基于NGS技术建立HLA-DPA1和DPB1连锁预测模型及验证其临床应用价值  

Establishment of an HLA-DPA1 and DPB1 linkage prediction model based on NGS technology and validation of its clinical application value

作　　者：张腾腾[1] 刘双 袁晓妮[1] 李杨[1] 姜雪 杨天杰[1] 鲍晓晶[1] 何军[1] Zhang Tengteng;Liu Shuang;Yuan Xiaoni;Li Yang;Jiang Xue;Yang Tianjie;Bao Xiaojing;He Jun(HLA Laboratory of Jiangsu Institute of Hematology,the First Affiliated Hospital of Soochow University,Suzhou215031,China;Department of Hematology,the First Affiliated Hospital of Soochow University,Suzhou215031,China)

机构地区：[1]苏州大学附属第一医院,江苏省血液研究所HLA配型实验室,苏州215031 [2]苏州大学附属第一医院血液科,苏州215031

出　　处：《中华检验医学杂志》2024年第11期1292-1298,共7页Chinese Journal of Laboratory Medicine

基　　金：国家自然科学基金(82070180);江苏高校血液学协同创新中心(SX21100121)。

摘　　要：目的建立人类白细胞抗原(HLA)DPA1~DPB1连锁预测模型, 并采用非血缘异基因造血干细胞移植供受者的临床资料和随访数据验证, 探讨该预测模型对判断移植预后的临床应用价值。方法回顾性研究。基于NetMHCⅡpan的人工神经网络算法、实验室已建立的中国人群DPA1~DPB1连锁单体型数据库、最新国际免疫遗传学/人类白细胞抗原(IPD-IMGT/HLA)数据库公布的序列已知的DPA1~DPB1氨基酸FASTA数据, 建立47种DPA1~DPB1连锁模型。采用基于杂交捕获文库构建法的二代测序技术, 对2016年1月至2021年9月间在苏州大学附属第一医院血液科进行无关供者造血干细胞移植的250对供受者进行HLA-A、-B、-C、DRB1、DQB1、DQA1、DRB3/4/5、DPB1、DPA1(9个位点)HLA基因分型检测。回顾性分析移植供受者的HLA分型数据和临床资料, 验证及预测采用模型分析的DPA1~DPB1连锁可允许错配与不可允许错配对移植预后的影响。采用Kaplan-Meier 法Log-Rank检验比较不同组间总生存(OS)率的生存曲线;采用竞争风险模型比较不同组之间Ⅱ~Ⅳ度急性移植物抗宿主病和非复发死亡(NRM)的累积发生率。采用ROC曲线下面积比较建立的预测模型与T细胞表位(TCE)模型的预测性能。结果根据氨基酸的亲水与疏水特性不同将模型分为Ⅰ~Ⅳ型:Ⅰ型P1~P8位疏水型加P9位亲水型6种, Ⅱ型为疏水型17种, Ⅲ型为两亲型9种, Ⅳ型为亲水型15种。按预测模型将供受者DPA1相合、DPB1错配病例分为P1相合或不相合者, 与DPA1和DPB1全相合者比较发现, P1不相合患者2年OS率为75%(12/16)比96.2%(25/26)(χ^(2)=4.13, P=0.04), NRM率为4/16比0(χ^(2)=7.05, P<0.01);而P1相合的患者与DPA1和DPB1全相合者相比, 2年OS率和NRM率差异无统计学意义(P>0.05)。本文建立的预测模型与DPB1 的TCE模型相比, 预测移植后2年OS率的ROC 曲线下面积更大(Z=0.71, P=0.48)。在供受者DPA1和DPB1均错配的病例中, 与DPA1和DPB1�Objective To establish a linkage prediction model for human leukocyte antigen(HLA)DPA1-DPB1 and validate it by using clinical data and follow-up data from unrelated allogeneic hematopoietic stem cell transplantation donors and recipients,and to explore the clinical application value of the prediction model in transplantation prognosis.Methods This is a retrospective study.Leveraging the artificial neural network algorithm of NetMHCⅡpan and the DPA1-DPB1 haplotype linkage database of the Chinese population established in our previous research,and incorporating the amino acid FASTA data of DPA1-DPB1 of all known sequences newly published by the Latest International Immunogenetics/Human Leukocyte Antigens,47 DPA1-DPB1 linkage models were established.Employing next-generation sequencing technology based on the hybridization capture library construction method,HLA genotyping tests for HLA-A,-B,-C,DRB1,DQB1,DQA1,DRB3/4/5,DPB1,and DPA1(9 loci)were performed on 250 donor-recipients pairs who underwent unrelated-donor hematopoietic stem cell transplantation in the Department of Hematology of the First Affiliated Hospital of Soochow University between January 2016 and September 2021.HLA typing data and clinical information of transplant donors and recipients were retrospectively analyzed to assess and predict the impact of permissive and non-permissive linkage mismatches of DPA1-DPB1 on transplantation prognosis.The Kaplan-Meier method with the log-rank test was applied to compare the survival curves of overall survival(OS)rates between different groups.Additionally,a competing risks model was utilized to compare the cumulative incidence of gradeⅡ-Ⅳacute graft-versus-host disease and non-relapse mortality(NRM)across groups.The area under the receiver operating characteristic curve was employed to compare the predictive performance of the established prediction model with that of the T-cell epitope(TCE)model.Results According to the different hydrophilic and hydrophobic properties of amino acids,the DPA1-DPB1 linkage

关 键 词：NGS 人类白细胞抗原 DPA1 DPB1 连锁 造血干细胞移植 

分 类 号：R457.7[医药卫生—治疗学]