Is there a harmonised standard to oncology trial data provision in data sharing initiatives?

Abstract

e13645 Background: Over the past decade the pharmaceutical industry has invested considerably in data sharing infrastructures to support independent researcher access to clinical trial individual-participant data (IPD), facilitating novel secondary analyses and the development of predictive models/simulation tools. Data sharing also helps in informing trial design and can accelerate the development of new therapies. To facilitate a data sharing ecosystem achieving these visions, researchers need to be able to access data, and when they do so, optimally, the data needs to be provided in a harmonised format that supports the research. Methods: In 2021 an audit confirmed 91 (of 203) clinical trials supporting medicines approved by the FDA in the last 10 years for the treatment of solid tumours as eligible (‘in scope’) for independent IPD request. This study reports on data provision for a planned clinical trial IPD meta-analysis summarising adverse events and therapeutic outcomes for new anticancer medicines according to race and sex. For the trials shared upon request, we report the heterogeneity in provisions of IPD and supporting documents. Results: Following SAP submission, the team obtained IPD packages from 76% (69 of 91) of requested trials – this includes data from 13 companies in a pooled cohort of over 40000 patients treated with new medicines against solid tumours. Time to data provision ranged from 117 to 312 days. In the IPD packages of 7 trials, overall survival or progression-free survival data were redacted, while for 9 trials there were partial redactions to adverse event data. For 3 trials, race data was removed, and for 1, sex was removed. Age was provided as a categorised variable in 17 trials. Weight was removed from 10 trials, and performance status was removed from 2. For 19 trials no clinical study report was provided, for 4 trials the data dictionary was missing, for 18 trials there were no data derivation specifications, while 5 IPD packages did not contain files orientating to anonymisation. Conclusions: While there is room to improve and standardize data sharing principles and procedures, overall, the acquired IPD was observed as a significant and high-quality resource for future scientific discovery. Future research could consider evaluating the ability to access data from non-industry trial sponsors, and strategies to harmonise data sharing practices between companies.