Definitive Chemoradiation Treatment Response Evaluation Using NI-RADS and ctHPVDNA for HPV-Associated Oropharyngeal Squamous Cell Carcinoma

Abstract

To compare the evaluation of treatment response among patients with human papillomavirus (HPV)-associated oropharyngeal squamous cell carcinoma (OPSCC) treated with definitive (chemo)radiotherapy (CRT). Patients with locally advanced HPV-associated OPSCC treated with definitive radiotherapy (RT) or CRT from 2019 to 2022 at a single institution were reviewed. Patients underwent standard 3-month post-CRT positron emission tomography/computed tomography (PET/CT) scan with or without contrast enhanced CT (CECT) of the head and neck. Plasma circulating tumor HPV DNA (ctHPVDNA) was collected from 2-8 months post-CRT. Equivocal findings on post-CRT imaging prompted repeat evaluation. Imaging response was assessed via NI-RADS (Neck Imaging Reporting and Data System) risk classification and independently reviewed by two board certified radiologists, both blinded to outcomes and ctHPVDNA values. Our cohort of 52 patients included: 87% males; median age 61.5; 63% never smokers, 31% former smokers, 6% current smokers; 44% tonsil primary, 46% base of tongue; 4% T0, 25% T1, 40% T2, 12% T3, 17% T4; 6% N0, 15% N1, 2% N2a, 44% N2b, 25% N2c, 8% N3 (AJCC 7th edition). Concurrent systemic therapy was received in 90%. During this period 71 PET/CTs and 15 CECTs were reviewed for treatment response evaluation; 44% (23/52) patients required additional imaging for equivocal findings; 62 ctHPVDNA blood samples were co-analyzed for treatment evaluation. The highest risk classification score between mucosa, primary, and/or neck site was: 42% NI-RADS 1, 48% NI-RADS 2, and 10% NI-RADS 3. Only patients with locoregional disease recurrence/progression were included for evaluation comparison between imaging and circulating biomarkers. No cancer events occurred without imaging and/or ctHPVDNA detection. Patients with NI-RADS score ≥2 during first post-CRT imaging evaluation more frequently underwent additional imaging (70% vs 30%, p<0.001). NI-RADS risk classification suggested 5 locoregional events (2 true positives, 3 false positives) resulting in 100% sensitivity, 94% specificity, 40% positive predictive value (PPV), and 100% negative predictive value (NPV). Circulating tumor HPV-DNA identified 2 locoregional events (2 true positives, 0 false positives) resulting in 100% sensitivity, 100% specificity, 100% PPV, and 100% NPV. Salvage operations were performed in 2 of 3 patients with false positive disease by NI-RADS classification without any evidence of cancer on final pathology. While limited by the small number of recurrence events in this cohort, ctHPVDNA for HPV-associated OPSCC in conjunction with post-treatment imaging evaluation may limit the need for repeat imaging and unwarranted salvage operations that increase patient worry, morbidity, and financial toxicity. Additional prospective study is warranted.