Association of XPC polymorphisms with breast cancer risk

Abstract

To the Editor, We read with great interest the article by Zheng W et al. [1], which assessed the association between Xeroderma Pigmentosum group C (XPC) polymorphisms and breast cancer risk with a meta-analysis. The results suggested that there was no significant association between either Lys939Gln or Ala499Val polymorphism and breast cancer risk in any genetic model or any subgroup based on ethnicity/source of controls. However, a significant association was observed for PAT ± polymorphism under a recessive model (odds ratio 1.41, 95% confidence interval 1.05–1.89). After carefully reading the article, we noted several methodological issues which should be considered. First, in the ‘‘Publication search’’, they did not state whether there was language restriction for the study identification. In addition, in our opinion, ‘‘Xeroderma Pigmentosum group C’’, the full name of ‘‘XPC’’ also should be used as one of the search terms, to search all relevant potential studies. Second, they did not assess Hardy–Weinberg equilibrium (HWE) in controls of each study. Evidence suggested that deviation from HWE might reflect the presence of genotyping errors, population stratification, and selection bias in the controls [2]. Thus, we assessed HWE in controls of included studies [3–10] in Table 1. Third, in the Table 1 of their study, it is better to present the genotype frequencies of cases and controls to be more readable. In addition, I do not know why they categorized the population in the study by Jorgensen et al. [5] as Caucasian. Indeed, the ethnicity was unknown for this study. In the table 2 of the study, it is also better to provide the number of studies for each subgroup analysis. Fourth, there was only one study with 583 cases and 583 controls, which investigated the association between PAT ± polymorphism and breast cancer risk [10]. Thus, the positive conclusion should be made with caution. At last, we excluded two studies which deviated from HWE in controls for Lys939Gln polymorphism and reassessed the association under a dominant genetic model based on ethnicity (Figs. 1, 2). For Lys939Gln polymorphism, there were 4,064 cases and 4,025 controls; for Ala499Val polymorphism, there were 1,838 cases and 2,007 controls. The association was not statically significant for either Lys939Gln or Ala499Val polymorphism. Since the sample size was relative small, especially in subgroups, further well-designed studies with enough statistical power are required to assess the associations. F. Xu D. Zhong (&) Z. Tang M. Zhou Department of Breast Surgery, The Second Xiangya Hospital of Central South University, Changsha, Hunan 410007, China e-mail: zhong_dewu@163.com