O-007 Biomarker analyses of second-line ramucirumab in patients with advanced gastric cancer from RAINBOW, a global, randomized, double-blind, phase 3 study

Abstract

Introduction: Adding ramucirumab to paclitaxel in the RAINBOW trial led to significant improvements in overall survival (OS), progression‐free survival (PFS), and response rate when compared to placebo and paclitaxel (Wilke H, et al. Lancet Oncol. 2014;15(11):1224‐35). Plasma markers were evaluated for association with efficacy outcomes with the intention of identifying potential predictive or prognostic biomarkers. Methods: Plasma was collected from all patients in the RAINBOW trial pre‐treatment; prior to cycle 2, day 15 (fourth ramucirumab/placebo infusion); prior to cycle 4, day 1 (seventh ramucirumab/placebo infusion); and at the 30‐day follow‐up visit. Two sets of assays, platform 1 and platform 2 (measuring 24 markers and 5 markers, respectively), were used to assess circulating factors in plasma from different subsets of patients during the course of the trial. Key markers assessed included vascular endothelial growth factor (VEGF)‐C and D; sVEGFR 1, 2, and 3; and platelet‐derived growth factor. Patient data were dichotomized into low and high marker subgroups for each marker using the lower limit of quantitation as the cutpoint for those markers with >20% of samples below the limit of quantitation, and the median marker level as the cutpoint for others. Markers were analyzed for predictive effects using interaction models and for prognostic effects using main‐effects models. Results: Both biomarker platform populations appeared to be fairly representative of the intent‐to‐treat population with regard to demographics and efficacy outcomes. No strong predictive relationships were identified for markers with OS and PFS. However, several pharmacodynamic trends were observed in the ramucirumab treatment arm compared to placebo. Levels of VEGF‐D and placental growth factor increased and angiopoietin 2 decreased in the ramucirumab arm, following initiation of study treatment. However, levels of these markers changed minimally in the placebo arm. Potential prognostic markers for which low baseline levels corresponded to both longer OS and PFS across the two treatment arms, using an alpha of 0.05, were: C‐reactive protein, hepatocyte growth factor, intercellular adhesion molecule‐3, interleukin 8, serum amyloid A, and vascular cell adhesion molecule‐1. Conclusion: Despite multiple approved anticancer treatments that target angiogenesis, there are no known consistently predictive biomarkers to guide patient selection. The exploratory plasma analyses available from the RAINBOW study also do not identify a predictive biomarker for ramucirumab, but did reveal several pharmacodynamic trends and potential markers for prognostic effects.