Radiomic Model Predicts Lymph Node Response to Induction Chemotherapy in Locally Advanced Head and Neck Cancer

• Published in: Diagnostics (Basel) Vol. 11; no. 4; p. 588
• Main Authors: Zhang, Michael H, Cao, David, Ginat, Daniel T
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 25.03.2021; MDPI
• Subjects: Age; Biomarkers; Chemotherapy; Feature selection; HNSCC; Human papillomavirus; induction chemotherapy; lymph node; Lymphatic system; Medical prognosis; Metastasis; Patients; radiomics; texture analysis; Tumors; CT; radiomics; HNSCC; lymph node; cancer; nodal response; induction chemotherapy; texture analysis
• ISSN: 2075-4418; 2075-4418
• DOI: 10.3390/diagnostics11040588

This study developed a pretreatment CT-based radiomic model of lymph node response to induction chemotherapy in locally advanced head and neck squamous cell carcinoma (HNSCC) patients. This was a single-center retrospective study of patients with locally advanced HPV+ HNSCC. Forty-one enlarged lymph nodes were found from 27 patients on pretreatment CT and were split into 3:1 training and testing cohorts. Ninety-three radiomic features were extracted. A radiomic model and a combined radiomic-clinical model predicting lymph node response to induction chemotherapy were developed using multivariable logistic regression. Median age was 57 years old, and 93% of patients were male. Post-treatment evaluation was 32 days after treatment, with a median reduction in lymph node volume of 66%. A three-feature radiomic model (minimum, skewness, and low gray level run emphasis) and a combined radiomic-clinical model were developed. The combined model performed the best, with AUC = 0.85 on the training cohort and AUC = 0.75 on the testing cohort. A pretreatment CT-based lymph node radiomic signature combined with clinical parameters was able to predict nodal response to induction chemotherapy for patients with locally advanced HNSCC.