VEGF 936C>T polymorphism and breast cancer risk: evidence needed further clarification.

We have read the meta-analysis by Dongying Gu et al. [1] with great interest, which was published in your journal recently. Through quantitative analysis, this meta-analysis suggested that VEGF 936C[T polymorphism might not contribute to breast cancer susceptibility. This meta-analysis was well done with perfect methodology. However, two points should be mentioned when we closely inspect this study. First, at least two eligible studies [2, 3] with 2335 cases and 2397 controls were not included in this meta-analysis according to the inclusion criteria. One study by Pharoah et al. [3] with 2016 cases and 2107 controls was published in 2007. And the other study by Jakubowska et al. [2] with 319 cases and 290 controls was published in 2008. Actually, the data of this study was not overlapped with the latter study by Jakubowska et al. [4] in 2009 when carefully reading the full texts. We believed that the bias would be introduced into the results of the meta-analysis due to this shortage. Therefore, we decided to perform a combined analysis based on the meta-analysis and the two original studies. Through quantitatively combined analysis, the odds ratio was 0.96 (0.78–1.16) for TT versus CC, 0.93 (0.84–1.03) for CT versus TT, 0.94 (0.85–1.04) for TT/CT versus CC and 0.99 (0.81–1.22) for TT versus CT/TT. The P value for the heterogeneity was 0.73 for TT versus CC, 0.19 CT versus TT, 0.18 TT/CT versus CC, and 0.81 TT versus CT/TT, respectively, and the fixed effects model was used. The results further suggested that polymorphism of VEGF ? 936C[T might not contribute to the risk of breast cancer susceptibility. Second, the definitions for the population-based case– control study (PCC) and hospital-based case–control study (HCC) were not clear in this meta-analysis. This point greatly influenced the results of this meta-analysis. For example, if the author defined the PCC as the controls from healthy population, and the HCC from patients, we could clearly ascertain that the study by Lin et al. [5] was not HCC at least.