摘要目的:探讨髓系白血病异基因造血干细胞移植（allo-HSCT）后髓外复发的危险因素及治疗策略。方法:回顾性分析2008年1月至2018年12月河南省肿瘤医院280例行allo-HSCT髓系白血病患者的临床资料，分析患者性别、原发病、移植前疾病状态、染色体、预处理方案、供者类型、移植前髓外浸润、移植物抗宿主病等指标。应用log-rank检验方法进行单因素分析，Cox比例风险模型进行多因素分析。结果:280例患者中髓外复发20例，复发率为7.14%（20/280），中位时间为移植后7.5 （1~123）个月。髓外复发患者16例死亡，病死率80%（16/20）。单因素分析中，疾病类别、第2次完全缓解（CR2）/进展期、移植前髓外浸润、预处理不含全身放疗、染色体异常与髓外复发显著相关（ P<0.05）。Cox回归显示，CR2/进展期（ RR=3.468， 95% CI 2.189~7.786）、染色体异常（ RR=1.494， 95% CI 1.020~2.189）、移植前髓外浸润（ RR=8.627， 95% CI 3.921~18.452）为髓外复发的独立危险因素。 结论:髓系白血病allo-HSCT后髓外复发的预后较差，联合治疗可能延长患者的生存期，早期评估预防至关重要。

abstractsObjective:To evaluate risk factors and available treatments of extramedullary relapse (EMR) after allogeneic hematopoietic stem cell transplantation (allo-HSCT) in patients with myeloid leukemia.Methods:A total of 280 patients were retrospectively analyzed from January 2008 to December 2018 in Affiliated Cancer Hospital of Zhengzhou University. Clinical data were collected including disease patterns, pre-transplantation status, chromosome karyotype, conditioning regimen, types of donor, extramedullary disease before transplantation and graft-versus-host disease (GVHD). The log-rank test and Cox proportional hazard model were uesd for univariate analysis and multivariate analysis, respectively.Results:Twenty patients developed EMR (7.14%). The median time of EMR was 7.5 (1-123) months after allo-HSCT. The mortality of EMR was 80% (16/20). Univariate analysis identified disease patterns, second complete remission (CR2) or progressive disease before transplantation, extramedullary disease, abnormal karyotype and conditioning regimen without total body radiation as significant factors correlated to EMR ( P<0.05). Multi-variable analysis revealed that CR2 or progressive disease ( RR=3.468,95% CI 2.189-7.786), abnormal karyotype ( RR=1.494,95% CI 1.020-2.189) and extramedullary disease before transplantation ( RR=8.627,95% CI 3.921-18.452) were independent risk factors of EMR. Conclusions:The clinical outcome of EMR after allo-HSCT is poor.It is crucial to comprehensively assess and identify EMR as early as possible.