SEARCH

SEARCH BY CITATION

Keywords:

  • genetics;
  • genome-wide association;
  • pathogenesis;
  • systemic lupus erythematosus

Abstract.

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. References

We review the systemic lupus erythematosus (SLE) human genetics literature, including the first wave of genome-wide associations scans (GWAS), to identify confirmed and candidate risk variants that meet stringent statistical criteria. The understanding of the genetic basis of SLE in humans has expanded dramatically over the past year, offering an early glimpse into the primary genetic factors and major dysregulated pathways. A meta-analysis of published candidate variants was performed incorporating data from a 1310 case and 7859 control GWAS. Our review of the literature and meta-analysis identifies a total of 17 well-validated common SLE risk variants, including four candidate variants that achieve our definition of a confirmed SLE risk locus. These variants account for a fraction of the total genetic contribution to SLE risk, with many risk loci remaining to be identified, but may provide insight into the pathways involved in SLE. Initial pathway analyses of the 17 confirmed SLE risk alleles indicate an important role for B-cell signalling and development, signaling through toll-like receptors 7 and 9, and neutrophil function.


Introduction

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. References

Systemic Lupus Erythematosus (SLE) is a chronic inflammatory disease mediated by autoantibodies to nuclear and tissue specific antigens. Risk to SLE has a significant genetic component, as evidenced by familial aggregation and twin-studies. However, the relative contribution of copy number variants (CNVs), rare (<1%) and common (>1%) single nuclear polymorphisms (SNPs) to the genetic component of SLE is unclear. Comprehensive studies of CNVs in SLE are expected in the coming years and are eagerly anticipated given the important role CNVs appear to play in several common diseases [1, 2]. In SLE, studies of CNVs in the Fc receptor region and Complement Factor 4 in the HLA region are compelling, but a definitive role for the CNV has not been convincingly disentangled from nearby, linked risk variants [3, 4]. A number of rare variants that cause SLE in a Mendelian manner have been identified throughout the years, including disruption of several complement pathway components [5]. The Mendelian forms of SLE shed light onto pathways critical in pathogenesis, but account for only a small portion of the overall disease incidence [5]. More recently, rare variants that alter the amino acid sequence of TREX1 were found to be enriched in some patients with sporadic SLE [6]. The finding of rare variants influencing risk to SLE has generated excitement for large-scale exonic re-sequencing via 2nd generation technologies in SLE cases.

The field has made significant progress recently in identifying common SLE risk alleles, with at least 13 reports of variants that reach a genome-wide level significance (P < 5 × 10−8) [7–17]. Prior to 2008, candidate gene studies identified a modest number of common genetic variants that were reproducibly associated with SLE, including the HLA-Class II haplotypes, Fc Receptor 2 alpha (FcR2A), Interferon Regulatory Factor 5 (IRF5), Protein Tyrosine Phosphotase 22 (PTPN22) and Signal Transducer and Activator of Transcription 4 (STAT4). Progress in technology over the past 5 years has resulted in the availability of genome-wide SNP genotyping systems that can interrogate ∼50–80% of the common genetic variation, depending on genotyping platform and study population. In 2008, a series of genome scans in SLE was published which identified multiple novel SLE risk loci [10–12, 18]. Here, we review the SLE human genetics literature, including recent genome-wide scans, and summarize confirmed and candidate risk variants. In addition, we perform a meta-analysis of published candidate variants incorporating data from a 1310 case and 7859 control GWAS [11], resulting in four additional confirmed SLE loci. Finally, we ask whether the newly identified SLE risk loci provide new insights into SLE pathogenesis.

Methods

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. References

Literature review of reported SLE risk variants

We systemically examined the SLE genetics literature to identify loci that meet strict criteria for disease association. Specifically, we defined a ‘confirmed’ locus as one where two independent studies have reported an association to SLE with a P ≤ 1 × 10−5. A meta-analysis of two reports with a P ≤ 1 × 10−5 corresponds to a P-value no >2.4 × 10−9 using Fisher’s combined probability test, a more conservative criterion than the generally accepted genome-wide significance value of P < 5 × 10−8. The identical variant (or proxy with r2 > 0.3) showing association to SLE with the same direction of effect was required. These criteria may exclude some true disease risk loci where replication data are not yet available.

In addition, we identified ‘candidate’ alleles for which there was a single publication reporting an association with a P ≤ 1 × 10−5. The search included several variants with genome-wide significant P-values from a recent GWAS, but where inconsistent evidence of replication in the original study was reported [10].

Meta-analysis of candidate SLE risk variants

For the ‘candidate’ SLE risk loci (one report with a P ≤ 1 × 10−5), a meta-analysis was performed using the original report and data from a GWAS consisting of 1310 SLE cases and 7859 controls [11]. Additional details of the samples and analysis of the GWAS are provided below.

The association summary statistics from the 1310 SLE case and 7859 control dataset was examined for the ‘candidate’ SLE risk alleles. Whenever possible, the exact variant was included in the meta-analysis, however variants with an R2 > 0.75 (as determined from the CEU population of the Phase II HapMap) to a genotyped variant were also examined. For four of the candidate loci (CRP, SELP, PDCD1, and TYK2; see Table 2), no suitable proxy variants were available in our dataset and these loci were excluded from the meta-analysis.

Table 2.   Systemic lupus erythematosus (SLE) risk loci with one published report with P ≤ 1 × 10−5. Loci with a Meta P ≤ 5 × 10−8 were considered confirmed
LocusbChromosomeReportCurrent cases seriesaMeta P
AlleleP-valueReferenceSNPr2 to allele in ReportP-value
  1. SNPs, single nuclear polymorphisms.

  2. aA total of 1310 SLE cases and 7859 controls genotyped using the Illumina 550K SNP array.

  3. bThe symbol for a single gene from each locus is indicated. In some cases, the block of linkage disequilibrium includes multiple genes.

PTTG15q33.3rs24316971.0 × 10−10[10]rs24316971.003.3 × 10−65.3 × 10−14
ATG56q21rs65684311.7 × 10−8[10]rs65684311.005.5 × 10−62.7 × 10−12
IRAK1Xq28rs20755962.8 × 10−7[17]rs22693680.791.1 × 10−51.4 × 10−11
TNFSF41q25.1rs120399044.3 × 10−7[7]rs104892650.918.7 × 10−61.0 × 10−10
PHRF111p15.5rs49631283.0 × 10−10[10]rs49631281.003.1 × 10−31.0 × 10−9
UBE2L322q11.21rs57542177.5 × 10−8[10]rs57542171.006.4 × 10−37.3 × 10−9
BANK14q24rs105164873.7 × 10−10[12]rs105164871.000.0961.0 × 10−8
PXK3p14.3rs64459757.1 × 10−9[10]rs64459751.000.0101.0 × 10−8
FCGR2A1q23.3rs18012746.8 × 10−7[10]rs18012741.004.1 × 10−43.9 × 10−8
NMNAT21q25.3rs20220131.1 × 10−7[10]rs20220131.000.155.1 × 10−6
ICA17p21.3rs101560911.9 × 10−7[10]rs101560911.000.0952.0 × 10−5
LYN8q12.1rs78298165.4 × 10−9[10]rs78298161.000.483.6 × 10−3
SCUBE122q13.2rs20717251.2 × 10−7[10]rs20717251.000.638.3 × 10−3
ITPR36p21.31rs37480792.9 × 10−8[19]rs37480791.000.95
CRP1q23.2rs30930616.4 × 10−7[36] 
SELP1q24.2rs39178155.7 × 10−6[37] 
PDCD12q37.3rs115688211.0 × 10−5[38] 
TYK219p13.2rs23042562.2 × 10−8[16] 

The corrected meta-analysis association statistic was determined by the summing of the Z-scores weighted for cohort size for the current case series (using methodology described in [11]) and the reports from Kozyrev et al., Sawalha et al. and Oishi et al. [12, 17, 19]. For the family-based study described by Cunninghame-Graham et al., [7] the meta-analysis was conducted using Fisher’s combined probability test. It should be noted that some of the control samples from iControlDB (∼3589 samples) and a small number of (∼70) SLE cases in this dataset overlap with samples used in the SLEGEN study [10]. The meta-analysis for these alleles was therefore conducted by merging the SLE cases from the Harley et al. report and the current cases series and calculating the association statistic relative to the 7859 controls described above.

Genome-wide association scan

Genotype data used in the meta-analysis was from 1310 SLE cases genotyped with the Illumina 550K genome-wide SNP platform [11]. The selection and genotyping of the SLE case samples were described previously [11]. All SLE cases were North Americans of European descent, as determined by self-report and confirmed by genotyping. The diagnosis of SLE (fulfillment of four or more of the American College of Rheumatology [ACR] defined criteria [20]) was confirmed in all cases by medical record review (94%) or through written documentation of criteria by treating rheumatologists (6%). Clinical data for these case series were presented elsewhere [15, 21–24]. In addition to the 3583 controls previously described [11], 4564 control samples from the publicly available Cancer Genetics Markers of Susceptibility (CGEMS) project were included after obtaining approval (http://cgems.cancer.gov/). The entire sample of 7859 controls was examined using data quality control filters and association testing methodology as previously described [11].

Sample and SNP filtering was conducted using analytical modules within the software programs PLINK [25] and EIGENSTRAT [26], as described below. The genome-wide SNP data were used in this study to facilitate close matching of cases and controls, and to provide genotypes at the confirmed and suspected SLE loci.

SLE cases, NYHP samples, and iControlDB samples.  The Illumina 550K SNP array, version 1 (HH550v1; Illumina Inc., San Diego, CA, USA) was used to genotype 464 cases and 1962 controls, and the Illumina 550K SNP array, version 3 (HH550v3) was used to genotype 971 cases and 1621 controls as described [11]. Samples where the reported sex did not match the observed sex (HH550v1: 10, HH550v3: 11) and samples with >5% missing genotypes (HH550v1: 25, HH550v3: 21) were excluded from the analysis. Cryptic relatedness between the SLE cases and controls was determined by the estimation of the identity-by-state (IBS) across the genome for all possible pair-wise sample combinations. A sample from each pair estimated to be duplicates or 1st–3rd degree relatives were excluded (Pi_hat ≥ 0.10 and Z1 ≥ 0.15; HH550v1: 88, HH550v3: 73).

Single nuclear polymorphisms with HWE P ≤ 1 × 10−6 in controls (HH550v1: 3176, HH550v3: 2240) and SNPs with >5% missing data (HH550v1: 12605, HH550v3: 7137) were removed. The SNPs were tested for a significant difference in the frequency of missing data between cases and controls, and SNPs with P ≤ 1 × 10−5 in the differential missingness test implemented in PLINK were removed (HH550v1: 5027, HH550v3: 2804). The SNPs were also tested for a significant allele frequency difference between genders; all SNPs had P ≥ 1 × 10−9 in controls. The data were examined for the presence of batch effects (for example, between ABCoN samples and all other cases), and SNPs with an allele frequency difference with a P < 1 × 10−9 were excluded (HH550v1: 18, HH550v3: 10). Variants with heterozygous haploid genotypes were set to missing (HH550v1: 2305, HH550v3: 875). In addition, variants with a minor allele frequency <0.0001 were removed (HH550v1: 97, HH550v3: 57).

CGEMS samples.  For the 2277 prostate cancer samples and, separately, 2287 breast cancer samples, heterozygous haploid genotypes were set to missing (prostate: 2717, breast: 0). Samples where the reported gender did not match the observed gender (prostate: 0, breast: 2) and samples with >5% missing data (prostate: 15, breast: 1) were excluded. Samples were tested for cryptic relatedness, as described above, and we removed one sample from each pair estimated to be duplicates or 1st–3rd degree relatives (Pi_hat ≥ 0.10 and Z1 ≥ 0.15; prostate: 12, breast: 7). SNPs with a MAF < 0.0001 (prostate: 3254, breast: 2166) were removed.

All samples.  Additional data quality filters were applied to the merged dataset consisting of all SLE cases and controls. SNPs with >5% missing data (n = 65 421) and samples with >5% missing data (n = 0) were removed. A test for duplicate samples was conducted using 957 independent SNPs with MAF ≥ 0.45, and no duplicate samples were found. SNPs with HWE P ≤ 1 × 10−6 in controls (n = 2174) and SNPs with >2% missing data (n = 5522) were removed. We tested the SNPs for a significant difference in the proportion of missing data between cases and controls and removed SNPs with excess missing data differential (P ≤ 1 × 10−5, n = 16080). SNPs were tested for a significant difference between genders and all SNPs had P ≥ 1 × 10−9 in controls. SNPs were also examined for the presence of batch effects; in particular, between CGEMS breast cancer samples and all other controls, and between CGEMS prostate cancer samples and all other controls and removed SNPs with P < 1 × 10−9 (n = 73). After application of the above quality filters, 480 831 SNPs remained.

The cases and controls were tested for the presence of population outliers using EIGENSTRAT. SNPs with MAF < 2% in cases (n = 16 068), HWE P ≤ 1 × 10−4 in controls (n = 977), or >1% missing data (n = 17 029); SNPs in regions of abnormal LD patterns due to structural variation on chromosomes 6 (from 24–36 Mb), 8 (8–12 Mb), 11 (42–58 Mb), and 17 (40–43 Mb); and SNPs in the pseudoautosomal region of chromosome X (n = 12) were excluded for the purpose of determining the principal components (EIGENSTRAT) of variation to detect population outliers. Samples with >6 SD from the mean along any of the top 10 principal components were removed (n = 148).

The final data set had 1310 cases, 7859 controls, and 480 831 SNPs. The final genomic control inflation factor (λgc) [27] was 1.06, indicating adequate matching of cases and controls.

Results

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. References

A systematic review of reported SLE risk variants

Over the past 35 years, hundreds of journal articles have been published investigating associations between candidate variants and SLE risk. Many loci are reported to be associated with SLE, however studies claiming a disease association had variable power to detect the modest effects typically seen in complex disease and may have been confounded by population stratification or technical artefacts. These experimental challenges and variable standards for reporting results resulted in a perceived irreproducibility of candidate gene association studies in complex disease [28], and led to calls for more stringent levels of statistical significance in the initial report and replication in independent cohorts and studies [29]. We therefore applied conservative criteria for our definition of a confirmed SLE risk locus, acknowledging that some true disease risk loci where replication data are not yet available may be excluded.

We first identified loci with two independent published reports in nonoverlapping SLE cohorts, each with a P ≤ 1 × 10−5 (corresponding to a P-value of 2.4 × 10−9 using Fisher’s combined probability test) (Table 1). Eight alleles fulfilled the requirements, including HLA-DRB1*0301 (HLA-DR3) [30, 31], HLA-DRB1*1501 (HLA-DR2) [30, 31], Protein Tyrosine Phosphatase Nonreceptor type 22 (PTPN22) [10, 32], Interferon Regulatory Factor 5 (IRF5) [8, 16], Signal Transducer and Activator of Transcription 4 (STAT4) [10, 15], B Lymphoid tyrosine Kinase (BLK) [10, 11], Integrin Alpha M (ITGAM) [11, 14], and Tumour Necrosis Factor Alpha Interacting Protein 3 (TNFAIP3) [18, 33]. The identical allele or best proxy (r2 > 0.85) in our genome-wide data set of 1310 SLE cases and 7859 controls was examined in the meta-analysis (Table 1). For the loci defined by ITGAM (Arg77His), PTPN22 (R620W) and IRF5 (multiple functional alleles), excellent evidence for the causal alleles exists, but their exact role in SLE biology has not been fully elucidated. For the remaining loci, additional resequencing and functional studies will be required to identify the causal allele(s).

Table 1.   Confirmed systemic lupus erythematosus (SLE) risk loci based on presence of two published reports with P ≤ 1 × 10−5
LocusChromosomeReport 1Report 2Additional referencesCurrent cases seriesa
AlleleP-valueReferenceAlleler2 to allele in Report 1P-valueReferenceSNPr2 to allele in Report 1
  1. SNPs, single nuclear polymorphisms.

  2. a1310 SLE cases and 7859 control.

PTPN221p13.2rs24766011.0 × 10−5[32]rs24766011.005.2 × 10−6[10][13]rs24766011.00
STAT42q32.2rs75748651.9 × 10−9[15]rs75748651.002.8 × 10−9[10][11]rs75748651.00
HLA-DR26p21.32DRB1*15011.0 × 10−5[30]DRB1*15011.001.0 × 10−7[31][5]rs31298600.97
HLA-DR36p21.32DRB1*03011.0 × 10−6[30]DRB1*03011.001.0 × 10−5[31][5, 10, 11, 34]rs21876680.87
IRF57q32.1rs20046405.2 × 10−8[16]rs20046401.004.4 × 10−16[8][9–11, 35]rs10488631
BLK8p23.1rs132771131.0 × 10−10[11]rs69851090.332.5 × 10−11[10] rs132771131.00
ITGAM16p11.2rs11436796.9 × 10−22[14]rs115746373.0 × 10−11[11][10]rs98887390.86
TNFAIP36q23.3rs50299372.9 × 10−12[18]rs22309260.953.0 × 104[33] rs50299371.00

Meta-analysis of published SLE loci

In an effort to confirm additional risk loci, a meta-analysis of candidate SLE risk loci identified from a literature search was performed using data from a large genome-wide association study [11]. We define a ‘candidate’ locus as a locus where a single study has reported a variant associated with disease with a P ≤ 1 × 10−5 (Table 2). A total of 18 independent variants met the criteria, including several variants with genome-wide significant P-values from a recent GWAS, but where inconsistent evidence of replication in the original study was reported [10]. The reported disease associated variant was compared with genotype data from 1310 SLE cases genotyped with the Illumina 550K genome-wide SNP platform. In addition to the 3583 controls previously described [11], 4564 control samples from the publicly available Cancer Genetics Markers of Susceptibility (CGEMS) project were included after obtaining approval (http://cgems.cancer.gov/). A total of 7859 controls were examined using the data quality control filters and association testing methodology described above. Whenever possible, the exact variant was included in the meta-analysis, however variants with an R2 > 0.75 (as determined from the CEU population of the Phase II HapMap) to a genotyped variant were also examined. For five of the candidate loci, no suitable proxy variants were available in our dataset and these loci were excluded from the meta-analysis (Table 2).

For five of the candidate loci, we found no evidence of association in our dataset (Tables 2 and 3). While a definitive exclusion of these loci is not possible with the available data, at a minimum these results suggest that the initial observation may have been an overestimate of the true effect size and could indicate a false positive result in the original study.

Table 3.   Association statistics for 17 confirmed systemic lupus erythematosus (SLE) risk alleles in a genome-wide association scan of 1310 SLE cases and 7859 controls
LocusbChromosomeSNPPositiona(Mb)Minor alleleAllele frequencyP-valueOdds ratio (95% CI)
CaseControl
  1. The variants are ordered by P-value.

  2. SNPs, single nuclear polymorphisms.

  3. aPositions are from NCBI Build 35.

  4. bThe symbol for a single gene from each locus is indicated. In some cases, the block of linkage disequilibrium includes multiple genes.

HLA-DR36p21.32rs218766832.714T0.1900.1179.5 × 10−251.76 (1.58–1.97)
IRF57q32.1rs10488631128.188C0.1700.1091.4 × 10−191.68 (1.50–1.89)
STAT42q32.2rs7574865191.790T0.3120.2352.5 × 10−141.48 (1.34–1.64)
ITGAM16p11.2rs988873931.221T0.1750.1272.3 × 10−111.46 (1.31–1.63)
BLK8p23.1rs1327711311.387A0.2940.2421.7 × 10−81.30 (1.19–1.43)
PTTG15q33.3rs2431697159.813C0.3890.4383.3 × 10−60.82 (0.75–0.89)
ATG56q21rs6568431106.695A0.4230.3765.5 × 10−61.22 (1.12–1.32)
TNFSF41q25.1rs10489265169.968C0.2780.2388.7 × 10−61.24 (1.09–1.30)
PTPN221p13.2rs2476601114.090A0.1160.0898.9 × 10−61.35 (1.18–1.54)
IRAK1Xq28rs2269368152.711T0.1750.1411.1 × 10−51.29 (1.15–1.45)
FCGR2A1q23.3rs1801274158.293A0.4630.5004.1 × 10−40.86 (0.79–0.94)
TNFAIP36q23.3rs5029937138.236G0.0500.0321.0 × 10−41.48 (1.12–2.20)
KIAA154211p15.5rs49631280.580T0.3030.3333.1 × 10−30.87 (0.80–0.96)
UBE2L322q11.21rs575421720.264T0.2150.1926.4 × 10−31.15 (1.04–1.27)
PXK3p14.3rs644597558.345G0.3050.2810.0101.13 (1.03–1.23)
HLA-DR26p21.32rs312986032.509A0.1600.1470.0921.10 (0.98–1.24)
BANK14q24rs10516487103.108A0.2880.3040.0960.93 (0.85–1.01)

Nine candidate loci had a genome-wide significant meta-P-value (<5 × 10−8) and were considered confirmed (Table 2). Of note, included in the nine confirmed loci were four candidate loci that did not reach genome-wide significance (P < 5 × 10−8) in the original report (Table 2). Loci defined by interleukin-1 receptor-associated kinase 1 (IRAK1), TNF Super Family 4 (TNFSF4, OX40L), ubiquitin-conjugating enzyme E2L 3 isoform 1 (UBE2L3) and Fc fragment of IgG, low affinity IIa receptor (FCGR2A), represent newly confirmed SLE risk variants under our criteria. We provide additional evidence for loci defined by PHD and ring finger domains 1 (PHRF1, aka KIAA1542), pituitary tumour-transforming protein 1 (PTTG1), autophagy protein 5 (ATG5), PX domain containing serine/threonine kinase (PXK) and B-cell scaffold protein with ankyrin repeats 1 (BANK1) (Table 2). A single gene from each associated locus was selected for clarity, however the reported risk variant may be linked to additional genes in the region. In Table 3, detailed summary statistics for the 17 confirmed SLE risk alleles are provided.

Discussion

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. References

Biological insights into disease pathogenesis

The first wave of genome-wide association scans has identified at least 17 ‘confirmed’ risk loci which provide an informative first glimpse of the primary common genetic factors that influence risk to SLE. However, there are several challenges to make biological insights from genetic studies of complex disease in humans. First, because only a limited set of variants are directly genotyped, the reported variant is unlikely to be the causal variant and rather is more likely to be in strong linkage disequilibrium with the biologically relevant variant. As a result, the disease-associated variant is linked to multiple variants that may include several genes. A complete catalogue of common human variation is currently unavailable, and will require deep re-sequencing of each locus to ensure discovery of all relevant variation. Complicating matters further, the disease-linked variants may not have an obvious biological impact and may require extensive experimentation to determine the causal allele and its role in SLE pathogenesis.

Despite these important challenges, several insights in SLE pathogenesis may be gained from the genome-wide association scans. With a small number of exceptions, the loci implicated by the genome-scans are immune related and can arguably be placed into several discrete pathways (Table 4). Our preliminary assessment of the 17 confirmed SLE risk alleles results in a nearly equal split between genes involved in the innate and adaptive immune system. A distinct clustering of genes involved in TLR7/9 signaling pathways and B/T cell signaling is evident (Table 4). The genetics appears to strongly support a primary role of dysregulated recognition of nucleic acid and the production of Type I-Interferons, B and T cell signaling and neutrophil function. The loci without an obvious immunological function are of significant interest and may lead to novel pathways and mechanisms at play in SLE (Table 4).

Table 4.   Preliminary assignment of systemic lupus erythematosus (SLE) risk loci to biological pathways
 ChromosomeGenesSNP
  1. SNPs, single nuclear polymorphisms.

Adaptive immunity
 Antigen presentation 6p21.32HLA-DR3rs2187668
6p21.32HLA-DR2rs3129860
 B and T cell receptor signaling1p13.2PTPN22rs2476601
4q24BANK1rs10516487
8p23.1BLK,C8orf13rs13277113
 Helper T cell regulation2q32.2STAT4rs7574865
1q25.1TNFSF4 (OX40L)rs10489265
 
Innate immunity
 Interferon and TLR7/9 Signaling7q32.1IRF5rs10488631
6q23.3TNFAIP3rs5029937
Xq28IRAK1, MECP2rs2269368
22q11.21UBE2L3rs5754217
11p15.5PHRF1, IRF7rs4963128
 Fc Receptor1q23.3FCGR2Ars1801274
 Neutrophil activity16p11.2ITGAMrs9888739
 
Unknown5q33.3PTTG1, SLU7rs2431697
6q21ATG5, PRDM1rs6568431
3p14.3PXK, RPP14rs6445975

The analysis presented here identifies 17 alleles that attain an extremely high bar of statistical significance from the first round of genome-wide association scans. This analysis suggests an important role for several pathways contributing to SLE susceptibility including B-cell signaling and development, signaling through toll-like receptors 7 and 9, and neutrophil function.

Conflict of interest statement

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. References

Timothy Behrens and Robert Graham are full-time employees of Genentech Inc.

References

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Conflict of interest statement
  8. References