Deep learning model of primary tumor and metastatic cervical lymph nodes from CT for outcome predictions in p16+ oropharyngeal squamous cell carcinoma

Primary tumor (PT) and metastatic cervical lymph nodes (LN) characteristics are associated with prognosis of OPSCC. A retrospective prognostic studie of 811 patients with p16+ OPSCC resulted in a deep learning model integrating pretreatment PT and LN computed tomography results. The model significantly stratified patients in low-versus high-risk groups for disease-free survival, overall survival, and locoregional failure. The model also identified patients who would derive addded benefit associated with concurrent chemoradiotherapy over definitive radiotherapy alone among patients with stage I disease. The authors conclude that the model is prognostic of survival outcomes and predictive of added benefit of chemotherapy among patients with p16+ OPSCC.