基于多参数MRI影像组学联合临床影像特征预测鼻咽癌肿瘤细胞增殖活性  

Prediction of tumor cell proliferation activity in nasopharyngeal carcinoma by nomogram based on multiparametric MRI radiomics combined with clinic-radiological features

作　　者：王卓 刘世莉 张少茹 周云舒 张若弟 陈志强 WANG Zhuo;LIU Shili;ZHANG Shaoru;ZHOU Yunshu;ZHANG Ruodi;CHEN Zhiqiang(Clinical Medicine School of Ningxia Medical University,Yinchuan 750004,China;Department of Radiology,the First Hospital Affiliated to Hainan Medical College,Haikou 570102,China;Department of Radiology,General Hospital of Ningxia Medical University,Yinchuan 750004,China)

机构地区：[1]宁夏医科大学临床医学院,银川750004 [2]海南医学院第一附属医院放射科,海口570102 [3]宁夏医科大学总医院放射科,银川750004

出　　处：《磁共振成像》2022年第11期30-36,41,共8页Chinese Journal of Magnetic Resonance Imaging

基　　金：宁夏回族自治区重点研发计划项目(编号:2019BEG03033);宁夏回族自治区自然科学基金(编号:2022AAC03472)。

摘　　要：目的 探讨基于多参数MRI的影像组学结合临床影像特征的列线图在预测鼻咽癌(nasopharyngeal carcinoma, NPC Ki-67)高表达中的价值。材料与方法 回顾性分析宁夏医科大学总医院2015年12月至2022年5月171例NPC患者的临床及MRI资料。根据Ki-67表达水平不同分为高表达组(n=105)和低表达组(n=66)。用3D-Slicer勾画感兴趣区并用“Pyradiomics”包提取特征。使用单-多因素logistic回归与绝对收缩与选择算法(least absolute shrinkage and selection operator,LASSO)筛选Ki-67高表达的独立危险因子并构建预测模型。采用受试者工作特征(receiver operating characteristic,ROC)曲线下面积(area under the curve, AUC)和DeLong检验评估和比较模型间的预测效能。通过决策曲线分析(decision curve analysis, DCA)观察列线图的临床实用性。结果 多因素logistic回归结果显示爱泼斯坦-巴尔病毒脱氧核糖核酸(Epstein-Barr virus deoxyribo nucleic acid, EBV-DNA)载量≥5000 IU/mL(OR=3.809,P=0.007)、肿瘤明显强化(OR=4.064,P=0.005)是Ki-67高表达的临床预测因子,以此建立临床模型。基于对比增强T1加权成像(contrast enhanced T1-weighted imaging, CE_T1WI)、T1加权成像(T1-weighted imaging, T1WI)、T2加权成像脂肪抑制序列(T2-weighted imaging fat suppression, T2WI_FS)三个序列分别中筛选出7、4、2个与Ki-67高表达显著相关的影像组学特征来计算影像组学评分(radiomics score, Rad-score)。用EBV-DNA载量、肿瘤强化程度、Rad-score三者构建联合模型,并绘制列线图将模型可视化。ROC曲线显示,列线图模型的AUC值均高于临床、影像组学模型(训练集:0.887 vs. 0.701、0.861;验证集:0.860 vs. 0.749、0.814)。训练集和验证集中,列线图模型的AUC值均显著高于临床模型,且差异均具有统计学意义(DeLong检验,P均<0.05)。结论 基于多参数MRI的影像组学结合临床影像特征的列线图模型在放化疗前预测NPC患者Ki-67表达状态方面具有较�Objective: To explore the value of nomogram based on multiparametric MRI radiomics combined with clinic-radiological features in predicting high expression of Ki-67 in nasopharyngeal carcinoma(NPC). Materials and Methods: Retrospectively analyzed the clinical and MRI data of 171 NPC patients from December 2015 to May 2022 in the General Hospital of Ningxia Medical University. According to the expression level of Ki-67, patients were divided into the high Ki-67 group(n=105) and the low Ki-67 group(n=66). We used 3D-Slicer to segment the region of interest and "Pyradiomic" package to extract features. Univariate and multivariate logistic regression and least absolute shrinkage and selection operator regression were performed to select the independent risk factors of high Ki-67 expression and then construct the predictive models.Predictive power among models was assessed and compared by using the area under the receiver operating characteristic(ROC) curve and DeLong test. The clinical utility of the nomogram was demonstrated by the decision curve analysis(DCA). Results: Logistic regression results reported that the markedly enhanced tumor focus(OR=4.064, P=0.005), Epstein-Barr virus deoxyribo nucleic acid(EBV-DNA)≥5000 IU/mL(OR=3.809, P=0.007) were significant clinical predictors of high Ki-67 expression, which can be applied to establish a clinical model. Seven, four,and two radiomics features significantly related to the high expression of Ki-67 from contrast enhanced T1-weighted imaging(T1WI_CE),T1-weighted imaging(TIWI) and T2-weighted imaging fat suppression(T2WI_FS) were selected to construct a radiomics model. EBV-DNA, the degree of enhancement and radiomics score(Rad-score) were used to develop the nomogram model. The ROC curve demonstrated that the AUCs of the nomogram model were higher than those of the clinical or radiomics models(training set: 0.887 vs. 0.701, 0.861;validation set: 0.860 vs.0.749, 0.814). In the training and validation sets, the AUCs of the nomogram model were statistically significant

关 键 词：鼻咽癌 KI-67 影像特征 多参数 磁共振成像 影像组学 列线图 

分 类 号：R445.2[医药卫生—影像医学与核医学] R739.63[医药卫生—诊断学]