Literature search and selection
We initially identified 3935 potentially relevant studies and retained 3089 for screening after removing the duplicated studies. Among these, 3003 studies were excluded after the title and abstract screening, and full text of 86 studies was read. We subsequently excluded 81 studies including 3 studies [43–45] that explored the effects of concurrent chemoradiotherapy in patients with biliary tract cancer and 1 study [46] did not provide sufficient original data in the manuscript on patients with locally advanced. Eventually, 5 studies with 2412 patients were included in this meta-analysis. Of these, two studies were conducted in the USA [37, 47], two in France [48, 49], and one in the multicenter [50]. The PRISMA flow chart of the systematic literature search is presented in Fig. 1.
Characteristics And Methodological Quality Assessment Of The Eligible Studies
The baseline characteristics and types of all included studies are shown in Table 1. Among the 2412 patients included in the meta-analysis, 640 were under the neoadjuvant chemotherapy followed by surgery group, while 1772 were under the upfront surgery group. The characteristics and general information of all included studies are summarized in Table 1.
Table 1
Characteristics of the studies included in the systematic review and meta-analysis
Author | Year | Study center | Type of study | Inclusion period | Treatment (Chemotherapy regimens) | No. of patients | Age (Mean) | Male sex (%) | Tumor stage | Hospital stay (Days) | Follow-up (Months) | Outcomes |
Mason et al. [37] | 2021 | NCDB, USA | RS | 2006 to 2016 | Neoadjuvant chemotherapy followed by surgery (NA) | 516 | 58.9 | 246 (47.7) | I–III | NA | NA | ③④ |
Upfront surgery | 516 | 59.9 | 259 (50.2) | I–III | NA |
Sutton et al. [47] | 2021 | OHSU Knight Cancer Institute Liver Database, USA | RS | 2004 to 2017 | Neoadjuvant chemotherapy followed by surgery (Gem-Cis: 9; FOLFOX: 1) | 10 | 58 | 3 (30) | I–III (I = 3, II = 3, III = 4) | 8 (6–10) | 27 | ①②③⑤⑥ |
Upfront surgery | 42 | 65 | 23 (54.8) | I–III (I = 17, II = 14, III = 11) | 8 (6–10) |
Le Roy et al. [48] | 2018 | Paul Brousse Hospital, France | RS | 2000 to 2013 | Neoadjuvant chemotherapy followed by surgery (mainly GEMOX; Other: Gem-based, Fluorouracil, Oxaliplatin and Irinotecan) | 39 | 60 | NA | Locally advanced | 14 (12–21) | 93.3 | ①②③④⑤⑥⑦⑧ |
Upfront surgery | 82 | 65 | NA | Resectable | 12 (10–21) |
Riby et al. [49] | 2020 | Pontchaillou Hospital, France | RS | 1997 to 2017 | Neoadjuvant chemotherapy followed by surgery (Seven chemotherapy regimens, mainly GEMOX and Cis-based) | 13 | 60 | 8 (61.5) | Initially unresectable | NA | 85.5 | ①②③④⑤⑥⑦⑧ |
Upfront surgery | 137 | 67 | 100 (73) | Resectable | NA |
Buettner et al. [50] | 2017 | Multicentre | RS | 1990 to 2016 | Neoadjuvant chemotherapy followed by surgery (Intra-arterial chemotherapy: 18; systemic chemotherapy: 44) | 62 | 60 | 37 (59.7) | More advanced | 9 | 27.6 | ①②③④⑤⑥⑦⑧ |
Upfront surgery | 995 | NA | NA | NA | 12 |
Abbreviations: |
NCDB National Cancer Database, OHSU Oregon Health & Science University, Gem Gemcitabine, Cis Cisplatin, FOLFOX 5-Fluorouracil, Oxaliplatin, Folinic Acid, GEMOX Gemcitabine and Oxaliplatin, NA Not Available, RS Retrospective Study, NOS Newcastle-Ottawa Scale |
①: 1-year OS; ②: 3-year OS; ③: 5-year OS; ④: R0 resection; ⑤: 1-year RFS; ⑥: 3-year RFS; ⑦: 5-year RFS; ⑧: Mortality in 90 days |
Two studies [37, 50] were classified as having a serious risk of bias, and the other three studies [47–49] were classified as having a moderate risk of bias. All of the above assessments were based on 5-year OS, the assessments based on all outcomes were shown in Table 2, and the quality of evidence of the included studies was shown in Table 3.
Table 2
Quality assessment of included studies according to ROBINS-I
Reference | Outcomes | Bias due to confounding | Bias in selection of participants into the study | Bias in classification of Interventions | Bias due to deviations from intended interventions | Bias due to missing data | Bias in measurement of outcomes | Bias in selection of the reported results | Overall risk of bias |
Mason et al. [37] | 5-year OS | Moderate | Moderate | Low | Serious | Moderate | Low | No information | Serious |
R0 resection | Moderate | Moderate | Low | Serious | Moderate | Moderate | No information | Serious |
Sutton et al. [47] | 1-year OS | Moderate | Moderate | Low | Low | Moderate | Low | No information | Moderate |
3-year OS | Moderate | Moderate | Low | Low | Moderate | Low | No information | Moderate |
5-year OS | Moderate | Moderate | Low | Low | Moderate | Low | No information | Moderate |
1-year RFS | Moderate | Moderate | Low | Low | Moderate | Low | No information | Moderate |
3-year RFS | Moderate | Moderate | Low | Low | Moderate | Low | No information | Moderate |
Le Roy et al. [48] | 1-year OS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
3-year OS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
5-year OS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
R0 resection | Moderate | Low | Low | Low | Low | Moderate | No information | Moderate |
1-year RFS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
3-year RFS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
5-year RFS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
Mortality in 90 days | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
Riby et al. [49] | 1-year OS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
3-year OS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
5-year OS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
R0 resection | Moderate | Low | Low | Low | Low | Moderate | No information | Moderate |
1-year RFS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
3-year RFS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
5-year RFS | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
Mortality in 90 days | Moderate | Low | Low | Low | Low | Low | No information | Moderate |
Buettner et al. [50] | 1-year OS | Serious | Moderate | Low | Low | Moderate | Low | No information | Serious |
3-year OS | Serious | Moderate | Low | Low | Moderate | Low | No information | Serious |
5-year OS | Serious | Moderate | Low | Low | Moderate | Low | No information | Serious |
R0 resection | Serious | Moderate | Low | Low | Moderate | Moderate | No information | Serious |
1-year RFS | Serious | Moderate | Low | Low | Moderate | Low | No information | Serious |
3-year RFS | Serious | Moderate | Low | Low | Moderate | Low | No information | Serious |
5-year RFS | Serious | Moderate | Low | Low | Moderate | Low | No information | Serious |
Mortality in 90 days | Serious | Moderate | Low | Low | Moderate | Low | No information | Serious |
Table 3
Assessment of certainty of evidence according to GRADE
Certainty assessment | No of patients | OR (95% CI) | Certainty |
Outcomes | Risk of bias | Inconsistency | Indirectness | Imprecision | Publication bias | NC | US |
1-year OS | Serious a | Not serious | Not serious | Not serious | Not assessed c | 97/124 | 994/1256 | 0.94 (0.59–1.50) | ⨁⨁⨁◯ |
3-year OS | Serious a | Not serious | Not serious | Not serious | Not assessed c | 63/124 | 623/1256 | 1.17 (0.80–1.72) | ⨁⨁⨁◯ |
5-year OS | Serious a | Not serious | Not serious | Not serious | Not assessed c | 234/640 | 616/1772 | 1.27(1.02–1.58) | ⨁⨁⨁◯ |
R0 resection | Serious a | Serious b | Not serious | Not serious | Not assessed c | 432/630 | 1381/1730 | 0.49 (0.26–0.91) | ⨁⨁◯◯ |
1-year RFS | Serious a | Not serious | Not serious | Not serious | Not assessed c | 71/124 | 781/1256 | 0.95 (0.64–1.40) | ⨁⨁⨁◯ |
3-year RFS | Serious a | Not serious | Not serious | Serious c | Not assessed c | 47/124 | 539/1256 | 1.02 (0.68–1.52) | ⨁⨁◯◯ |
5-year RFS | Serious a | Not serious | Not serious | Serious c | Not assessed c | 34/114 | 476/1214 | 0.89 (0.57–1.39) | ⨁⨁◯◯ |
Mortality in 90 days | Serious a | Not serious | Not serious | Serious c | Not assessed c | 5/114 | 50/1214 | 0.76 (0.28–2.01) | ⨁⨁◯◯ |
CI confidence interval; NC Neoadjuvant chemotherapy followed by surgery; US Upfront surgery |
a High risk of bias |
b High heterogeneity |
c the sample size was smaller |
Meta-analysis
One-year OS
Four studies [47–50] reported 1-year OS of the two procedures, and a fixed effects model was used for the analysis because no significant heterogeneity was observed between the studies (χ2 = 1.66, P = 0.56, I2 = 0). We found no significant differences between the two groups (OR = 0.94, 95% CI: 0.59–1.50) (Fig. 2).
Three-year Os
Similarly, four studies [47–50] reported 3-year OS of the two procedures, and a fixed effects model was used for the analysis (χ2 = 2.79, P = 0.43, I2 = 0), which revealed that there was no significant difference between the two groups (OR = 1.17, 95% CI: 0.80–1.72) (Fig. 3).
Five-year Os
All studies reported 5-year OS of the two procedures, and the meta-analysis revealed a statistically significant difference between the two groups (OR = 1.27, 95% CI: 1.02–1.58), with a moderate heterogeneity (χ2 = 7.54, P = 0.11, I2 = 47%) (Fig. 4).
R0 Resection
Four studies [37, 48–50] reported R0 resection of the two procedures. We used a random effects model for the analysis because a high heterogeneity was observed between the studies (χ2 = 11.99, P = 0.007, I2 = 75%). The meta-analysis also showed a significant difference between the two groups (OR = 0.49, 95% CI: 0.26–0.91) (Table 4).
Table 4
Meta-analysis results of the secondary outcomes
Outcomes | No. of Studies | Assessment of Heterogeneity | No. of patients | Meta-analysis Results |
I2 (%) | P | NC | US | OR ( 95% CI) | P |
R0 resection | 4 [37, 48–50] | 75 | 0.007 | 630 | 1381 | 0.49 (0.26–0.91) | 0.02 |
One-year RFS | 4 [47–50] | 0 | 0.41 | 124 | 1256 | 0.95 (0.64–1.40) | 0.79 |
Three-year RFS | 4 [47–50] | 0 | 0.72 | 124 | 1256 | 1.02 (0.68–1.52) | 0.94 |
Five-year RFS | 3 [48–50] | 0 | 0.70 | 114 | 1214 | 0.89 (0.57–1.39) | 0.60 |
Mortality in 90 days | 3 [48–50] | 0 | 0.76 | 114 | 1214 | 0.76(0.28–2.01) | 0.58 |
NC Neoadjuvant chemotherapy followed by surgery; US Upfront surgery |
One-year Rfs
One-year RFS of the two procedures was reported by four studies [47–50]. The meta-analysis revealed no significant heterogeneity (χ2 = 2.88, P = 0.41, I2 = 0%) and difference (OR = 0.95, 95% CI: 0.64–1.40) between the two groups (Table 4).
Three-year Rfs
Three-year RFS of the two procedures was also reported by four studies [47–50], and no significant difference between the two groups (OR = 1.02, 95% CI: 0.68–1.52) and heterogeneity (χ2 = 1.32, P = 0.72, I2 = 0%) between the studies was observed (Table 4).
Five-year Rfs
Three studies [48–50] reported 5-year RFS of the two procedures, and our meta-analysis revealed no statistically significant difference between the two groups (OR = 0.89, 95% CI: 0.57–1.39) and heterogeneity between the studies (χ2 = 0.71, P = 0.70, I2 = 0) (Table 4).
Mortality In 90 Days
The 90 days mortality of the two procedures was reported by three studies [48–50]. The meta-analysis revealed no statistically significant difference between the two groups (OR = 0.76, 95% CI: 0.28–2.01) and heterogeneity among the studies (χ2 = 0.54, P = 0.76, I2 = 0) (Table 4).
Sensitivity Analysis
Sensitivity analysis of the primary outcomes (1-year, 3-year, and 5-year OS, and R0 resection) and secondary outcomes (1-year, 3-year, and 5-year RFS, and mortality in 90 days) was performed by removing one study at a time from the meta-analysis using the Review Manager 5.3 and testing their heterogeneity differences. The results indicated that, for R0 resection, the heterogeneity reduced significantly when the study by Mason et al. [37] was excluded (χ2 = 0.18, P = 0.92, I2 = 0); however, the recalculated results were consistent with those obtained when all studies were included. No significant change was found in the overall statistical significance of the model.