KEY POINTS
- This retrospective single-center pilot study from University of Tsukuba, Japan, included 32 patients receiving definitive IMRT for head and neck cancer between January 2024 and November 2025. Thirteen patients had oropharyngeal cancer, while 19 had hypopharyngeal or laryngeal cancer. All received 70 Gy in 35 fractions with bilateral neck irradiation, and 15 patients (47%) received concurrent cisplatin-based chemotherapy.
- Taste dysfunction was assessed repeatedly during radiotherapy using routine patient diaries. Patients rated sweet, salty, and bitter taste perception at 10-Gy prescribed-dose intervals on a three-level scale, producing a total score from 0 to 6. Of 256 planned assessments, 248 (96.9%) were available. The taste-bud-bearing tongue mucosa was represented by a 5-mm-thick superior tongue structure on planning CT, with cumulative mean dose estimated at each assessment.
- Mean tongue mucosal exposure differed substantially by primary tumor location. At treatment completion, median taste-bud-bearing mucosal dose was 42.2 Gy in oropharyngeal cancer versus 21.2 Gy in hypopharyngeal/laryngeal cancer. Tongue mucosal overlap with the PTV occurred in all 13 oropharyngeal patients, compared with 3/19 during initial treatment and 2/19 during boost planning for hypopharyngeal/laryngeal disease. The median cumulative mean dose across the entire cohort was 27.1 Gy.
- A nonlinear dose-response model provided better statistical fit than linear or quadratic models. The exploratory piecewise model used a 20-Gy knot and achieved AIC 901.2, compared with 905.2 for a quadratic and 910.7 for a linear model. Below 20 Gy, taste dysfunction increased by 1.99 points per 10 Gy (95% CI 1.58–2.41). Above 20 Gy, the estimated slope decreased to 0.95 points per 10 Gy (95% CI 0.42–1.48), with a significant change in slope (p<0.001).
- The flattening of the dose-response relationship should not be interpreted as a biological tolerance threshold. The 20-Gy knot was selected by visual inspection, and alternative knot positions between 17.5 and 25 Gy produced similar qualitative results. Taste symptoms also increased with treatment duration, potentially reflecting mucosal inflammation, chemotherapy exposure, and cumulative treatment burden. The bounded 0–6 score could produce ceiling effects, making symptom increases appear smaller at higher doses.
- Taste-bud dose did not outperform conventional oral cavity dose as a predictor. Cumulative mean tongue mucosal and extended oral cavity doses were highly correlated, with Spearman ρ ranging from 0.921 to 0.964 across treatment milestones. Exploratory models using either structure had similar goodness of fit and cross-validated prediction errors. Estimated optimal knots were 23.0 Gy for tongue mucosa and 25.5 Gy for oral cavity dose, with overlapping uncertainty ranges.
- The findings do not yet support an additional mandatory OAR constraint. The study included only 32 patients, used a nonvalidated symptom scale, and modeled ordinal taste scores as continuous outcomes. No standardized CTCAE dysgeusia assessments, clinical taste testing, or prospective dose-sparing intervention were performed. Consequently, the results cannot establish whether reducing tongue mucosal dose improves taste preservation independently of broader oral cavity sparing.
CLINICAL TAKEAWAY
Cumulative dose to taste-bearing tongue mucosa was associated with patient-reported dysgeusia during head and neck IMRT, with a steeper symptom increase at lower observed dose levels. However, the exploratory 20-Gy inflection point is not a validated planning constraint, and tongue-specific dose did not provide a clear predictive advantage over oral cavity dose. Prospective dose optimization and standardized taste assessments are needed before incorporating this structure into routine clinical constraints.
SOURCE
Technical Innovations & Patient Support in Radiation Oncology