Benefit, risk, and optimization by ROC analysis in cancer radiotherapy

The objective of definitive cancer radiation therapy is cure or control. The attainment of that objective is not without risk of treatment-induced radiation injury. The optimum treatment is, therefore, that level of radiotherapeutic effect with the maximum probability of benefit and the minimum associated probability of injury. An objective of radiotherapy research is the formulation of a model of optimization that is independent of a consensus of what constitutes optimization. Receiver operating characteristic (ROC) analysis is such a model, for it can relate probabilities of benefit and injury yielding a graphical determination of the optimum level of radiotherapeutic effect. ROC analysis is explained, an example taken from the contemporary radiotherapy literature is presented, and clinical research requirements for a general application of ROC analysis to the optimization of definitive cancer radiotherapy are developed.