Transcriptome-wide association study for persistent hepatitis B virus infection and related hepatocellular carcinoma
Transcriptome-wide association study for persistent hepatitis B virus infection and related hepatocellular carcinoma
复制标题
持续性乙型肝炎病毒感染与相关肝细胞癌的全转录组相关性研究
DOI：
10.1111/liv.14577
复制
发表时间：
2020-07-13
影响因子：
6.7
通讯作者：
Hu, Zhibin
中科院分区：
文献类型：
--
作者：
Han, Jing;Chen, Congcong;Hu, Zhibin
背景与目标 先前的全基因组关联研究 (GWAS) 已鉴定出与持续性乙型肝炎病毒 (HBV) 感染相关的多种易感变异。然而，这些变异大多数位于非编码区，这使得很难确定这些关联背后的有效基因。我们进行了两个阶段的研究，第一阶段我们将肝组织的RNA测序数据和来自基因型组织表达（GTEx）项目的高密度基因分型数据与我们之前的GWAS数据相结合，进行乙型肝炎病毒感染的全转录组关联研究（TWAS）。首先，通过全基因组复杂性状分析（GCTA）的遗传相关性矩阵从 GTEx 数据中筛选顺式遗传基因。然后，在 951 例 HBV 携带者病例和 937 例 HBV 清除对照的 GWAS 数据中，通过全基因组有效混合模型关联 (GEMMA) 的限制最大似然 (REML) 预测了 2587 个顺式遗传基因的基因表达。接下来，我们研究了预测表达水平与持续 HBV 感染风险之间的关联。应用基因集富集分析（GSEA）来推断已识别基因的功能。为了确定 HBV 感染风险的因果单核苷酸多态性 (SNP)，我们在这些基因的 1 Mb 左右区域进行了基于表达数量性状位点 (eQTL) 的逐步逻辑回归分析，并通过基因分型实验验证了 994 名健康对照者与 994 名 HBV 持续感染病例之间的关联。第二阶段，收集了 1538 例 HBV 相关肝细胞癌 (HCC) 病例和 1465 例持续性 HBV 感染对照，以确定这些变异对 HBV 相关 HCC 的影响，并通过逻辑回归分析中的加性模型进行检验。我们鉴定了七个与 HBV 感染相关的基因。在经典人类白细胞抗原（HLA）区域，以往的GWAS报道了3个新基因BAK1、HLA-DOB和C4A（Z范围为-3.95至-3.64，P范围为7.84 x 10(-5)至2.00 x 10(-4)），以及2个基因（HLA-DPA1和HLA-DPB1）。在非HLA区域，在新识别的位点PARP9(Z = 3.69,P = 2.20 x 10(-4)) at3q21.1处与免疫相关。在22q11.21，除了报道的非顺式遗传基因UBE2L3之外，我们还鉴定了TMEM191A(Z = 3.55,P = 3.80 x 10(-4))作为目标基因。经过进一步的逐步逻辑回归分析和验证，我们确定了与持续性 HBV 感染独立相关的八种变异。其中，加法模型显示两个SNP与HBV相关HCC风险相关（rs9272714和rs9394194，OR范围为1.20至1.25，P范围为1.19 x 10(-4)至3.97 x 10(-4)）。通过整合转录组数据，我们的研究不仅确定了持续性乙型肝炎病毒感染的新易感位点，还确定了已报告位点的潜在靶基因，从而深入了解持续性乙型肝炎病毒感染和相关肝癌的遗传病因学。
Background & Aims Previous genome-wide association studies (GWAS) have identified multiple susceptible variants associated with persistent hepatitis B virus (HBV) infection. However, most of these variants are located in the noncoding regions, which make it difficult to determine the effective genes underlying these associations. We performed a two-stage study, in the first stage we integrated RNA sequencing data of liver tissues and high-density genotyping data from the Genotype-Tissue Expression (GTEx) project with our previous GWAS data to conduct a transcriptome-wide association study (TWAS) on HBV infection. Firstly, the cis-heritable genes were screened by a genetic relatedness matrix of genome-wide complex trait analysis (GCTA) from GTEx data. Then, the genetic expression of 2587 cis-heritable genes was predicted by restricted maximum likelihood (REML) of genome-wide efficient mixed-model association (GEMMA) in our GWAS data with 951 HBV carrier cases and 937 HBV cleared controls. Next, we investigated the associations between predictive expression levels and persistent HBV infection risk. Gene set enrichment analysis (GSEA) was applied to infer the function of the identified genes. To identify the causal single nucleotide polymorphisms (SNPs) of HBV infection risk, we conducted the expression quantitative trait loci (eQTL)-based stepwise logistic regression analysis in the regions around 1 Mb of these genes and validated the association between 994 health controls and 994 HBV-persistent infection cases by genotyping experiment. In the second stage, 1538 HBV-related hepatocellular carcinoma (HCC) cases and 1465 persistent HBV infection controls were collected to determine the effect of these variants on HBV-related HCC as well, which were examined by the additive model in logistic regression analysis. We identified seven genes associated with HBV infection. In the classic human leukocyte antigen (HLA) region, three novel genesBAK1, HLA-DOBandC4A(Z range from -3.95 to -3.64,Prange from 7.84 x 10(-5)to 2.00 x 10(-4)), as well as two genes (HLA-DPA1andHLA-DPB1) were reported by previous GWAS. In the non-HLA region, immune related at newly identified loci,PARP9(Z = 3.69,P = 2.20 x 10(-4)) at3q21.1. At22q11.21, we identifiedTMEM191A(Z = 3.55,P = 3.80 x 10(-4)) as a target gene in addition to the reported non-cis-heritable geneUBE2L3. After further stepwise logistic regression analysis and validation, we identified eight variants independently associated with persistent HBV infection. Among those variants, the additive model showed that two SNPs associated with HBV-related HCC risk (rs9272714 and rs9394194, OR range from 1.20 to 1.25,Prange from 1.19 x 10(-4)to 3.97 x 10(-4)). By integrating transcriptome data, our study not only identified new susceptibility loci of persistent HBV infection but also determined the potential target genes at reported loci, which provided insight into the genetic aetiology of persistent HBV infection and related HCC.