Correlation between imaging-detected and pathological extranodal extension in a randomised trial in Human Papillomavirus-positive oropharyngeal cancer

Background Imaging-detected and pathological extranodal extension (iENE, pENE) negatively impact prognosis in Human Papillomavirus (HPV)-positive oropharyngeal cancer (OPSCC), as reflected in future TNM staging updates. Correlation between iENE and pENE in HPV-positive OPSCC is currently unknown yet is vital to determine how iENE should be used to influence treatment decisions. Methods PATHOS is a trial of de-intensified adjuvant treatment after transoral surgery for HPV-positive OPSCC. 291 consecutively recruited patients undergoing surgery at three UK centres were included. Pre-operative cross-sectional imaging (CT and/or MRI) was independently scored for iENE by 2 expert radiologists; pENE was scored by 2 expert pathologists. Results Inter-rater agreement for iENE was fair in round 1 (Gwet’s AC: 0.34 (95%CI:0.26–0.41)) but improved to very good after second review (Gwet’s AC: 0.88 (95%CI:0.85–0.93), Agreement: 0.91 (95%CI:0.87–0.94)). Sensitivity of iENE for predicting pENE was relatively low (at best: 56.4% (95%CI:42.3–69.7) and specificity was high (at worst: 70.9% (95%CI:65.0–76.3)). Excluding cases with suboptimal image quality and recent core biopsy produced modest improvements in sensitivity (up to 59.4% (95%CI:40.6–76.3)) and specificity (up to 87.8% (95%CI:80.4–93.2)). Discussion The high specificity could help select iENE-negative patients for surgery, but higher sensitivity is required before excluding surgery based solely on iENE positivity.