基于超声特征的列线图模型鉴别诊断膀胱隆起样病变良恶性的价值

Nomogram Based on Ultrasonographic Features in Differentiating Benign from Malignant Bladder Neoplasms

目的构建基于超声特征的列线图模型,探讨其鉴别诊断良、恶性膀胱隆起样病变的价值。资料与方法回顾性分析2016年1月—2022年1月四川省人民医院经手术病理证实的膀胱隆起样病变538例(良性84例,恶性454例)的超声资料,对膀胱病变超声特征(病灶部位、数目、最大径线、回声、形态、基底、钙化、彩色多普勒血流显像信号)及患者简要临床指标(性别、年龄、泌尿系恶性肿瘤史、肉眼血尿)行Logistic单因素及多因素回归分析,筛选出独立预测因子,并构建预测模型。通过Bootstrap重抽样进行内部验证。绘制受试者工作特征曲线、校正曲线、临床决策曲线评估模型。结果单因素及多因素Logistic回归分析结果显示,性别(OR=1.822,P=0.038)、年龄(OR=1.044,P=0.000)、病灶部位(OR=0.359,P=0.000)、血流信号(OR=2.052,P=0.007)是预测恶性膀胱隆起样病变的独立因素,基于单因素结果构建的列线图预测模型的曲线下面积为0.780,敏感度为72.91%,特异度为71.43%,准确度为72.68%。校正曲线显示模型的一致性较好。临床决策曲线显示临床净获益良好。结论基于超声特征和简要临床指标构建的列线图模型鉴别诊断良、恶性膀胱隆起样病变具有较高的准确度和潜在的临床应用价值。

Purpose To construct a nomogram model based on ultrasonographic features and to evaluate its value in differentiating benign from malignant bladder neoplasms.Materials and Methods A total of 538 consecutive bladder neoplasm patients(including 84 benign cases and 454 malignant cases)confirmed by surgery or biopsy pathology from January 2016 to January 2022 were retrospectively enrolled,the ultrasonographic features(including lesion number,location,maximum diameter,echogenicity,morphology,basement,calcification,color Doppler flow imaging signal)and brief clinical data(gender,age,urinary tract malignant tumors history and gross haematuria)were all collected for univariate and multivariate Logistic regression analysis.Independent predictors for malignant bladder neoplasm were screened and nomogram model based on univariate Logistic regression analysis was constructed.Internal validation was performed by Bootstrap resampling.Meanwhile,the receiver operating characteristic curve,calibration curve and decision curve were drawn.Results Univariate and multivariate Logistic regression analysis showed patient gender(OR=1.822,P=0.038),age(OR=1.044,P=0.000),lesion location(OR=0.359,P=0.000)and color Doppler flow imaging signal(OR=2.052,P=0.007)were independent factors in predicting the malignancy of bladder lesions.Area under the curve of the nomogram prediction model based on univariate Logistic regression analysis were 0.780,the sensitivity,specificity and accuracy of the prediction model were 72.91%,71.43% and 72.68%,respectively.The calibration curve and decision curve showed good consistency and clinical practicability of the model.Conclusion The nomogram model based on ultrasonographic features and simple clinical characteristics shows good predictive accuracy in differentiating bladder neoplasms and has potential clinical application value.