Eighteen RCTs involving a total of 2,933 participants were included.
Most of the studies did not report details of the randomisation method; it was assumed nevertheless that allocation concealment was not compromised.
A meta-analysis of 10 RCTs (n=1,209) showed a statistically significant reduction in all-cause mortality with neoadjuvant chemoradiotherapy compared with surgery alone (HR 0.81, 95% CI: 0.70, 0.93, p=0.002). There was no statistically significant heterogeneity between the trials. Th exclusion of two unpublished studies, and other sensitivity analyses, did not change the findings. The absolute difference in survival at 2 years was 13%, based on data from 8 RCTs. The number-needed-to-treat (NNT) to prevent one death was 8. The pooled effect estimates for SCC and adenocarcinoma were similar to the overall results.
A meta-analysis of 8 RCTs (n=1,724) showed a reduction in all-cause mortality with neoadjuvant chemotherapy compared with surgery alone that just reached statistical significance (HR 0.90, 95% CI: 0.81, 1.00, p=0.05). There was no statistically significant heterogeneity between the trials. The absolute difference in survival at 2 years was 7%. The NNT was 15. The pooled effect estimate for SCC showed no significant difference between neoadjuvant chemotherapy and surgery alone, but the one RCT restricted to adenocarcinoma showed a statistically significant benefit (HR 0.78, 95% CI: 0.64, 0.95, p=0.014).
Additional analyses were presented in the report.
Funnel plot evidence for publication bias was inconclusive. The number of unpublished studies was estimated to be nine for neoadjuvant chemoradiotherapy and none for neoadjuvant chemotherapy.