Isoeffect models and fractionated radiation therapy.

Abstract

Since the advent of the Ellis formula, a number of isoeffect formulae have been proposed for estimating the total dose required in different fractionation schedules to produce equivalent biological effects.‘*3*” Most often these formulae are applied retrospectively to clinical data thus making it impossible to control factors such as the endpoint for assessing damage, non-biased scoring of the damage by (preferably) more than one independent observer, the criteria for patient eligibility, and adequate follow-up records for assessment. Thus, these formulae often are viewed skeptically by clinicians. However, there is an urgent need to accurately and concisely define isoeffect relationships for human normal tissues in vivo because of the implied steepness of dose-response curves as derived from animal data. Also, the initiation of new fractionation schedules, for example, hyperfractionation, are all based on experimental data and, although a comparison of mouse and human data indicates that the shape of the dose-response curves for the two species is similar, the absolute dose level for response varies, generally in the direction of a lower isoeffect dose in humans. The lung is one dose-limiting normal tissue for which there are some dose-response data available for humans. The lung is dose-limiting not only in the treatment of malignant diseases of the thorax but in the use of total body irradiation as a preparative regimen for bone marrow transplantation. This latter treatment and the use of upper half body irradiation have resulted in published dose-response curves for the incidence of radiation pneumonitis in humans, at least after large single doses of radiation. However, limited data are available after fractionated doses and these are, again, from retrospective studies.