Purpose: To assess efficacy of prospectively sparing parotid ducts during treatment planning.
Methods: 38 patients prospectively underwent MRI sialography-guided parotid ductal sparing (PDS) and had their patient reported xerostomia (PRX) evaluated at 6 and 12 months. 28/38 patients received de-escalated chemoradiotherapy. A parotid duct atlas was created, evaluated and used to generate parotid duct contours on a baseline cohort who did not receive PDS but were on the same de-escalation trials. PRX, defined as moderate to very severe symptoms, in the de-escalated PDS patients was compared to the aforementioned baseline cohort. Nested multivariate models including mean dose to the contralateral submandibular and parotid glands with and without mean parotid ductal doses were generated to determine if ductal dose improved model fit for PRX.
Results: The atlas-generated parotid duct contours had a mean distance-to-agreement of 3.6mm compared to the MRI sialography imaged contours and an average absolute dose difference of 3.5Gy. Atlas based contours found the mean dose to the contralateral and ipsilateral ducts was lower in patients receiving PDS by an average of 11.4Gy and 17.2Gy, respectively. PRX was lower in patients receiving PDS at 6 months post-CRT (46% vs 67%, p=0.03), and trended lower at 12 months post-CRT (32% vs 49%, p=0.11). The nested models were compared and the addition of mean parotid ductal dose significantly improved model fit at 6 and 12 months post CRT (p=0.030 and 0.047, respectively).
Conclusion: Atlas-based contours of parotid ducts demonstrated a substantial reduction of parotid ductal dose is possible with prospective sparing. This dose reduction yielded an approximately 20% absolute reduction in rates of PRX at 6 and 12 months. The significant improvement of model fit upon including mean parotid ductal dose suggests that the dose received by the parotid ducts relates to PRX even when considering traditional dosimetric predictors.