Table_3_Nomogram Based on CT Radiomics Features Combined With Clinical Factors to Predict Ki-67 Expression in Hepatocellular Carcinoma.docx

ObjectivesThe study developed and validated a radiomics nomogram based on a combination of computed tomography (CT) radiomics signature and clinical factors and explored the ability of radiomics for individualized prediction of Ki-67 expression in hepatocellular carcinoma (HCC).
MethodsFirst-order, second-order, and high-order radiomics features were extracted from preoperative enhanced CT images of 172 HCC patients, and the radiomics features with predictive value for high Ki-67 expression were extracted to construct the radiomic signature prediction model. Based on the training group, the radiomics nomogram was constructed based on a combination of radiomic signature and clinical factors that showed an independent association with Ki-67 expression. The area under the receiver operating characteristic curve (AUC), calibration curve, and decision curve analysis (DCA) were used to verify the performance of the nomogram.
ResultsSixteen higher-order radiomic features that were associated with Ki-67 expression were used to construct the radiomics signature (AUC: training group, 0.854; validation group, 0.744). In multivariate logistic regression, alfa-fetoprotein (AFP) and Edmondson grades were identified as independent predictors of Ki-67 expression. Thus, the radiomics signature was combined with AFP and Edmondson grades to construct the radiomics nomogram (AUC: training group, 0.884; validation group, 0.819). The calibration curve and DCA showed good clinical application of the nomogram.
ConclusionThe radiomics nomogram developed in this study based on the high-order features of CT images can accurately predict high Ki-67 expression and provide individualized guidance for the treatment and clinical monitoring of HCC patients.

研究目的：本研究构建并验证了一种联合计算机断层扫描（computed tomography, CT）放射组学特征与临床因素的放射组学列线图，并探讨了放射组学特征在肝细胞癌（hepatocellular carcinoma, HCC）Ki-67表达个体化预测中的应用价值。
研究方法：本研究从172例肝细胞癌患者的术前增强CT影像中提取一阶、二阶及高阶放射组学特征，筛选出对高Ki-67表达具有预测价值的放射组学特征以构建放射组学特征预测模型。基于训练队列，结合与Ki-67表达存在独立关联的放射组学特征与临床因素构建放射组学列线图。采用受试者工作特征曲线下面积（area under the receiver operating characteristic curve, AUC）、校准曲线及决策曲线分析（decision curve analysis, DCA）验证该列线图的性能。
研究结果：本研究筛选出16个与Ki-67表达相关的高阶放射组学特征以构建放射组学特征模型，其受试者工作特征曲线下面积在训练队列中为0.854，验证队列中为0.744。多因素logistic回归分析显示，甲胎蛋白（alpha-fetoprotein, AFP）与埃德蒙森分级（Edmondson grades）为Ki-67表达的独立预测因子。因此，将放射组学特征与AFP、埃德蒙森分级结合构建放射组学列线图，其AUC在训练队列中为0.884，验证队列中为0.819。校准曲线与决策曲线分析结果表明，该列线图具备良好的临床应用价值。
研究结论：本研究基于CT影像高阶特征构建的放射组学列线图可精准预测肝细胞癌患者的高Ki-67表达状态，可为肝细胞癌患者的治疗决策与临床监测提供个体化指导方案。