常规超声特征早期预测乳腺癌Ki-67高表达的模型构建  

Conventional ultrasound-based prediction model in early predicting Ki-67 expression in breast cancer

作　　者：胡巧洪[1] 侯春杰[1] 朱虎威 张婷婷 杜佩[1] 范小明[1] 张瑛 吕振晔[2] HU Qiaohong;HOU Chunjie;ZHU Huwei;ZHANG Tingting;DU Pei;FAN Xiaoming;ZHANG Ying;LYU Zhenye(Department of Ultrasound Medicine,Zhejiang Provincial People's Hospital(Affiliated People's Hospital of Hangzhou Medical College),Hangzhou 310014,China)

机构地区：[1]浙江省人民医院(杭州医学院附属人民医院)超声医学科,杭州310014 [2]浙江省人民医院(杭州医学院附属人民医院)乳腺外科,杭州310014 [3]蚌埠医学院研究生院 [4]宁波市医疗中心李惠利医院超声科

出　　处：《浙江医学》2023年第18期1915-1920,共6页Zhejiang Medical Journal

基　　金：浙江省教育厅一般科研项目(Y202044601);浙江省医药卫生科技计划项目(2021KY312)。

摘　　要：目的基于常规超声特征构建Ki-67表达水平预测模型,探讨模型对乳腺癌Ki-67高表达的早期无创诊断效能。方法回顾1999年5月至2022年5月浙江省人民医院行常规超声检查且手术病理行Ki-67检测的337例女性乳腺癌患者的影像学和临床病理资料。按照1∶4随机抽样分为训练集269例和验证集68例,并以20%作为Ki-67高表达状态的阈值分为高表达组(训练集169例,验证集43例)和低表达组(训练集100例,验证集25例)。应用单因素及多因素logistic回归分析常规超声特征与Ki-67表达水平的关系,构建基于常规超声特征的乳腺癌Ki-67高表达早期预测模型,并验证模型预测的准确性和效能。结果单因素分析显示,肿瘤最大径、边缘、血供、腋窝淋巴结异常在Ki-67高表达组与低表达组间比较差异均有统计学意义(均P<0.05)。多因素logistic回归分析显示,肿瘤最大径(OR=1.043,P<0.05)、边缘(OR=3.044,P<0.05)、腋窝淋巴结异常(OR=2.935,P<0.01)是乳腺癌Ki-67高表达的独立预测因素。基于上述特征构建乳腺癌Ki-67表达水平早期预测模型:方程Logit(p)=-1.891+1.077×腋窝淋巴结(正常=0,异常=1)+0.042×肿瘤最大径(mm)+1.113×边缘(光整=0,不光整=1),AUC为0.735(95%CI:0.672~0.798),灵敏度为0.746,特异度为0.660。验证集验证预测模型的AUC为0.794(95%CI:0.684~0.904),灵敏度为0.651,特异度为0.840。结论肿瘤最大径、边缘、腋窝淋巴结异常是乳腺癌Ki-67高表达的独立预测因素,基于常规超声特征构建的Ki-67高表达预测模型可以对乳腺癌Ki-67表达情况进行早期快速无创评估。Objective To investigate the clinical value of the conventional ultrasound-based prediction model in early predicting Ki-67 expression level in breast cancer.Methods The study retrospectively analyzed the imaging and patholo gical data of 337 female patients with breast cancer who underwent conventional ultrasonography and surgery for Ki 67 pathological detection in Zhejiang Provincial People's Hospital from May 1999 to May 2022.The enrolled patients were randomly assigned into a training set(269 cases)and a validation set(68 cases)according to 1∶4 and divided into the high expression group(169 cases in the training set and 43 cases in the validation set)and the low expression group(100 cases in the training set and 25 cases in the validation set)by using 20%as the threshold of Ki-67 high expression.In the training set,univariate and multivariate logistic regression analysis were used to construct an early prediction model of Ki 67 expression level in breast cancer based on conventional ultrasound features,followed by the evaluation of accuracy and efficiency of the prediction model in the validation set.Results In the training set,univariate analysis showed that there were statistically significant differences in maximum tumor diameter,morphology,blood supply,and axillary lymphatic abnormality between the high and low Ki-67 expression groups(all P<0.05).Multivariate logistic regression analysis showed that maximum tumor diameter(OR=1.043,P<0.05),morphology(OR=3.044,P<0.05),and axillary lymphatic abnormality(OR=2.935,P<0.01)were independent predictors of Ki-67 high expression.An early prediction model for Ki-67 expression levels in breast cancer was constructed based on the above characteristics.Equation:Logit(p)=-1.891+1.077×axillary lymph node(normal=0,abnormal=1)+0.042×maximum tumor diameter(mm)+1.113×morphology(smooth=0,unsmooth=1).The AUC was 0.735(95%CI:0.672-0.798),the sensitivity was 0.746,and the specificity was 0.660.The AUC,sensitivity,and specificity of the prediction model validated using the va

关 键 词：超声 KI-67 乳腺癌 预测模型 

分 类 号：R737.9[医药卫生—肿瘤] R445.1[医药卫生—临床医学]