202P Systematic review and meta-analysis of recommended 2nd-line therapies for advanced gastric cancer (GC)

Abstract

Aim/Background:With several approaches to treatment of patients with GC after 1st-line therapy, a meta-analysis was conducted to compare outcomes among these options. Methods: Electronic databases and relevant congress abstracts were systematically searched. Inclusion criteria included English language, randomized controlled trials (RCTs) of adult patients who received prior chemotherapy for GC or gastroesophageal junction cancer, published through 28 May 2014. Interventions were limited to those in NCCN or ESMO guidelines. For all outcomes, Bucher indirect comparisons were conducted. A fixed-effect Bayesian network meta-analysis was conducted for overall survival (OS). Risk of bias and heterogeneity were evaluated. Results: Ten RCTs met eligibility criteria and evaluated best supportive care (BSC), docetaxel, FOLFIRI, irinotecan (IRI), IRI + cisplatin, paclitaxel (PAC), ramucirumab (RAM) and RAM + PAC. Two studies were excluded from the base-case analysis of OS due to design or reporting limitations but were included in a sensitivity analysis. 2 studies were excluded from the analysis of discontinuations (DCs) due to adverse events (AEs) due to design or lack of data. Table 1 reports results for these outcomes relative to BSC. Results were consistent across all sensitivity analyses. Sources of heterogeneity included geographic participation, primary tumor site and dosing regimen; however, a formal test of heterogeneity was not feasible as single studies were used for almost all treatment comparisons. Conclusions: Compared to BSC, OS was significantly improved for all interventions and DCs due to AEs were significantly higher for all interventions except RAM. Results should be interpreted with caution due to a limited number of studies, wide confidence intervals and between-study heterogeneity. Odds ratios could not be reliably estimated due to data scarcity and should be interpreted relative to each other, not in terms of absolute values. (Table presented).