P48.05 Is Relapse-Free Survival at 2-Years an Appropriate Surrogate for Overall Survival at 5-Years in EGFR-mutated Resected NSCLC?

Abstract

The phase 3 ADAURA trial identified a highly significant hazard ratio (HR) of 0.17 for disease-free (i.e., relapse-free) survival (RFS) favoring adjuvant osimertinib, but a non-significant immature overall survival (OS) HR of 0.40. However, is RFS an appropriate surrogate for OS in this setting? OS is still the gold standard in health technology assessments, but it will take years for ADAURA’s OS data to mature. In the ADAURA-eligible setting, we compared HRs for clinico-demographic-treatment variables between RFS censored at 2 years (RFS-2YRS) with OS censored at 5 years (OS-5YRS), and explored reasons why such a relationship may exist. A retrospective analysis from Princess Margaret Cancer Centre of all patients with EGFR-mutated resected NSCLC (diagnosed 2012-2018) utilized Kaplan-Meier curves and Cox regression to compare RFS-2YRS and OS-5YRS for common clinico-demographic-treatment variables. We explored how these patients were managed after relapse, and compared their OS outcomes to patients with de novo advanced/metastatic EGFR-mutated disease. Among 104 patients with EGFR-mutated resected NSCLC, female=66%; median age=67 years; never-smokers=73%; Stage IB=41%; Stage II=34%; Stage IIIA=25%. Univariable Cox regression results comparing RFS-2YRS and OS-5YRS are presented in the table. Overall, HRs were either similar or slightly attenuated at OS-5YRs when compared to RFS-2YRS. Among the n=63 (61%) patients who relapsed, 81% were due to distant metastases. Only 8% were treated with curative surgery or chemoradiation at the time of relapse (an additional 3% were offered but declined curative treatment). The remaining patients (92%) were managed as advanced/incurable cancer. Of those treated at Princess Margaret, 75% received first-line EGFR-TKI therapy, while 4% received other systemic therapies; 8% had developed multifocal disease that was monitored rather than treated systemically; and 8% died before systemic therapy could be started or had concurrent comorbid conditions. Starting with diagnosis of advanced/metastatic disease, there were no differences in the OS of this relapsed cohort (n=63) versus 252 de novo advanced/metastatic NSCLC (log-rank p=0.29). The adjusted HR for OS was 0.82 (95% CI: 0.55-1.24 ). Among early-stage patients with EGFR-mutated resected NSCLC, 92% were treated as advanced/metastatic disease at the time of relapse. Their survival (from the time of relapse) was similar to those diagnosed initially with EGFR-mutated advanced/metastatic cancer. This information helps explain why relapse-free survival censored at 2-years represents an acceptable surrogate endpoint for overall survival at 5-years.