Genome-wide Association Study for AKI

BACKGROUND
While common genetic risks for chronic kidney disease are well established, genetic factors influencing risk for acute kidney injury (AKI) in hospitalized patients are poorly understood.


METHODS
We conducted a genome-wide association study in 1,369 participants in the Assessment, Serial Evaluation, and Subsequent Sequelae of AKI (ASSESS-AKI) Study; a multi-ethnic population of hospitalized participants with and without AKI matched on demographics, comorbidities and kidney function prior to hospitalization. We then completed functional annotation of top-performing variants for AKI using single cell RNA sequencing data from kidney biopsies in 12 AKI patients and 18 healthy living donors from the Kidney Precision Medicine Project (KPMP).


RESULTS
No genome-wide significant associations with AKI risk were found in ASSESS-AKI (p < 5 x 10-8). The top two variants with the strongest association with AKI mapped to the dispatched RND transporter family member 1 (DISP1) gene and toll like receptor 5 (TLR5) gene locus, rs17538288 (OR=1.55, 95% CI:1.32-182, p=9.47 x 10-8) and rs7546189 (OR=1.53, 95% CI:1.30-1.81, p=4.60 x 10-7). In comparison to kidney tissue from healthy living donors, kidney biopsies in patients with AKI showed differential DISP1 expression in proximal tubular epithelial cells (adjusted p=3.9 x 10-2) and thick ascending limb of the loop of Henle (TAL) (adjusted p=8.7 x 10-3) and differential TLR5 gene expression in TAL (adjusted p=4.9 x 10-30).


CONCLUSIONS
AKI is a heterogeneous clinical syndrome with various underlying risk factors, etiologies and pathophysiology that may limit the identification of genetic variants. While no variants reached genome wide significance, we report two variants in the intergenic region between DISP1 and TLR5, suggesting this region as a novel risk for AKI susceptibility.

. Demographics and kidney function in living donor and AKI participants who underwent kidney biopsy for single-cell RNAseq analyses Table S7. Differential DISP1 gene expression between living donor and AKI biopsies. Table S8. Differential TLR5 gene expression between living donor and AKI biopsies.
The SNP rs80052123 (hg19 chr1:89493224) is in the 5' region of guanylate binding protein 3 (GBP3). GBP3 is expressed in the GTEx kidney cortex data and significantly differentially expressed in association with this SNP in 21 tissues, the most significant is Adipose Viseral (Omentum) Tissue (NES 0.39, p=6.1 x 10 -11 ). rs80052123 is also associated with differential expression of GBP7 in 12 tissues with the most significant also being Adipose Viseral (Omentum) Tissue (NES 0.45, p=1.1x10 -11 ). The gene KYAT3 is also differentially expressed in association with rs80052123 in Brain Putamen basal ganglia (NES 0.45, p=0.000011) and Brain Caudate basal ganglia (NES 0.4, p=0.000018) tissues. See Supplemental Excel file (GTEx_AKI_GWAS_hits_eQTLs.xlsx) for all tissue results and Figure S5 for a selection of tissue specific GTEx boxplots. In the NephQTL data rs80052123 is associated with differential expression of GBP3 (Beta 0.39, p=0.016) and GBP4 (Beta 0.26, p=0.032) in the Glomerulus (See Figure S6 for boxplots) and GBP3 is suggestive (Beta 0.22, p=0.058) in tubulointerstitium.
The SNP rs7546189 (hg19 chr1:223224864) is also in the 3' region of DISP1. DISP1 is significantly (NES 0.16, p=0.000099) differentially expressed in the Left Heart Ventricle tissue in association with rs7546189 (See Supplemental Figure S3 for Boxplots). rs7546189 is also associated with differential expression of the other nearby gene TLR5 (NES -0.29, p=3.50x10 -10 ) in Left Heart Ventricle tissue.
The SNP rs1368999 (hg19 chr9:28879870) is in an intron of leucine rich repeat and Ig domain containing 2 (LINGO2). LINGO2 is expressed in the GTEx kidney cortex data and significantly differentially expressed in association with this SNP in Nerve Tibial tissue (NES 0.26, p=1.3x10 -6 ). See Figure S9 for rs1368999 GTEx boxplots. In the NephQTL data rs1368999 is associated with differential expression of the nearby (~9kb) gene MIR873 (Beta 0.312, p=0.0162) in glomerulus ( Figure S10). LINGO2 is genomically overlapping with MIR873 sharing sequence.
Among our most associated SNP loci, all showed significant cis eQTL effects except rs4414368 (hg19 chr13:83974202) which is in a gene desert and not associated with differential expression of current GTEx and NepthQTL annotated transcripts. The gene SLIT and NTRK-like family, member 1 (SLITRK1) is 478,855 bp and the lincRNA LINC00377 is ~2,360,000 bp away toward the centromere. SNP -single nucleotide polymorphism, MA -minor allele, MAF -minor allele frequency, OR -odds ratio, CI-confidence interval.
The outcome is modeled as an ordinal variable with (0=No AKI),1=KDIGO Stage 1, 2=KDIGO Stage 2, and 3=KDIGO Stage 3)      Adjusted p-value based on a Bonferroni correction using all genes. Adjusted p-value based on a Bonferroni correction using all genes.