BASIC RESEARCH www.jasn.org Lupus Nephritis Susceptibility Loci in Women with Systemic Lupus Erythematosus †‡ | Sharon A. Chung,* Elizabeth E. Brown, Adrienne H. Williams,§ Paula S. Ramos, ††‡‡ Celine C. Berthier,¶ Tushar Bhangale,** Marta E. Alarcon-Riquelme, Timothy W. || Behrens,§§ Lindsey A. Criswell,* Deborah Cunninghame Graham, F. Yesim Demirci,¶¶ ‡ †† †† †††‡‡‡ Jeffrey C. Edberg, Patrick M. Gaffney, John B. Harley, *** Chaim O. Jacob,§§§ †† ||| †† M. Ilyas Kamboh,¶¶ Jennifer A. Kelly, Susan Manzi, ¶¶¶ Kathy L. Moser-Sivils, **** †††† ‡‡‡‡ || Laurie P. Russell,§ Michelle Petri, Betty P. Tsao, Tim J. Vyse, Raphael Zidovetzki,§§§§ ‡ |||| Matthias Kreztler,¶ Robert P. Kimberly, Barry I. Freedman, Robert R. Graham,§§ and Carl D. Langefeld§ for the and International Consortium for Systemic Lupus Erythematosus Genetics Due to the number of contributing authors, the affiliations are listed at the end of this article. ABSTRACT Lupus nephritis is a manifestation of SLE resulting from glomerular immune complex deposition and inflammation. Lupus nephritis demonstrates familial aggregation and accounts for significant morbidity and mortality. We completed a meta-analysis of three genome-wide association studies of SLE to identify lupus nephritis–predisposing loci. Through genotyping and imputation, .1.6 million markers were assessed in 2000 unrelated women of European descent with SLE (588 patients with lupus nephritis and 1412 patients with lupus without nephritis). Tests of association were computed using logistic regression adjusting for population substructure. The strongest evidence for association was observed outside the 2 MHC and included markers localized to 4q11-q13 (PDGFRA, GSX2; P54.5310 7), 16p12 (SLC5A11; 2 2 2 P55.1310 7), 6p22 (ID4; P57.4310 7), and 8q24.12 (HAS2, SNTB1; P51.1310 6). Both HLA-DR2 and HLA-DR3, two well established lupus susceptibility loci, showed evidence of association with lupus ne- 2 phritis (P=0.06 and P53.7310 5, respectively). Within the class I region, rs9263871 (C6orf15-HCG22)had the strongest evidence of association with lupus nephritis independent of HLA-DR2 and HLA-DR3 2 (P58.5310 6). Consistent with a functional role in lupus nephritis, intra-renal mRNA levels of PDGFRA and associated pathway members showed significant enrichment in patients with lupus nephritis (n=32) compared with controls (n=15). Results from this large-scale genome-wide investigation of lupus nephritis provide evidence of multiple biologically relevant lupus nephritis susceptibility loci. J Am Soc Nephrol 25: ccc–ccc, 2014. doi: 10.1681/ASN.2013050446 SLE is a prototypic autoimmune disease that dispro- of women of European descent with SLE and ac- portionately affects women (ratio of female to male counting for significant morbidity and mortality.1 patients, 9:1). It is characterized by the development CKDs are heavily influenced by both genetic and ofautoantibodiesdirectedagainstnuclearandcellular environmental factors.2 A genetic component to LN components and the activation of inflammatory susceptibility is supported by an over-representation cascades, resulting in multisystem organ damage. of LN among children of parents with SLE, familial Lupusnephritis(LN)iscommon,affecting30%–40% aggregation of ESRD in African Americans with Received May 2, 2013. Accepted March 14, 2014. Correspondence: Dr. Carl D. Langefeld, Division of Public Health Sciences, Department of Biostatistical Sciences, Wake S.A.C. and E.E.B. are co-first authors. Forest School of Medicine, Medical Center Boulevard, Winston- Salem, NC 27157. Email: [email protected] Published online ahead of print. Publication date available at www.jasn.org. Copyright © 2014 by the American Society of Nephrology J Am Soc Nephrol 25: ccc–ccc, 2014 ISSN : 1046-6673/2511-ccc 1 BASIC RESEARCH www.jasn.org LN,3 linkage studies and candidate gene studies.4–7 Genome- To test whether the MHC associations observed with LN wide association studies (GWAS) have been extremely successful were influenced by linkage disequilibrium with HLA-DR2 and in identifying susceptibility loci for many disease phenotypes, HLA-DR3, each SNP in the MHC region was tested for asso- including SLE8–15 and related endophenotypes.16,17 Herein, we ciation while including the HLA-DR2 and HLA-DR3 tagSNPs report the first large-scale GWAS to identify loci predisposing as covariates in the logistic regression model (Figure 2B, Sup- to LN among women with SLE. plemental Table 3). In these analyses, the previously most significantly associated MHC SNP, rs9267972, was weakly asso- ciated with LN (OR, 1.35; 95% CI, 1.00 to 1.84; Padjusted=0.06). RESULTS The most significantly associated MHC locus independent of HLA-DR2 and HLA-DR3 was rs9263871, located within 26 Table 1 summarizes the clinical characteristics of the 2000 un- HCG27 (OR, 1.7; 95% CI, 1.35 to 2.13; Padjusted=8.5310 ; related women of European ancestry who meet the American Figure 2B, Supplemental Table 3) in the MHC class I re- College of Rheumatology (ACR) classification criteria for SLE gion. No other MHC SNPs were associated with LN at 25 and are included in this study. Of the 2000 patients with SLE, Padjusted,1.0310 . 588 had LN defined by ACR criteria and 1412 had SLE with no evidence of LN. Sets I and II are the largest contributors to the LN Associations outside the MHC meta-analysis and had comparable proportions of patients with Five loci outside the MHC region showed comparable or LN. Interestingly, the smallest sample, set III, had nearly double stronger evidence of association than those within the MHC 2 the proportion of LN and a slightly earlier median age of SLE (approximate P=1.0310 6;Figure3,Table2).Thestron- onset. Set II had the highest proportion of patients with SLE from gest statistical evidence of association with LN was with 2 multiplex pedigrees (set I, 0.08; set II, 0.82; and set III, 0.43). rs1364989 (OR, 3.41; 95% CI, 2.10 to 5.54; P=4.5310 7) After applying quality control measures, 1,621,689 single on 4q11-q13. This association localized to an intergenic re- nucleotide polymorphisms (SNPs) with genotype or imputed gion 83.5 kb upstream of PDGF receptor-a (PDGFRA). The genotype data were available for these 2000 unrelated indi- next strongest statistical evidence of association with LN was viduals. After adjusting for potential population substructure with rs274068, located within an intronic region of sodium- in each set via principal components, the meta-analysis infla- dependent glucose cotransporter SCL5A11 at 16p12.1 (OR, 2 tion factor was 1.05; sets I–III had inflation factors of 1.16, 2.85; 95% CI, 1.93 to 4.22; P=5.1310 7; Table 2). A potentially 1.09, and 1.03, respectively. The inflation factor for the best intriguing region of association was within 8q24.12 2 P value (minimum of dominant, additive, and recessive) was (rs7834765; OR, 3.15; 95% CI, 1.97 to 5.03; P=1.1310 6). 1.70. Supplemental Figure 1 shows the corresponding probability- Although the associated SNPs in this region are intergenic probability plot. Given the exploratory nature of the study, we and are not in linkage disequilibrium with any specific gene, report the results considering all genetic models (Table 2) as the region contains a DNase I hypersensitivity cluster and well as only the additive model (Supplemental Table 1). P values HAS2 (see the Discussion). The remaining two regions with 2 were genomic control adjusted based on the best P value infla- an approximate P=1310 6 were in intergenic regions on 6p22 tion factor. Supplemental Figure 2 presents the power analysis near ID4 and 9p21 (Table 2). Supplemental Tables 1 and 4 for the 588 patients with LN versus 1412 patients with SLE with- provide the summary statistics for the top non-MHC associa- out nephritis. Figure 1 shows a Manhattan plot of the associa- tions for the additive and all genetic models, respectively. tions across the genome. SLE Susceptibility Loci LN Associations within the MHC Given the frequency of LN among patients with SLE and the The MHC (chromosome 6: 25–32 Mb) is the most important manner in which SLE loci have been discovered (i.e., SLE cases region for SLE susceptibility, but its role in LN susceptibility is versus unaffected controls), some previously discovered SLE not established. In this study, the SNP most strongly associated susceptibility loci may be LN loci. This hypothesis was tested with LN within the MHC was rs9267972 (odds ratio [OR], 1.85; by taking 31 previously identified SLE susceptibility loci and 2 95% confidence interval [95% CI], 1.46 to 2.33; P=1.5310 6; testing for association with LN under the above case-only de- Figure 2A, Supplemental Table 2), which is located between sign (i.e., patients with LN versus patients with SLE without NOTCH4 and C6orf10. nephritis; Table 3). Three loci met the Bonferroni corrected 2 Because HLA-DRB1*1501 (HLA-DR2)andHLA- significance level (a=0.05/31=1.61310 3). The most strongly DRB1*0301 (HLA-DR3) are well established SLE susceptibility associated SLE susceptibility locus was rs2187668 within the 2 loci, these two loci were examined using their tagSNPs HLA region (OR, 1.55; 95% CI, 1.25 to 1.92; P=3.67310 5). rs9271366 (HLA-DR2) and rs2187668 (HLA-DR3).18 As as- This SNP tags the well known HLA-DR3 lupus-risk haplotype.18 sessed by these proxies, both HLA-DR2 (OR, 1.37; 95% CI, However, the magnitude of the OR for LN was significantly 1.09 to 1.71; P=0.06) and HLA-DR3 (OR, 1.55; 95% CI, 1.25 to lower than for SLE. The second most strongly associated SNP 2 1.92; P=3.7310 5) showed evidence of association with LN. was rs2205960 within the 1q25.1 region near TNFSF4 (OR, 2 Supplemental Table 2 summarizes the MHC associations.