The aim of the study was to compare the sensitivity and specificity of intraoperative identification of primary tumours in patients with unknown primary head and neck squamous cell carcinoma (UP HNSCC) using transoral laser microsurgery (TLM) with frozen sections, with PET-CT. Other aims include comparing intraoperative frozen section margins to final histopathology. After a PET-CT, patients underwent partial oropharyngectomy of the most likely primary tumour site via TLM. The ipsilateral palatine tonsil was excised; if no tumour was identified on frozen section, the ipsilateral lingual tonsil was excised. If this was negative on frozen, the contralateral palatine tonsil and lingual tonsil were excised sequentially. If the primary tumour was identified, margins were assessed using frozen section and if needed, re-resection was performed. This protocol has the aim of reducing surgical morbidity and the need for (chemo)radiotherapy. Sixty-two patients were included. Of 59 who underwent a PET-CT, 34 (57.6%) localised a potential primary. Twenty-seven PET-CTs correctly identified a primary and were truly negative for uninvolved subsites. Seven incorrectly identified a potential primary and were also falsely negative for the true primary site. Nineteen PET-CTs were falsely negative and six true negatives.
Sensitivity and specificity of PET-CT was 50.9% and 82.5% respectively. Intraoperatively, 50 of 61 (82.0%) patients had their primary identified. Overall, 55 of 61 (90.2%) patients had their primary tumour identified on final histopathology; 49 of 50 (98.0%) with intraoperatively identified tumours had negative margins on frozen section; 45/50 (88.2%) margins remained negative on final histopathology, and 4(8.2%) were positive, leading to a negative predictive value of 91.8% for margin status on frozen section.
Of 61 patients, 57 had p16-positive tumours (93.4%). Of 60 patients, 18 (30%) received no further treatment, 28 (46.7%) postoperative radiotherapy, and 14 (23.3%) received postoperative chemoradiotherapy. Of the six patients whose primary tumour remained unknown, one had no further treatment, three radiotherapy, and one chemoradiotherapy based on their lymph node status. One had unknown adjuvant therapy status. The study suggests that the above Dalhousie PET-CT TLM protocol has an excellent detection rate of primary tumours in UP HNSCC.