A Bayesian Approach for Prediction of Patient Radiosensitivity.

Abstract

A priori identification of the small proportion of radiation therapy patients who prove to be severely radiosensitive is a long-held goal in radiation oncology. A number of published studies indicate that analysis of the DNA damage response after ex vivo irradiation of peripheral blood lymphocytes, using the γ-H2AX assay to detect DNA damage, provides a basis for a functional assay for identification of the small proportion of severely radiosensitive cancer patients undergoing radiotherapy. We introduce a new, more rigorous, integrated approach to analysis of radiation-induced γ-H2AX response, using Bayesian statistics. This approach shows excellent discrimination between radiosensitive and non-radiosensitive patient groups described in a previously reported data set. Bayesian statistical analysis provides a more appropriate and reliable methodology for future prospective studies.