基于CT征象构建肾透明细胞癌WHO/ISUP分级积分评价系统  

Construction of a scoring system for predicting WHO/ISUP grading of clear cell renal cell carcinoma based on CT features

作　　者：赵才勇 陈超[2] 陈文 严志强 康书朝 崔凤[1] ZHAO Caiyong;CHEN Chao;CHEN Wen;YAN Zhiqiang;KANG Shuchao;CUI Feng(Department of Radiology,Hangzhou Hospital of Traditional Chinese Medicine,Hangzhou 310007,China;不详)

机构地区：[1]杭州市中医院放射科,310007 [2]浙江大学医学院附属邵逸夫医院放射科

出　　处：《浙江医学》2022年第22期2383-2387,共5页Zhejiang Medical Journal

基　　金：杭州市生物医药和健康产业发展扶持科技项目(2021WJCY355)。

摘　　要：目的探讨基于CT征象构建的积分评价系统对术前预测肾透明细胞癌(ccRCC)WHO/国际泌尿病理学会(ISUP)分级的评估价值。方法收集2017年1月至2022年9月杭州市中医院(87例)和浙江大学医学院附属邵逸夫医院(53例)经病理检查证实的ccRCC患者共140例,依据WHO/ISUP分级标准分为低级别组(98例,Ⅰ级13例、Ⅱ级85例)和高级别组(42例,Ⅲ级34例、Ⅳ级8例),并以分层抽样法按照8∶2的比例将患者分为训练集(112例)与验证集(28例)。通过Mann-Whitney U检验、χ2检验、多因素logistic回归分析CT征象筛选组间差异有统计学意义的因素,并进行加权赋分得到积分模型;绘制ROC曲线评价模型预测效能;最后将积分模型分为3个积分区间。结果多因素logistic回归分析显示,形状与边界、坏死及皮髓质期强化程度为预测ccRCC分级的独立危险因素,该模型的AUC为0.851(95%CI:0.762~0.941),灵敏度为0.794,特异度为0.846。积分模型包括形状与边界(分叶型2分或浸润型3分)、大量坏死(3分)及皮髓质期轻中度强化(2分)。积分模型的AUC为0.850(95%CI:0.760~0.940),应用Youden指数确定最佳阈值(3.5),灵敏度为0.765,特异度为0.859。将积分模型分为3个积分区间:0~3分、4~6分、7~8分。随着积分增加,训练集、验证集各积分区间高级别ccRCC的发生率逐渐增高。结论基于CT征象构建的积分评价系统对术前预测ccRCC WHO/ISUP分级具有较高的临床应用价值。Objective To construct a scoring system for predicting the WHO/International Society of Urological Pathology(ISUP)grading of clear cell renal cell carcinoma(ccRCC)based on CT imaging features.Methods One hundred and forty patients diagnosed as ccRCC by pathology from January 2017 to September 2022 in Hangzhou Hospital of Traditional Chinese Medicine and the Sir Run Run Shaw Hospital Zhejiang University School of Medieine were retrospectively analyzed.According to the WHO/ISUP grading system,the patients were divided into low grade group(total 98 cases,including 13 cases of grade Ⅰ and 85 cases of grade Ⅱ)and high grade group(total 42 cases,including 34 cases of grade Ⅲ and 8 cases of grade Ⅳ).Patients were assigned to training set(n=112)and validation set(n=28)in a ratio of 8∶2 using stratified sampling.Mann-Whitney U test,Chi squared test and multivariate binary logistic regression analysis were used to determine the independent predictor of CT imaging features and establish the score model.The receiver operating characteristic(ROC)curve was used to assess the discriminatory power of the models.Results Multivariate binary logistic regression analysis showed that shape and margin,necrosis and enhancement degree in corticomedullary phase were independent predictors for the WHO/ISUP grading of ccRCC.The area under ROC curve(AUC)of the primary predictive model was 0.851(95%CI:0.762-0.941),the sensitivity and specificity were 0.794 and 0.846.Three CT features were included in the score model:lobulated or infiltrated shape and margin(2 points;3 points,respectively),massive necrosis(3 points),and mild to moderate enhancement in corticomedullary phase(2 points).The AUC of the score model was 0.850(95%CI:0.760-0.940).This scoring system presented with a sensitivity of 0.765 and a specificity of 0.859 when using 3.5 points as cut-off value.Three score ranges were also proposed as follows:0-3 points,4-6 points,7-8 points.The number of patients with high grade ccRCC in the three ranges significantly increased with

关 键 词：肾透明细胞癌 病理分级 计算机断层扫描 积分评价系统 

分 类 号：R737.11[医药卫生—肿瘤] R730.44[医药卫生—临床医学]