Final results of the randomized phase III trial of sorafenib in advanced renal cell carcinoma: Survival and biomarker analysis

5023 Background: Based on the significant PFS benefit of sorafenib (SOR) vs placebo (P) in a Phase III advanced RCC trial, P patients were unblinded and crossed over to SOR in May 2005. Final OS and biomarker data are reported. Methods: Final OS analysis was planned at ∼540 events (a=0.037 after adjusting for previous analyses). To minimize effect of crossover on OS, a secondary analysis was planned censoring P data on June 30, 2005 (a=0.037). Plasma VEGF and sVEGFR2 were measured by ELISA at baseline (BL), cycle (C) 1 day (D) 21, and C3D1. pERK was assayed by IHC. Results: 903 patients were randomized (SOR, 451; P, 452). The only OS analysis before crossover (May 2005) showed an estimated 39% OS improvement for SOR vs P (HR=0.72; p=0.018) (ECCO 2005); 216 P patients had crossed to SOR. OS analysis 6 months after crossover (Nov 2005) showed a 30% improvement in OS for SOR vs P (HR=0.77, p=0.015) (ASCO 2006). These OS differences did not reach prespecified O’Brien-Fleming statistical boundaries. Final OS (Sep 2006) at 561 deaths showed an improvement of 13.5% for SOR vs P and was not significant (median 17.8 vs 15.2 months; HR=0.88; p=0.146; a=0.037). Secondary analysis censoring P data (June 2005) showed a significant OS benefit for SOR vs P (HR=0.78, 95% CI: 0.62, 0.97; p=0.0287; a=0.037), suggesting crossover had confounded OS. Changes in VEGF (n=712) and sVEGFR2 (n=717) were observed after SOR treatment (AACR 2006); VEGF increased 32% (n=279) at C1D21 and 47% (n=203) at C3D1, and sVEGFR2 decreased 18% (n=282) and 24% (n=206). Using a COX proportional hazards model, BL VEGF was an independent prognostic factor (p=0.014); patients with high BL VEGF (>131 pg/ml) had poorer prognosis and a trend towards greater PFS benefit with SOR (SOR vs P, HR=0.48 vs 0.64 for high vs low VEGF, p=0.096). BL sVEGFR2, changes in VEGF or sVEGFR2 at C1D21, and pERK levels in limited diagnostic tumor biopsies were not predictive of SOR response. Conclusion: SOR demonstrated a PFS benefit in advanced RCC, although ITT final OS analysis showed a confounding effect of crossover. Significant OS benefit of SOR was seen in a planned secondary analysis adjusting for crossover. VEGF levels have prognostic importance, and SOR-associated changes in VEGF and sVEGFR2 are consistent with inhibition of VEGF signaling. No significant financial relationships to disclose.