals (CIs); (4) all the breast cancer subjects in case groups must be pathologically confirmed. The exclusion criteria were: (1) not case-control studies; (2) review article or commentary; (3) duplicate studies; (4) studies lacking relevant data.

Data extraction
Two reviewers independently extracted the relevant data from the included studies, and discrepancies were resolved during a discussion with a third author. The following information was extracted: the first author, year of publication, country, ethnicity, source of controls, number of cases and controls, and P value for Hardy-Weinberg equilibrium (HWE). In addition, we also evaluated the methodological quality of included studies based on Newcastle-Ottawa Scale (NOS), which scored studies according to three aspects: selection, comparability, and exposure. Therefore, all studies could be divided into three categories: "low quality" studies (score 0-3); "moderate quality" studies (score 4-6); "high quality" studies (score 7-9).

Statistical analysis
The association between CYP17 T-34C polymorphism and BC susceptibility was measured by pooled odds ratios (ORs) and 95% confidence intervals (CIs) in five genetic models, including an allele contrast genetic model, a homozygote genetic model, a heterozygote genetic model, a dominant genetic model, and a recessive genetic model. Pooled ORs were performed for homozygote comparison (TT vs. CC for rs743572), heterozygote comparison (TC vs. CC for rs743572), dominant model (TT/TC vs. CC for rs743572), recessive model (TT vs. TC/CC for rs743572) and allelic model (T vs. C for rs743572) respectively. Statistical heterogeneity was evaluated by I 2 test and Q test, P < 0.05 was considered statistically significant. For I 2 test, the criteria for heterogeneity were as follows: I 2 < 25%, no heterogeneity; 25%-75%, moderate heterogeneity; I 2 > 75%, high heterogeneity. If the P value of Q test was < 0.1, the random-effects model was used; otherwise, the fixed-effects model was applied. Sensitivity analysis was performed by excluding one study at a time to assess the influence of each study on the pooled ORs. Begg's funnel plot and Egger's tests were used to examine publication bias and to evaluate the stability of the results by sensitivity analysis. The P value for Hardy-Weinberg equilibrium (HWE) in controls of every included study was calculated by Chi-square test. Subgroup analysis was performed according to ethnicity. All statistical analyses were performed using STATA version 10.0 software (StataCorp LP, College Station, TX, USA). All P values were two sided, and P < 0.05 was considered statistically significant.