基于MRI、钼靶和病理的列线图预测肿块型乳腺浸润性导管癌前哨淋巴结转移的价值

Value of a nomogram based on MRI,mammography and pathology for predicting sentinel lymph node metastasis of mass-type breast invasive ductal carcinoma

目的探讨基于MRI、钼靶影像特征联合临床病理因素构建的列线图在预测肿块型乳腺浸润性导管癌前哨淋巴结(sentinel lymph node,SLN)转移中的价值。材料与方法回顾性分析经病理证实为浸润性导管癌患者的临床病理及影像资料312例,按3∶1随机分成训练组(234例)与验证组(78例),两组间比较采用χ2检验或Fisher精确检验。在训练组中,SLN阴性组158例,阳性组76例,对两组患者的临床病理因素、MRI、钼靶影像特征进行分析。通过多因素Logistic回归分析筛选出独立预测因子,构建预测SLN转移的列线图模型。使用受试者操作特征(receiver operating characteristic,ROC)曲线、校准曲线、Hosmer-Lemeshow检验拟合优度对模型进行评价。结果临床病理因素、MRI、钼靶影像特征在训练组及验证组间差异无统计学意义(P>0.05)。在训练组中,肿瘤最大径、临床T分期、淋巴结触诊、孕激素受体、人类表皮生长因子受体2、脉管浸润、MRI[肿块形状、乳腺影像报告和数据系统(Breast Imaging Reporting and Data System,BI-RADS)分类、腋窝淋巴结状态]、钼靶(BI-RADS分类、腋窝淋巴结状态)这11个变量在SLN阴性组和阳性组间差异有统计学意义(P<0.05)。通过多因素Logistic回归分析得到,肿瘤最大径、淋巴结触诊、MRI(腋窝淋巴结状态)、钼靶(腋窝淋巴结状态)以及脉管浸润为预测SLN转移的独立危险因素。基于这5个变量构建模型,训练组及验证组ROC曲线下面积分别为0.908和0.897;Hosmer-Lemeshow检验拟合优度P值分别为0.883和0.579(P>0.05)。结论基于MRI及钼靶的术前影像学特征联合临床病理因素构建的列线图模型能较好地预测肿块型浸润性导管癌患者SLN转移情况。

Objective:To explore the value of nomogram based on MRI,mammography imaging features and clinicopathological factors for predicting sentinel lymph node(SLN) metastasis in mass-type breast invasive ductal carcinoma.Materials and Methods:The clinicopathological and imaging data of patients with invasive ductal carcinoma confirmed by pathology were analyzed retrospectively.A total of 312 cases were included,and randomly divided into training group(234 cases) and verification group(78 cases) according to the ratio of 3∶1.χ2test or Fisher exact test were used for comparison between the two groups.In the training group,there were 158 cases with negative SLN and 76 cases with positive SLN.The clinicopathological factors,MRI and mammography imaging features were analyzed in these two groups.Multivariate Logistic regression analysis was used to select independent predictors to build a nomogram model for predicting SLN metastasis.Receiver operating characteristics(ROC) curve,calibration curve and Hosmer-Lemeshow goodness of fit test were used to evaluate the model.Results:There was no significant difference in clinicopathological factors,MRI and mammography imaging features between the training group and the verification group(P>0.05).In the training group,there were statistical significances in 11 variables including greatest tumor diameter,clinical T stage,lymph node palpation,progesterone receptor,human epidermal growth factor receptor 2,lymphovascular invasion,MRI [tumor shape,Breast Imaging Reporting and Data System(BI-RADS) classification,axillary lymph node status],mammography(BI-RADS classification,axillary lymph node status) between negative SLN group and positive SLN group(P<0.05).Multivariate Logistic regression analysis showed that greatest tumor diameter,lymph node palpation,MRI(axillary lymph node status),mammography(axillary lymph node status) and lymphovascular invasion were independent risk factors for predicting SLN metastasis.The Nomogram model was constructed based on these five variables,the area under ROC curve in the training group and verification group was 0.908 and 0.897,respectively;the P values in Hosmer-Lemeshow goodness-of-fit test were 0.883 and 0.579,respectively(P>0.05).Conclusions:The nomogram model based on MRI and mammography imaging features combined with clinicopathological factors can be used to predict the metastasis of SLN in patients with mass-type breast invasive ductal carcinoma.