Forty-five randomised controlled trials (RCTs) were included in the meta-analysis (n=11,867 patients). Twenty-five were phase III and 20 were phase II trials. Trials excluded patients for various reasons including no treatment or ineligible (12 RCTs), discontinued treatment early (two RCTs), and missing data or insufficient follow-up (two RCTs). Eight trials did not report the reasons for exclusion.
Response rates: No differences were observed in the response rates between regimens containing gemcitabine (32 RCTs, n=8,136), vinorelbine (32 RCTs, n=7,333), docetaxel (16 RCTs, n=4,935), or paclitaxel (24 RCTs, n=6,826) and regimens that did not contain these drugs. No significant between-trial heterogeneity was observed.
Progressive disease: There was a significantly lower risk of disease progression for patients receiving gemcitabine compared with those receiving regimens that did not contain gemcitabine (OR 0.86, 95% CI 0.77 to 0.95; 23 RCTs, n=6,681). There was a significantly increased risk of disease progression for patients receiving paclitaxel compared with those receiving regimens not containing paclitaxel (OR 1.22, 95% CI 1.09 to 1.37; 16 RCTs, n=5,536). There were no significant associations between disease progression and the use of docetaxel (12 RCTs, n=4,642) and vinorelbine regimens (23 RCTs, n=6,048). No significant between-trial heterogeneity was observed.