Abstract

Cell lines are indispensable tools in prostate cancer research yet their suitability as models for distance metastasis remains incompletely characterized. Here, we conduct a systematic evaluation study using large-scale public multi-omics data. We reveal substantial genomic differences between cell lines and metastatic patient samples, and meanwhile pinpoint cell lines which more closely resemble metastatic prostate cancer. Notably, hypermutation significantly influences the tumor microenvironment, underscoring the importance of considering mutational burden in model selection. Surprisingly, the widely used PC3 cell line exhibits poor transcriptomic and epigenomic similarity to any prostate cancer subtype, revealing a previously unrecognized limitation. Furthermore, we find existing engineered stem-like cell lines fail to faithfully recapitulate the transcriptomic profiles of mesenchymal stem-like prostate cancer, whereas selected organoids exhibit higher fidelity. Our study provides guidance for cell line selection in metastatic prostate cancer research and highlights the urgent need to develop improved cell lines for the mesenchymal stem-like subtype. Systematic multi-omics evaluation reveals substantial differences between prostate cancer cell lines and metastatic patient samples, and identifies models that more closely resemble metastatic disease.