Patterns of failure in patients receiving definitive and postoperative IMRT for head-and-neck cancer.

PURPOSE To analyze the patterns of locoregional failure in patients with head-and-neck cancer treated with inverse planning intensity-modulated radiotherapy (IMRT). METHODS AND MATERIALS Between February 1997 and December 2000, 165 patients with histologically confirmed head-and-neck cancer were treated using a parotid-sparing inverse planning IMRT protocol. Thirty-nine patients who received either palliative repeat irradiation or IMRT as a boost were excluded from this analysis, leaving 126 patients for this analysis. Of the 126 patients, 30 were women and 96 were men (median age 56 years, range 13-84). Fifty-two patients (41%) received definitive IMRT. Of the 52 patients, 17 were treated with RT alone and 35 with concurrent cisplatin-based chemotherapy regimens. Seventy-four patients (59%) received postoperative IMRT. The median follow-up was 26 months (range 12-55). IMRT was used only in the upper neck for salivary sparing. The lower neck was treated with a conventional AP low-neck port abutted to the inferior IMRT dose distribution border. The radiation dose was prescribed to the two clinical target volumes (CTVs) according to the assumed risk of containing disease. The mean dose for definitive IMRT patients was 72.64 +/- 4.83 Gy to CTV1 and 64.34 +/- 5.15 Gy to CTV2. The mean dose to CTV1 and CTV2 in postoperative cases was 68.53 +/- 4.71 Gy and 60.95 +/- 5.33 Gy, respectively. The locations of failure were analyzed. RESULTS Seventeen locoregional failures (persistent or recurrent disease) were found. Of these 17 failures, 9 (53%) were inside CTV1. One failure (6%) was marginal to CTV1 but inside CTV2. One failure (6%) occurred outside CTV1 but inside CTV2. Another failure was marginal to CTV2. Of the 17 failures, 5 (28%) were found outside of the IMRT field and in the lower neck. Dose-volume histogram analysis revealed that for all but 1 patient, the recurrent/persistent disease within the CTVs received comparable or superior dose coverage relative to the CTV. The 2-year actuarial locoregional control rate was 85%, and the ultimate locoregional control rate after surgical salvage was 89%. We observed no dermal failure and only one marginal failure in the region adjacent to the spared parotid glands. CONCLUSION We have shown that the target definition and coverage for patients treated with IMRT for parotid sparing is adequate. The predominant tumor failure within CTV1 may imply the need to identify patients with radioresistant tumor subvolumes (such as hypoxic regions) within the CTV. This information would assist in discriminating a subgroup of tumors for a more aggressive and target-specific therapeutic approach.