定量参数联合膀胱影像报告和数据系统诊断肌层浸润性膀胱癌的临床价值

The Application Value of Quantitative Parameters Combined with Vesical Imaging-Report and Data System in Distinguishing Muscle-Invasive Bladder Cancer

目的探讨定量参数联合膀胱影像报告与数据系统(VI-RADS)诊断膀胱癌肌层浸润的应用价值。方法回顾性分析2015年5月至2021年5月本院收治的74例膀胱癌患者资料。由两名不同年资的影像医师(2年和15年)对多参数MRI图像分别进行VI-RADS评分,并测量和记录定量参数,包括肿瘤与膀胱壁接触长度(TCL)、肿瘤最大直径(D_(max))、TCL/D_(max),采用组内相关系数(ICC)评估两名影像医师测量定量参数的一致性;利用LASSO回归筛选出诊断肌层浸润性膀胱癌(MIBC)最具有显著意义的参数;采用受试者工作特征(ROC)曲线评价单独使用VI-RADS评分、单独使用定量参数以及VI-RADS评分联合定量参数的Logistic回归模型对于MIBC的诊断效能;采用Z检验比较三者间诊断效能的优劣。结果两名影像医师对测量定量参数的一致性良好(ICC 0.806~0.960,P<0.05);LASSO回归分析筛选出VI-RADS评分及TCL/D_(max)是诊断膀胱癌肌肉浸润最具鉴别意义的参数;VI-RADS评分诊断膀胱癌肌层浸润的曲线下面积(AUC)为0.929(95%CI:0.845~0.976),使用阈值为VI-RADS>3分时,灵敏度和特异度分别为70.6%、96.5%。TCL/D_(max)的AUC为0.835(95%CI:0.730~0.911),以TCL/D_(max)>1.027为阈值时,灵敏度和特异度分别为94.1%、71.9%。VI-RADS联合TCL/D_(max)的AUC为0.979(95%CI:0.913~0.999),灵敏度和特异度分别为100.0%、93.0%,诊断效能高于单独使用VI-RADS评分或者TCL/D_(max),结果具有统计学意义(P<0.05)。VI-RADS评分为3分的25例患者中,20例经病理证实为非肌层浸润性膀胱癌(NMIBC),VI-RADS评分诊断准确率为80.0%;VI-RADS评分联合TCL/D_(max)诊断MIBC 5例,NMIBC 18例,诊断准确率提高至92.0%。结论定量参数和VI-RADS评分评估膀胱癌是否存在肌层浸润具有良好的诊断效能,对于VI-RADS评分为3分的肿瘤联合应用可以进一步提高诊断准确率。

Objective To evaluate the diagnostic performance of quantitative MRI parameters combined with Vesical Imaging Reporting and Data System(VI-RADS)in predicting muscle invasive bladder cancer(MIBC).Methods Seventy-four patients with pathologically proven bladder cancer in our hospital between May2015 and May 2021 were retrospectively enrolled and analyzed.All patients were performed multi-parametric magnetic resonance imaging(mp-MRI)before the trans-urethral resection of bladder tumor(TURBT)or radical cystectomy.Two radiologists(with 2 and 15 years’experience)scored all mp-MRI examinations according to the VI-RADS criteria.The quantitative parameters of mp-MRI,including the tumor contact length(TCL),maximum tumor diameter(D_(max))and TCL/D_(max) were measured and calculated independently.Interobserver agreement was assessed by intraclass correlation coefficient(ICC)analysis.LASSO regression was used to select the quantitative parameters that were significant for distinguishing MIBC.ROC curve was used to evaluate the diagnostic efficacy of Logistic regression model using VIRADS score alone,quantitative parameters alone and combined quantitative parameters of VIRADS score for MIBC,and Z test was used to compare the diagnostic efficacy among the three.Results Interobserver agreement was almost perfect in the measurement of quantitative parameters(ICC:0.806~0.960,P<0.05).LASSO regression was used to extract VI-RADS and TCL/D_(max) which were the most significant parameters for predicting MIBC.The cut-off value of VI-RADS was 3,which yielded an area under the curve(AUC)of 0.929(95%CI:0.845~0.976),with the sensitivity and specificity of 70.6%and 96.5%respectively.The cut-off value for TCL/D_(max) was 1.027,which yielded an area under the curve(AUC)of 0.835(95%CI:0.730~0.911),with the sensitivity and specificity of 94.1%and 71.9%respectively.The combined cut-off value of VI-RADS and TCL/D_(max) yielded an area under the curve(AUC)was 0.979(95%CI:0.913~0.999),which were superior to the individual values(P<0.05),with the sensitivity and specificity of 100.0%and 93.0%respectively.In patients with VI-RADS of 3,the diagnostic accuracy was improved from 80.0%(20/25)to 92.0%(23/25)when combined VI-RADS and TCL/D_(max).Conclusion Quantitative parameters and VI-RADS score have a good diagnostic performance in distinguishing MIBC.The combined use could improve the diagnostic accuracy with a VI-RADS score of 3.