A Procedure for Estimating the Dose Modifying Effect of Chemotherapy on Radiation Response

Abstract

A procedure based on a logistic regression model was used to estimate the dose-modifying effect of chemotherapy on the response of normal tissues to radiation. The magnitude of this modifying effect is usually expressed as a dose–effect factor (DEF). Values > 1·0 indicate an enhancing effect of chemotherapy while values < 1·0 suggest a ‘protective’ effect. The DEF in the proposed procedure is expressed as a function of logistic regression coefficients, response levels and values of covariates in the model. The proposed procedure is advantageous as it allows consideration of both the response levels and the values of covariates in calculating the DEF. A plot of the DEF against the response or a covariate describes how the DEF varies with the response levels or the covariate values. Confidence intervals of the DEF were obtained based on the normal approximation of the distribution of the estimated DEF and on a non-parametric Bootstrap method. An example is given to illustrate the proposed procedure.