Genome-Scale Signatures of Gene Interaction from Compound Screens Predict Clinical Efficacy of Targeted Cancer...

Area of Science:

• Computational biology
• Genomics
• Cancer research

Background:

• Identifying reliable drug response biomarkers is crucial for effective cancer treatment.
• Current methods often overlook complex gene interactions influencing drug efficacy.

Purpose of the Study:

• To introduce Computational Analysis of Resistance (CARE), a novel method for inferring transcriptomic signatures of drug efficacy.
• To assess CARE's ability to predict therapy outcomes and identify resistance mechanisms.

Main Methods:

• CARE analyzes cell line compound screen data to generate genome-scale scores of drug target gene interactions.
• The method infers transcriptomic signatures associated with drug efficacy.
• CARE's predictions were validated using clinical transcriptome data.

Main Results:

• CARE accurately predicts therapy outcomes, outperforming existing computational and genomics approaches.
• CARE signatures for a BRAF inhibitor correlated with anti-PD-1 immunotherapy response, suggesting shared efficacy pathways.
• CARE identified PRKD3 as a lapatinib resistance gene; its inhibition sensitized cancer cells to lapatinib.

Conclusions:

• CARE enables large-scale inference of drug response biomarkers from compound screen data.
• The method facilitates the discovery of predictive biomarkers and potential drug combinations for targeted therapies.
• CARE highlights potential links between targeted therapy and immunotherapy efficacy.