Impact of Preoperative Imaging on Identifying Rapid Recurrence in Head and Neck Cancer

Abstract

Rapid recurrence, defined as detectable tumor recrudescence after an oncologically sound primary operation but prior to initiating postoperative radiation therapy (PORT), is now recognized to be an underreported poor prognostic factor in head and neck cancer (HNC), occurring at a rate of 15-30% across different series. We sought to determine and analyze the impact of preoperative imaging on detection of rapid recurrence at the time of Computer Tomography (CT) simulation. CT simulation images of HNC patients treated with primary surgery and PORT between 2009 and 2017 were retrospectively reviewed by a neuroradiologist blinded to patient outcome but provided with pre-operative imaging and the surgical pathology report. Findings in the operative bed, ipsilateral neck and/or contralateral neck considered either suspected or definite recurrence were categorized as rapid recurrence. Multivariable analysis (MVA) was conducted to ascertain risk factors for rapid recurrence. Variables included were: days between imaging and surgery, number of imaging modalities, extranodal extension (ENE), lymphovascular invasion (LVI), perineural invasion (PNI), and pathologic stage. A total of 188 HNC patients were included with oral cavity and squamous cell carcinoma being the most common subsite (58%) and histology (84%). Overall, 22% had ENE, 24% had LVI, 45% had PNI, and 77% were AJCC 7th edition pathologic stage III/IV. Thirty-eight patients (20%) were identified to have rapid recurrence, of which 21 were at the operative bed, 17 were at the ipsilateral neck and 13 were at the contralateral neck. 13 patients (34%) had rapid recurrence at multiple sites. The median time between initial imaging and day of surgery for patients with and without rapid recurrence was 24.5 and 27 days, respectively (p= 0.28). Pre-operative imaging modality included CT, Magnetic Resonance Imaging (MRI) and Positron Emission Tomography fused to CT dataset (PET/CT). 107 patients (57%) had a single imaging modality and among them 86 had CT, 13 had MRI, and 8 had PET/CT. 81 patients (43%) had > 1 imaging modality; most frequently anatomic imaging (CT and/or MRI) and PET/CT (81%). The median time from initial imaging and day of surgery in patients with 1 and > 1 imaging modality were 24 and 31 days, respectively (p= 0.02). On MVA, patients with pre-operative PET/CT were less likely to have local or regional rapid recurrence (OR=0.30, 95% CI 0.13-0.73, p< 0.01). ENE was a predictor of rapid recurrence (OR 2.7, 95% CI 1.01-7.4, p= 0.05). There were no difference in ENE, LVI, PNI or pathologic stage in patients with and without PET/CT. The time between the initial imaging and the date of operation was not associated with rapid recurrence. Patients evaluated with functional preoperative imaging modalities were less likely to have rapid recurrence at the primary site and/or neck(s) at the time of CT simulation.