Personalized cancer therapy prioritization based on driver alteration co-occurrence patterns

Area of Science:

• Oncology
• Genomics
• Pharmacology

Background:

• Precision oncology relies on identifying actionable genomic vulnerabilities.
• Patient-derived xenografts (PDXs) are valuable models for cancer research.

Purpose of the Study:

• To uncover connections between driver gene alterations and drug sensitivity.
• To develop predictive models for drug response in cancer patients.

Main Methods:

• Large-scale drug screening in patient-derived xenografts.
• Derivation of driver co-occurrence (DCO) networks.
• Development and validation of drug-response predictors.

Main Results:

• A collection of 53 drug-response predictors achieved 58% average balanced accuracy in cross-validation.
• High-confidence predictions reached 66% accuracy.
• 12 out of 14 predictions were experimentally validated in mice.

Conclusions:

• The strategy links oncogenic alterations to drug sensitivity, enhancing precision oncology.
• The approach can be adapted to utilize patient progression-free survival data for drug-response modeling.