中国实用外科杂志

中国实用外科杂志 ›› 2023, Vol. 43 ›› Issue (07): 796-802.DOI: 10.19538/j.cjps.issn1005-2208.2023.07.17

• 论著 • 上一篇    下一篇

HER2阳性乳腺癌病人肿瘤特异性死亡风险预测模型构建及验证

张    华1,2，李    志2，袁    杰2，贾璐瑶1，赵    菁1，赵    烨1，康    骅1   

  1. 1首都医科大学宣武医院普通外科   甲状腺乳腺疾病诊疗中心，北京 100053；2湖北医药学院太和医院乳腺甲状腺血管外科，湖北十堰 442012
  • 出版日期:2023-07-01 发布日期:2023-07-23

  • Online:2023-07-01 Published:2023-07-23

摘要： 目的    筛选影响HER-2阳性乳腺癌病人特异性死亡的预后因素，并建立预测模型列线图。方法    从监测、流行病学和最终结果（SEER）数据库中，选择2010—2015年诊断为HER-2阳性浸润性导管癌病人的临床病理资料。比较Kaplan-Meier分析和竞争风险模型计算的肿瘤特异性病死率的差异。采用竞争风险模型识别独立预后因素，并构建预测模型。采用一致性指数（C-index）、受试者工作特征（ROC）曲线下面积（AUC）和校正曲线对模型的效率进行评估。结果    共纳入31 282例HER-2阳性浸润性导管癌乳腺癌病人，采用随机抽样方法按7∶3比例分为建模组（21 897例）和测试组（9385例），组间临床病理特征差异无统计学意义（P均＞0.05）。与Kaplan-Meier生存分析相比，竞争风险模型计算出的肿瘤特异性病死率较低。年龄、种族、婚姻、组织学分级、T分期、N分期、手术方式、化疗、第一原发恶性肿瘤和孕激素受体状态是HER-2乳腺癌肿瘤特异性死亡的独立危险因素。建模组和测试组预测列线图的C-index分别为0.789和0.797。在建模组中预测肿瘤特异性死亡的1、3、5年AUC值分别为0.774、0.810、0.787，测试组中为0.776、0.814、0.793。校正曲线证实列线图与实际观测结果有较好的一致性。结论    利用竞争风险模型构建的HER-2阳性乳腺癌肿瘤特异性死亡风险预测列线图，可较好地预测HER-2阳性乳腺癌病人肿瘤特异性死亡风险，对临床决策具有一定的参考价值。

关键词: HER-2阳性乳腺癌, 竞争风险模型, 肿瘤特异性死亡, 列线图

Abstract: Establishment and verification of a nomogram to predict cancer-specific mortality risk in HER-2-positive breast cancer        ZHANG Hua*, LI Zhi, YUAN Jie, et al. *Center for Thyroid and Breast Surgery, Department of General Surgery, Xuanwu Hospital, Capital Medical University, Beijing 100053, China; Department of General Surgery, Taihe Hospital, Hubei University of Medicine, Shiyan 442012, China
Corresponding author:KANG Hua, E-mail:kanghuamd@163.com
Abstract    Objective    To identify the prognostic factors of cancer-specific mortality of HER-2-positive breast cancer and establish a cancer-specific mortality nomogram. Methods    Data of the HER-2-positive patients was extracted from the Surveillance, Epidemiology, and End Results (SEER) database. Cancer-specific mortality was calculated by Kaplan-Meier analysis and the competing risk model, respectively. The independent prognostic factors were screened out by the competing risk model. A CSM nomogram was developped. The nomogram was evaluated by concordance index，receiver operating characteristic (ROC) area under curves (AUC), and the calibration curve, respectively. Results    The study had 31282 HER-2-positive breast cancer patients after data cleansing. Patients were assigned to the training set and the testing set at random, with 21897 patients in the training set and 9385 patients in the testing set. Comparing the competing risk model and the Kaplan-Meier survival analysis, CSM calculated by the former is lower (P<0.01). The independent prognostic factors were age, race, grade, T stage, N stage, surgery, chemotherapy, marital status, history of malignancy, PR state. A nomogram was developed by the independent prognostic factors. The C-index of the nomogram was 0.789 and 0.797 in the training and testing sets, respectively. In the training set, the 1-year, 3-year, and 5-year AUC of the ROC values were 0.774, 0.810, and 0.787, respectively, and the corresponding values in the testing set were 0.776, 0.814, and 0.793. The calibration curves showed good consistency in predicting CSM. Conclusion    A prediction nomogram was created with good performance. These results may provide a reference value of decision-making in clinical practice.

Key words: HER-2-positive breast cancer, competing risk model, cancer-specific mortality, nomogram