Development of Prediction Model for Mean Parotid Dose of HNC Undergoing Radiotherapy - A Single Institutional Study

Area of Science:

• Radiation Oncology
• Medical Physics
• Head and Neck Cancer Treatment

Background:

• Head-and-neck cancer (HNC) patients undergoing radiotherapy require careful planning to spare organs at risk, particularly the parotid glands.
• Accurate prediction of mean parotid dose is crucial for minimizing xerostomia and other radiation-induced toxicities.
• Existing methods may not always be efficient or precise in optimizing dose distribution for parotid glands.

Purpose of the Study:

• To develop a simple prediction model for estimating the mean parotid dose in HNC patients.
• To utilize previous treatment plans to create a predictive tool for radiotherapy planning.
• To improve the efficiency and accuracy of dose sparing for parotid glands during HNC treatment.

Main Methods:

• Developed and compared two prediction models (linear regression and exponential) using data from 95 HNC patients treated with VMAT.
• The predictor variable used was fractional overlapping parotid volume with the planning target volume (PTV) plus a 2 mm margin.
• Statistical analysis included outlier exclusion, R-squared, root-mean-square error calculation, and Shapiro-Wilk test for residual normality.

Main Results:

• Both linear and exponential models showed strong predictive capabilities, with R-squared values of 0.85 and 0.82, respectively.
• The models predicted a mean parotid dose of 26 Gy with fractional overlaps of 16.4% (linear) and 18.9% (exponential).
• Prospective implementation in 12 patients demonstrated significant improvements in minimizing dose to the parotid glands.

Conclusions:

• A simple prediction model for mean parotid dose in HNC radiotherapy was successfully developed.
• The model shows promise for assisting planners in optimizing treatment, reducing toxicity, and enhancing planning efficiency.
• Further validation with external datasets and clinical integration are recommended for future studies.