摘要
目的:探讨溃疡性结肠炎(UC)患者发生结直肠癌变的影响因素,构建决策树预测模型,并评估药物干预的临床价值。方法:采用回顾性队列研究设计,纳入2018年1月—2021年12月于我院确诊并随访至2024年12月31日的UC患者。根据是否发生结直肠癌变将其分为癌变组(n=71)和非癌变组(n=285),收集所有患者的基线临床资料、疾病特征及用药史,采用单因素与多因素logistic回归分析UC癌变的危险因素,应用SPSS Modeler软件构建决策树模型,采用受试者工作特征曲线评估决策树模型的预测效能,基于决策树模型得出的风险分层,采用多因素logistic回归分析评估规律使用5-氨基水杨酸(5-ASA)与生物制剂在不同风险层级中的预防价值。结果:截至2024年12月31日,本研究中UC癌变发生率为19.94%(71/356);logistic回归分析显示,年龄≥50岁、病程≥10年、病变范围广泛、合并原发性硬化性胆管炎、白介素-6≥40 ng·L^(-1)、肿瘤坏死因子-α≥30 ng·L^(-1)、错配修复基因缺乏及c-myc基因阳性表达是UC癌变的独立危险因素(P<0.05);决策树模型筛选出6个关键预测变量,其中合并原发性硬化性胆管炎为最重要节点,模型预测UC癌变的曲线下面积为0.793,灵敏度为73.24%,特异度为70.88%;基于决策树风险分层的药物干预价值评估显示,规律使用5-ASA与使用生物制剂在整体人群中保护趋势无统计学意义(P>0.05),但在模型判定的高风险层中,两者均显示出显著的预防价值。结论:本研究成功构建的决策树模型能有效预测UC患者的癌变风险,且效能良好,模型揭示的药物干预价值具有明显的风险层级异质性,规律使用5-ASA与生物制剂对于决策树模型识别出的高危患者具有显著的预防作用。该模型为实现UC癌变的精准风险分层与预防提供了重要的循证依据。
Objective:To explore the influencing factors of colorectal cancer in patients with ulcerative colitis(UC),construct a decision tree prediction model,and evaluate the clinical value of drug intervention,providing evidence-based evidence for individualized cancer prevention.Methods:A retrospective cohort study design was adopted to include patients with UC who were diagnosed in our hospital from January 2018 to December 2021 and followed up until December 31,2024.According to whether colorectal cancer occurred,they were divided into cancer group(n=71)and non-cancer group(n=285).The baseline clinical data,disease characteristics and medication history of all patients were collected and analyzed.Univariate and multivariate logistic regression were used to analyze the risk factors of UC cancer.The decision tree model was constructed using the SPSS Modeler software.The predictive efficacy of the decision tree model was evaluated using the receiver operating characteristic curve.Based on the risk stratification derived from the decision tree model,multivariate logistic regression analysis was conducted to assess the preventive value of regular use of 5-aminosalicylic acid and biological agents at different risk levels.Results:As of December 31,2024,the incidence of UC canceration in this study was 19.94%(71/356).Univariate and multivariate logistic regression analyses showed that Age≥50 years old,disease course≥10 years,wide lesion range,combined with primary sclerosing cholangitis,IL-6≥40 ng·L^(-1),TNF-α≥30 ng·L^(-1),mismatch repair gene deficiency and positive expression of c-myc gene were independent risk factors for UC carcinogenesis(P<0.05);The decision tree model screened out six key predictive variables.Among them,the combination of primary sclerosing cholangitis was the most important node.The area under the curve of the model for predicting UC canceration was 0.793,with a sensitivity of 73.24%and a specificity of 70.88%.The value assessment of drug intervention based on decision tree risk stratification showed that there was no statistically significant protective trend of regular use of 5-aminosalicylic acid and biological agents in the overall population(P>0.05),but in the high-risk layer determined by the model,both demonstrated significant preventive value.Conclusion:The decision tree model successfully constructed in this study can effectively predict the cancer risk of UC patients with good performance.The drug intervention value revealed by the model has obvious heterogeneity at risk levels.Regular use of 5-aminosalicylic acid and biological agents has a significant preventive effect on high-risk patients identified by the decision tree model.This model provides an important evidence-based basis for achieving precise risk stratification and individualized chemoprevention of UC carcinogenesis.
作者
翟美娟
朱瑞芳
刘清珍
季士亮
王良晶
江翊国
周丁霞
ZHAI Meijuan;ZHU Ruifang;LIU Qingzhen;JI Shiliang;WANG Liangjing;JIANG Yiguo;ZHOU Dingxia(Department of Pharmacy,the Affiliated Suzhou Hospital of Nanjing Medical University,Suzhou Municipal Hospital,Suzhou 215002,China;Department of Pharmacy,Suzhou Hospital,Affiliated Hospital of Medical School,Nanjing University,Suzhou 215153,China;Department of Anesthesiology,Jinling Hospital,Affili-ated Hospital of Medical School,Nanjing University,Nanjing 210002,China)
出处
《药学与临床研究》
2026年第1期45-51,共7页
Pharmaceutical and Clinical Research
基金
江苏省中医药局项目(YB2020064)。
关键词
决策树算法
溃疡性结肠炎
癌变
风险预测模型
药物干预
价值评估
Decision tree algorithm
Ulcerative colitis
Cancerous transformation
Risk prediction model
Drug intervention
Value assessment
