Optimization of radiotherapy. Some notes on the principles and practice of optimization in cancer treatment and implications for clinical research.

Abstract

Two methods dominate cancer treatment--one, the traditional best practice, individualized treatment method and two, the a priori determined decision method of the interinstitutional, cooperative, clinical trial. In the first, choices are infinite and can be made at the time of treatment; in the second, choices are finite and are made in advance of treatment on a random basis. Neither method systematically selects, identifies, or formalizes the optimum level of effect in the treatment chosen. Of the two, it can be argued that the first, other things being equal, is more likely to select the optimum treatment. The determination of level of effect for the optimization of cancer treatment requires the generation of dose-response relationships for both benefit and risk and the introduction of benefit and risk considerations and judgements. The clinical trial, as presently constituted, doses not yield this kind of information, it being, generally, of the binary yes or no, better or worse type. The best practice, individualized treatment method can yield, when adequately documented, both a range of dose-response relationships and a variety of benefit and risk considerations. The presentation will be limited to a consideration of a single modality of cancer treatment, radiation therapy, but an analogy with other modalities of cancer treatment will be inferred. Criteria for optimization will be developed and graphic means for its identification and formalization will be demonstrated with examples taken from the radiotherapy literature. The general problem of optimization theory and practice will be discussed; the necessity for its exploration in relation to the increasing complexity of cancer treatment will be developed; and recommendations for clinical research will be made including a proposal for the support of clinics as an alternative to the support of programs.