Optimization of dose-time factors for a tumor and multiple associated normal tissues

This study explores the feasibility of identifying, for a given tumor and its associated normal tissues, that combination of dosimetric and technical factors which will provide the best chance of achieving local control of the tumor without significant complications. Tumor control probability depends on the number of clonogenic cells surviving the course of treatment, which is a function of the initial tumor volume and the computed cellular surviving fraction. This in turn depends on the physical dose, number of fractions, and overall time. Similarly, the risk of injury to any one associated normal tissue is also a function of dose, field-size, fractions, and time. The best treatment scheme is assumed to be one which maximizes the chance both of tumor control and of avoiding complications. Computer programs can achieve this end by searching a wide range of fractionation schemes and selecting that with the largest therapeutic ratio or probit difference (difference between tumor control and normal tissue damage probabilities). When two or more normal tissues are involved, this process is much more complex. Fractionation schemes that reduce the risk of injury in one tissue may be worse for another. The safest procedure is then one which reconciles these conflicting requirements, searching for that combination of factors which will maximize the conditional probability of controlling the tumor and avoiding injury in any of the several tissues concerned. Results of a computer-derived solution for this type of problem: control of a hypothetical lung cancer associated with lung parenchyma, late-reacting stroma, and acutely reacting esophageal epithelium, are presented.