A CT-based radiomics signature for preoperative discrimination between high and low expression of programmed death ligand 1 in head and neck squamous cell carcinoma. Issue 146 (January 2022)

• Record Type: Journal Article
• Authors: Zheng, Ying-mei; Zhan, Jin-feng; Yuan, Ming-gang; Hou, Feng; Jiang, Gang; Wu, Zeng-jie; Dong, Cheng
• Is Part Of: European journal of radiology. Issue 146(2022)
• Publication Date: 2022-01
• Subjects: Squamous Cell Carcinoma of Head and Neck -- PD-L1 expression -- Tomography -- X-ray computed -- Radiomics; Medical radiology -- Periodicals; Radiology -- Periodicals; Radiologie médicale -- Périodiques; Medical radiology; Periodicals; 616.075705
• DOI: 10.1016/j.ejrad.2021.110093
• Languages: English
• ISSNs: 0720-048X

Abstract

Highlights: Accurate prediction of the PD-L1 expression level in HNSCC is crucial. We developed a CECT-based radiomics model for prediction of PD-L1 expression level. The model showed favorable efficacy for prediction of PD-L1 expression level. Abstract: Purpose: Accurate prediction of the expression level of programmed death ligand 1 (PD-L1) in head and neck squamous cell carcinoma (HNSCC) is crucial before immunotherapy. The purpose of this study was to construct and validate a contrast-enhanced computed tomography (CECT)-based radiomics signature to discriminate between high and low expression status of PD-L1. Methods: A total of 179 HNSCC patients who underwent immunohistochemical examination of tumor PD-L1 expression at one of two centers were enrolled in this study and divided into a training set (n = 122; 55 high PD-L1 expression and 67 low PD-L1 expression) and an external validation set (n = 57; 26 high PD-L1 expression and 31 low PD-L1 expression). The least absolute shrinkage and selection operator method was used to select the key features for a CECT-image-based radiomics signature. The performance of the radiomics signature was assessed using receiver operating characteristics analysis. Results: Six features were finally selected to construct the radiomics signature. The performance of the radiomics signature in the discrimination between high and low PD-L1 expression status was good in both the training and validation sets, with areas under the receiver operating characteristics curve of 0.889 and 0.834 for the training and validation sets, respectively. Conclusions: The constructed CECT-based radiomics signature model showed favorable performance for discriminating between high and low PD-L1 expression status in HNSCC patients. It may be useful for screening out those patients with HNSCC who can best benefit from anti-PD-L1 immunotherapy.