Better cancer biomarker discovery through better study design

Area of Science:

• Biomarker discovery
• Epidemiology
• Bioinformatics

Background:

• High-throughput technologies and bioinformatics offer potential for novel biomarker discovery.
• Biomarker discovery often uses case-control designs, particularly nested case-control designs.
• Methodological challenges exist in integrating high-throughput biomarker discovery with case-control study designs.

Purpose of the Study:

• To review principles of nested case-control study design for biomarker discovery.
• To describe how study design choices impact biomarker discovery efficiency.
• To illustrate the influence of study design on biomarker discovery using simulated prostate cancer data.

Main Methods:

• Review of nested case-control study design principles.
• Development of a simulated prostate cancer cohort dataset.
• Creation of nested case-control studies within the cohort for analysis.

Main Results:

• Standard nested case-control elements (incidence density sampling, matching) can bias biomarker discovery analyses.
• Incidence density sampling and matching reduce the specificity of discovered biomarkers.
• A novel "antimatching" protocol enhances biomarker discovery efficiency.

Conclusions:

• Discrepancies between epidemiologic study design and biomarker discovery practice hinder the identification of true biomarkers.
• Current practices increase the false-positive discovery rate in biomarker research.
• The proposed "antimatching" protocol aims to improve the validity and efficiency of biomarker discovery.