初治急性髓系白血病早期并发症状性心力衰竭风险的列线图模型构建及其效果的初步评估

Establishment and preliminary assessment of a nomogram model for the risk of early complication of symptomatic heart failure in de novo patients with acute myeloid leukemia

目的:探讨初治急性髓系白血病早期并发症状性心力衰竭(心衰)的危险因素,建立预测症状性心衰风险的列线图模型。方法:回顾性分析苏州大学附属第一医院血液科2017年9月—2021年10月初治急性髓系白血病288例,根据3个月内是否发生症状性心衰分为心衰组(59例)和非心衰组(229例)。比较两组临床资料,采用多因素logistic回归分析筛选患者并发症状性心衰的危险因素;应用R语言建立预测症状性心衰风险的列线图模型,采用受试者工作曲线(ROC)分析列线图模型对初治急性髓系白血病早期并发症状性心力衰竭的预测效果。结果:心力衰竭组和非心力衰竭组患者的年龄、舒张压、丙氨酸氨基转移酶、乳酸脱氢酶、人血白蛋白、前白蛋白、3个月内地西他滨的使用率、降钙素原或G/GM试验阳性或败血症均差异有统计学意义(均P<0.05)。3个月后随访两组间体重变化及E峰、左心房大小变化均差异有统计学意义(均P<0.05)。多因素logistic回归分析结果显示,年龄(OR=1.026,95%CI:1.004~1.049,P=0.020)、舒张压(OR=0.958,95%CI:0.926~0.992,P=0.958)、前白蛋白(OR=0.995,95%CI:0.991~1.000,P=0.041)、感染指标或败血症(OR=2.590,95%CI:1.367~4.905,P=0.004)是初治急性髓系白血病并发症状性心衰的危险因素。列线图模型预测初治急性髓系白血病早期并发症状性心衰的一致性指数为0.714。ROC曲线分析显示,模型曲线下面积(AUC)为0.714(95%CI:0.645~0.783),模型截断值为190.1分,灵敏度为0.831,特异度为0.359。结论:基于初治急性髓系白血病早期并发症状性心衰风险的列线图模型具有一定的区分度和预测效果,可辅助临床诊疗决策。

Objective:To explore the risk factors of early complication of symptomatic heart failure in de novo patients with acute myeloid leukemia,and establish a nomogram model to predict the risk of symptomatic heart failure.Methods:A total of 288 patients of de novo patients with acute myeloid leukemia in First Affiliated Hospital of Soochow University from September 2017 to October 2021 were retrospectively enrolled.Patients were divided into the heart failure group(n=59) and non-heart failure group(n=229) according the presence or absence of heart failure in three mouths.The risk factors of heart failure were analyzed by multivariate logistic regression analysis.R language was used to establish a nomogram model to predict the risk of early complication of symptomatic heart failure.Receiver operating characteristic(ROC) curve was used to explore the prediction efficiency of the nomogram model for symptomatic heart failure in de novo patients with acute myeloid leukemia.Results:There were statistical significance differences in age,diastolic blood pressure,alanine aminotransferase,lactic dehydrogenase,albumin,prealbumin,decitabine usage,procalcitonin,G test/GM test or septicemia between the heart failure group and the non-heart failure group(all P<0.05).There were significant differences in body weight,E peak and left atrial size between the two groups after 3 months follow-up(all P<0.05).Multivariate logistic regression analysis showed that age(OR=1.026,95%CI:1.004-1.049,P=0.020),diastolic blood pressure(OR=0.958,95%CI:0.926-0.992,P=0.958),prealbumin(OR=0.995,95%CI:0.991-1.000,P=0.041) and G test/GM test or septicemia(OR=2.590,95%CI:1.367-4.905,P=0.004) were independent risk factors for symptomatic heart failure in de novo patients with acute myeloid leukemia.The consistency index of the nomogram model for predicting symptomatic heart failure in de novo patients with acute myeloid leukemia was 0.714.ROC curve showed that the area under the curve predicted by nomogram model was 0.714(95%CI:0.645-0.783),the cutoff value was 190.1 points,and the sensitivity and specificity were 0.831 and 0.359,respectively.Conclusion:Based on the risk factors for symptomatic heart failure in de novo patients with acute myeloid leukemia,a nomogram model for predicting the risk of heart failure is established,which has a certain differentiation degree and prediction effect,which is helpful to guide clinical diagnosis and treatment.