扩散峰度成像联合动态增强磁共振成像对乳腺癌不同分子分型的诊断价值

Prediction of the Expression of Molecular Subtypes in Breast Cancer Based on Diffusion Kurtosis Imaging and Dynamic Contrast Enhancement Imaging

目的探讨扩散峰度成像(DKI)联合动态增强磁共振成像(DCE-MRI)对乳腺癌分子分型的诊断价值。方法选取经病理证实的乳腺癌患者148例,所有病例均行DKI及DCE-MRI检查。分析两组间的临床病例资料和平均扩散峰度值(MK)、平均扩散率(MD)、容量转移常数(K^(trans))、血管外细胞外间隙容积比(V_(e))和速率常数(K_(ep)),绘制受试者工作特征曲线比较其诊断效能。结果Luminal A型和Luminal B型的MD值小于HER-2阳性型、三阴型(P<0.05);Luminal B型的MK值大于HER-2阳性型(P=0.021);HER-2阳性型的K^(trans)和V_(e)值均大于其他分子分型(P<0.05);三阴型组V_(e)值均小于其他分子分型(P<0.05)。当MD≤1.96×10^(-3)mm^(2)/s、MK≥0.70/min、K^(trans)≤0.36(min^(-1))、K_(ep)≤0.41(min^(-1))时,对鉴别Luminal A/B型与非Luminal型有价值,联合后受试者工作特征曲线下面积(AUC)为0.811,诊断效能高于单一参数(Z=2.040~3.352,P=0.001~0.041)。当MD≥2.28×10^(-3)mm^(2)/s、MK≤0.66/min、K^(trans)≥0.36(min^(-1))、V_(e)≥0.49时,对鉴别HER-2阳性型与非HER-2阳性型有价值,联合后AUC为0.872,诊断效能高于单一参数(Z=2.038~3.397,P=0.001~0.048)。当MD≥1.96×10^(-3)mm^(2)/s、K_(ep)≥0.39(min^(-1))、V_(e)≤0.59时,对鉴别三阴型与非三阴型有价值,V_(e)的诊断效能高于MD及K_(ep)值(Z=2.032,P=0.042;Z=2.562,P=0.01);三者联合后AUC为0.834,与单一参数差异有统计学意义(Z=2.051~3.852,P=0.001~0.042)。结论DKI联合DCE-MRI可用于乳腺癌分子分型的预测,有利于患者的个体化治疗。

Objective To investigate predictive value of diffusion kurtosis imaging and dynamic contrast-enhanced magnetic resonance imaging for the expression of molecular subtypes in breast cancer.Methods A total of 148 cases of breast cancer patients confirmed by pathology were collected.DKI and DCE-MRI scanning were performed in all patients.The mean diffusion rate(MD),mean kurtosis value(MK),volume transfer constant(K^(trans)),extravascular extracellular space distribute volume per unit tissue volume(V_e)and rate constant(K_(ep))were analyzed.The ROC was plotted to compare the diagnostic efficiency.Results MD values of Luminal A and Luminal B were lower than HER-2 positive and triple negative(P<0.05).MK value of Luminal B was greater than HER-2 positive(P=0.021).K^(trans)and V_(e)values of HER-2 positive were higher than other molecular types(P<0.05).V_(e)value of non-triple subtype was lower than other molecular types(P<0.05).When MD≤1.96×10^(-3)mm^(2)/s,MK≥0.70,K^(trans)≤0.36 min^(-1),K_(ep)≤0.41min^(-1),it was valuable to distinguish between the LuminalA/B and non-Luminal,and The combined model(MD+MK+K^(trans)+K_(ep))showed significantly higher AUC than single parameter(Z=2.040-3.352,P=0.001-0.041).When MD≥2.28×10^(-3)mm^(2)/s,MK≤0.66/min,K^(trans)≥0.36 min^(-1),V_e≥0.49,it was valuable to identify HER-2 positive and other molecular subtypes,and the combined model(MD+MK+K^(trans)+V_e)showed significantly higher AUC than single parameter(Z=2.038-3.397,P=0.001-0.048).When MD≥1.96×10^(-3)mm^(2)/s,K_(ep)≥0.39 min^(-1),V_e≤0.59,it was valuable for identifying triple and non-triple subtypes.The diagnostic efficacy of V_(e)is higher than MD and K_(ep)values(Z=2.032,P=0.042;Z=2.562,P=0.01).The combined model(MD+K_(ep)+V_e)showed better diagnostic performance than a single parameter(Z=2.051-3.852,P=0.001-0.042).Conclusion DKI combined with DCE-MRI can be used to predict the expression of molecular subtypes in breast cancer,which is beneficial to individualized treatment of patients.