Fifty-one RCTs (n=9,970) met the inclusion criteria, and of these 33 RCTs(n=6,026) provided sufficient data to be included in the meta-analyses.
Compared with best supportive care, chemotherapy significantly improved overall survival (HR 0.64, 95% CI: 0.42, 0.98; 6 RCTs), but there was evidence of significant heterogeneity. There were no significant differences in survival between FU-based combination therapy and FU-single agent chemotherapy (HR 0.94, 95% CI: 0.82, 1.08; 5 RCTs), or between gemcitabine and FU (HR 0.75, 95% CI: 0.42, 1.31; 2 RCTs), although there was significant evidence of heterogeneity for the latter result. Gemcitabine-based combination therapy had significantly improved survival in comparison with gemcitabine alone (HR 0.91, 95% CI: 0.85, 0.97; 14 RCTs).
Subgroup analyses suggested that platinum-based therapies (HR 0.85, 95% CI: 0.74, 0.96; 3 RCTs) and capecitabine (HR 0.83, 95% CI: 0.72, 0.96; 3 RCTs) in combination with gemcitabine reported significantly better survival than single-therapy gemcitabine. Irinotecan-based therapies (HR 1.01, 95% CI: 0.84, 1.22; 2 RCTs) and FU-based combination therapies were not significantly better than single-therapy gemcitabine.
There was evidence of publication bias for all comparisons.