Multivariate analysis including biomarkers in the phase III RADIANT-2 study of octreotide LAR plus everolimus (E+O) or placebo (P+O) among patients with advanced neuroendocrine tumors (NET).

Abstract

157Background: In this large phase III trial, median progression-free survival (PFS) improved by 5.1 mo with E+O compared to P+O in patients (pts) with NET associated with carcinoid syndrome. Randomization imbalances including WHO performance status (PS), and primary site favoring P+O confounded primary analysis. Chromogranin A (CgA) and 5-hydroxyindoleacetic acid (5-HIAA) are important biomarkers in NET. Analyses were performed to identify prognostic factors and adjust for randomization imbalances.Methods: Pts were randomized to E+O (n=216) or P+O (n=213). Potential prognostic factors including baseline CgA (≤2×ULN vs >2×ULN), baseline 5-HIAA (≤median vs >median), age (<65 vs ≥65), gender, race, WHO PS (0 vs 1, 2), primary site (lung vs other), prior somatostatin analog use (yes vs no), duration from diagnosis (<6 mo, 6-24 mo, 2-5 yr, >5 yr), and organs involved (liver, bone) were assessed in univariate analysis using the log rank test and a stepwise regression using Cox proportional hazards model.Results: Randomization resulted in significant imbalance in baseline CgA (median [ng/mL], 251 E+O vs 137 P+O). Median PFS (mo) was significantly longer for pts with nonelevated CgA (27 vs 11; P