Recurrences near base of skull after IMRT for head-and-neck cancer: implications for target delineation in high neck and for parotid gland sparing.

PURPOSE: Locoregional (LR) failures near the base of the skull, and their relationships to the targets in the high neck, were examined in a series of patients who underwent intensity-modulated radiotherapy (IMRT) for head-and-neck cancer. METHODS AND MATERIALS: Between 1994 and 2002, 133 patients with non-nasopharyngeal head-and-neck squamous cell carcinoma completed a course of curative, parotid-sparing RT. Treatment was delivered until 1996 with conformal three-dimensional techniques and thereafter with multisegmental static IMRT. Of the 133 patients, 80 had oropharyngeal, 27 oral cavity, 12 hypopharyngeal, 11 laryngeal, and 3 unknown primary cancer. The AJCC stage was I in 1, II in 6, III in 26, IVA in 83, and IVB in 12; cancer was recurrent in 5. Sixty patients received primary and 73 postoperative IMRT. 86% of patients (n = 115) had ipsilateral neck metastasis. In all patients, the contralateral neck was clinically node negative but was judged to be at high risk of subclinical disease. Delineation of the superior-most extent of the nodal targets was consistent and corresponded with Rouviere's observations. In the contralateral clinically node-negative neck, the uppermost Level II nodal target was the subdigastric (SD) nodes. To ensure coverage, the uppermost clinical target volume was delineated at the axial CT image in which the posterior belly of the digastric muscle crossed the jugular vein. In the ipsilateral neck, which was node positive in most patients, the uppermost Level II clinical target volume was delineated through the base of the skull. The uppermost retropharyngeal (RP) nodal target was delineated at the level of the top of the C1 vertebral body, accommodating Rouviere's description of the location of the lateral RP nodes. The dose prescription was 70 Gy for the primary planning target volumes (PTVs); 64 Gy and 60 Gy for PTV of the postoperative beds with and without extracapsular extension, respectively; and 50-54 Gy for PTVs of nonoperated subclinical disease, at 1.8-2.0 Gy fractions. In-field or marginal recurrences were defined as those occurring when >95% or 20-95% of the recurrence volume, respectively, had received >95% of the prescribed dose.
RESULTS: At a median follow-up of 32 months (range, 6-107 months), 21 patients (16%) had locoregional recurrence. Seventeen recurrences were in-field and four were marginal. The most prevalent nodal recurrence was in Level II bilaterally, where all failures were in-field. In-field failures were observed in the ipsilateral high neck cranial to the SD nodes (all in initially node-positive neck). No recurrences occurred in the contralateral high neck, cranial to the SD nodes, which was not included in the targets. The 95% confidence interval for the risk of recurrence in that region was 0-2.7%. Three RP nodal failures were observed; two were marginal, occurring ipsilateral and contralateral to the primary tumors, and centered cranial to the top of C1. After these recurrences, the RP nodal clinical target volumes were delineated bilaterally through the base of skull, without subsequent RP recurrence. These target delineation guidelines allowed the achievement of a mean contralateral parotid dose of < or =26 Gy (found previously to preserve salivary output significantly) in 82% of the patients. The 3-year actuarial LR recurrence-free survival rate of primary and postoperative IMRT patients was similar (81% and 84%, respectively). Oropharyngeal cancer patients had the greatest LR recurrence-free survival rate (94%, p <0.001). No statistically significant differences were found in the dose delivered to the PTVs or the in-field recurrence volume between patients who had or did not have LR failure.
CONCLUSION: These results suggest that when the contralateral node-negative side of the neck has a high risk of subclinical metastasis, it is adequate to include the SD nodes as the cranial-most Level II nodal target in non-nasopharyngeal head-and-neck cancer. In the node-positive side of the neck, this nodal level should be delineated more cranially. The RP nodal targets should be delineated more cranially. The RP nodal targets should be delineated bilaterally and should extend to the base of the skull, rather than to the top of C1. These guidelines allowed substantial sparing of the contralateral parotid gland. The results of this series validate a consensus for target delineation adopted recently by cooperative radiotherapy groups.