The association between two polymorphisms in the TYMS gene and breast cancer risk: appraisal of a recent meta-analysis

To the Editor, Recently, we read with great interest the article by Wang et al. [1]. They performed a meta-analysis of 10 case– control studies (4,443 cases and 5,447 controls) to examine the association between thymidylate synthase (TYMS) gene polymorphisms and breast cancer (BC). Their metaanalysis suggests that the TSER polymorphism may increase susceptibility to BC in the Caucasian population, and the TS30-UTR polymorphism may be a genetic determinant for developing BC in the Asian population. However, the authors have several concerns related to the article. First, the results of the meta-analysis should be interpreted with caution. In the conclusion, they stated that ‘‘ethnic background should be carefully considered in further studies.’’ For the TSER polymorphism, the overall odds ratios (ORs) and the corresponding 95% confidence interval (CI) showed a statistical association between this polymorphism and BC risk under homozygote comparison, allele comparison and the recessive model. In the subgroup analysis by ethnicity, the association was detected in the Caucasian population, but not in the Asian population. While, only two studies were conducted in the Asian population, and null findings may be due to a lack of statistical power in the Asian population. Similarly, for the TS30-UTR polymorphism, the overall ORs and the corresponding 95%CI showed a statistical association between this polymorphism and BC risk under homozygote comparison. Subgroup analysis revealed that there was only a marginal association in the Asian population for homozygote comparison (P = 0.05). Nevertheless, the meta-analysis showed a modest association in the overall population (All in HWE) when the Caucasian study by Justenhoven et al. [2] was included in the analysis (P = 0.03). Furthermore, for the TS30-UTR polymorphism, only two studies were conducted in the Caucasian population under homozygote comparison, and, of the two studies, the genotype distributions of controls in one study were not in agreement with Hardy–Weinberg equilibrium (HWE) [3]. There may be no enough power to detect a significant association of the TS30-UTR polymorphism with BC in the Caucasian population. Therefore, the data included in the meta-analysis are insufficient to support the conclusion that ethnic background should be carefully considered in further studies. Second, there are some issues in the methods. In the results, they stated that ‘‘A total of 10 studies (4,443 cases and 5,447 controls) met the inclusion criteria.’’ However, in the methods, they stated that ‘‘The following criteria were used to select the eligible studies... (d) a genotype distribution among the control populations consistent with Hardy–Weinberg Equilibrium (HWE).’’ Of the ten studies, the genotype distributions among the control populations in two studies were not in agreement with HWE [3, 4]. Thus, the two studies did not meet the inclusion criteria and should be excluded. Poor reporting of meta-analysis diminishes their value to clinicians, policy makers, and other users. An international group that included experienced authors and methodologists developed PRISMA (preferred reporting items for systematic reviews and metaanalyses) as a guideline for systematic reviews and metaanalyses. The PRISMA statement consists of a 27-item Y.-F. Zou X.-L. Feng Department of Epidemiology and Biostatistics, School of Public Health, Anhui Medical University, Hefei 230032, Anhui, China