Search results and study selection
385 researches were obtained from the original search,21 duplicate records were removed. Following screening title and abstract, 291 publications were excluded owing to case reports, case series, comments, reviews, letters, editorials, animal models. The remaining 46 records were obtained for full-text assessment. Whereafter,32 publications were removed according to the items of exclusion criteria after checking the full-text. Such as veteran studies were excluded for its very high male proportion (almost 100%) which may cause bias to the result. Besides, publications without enough data for calculating pooled HRs for PFS and OS were also excluded. Eventually, 14 studies were eligible for this meta-analysis. A PRISMA flow chart describing the process of study screen and selection was presented in Figure 1.