Eleven trials were eligible, but the patient data could not be obtained for one. Therefore, 10 trials (n=2,492) were included in the meta-analysis.
Overall survival (all 10 trials, n=2,492): the confidence intervals (CI) of the effect sizes for the individual studies were wide and the results of individual trials were inconclusive. There was no statistical heterogeneity and the pooled HR was 0.91 (95% CI: 0.83, 1.01) (i.e. no statistically significant benefit).
Overall survival with single-agent chemotherapy (i.e. cisplatin) (3 trials, n=376): the HR was 1.15 (95% CI: 0.90, 1.47), indicating no benefit of neoadjuvant chemotherapy.
Overall survival with combination neoadjuvant chemotherapy (6 trials, n=2,116): the HR was 0.87 (95% CI: 0.78, 0.97), indicating a statistically significant reduction in risk of death. A sensitivity analysis that excluded the largest trial found this result to be robust and not driven by that one trial. The difference between the results for combination chemotherapy and single-agent chemotherapy was statistically significant.
A further analysis, which included the summary data of the one extra trial for which individual patient data could not be obtained, did not alter the findings of the main meta-analysis but did narrow the overall CI.
Findings for the secondary outcomes closely reflected those for overall survival.