Dose, volume, and function relationships in parotid salivary glands following conformal and intensity-modulated irradiation of head and neck cancer.

PURPOSE: To determine the relationships between the three-dimensional dose distributions in parotid glands and their saliva production, and to find the doses and irradiated volumes that permit preservation of the salivary flow following irradiation (RT). METHODS AND MATERIALS: Eighty-eight patients with head and neck cancer irradiated with parotid-sparing conformal and multisegmental intensity modulation techniques between March 1994 and August 1997 participated in the study. The mean dose and the partial volumes receiving specified doses were determined for each gland from dose-volume histograms (DVHs). Nonstimulated and stimulated saliva flow rates were selectively measured from each parotid gland before RT and at 1, 3, 6, and 12 months after the completion of RT. The data were fit using a generalized linear model and the normal tissue complication probability (NTCP) model of Lyman-Kutcher. In the latter model, a "severe complication" was defined as salivary flow rate reduced to < or =25% pre-RT flow at 12 months.
RESULTS: Saliva flow rates data were available for 152 parotid glands. Glands receiving a mean dose below or equal to a threshold (24 Gy for the unstimulated and 26 Gy for the stimulated saliva) showed substantial preservation of the flow rates following RT and continued to improve over time (to median 76% and 114% of pre-RT for the unstimulated and stimulated flow rates, respectively, at 12 months). In contrast, most glands receiving a mean dose higher than the threshold produced little saliva with no recovery over time. The output was not found to decrease as mean dose increased, as long as the threshold dose was not reached. Similarly, partial volume thresholds were found: 67%, 45%, and 24% gland volumes receiving more than 15 Gy, 30 Gy, and 45 Gy, respectively. The partial volume thresholds correlated highly with the mean dose and did not add significantly to a model predicting the saliva flow rate from the mean dose and the time since RT. The NTCP model parameters were found to be TD50 (the tolerance dose for 50% complications rate for whole organ irradiated uniformly) = 28.4 Gy, n (volume dependence parameter) = 1, and m (the slope of the dose/response relationship) = 0.18. Clinical factors including age, gender, pre-RT surgery, chemotherapy, and certain medical conditions were not found to be significantly associated with the salivary flow rates. Medications (diuretics, antidepressants, and narcotics) were found to adversely affect the unstimulated but not the stimulated flow rates.
CONCLUSIONS: Dose/volume/function relationships in the parotid glands are characterized by dose and volume thresholds, steep dose/response relationships when the thresholds are reached, and a maximal volume dependence parameter in the NTCP model. A parotid gland mean dose of < or =26 Gy should be a planning goal if substantial sparing of the gland function is desired.