Factors influencing the association between CYP17 T34C polymorphism and the risk of breast cancer: meta-regression and subgroup analysis

Abstract

A number of studies have been investigated the association between CYP17 T34C polymorphism and the risk of breast cancer; the results of these studies are inconsistent, however. This fact implies that the effect of CYP17 T34C polymorphism on susceptibility to breast cancer may be modified by other risk factors. In order to provide a more definitive conclusion, a full meta-analysis combining and summarizing 24 studies was first performed. Both traditional method and Bayesian approach were applied. Odds ratio was estimated using a dominant mode of inheritance after a biological justification for the choice of genetic model. The results of homogeneity analysis (H = 1.16, I (2) = 25.4%, and P = 0.127) suggested the presence of heterogeneity across the studies. Thus, random effects models simulated by the DerSimonian-Laird method were employed. The capability of a Bayesian approach was highlighted in the estimation of a pooled odds ratio and 95% confidence interval. The results of meta-analysis (OR = 1.001, CI = 0.832-1.208) suggest no significant association in the combined populations. Furthermore, Bayesian meta-regression and subgroup analysis were conducted to investigate the sources of heterogeneity. The risk factors evaluated in the study were menopausal status, ethnicity, age at menarche, age at first birth, parity, use of oral contraceptives, body mass index (BMI), and use of hormone repair therapy (HRT). After these population stratifications, there was evidence indicating that a possible impact of menopausal status, age at menarche, and BMI on the association between CYP17 T34C polymorphism and the risk of breast cancer.

Evaluation of <b><i>CYP17A1</i></b> and <b><i>LEP</i></b> Gene Polymorphisms in Breast Cancer

“…There are 3 meta-analyses that have reported an association between the CYP17A1 T27C polymorphism and breast cancer risk, all published in 2010 [12,22,23]. These meta-analyses included 24-43 studies from different populations and showed no association between the CYP17A1 T27C polymorphism and breast cancer, which is in line with our results.…”

Possible risk modifications in the association between MnSOD Ala-9Val polymorphism and breast cancer risk: subgroup analysis and evidence-based sample size calculation for a future trial

“…Therefore, a biological justification for the choice of genetic model was required [27]. The procedure to capture the inheritance mode in association studies has been described previously [28]. In brief, k (the ratio of log OR Val/Ala and log OR Ala/Ala ) was determined; the pairwise estimates of the OR of Ala/Ala versus Val/Val (OR Ala/Ala ) and the OR of Val/Ala versus Val/Val (OR Val/Ala ) were performed by carrying out two separate meta-analyses [27].…”

“…The traditional and Bayesian meta-analysis procedures have been described previously [28,34]. In traditional metaanalysis, Mantel and Haenszel method and DerSimonian and Laird method were employed in fixed effects and random effects models, respectively.…”

“…The estimation process was performed using Markov Chain Monte Carlo (MCMC) and Gibbs sampling algorithms through Bayesian computation software WinBUGS [37]. The code for the model has been described elsewhere [28]. Ten thousand iterations were dismissed as burn-in, and the subsequent 100,000 iterations were used for parameter estimation.…”

“…The full Bayes model also provided estimates of s 2 (i.e., between-study variance) as 0.008 (CI = 0.0005-0.030). As previously described, the prior distribution for s 2 allows the CI wider than that derived from traditional model [28]. In addition, the overall goodness of fit of the model was assessed and the result was satisfactory (see Supplemental Data).…”