A total of 2962 manuscripts were initially identified through 5 databases. After removal of duplicates, titles and abstracts of the remaining 1565 manuscripts were scanned, and 48 studies were included for full-text screening. Eventually, total 14 cohort studies [14,15,16,17,18,19,20,21,22,23,24,25,26,27] were included in the meta-analysis, including 1072 patients of 447 LH and 625 RH (Fig. 1). Four studies [14, 21, 22, 25] originated from western centers (Germany, USA, and Italy) and ten studies [15,16,17,18,19,20, 23, 24, 26, 27] from eastern centers (Japan, Korea, and India) with all of them being single-center studies. The NOS score of all included studies were 7.0 ± 0.76 with low risk of bias and high quality of evidence. The basic characteristics of all studies were shown in Table 1.

PRISMA flow diagram of the search strategy for studies included in this meta-analysis

Thirteen cohort studies [14,15,16,17,18,19,20,21,22,23, 25,26,27] comprising 900 patients reported data of OS, and HR could be either directly extracted from 2 studies [21, 22] or calculated from K-M curve by using the method described above from another 11 studies [14,15,16,17,18,19,20, 23, 25,26,27]. The fixed-effects model was used and pooled HR reveled that there was no significant difference between LH and RH in OS (HR = 1.03, 95%CI 0.86–1.23, I2 = 30.3%, P = 0.73, Fig. 2A). By subsequent cumulative meta-analysis of publication year, the pooled HR was found to cross the invalidation line (HR = 1) at around year 2009 (Fig. 3). Therefore, a subgroup analysis using boundary year of publication as 2009 was conducted, and three studies published earlier than 2009 [25,26,27] showed a better OS in the RH group than in the LH group (HR = 3.08, 95%CI 1.43–6.66, I2 = 0.5%, P = 0.004), while the other 10 studies published after 2009 showed a comparative OS between LH and RH (HR = 0.97, 95% CI 0.81–1.17, I2 = 0%, P = 0.749, Table S1). The results concerning other subgroup analyses revealed no statistical difference for OS between LH and RH comparing eastern vs. western centers, and studies with higher number of cases (> 100 cases) vs. lower number of cases (< 100 cases) (Table S1). Data of DFS were also reported in 5 studies [14,15,16,17, 19] and HR could be extracted with a low heterogeneity (I2 = 0%), and no statistical difference was found between LH and RH groups (HR = 1.12, 95%CI 0.90–1.39, P = 0.31, Fig. 2B).

Forest plots for A OS and B DFS of patients with pCCA between LH and RH

Forest plot of A time cumulative meta-analysis and bubble plot of B meta-regression according to the year of publication

In stratified analyses, 6 studies [15,16,17, 19, 22, 25] reported 1-year survival rate, 9 studies [14,15,16,17,18,19, 21, 24, 25] reported 3-year survival rate, and 9 studies [14,15,16, 18, 19, 21, 22, 24, 25] reported 5-year survival rate. Pooled OR revealed that LH and RH exhibited comparable 1-year (OR = 1.01, 95%CI 0.68–1.50, I2 = 49.9%, P = 0.96), 3-year (OR = 0.86, 95%CI 0.65–1.14, I2 = 0%, P = 0.30), and 5-year survival (OR = 0.76, 95%CI 0.57–1.01, I2 = 0%, P = 0.06) in pCCA patients (Fig. S1). Similar results were observed in 1-year, 3-year, and 5-year DFS (1-year OR = 0.99, 95%CI 0.64–1.52, I2 = 0%, P = 0.95 [15,16,17, 19]; 3-year OR = 0.68, 95%CI 0.46–0.99, I2 = 27.3%, P = 0.05 [14,15,16,17, 19]; 5-year OR = 0.83, 95%CI 0.50–1.37, I2 = 0%, P = 0.46 [15, 16, 19], Fig. S2). Subgroup analysis showed a better 5-year survival rate of RH group in western centers than in eastern centers, and no statistic difference was observed in other subgroup analyses concerning year of publication and case numbers (Table S1).

Preoperative total bilirubin levels were reported in 8 studies [14,15,16,17, 19, 21, 24, 27] with no statistical difference between LH and RH groups, although a random-effects model was used (WMD =  − 0.38, 95%CI − 1.42–0.66, I2 = 90.63%, P = 0.47, Fig. S3A). Due to various approaches of biliary drainage were used among studies, data of preoperative biliary drainage could only be aggregated from 5 studies [15, 16, 19,20,21], and results did not reveal statistical difference between LH and RH groups (RR = 0.91, 95%CI 0.81–1.02, I2 = 0%, P = 0.10, Fig. S3B). PVE was reported in most studies [14,15,16, 18,19,20,21,22, 24], and data demonstrated that it was broadly performed to increase FLR volume in RH group (RR = 0.07, 95%CI 0.04–0.12, I2 = 0%, P < 0.01, Fig. S3C). Subgroup analysis did not change the final results.

A total of 12 studies [14,15,16,17,18,19,20,21,22, 24, 25, 27] involving 957 patients provided information on R0 resection rate of two different surgical methods. The pooled RR showed that R0 resection rate was comparable between LH and RH (RR = 0.95, 95%CI 0.88–1.02, I2 = 0%, P = 0.12, Fig.