Accepted Manuscript
Total Page:16
File Type:pdf, Size:1020Kb
Accepted Manuscript Genome-wide association study in Guillain-Barré syndrome Stefan Blum, Ying Ji, David Pennisi, Zhixiu Li, Paul Leo, Pamela McCombe, Matthew A. Brown PII: S0165-5728(18)30033-X DOI: doi:10.1016/j.jneuroim.2018.07.016 Reference: JNI 476822 To appear in: Journal of Neuroimmunology Received date: 23 January 2018 Revised date: 27 July 2018 Accepted date: 27 July 2018 Please cite this article as: Stefan Blum, Ying Ji, David Pennisi, Zhixiu Li, Paul Leo, Pamela McCombe, Matthew A. Brown , Genome-wide association study in Guillain-Barré syndrome. Jni (2018), doi:10.1016/j.jneuroim.2018.07.016 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. ACCEPTED MANUSCRIPT Genome-wide Association Study in Guillain-Barré Syndrome Stefan Bluma,1, Ying Jib, c,1, David Pennisib, Zhixiu Lib, Paul Leob, Pamela, McCombea,2 and Matthew A. Brownb,2 1 These authors contributed equally. 2 These authors contributed equally. aThe University of Queensland, UQ Centre for Clinical Research, Royal Brisbane & Women's Hospital, Brisbane, Australia. bTranslational Genomics Group, Institute of Health and Biomedical Innovation, School of Biomedical Sciences, Queensland University of Technology (QUT) at Translational Research Institute, Brisbane, Australia. cDepartment of Neurology, Peking University Third Hospital, Beijing, P. R. China. *Correspondence to: Professor Matthew A. Brown, Translational Genomics Group, Institute of Health and Biomedical Innovation, School of Biomedical Sciences, Queensland University of Technology (QUT) at Translational Research Institute, Brisbane, Australia Telephone: +61ACCEPTED 7 34437017 MANUSCRIPT Fax: +61 7 34437099 [email protected] ACCEPTED MANUSCRIPT ABSTRACT Guillain-Barré syndrome (GBS) is considered to have an immune-mediated basis, but the genetic contribution to GBS is unclear. We conducted a GWAS involving 215 GBS patients and 1,105 healthy controls. No significant associations of individual SNPs or imputed HLA types were observed. We performed a genome-wide complex trait analysis for evaluation of the heritability of GBS, and found that common SNPs contribute up to 25% of susceptibility to the disease. Genetic risk score analysis showed no evidence of overlap in genetic susceptibility factors of GBS and multiple sclerosis. Given the unexplained heritability of the trait further larger GWAS are indicated. KEYWORDS: Guillain-Barré syndrome; case-control study; genome-wide association study; heritability; genetic risk score; human leukocyte antigen (HLA) genes. ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT 1. INTRODUCTION Guillain-Barré syndrome (GBS) is a severe acute ascending peripheral neuropathy causing weakness of the limbs. GBS with weakness can be subdivided into acute inflammatory demyelinating polyradiculoneuropathy (AIDP), acute motor axonal neuropathy (AMAN), and acute motor and sensory axonal neuropathy (AMSAN). There are also disorders that are regarded as GBS variants, such as Miller Fisher syndrome (MFS) or focal variants of GBS; [1]. The pathogenesis of GBS is incompletely understood. It is considered to be mediated by immune-mediated responses, triggered by infection, with two-thirds of patients experiencing a preceding bacterial or viral infection [2, 3]. However, unlike classical autoimmune diseases, GBS is typically monophasic, and is more common in males, and does not have a clear HLA association, as required by the Rose and Bona criteria for autoimmune diseases [4]. GBS has pathological similarities to multiple sclerosis (MS), another inflammatory demyelinating neurological disease. However, in MS the central rather than peripheral nervous system is targeted [5]. MS is a highly heritable disease, found to be associated with more than 140 genetic variants [6]. A potential genetic basis for GBS, however, is less clear. No systematic twin study has been performed. Increased familiality of the disease has been reported, as well as evidence of anticipation with younger generations developing disease at an earlier age, suggesting a genetic mechanism [7ACCEPTED-10]. MANUSCRIPT Many candidate gene studies have been performed in GBS, most focusing on human leukocyte antigen (HLA) genes, the most important genetic regulators of the immune system, as well as other genes [11]. HLA genes are encoded within the Major Histocompatibility Complex (MHC), a highly polymorphic region that is essential in ACCEPTED MANUSCRIPT the regulation of immunity and infection, and is closely linked with most autoimmune diseases [12]. However, to date, most studies of HLA associations with GBS have been negative [13-15], with no positive findings being consistently replicated between studies. A recent meta-analysis has suggested association of TNF genetic variants with disease, but this locus is known to be markedly affected by population stratification, which was not controlled for in the meta-analysis [16]. Similarly, multiple association studies of non-MHC candidate genes have been reported but none has achieved definitive levels of association, nor been consistently replicated. None of these studies has controlled for population stratification, and all have been modest in size. Therefore, we conducted the present genome-wide association study (GWAS) to explore the possible genetic basis of GBS more extensively. ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT 2. METHODS 2.1 Demographics A total of 215 patients with confirmed diagnosis of GBS were included in our study (Table 1). GBS was defined according to the National Institute of Neurological Disorders and Stroke criteria [17], and the majority of patients had level 1 or level 2 certainty of diagnosis according to the Brighton criteria [18](Table 1). 1,105 samples were collected as control group. All samples came from centres in Brisbane, Townsville, Sunshine Coast and Sydney, Australia. Both case and control subjects were of Caucasian ethnicity. Informed written consent was obtained from all participants. This study was approved by the Human Research Ethics Committee of the Royal Brisbane and Women's Hospital (approval number HREC/QRBW/31). 2.2 Quality control and association analysis Genotyping was performed in an ISO15189-accredited clinical genomics facility, Australian Translational Genomics Centre (ATGC), Queensland University of Technology. All samples were genotyped by Illumina HumanOmniExpress (OmniExpress) BeadChip. Quality control (QC) was performed using the PLINK 1.9 package (https://www.cog-genomics.org/plink2)[19, 20] and Shellfish (http://www.stats.ox.ac.uk/~davison/software/shellfish). Detailed information regarding QC filtering is described below (Table 2). We included all 22 autosomal chromosomes in our study. For individuals, samples with a missing genotype rate higher than 10%ACCEPTED or extreme heterozygosity MANUSCRIPT (±3SD from the mean) were excluded. Pairwise identity by descent (IBD) was used to detect cryptic relatedness, with samples with an IBD > 0.185 excluded from further analysis. For markers, SNPs with a call rate less than 95% or with significant differences (P < 10−5) in missing genotype rate between cases and controls were removed. SNPs identified with extensive deviation (P < 10−6) from Hardy-Weinberg equilibrium (HWE) and minor allele ACCEPTED MANUSCRIPT frequency (MAF) of <5% were excluded. Differential ethnicity was detected by principal component analysis (PCA) using Shellfish (Supplementary Figure S1). After merging with HapMap genotypes, samples not clearly consistent with a Caucasian ethnicity were excluded. Then a second PCA was conducted to identify outliers, defined as more than ±6SD from the mean for the first principal component, to correct for population stratification. After QC steps were performed, the genotype-phenotype association was tested using logistic regression analysis with the first principal component as the covariant using PLINK. Genomic inflation was assessed by quantile-quantile (Q-Q) plot and the genomic inflation factor, λ1000. 2.3 Imputation of SNPs in the human leukocyte antigen region SNPs in the HLA region were specifically imputed by the SNP2HLA software [21], with The National Institute of Diabetes and Digestive and Kidney Diseases Type 1 Diabetes Genetics consortium dataset (https://www.niddkrepository.org/studies/t1dgc/) used as the reference panel. The logistic regression analysis was performed with dosage data after removing poorly imputed markers and sample outliers. All analyses were performed following the standard procedures according to the SNP2HLA instructions (http://broadinstitute.rog/mpg/snp2hla/). 2.4 Estimation of the narrow-sense heritability (h2) Genome-wide Complex Trait Analysis (GCTA) was applied to estimate the variance explained by allACCEPTED the SNPs [22]. Since we MANUSCRIPTwere conducting a case-control study, we performed a further, more stringent, QC for this analysis. Markers with a call rate < 95%, or individuals with a genotype missing rate > 1%, were excluded, and 0.05 was chosen as the cutoff for IBD and the P-value for HWE [23]. Genetic relationship matrix (GRM) from all autosomal SNPs was assessed to remove samples with cryptic relatedness (0.025