Pancreatic Cancer Organoids from the Human Cancer Models Initiative Biobank Reflect Disease Genotypes, Capture Patient Heterogeneity, and are Amenable to Therapeutic Screening

Background

Pancreatic cancer is the fourth leading cause of cancer-related death in the United States (~3% of all cancer cases) and has one of the lowest 5-year relative survival rates. There is a lack of clinically representative, easily available, validated, in vitro pancreatic cancer models that reflect the genomic and phenotypic diversity of the disease. The Human Cancer Models Initiative (HCMI) is an international collaborative project devoted to the development and distribution of approximately 1,000 novel human primary tissue–derived tumor models supported with clinical and molecular annotation. These models were manufactured, characterized, and validated at a not-for-profit biobank and made available for academic and commercial use in basic research, compound screening, and drug development.

Methods

Available pancreatic cancer organoids (PCOs) from the HCMI biobank (n=46 unique models, n=41 unique donors) were acquired and subjected to a targeted next-generation sequencing hybridization capture, pan-cancer panel of 663 genes. Data were compared to publicly available clinical and genotyping results from the patient’s tumor (n=42) that the model was derived from as well as cancer datasets such as The Cancer Genome Atlas (TCGA). A subset of models was cultured following standard procedures and tested for response to chemotherapeutics (Cisplatin, Abraxane), KRAS-targeting inhibitors (MRTX1133, KR-pep-2D), or PARP inhibitors (PJ34, Olaparib) at three time points. A luminescent ATP viability assay was used to assess response.

Results

PCOs from the HCMI were predominantly derived from primary tumors (75%), had an adenocarcinoma ductal histological type, and were from female patients (56%) with an average age of 65 years at diagnosis. Models had mutations in key disease-relevant driver genes including KRAS, TP53, SMAD4, and/or CDKN2A at frequencies similar to the TGCA cohort. Sequencing revealed mutations in genes less commonly associated with pancreatic cancer, including BRCA, APC, BRAF, TGFBR2, and ARID1A, albeit at lower frequencies. Most models had mutations in KRAS including G12D, G12V, G12C, and G12R variants. Significant but not complete concordance was seen in comparison with the mutation status of the patient’s respective tumor, indicating potential genotypic drift had occurred. Response to acute treatment varied by model, genotype, timepoint, and drug. Chemotherapeutics like Cisplatin were broadly cytotoxic while MRTX1133 exhibited greater toxicity in models with KRAS G12D variants as compared with KRAS wild type.

Conclusion

Patient-derived cancer organoids from the HCMI recapitulate genotypes seen in patient populations and respond to therapeutic treatment in vitro.