Incidence of radiographically occult nodal metastases in HPV+ oropharyngeal carcinoma: Implications for reducing elective nodal coverage

PURPOSE:Initial deescalation studies for human papilloma virus (HPV)-positive driven oropharyngeal squamous cell carcinomas (HPV+ OPSCC) altered radiation therapy dose or the systemic agent used. Newer trials examine the disease control achieved with a reduced elective nodal field. We examined patterns of nodal involvement in patients with HPV+ OPSCC with a focus on implications for radiation field design for treatment deescalation. METHODS AND MATERIALS:Records of patients with HPV+ OPSCC with preoperative imaging (computed tomography or fludeoxyglucose positron emission tomography/computed tomography) who underwent neck dissection without neoadjuvant therapy from 2010 to 2017 were retrospectively reviewed. The number and location of clinically positive lymph nodes on preoperative imaging were compared with those documented on pathology. These data were then used to establish the probability of missing nodal disease in 3 modified radiation field designs. RESULTS:One hundred patients were included. The median time between imaging and surgery was 22 days. The most common clinical N stage was cN2a (35%), whereas the most common pathologic N stage was pN2b (45%). The median number of radiographically and pathologically involved nodes was 1 (range, 0-6) and 2 (range, 0-11), respectively. Forty-three percent of patients had more pathologically involved nodes than predicted on imaging, whereas 21% had pathologic involvement at an additional nodal level not predicted on imaging. Of the 21 patients with additional pathologically involved nodal levels, 14 had involvement of a directly adjacent station, 4 were patients with a cN0 hemineck with pathologically positive level II disease, and 3 had pathologic involvement of level 2 echelons removed from that predicted on imaging. CONCLUSION:Our study suggests that radiation fields encompassing only clinically involved nodes or levels has an unacceptably high likelihood of missing subclinical disease. Alternatively, treating the first uninvolved echelon nodes in addition would cover pathologic sites of disease in 97% of patients. This approach merits further study in prospective trials.