Overall Survival Analyses of First-Line Erlotinib Versus Chemotherapy in the Eurtac Study Population Controlling for the Use of Post-Study Therapy

Abstract

ABSTRACT Aim: Overall survival (OS) as an endpoint in trials of first-line therapy is challenging due to the effect of post-study therapy (PST). OS benefit may not be observed despite a clear benefit in progression-free survival. This exploratory analysis of the first-line EURTAC study assesses the effect of erlotinib on OS, using two statistical models to control for second-line (2L) PST use. Methods: Patients with EGFR mutation-positive non-small-cell lung cancer were randomised to receive once-daily erlotinib (E) or 4–6 cycles of chemotherapy (C). OS, calculated from date of randomisation to death, was assessed by Kaplan–Meier methodology. In this analysis, two statistical methods controlled for 2L PST use; method 1: the original Cox proportional hazard model was used, with OS censored at the time of 2L PST initiation, to evaluate OS only when patients were not receiving 2L PST; method 2: a Cox model adjusting for 2L PST exposure in the OS analyses where ‘2L PST exposure’ was incorporated as a time-dependent covariate (allowing for differential effects for E and C). Results: Baseline characteristics were balanced between the arms (E, n = 86; C, n = 88). In the protocol-specified analysis (data cut-off 11 April 2012), the hazard ratio (HR) for OS was 0.92 (95% CI 0.63–1.35, p = 0.68; median OS 22.9 vs 19.6 months for E vs C). When adjusted for baseline factors (age, gender, ECOG PS, smoking history and EGFR mutation type), the HR was 0.86 (95% CI 0.58–1.27). A total of 58% of patients received 2L PST (tyrosine-kinase inhibitors: 31% E, 58% C; platinum compounds: 44% E, 5% C; antimetabolites: 48% E, 19% C). Median time to initiation of 2L PST from time of disease progression (n = 102) was 0.85 months for E and 0.36 months for C. Using method 1, the HR was 0.68 (95% CI 0.37–1.25, p = 0.21; median OS was not reached with E vs 20.8 months with C). The adjusted HR was 0.63 (95% CI 0.34–1.19). Using method 2, the HR was 0.69 (95% CI 0.38–1.26), with an adjusted HR of 0.65 (95% CI 0.35–1.20). Conclusions: These exploratory results suggest a trend towards a survival benefit for first-line E vs C, which was previously unobserved due to the confounding factor of non-randomised 2L PST use. Disclosure: L.F. Leon: Stock ownership: Yes, Genentech/Roche, Employee Stock Options Advisory board: No Board of directors: No Corporate sponsored research: Yes, Genentech/Roche, Employee Other substantive relationships: Yes, Genentech/Roche, Employee; A. Golsorkhi: Stock ownership: Yes, Genentech/Roche, Employee Stock Options Advisory board: No Board of directors: No Corporate sponsored research: Yes, Genentech/Roche, Employee Other substantive relationships: Yes, Genentech/Roche, Employee; S. Liu: Stock ownership: Yes, Genentech/Roche, Employee Stock Options Advisory board: No Board of directors: No Corporate sponsored research: Yes, Genentech/Roche, Employee Other substantive relationships: Yes, Genentech/Roche, Employee; A. Drozdowskyj: Stock ownership: No Advisory board: No Board of directors: No Corporate sponsored research: No Other substantive relationships: Yes, Statistical Consultant for Roche. All other authors have declared no conflicts of interest.