Anti-neoplastic drugs for malignant pleural effusion in non-small cell lung cancer: A meta-analysis.

e21060 Background: Malignant pleural effusion (MPE) occurs in 7-23% of advanced non-small cell lung cancer (NSCLC) patients with overall survival (OS) of 8.5 months only. The optimum anti-neoplastic regimen in MPE is yet to be determined. Thus, we conducted a meta-analysis to determine the optimum anti-neoplastic agent in managing MPE in NSCLC. Methods: We searched PubMed, Scopus, CENTRAL, ASCO, ESMO, and IASLC meetings libraries to retrieve relevant trials. We included single-arm and controlled trials in which patients with MPE were treated by anti-neoplastic drugs such as chemotherapy and/or targeted therapy. Patients treated with radiotherapy were excluded. Primary and secondary endpoints were efficacy and safety, respectively. Meta-analysis was conducted for the primary endpoint estimated by OS and objective response rate (ORR). In addition, subgroup analysis was conducted to investigate the effect of route of administration (ROA) and randomization on ORR. Results: We included 11 trials with a total of 542 patients. There were 5 randomized clinical trials, 5 phase II trials, and 1 phase I trial with a total of 16 relevant arms. Five arms used single-agent chemotherapy (SC), 11 combined therapy (CT) and 6 arms bevacizumab-based (BEVA) therapy. The ROA was intrapleural (IP) in 10 arms and intravenous (IV) in 6 arms. We pooled 124 patients from 5 trials that reported ORR results for SC patients. The pooled ORR was 41.7% (95% CI: 32.8 to 50.8%) with no heterogeneity ( I 2 = 0%, p = 0.47). Pooling of 396 patients who underwent CT in 10 studies results in an ORR of 64.1% (95% CI: 52.7 to 74.8%) with a high heterogeneity rate ( I 2 = 84%, p< 0.01). Analysis of BEVA studies yielded an ORR of 62.2% (95% CI: 43.5 to 79.2%), representing 133 patients from 5 trials. The table shows the effect of randomization and ROA on ORR. One-year OS was reported in 4 studies that used CT; pooled 1-year OS estimate for 216 patients was 81.2% (95% CI: 75.5 to 86.4%). As for grade 3/4 AEs, the occurrence of nausea/vomiting (N/V) was higher in CT than SC (14% vs 2%). Chest pain rate was similar (9.8% vs 7.0%), leukopenia was higher in BEVA than SC (19.1% vs 9.8%), and neutropenia was more prevalent in BEVA than CT (22.1% vs 8.1%). Incidence of any-grade N/V was more common in BEVA than SC (42.9% vs 19.3%). Conclusions: Both CT and BEVA-based regimens were effective and safe in treating MPE. IP was superior to IV in CC and BEVA based regimens. ORR was higher in randomized trials, suggesting a better response of MPE under optimum clinical conditions. Further trials of these agents are needed to confirm these results.[Table: see text]