OBJECTIVES: The latest technology appraisal guidance released by National Institute for Health and Care Excellence (NICE) for treatment strategies in malignant pleural mesothelioma (MPM) has highlighted the need for trials comparing alternative chemotherapy regimens beyond the current recommended combination of Cisplatin and Pemetrexed. This study aims to apply methods to compare treatment effectiveness using synthetic real-world data (RWD) from Systemic Anti-Cancer Therapy dataset (Simulacrum) and combine RWD with existing trial data to draw meaningful inferences.

METHODS: Synthetic patient data were extracted from Simulacrum and treatments of interest were identified following clinician guidance. Trials were emulated using the “target trial protocol” approach described by Hernan & Robins et al (2016). Survival estimates were obtained using a Kaplan-Meier estimator and time-to-event analysis using Cox proportional hazards models adjusted for prognostic factors. Results from emulated trials were combined with existing trials and single arm studies in a network meta-analysis framework. Relevant single arm studies were included in the analysis using the “distance metric method” proposed by Schmitz et al (2018).

RESULTS: Four treatments of interest were identified from Simulacrum and 3 trials were emulated. Patients receiving Vinorelbine and Gemcarbo had better median survival as compared to Cisplatin+Pemetrexed (15.3 vs 10.9 months; 18.9 vs 12.2 months). 5 existing trials were identified through literature and information was combined using a hierarchical network model to account for differences in study design. The model was further extended to include 6 single arm studies as prior information to obtain more stable estimates by varying the degree of RWD inclusion.

CONCLUSIONS: This study evaluated first-line treatments for MPM with promising results for Vinorelbine and Gemcarbo based regimens using target trial emulation. This study also demonstrated how integration of multiple sources of evidence using existing Bayesian methods can be used to increase the precision of relative treatment effects and derive more generalizable inferences.