鼻咽癌同步放化疗病人营养不良风险预测模型的构建及验证

Building and validating a risk prediction model for malnutrition during concurrent chemoradiotherapy in patients with nasopharyngeal carcinoma

目的:建立并验证鼻咽癌同步放化疗病人营养不良风险预测模型。方法:2022年4月~2023年8月,便利选取广西医科大学第一附属医院放疗科收治的430例初治鼻咽癌同步放化疗病人作为研究对象,按入组先后顺序以7:3比例将其分为建模组300例和内部验证组130例,选取广西医科大学附属肿瘤医院收治的61例初治鼻咽癌同步放化疗病人作为外部验证组。采用Logistic回归建立风险预测模型并绘制列线图,采用HosmerLemeshow、校准曲线及ROC验证模型拟合优度及预测效果,采用临床决策曲线评估临床效用。结果:Logstic回归分析显示,骨骼肌质量指数、自评焦虑量表评分、匹兹堡睡眠质量问卷评分、中国膳食宝塔评分、规律运动锻炼、消化系统症状群组是鼻咽癌同步放化疗病人发生营养不良的影响因素。建模组中ROC曲线下面积为0.853(95%CI:0.81~0.89),Youden最大值为0.600,对应的特异度为0.764、灵敏度为0.836。Hosmer-Lemeshow检验χ^(2)=4.040,P=0.853,表明模型具有良好的预测能力。校准曲线校显示模型的预测效果与实际概率拟合较好,平均绝对误差为0.024。临床决策曲线的阈值概率为0.05~0.85时,临床净收益率较高。内部验证组受试者操作曲线下面积为0.891,灵敏度为77.36%,特异度为89.61%,实际应用正确率84.62%。外部验证组受试者操作曲线下面积为0.886,灵敏度为76.00%,特异度为83.33%,总体正确率为80.33%。结论:本研究构建的风险预测模型效果较好,可有效预测鼻咽癌同步放化疗病人营养不良的发生率,为临床工作人员制订和实施营养干预措施提供参考依据。

Objective:To develop and validate a model to predict the risk of malnutrition in patients with nasopharyngeal carcinoma receiving concurrent chemoradiotherapy.Methods:From April 2022 to August 2023,430patients with nasopharyngeal carcinoma who were admitted to the department of radiotherapy of the first affiliated hospital of Guangxi medical university in Nanning were conveniently selected as the study subjects,and they were divided into the modelling group (300 cases) and the internal validation group (130 cases) in the internal validation group in the ratio of 7:3,and 61 patients with nasopharyngeal carcinoma admitted to the affiliated cancer hospital of Guangxi medical university in Nanning City were selected as the external validation group.Logistic regression was used to establish the risk prediction model and draw nomograms,Hosmer-Lemeshow,calibration curve and ROC were used to verify the goodness of fit and predictive power of the model,and clinical decision curve was used to assess the clinical utility.Results:Logistic regression analysis showed that skeletal muscle mass index,self-rated anxiety scale score,Pittsburgh sleep quality questionnaire score,Chinese diet pagoda score,regular exercise,and digestive symptom groups were the influencing factors for malnutrition in patients with nasopharyngeal carcinoma receiving concurrent chemoradiotherapy.In the modelling group,the area under the ROC curve was 0.853 (95%CI:0.81~0.89),the maximum Youden was 0.600,and the corresponding specificity was 0.764 and the sensitivity was 0.836.The HosmerLemeshow test=4.040 and P=0.853 indicated that the model had good predictive ability.Calibration curve of the calibration showed that the predictive effect of the model matched actual probability well,with an average absolute error was 0.024.When the threshold probability of the clinical decision curve is 0.05~0.85,the clinical response rate is higher.The area under the operating curve of the subjects in the internal validation group was 0.891,the sensitivity was77.36%,the specificity was 89.61%,and the practical application accuracy was 84.62%.The area under the operating curve of the subjects in the external validation group was 0.886,the sensitivity was 76.00%,the specificity was 83.33%,and the overall accuracy was 80.33%.Conclusion:The risk prediction model constructed in this study has a good effect,which can effectively predict the incidence of malnutrition in patients receiving concurrent radiotherapy and chemotherapy for nasopharyngeal carcinoma,and provide a reference for clinical staff to formulate and implement nutritional interventions.