Randomized proteomic stratified phase III study of second-line erlotinib (E) versus chemotherapy (CT) in patients with inoperable non-small cell lung cancer (PROSE).

Abstract

LBA8005 Background: Second-line therapy for advanced NSCLC patients (pts) after progression on platinum-based regimens typically employs CT or E. Improved PFS in E-treated pts is associated with EGFR sensitizing mutations. However, a test for optimizing treatment in pts with wild-type or unknown EGFR mutation status or squamous histology is of clinical value. VeriStrat (VS) is a serum protein test that assigns "good" (VSG) or "poor" (VSP) classification and has demonstrated prognostic and predictive utility. PROSE is a multicenter prospective randomized biomarker validation trial, designed to evaluate the ability of VS to predict survival in second-line NSCLC pts treated with E or CT. It is the first completed prospective randomized biomarker validation trial following the MARVEL design (Freidlin et al. JNCI. 2010). Methods: 285 pts, stratified by ECOG-PS, smoking, and blinded pretreatment VS classification, were randomized 1:1 to receive E or CT at standard doses. Primary endpoint was overall survival (OS) and the primary hypothesis was significant interaction between VS status and treatment. Sample size was calculated based on an estimated 65%/35% VSG:VSP ratio and hazard ratio (HR) for interaction of 2.35, with a 2-sided α=0.05 and 90% power. Results: 285 pts were randomized and 263 (129 CT, 134 E) included in the per protocol primary analysis. 68% of pts in CT arm and 72% in E arm were classified as VSG. Analysis was performed at 226 events. The trial reached its primary objective of significant interaction between treatment and VeriStrat classification with an interaction p value of 0.037. Pts in the VSP group performed worse on E compared to CT (HR: 1.72, 95% CI: 1.08-2.74); there was no significant difference in OS between treatments in the VSG group (HR: 1.09, 95% CI:0.79-1.50). 194/198 pts with histologic diagnosis had tissue available for EGFR and KRAS mutations. Conclusions: The results suggest that VS status is predictive of differential OS benefit for E versus CT in second line setting, complementing the result from a retrospective analysis of NCIC BR.21 where the prognostic behavior of VS was established (Carbone et al. JTO. 2012). Clinical trial information: NCT00989690.