添加胆汁酸作为风险因素用于新型结直肠癌预测模型的开发OA北大核心
Developing of a novel predictive model for colorectal cancer by adding bile acids as risk factors
为探究结直肠癌发病的风险因素,并联合血清多种胆汁酸浓度构建结直肠癌发病的预测模型,前瞻性纳入2020-2024年于扬州大学附属江都人民医院胃肠外科就诊的患者,并获取相应的病历资料与实验室检查(含液相色谱检测的血清胆汁酸亚型浓度),对结直肠癌患者与对照组患者进行比较,分析差异变量.并将所有患者随机抽样(比例7∶3)分组为训练集与验证集,使用单因素与多因素logistic回归法,以训练集逐步构建预测模型,并绘制列线图、ROC曲线、校准曲线与DCA曲线,随后使用验证集进行验证.结果表明:共收集了 421例患者资料(结直肠癌组217例、对照组204例),分析差异变量发现:年龄增加、性别男、糖尿病史、不规律结肠镜检查、结直肠癌家族史、幽门螺旋杆菌感染史、高BMI、低ALB、高GCA、高GCDCA、高TCDCA、高TBA在结直肠癌组更常见.经过单因素与多因素logistic回归分析,年龄、结直肠癌家族史、糖尿病史、结肠镜检查、幽门螺旋杆菌感染史、BMI、ALB、GCA被识别为结直肠癌发病的独立危险因素(P＜0.05),训练集/验证集ROC曲线显示AUC为0.79/0.75,校准曲线一致性良好,DCA曲线显示预测模型具有良好的临床净获益.综上,多个因素共同影响了结直肠癌的发生,该研究联合血清胆汁酸亚型构建的模型具有良好的预测价值.
To explore the risk factors for colorectal cancer and to develop a predictive model for colorectal cancer by combining serum levels of various bile acids.A prospective study was conducted on patients who visited the Department of Gastrointestinal Surgery at the Affiliated Jiangdu People's Hospital of Yangzhou University between 2020 and 2024.Rele-vant medical records and laboratory test results,including serum bile acid subtypes measured by liquid chromatography,were collected.Patients with colorectal cancer were compared with control patients to analyze the differences in variables.All patients were randomly divided into training and validation sets in a 7∶3 ratio.Using univariate and multivariate lo-gistic regression analysis,a predictive model was gradually constructed in the training set,and nomograms,ROC curves,calibration curves,and DCA curves were plotted.The validation set was then used to verify the model.Results showed that:A total of 421 patients were included(217 in the colorectal cancer group;204 in the control group).Analysis of dif-ferential variables revealed that increased age,male gender,history of diabetes,irregular colonoscopy,family history of colorectal cancer,history of Helicobacter pylori infection,high BMI,low ALB,high GCA,high GCDCA,high TCDCA,and high TBA were more common in the colorectal cancer group.Through univariate and multivariate logistic regression analysis,age,family history of colorectal cancer,history of diabetes,colonoscopy frequency,history of Helicobacter py-lori infection,BMI,ALB,and GCA were identified as independent risk factors for colorectal cancer(P＜0.05).The ROC curves of the training and validation sets showed AUC values of 0.79 and 0.75,respectively.The calibration curves demonstrated good agreement,and the DC A curves indicated that the predictive model had good clinical net benefit.Mul-tiple factors contributed to the development of colorectal cancer.This study,by incorporating serum bile acid subtypes,constructed a model with good predictive value.
黄永;宋航宇;陈奕名;丁毅;石立新;杨炎;韦伟
扬州大学附属江都人民医院,江苏扬州 225100扬州大学附属江都人民医院,江苏扬州 225100扬州大学医学院,江苏扬州 225009扬州大学医学院,江苏扬州 225009扬州大学附属江都人民医院,江苏扬州 225100扬州大学附属江都人民医院,江苏扬州 225100扬州大学附属江都人民医院,江苏扬州 225100
基础医学
结直肠癌预测模型胆汁酸风险因素
colorectal cancerpredictive modelbile acidsrisk factors
《扬州大学学报（农业与生命科学版）》 2025 (2)
38-47,76,11
扬州市社会发展项目(YZ2022089、YZ2023094)扬州市卫生健康委员会医学重点科研项目(2023-1-04)
评论