乳腺X线摄影、MRI及病理联合诊断乳腺癌分子分型

目的 观察乳腺X线摄影(MG)、MRI及病理联合诊断乳腺癌分子分型的价值。方法 收集600例乳腺癌女性患者,包括Luminal A型147例,Luminal B型277例,人表皮生长因子受体2(HER-2)过表达型65例及三阴型111例;比较不同分子分型乳腺癌的MG、MRI及病理特征,行logistic回归分析,观察不同分子分型的独立影响因素;构建病理模型(模型1)、MG+病理模型(模型2)、MRI+病理模型(模型3)及MG+MRI+病理模型(模型4),评估其诊断价值,并以SHAP分析评估最佳诊断模型中各参数的贡献价值。结果 组织学低分级浸润性导管癌、无腋窝淋巴结转移、MG假阴性、无恶性钙化及“流入型”时间-信号强度曲线(TIC)为Luminal A型乳腺癌的独立影响因素(P均<0.05),腋窝淋巴结转移、恶性钙化及毛刺征为Luminal B型的独立影响因素(P均<0.05),组织学高分级浸润性导管癌、恶性钙化、毛刺征、非肿块样强化(NME)及较高表观弥散系数为HER-2过表达型的独立影响因素(P均<0.05),组织学高分级浸润性导管癌、无恶性钙化及无毛刺征为三阴型的...

Combination of mammography, MRI and pathology for diagnosis of molecular subtypes of breast cancer

Objective To observe the value of combination of mammography (MG), MRI and pathology for diagnosis of molecular subtypes of breast cancer. Methods Totally 600 female patients with breast cancer were enrolled, including 147 cases of Luminal A type, 277 cases of Luminal B type, 65 cases of human epidermal growth factor receptor 2 (HER-2) enriched type and 111 cases of triple negative type. MG, MRI and pathological features of different molecular subtypes of breast cancer were compared, and logistic regression analysis was performed to explore the independent impact factors of different molecular subtypes of breast cancer. Then pathology model (model 1), MG+pathology model (model 2), MRI+pathology model (model 3) and MG+MRI+pathology model (model 4) were constructed, and the diagnostic values were observed. SHAP was used to analyze the contribution value of each parameter in the best diagnostic model. Results Low histological grade invasive ductal carcinoma, no axillary lymph node metastasis, false negative MG, no malignant calcification and "progressive" time-signal intensity curve (TIC) were independent impact factors of Luminal A breast cancer (all P<0.05). Axillary lymph node metastasis, malignant calcification and spicule sign were the independent impact factors of Luminal B breast cancer (all P<0.05). High histological grade invasive ductal carcinoma, malignant calcification, spicule sign, non mass enhancement (NME) and high apparent diffusion coefficient were independent impact factors of HER-2 enriched breast cancer (all P<0.05). High histological grade invasive ductal carcinoma, no malignant calcification and no spicule sign were independent impact factors of triple negative breast cancer (all P<0.05). Model 2 was the best model for diagnosing Luminal A and HER-2 enriched breast cancer (the area under the curve=0.663, 0.621), while model 4 was the best model for diagnosing Luminal B and triple negative breast cancer (AUC=0.649, 0.642). Malignant calcification, axillary lymph node metastasis and spicule sign were of great value for diagnosis molecular subtypes of breast cancer. Conclusion The combined model of MG, MRI and pathology had certain value for diagnosing molecular subtypes of breast cancer. Malignant calcification, axillary lymph node metastasis and spicule sign were of high diagnostic value.