乳腺浸润性导管癌不同分子亚型MRI特征

目的探讨乳腺浸润性导管癌分子亚型与MRI表现的关系。方法根据免疫组化结果,将经病理证实的61例乳腺浸润性导管癌患者分为Luminal A型(n=18)、Luminal B型(n=24)、HER-2过表达型(n=8)和三阴型(n=11)4种分子亚型。回顾性分析4种分子亚型乳腺癌的MRI特征,包括病灶最大径、形态、边缘、边界、强化方式、病灶分布、时间-信号强度曲线(time-signal intensity curve,TIC)、表观扩散系数(ADC)值、病理分级及淋巴结转移情况。采用卡方检验、Fisher确切概率法及单因素方差分析得出有统计学差异的MRI特征,并根据这些特征建立基于支持向量机(support vector machine,SVM)分类模型,计算ROC下面积(AUC),得出诊断效能。结果61例乳腺浸润性导管癌的4种分子亚型间,病灶最大径(<2cm、≥2cm)和淋巴结转移差异均无统计学意义(P均>0.05),病理分级(I、II、III级)差异有统计学意义(P<0.01)。4种分子亚型间5个MRI征象(形态、边缘、边界、强化方式、病灶分布)差异有统计学意义(P均<0.01),TIC类型及ADC值无统计学差异(P均>0.05)。基于SVM分类模型,得出四分类诊断效能AUC为0.817。结论不同分子亚型乳腺浸润性导管癌MRI表现有其特征性,利用SVM分类模型可以有效区分分子亚型。

MRI features of different molecular subtypes of breast invasive ductal carcinoma

Objective To explore the relationship between MRI findings and different molecular subtypes of breast invasive ductal carcinoma.Methods According to immunohistochemical results,61patients with breast invasive ductal carcinoma were divided into 4 molecular subtypes,i.e.Luminal type A(n=18),Luminal type B(n=24),HER-2expresion type(n=8)and triple-negative type(n=27).The MRI features of the 4 molecular subtypes breast carcinoma were retrospectively analyzed,including the tumor sizes,shape,edge,boundary,enhancement pattern,tumor distribution,type of time-signal intensity curve(TIC),ADC value,histopathological grade and lymph node metastasis.Chi-square test,Fisher exact probability and single factor analysis of variance were used to obtain MRI features with statistical differences,and SVM classification model was established according to these features to calculate the area under ROC(AUC)and to obtain the diagnostic efficiency.Results There was breast invasive ductal carcinoma in the tumor sizes(<2 cm,≥2 cm)nor lymph node metastasis(with or without)among them(both P>0.05),but in the histopathological grade(grade I,II,III),there was a statistically significant difference(P<0.01).Totally,5 MRI fatures(tumor sizes,shape,edge,boundary,enhancement pattern,tumor distribution)had statistically significant differences among 4 subtypes(all P<0.01),while the TIC type and ADC value had no statistically significant difference(both P>0.05).Based on SVM classification model,the diagnostic efficiency of four classification AUC was 0.817.Conclusion The MRI features of different molecular subtypes of breast invasive ductal carcinoma have certain characteristics.Molecular subtypes can be distinguished effectively by SVM classification model.