Room: Exhibit Hall
Purpose: To investigate an innovative heterogeneous dose based predictive model of normal tissue complication and its correlation to clinical endpoints for intensity modulated radiotherapy (IMRT) of head and neck (HN) cancer.
Methods: Fifty IMRT plans of HN patients are employed for a phenomenological study of the heterogeneous dose distribution with the cluster model which represents the aggregation of higher dose in the organ at risk (OAR) close to the planning target volumes (PTVs) and is characterized with different connectivity choices. Cluster structures and sizes inside the OARs are investigated to examine the pattern consistency among the first twenty-five plans where similar PTV coverage and OAR sparing are ensured. In the second group of twenty-five plans, clusters of the esophagus are scrutinized to correlate their sizes to dosimetric and radiobiological indices. A multivariate analysis based on logistic regression model is performed to quantitatively evaluate the correlations with varying reference doses. Finally, an extensive Monte Carlo (MC) procedure is developed to gain deeper insights into the behavioral properties of the cluster formation.
Results: Clinical studies show that the OAR clusters have substantial variations in spite of similar PTV coverage among different patients. Minimal difference is observed with clinically similar plans from the same patient, indicating that clustering pattern is sensitive to geographic locations between OARs and PTV. The radiobiological parameters fail to positively correlate with the cluster sizes. MC studies demonstrate the inverse relationship between the mean size of the largest cluster and the cluster connectivity and the nonlinear changes in cluster size with fractional density regardless of connectivity types.
Conclusion: Dose-based cluster models can potentially be a useful index to evaluate normal tissue complication. Clinical results demonstrated that cluster parameters are not necessarily consistent with conventional NTCP model predictions, and the clinical outcome of the same dose volume might be different.
Not Applicable / None Entered.