基于病理和超声图像特征的列线图模型预测乳腺癌腋窝淋巴结转移的临床价值

Clinical value of nomogram model based on pathological and ultrasound image features in predicting axillary lymph node metastasis of breast cancer

目的基于浸润性乳腺癌原发灶的病理和超声图像特征构建列线图模型,探讨其预测腋窝淋巴结转移的临床价值。方法回顾性分析经病理证实的浸润性乳腺癌女性患者369例患者,以7∶3比例随机分为训练集(258例)和验证集(111例)。根据是否发生腋窝淋巴结转移将训练集患者分为转移组(116例)和未转移组(143例),比较两组血清肿瘤标志物、病理和超声图像特征的差异;采用二元Logistic回归分析筛选预测浸润性乳腺癌患者发生腋窝淋巴结转移的独立影响因素,并构建预测腋窝淋巴结转移的列线图模型;绘制受试者工作特征(ROC)曲线和校准曲线分别评估该模型的区分度和校准度。结果训练集中转移组与未转移组原发灶病理类型、组织学分级、血清肿瘤标志物[糖类抗原153(CA153)和癌胚抗原(CEA)]水平,以及超声图像特征(最大径、位置、有无高回声晕环)比较,差异均有统计学意义(均P<0.05)。二元Logistic回归分析显示,原发灶位置(内下象限)、高回声晕环、最大径、组织学分级均为预测浸润性乳腺癌患者发生腋窝淋巴结转移的独立影响因素(OR=0.064、13.278、1.049、9.277,均P<0.05)。以此构建的列线图模型预测训练集和验证集浸润性乳腺癌患者发生腋窝淋巴结转移的ROC曲线下面积分别为0.845、0.823,校准曲线与理想曲线高度吻合,其区分度和校准度均高。结论基于浸润性乳腺癌原发灶的病理和超声图像特征构建的列线图模型可预测腋窝淋巴结转移风险,可为临床精准诊疗提供参考。

Objective To construct a nomogram model based on the pathological and ultrasound image features of the primary lesions of invasive breast cancer,and to investigate its clinical value in predicting axillary lymph node metastasis.Methods A total of 369 female patients with pathologically confirmed invasive breast cancer were retrospectively analyzed,and were randomly divided into the training set(n=258)and verification set(n=111)at a ratio of 7∶3.The patients in the training set were divided into metastatic group(n=116)and non-metastatic group(n=143)according to whether axillary lymph node metastasis occurred.The differences of serum tumor marker,pathological and ultrasound image features between the two groups were compared.Binary Logistic regression analysis was used to screen the independent influencing factors for predicting axillary lymph node metastasis in patients with invasive breast cancer,and a nomogram model for predicting axillary lymph node metastasis was constructed.Receiver operating characteristic(ROC)curve and calibration curve were drawn to evaluate the discrimination and calibration,respectively.Results The differences in pathological type,histological grade,serum tumor marker(carbohydrate antigen 153 and carcinoembryonic antigen)levels,ultrasound image features(maximum diameter,location and presence or absence of hyperechoic halo)between metastatic group and non-metastatic group in the training set were statistically significant(all P<0.05).Binary Logistic regression analysis showed that the location of the primary lesions(inner lower quadrant),hyperechoic halo,maximum diameter and histological grade were independent influencing factors in predicting axillary lymph node metastasis in patients with invasive breast cancer(OR=0.064,13.278,1.049,9.277,all P<0.05).The area under ROC curve of the constructed nomogram model for predicting axillary lymph node metastasis in the training set and verfication set were 0.845 and 0.823,respectively.The calibration curve was highly consistent with the ideal curve,and the nomogram model had good discrimination and calibration.Conclusion The nomogram model based on the pathological and ultrasound image features of the primary invasive breast cancer can effectively predict the risk of axillary lymph node metastasis,and provide reference for accurate diagnosis and treatment.