Enhanced CT-based radiomics model to predict natural killer cell infiltration and clinical prognosis in non-small cell lung cancer

作者：Xiangzhi Meng, Haijun Xu, Yicheng Liang, Mei Liang, Weijian Song, Boxuan Zhou, Jianwei Shi, Minjun Du, Yushun Gao · 发表于：Frontiers in Immunology · 年份：2024 · DOI：10.3389/fimmu.2023.1334886 · 被引用次数：13 · 研究领域：Radiomics and Machine Learning in Medical Imaging、Cancer Immunotherapy and Biomarkers、Ferroptosis and cancer prognosis

Background Natural killer (NK) cells are crucial for tumor prognosis; however, their role in non-small-cell lung cancer (NSCLC) remains unclear. The current detection methods for NSCLC are inefficient and costly. Therefore, radiomics represent a promising alternative. Methods We analyzed the radiogenomics datasets to extract clinical, radiological, and transcriptome data. The effect of NK cells on the prognosis of NSCLC was assessed. Tumors were delineated using a 3D Slicer, and features were extracted using pyradiomics. A radiomics model was developed and validated using five-fold cross-validation. A nomogram model was constructed using the selected clinical variables and a radiomic score (RS). The CIBERSORTx database and gene set enrichment analysis were used to explore the correlations of NK cell infiltration and molecular mechanisms. Results Higher infiltration of NK cells was correlated with better overall survival (OS) ( P = 0.002). The radiomic model showed an area under the curve of 0.731, with 0.726 post-validation. The RS differed significantly between high and low infiltration of NK cells ( P < 0.01). The nomogram, using RS and clinical variables, effectively predicted 3-year OS. NK cell infiltration was correlated with the ICOS and BTLA genes ( P < 0.001) and macrophage M0/M2 levels. The key pathways included TNF-α signaling via NF-κB and Wnt/β-catenin signaling. Conclusions Our radiomic model accurately predicted NK cell infiltration in NSCLC. Combi...