Background: In this randomised phase III study (VEG105192; NCT00334282), pazopanib previously demonstrated statistically and clinically meaningful improvement of progression-free survival versus placebo in patients with advanced/metastatic renal cell carcinoma (mRCC). Final overall survival (OS) and updated safety results are now reported. Methods: Treatment-naive or cytokine-pretreated mRCC patients (n = 435) stratified and randomised (2:1) to pazopanib 800 mg daily or placebo, were treated until disease progression, death or unacceptable toxicity. Upon progression, placebo patients could receive pazopanib through an open-label study. Final OS in the intent-to-treat population was analysed using a stratified log-rank test. Rank-preserving structural failure time (RPSFT) and inverse probability of censoring weighted (IPCW) analyses were performed post-hoc to adjust for crossover. Findings: The difference in final OS between pazopanib- and placebo-treated patients was not statistically significant (22.9 versus 20.5 months, respectively; hazard ratio [HR] = 0.91; 95% confidence interval [CI], 0.71-1.16; one-sided P =.224). Early and frequent crossover from placebo to pazopanib and prolonged duration of crossover treatment confounded the OS analysis. In IPCW analyses, pazopanib decreased mortality (HR = 0.504; 95% CI, 0.315-0.762; two-sided P =.002). Similar, albeit non-significant, results were obtained in RPSFT analyses (HR = 0.43; 95% CI, 0.215-1.388; two-sided P =.172). Since the last cutoff, cumulative exposure to pazopanib increased by 30%. The pazopanib safety profile showed no new safety signals or changes in the type, frequency and severity of adverse events. Interpretation: Although no significant difference in OS was observed in this study, extensive crossover from placebo to pazopanib confounded final OS analysis. Post-hoc analyses adjusting for crossover suggest OS benefit with pazopanib treatment for mRCC patients.
- Inverse probability of censor weighting
- Overall survival
- Rank-preserving structural failure time model
- Renal cell carcinoma