Prediction model for cervical lymph node metastasis in human papillomavirus-related oropharyngeal squamous cell carcinomas

Abstract

Objectives To develop and validate a risk scoring system based on clinical and imaging findings to predict lymph node metastasis from HPV-related oropharyngeal squamous cell carcinomas. Methods This study population who had undergone neck dissections or lymph node biopsies in patients with HPV+ OPSCC was obtained from a historical cohort from two tertiary referral hospitals. The training set from one hospital included 455 lymph nodes from 82 patients, and the test set from the other hospital included 150 lymph nodes from 42 patients. The baseline clinical and imaging findings on pretreatment CT or MR were investigated and the reference standards were the histopathologic results. A risk scoring system was constructed based on logistic regression and validated both internally and externally. Results A 7-point risk scoring system was developed based on the following variables: central necrosis, infiltration of adjacent planes, lymph node level, and the maximal axial diameter of the lymph node. This risk scoring system showed good discriminative ability for metastasis in the training set (C-statistic 0.952; 95% CI, 0.931-0.972) and test set (C-statistic 0.968, 95% CI, 0.936-0.999) and good calibration ability in the training set (p = 0.723) and test set (p = 0.253). Conclusions We developed and validated a reliable risk scoring system that predicts lymph node metastasis from HPV+ OPSCCs based on the clinical data and pretreatment imaging findings. We expect this risk scoring system to be a useful guide for better decision-making in practice.