Logistic regression of ultrasound signs in differentiating female breast metaplastic carcinoma from invasive ductal car-cinoma

FU Fengmei; FANG Fengkai; FU Meichui

Abstract：Objective To investigate the value of ultrasonographic features of breast lesions in the differential diagnosis of metaplastic carcinoma and invasive ductal carcinoma of the breast.Methods The ultrasound images of 117 patients with breast tumors confirmed by operation and pathology in Hainan West Central Hospital were analyzed retrospectively.According to the pathological results,the patients were divided into metaplastic carcinoma group(n=36)and invasive ductal carcinoma group(n=81).The ultrasound signs were analyzed by univariate analysis,and the statistically significant parameters of ultrasound signs were set as independent variables for Logistic regression analysis.Results Univariate analysis showed that there were sig-nificant differences in tumor size,echo type,posterior echo and growth orientation in the differential diagnosis of breast meta-plastic carcinoma and invasive ductal carcinoma(P<0.05).Tumor size,echo type,rear echo and growth orientation were inde-pendent factors to distinguish the two.The regression equation was Logistic(Y)=-2.209+1.335X1+2.186X2-1.149X3-1.734X4(X1:tumor size;X2:echo type;X3:rear echo).X4:orientation),the area under the ROC curve was 0.901.The sensitivity,specificity and accuracy of the model for distinguishing metaplastic breast cancer from invasive ductal carcinoma were 61.1%,96.3%and 87.2%,respectively.Conclusion Multivariate analysis is effective in the differential diagnosis of breast metaplas-tic carcinoma and invasive ductal carcinoma.Ultrasound can be used as a reliable index to comprehensively judge the size,echo type,posterior echo and orientation of breast metaplastic carcinoma and invasive ductal carcinoma.