Predicting Normal Tissue Complications after External Beam Radiation Therapy

Although modern external beam radiation (EBRT) therapy has the ability to conform the dose tightly around the volume to be treated, unwanted irradiation to surrounding normal tissue is still a problem. The probability of a side effect arising in normal tissue after EBRT is commonly modelled by an s-shaped dose-volume response curve where dose or volume are plotted against response (i.e. the frequency with which the side effect occurs). The models are based on sigmoid mathematical functions and are fitted to input data representing the outcome (absence or presence of the studied symptom associated with the side effect) and the dose distribution for potentially injured organs using statistical methods. The purpose of this thesis is to give an overview of the current forms of outcome and dose data, how they are generated and used to model side effects today, some of their limitations, and potential future directions. The results are based on concepts from the literature as well as from the three appended papers. The first two present questionnaire-collected outcome data and 2D dose-volume histogram data of the pubic bone for 650 long-term gynecological cancer survivors treated with pelvic radiation therapy between 1991 and 2003 where one in every ten women reported pubic bone pain. The mean dose to the pubic bone proved critical for pain occurrence and the risk to experience pubic bone pain two or more years after pelvic radiation therapy is expected to be 5 % for a mean absorbed dose of 25 Gy. The third paper presents dose differences between non-corrected and fractionation-corrected combined doses in sequential two-phase treatments using 16 combined dose distributions over a model organ at risk (OAR) “irradiated” with a conventional fractionation schedule (2 Gy per fraction; 46 Gy + 22 Gy). Dose differences up to 6 Gy (50 % of a reference total dose