Evaluating the molecule-based prediction of clinical drug responses in cancer

Area of Science:

• Oncology
• Bioinformatics
• Genomics

Background:

• Precision oncology aims to predict patient drug response using molecular data.
• Large-scale cancer genomic datasets like The Cancer Genome Atlas (TCGA) offer opportunities to evaluate molecular predictors.
• Evaluating the predictive utility of molecular data for clinical drug responses across multiple cancer types is crucial.

Purpose of the Study:

• To curate drug treatment data from TCGA for four chemotherapeutic drugs.
• To develop and apply a computational framework for predicting clinical drug responses using molecular data.
• To identify molecular signatures associated with drug response and evaluate prediction performance across cancer types.

Main Methods:

• Curated drug treatment records from TCGA for over 180 patients for four chemotherapeutic drugs.
• Developed a computational framework to assess molecule-based prediction of clinical drug responses.
• Utilized mRNA and miRNA expression data to identify predictive molecular signatures.

Main Results:

• mRNA and miRNA expressions significantly predicted drug responses better than random chance in specific cancer types.
• Identified signature genes like DDB1 (DNA repair) and DLL4 (Notch signaling) involved in drug response pathways.
• Prediction performance for cisplatin improved when using miRNA expressions across multiple cancer types.

Conclusions:

• Integrative analysis of clinical and molecular data can uncover predictive markers for cancer drug response.
• The developed framework provides a method for objectively evaluating molecule-based drug response predictions.
• This study serves as a foundation for advancing precision oncology through molecular data-driven predictions.