In the present study we identified four independent SNPs (rs9271366, rs204890, rs2071349, and rs2844580) in the MHC region associated with risk of SLE in a population of African American women. The four identified SNPs extend over a region of 1.7 Mb and they point to different signals that may contribute to SLE risk in African American women.
SNPs rs9271366 (the most significant signal in the present study) and rs2071349 are both located inside the HLA class II region that has been consistently associated with SLE in European, East Asian and African American populations (Fernando et al. 2007
; Han et al. 2009
; Hom et al. 2008
; Reveille et al. 1998
; Rioux et al. 2009b
; Uribe et al. 2004
). It is noteworthy that a recent GWAS found the rs9271366 SNP to be associated with SLE in Chinese Han subjects at genome-wide significance (Han et al. 2009
). The rs9271366 was part of a group of genome-wide significant SNPs near the HLA-DRB1
gene that belong to the same haplotype block and seem to be pointing to the same causal signal (Han et al. 2009
In populations of European ancestry, a major part of the signal from the HLA class II region comes from the presence of haplotypes carrying the HLA-DRB1
alleles (Graham et al. 2007
; Graham et al. 2002
). In the present study, both HLA class II alleles (as determined by tagging SNPs) were in low frequency and not associated with SLE, suggesting that the observed associations in the HLA class II region in African American women are not explained by classic class II alleles previously identified as risk factors in populations of European ancestry.
The rs204890 SNP is located in intron 6 of the ATF6B
gene that codes for a transcription factor involved in the unfolded protein response pathway during endoplasmic reticulum stress. It is noteworthy that the rs8283 SNP identified in a previous high-density screening of the MHC region (Barcellos et al. 2009
) is located in the 3′UTR region of the ATF6B
gene. Although our best proxy for rs8283, SNP rs429150, r2
=0.94, was not associated with SLE in the present study, both rs204890 and rs8283 are located in the same LD block in the HapMap CEU population suggesting that a single causal variant may be responsible for the rs204890 association in the present study and the rs8283 association in European ancestry subjects.
The rs2844580 SNP is located ~9 kb from the HLA-B
gene and ~38 kb from the MHC class I chain-related A (MICA
) gene. Genetic variation in both genes has been found to be associated with SLE in European ancestry populations (Gambelunghe et al. 2005
; Price et al. 1999
; Sanchez et al. 2006
; Smerdel-Ramoya et al. 2005
), the most consistent being the association with 8.1 ancestral haplotype (AH) (Candore et al. 2002
; Price et al. 1999
). The 8.1 AH consists of the HLA-A
, and DQA1
alleles (Ide et al. 2005
) and has been associated with different immune diseases in European populations (see review in (Candore et al. 2002
)). Because the 8.1 AH is specific to European populations, our findings show that other genetic variants in the HLA-B
neighborhood may contribute to SLE risk.
Our combined analysis suggests that the four newly identified SNPs (rs9271366, rs204890, rs2071349, and rs2844580) are independent signals of SLE risk in African American women. This conclusion is supported by the low linkage disequilibrium values among the four newly identified SNPs. In an analysis summing the risk of the four newly identified SNPs into a genetic risk score, the risk of SLE increased in a linear way according to the number of risk alleles. It is noteworthy that the estimated OR was robust to the use of different sets of SLE cases, suggesting that our results are not being distorted by bias related to survival from SLE or to specificity of case definition.
Comparison with previously published results suggests that some genetic risk factors for SLE may be shared among African Americans and other ethnic groups such as East Asian and European populations. Our most significant result, the rs9271366, was also found to be associated with SLE risk in Chinese subjects (Han et al. 2009
); the second SNP from the Chinese GWAS (rs3997854) was also associated with SLE risk in the BWHS although in an opposite direction: the C-allele was protective in Chinese subjects but harmful in the BWHS. Differences in allele frequencies and haplotype background may explain the opposite results for the rs39978534 SNP. The C-allele of the rs39978534 SNP has a frequency of 26% in the BWHS as compared to just 8% in Chinese Han subjects (Han et al. 2009
), suggesting that different C-allele carrying haplotypes may be present in the BWHS and they may have different effects on risk of SLE. We also found that the two reported SNPs from the SLEGEN GWAS (Harley et al. 2008
), rs3131379 and rs1270942 (tested with the perfect proxy rs440454), were also associated with increased risk of SLE in the present study. However, these two SNPs did not show independent effects in the BWHS after conditioning for the rs9271366 SNP. Although none of the SNPs from the high-density screening of the MHC region (Barcellos et al. 2009
) was replicated in the present study, some of the genetic regions were shared with BWHS subjects. For example, the rs204890 SNP reported in the present study and the rs8283 from the high-density screening are located both in the ATF6B
gene. Thus, genetic variation in the ATF6B
gene seems to affect risk of SLE in African American and European ancestry subjects.
To our knowledge this is the first screening of the MHC region in relationship to SLE in a population-based cohort of African American subjects. Our case-control study was derived from a well-characterized cohort, adding to the strength of the reported results. Control for degree of European ancestry and the case-control matching by region of residence make it unlikely that our results are due to population stratification. The classification of cases of SLE was based on the judgment of the women’s physicians as to the presence of ACR criteria for the illness. While these decisions may have varied across physicians, the fact that our most significant result (the rs9271366 SNP) was also reported in a previous GWAS in Chinese subjects and that we reproduced findings in significant SNPs from the SLEGEN study supports the validity of the BWHS case definition and results. Our study had 80% statistical power to identify ORs of 1.4 or greater for SNPs with minor allele frequencies >10% at MHC-wide significance level. Therefore, it is likely that rare variants or SNPs with smaller effects on the risk of SLE in the MHC region remain to be found.
In summary, our results suggest the presence of at least four independent signals in the MHC region associated with SLE in African American women, and they may be shared in part with other ethnic groups. Taken together, our results and previous GWAS in European and East Asian ancestry populations show that women of different ancestral origins may share some genetic components for the risk of SLE. The identity of the causal variants that are being tagged by the reported SNPs is still unknown. Further studies are needed to narrow the position of the potential causal variants.