Pan-Cancer Pharmacogenomic Analysis of Patient-Derived Tumor Cells Using Clinically Relevant Drug Exposures

Area of Science:

• Oncology
• Genomics
• Pharmacology

Background:

• Tumor heterogeneity and multiple molecular defects in solid cancers necessitate comprehensive evaluation of genetic and pharmacologic factors for effective precision medicine.
• Patient-derived xenograft (PDX) and patient-derived xenograft culture (PDXC) models closely mimic patient tumors, enabling detailed pharmacogenomic analysis.

Purpose of the Study:

• To evaluate the utility of PDX and PDXC models in predicting patient response to antitumor agents.
• To develop logic models for predicting drug response based on somatic mutations and drug sensitivity data.

Main Methods:

• Next-generation sequencing (NGS) was used to analyze somatic mutations in 23 matched patient tumor and PDX samples across four cancer types.
• 19 antitumor agents were screened against 78 patient-derived tumor cultures (PDXC) using clinically relevant drug exposures.
• A binarization threshold sensitivity classification in PDXC was used to predict drug response in vivo (PDX), and logic models were developed to correlate mutations with response.

Main Results:

• Concordance of somatic mutations between patient tumors and PDX models increased with variant allele frequency.
• Specific drug responses were identified in individual PDXC models and across cancer lineages.
• Drug responses observed in PDXC were successfully recapitulated in vivo in PDX models, and logic modeling identified key somatic mutations predictive of response.

Conclusions:

• Integrating NGS of patient tumors, high-throughput drug screening, and logic modeling provides a robust platform for understanding and predicting therapeutic drug response in cancer.
• PDX and PDXC models are valuable tools for comprehensive pharmacogenomic analysis and advancing precision oncology.