Parotid Dose–Response and Xerostomia Risk In Head and Neck Cancer Treated with IMPT and IMRT
Abstract
Purpose
To evaluate and compare the incidence of xerostomia (“dry mouth”) in head and neck (H&N) cancer patients treated with intensity-modulated proton therapy (IMPT) and intensity-modulated radiation therapy (IMRT), across unilateral (UT) and bilateral (BT) treatment plans, while accounting for parotidectomy status and characterizing parotid gland dose-response relationships using dosimetric and NTCP analyses.
Methods
A total of 173 H&N patients treated at the University of Miami with IMPT or IMRT were included. In patients with parotidectomy, parotid dose data were unreliable, therefore, analyses focused on comparing xerostomia incidence across modalities (IMPT and IMRT). Patients without parotidectomy were further grouped by modality and laterality (UT and BT). UT patients were assessed by analyzing the ipsilateral parotid doses and BT patients were assessed by analyzing the combined parotid doses. Boxplots and normal tissue complication probability (NTCP) curves were generated for each modality and treatment plan for analysis.
Results
Patients with or without parotidectomy do not show a significant toxicity difference when treated with IMPT (p=0.32) or IMRT (p=0.15). Among parotidectomy patients, 62% treated with IMPT developed xerostomia, compared with 91% treated with IMRT (p=0.032). Among BT patients, cumulative mean parotid dose was higher in patients with xerostomia; this difference was significant for IMRT (p<0.001) but not for IMPT (p=0.267), likely due to limited sample size. NTCP curve analysis demonstrated left-shifted curves for IMPT in both laterality groups, indicating lower tolerance doses for 50% complication probability (TD50) compared with IMRT. The derived relative biological effectiveness (RBE) was 1.55 for the UT cohort and 1.57 for the BT cohort.
Conclusion
QUANTEC-based parotid dose constraints may inadequately classify xerostomia risk. The RBE for salivary gland toxicity in IMPT is substantially higher than the conventional constant value of 1.1.