Jiangsu Institute of Hematology, The First Affiliated Hospital of Soochow University, No.188 Shizi Street, Suzhou, 215000, China.
Collaborative Innovation Center of Hematology, Soochow University, Suzhou, China.
Ann Hematol. 2021 Sep;100(9):2351-2361. doi: 10.1007/s00277-021-04520-0. Epub 2021 Apr 12.
Severe bleeding is a major cause of death in acute leukemia (AL) patients with graft-versus-host disease (GVHD) after allogene hematopoietic stem-cell transplantation (allo-HSCT). However, the prognostic value and prediction of HSCT-associated severe bleeding in GVHD patients have not been reported in cohort studies. We did a retrospective analysis of 200 AL patients with GVHD after allo-HSCT from Feb 1, 2014, to Dec 1, 2015. Multivariate analysis showed that the severe bleeding class was associated with the risk of death (HR 2.26, 95% CI 1.31-3.92, p<0.001***). In order to predict severe bleeding and figure out the solution to bleeding events, we established a multiple logistic regression model. HLA-DQB1 unmatching, megakaryocyte reconsititution failure, and III or IV GVHD were the independent risk factors for severe bleeding. Among all the variations above, OR of HLA-DQB1 was the highest (OR: 16.02, 95% CI: 11.54-48.68). Adding HLA-DQB1 to other factors improved the reclassification for predicting severe bleeding (NRI=0.195, z=2.634, p=0.008**; IDI=0.289, z=3.249, p<0.001***). Lasso regression was used to select variants. A nomogram of the logistic model was generated and displayed. Calibration curve demonstrated excellent accuracy in estimating severe bleeding (C index of 0.935). HLA-DQB1 showed excellent efficacy of predicting severe bleeding in HSCT patients.
严重出血是异基因造血干细胞移植(allo-HSCT)后移植物抗宿主病(GVHD)导致急性白血病(AL)患者死亡的主要原因。然而,在队列研究中尚未报道与 HSCT 相关的 GVHD 患者严重出血的预后价值和预测。我们对 2014 年 2 月 1 日至 2015 年 12 月 1 日接受 allo-HSCT 后发生 GVHD 的 200 例 AL 患者进行了回顾性分析。多变量分析表明,严重出血等级与死亡风险相关(HR 2.26,95%CI 1.31-3.92,p<0.001***)。为了预测严重出血并找到出血事件的解决方案,我们建立了一个多因素逻辑回归模型。HLA-DQB1 不匹配、巨核细胞重建失败和 III 或 IV 级 GVHD 是严重出血的独立危险因素。在所有这些变异中,HLA-DQB1 的比值比(OR)最高(OR:16.02,95%CI:11.54-48.68)。将 HLA-DQB1 添加到其他因素中可以提高预测严重出血的重新分类(NRI=0.195,z=2.634,p=0.008**;IDI=0.289,z=3.249,p<0.001***)。Lasso 回归用于选择变量。生成并显示了逻辑模型的列线图。校准曲线在估计严重出血方面表现出优异的准确性(C 指数为 0.935)。HLA-DQB1 显示出对 HSCT 患者严重出血预测的优异效果。