Implementation of a Delineation Guideline for Dorsal Tongue Mucosa and ROI-specific Analysis of Dose-related Dysgeusia

Abstract

Only few studies have addressed dysgeusia dose-toxicity relationship in head and neck cancer patients treated with definitive RT, all without ROI-specific analysis. To this end, we aim to assess the impact of different delineation methods of taste bud-bearing tongue mucosa on the dosimetric parameters and toxicity outcomes. A total of 42 T1-3 NX-2 oropharyngeal cancer patients treated with IMRT alone were prospectively enrolled under an IRB-approved protocol. All patients were disease-free and without cancer-directed therapy for the last 6 months prior to enrollment. Three ROIs were manually delineated on the planning CT: A) the whole tongue (extrinsic and intrinsic muscles), B) the dorsal mucosa of the oral tongue using a highly reproducible approach by adapting the whole tongue structure in a defined way and C) the dorsal mucosa using a much faster method with axial adaption of a midsagittal contour. Both, B and C, consisted of a 5mm layer reaching from the apex along the upper tongue surface to the base of tongue. Dosimetric parameters were calculated from the original RT plan. The single item Taste assessment of the MD Anderson Symptom Inventory – Head and Neck module (MDASI-HN), rating taste changes between 0 (“not at all”) and 10 (“the worst you can imagine”), was collected after a median of 5 years following IMRT (range 2-10). Wilcoxon signed rank test has been used to test for differences between the contouring methods. TD50 of dysgeusia (scores of 1+) was calculated using logistic regression. Multivariate analysis was done to assess for correlation with clinical parameters. A p-value of <0.05 was considered significant. The mean volume of the contours differed significantly between the three methods, with highest volume for the whole tongue (method A, 122.2cc, SD 26.0), followed by the mucosa contoured according to method B (20.0cc, SD 5.8) and C (18.8cc, SD 4.9). Dose difference was statistically significant between method A / B and method A / C regarding min, max, mean doses as well as V30, 40, 50 (all p<0.001). The difference between the mean dose in method A (49.3 Gy) and B (46.6 Gy) accounted on average +2.6 Gy (SD 4.5, range -6.3 – 19.7). There was also a difference for V20 (p<0.06) and V60 (p<0.05), but in a lesser extent. No significant difference in any of the dose parameters was found between method B and C. Mean MDASI Taste score was 2.3 (SD 3.1, range 0-10) with 57% of patients having dysgeusia. Patients with higher mean doses in the ROIs were more likely to exhibit dysgeusia. TD50 for mean dose was 30.8 Gy with method A, and much higher with method B (40.4 Gy) and C (39.8 Gy). No significant correlation was found between dysgeusia and age, smoking status or T stage. Dosimetric results differed significantly for nearly all tested parameters with the different delineation methods contouring the whole tongue or the dorsal mucosa only. This is highly relevant and should be considered in the evaluation of any normal tissue complication probability of dysgeusia.