Subjects
In total, the data of our previous vitiligo GWAS from 1117 vitiligo patients and 1701 controls were used [11,12], while 2069 cases and 1370 controls were recruited for replication in this study (Table 1). All (the Chinese Han population) were collected from the multiple hospitals in China. After the written informed consent was obtained from all individuals, the blood samples, clinical and demographic information (a previous designed questionnaire) were collected from cases and controls. Based on the Declaration of Helsinki principles, this study was approved by the Institutional Ethical Committee of Anhui Medical University. According to the diagnostic criteria of the Vitiligo European Task Force [14], the clinical diagnosis of all cases were confirmed by at least two dermatologists. All controls were healthy individuals without vitiligo, family history (including first-, second- and third-degree relatives) of vitiligo, and any other autoimmune diseases or systemic disorders. Common factors (ethnicity, age and gender) were matched in case and control groups. According to the manufacturer’s instructions, genomic DNA was extracted from peripheral blood lymphocytes by using QIAamp DNA Blood kit (Qiagen, Valencia, CA, USA) and was diluted to working concentrations of 20-25 ng/ml for the replication study.
SNP selection for replication
Based on our previous vitiligo GWAS [11,12] and the reference panel of 1000 Genomes Project (Mar.2012 release) [15], the imputation for a expanded region of 22q12 locus (expansion of 5 Mb of upstream and downstream for this locus, chr 22: 20000000-42000000, hg19) was performed in 1117 vitiligo patients and 1701 controls by using Impute v2.0 [16]. The following conditions will be excluded from further analysis: 1) SNPs with low imputation confidence (INFO score ≤0.5), 2) Significant Hardy-Weinberg disequilibrium (P < 0.05), 3) A MAF < 1%, 4) A low call rate < 95%. At last, 8 SNPs at the region showed nominal significance of association (Pinitial < 0.01) in our vitiligo GWAS and were then selected for this replication study.
Genotyping analysis of the replication study
The detection primers for the 8 SNPs were designed using the MassARRAY Assay Design 3.0 software (Sequenom). Approximately 15-20 ng of genomic DNA for each sample was used to genotype. Eight SNPs were genotyped by using the Sequenom MassArray iPLEX1 system (the Key Laboratory of Dermatology, Ministry of Education, China). The DNA samples were amplified by multiplex PCR reactions, and then the PCR products were used for locus-specific single-base extension reactions. The resulting products were desalted and transferred to a 384-element SpectroCHIP array. Allele detection was performed using MALDI-TOF MS. The mass spectrograms were analyzed by the MassARRAY Typer software (Sequenom). SNPs with MAF <1%, a call rate <95%, and deviation from Hardy-Weinberg equilibrium (P < 0.05) in the control subjects were excluded from further association analysis.
Statistical analysis
Association of each SNP with disease phenotype was tested in the discovery and replication samples using PLINK version 1.07 [17]. Principal component analysis was used to assess population outlier and stratification in the dataset [11]. GWAS and replication association analyses were carried out by using the Cochran-Armitagetrend test. Conditional logistic regression was used to determine whether independent effects existed and was carried out by SNPTEST (V2). The regional plot of association result was generated using LocusZoom based on the information of JPT and CHB populations from hg19/HapMap Phase II. The level of associated significance was assigned at P values of less than 5.0×10-8 (criterion for genome-wide significance). The sub-phenotype analyses of the associated SNPs were also performed according to the clinical features which included gender, family history, age of onset (early onset <=20 years and late onset >20 years), clinical classification (non segmental and segmental) and autoimmune disease involvement (such as, systemic lupus erythematosus, alopecia areata, thyroid disease, rheumatoid arthritis, myasthenia gravis, and scleroderma), P values of less than 5.0×10-2 were considered to be statistically significant.
Bioinformatics analysis
Several bioinformatics tools were utilized in this study. The HaploReg4.1 (https://pubs.broadinstitute.org/mammals/hap loreg/haploreg.php) was adopted to select the strong linked SNPs and evaluated the potential biological significance for targeted SNPs. Single Nucleotide Polymorphism database (dbSNP) was used for gene mapping (http://www.ncbi. nlm.nih.gov/snp). In addition, the expression quantitative trait loci (eQTL) studies data based on the Genotype-Tissue Expression (GTEx) database, version 7 (http://www.gtexportal.org/home/) were adopted (Consortium, 2013).