Do predictive signatures really predict response to cancer chemotherapy?

Area of Science:

• Oncology
• Genomics
• Bioinformatics

Background:

• Predictive signatures aim to forecast tumor response to chemotherapy.
• Current methods using cell line data show controversial clinical utility.
• Published signatures often rely on unreliable data, limiting their application.

Purpose of the Study:

• To discuss the challenges in deriving predictive gene expression signatures from cell line panels.
• To argue against the feasibility of obtaining fully predictive signatures for classical chemotherapy using oligo-based gene expression arrays.
• To propose alternative approaches for developing reliable predictive markers.

Main Methods:

• Critical review of existing methodologies for deriving predictive signatures.
• Analysis of limitations in using cell line panels for chemotherapy response prediction.
• Exploration of alternative strategies for biomarker discovery.

Main Results:

• Difficulty in obtaining meaningful predictive signatures from cell line panels.
• Oligo-based gene expression arrays are unlikely to yield fully predictive signatures for classical chemotherapy.
• Subtle genetic alterations causing chemotherapy resistance may not be reliably detected by standard gene expression profiling.

Conclusions:

• Current approaches to predictive signature development are flawed.
• Alternative methods are necessary to identify reliable predictive markers for patient treatment.
• Optimizing patient treatment requires more robust predictive tools.