列线图预测模型结合超声多参数评分预测乳腺癌新辅助化疗效果及受试者工作特征分析

Analysis on the nomogram prediction model combined with the score of multi-parameter of ultrasound in predicting the effect of NAC on breast cancer and the ROC

目的:基于乳腺癌患者临床资料建立预测模型,分析列线图预测模型结合多参数半定量超声对新辅助化疗(NAC)后病理完全缓解(pCR)的预测价值。方法:选取医院收治的126例行NAC女性乳腺癌患者,根据是否pCR将其分为pCR组(33例)和非pCR组(93例)。采用单因素分析和多因素Logistic回归分析临床资料与NAC后pCR的关系,根据多因素Logistic回归分析结果建立列线图预测模型,绘制受试者工作特征(ROC)曲线,分析列线图预测模型、列线图预测模型结合超声多参数评分在NAC疗效评估中的预测价值。结果:Logistic回归分析结果显示,病理分型、增殖细胞核抗原-67(Ki-67)及肿瘤直径是乳腺癌NAC后pCR的独立预测因素(OR=1.907,OR=0.518,OR=2.294;P<0.05)。列线图预测模型的ROC曲线下面积(AUC)为0.806,列线图预测模型结合超声多参数的AUC为0.881。结论:以病理分型、Ki-67及肿瘤直径为基础构建的列线图模型可较准确地预测乳腺癌NAC效果,结合超声多参数评分后效果更为突出,且联合预测效果更好。

Objective:To analyze the predictive value of nomogram prediction model combined with multiparameter semi-quantitative ultrasound on pathological complete response(pCR)after neoadjuvant chemotherapy(NAC)based on the predictive model that was established on the basis of the clinical data of patients with breast cancer.Methods:A retrospective analysis was performed on 126 patients with breast cancer who received NAC and surgical treatment in hospital.All of them were divided into pCR group(33 cases)and non-pCR group(93cases)according to whether occurred pCR.Univariate analysis and multivariate Logistic regression were used to analyze the relationship between clinical data and pCR after NAC,and a nomogram prediction model was established according to the results of multivariate Logistic regression analysis.In addition,a receiver operating characteristic(ROC)curve was constructed to analyze the predictive value of the nomogram prediction model and the combination of the nomogram prediction model and the score of multi-parameter of ultrasound in the assessment of NAC curative effect.Results:The results of Logistic regression analysis showed that pathological type,proliferating cell nuclear antigen(KI-67)and tumor diameter were independent predictors of pCR after NAC of breast cancer(OR=1.907,OR=0.518,OR=2.294,P<0.05).The analysis of ROC curve showed that the area under curve(AUC)of the nomogram prediction model was 0.806,and the AUC of the nomogram prediction model combined with multi-parameter of ultrasound was 0.881.Conclusion:The constructed nomogram prediction model based on pathological classification,KI-67 and tumor diameter can accurately predict the effect of NAC for breast cancer,and the effect of that is more prominent after combines with the score of multi-parameter of ultrasound,and the effect of the combined detection is better.