The value of predicting axillary lymph node metastasis in patients with breast cancer based on multi-modal magnetic resonance imaging

DANG Yan; WEI Yanyan; ZHU Laimin; SHAO Kai; HU Xibin; WANG Weiwei

Abstract：Objective To assess the value of dynamic contrast-enhanced magnetic resonance imaging(DCE-MRI)and Intra-voxel incoherent motion(IVIM),diffusion kurtosis imaging(DKI)for predicting the risk of axillary lymph node(ALN)metasta-sis in patients with breast cancer.Methods A total of 165 cases of breast cancer confirmed by pathology were collected.Seventy-seven patients had an axillary lymph node metastasis(axillary lymph node,ALN),and 88 patients had no axillary lymph node metastasis.DCE-MRI,IVIM and DKI scanning were performed in all patients.The clinicopathological characteris-tics and conventional MRI features and volume transfer constant(Ktrans),extravascular extracellular space distribute volume per unit tissue volume(Ve)and rate constant(Kep)and true diffusion coefficient(D),perfusion-related diffusion coefficient(D*),perfusion fraction(f)and mean diffusion rate(MD),mean kurtosis value(MK)were analyzed.The diagnostic performance of these models was evaluated by Receiver Operating Characteristic Curve(ROC)analysis.Results The edge of the ALN metas-tasis group was more blurry than that of the non-ALN metastasis group(P=0.032).The ALN metastasis group had higher propor-tion of patients with heterogeneous enhancement and rim enhancement(P=0.012).The D and MD values of the ALN metastasis group were lower than those of the non-ALN metastasis group(P<0.01),showing a low degree of negative correlation(r=-0.441,-0.259).The D*,Ktrans,Kep and MK values of the ALN metastasis group were higher than those of the non-ALN metastasis group(P<0.01).The Ktrans value was moderately positively correlated with ALN metastasis(r=0.621),and the Kep,D * and MK val-ues were low-degree positively correlated with ALN metastasis(r=0.337;r=0.255,r=0.398).The Ktrans value of the DCE-MRI model had the highest diagnostic efficacy,with the area under the ROC curve(AUC)of 0.859.The DCE-MRI model was Ktrans+Kep,with an AUC of 0.868.The IVIM model was D+D *,with an AUC of 0.781.The DKI model was MD+MK,with an AUC of 0.730.The diagnostic efficiency of the DCE-MRI model(AUC=0.868)was higher than that of the IVIM(AUC=0.781)and DKI(AUC=0.730)models,and the differences were statistically significant(Z=2.001,P=0.046;Z=3.156,P=0.002).When DCE-MRI,IVIM and DKI models were combined,the AUC was 0.913,the diagnostic efficacy was higher than that of the single model(Z=2.483～4.838,P<0.05),the sensitivity was 90.9%,the specificity was 81.8%,and the accuracy was 86.1%,which was also higher than that of the single model.Conclusion DCE-MRI,IVIM and DKI models can be used to predict the axillary lymph node metastasis of breast cancer.Among them,DCE-MRI model and its Ktrans value have higher diagnostic efficiencies,and DCE-MRI+IVIM+DKI model is superior to a single model in the differentiation of axillary lymph node properties of breast cancer.