Seven trials (2,107 participants) were included. Three trials had adequate methods of randomisation. None of the trials reported use of allocation concealment or blinding. All seven trials reported losses to follow-up. Six trials used intention-to-treat analyses.
Survival: OXA-based chemotherapy improved overall survival compared to the IRI-based regimen (HR 1.28 for IRI versus OXA, 95% CI 1.13 to 1.45; five trials with p=0.064 for heterogeneity). There was no evidence of publication bias. Six trials reported median survival, which ranged from 14 to 17.6 months for IRI-based chemotherapy compared to 13.7 to 19.5 months for OXA-based chemotherapy. Survival tended to be longer for OXA-based chemotherapy and this was statistically significant in three trials.
Progression-free survival: Six trials reported time to progression and this was longer for OXA-based chemotherapy in four trials. Time to progression ranged from 5.5 to 8.9 months for IRI-based and 7.0 to 9.7 months to OXA-based chemotherapy. The difference was statistically significant in two trials.
Toxicity: Combined results of seven trials showed that nausea/vomiting/emesis and diarrhoea were significantly more likely with IRI-based chemotherapy and paraesthesia, sensory neuropathy and thrombocytopenia were significantly more likely with the OXA-based regimen. Full numerical results for 14 types of toxicity were reported in the paper.