首都医科大学学报 ›› 2023, Vol. 44 ›› Issue (2): 295-301.doi: 10.3969/j.issn.1006-7795.2023.02.017

• 临床研究 • 上一篇    下一篇

结直肠癌术后患者异时性进展期腺瘤风险筛查模型的构建及验证

郭春梅1,刘红1,焦月2,张倩2,王沧海1,王亚丹1,蔺武1,魏南1,张澍田2,吴静1,2*
  

  1. 1.首都医科大学附属北京世纪坛医院消化内科,北京,100038;        2.首都医科大学附属北京友谊医院消化内科,北京,100050
  • 收稿日期:2022-09-26 出版日期:2023-04-21 发布日期:2023-04-18
  • 通讯作者: 吴静 E-mail:wujing36youyi@ccmu.edu.cn
  • 基金资助:
    北京市属医院科研培育项目(PX2021030),首都卫生发展科研专项 (2020-4-2085)

Design and verification of screening model of metachronous advanced adenoma for postoperative colorectal cancer patients

Guo Chunmei1, Liu Hong1, Jiao Yue2, Zhang Qian2,  Wang Canghai1, Wang Yadan1, Lin Wu1, Wei Nan1, Zhang Shutian2, Wu Jing1,2*   

  1. 1. Gastroenterology Department, Beijing Shijitan Hospital, Capital Medical University, Beijing 100038, China;      2. Gastroenterology Department, Beijing Friendship Hospital, Capital Medical University, Beijing 100050, China
  • Received:2022-09-26 Online:2023-04-21 Published:2023-04-18
  • Supported by:
    This study was supported by Beijing Municipal Administration of Hospitals Incubating Program(PX2021030), Capitals Funds for Health Improvement and Research (2020-4-2085)

摘要: 目的  筛选结直肠癌术后患者发生异时性进展期腺瘤的危险因素,构建个体化风险筛查模型并验证模型的筛查效果。方法  将2007年4月至2017年10月行结直肠癌根治术的734例患者分为建模组和验证组,术后3年结肠镜随访。建模组采用单因素及多因素Logistic回归分析,建立风险筛查模型。利用验证组验证模型的筛查效果。结果  734例患者包括建模组485例,验证组249例。平均年龄为(64.6±11.5)岁,男性440例,女性294例。总的异时性进展期腺瘤发生率为11.9%(建模组12.0%,验证组11.6%)。多因素Logistic回归分析显示,男性、糖尿病病史、右半结肠癌、中低分化腺癌、腺癌合并腺瘤,腺癌合并进展期腺瘤为异时性进展期腺瘤的独立预测因素。根据Logistic多因素分析结果,建立筛查结直肠癌术后出现异时性进展期腺瘤的模型,并建立风险筛查列线图。受试者工作特征 (receiver operating characteristic,ROC)曲线分析结果显示,该模型的受试者工作特征曲线下面积 (area under the curve,AUC)=0.957(95%CI:0.935~0.973),模型筛查临界值为0.938 7。验证结果显示,构建的筛查模型的灵敏度=93.1%,特异度= 89.1%,并且筛查结果与实际结果的一致性较高(Kappa=0.62)。结论  以男性、糖尿病病史、右半结肠癌、中低分化腺癌、腺癌合并腺瘤、腺癌合并进展期腺瘤七个因素,构建的结直肠癌术后出现异时性进展期腺瘤筛查模型,具有较高的真实性和与实际结果的一致性,有临床应用价值。

关键词: 结直肠癌, 异时性进展期腺瘤, 筛查模型, 验证

Abstract: Objective  To analysis the potential risk factors for metachronous advanced adenoma (MAA) in postoperative colorectal cancer (CRC) patients, so as to develop an individual surveillance scheme and verify it. Methods  Totally 734 CRC patients who underwent surgery from April 2007 to October 2017 were divided into derivation group and validation group, and underwent surveillance colonoscopy for 3 years. Univariate and multivariate logistic analysis were conducted to establish screening model, and verify it in validation group. Results  Totally 734 patients (male/female: 440/294) were finally included. The average age was (64.6 ±11.5) years, and the overall MAA incidence was 11.9% (12.0% in derivation group and 11.6% in validation group). Male, diabetes mellitus, right-sided colon cancer, moderately and poorly differentiated adenocarcinoma, synchronous adenoma, synchronous advanced adenoma were independent risk factors for MAA. According to the results of multivariate logistic analysis, MAA screening model and nomogram were established. The area under the curve (AUC) of MAA screening model was 0.957(95% CI: 0.935-0.973), and critical value was 0.938 7. External validation showed that the sensitivity was 93.1%, specificity was 89.1%, and the consistency between screening results and actual results was high (Kappa=0.62). Conclusions  The screening model of metachronous advanced adenoma for postoperative CRC patients was constructed based on seven factors: male, diabetes history, right-sided colon cancer, moderately and poorly differentiated adenocarcinoma, synchronous adenoma, and synchronous advanced adenoma. It had high authenticity and consistency with actual results, and showed clinical application value.

Key words: colorectal cancer, metachronous advanced adenoma, screening model, verification

中图分类号: