Predicting cancer drug TARGETS - TreAtment Response Generalized Elastic-neT Signatures

Area of Science:

• Oncology
• Genomics
• Pharmacogenomics

Background:

• Molecular medicine enables targeted cancer therapies based on specific DNA alterations.
• Currently, few clinically validated predictive biomarkers exist in oncology.
• Developing novel predictive biomarkers is crucial for advancing personalized cancer care.

Purpose of the Study:

• To develop and validate a computational approach called TARGETS (TreAtment Response Generalized Elastic-neT Signatures) for predicting patient response to cancer drugs.
• To assess the performance of TARGETS models using in vitro and clinical data.
• To establish TARGETS as a potential tool for patient stratification and drug repurposing in oncology.

Main Methods:

• Utilized in vitro DNA/RNA sequencing and drug response data from the Genomics of Drug Sensitivity in Cancer (GDSC) database.
• Trained predictive models using Elastic-Net regression.
• Validated models on independent datasets including the Cancer Cell Line Encyclopedia (CCLE) and The Cancer Genome Atlas (TCGA), as well as clinical samples from the WCDT.

Main Results:

• TARGETS models accurately predicted treatment response in the CCLE dataset.
• Predictions for FDA-approved biomarker-based drug indications in TCGA were concordant with established clinical uses.
• The TARGETS AR signaling inhibitors (ARSI) signature demonstrated significant clinical validation in predicting treatment response in metastatic castration-resistant prostate cancer.

Conclusions:

• TARGETS provides a pan-cancer, platform-independent method for predicting oncologic therapy response.
• This approach can aid in selecting appropriate patients for current therapies and identifying new therapeutic indications.
• TARGETS holds potential for improving clinical trial design and patient outcomes in cancer treatment.