APOL1 gene variants and kidney disease in whites: the cardiovascular health study

Two coding variants in the APOL1 gene confer a substantially increased risk of nondiabetic kidney disease in African Americans [1]. These variants, a set of two amino acid substitutions termed G1 and a 6-base pair deletion termed G2, are present in about 23 and 14% of African American chromosomes, respectively, and are absent in those without recent African ancestry [2]. Recent reports have described an association between these APOL1 kidney risk variants and cardiovascular disease, but results are inconsistent [3].

Substantial coding variation exists in APOL1, located on chromosome 22 and encompassing 14 522 base pairs and seven exons, but no variants beyond G1 and G2 have been found to be associated with focal segmental glomerulosclerosis, human immunodeficiency virus-associated nephropathy or end-stage renal disease [4, 5]. However, it remains unknown if other variants in APOL1 are associated with subclinical degrees of kidney dysfunction, which may be more sensitive markers, or with other potential phenotypes among individuals without recent African ancestry. Here we describe a project examining variations in APOL1 and risk for kidney and cardiovascular disease. To address these questions, we tested the associations between a set of candidate single-nucleotide polymorphisms (SNPs) in the APOL1 gene among white participants of the Cardiovascular Health Study (CHS), a richly phenotyped cohort of community-dwelling older adults and validated all nominally significant findings in external genome-wide association studies.

We included 3943 white CHS participants who had consented for genetic information and for whom genotyping was available on the 50 K ITMAT-Broad-CARe (IBC) array. The IBC array was previously designed to assay SNPs in candidate genes and pathways for cardiovascular, inflammatory and metabolic phenotypes [6]. Among whites, the array genotyped 16 SNPs in APOL1. To reduce collinearity and increase power to detect possible associations (by reducing multiple testing), we used the 1000 Genomes Pilot 1 to identify linkage disequilibrium (LD) [7]. Using an R² threshold of >0.7 and a minor allele frequency (MAF) of >1%, tag SNPs with the highest MAF were chosen from four linkage disequilibrium (LD) groups for analysis: rs4820224, rs136175, rs9610467 and rs4419330. Detailed methods are available in Supplementary data, Item S1. Assuming an additive mode of inheritance, we used linear regression to test the association of each SNP with log-transformed albuminuria, estimated glomerular filtration rate (eGFR)–cystatin C and ankle–brachial index (ABI), with adjustment for age, sex, clinic site and ancestry, using the first three principal components for ancestry. Multivariable Cox proportional hazards analyses with sex-specific hazards were used to test the association of each SNP with incident cardiovascular disease and mortality.

In the experimental sample, the average age was 72.8 ± 5.6 years, 43.8% were male, 63.5% had prevalent hypertension and 14.6% had prevalent diabetes (see Supplementary data, Table SA). The baseline eGFR was 73.8 ± 18.2 mL/min/1.73 m² and albuminuria was 55.6 ± 260.4 mg/g of creatinine. In fully adjusted models, each minor allele of rs4820224 was associated with a 12% decrease in albuminuria [odds ratio (OR) 0.88, P = 0.039] and each minor allele of rs136175 was associated with a 10% decrease in albuminuria (OR 0.10, P = 0.032) (Table 1). No SNP was associated with eGFR or ABI. There were no significant associations with incident cardiovascular disease or all-cause mortality (see Supplementary data, Table SC).

Table 1.

Association of APOL1 SNPs with kidney and cardiovascular disease traits in white CHS participants

	Gene-wise P-value	rs9610467	rs4820224	rs4419330	rs136175
Albuminuria, β	0.0330	0.0382	−0.1276	−0.0047	−0.1017
(n = 2070)		(−0.0580–0.1344)	(−0.2491 to −0.0061)	(−0.1586–0.1493)	(−0.1948 to −0.0086)
P-value		0.436	0.039	0.953	0.032
eGFR, β	0.3171	−0.2016	0.5825	−1.3596	0.6487
(n = 3442)		(−1.2095–0.8064)	(−0.7002–1.8653)	(−3.0160–0.2969)	(−0.3182–1.6176)
P-value		0.695	0.373	0.108	0.188
ABI (continuous), β	0.3895	0.0047	0.0024	0.0023	−0.0077
(n = 3669)		(−0.0040–0.0133)	(−0.0085–0.0133)	(−0.0164–0.0118)	(−0.0159–0.0006)
P-value		0.288	0.665	0.746	0.068
ABI <0.9	0.7252	0.9041	0.9669	1.1098	1.1075
OR (n = 3669)		(0.7483–1.0924)	(0.7672–1.2188)	(0.8336–1.4776)	(0.9053–1.2774)
P-value		0.296	0.776	0.476	0.408

Adjusted for age, sex, clinic site and first three principal components of ancestry.

We sought to validate the associations between rs4820224 and rs136175 and albuminuria with existing genome-wide association studies (GWAS) data. We identified two GWAS of albuminuria [8, 9] in the genome-wide repository of associations between SNPs and phenotypes [10], including a meta-analysis of data from participants of European ancestry in the CKDGen and CARe Consortia [9]. Neither rs4820224 nor rs136175 was associated with urine albumin excretion (see Supplementary data, Table SD). In addition, none of the other 31 SNPs in the APOL1 gene region were associated with urine albumin excretion (see Supplementary data, Table SD).

The strengths of this study include the use of a well-phenotyped cohort and a large GWAS meta-analysis for replication. Limitations include missing data for albuminuria in a subset of the CHS sample and the use of previously genotyped SNPs rather than whole-exome sequencing of the APOL1 gene.

In conclusion, we identified two SNPs in APOL1 that were significantly associated with albuminuria among CHS participants of European ancestry, but neither SNP was validated in an existing GWAS (nor were any other SNPs identified) and no associations were observed with clinical or subclinical cardiovascular disease. Our findings cast doubt on the possibility of common genetic variants in APOL1 in persons of European ancestry that associate with kidney or cardiovascular disease. Additional studies are required to evaluate other variants not tested in this study as well as rarer variants.

SUPPLEMENTARY DATA

Supplementary data are available at ndt online.

FUNDING

This research was supported by contracts HHSN268201200036C, HHSN268200800007C, HHSN268201800001C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083 and N01HC85086 and grants U01HL080295 and U01HL130114 from the National Heart, Lung, and Blood Institute, with additional contributions from the National Institute of Neurological Disorders and Stroke. Additional support was provided by R01AG023629 from the National Institute on Aging. A full list of principal CHS investigators and institutions can be found at CHS-NHLBI.org

AUTHORS’ CONTRIBUTIONS

E.R.D. and K.J.M. developed the research idea and study design and completed statistical analyses. D.J.F. and M.R.P. provided input. All authors approved the study and analysis plan and contributed to the interpretation and discussion of data. E.R.D. drafted the manuscript and all authors provided review and final approval of the manuscript to be published. This content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

CONFLICT OF INTEREST STATEMENT

The results presented in this article have not been published previously in whole or part, except in abstract format. M.R.P. and D.J.F. have filed patents related to ApoL1-associated kidney disease and own equity in ApoLo1 Bio, LLC. All other authors have no conflicts to declare.