自动乳腺全容积成像联合声辐射力脉冲成像的列线图模型鉴别诊断乳腺导管原位癌与浸润癌的价值

Value of nomogram model constructed based on automated breast volume scanner combined with acoustic radiation force impulsed imaging for differentiating breast ductal carcinoma in situ from invasive carcinoma

目的探讨基于自动乳腺全容积成像(ABVS)联合声辐射力脉冲成像(ARFI)构建的列线图模型在乳腺导管原位癌(DCIS)与浸润性导管癌(IDC)中的鉴别诊断价值。方法回顾性分析我院经手术病理证实为DCIS或IDC的150例患者(150个病灶)的病历资料,包括DCIS组39例(39个)和IDC组111例(111个),比较两组临床及影像学表现的差异;采用多因素Logistic回归分析鉴别DCIS与IDC的独立影响因素。建立各参数鉴别DCIS与IDC的列线图模型;绘制受试者工作特征(ROC)曲线分析该模型的诊断效能;3折交叉验证评估该模型的稳定性及泛化能力;Hosmer-Lemeshow检验并绘制校准曲线评价该模型的拟合优度及校准度;决策曲线分析该模型的临床获益。结果多因素Logistic回归分析结果显示,纵横比、微分叶、冠状面特征、声触诊组织定量(VTQ)值是鉴别DCIS与IDC的独立影响因素(均P<0.05)。基于上述因素构建的列线图模型鉴别DCIS与IDC的ROC曲线下面积(AUC)为0.787[95%可信区间(CI):0.703~0.871],灵敏度为82.9%,特异度为64.1%。3折交叉验证显示该模型的AUC分别为0.774(95%CI:0.621~0.926)、0.780(95%CI:0.616~0.943)、0.749(95%CI:0.604~0.894),灵敏度分别为78.6%、100%、52.9%,特异度分别为77.8%、47.6%、87.9%,平均AUC为0.768。Hosmer-Lemeshow检验显示该模型具有良好的拟合优度;校准曲线显示该模型预测概率与病理结果一致性较好(C-index:0.787);决策曲线分析表明该模型具有临床使用价值。结论基于ABVS联合ARFI构建的列线图模型可用于术前鉴别DCIS与IDC,具有良好的临床应用价值。

Objective To investigate the value of nomogram model constructed based on automated breast volume scanner(ABVS)combined with acoustic radiation force impulse(ARFI)imaging in the differential diagnosis of breast ductal carcinoma in situ(DCIS)and invasive ductal carcinoma(IDC).Methods The medical records of 150 patients(150 lesions)with DCIS or IDC confirmed by surgery and pathology in our hospital were retrospectively analyzed,including 39 patients(39lesions)in DCIS group and 111 patients(111 lesions)in IDC group.The clinical and imaging data of the two groups were analyzed and compared.Multi-factor Logistic regression analysis was used to identify the independent influencing factors of DCIS and IDC.A nomogram model was established for each parameter to differentiate DCIS from IDC,and the diagnostic efficacy of the model was analyzed by receiver operating characteristic(ROC)curve,the stability and generalization ability of the model was assessed by 3-fold cross-validation,the goodness of fit and calibration of the model were evaluate by Hosmer-Lemeshow test and calibration curve,and the clinical benefit of the model was analyzed by decision curves.Results The results of multi-factor Logistic regression analysis showed that the aspect ratio of nodules,differential lobes,coronal features,and virtual touch tissue quantification(VTQ)values were independent factors in differentiating DCIS from IDC(all P<0.05).The area under curve(AUC)of the nomogram model constructed based on the above factors to differentiating DCIS from IDC was 0.787[95%confidence interval(CI):0.703~0.871],with a sensitivity of 82.9%and specificity of 64.1%.3-fold cross-validation showed that the AUCs of the model were 0.774(95%CI:0.621~0.926),0.780(95%CI:0.616~0.943),and 0.749(95%CI:0.604-0.894),with sensitivities of 78.6%,100%,52.9%,and specificities of 77.8%,47.6%,87.9%,respectively,with a mean AUC of 0.768.The Hosmer-Lemeshow test shows that the model has good goodness of fit.The calibration curve showed that the prediction probability of the model was in good agreement with the pathological results(C-index:0.787),and the decision curve analysis showed that the model was of clinical use.Conclusion The nomogram model constructed based on ABVS combined with ARFI can be used to differentiating DCIS from IDC preoperatively,which has good clinical application value.