Overview
- Nature papers published Aug. 5 describe an international, decade-long Human Cancer Models Initiative that produced about 665–700 patient-derived models covering 25 cancer types and deposited the collection with ATCC for distribution.
- Analysis of 421 matched tumor–model pairs shows high fidelity between models and source tumors with 97.8% genetic concordance, 95% epigenetic concordance, and 92% transcriptional similarity.
- Teams adapted suspension culture and a compact CRISPR guide library to run large-scale loss-of-function screens across more than 162 organoids and to profile 300+ models, yielding roughly 1,700 links between gene dependencies and clinical or genomic features.
- Key findings include allele-specific dependencies such as differing KRAS G12 versus Q61 behavior, and the new data have been added to community resources like DepMap so researchers can search, download, and request models.
- Although formal HCMI activity is winding down, investigators and stewards plan continued model generation, wider use in drug-discovery screens, and public access so clinicians and scientists can study tumor evolution, therapy resistance, and patient-focused treatment hypotheses.