A frequency-based gene selection method to identify robust biomarkers for radiation dose prediction.

Area of Science:

• Molecular Biology
• Radiation Biology
• Genomics

Background:

• Accurate radiation dose prediction is crucial for medical management of exposed individuals.
• Low linear energy transfer (LET) radiation exposure requires specific dose assessment methods.

Purpose of the Study:

• To identify minimal gene signatures for discriminating low and medium radiation doses.
• To develop a fast and accurate method for radiation dose prediction.

Main Methods:

• Utilized microarray data and a frequency-based gene selection approach (p-value, fold-change).
• Employed repeated cross-validation for unbiased performance assessment.
• Irradiated human blood ex vivo with varying doses (0.5-4 Gy) and analyzed lymphocyte gene expression at multiple time points.

Main Results:

• Identified radiation-responsive genes involved in apoptosis, DNA damage, and cell-cycle regulation.
• Developed small gene subsets achieving 95.7% prediction accuracy irrespective of post-irradiation time.
• Validated seven key genes using Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR).

Conclusions:

• Identified robust gene biomarkers for radiation dose level discrimination.
• These biomarkers are suitable for timely medical triage in radiation exposure incidents.