结直肠癌化疗后病人认知功能障碍预测模型的建立

Establishment of a predictive model for cognitive dysfunction in patients with colorectal cancer after chemotherapy

目的:分析结直肠癌幸存者癌症相关认知障碍(cancer-related cognitive impairment,CRCI)的影响因素,建立列线图预测模型。方法:将医院门诊及病房复诊的356例结直肠癌幸存者纳入研究,按照1∶3比例分为建模组(267例)和验证组(89例)。采用LASSO回归分析以及交叉验证筛选建模组CRCI的相关变量,纳入多因素Logistic回归分析确认CRCI的独立危险因素。建立列线图预测模型并评估其预测的准确性。结果:所有结直肠癌幸存者CRCI发生率为29.5%。建模组中,年龄、教育水平、放疗、抑郁为CRCI的独立危险因素(P<0.05)。采用R软件建立CRCI列线图,建模组和验证组受试者工作特征曲线下面积分别为0.750[95%CI(0.681,0.819)]和0.720[95%CI(0.641,0.800)]。两组的校准曲线均表明鉴别能力良好,临床决策曲线显示净收益率高。结论:本研究建立的结直肠癌幸存者CRCI预测模型具有较好的预测能力,有助于甄别高风险病人并及早制定预防措施。

Objective:To analyze the influencing factors of cancer-related cognitive impairment(CRCI)in colorectal cancer survivors and to establish a nomogram prediction model.Methods:A total of 356 colorectal cancer survivors who revisited outpatient clinics and wards of our hospital were included in the study and divided into a modeling group(267 cases),and a verification group(89 cases)according to the ratio of 1∶3.LASSO regression analysis and cross-validation were used to screen the relevant variables of CRCI in the modeling group.Multivariate Logistic regression analysis was included to confirm the independent risk factors of CRCI.Anomogram prediction model was established,and its prediction accuracy was evaluated.Results:The incidence of CRCI among all colorectal cancer survivors was 29.5%.In the modeling group,age,education level,radiotherapy,and depression were independent risk factors for CRCI(P<0.05).R software was used to establish the CRCI nomogram,and the area under the receiver operating characteristic curve of the modeling group and the verification group were:0.750[95%CI(0.681,0.819)]and 0.720[95%CI(0.641,0.800)].Calibration curves for both groups showed good discriminatory power,and clinical decision curves showed high net yield.Conclusions:The CRCI prediction model for colorectal cancer survivors established in this study has good predictive ability,which is helpful to identify high-risk patients and formulate preventive measures early.