含营养指标的结直肠癌患者化疗后不良反应预测模型构建及验证  

Construction and verification of a nutritional index-containing adverse reaction prediction model in colorectal cancer patients after chemotherapy

作　　者：陈作伟 陈彬 王韩英 Zuo-Wei Chen;Bin Chen;Han-Ying Wang(Department of Gastroenterology,Qiushi Cardiova-scular Hospital,Hangzhou 310012,Zhejiang Province,China;Department of Gastroenterology,Lishui People’s Hospital,Lishui 323000,Zhejiang Province,China;Department of Oncology Radiotherapy,Lishui People’s Hospital,Lishui 323000,Zhejiang Province,China)

机构地区：[1]江求是心血管病医院消化内科,浙江省杭州市310012 [2]浙江省丽水市人民医院消化内科,浙江省丽水市323000 [3]浙江省丽水市人民医院肿瘤放疗科,浙江省丽水市323000

出　　处：《世界华人消化杂志》2025年第2期148-157,共10页World Chinese Journal of Digestology

摘　　要：背景结直肠癌患者化疗后易出现不良反应,但缺乏早期预测模型,基于营养状况与消化系统恶性肿瘤化疗、放化疗不良反应具有一定相关性,推测其可作为预后模型相关因素.目的探究结直肠癌患者化疗后不良反应发生影响因素,并构建含营养指标的预测模型.方法选取2017-07/2021-03浙江求是心血管病医院接受化疗的结直肠癌患者210例作为模型建模集,统计患者化疗后不良反应发生率,采用Lasso初筛结直肠癌患者化疗后不良反应的特征变量,Logistic分析结直肠癌患者化疗后不良反应因素;构建列线图模型,并采用校准曲线、受试者工作特征(receiver operating characteristic,ROC)曲线评价该模型准确度及区分度.另选取2021-04/2024-03接受化疗的结直肠癌患者90例作为模型验证集,对构建的列线图模型进行验证.结果化疗1个疗程结束后,建模集210例结直肠癌患者中,85例(40.48%)发生不良反应(≥Ⅲ级)纳入发生组,125例(59.52%)未发生不良反应纳入未发生组;Lasso回归分析显示,当惩罚系数λ=0.072时模型性能优良且影响因素最少,最终筛选出6个预测变量为年龄、乳酸脱氢酶、ECOG评分及营养指标白蛋白、总淋巴细胞计数、血红蛋白;Logistic分析显示,年龄、美国东部肿瘤协作组(Eastern Cooperative Oncology Group,ECOG)评分、乳酸脱氢酶及营养指标白蛋白、血红蛋白、总淋巴细胞计数均为结直肠癌患者化疗后不良反应影响因素(P<0.05);根据影响因素(年龄、乳酸脱氢酶、ECOG评分及营养指标白蛋白、总淋巴细胞计数、血红蛋白)建立列线图预测模型,该列线图模型在建模集中的ROC曲线的曲线下面积为0.811(95%CI:0.754-0.869),预测结直肠癌患者化疗不良反应发生相关因素的一致性指数为0.839(95%CI:0.778-0.902),与实际观察结果相关性较高,提示该模型具有良好区分度及准确度;使用验证集患者数据对构成化疗不良反应的列�BACKGROUND Patients with colorectal cancer are prone to adverse reactions after chemotherapy,but there is a lack of models for early prediction of these adverse reactions.Based on the correlation between nutritional status and chemotherapy and radiochemotherapy adverse reactions in malignant tumors of the digestive system,it is speculated that nutritional status can be used as a factor involved in prognostic models.AIM To explore the influencing factors of adverse reactions in patients with colorectal cancer after chemotherapy,and establish a prediction model containing nutritional indexes.METHODS A total of 210 patients with colorectal cancer who received chemotherapy in Zhejiang Qiushi Cardiovascular Hospital from July 2017 to March 2021 were selected as a model training set to analyze the incidence of adverse reactions after chemotherapy.The characteristic variables of adverse reactions after chemotherapy in colorectal cancer patients were initially screened by Lasso regression.Logistic regression analysis was performed to identify factors related to the development of adverse reaction in patients with colorectal cancer after chemotherapy.Based on the factors identified,a prediction model was developed.The calibration curve and receiver operating characteristic(ROC)curve were used to evaluate the accuracy and differentiation of the model.In addition,90 patients with colorectal cancer who received chemotherapy in Zhejiang Qiushi Cardiovascular Hospital from April 2021 to March 2024 were selected as a model validation set to verify the constructed nomogram model.RESULTS After one cycle of chemotherapy,of 210 patients with colorectal cancer in the modeling set,85(40.48%)patients with colorectal cancer in the model training group had adverse reactions(≥GradeⅢ)and 125(59.52%)had no adverse reactions and were included in the no-occurrence group.Lasso regression analysis showed that when the penalty coefficientλ=0.072,the model had good performance and the least influencing factors.Six predictive variables were

关 键 词：结直肠癌 化疗 营养指标 不良反应 列线图预测模型 

分 类 号：R73[医药卫生—肿瘤]