Revisiting inconsistency in large pharmacogenomic studies

Area of Science:

• Pharmacogenomics
• Cancer Cell Line Research
• Biomarker Discovery

Background:

• Previous analysis in 2013 found inconsistencies in drug sensitivity data between GDSC and CCLE, despite concordant gene expression profiles.
• Extensive feedback and new data prompted a re-analysis to address limitations in the initial study.

Purpose of the Study:

• To re-evaluate and compare mutation and gene expression profiles and drug sensitivity data from GDSC and CCLE using updated datasets and improved methodologies.
• To assess the consistency of drug sensitivity measurements, differentiating between targeted therapies and broad cytotoxic drugs.
• To improve the usability of analysis tools for researchers working with GDSC and CCLE data.

Main Methods:

• Performed a comparative analysis of mutation and gene expression profiles and drug sensitivity data for 15 drugs across 471 cancer cell lines.
• Applied new statistical methods to assess data consistency, distinguishing between drug types and comparing profiles across cell lines.
• Utilized expanded datasets from GDSC and CCLE for the re-analysis.

Main Results:

• Gene expression data showed significantly higher consistency between GDSC and CCLE compared to drug sensitivity measurements.
• Identified two broad-effect and three targeted drugs with moderate to good drug sensitivity consistency.
• Found inconsistencies in pharmacological phenotypes for eight drugs, with insufficient data for three other targeted drugs.

Conclusions:

• Drug sensitivity data in GDSC and CCLE continue to pose challenges for reliable biomarker discovery.
• Experimental standardization and validation of pharmacogenomic responses are crucial for advancing the use of large-scale pharmacogenomic screens.