基于临床常用指标建立急性髓系白血病死亡风险模型

Establishment of death risk model for acute myeloid leukemia based on conventional clinical indexes

目的应用临床常规指标建立急性髓系白血病(acute myelocytic leukemia,AML)死亡风险模型,为AML患者预后判断提供参考。方法回顾性分析165例AML患者的临床资料,以5a生存情况为结果变量,对可能影响预后的因素进行Cox比例风险回归分析,建立死亡风险模型。结果单变量Cox比例风险回归显示,年龄>60岁、体能状态评分高、确诊时外周血白细胞计数>50×109/L、放弃治疗、诱导治疗(柔红霉素+阿糖胞苷方案)4周失败及姑息治疗增加5a死亡风险,确诊时血红蛋白>80g/L是预后的保护因素;多因素Cox比例风险回归显示,年龄(RR=1.872)、体能状态评分(RR=1.269)、确诊时血红蛋白(RR=0.585)、外周血白细胞计数(RR=1.838)和治疗方案(RR=1.651)是影响预后的独立因素,死亡风险模型:h(t,X)/h0(t)=Exp(0.627年龄+0.238体能状态评分-0.536血红蛋白+0.609白细胞计数+0.501治疗方案);此模型预测死亡的AUC为0.867,大于年龄(0.725)、血红蛋白(0.698)、体能状态评分(0.682)、白细胞计数(0.634)和治疗方案(0.587)。结论以年龄、血红蛋白、体能状态评分、白细胞计数和治疗方案建立的死亡风险模型可为AML预后提供中等强度的诊断能力。

Objective To establish death risk model for acute myeloid leukemia（AML）based on conventional clinical indexes in order to provide references for predicting the prognosis of the patients with AML.Methods The clinical data of 165 patients with AML were retrospectively analyzed.Cox’s proportional hazard regression analysis was used to select influencing factors for prognosis,with 5-year survival as variable,and to establish death risk model.Results Univariate Cox’s proportional hazard regression analysis showed that age〉60years,high performance status score,peripheral white blood cell counts 〉50×109/L at the diagnosis of AML,abandoning treatment,failure of inductive treatment with daunorubicin plus cytarabin for 4weeks,and palliative treatment increased the 5-year death risk for patients with AML.And hemoglobin 〉80g/L at the diagnosis of AML was a protective factor for the prognosis.Multivariable Cox’s proportional hazard regression analysis showed that the age（RR=1.872）,physical performance status score（RR=1.269）,hemoglobin level at diagnosis（RR=0.585）,peripheral white blood cell counts（RR=1.838）and treatment scheme（RR=1.651）were the independent factors for the prognosis.The death risk model was that h（t,X）/h0（t）=Exp（0.627age＋0.238 performance status score-0.536hemoglobin＋0.609 white blood cell counts＋0.501 treatment scheme）.By using the model for the prognosis of death,AUC was 0.867,larger than the age（0.725）,hemoglobin（0.698）,physical performance status score（0.682）,white blood cell counts（0.634）and treatment scheme（0.587）.Conclusion The death risk model established with covariates of age,hemoglobin,performance status score,white blood cell counts and treatment scheme can provide diagnostic value for the prognosis of AML.