Radiation Doses to All Salivary Glands Affect Patient-Reported Xerostomia and Quality of Life: A Longitudinal Study

Abstract

Randomized studies comparing 2-dimensional (2D) radiotherapy (RT) and parotid-sparing intensity-modulated RT (IMRT) in patients with head and neck cancer (HNC) have demonstrated improvements in saliva output and observer-rated xerostomia with IMRT. However, patient-reported xerostomia (PRX) has either not differed (Kam MK et al., JCO, 2007) or been only marginally better with IMRT in the first year post-therapy (Nutting CM et al., Lancet Oncology, 2011). We hypothesized that doses to all salivary glands, including the mucin-producing submandibular and minor glands within the oral cavity, are important determinants of PRX. We sought to assess the effect of doses to these structures in patients with HNC receiving IMRT aiming to spare all glands. Patients with HNC receiving IMRT to the bilateral neck answered a validated xerostomia questionnaire (XQsum) consisting of 4 questions inquiring about oral dryness during eating and speaking (XQeat), and 4 about dryness at rest (XQrest), as well as a validated, multi-domain HN quality of life questionnaire (HNQOL). The scale for XQ ranges from 0-100, with higher scores denoting worse xerostomia. PRX and HNQOL data were obtained at each follow-up visit through 48 months. Univariate and multivariate Cox-regression were used to correlate PRX and HNQOL scores with clinical factors, time since therapy, and mean RT doses to the parotid glands (PG), contralateral submandibular gland (SMG), and oral cavity (OC). These factors were combined into random effect models to account for within-person correlations arising from repeated measures. Two-hundred and fifty-two HNC patients treated with IMRT completed a total of approximately 600 PRX and HNQOL questionnaires through 48 months post-RT. Median mean doses to both parotids, contralateral SMG, and OC, were 36 (range 3-64), 31 (6-57), and 37 (8-70) Gy, respectively. Mean OC doses correlated moderately with both mean PG (r=0.42, p<0.001) and SMG (r=0.34, p<0.001) doses, as did mean PG and SMG doses with each other (r=0.33, p<0.001). Mean XQsum scores were 39 (standard deviation 21), 31 (22), 29 (22), and 27 (21) at 1, 12, 24, and 48 months post-RT, respectively. On univariate analysis, N stage, and mean PG, SMG, and OC doses statistically significantly correlated with XQsum, XQeat, XQrest, HNQOL Summary, and HNQOL Eating domain scores (all p<0.01). In all instances, higher mean doses were associated with higher (worse) scores. Scores improved significantly over time, with steeper improvement of XQsum at lower SMG dose. On multivariate analysis, mean PG and OC doses significantly correlated with XQtotal, XQeat, and HNQOL Eating scores, and mean SMG dose with HNQOL summary. These results suggest that reducing doses to all salivary glands, including PG, SMG, and minor glands within the OC, is likely to improve PRX and HNQOL compared with PG-sparing alone.