三阴性乳腺癌全数字化乳腺X线摄影(FFDM)影像特征与腋窝淋巴结转移相关性的logistic回归分析

Logistic regression analysis of the correlation between FFDM imaging features of triple-negative breastcancer and axillary lymph node metastasis

目的:探讨三阴性乳腺癌(TNBC)患者腋窝淋巴结转移(ALNM)的危险因素,并基于全视野数字乳腺X线摄影(FFDM)影像特征构建预测ALNM模型。方法:回顾分析227例经病理学检查证实为TNBC患者的临床病理资料。根据病理结果分为腋窝淋巴结转移组(90例)和非转移组(137例)。所有病例术前均行常规乳腺FFDM检查。比较两组TNBC患者临床病理特征(年龄、绝经状态、组织学分级、Ki-67表达、病理类型)和肿瘤FFDM征象(腺体类型,肿瘤最大径、密度、形态、边缘、有无可疑钙化,FFDM显示有无ALNM及有无邻近皮肤增厚或凹陷)的差异,采用单因素分析及二元logistic逐步回归分析筛选TNBC淋巴结转移的独立危险因素,并构建联合预测模型。采用受试者工作特征(ROC)曲线评价模型预测腋窝淋巴结转移的价值。结果:单因素分析显示两组TNBC患者绝经状态,肿瘤的最大径、密度、形态、边缘、有无可疑钙化,FFDM显示有无ALNM、有无皮肤增厚或凹陷差异具有统计学意义,与腋窝淋巴结转移具有相关性(P<0.05);logistic回归分析显示肿块边缘呈星芒状(OR=12.362,P<0.001)、肿块伴可疑钙化(OR=4.249,P=0.001)、FFDM显示ALNM(OR=34.428,P<0.001)是TNBC腋窝淋巴结转移的独立危险因素。建立预测淋巴结转移的联合诊断模型为Logit(P)=-5.349+2.515X肿块边缘呈星芒状+1.447X肿块伴可疑钙化+3.539X FFDM显示ALNM。联合预测模型的曲线下面积(AUC)为0.880,符合率81.5%,敏感度86.7%,特异度73.7%。结论:FFDM影像显示肿块边缘呈星芒状、肿块伴可疑恶性钙化、FFDM显示ALNM是TNBC患者腋窝淋巴结转移的危险因素,据此构建的联合预测模型对TNBC淋巴结转移具有良好的预测价值。

Objective:To investigate the risk factors of axillary lymph node metastasis(ALNM)in patients with triple-negative breast cancer(TNBC),and construct a model to predict ALNM based on full-field digital mammography(FFDM)image features.Methods:The clinicopathologic data of 227 patients with TNBC confirmed by pathological examination were retrospectively analyzed.The patients were divided into axillary lymph node metastasis group(90 cases)and non-metastatic group(137 cases).All cases underwent routine breast FFDM examination before operation.The clinicopathologic features(age,menopausal status,histological grade,Ki-67 expression,pathological type)and tumor FFDM signs(glandular type,tumor maximum diameter,density,shape,margin,suspicious calcification,FFDM indicated ALNM and adjacent skin thickening or depression)were compared between the two groups of TNBC patients.Univariate analysis and binary logistic stepwise regression analysis were used to identify independent risk factors for TNBC lymph node metastasis,and a combined prediction model was constructed.Receiver operating characteristic(ROC)curve was used to evaluate the value of the model in predicting axillary lymph node metastasis.Results:Univariate analysis showed that there were significant differences in menopausal status,maximum diameter,density,shape,margin and suspicious calcification of tumors,ALNM and skin thickening or indentation indicated by FFDM between two groups of TNBC patients,which were correlated with axillary lymph node metastasis(P<0.05).Logistic regression analysis showed that tumor with spiculate margin(OR=12.362,P<0.001),tumor with suspicious calcification(OR=4.249,P=0.001),and FFDM indicated ALNM(OR=34.428,P<0.001)were independent risk factor for axillary lymph node metastasis in TNBC.The combined diagnostic model for predicting lymph node metastasis was established as Logit(P)=-5.349+2.515×tumor with spiculate margin+1.447×tumor with suspicious calcification+3.539×FFDM indicated ALNM.The area under the curve(AUC)of the combined prediction model was 0.880,with a coincidence rate of 81.5%,sensitivity of 86.7%,and specificity of 73.7%.Conclusion:FFDM imaging showed that tumor with spiculate margin,tumor with suspicious calcification,and FFDM indicated ALNM was a risk factor for axillary lymph node metastasis in patients with TNBC.The combined prediction model constructed based on these factors has good predictive value for TNBC lymph node metastasis.