三阴性乳腺癌与非三阴性乳腺癌mp-MRI征象的Logistic回归分析模型的建立及其预测价值

Establishment of Logistic Regression Analysis Model for mp-MRI Signs of Triple-Negative Breast Cancer and Non-Triple-Negative Breast Cancer and its Predictive Value

目的探究基于多参数MRI乳腺动态增强特征Logistic回归分析模型鉴别三阴性乳腺癌与非三阴性乳腺癌的临床价值。方法搜集2014年1月至2020年6月行乳腺动态增强MRI检查,病理证实为乳腺癌的194例病例纳入研究,并根据病理结果分为68例三阴性组和126例非三阴性组。比较分析两组间多参数MRI征象,用秩和检验进行单因素分析,将有统计学意义的指标作为主要风险预测因子,用多因素Logistic回归建立风险预测模型,运用受试者工作特征曲线(ROC)分析Logistic回归模型的诊断效能。结果单因素分析显示,肿瘤更大、边界不清晰、乳头受累、伴有同侧淋巴结单发转移以及乳腺影像报告和数据系统(BI-RADS)分级更高时等因素具有统计学意义(P<0.05),多因素分析表明BI-RADS分类、乳腺肿块边界不清、伴同侧淋巴结单发转移、有乳头受累是鉴别三阴性乳腺癌与非三阴性乳腺癌的独立预测指标(P<0.05),运用独立预测因子绘制的曲线下面积(AUC)为0.907,用Logistic回归模型预测三阴性乳腺癌的准确率、敏感度、特异度、阳性预测值、阴性预测值分别为0.887、0.960、0.750、0.877、0.911。结论多参数MRI征象的Logistic回归分析模型对三阴性乳腺癌与非三阴性乳腺癌鉴别有重要的诊断价值。

Objective To explore the clinical value of Logistic regression analysis model of breast dynamic enhancement characteristics based on multi-parameter MRI(mp-MRI)in differentiating triple-negative breast cancer from non-triple-negative breast cancer.Methods A total of 194 patients with pathologically confirmed breast cancer who underwent dynamic contrast-enhanced MRI of the breast in our hospital from January 2014 to June 2020 were included in the study and divided into 68 cases of triple-negative 126 cases of triple-negative group according to the pathological results.Multivariate MRI signs were compared between the two groups,univariate analysis was performed with rank sum test,statistically significant indicators were used as the main risk predictors,risk prediction models were established with multivariate Logistic regression,and receiver operating characteristic curve(ROC)was used to analyze the diagnostic efficacy of Logistic regression models.Results Univariate analysis showed that tumor,unclear boundary,nipple involvement,single ipsilateral lymph node metastasis and higher MRI-BI-RADS grade were statistically significant(P<0.05).Multivariate analysis showed that MRI-BI-RADS grade,unclear boundary of breast mass,single ipsilateral lymph node metastasis and nipple involvement were independent predictors for differentiating triple-negative breast cancer from non-triple-negative breast cancer(P<0.05).The ROC curve area drawn by independent predictors was 0.907.The accuracy,sensitivity,specificity,positive predictive value and negative predictive value of Logistic regression model in predicting triple-negative breast cancer were 0.887,0.960,0.750,0.877 and 0.911,respectively.Conclusion Logistic regression analysis model of multi-parameter MRI features has important diagnostic value in differentiating triple-negative breast cancer from non-triple-negative breast cancer.