Abstract
Cell line experiments arguably remain the most used tool in preclinical cancer research. Given that PCa cell lines form a major part of this preclinical framework, evaluating their relevance is crucial to improving the predictive power of early-stage research. A range of in silico, in vitro, in vivo and ex vivo approaches are gaining popularity. However, the plethora of historical data from cell lines cannot be ignored, nor can their future use. This study integrated clinical and multiomics data to assess the relevance of 46 PCa cell lines. Data from over 10,000 patient cases across 24 studies in nine countries were harmonized using cBioPortal, COSMIC, CCLE, and peer-reviewed literature. Multiomics analysis covered mutations, copy number alterations, structural variants, microsatellite instability, mRNA, and protein expression. Statistical comparisons normalized and cross-validated data using bioinformatics tools. Clinical variables, such as age at sample collection, geographical location, ethnicity, race, cancer type, Gleason grade, and metastatic stage, were included to evaluate the population relevance of cell line representation. This study found that the most used cell lines accurately represent a small subset of patients. A data-driven pipeline was proposed to guide selection of representative cell lines so more preclinical research can be translated into practice.