P-28 Exploratory Analysis of Patients with Gastric/gej Adenocarcinoma with or Without Liver Metastasis from the Phase 3 RAINBOW Study

The prognosis of patients with advanced/metastatic GEJ adenocarcinoma (GEA) remains poor, and development of liver metastasis (LM) remains a major cause of high mortality. LM is one of the most common distant metastases in metastatic gastric cancer (mGC), and reported in nearly half of patients with mGC. This post-hoc analysis from the RAINBOW trial reports efficacy, safety, and biomarker outcomes from ramucirumab and paclitaxel treatment (RAM+PAC) in patients with(+) and without(-) LM at baseline. Patients (N=665) were randomised on a 1:1 basis to receive RAM+PAC (LM+: n=150, LM-: n=180) or placebo and paclitaxel (PL+PAC) (LM+: n=138; LM-: n=197). Overall survival (OS) and progression-free survival (PFS) were evaluated by stratified Kaplan-Meier (KM) and Cox regression models. Correlation of dichotomized biomarkers (VEGF-C, D; VEGFR-1,2) with efficacy in LM+ vs LM- subgroups were analyzed using Cox regression model and interaction p-values reported. Baseline demographics were balanced among treatment groups, with ECOG differences noted in LM- subgroup (ECOG-0: 34% vs. 46%; ECOG-1: 66% vs. 54% in RAM+PAC and PL+PAC, respectively). LM+ subgroup exhibited tumour baseline characteristics with higher incidences of metastatic disease, measureable, intestinal subtype, and ≥3 metastatic sites than LM-. LM+ subgroup treated with PL+PAC had a poorer OS (HR: 1.146) compared to LM-. In addition, the HR for PFS is significantly greater than 1 (HR: 1.680) at nominal level of 5% indicating LM is a prognostic factor. RAM+PAC could prolong OS and PFS in LM+/- groups. Median(m) OS for LM+ subgroup treated with RAM+PAC versus PL+PAC was 9.6 vs. 6.8 months (HR: 0.71, 95% CI:0.544-0.926), and LM- was 9.6 vs. 7.5 months (HR: 0.876, 95% CI:0.691-1.110). mPFS for LM+ was 4.6 vs. 2.8 months (HR: 0.466, 95% CI:0.359-0.604), and LM- was 4.3 vs. 3.6 months (HR: 0.762, 95% CI:0.607-0.958). Baseline Intertek assay results were analysed for a predictive relationship between biomarker level (VEGF-C, D; VEGFR-1,2) and efficacy outcome (OS, PFS) using interaction models in LM+ vs LM- subgroups. No significant interactions indicative of a predictive relationship were found. ORR for RAM+PAC treated LM+ subgroup was increased compared to LM- (38.0% vs. 19.4%). DCR was similar between subgroups (LM+: 79.3% vs. LM-: 80.6%). A benefit among patients treated with RAM+PAC versus PL+PAC for ORR and DCR were as follows, LM+: ORR 38.0% vs. 23.9%, DCR: 79.3% vs. 55.8% and LM-: ORR 19.4% vs. 10.7%, DCR: 80.6% vs. 69.0%. A higher incidence of grade ≥3 hypertension was observed in RAM+PAC versus PL+PAC subgroups (LM+: 18.4% vs. 2.2%; LM-: 11.7% vs. 3.1%). Treatment-emergent adverse events were generally well balanced between patients with and without LM. In this exploratory analysis, presence of LM was a negative prognostic factor, and RAM+PAC demonstrated benefits for patients with GEA irrespective of LM status, however, a more positive trend was observed in the LM+ subgroup. Biomarker data was limited, with insufficient evidence for a predictive relationship between any marker and treatment effect. No new safety signals were observed. These results indicate RAM+PAC is a viable therapeutic option for patients with GEA and LM.