多发性骨髓瘤并发带状疱疹的预测模型构建研究  

Construction of a predictive model for multiple myeloma complicated with herpes zoster

作　　者：郭丽莎 俞颖 Guo Lisha;Yu Ying(Department of Hematology and Oncology,the First People's Hospital of Linping District,Hangzhou 311201,China)

机构地区：[1]杭州市临平区第一人民医院血液肿瘤科,浙江杭州311201

出　　处：《中国医院统计》2024年第5期345-350,共6页Chinese Journal of Hospital Statistics

摘　　要：目的分析多发性骨髓瘤(MM)并发带状疱疹(HZ)的危险因素,构建列线图预警模型。方法选取2020年4月—2022年3月某院血液内科收治的MM患者作为研究对象,根据是否发生HZ将其分为HZ组和非HZ组。分析所选MM患者的临床资料,采用logistic回归分析筛选出MM患者并发HZ的危险因素,根据独立危险因素构建列线图并进行拟合优度检验。结果本研究共纳入229例MM患者,经检测确认HZ发生率为24.02%;年龄≥65岁(OR=3.242,95%CI:1.456~7.218)、使用硼替佐米(OR=3.902,95%CI:1.725~8.830)、使用环磷酰胺(OR=2.862,95%CI:1.349~6.073)、骨髓瘤状态未缓解(OR=3.004,95%CI:1.394~6.475)、粒细胞减少(OR=2.496,95%CI:1.147~5.433)、白球蛋白比值<0.5(OR=4.440,95%CI:1.933~10.196)、淋巴细胞减少(OR=3.145,95%CI:1.481~6.679)是MM患者并发HZ的独立危险因素(P<0.05);列线图模型显示,年龄≥65岁为79分、使用硼替佐米为92分、使用环磷酰胺为71分、骨髓瘤状态未缓解为73分、粒细胞减少为62分、白球蛋白比值<0.5为99分、淋巴细胞减少为76分,验证结果显示C-index为0.846,校正曲线趋近于理想曲线,ROC曲线下面积为0.815,阈值概率在5%~84%内净获益值较高。结论本文构建的列线图预测模型准确性较高,可为临床评估MM患者HZ发生风险做出有利贡献,也满足了临床上对于整合模型的需求。Objective To analyze the risk factors of multiple myeloma(MM)complicated with herpes zoster(HZ),and to construct a nomogram early warning model.Methods The patients with MM who were admitted to the Department of Hematology in a hospital from April 2020 to March 2022 were selected as research objects,and they were divided into the HZ group and the non-HZ group according to whether HZ occurred.The clinical data of selected MM patients were analyzed,and logistic regression analysis was used to screen out the risk factors for MM patients complicated with HZ.A nomogram was constructed based on independent risk factors and a goodness-of-fit test was performed.Results A total of 229 MM patients were included in this study,and the incidence rate of HZ confirmed by testing was 24.02%(55/229);age≥65 years old(OR=3.242,95%CI:1.456-7.218),taking bortezomib(OR=3.902,95%CI:1.725-8.830),using cyclophosphamide(OR=2.862,95%Cl:1.349-6.073,P=0.006),myeloma status(0R=3.004,95%CI:1.394-6.475),neutropenia(0R=2.496,95%Cl:1.147-5.433),white globulin ratio<0.5(OR=4.440,95%Cl:1.933-10.196),lymphopenia(OR=3.145,95%Cl:1.481-6.679)were independent risk factors for MM patients with HZ(P<0.05);the nomogram model showed that age≥65 years was 79 points,bortezomib was 92 points,cyclophosphamide was 71 points,myeloma status was 73 points,granulocytopenia was 62 points,globulin ratio<0.5 was 99 points,and lymphocytopenia was 76 points.The verification results showed that the Cindex was 0.846,the calibration curve was close to the ideal curve,the area under the R0C curve was 0.815,the threshold probability was 5%-84%,and the net benefit value was higher.Conclusion The nomogram prediction model constructed in this paper has high accuracy,which can make a favorable contribution to the clinical assessment of the risk of HZ in MM patients,and also meets the clinical needs for the integrated model.

关 键 词：多发性骨髓瘤 带状疱疹 危险因素 预测模型 列线图 

分 类 号：R73[医药卫生—肿瘤]