Dose-Response Kinematics of Salivary Function Loss and Recovery after Radical Head and Neck Radiotherapy
Abstract
Purpose
Xerostomia is a common toxicity after radical head and neck cancer (HNC) radiotherapy (RT), causing long-term quality-of-life impairment. Although salivary function may recover, dose–response relationships governing long-term recovery remain unclear. This study defines data-driven planning constraints associated with preservation and recovery of salivary function using longitudinal clinical measurements and treatment planning data.
Methods
We retrospectively analyzed 1,098 HNC patients treated with curative-intent RT (46–70 Gy). Unstimulated salivary output was measured using the modified Schirmer Test (MST) at baseline and at 1, 6, 12, and 24 months post-RT. Submandibular glands (SMG), parotid glands, and oral cavity were retrospectively auto-contoured, and dose-volume metrics (V20–60 Gy) were extracted. Receiver operating characteristic (ROC) analysis identified the structure and dose metric most predictive of a >10 mm MST decline. A recovery threshold was defined as the maximum dose permitting return to baseline MST at 24 months within a 95% confidence interval (95CI). Recovery trajectories were compared between low- and high-dose groups.
Results
Total SMG V40Gy was the strongest predictor of salivary decline (ROC-AUC 0.65–0.83). Only SMG AUC was sensitive to the chosen volume parameter. A recovery threshold of SMG V40 < 85% was identified. Patients below this threshold demonstrated near-complete recovery (98% of baseline) by 24 months, while those above showed limited recovery (46%). Recovery was linear from 6–24 months in both groups (p<0.05), with no evidence of plateau. Rate of recovery was 17.8%/yr (95CI: 5.7 – 30.0) and 8.6%/yr (95CI: 5.5 – 11.8) in the low- and high-dose groups respectively.
Conclusion
Limiting total SMG V40 to < 85% is associated with near-complete salivary recovery at 24 months, supporting total SMG V40 as a novel, data-driven planning constraint for preserving long-term salivary function. External validation of this finding is required.