Recently, we read with great interest the article entitled “Association between BRIP1 (BACH1) polymorphisms and breast cancer risk: a meta-analysis” published online in Breast Cancer Res Treat, 2013, 137: 553–558 [1]. Pabalan et al. conducted a meta-analysis to examine the association between the Pro919Ser polymorphisms in the BRCA1 interacting protein 1 (BRIP1) gene and breast cancer risk on the basis of eight case–control studies with 5122 cases and 5735 controls. They also studied the risk associated with the two additional BRIP1 C47G and G64A polymorphisms and breast cancer risk on the basis of 1539 cases and 1183 controls, and 667 cases and 782 controls, respectively. The authors found that the association was lacking between the Pro919Ser polymorphisms and breast cancer risk in overall analysis [odds ratio (OR) 0.98–1.02], materially unchanged when confined to subjects of European ancestry (OR 0.96–1.03) or even in the high-powered studies (OR 0.97–1.03). In the menopausal subgroups, premenopausal women followed the null pattern (OR 0.94–0.98) for the Pro and Ser allele contrasts, but not for the Pro-Ser genotype comparison where significant increased risk was observed (OR 1.39, P = 0.002). The G64A polymorphism effects were essentially null (OR 0.90–0.98), but C47G was found to confer nonsignificantly increased risk under all genetic models (OR 1.27–1.40). It is an interesting study.

Nevertheless, careful examinations of the data provided by Pabalan et al. [1] (shown in Table 1 in their original text) reveal four key issues that are worth noticing. Firstly, the data reported by Pabalan et al. [1] for the study of Rutter et al. [2] did not seem in line with the data provided by Rutter et al. [2] in their original publication. The numbers reported by Rutter et al. for cases and controls, are 58 and 30, respectively [2]. Interestingly enough, after carefully examining the data reported by Pabalan et al. [1], the numbers are 116 in cases and 60 in controls, respectively. Secondly, Rutter et al. [2] also reported the association of BRIP1 G64A polymorphisms with breast cancer risk. But the data were not included in Pabalan et al’s study [1]. Thirdly, one eligible paper [3] focusing on the association of BRIP1 G64A polymorphisms with breast cancer risk was not included in Pabalan et al’s study [1]. Fourthly, one eligible paper [4] focusing on the association of BRIP1 Pro919Ser polymorphisms with breast cancer risk was not included in Pabalan et al’s study [1]. Therefore, the conclusions by Pabalan et al. [1] are not entirely reliable. It is required to clarify the association between BRIP1 polymorphisms and the risk of breast cancer comprehensively and objectively. We re-evaluated this association by performing an updated meta-analysis on the basis of a total of ten studies with 6491 cases and 7181 controls for Pro919Ser, three studies with 1481 cases and 1154 controls for C47G and five studies with 3214 cases and 3381 controls for G64A. Further subgroup analysis was also conducted in this study stratified by source of control and ethnicity. In addition, cumulative meta-analysis was performed to investigate the tendency of results by accumulating single study year by year, which could be used to determine whether new relevant studies are needed or not. We believe that our results will provide objective and comprehensive evidence for the association between BRIP1 polymorphisms and breast cancer risk.

Table 1 Characteristics of the included studies associating BRIP1 polymorphisms in breast cancer
Full size table

Table 1 listed the general information of selected studies in this meta-analysis. Table 2 listed the summary odds ratios of the association between BRIP1 polymorphisms and breast cancer risk. Overall, we did not observe significant association between BRIP1 Pro919Ser polymorphisms and breast cancer risk under the genetic model of Ser-allele versus Pro-allele (OR = 0.99, 95 % CI 0.97–1.01) (Fig. 1a). We did not observe the association of BRIP1 C47G polymorphisms with breast cancer risk under the genetic model of G-allele versus C-allele (OR = 1.02, 95 % CI 0.99–1.05) (Fig. 1b). We also did not observe the association of BRIP1 G64A polymorphisms with breast cancer risk under the genetic model of A-allele versus G-allele (OR = 0.99, 95 % CI 0.97–1.02) (Fig. 1c). The cumulative meta-analysis accumulated the studies in accordance with the year of publications and the results showed that there was still no significant association between BRIP1 polymorphisms and breast cancer risk under allele models, the cumulative ORs were 0.99 with 95 % CI 0.93–1.03 for Pro919Ser, 1.07 with 95 % CI 0.95–1.22 for C47G and 0.99 with 95 % CI 0.92–1.06 for G64A, respectively (Fig. 2a, b, c). In subgroup analysis by source of control, we did not observe a significant association between BRIP1 Pro919Ser polymorphisms and breast cancer under the allele model of Ser-allele versus Pro-allele on the basis of population-based controls (OR = 1.00, 95 % CI 0.98–1.02) (Table 2). We did not observe any association between BRIP1 Pro919Ser polymorphisms and breast cancer risk among Europeans when stratified by ethnicity (Table 2).

Table 2 Summary effects of BRIP1 polymorphisms in breast cancer
Full size table
Fig. 1
figure 1

Forest plots for the odds ratio of the association between BRIP1 polymorphisms and breast cancer risk (a Ser-allele vs. Pro-allele of Pro919Ser; b G-allele vs. C-allele of C47G; c A-allele vs. G-allele of G64A)

Full size image
Fig. 2
figure 2

Cumulative meta-analysis for the association between BRIP1 polymorphisms and breast cancer risk (a Ser-allele vs. Pro-allele of Pro919Ser; b G-allele vs. C-allele of C47G; c A-allele vs. G-allele of G64A)

Full size image

The shape of funnel plots seemed to be approximately symmetrical among total population (Fig. 3a, b, c). Egger’s test and Begg’s test suggested that there was no obvious publication bias in this meta-analysis excerpt in the model of Ser-allele versus Pro-allele (Table 2). To evaluate the stability of the results of this current meta-analysis, a sensitivity analysis was conducted through sequentially removing each individual study. The sensitivity analysis showed that our results were robust and were not influenced by any single study (Fig. 4a, b, c).

Fig. 3
figure 3

Funnel plots for the association between BRIP1 polymorphisms and breast cancer risk (a Ser-allele vs. Pro-allele of Pro919Ser; b G-allele vs. C-allele of C47G; c A-allele vs. G-allele of G64A)

Full size image
Fig. 4
figure 4

Sensitivity analysis for the association between BRIP1 polymorphisms and breast cancer risk (a Ser-allele vs. Pro-allele of Pro919Ser; b G-allele vs. C-allele of C47G; c A-allele vs. G-allele of G64A)

Full size image

In conclusion, the results of the study by Pabalan et al. [1] should be explained with caution. To reach a definitive conclusion, well-designed studies with large sample size are required to verify the association between BRIP1 polymorphisms and breast cancer risk. We hope that this remark will contribute to more accurate elaboration and substantiation of the results presented by Pabalan et al. [1].