Pembrolizumab versus the standard of care for cancer therapy: A meta-analysis of 12 KEYNOTE trials comparing overall survival.
e14159 Background: Pembrolizumab (P) is an antineoplastic agent approved for multiple different tumor origins as well as tumor agnostic based on MSI status. Thus understanding the tumor characteristics most predictive of response is essential. Living meta-analysis provides a method to continuously assimilate emerging trials. In this study, we created a living meta-analysis to compare the effect of P on overall survival (OS). Methods: Meta-analysis was conducted according to the PRISMA guidelines. PubMed and Cochrane databases, and conference abstracts (e.g. ASCO, ESMO), were searched for phase III KEYNOTE RCT’s that reported OS among cancer patients receiving P. Results: 12 phase III trials involving 7,319 patients (3,861 in P arms) treated for 6 different types of cancer were retrieved. Median follow up was 12.9 months [range: 0.1-35.5]. Treatment was P alone, SOC chemo, and P + chemo in 11 (40%), 13 (46%), and 4 (14%) arms, respectively. P significantly improved OS (HR 0.71, 95% CI: 0.59-0.85, p <0.001, I2 = 0%) when used in any cancer type, setting, or therapy for advanced refractory or chemo-naïve cancer patients. The mean median OS was 12.5 vs 11.1 months in the total populations for all P and control arms, respectively (Table). Conclusions: Among all trials, P was associated with improved OS. Additional meta-analysis will be performed with R software (version 3.3.3; R Fdn.). Random-effects models will be used to compare OS. Heterogeneity will be assessed with Cochrane Q -statistic and quantified with I2test via subgroup analyses for cancer type, setting, therapy, and PD-L1%. This meta-analysis is continuously updated at http://openmetaanalysis.github.io/checkpoint-inhibitors. [Table: see text]