基于超声影像组学评分构建浸润性乳腺癌HER2表达预测模型  

The construction of a predictive model for HER2 expression in invasive breast cancer based on ultrasonographic radiomics scoring

作　　者：赵安莉 吴江锋 杜艳红 吴琴 胡丽艳 王正平 Zhao Anli

机构地区：[1]浙江省东阳市人民医院,322100

出　　处：《浙江临床医学》2025年第3期346-348,352,共4页Zhejiang Clinical Medical Journal

基　　金：浙江省金华市公益性技术应用研究项目(2023-4-225)。

摘　　要：目的构建基于超声影像组学的机器学习模型用于术前预测乳腺癌(BC)HER2状态。方法收集2020年6月至2024年6月病理确诊的BC患者436例,采用免疫组织化学和荧光原位杂交技术分析术后肿瘤病理标本中HER2的状态;LASSO回归选择不同变量,采用Logistic回归分析HER2状态影响因素;校准曲线分析列线图模型的效能。结果436例乳腺癌患者中有97例HER2结果为阳性。不同HER2状态患者孕激素受体表达、Ki-67水平和雌激素受体表达差异有统计学意义(P<0.05)。LASSO回归从786个特征池中确定6个与HER2状态密切相关的超声变量,从而为每个患者生成影像组学评分。多变量Logistic回归分析显示,PR(OR=0.15,95%CI:0.06~0.36,P<0.001)、Ki-67(OR=1.02,95%CI:1.00~1.03,P=0.012)和影像学评分(OR=5.89,95%CI:2.58~13.45,P<0.001)是HER2状态的独立预测因素,列线图模型在训练集和验证集中的ROC曲线下面积分别为0.823(95%CI:0.772~0.874)和0.812(95%CI:0.717~0.906)。结论基于超声影像组学评分构建的深度学习模型可以有效预测乳腺癌患者HER2表达状态,为临床治疗决策提供重要依据。Objective To construct a machine learning model based on ultrasound radiomics for the preoperative prediction and assessment of HER2 status in breast cancer.Methods A total of 436 patients with breast masses and definitive pathological results were collected from June 2020 to June 2024.The HER2 status of postoperative tumor pathological specimens was analyzed using immunohistochemistry and fluorescence in situ hybridization.LASSO regression was employed to select different variables,and logistic regression was used to analyze factors influencing HER2 status.The performance of the nomogram model was evaluated using calibration curves.Results Among the 436 breast cancer patients,97 had positive HER2 results.There were statistically significant differences in progesterone receptor expression,Ki-67 levels,and estrogen receptor expression among patients with different HER2 statuses(P<0.05).LASSO regression identified 6 ultrasound variables closely related to HER2 status from a pool of 786 features,generating a radiomics score for each patient.Multivariate logistic regression analysis revealed that PR(OR=0.15,95%CI:0.06~0.36,P<0.001),Ki-67(OR=1.02,95%CI:1.00~1.03,P=0.012),and radiomics score(OR=5.89,95%CI:2.58~13.45,P<0.001)were independent predictors of HER2 status.The area under the ROC curve(AUC)of the nomogram model was 0.823(95%CI:0.772~0.874)in the training set and 0.812(95%CI:0.717~0.906)in the validation set.Conclusion The deep learning model constructed based on ultrasound radiomics scores can effectively predict HER2 expression status in breast cancer patients,providing an important basis for clinical treatment decision-making.

关 键 词：乳腺癌 人表皮因子受体2 超声影像 人工智能 预测模型 

分 类 号：R737.9[医药卫生—肿瘤]