Genome-Wide Association Study of Glomerular Filtration Rate in a Cohort of Sickle Cell Disease Patients

BACKGROUND: Nephropathy is a common and devastating complication of sickle cell disease (SCD) associated with mortality (Elmariah et al, 2014). However, early detection of SCD nephropathy (SCDN) has proven difficult. Discovery of genetic markers associated with SCDN could greatly improve our ability to identify patients at risk for renal decline. To that end, we have performed a genome-wide association study (GWAS) of glomerular filtration rate (GFR) in our adult SCD cohort.

METHODS: Medical history, laboratory values and DNA for genotyping were collected as part of a multicenter study of outcome-modifying genes in SCD. Participating institutions included sickle cell centers at Duke University, University of North Carolina at Chapel Hill, Emory University, and East Carolina University. GFR was estimated using the ‘Modification of Diet in Renal Disease’ (MDRD) study definition (Levey et al, 1999) and dichotomized at the clinically relevant threshold of 90 ml/min/1.73m². 463 patients with complete data were included in the analysis (54.6% female, mean age 34.2 ± 12 years, 17.5% GFR < 90 ml/min/1.73m²). Genotyping was performed using the Illumina Human610-Quad BeadChip (Illumina, San Diego, CA), and a global reference panel from the 1000 Genomes project was used to impute genotypes not covered on the GWAS chip. Samples were pre-phased with SHAPEIT (Delaneau et al, 2012) and genotypes inferred using IMPUTE2 (Howie et al, 2009). After data cleaning and removal of single nucleotide polymorphisms (SNPs) with minor allele frequency (MAF) < 5%, 3,021,990 SNPs were available for analysis. Principal component (PC) analysis was performed to obtain measures of population stratification in our data set using EIGENSOFT (Patterson et al, 2006). Logistic regression was utilized to test for association between each SNP and dichotomized GFR, controlling for genome-wide PCs using PLINK (Purcell et al, 2008). False discovery rate (FDR) p-values were generated using PROC MULTTEST in SAS (SAS Systems, Cary, NC).

RESULTS: Because GWAS chips capture common variation, several of the most highly associated SNPs were intergenic. The SNP with the most evidence for association that also lies within a gene was rs72777730 in XYLT1 (p=6.8E^-6). For each additional C allele, the odds of having GFR < 90 ml/min/1.73m² increased by 2.7. Other top hits included rs1990628 in ZNF423 (p=2.7E^-5) and rs6449202 in SLC2A9 (p=3.8E^-5). None of the findings met FDR significance. To further investigate our findings, we assessed the role of rare variants on renal function in our data set. We observed SNPs in XYLT1 from the Illumina HumanExome BeadChips associated with proteinuria (p=0.008) using the gene-based SKAT test (Wu et al, 2011). Further, we performed whole-exome sequencing of 15 patients with disparate GFR levels using VCRome 2.1 capture and 100bp paired-end reads generated from an Illumina HiSeq. One novel non-synonymous SNP (chr16:17353261, C>T) was identified in XYLT1 in 6 of 7 patients with low GFR (≤ 70 ml/min/1.73m²) but was absent in patients with normal GFR (> 90 ml/min/1.73m²). Additionally, four known non-synonymous SNPs were observed in low GFR patients but were absent in those with normal GFR levels.

DISCUSSION: Xylosyltransferase 1 (XYLT1, also known as XT-1) encodes the enzyme responsible for biosynthesis of heparin sulfate proteoglycans, which affect permeability of the glomerular basement membrane. The c.343G>T polymorphism in XYLT1 has been implicated in diabetic nephropathy (Schön et al, 2006), and XT-1 is up-regulated in rats with adriamycin nephropathy (Sichuan et al, 2007). Zinc finger protein 423 (ZNF423) is involved in DNA damage response signaling, and mutations in ZNF423 have been identified in patients with nephronophthisis and Joubert syndrome (Chaki et al, 2012). Finally, solute carrier family 2 member 9 (SLC2A9, also known as GLUT9) is a transporter involved in urate reabsorption in the renal tubules. Mutations in this gene have been identified in patients with renal hypouricemia-2 (Matsuo et al, 2008). Thus, while none of the SNPs reached genome-wide significance, likely due to the modest size of our data set, we have identified several promising candidate genes nominally associated with renal function in our cohort of SCD patients and demonstrate that both common and rare variation in these genes likely contributes to risk for renal decline.