Evaluation of radiobiological models in prediction of acute esophagus complication probability following radiation therapy of thorax and neck tumors

Abstract

Evaluation of treatment plans is a process that can increase the accuracy and validity of radiotherapy and achieve the aim of delivering a higher dose to the tumor and obtaining the least normal tissue complication in radiotherapy. Our purpose in this thesis was to predict acute esophagitis (AE) grade2 following 3 dimensional conformal radiotherapy of head, neck, thoracic tumors as well as spinal metastases using radiobiological modeling, regression models, and Cox models according to clinical outcomes. Methods and materials: 100 patients were studied in two groups of concurrent chemo radiotherapy (group I) and sequential chemo radiotherapy and only radiotherapy (group II). AE grading was performed according to Radiation Therapy Oncology Group (RTOG) criteria. The studied radiobiological models were included models (Logistic, Logit, Nimireko, LKB, MD). The performance and ranking of the models were determined using Akaike criteria. Regression and Cox modeling was performed using SPSS software. Results: For all patients and patients in the first group MD model and for patients in the second group LKB model according to Akaike's information criterion (AIC) criteria were most consistent with clinical results. In multivariate Cox analysis, concurrent chemo radiotherapy and mean dose were significantly associated with acute esophagitis. In univariate logistic regression, volume and length variables were the most important predictors of acute esophagitis.