A small number of subjects do not always indicate that they are minor variants data for inclusion in a pooled analysis

Abstract

We have read with great interest the article ‘‘Three common TP53 polymorphisms in susceptibility to breast cancer: evidence from meta-analysis’’ published in April 2010 issue of ‘‘Breast Cancer Research and Treatment’’ [1]. The results suggest that IVS3 16 bp Del/Ins is likely an important genetic marker contributing to the susceptibility of breast cancer, and codon 72 has a potential role in the association with breast cancer risk within certain populations or regions. Thus, some methodological issues need to be addressed concerning the data provided in the meta-analysis by Hu et al. [1]. We would like to comment and raise some concerns. The minor variants of IVS3 16 bp (rs17878362) polymorphism in the cases and controls were three and one, respectively [2]. However, a small number of subjects do not always indicate that they are minor variants in the meta-analysis. The data reported by Hu et al. [1] for the study did not seem in accordance with the data provided by Weston et al. [2] in their original publication. The numbers reported by Hu et al. [1] for RR, RP, and PP of codon 72 (rs1042522) polymorphism in the cases and controls were 32-27-6 and 72-42-3. Interestingly enough, after carefully studying the data presented by Weston et al. [2], the frequencies we retrieved were 6-27-32 and 3-42-72, which corresponded to the genotypes of 1-1, 1-2, and 2-2 [2]. In Weston’s research, the minor variants data were 32 and 72, indicating the large number of subjects. Similarly, the numbers reported by Hu et al. [1] for GG, GA, and AA of IVS6 ? 62A[G (rs1625895) polymorphism in the cases and controls were 43-20-2 and 95-22-0. The frequencies corresponding to the genotypes of 1-1, 1-2, and 2-2 [2] were 2-20-43 and 0-22-95. We can conclude that a small number of subjects do not always indicate that they are minor variants data for inclusion in a pooled analysis. Hu et al. [1] confused the original data and made the results not entirely credible. Thus, the odds ratio and the 95% confidence interval for codon 72 (rs1042522) and IVS6 ? 62A[G (rs1625895) polymorphisms in the metaanalysis significantly differed from those calculated by the original data of Weston’s research [2]. It would be valuable if the authors could provide a new and more accurate estimation of the pooled odds ratio after considering our comment. More importantly, the research by Weston et al. [2] was a nested case–control study involving 182 Caucasians, 56 Hispanics, and 46 African-Americans. This implies that the genotypes of 1-1, 1-2, and 2-2 by Hu et al. [1] were only subjects for Caucasians. The genotypes of 56 Hispanics and 46 African-Americans were not included even if they Pei-Hua Lu, Min-Bin Chen and Mu-Xin Wei contributed equally to this work and should be considered as co-first authors.