Observational study examining the feasibility of generating real-world evidence (RWE) for new drug applications from the clinical trials database (DB; RELIASE study).

TPS100 Background: The extraction of the data that contributes to regulatory approval from real-world data (RWD) are difficult due to the lack of standardized data format as well as the extraction methodology. In addition, when RWE is used as an external control group, the similarity between internal and external control data has not been evaluated. Methods: In our ‘REALISE’ study, we will investigate the data extraction methodology as external control data of rare molecular subtype. We also elucidate the 'quality' and 'reliability' of RWD/RWE necessary for regulatory submissions. In addition, as most databases are not designed for regulatory use in the creation phase, we will investigate the retrospective methodologies to ensure reliability of RWD/RWE. We compare ‘data quality’ and ‘data reliability’ of ARCAD global DB (N=45,224), SCRUM-Japan registry (N=546), SCRUM-Japan observational study (N=14,325), and the Flatiron health RWD in Japan (N=650) and statistically analyze the differences and similarities among 4 databases. All analyses will be summarized descriptively. Similarity of patient characteristics and efficacy endpoints such as response rate, progression-free survival, overall survival, and all available endpoints will be evaluated by cancer type, treatment line, and standard treatment. For colorectal cancer, data extracted from the SCRUM-Japan registry, SCRUM-Japan observational study, and the Flatiron health RWD will be compared with data extracted from randomized controlled trials in the ARCAD global DB to examine the availability as an external control of RWD. The Kaplan-Meier method will be applied to estimate the survival distribution of time-to-event data. No imputation method for missing data will be used, and the frequency of missing data will also be summarized. We will also examine the methodology for extracting data with sufficient quality from the SCRUM-Japan observational study. In addition, if the reliability of RWD/RWE does not reach the required level, we will also examine the methodologies to retrospectively assure the reliability of the SCRUM-Japan observational study for regulatory submissions. Table: see text