Prognostic and predictive factors for overall survival (OS) in metastatic esophagogastric cancer (EGC): A meta-analysis.

Abstract

4048 Background: Prognostic and predictive factors for metastatic EGC are important to estimate prognosis, inform clinical decision-making and design future trials. We performed a systematic review with meta-analysis to identify these factors. Methods: We searched Medline, EMBASE and CENTRAL for phase 2/3 randomized controlled trials (RCTs) until January 2016 on palliative chemotherapy and targeted therapy for metastatic EGC. Prognostic and predictive factors were identified from respectively multivariate cox regressions and stratified treatment comparisons. Hazard Ratio’s (HR) for OS were extracted and pooled with meta-analysis if possible. Prognostic factors were considered independent if the multivariate HR was significant (P≤0.05). Predictive factors were clinically relevant if P for subgroup interaction was ≤0.20 and the HR in one of the subgroups was significant (P≤0.05). Results: We identified 47 RCTs (14,853 patients), wherein 54 potential prognostic and 40 predictive factors were reported. Eight independent prognostic factors for poor OS reported in ≥2 RCTs based on ≥300 patients were: performance status of ≥1 vs 0 (pooled HR, 95% confidence interval: 1.47, 1.25-1.73) or 2 vs 0-1 (1.52, 1.32-1.76); metastatic vs locally advanced disease (1.55, 1.39-1.72); diffuse vs intestinal/other histology (1.38, 1.12-1.71); ≥3 vs < 2 metastatic sites (1.35, 1.07-1.70); presence of metastases in peritoneum (1.24, 1.01-1.51) or liver (1.45 (1.28-1.64); measurable vs non-measurable disease (1.31, 1.04-1.66); and no prior vs prior surgery (1.33, 1.16-1.53). Predictive factors for specific treatment comparisons based on ≥300 patients were: age (≥65 vs < 65); performance status; tumor location (GEJ vs stomach); disease stage; number of metastatic sites; peritoneal metastasis; measurable disease; histology; HER2; KRAS; VEGF A; and Neuropilin-1 for first line treatments; and time to progression on first line therapy ( < 3, 3-6 or ≥6 months) for second-line treatments. Conclusions: Eight independent prognostic factors for OS and thirteen clinically relevant predictive factors for treatment efficacy of EGC were found.