Browse the corpus

APOL1 Bi- and Monoallelic Variants and Chronic Kidney Disease in West Africans. BACKGROUND: Apolipoprotein L1 gene (APOL1) variants are risk factors for chronic kidney disease (CKD) among Black Americans. Data are sparse on the genetic epidemiology of CKD and the clinical association of APOL1 variants with CKD in West Africans, a major group in the Black population. METHODS: We conducted a case-control study involving participants from Ghana and Nigeria who had CKD stages 2 through 5, biopsy-proven glomerular disease, or no kidney disease. We analyzed the association of CKD with APOL1 variants among participants with high-risk genotypes (two APOL1 risk alleles) and those with low-risk genotypes (fewer than two APOL1 risk alleles) by fitting logistic-regression models that controlled for covariates, including clinical site, age, and sex. RESULTS: Among 8355 participants (4712 with CKD stages 2 through 5, 866 with glomerular diseases, and 2777 with no kidney disease), the prevalence of monoallelic APOL1 variants was 43.0% and that of biallelic APOL1 variants was 29.7%. Participants with two APOL1 risk alleles had higher odds of having CKD than those with one risk allele or no risk alleles (adjusted odds ratio, 1.25; 95% confidence interval [CI], 1.11 to 1.40), as well as higher odds of focal segmental glomerulosclerosis (adjusted odds ratio, 1.84; 95% CI, 1.30 to 2.61). Participants with one APOL1 risk allele had higher odds of having CKD than those with no risk alleles (adjusted odds ratio, 1.18; 95% CI, 1.04 to 1.33), as well as higher odds of focal segmental glomerulosclerosis (adjusted odds ratio, 1.61; 95% CI, 1.04 to 2.48). The inclusion of covariates did not modify the association of monoallelic and biallelic APOL1 variants with CKD or focal segmental glomerulosclerosis. CONCLUSIONS: In this study, monoallelic APOL1 variants were associated with 18% higher odds of CKD and 61% higher odds of focal segmental glomerulosclerosis; biallelic APOL1 variants were associated with 25% higher odds of CKD and 84% higher odds of focal segmental glomerulosclerosis. (Funded by the National Human Genome Research Institute and others.).

The Human Health and Heredity in Africa (H3Africa) Kidney Disease Research Network is a member of the H3Africa consortium. Recruitment and data collection were described in detail earlier.16 For a summary, and the full study protocol and statistical analysis plan, see the Supplementary Appendix at NEJM.org. The study was approved by the Institutional Review Board of all participating Institutions; all participants gave written informed consent. The authors vouch for the accuracy and completeness of the data and the reliability of the study protocol. The contributions of the authors are described in the Supplementary Appendix. Persons with CKD were aged 1–74 years with CKD-EPI estimated glomerular filtration rate (eGFR) <90ml/min/1.73m2, and/or urine albumin/creatinine ratio (ACR) ≥3.0 mg/mmol (30 mg/g), and persons with biopsy confirmed glomerular diseases, pregnant women were excluded from the study (Supplementary Appendix). Study controls were participants without CKD defined as having an eGFR ≥90ml/min/1.73m2 and urine ACR <3.0mg/mmol (<30 mg/g). The study population is representative of the West African population affected by CKD with and without APOL1 CKD risk variants, and our findings are likely to be generalizable (Supplementary Appendix Table S1). APOL1 kidney risk variants G1 (rs73885319, p.S342G and rs60910145, p.I384M) and G2 (rs71785313, p.N388_Y389del) were genotyped using custom TaqMan assays (Applied Biosystems; see the Supplementary Appendix).

Persons with CKD were aged 1–74 years with CKD-EPI estimated glomerular filtration rate (eGFR) <90ml/min/1.73m2, and/or urine albumin/creatinine ratio (ACR) ≥3.0 mg/mmol (30 mg/g), and persons with biopsy confirmed glomerular diseases, pregnant women were excluded from the study (Supplementary Appendix). Study controls were participants without CKD defined as having an eGFR ≥90ml/min/1.73m2 and urine ACR <3.0mg/mmol (<30 mg/g). The study population is representative of the West African population affected by CKD with and without APOL1 CKD risk variants, and our findings are likely to be generalizable (Supplementary Appendix Table S1). APOL1 kidney risk variants G1 (rs73885319, p.S342G and rs60910145, p.I384M) and G2 (rs71785313, p.N388_Y389del) were genotyped using custom TaqMan assays (Applied Biosystems; see the Supplementary Appendix). Kidney biopsies were performed when the GFR was ≥15 and the ACR was greater than 50 mg/mmol or albuminuria ›500 mg/24 hours, renal biopsy samples were read after staining for light, immunofluorescence and electron microscopy at the University of Michigan and Massachusetts General Hospital, as detailed in theSupplementary Appendix.

ies were performed when the GFR was ≥15 and the ACR was greater than 50 mg/mmol or albuminuria ›500 mg/24 hours, renal biopsy samples were read after staining for light, immunofluorescence and electron microscopy at the University of Michigan and Massachusetts General Hospital, as detailed in theSupplementary Appendix. APOL1 risk status was defined according to the number of copies of the risk alleles (0, 1, or 2 copies). Our analyses included participants with missing data on age, sex and urine ACR; body mass index (BMI) and mean arterial pressure (MAP) used as covariates in association models. Proportion of missingness for any variable was less than 10% (Table 1). We performed multivariate imputation using the method of fully conditional specifications also known as multivariate imputation by chained equations. Logistic regression model was used for categorical variables, and the predictive mean matching method based on linear regression which imputes a value randomly from a set of observed values was used for imputing continuous variables each with specified predictors. For each imputation variable, 25 imputations were performed. We then fitted recessive genetic models to examine the association between APOL1 high risk variants and CKD, comparing participants with two copies of APOL1 risk alleles (high-risk) with participants with one or no copy of APOL1 risk allele (low risk). Adjustment covariates in the models include age, sex, clinical site and the baseline BMI, MAP, HIV status, diabetes mellitus status, and history of tobacco use Albuminuria was not treated as a covariate because it is part of the definition of CKD. We used multiple imputation with 25 imputations to replace missing values for BMI and MAP. Effect estimates from the imputed datasets were combined using Rubin’s rules.17 Stratified analyses were performed to evaluate potential effect modification by socio-demographic variables and clinical comorbidities. All statistical analyses were performed by using SAS 9.4 software (Supplementary Appendix).

Persons with CKD were aged 1–74 years with CKD-EPI estimated glomerular filtration rate (eGFR) <90ml/min/1.73m2, and/or urine albumin/creatinine ratio (ACR) ≥3.0 mg/mmol (30 mg/g), and persons with biopsy confirmed glomerular diseases, pregnant women were excluded from the study (Supplementary Appendix). Study controls were participants without CKD defined as having an eGFR ≥90ml/min/1.73m2 and urine ACR <3.0mg/mmol (<30 mg/g). The study population is representative of the West African population affected by CKD with and without APOL1 CKD risk variants, and our findings are likely to be generalizable (Supplementary Appendix Table S1).

APOL1 kidney risk variants G1 (rs73885319, p.S342G and rs60910145, p.I384M) and G2 (rs71785313, p.N388_Y389del) were genotyped using custom TaqMan assays (Applied Biosystems; see the Supplementary Appendix).

Kidney biopsies were performed when the GFR was ≥15 and the ACR was greater than 50 mg/mmol or albuminuria ›500 mg/24 hours, renal biopsy samples were read after staining for light, immunofluorescence and electron microscopy at the University of Michigan and Massachusetts General Hospital, as detailed in theSupplementary Appendix.

APOL1 risk status was defined according to the number of copies of the risk alleles (0, 1, or 2 copies). Our analyses included participants with missing data on age, sex and urine ACR; body mass index (BMI) and mean arterial pressure (MAP) used as covariates in association models. Proportion of missingness for any variable was less than 10% (Table 1). We performed multivariate imputation using the method of fully conditional specifications also known as multivariate imputation by chained equations. Logistic regression model was used for categorical variables, and the predictive mean matching method based on linear regression which imputes a value randomly from a set of observed values was used for imputing continuous variables each with specified predictors. For each imputation variable, 25 imputations were performed. We then fitted recessive genetic models to examine the association between APOL1 high risk variants and CKD, comparing participants with two copies of APOL1 risk alleles (high-risk) with participants with one or no copy of APOL1 risk allele (low risk). Adjustment covariates in the models include age, sex, clinical site and the baseline BMI, MAP, HIV status, diabetes mellitus status, and history of tobacco use Albuminuria was not treated as a covariate because it is part of the definition of CKD. We used multiple imputation with 25 imputations to replace missing values for BMI and MAP. Effect estimates from the imputed datasets were combined using Rubin’s rules.17 Stratified analyses were performed to evaluate potential effect modification by socio-demographic variables and clinical comorbidities. All statistical analyses were performed by using SAS 9.4 software (Supplementary Appendix).

Among the 8,355 participants with available clinical and genotypic data, 5,578 had CKD (4,712 patients with CKD stages 2–5, and 866 with biopsy-proven glomerular diseases) and 2,777 had no CKD. Table 1 displays baseline characteristics of cases, stratified by CKD stage/biopsy. The study population comprised of 63.3% Nigerians and 36.7% Ghanaians. Participants with CKD had higher blood pressure and were more likely to have hypertension and diabetes mellitus (Table 1). Baseline characteristics of participants with biopsy-proven glomerular diseases are depicted in Table S2, which shows histologic findings among the 866 biopsied patients minimal change disease (MCD; n = 300, 34.6%); focal segmental glomerulosclerosis (FSGS; n= 214, 24.7%); membranous nephropathy (MN; n = 88, 10.2%); lupus nephritis (LN; n = 101, 11.7%) and others (n= 161, 18.6%, Table S3). Overall, 43.0% and 29.7% participants carried 1 and 2 APOL1 risk variants, respectively (Table S4). The G1 haplotype was more frequent (40.7%) than the G2 haplotype (13.9%). The prevalence of APOL1 high-risk carriers was higher among persons with CKD compared to those without CKD (31.6% vs. 25.7%). The frequency of G1 or G2 risk haplotypes in Ghana and Nigeria, and in the major ethnic groups in the two countries are shown in Tables S5 and S6, and Figure S1.

) than the G2 haplotype (13.9%). The prevalence of APOL1 high-risk carriers was higher among persons with CKD compared to those without CKD (31.6% vs. 25.7%). The frequency of G1 or G2 risk haplotypes in Ghana and Nigeria, and in the major ethnic groups in the two countries are shown in Tables S5 and S6, and Figure S1. Among participants carrying 2 APOL1 risk alleles, odds ratio of CKD was 1.25 (95% CI 1.11–1.40) compared to low-risk carriers after adjusting for age, sex, mean arterial pressure, HIV status, diabetes mellitus, clinical sites, tobacco use, and ethnic group (Table 2, Figure 1). Odds of having CKD were 1.37 higher (95% CI 1.16–1.61), 2.05 higher (95% CI 1.35–3.13), 1.34 higher (95% CI 1.12–1.61) for G1/G1, G2/G2, and G1/G2 genotype, respectively. In APOL1 high-risk carriers, there is a graded increased odds by CKD stage; adjusted OR (95% CI) of 1.20 (1.04–1.38), 1.32 (1.12–1.56), 1.37 (1.18–1.59) for CKD stages 2, 3, and 4/5 respectively (Table 3). There were no differences by sex, age, and hypertension, the number of participants with history of diabetes and individuals positive for HIV were too small for meaningful interpretation of the association between CKD and high risk APOL1 genotype in these subgroups. (Figure S2). Secondary analysis excluding participants with CKD stage 2 or history of diabetes had minimal impact on the effect size of association between CKD and 2 APOL1 variants (Tables S7 and S8).

too small for meaningful interpretation of the association between CKD and high risk APOL1 genotype in these subgroups. (Figure S2). Secondary analysis excluding participants with CKD stage 2 or history of diabetes had minimal impact on the effect size of association between CKD and 2 APOL1 variants (Tables S7 and S8). To determine the impact of one APOL1 variant with disease risk, we compared the odds of CKD in participants with a single disease risk variant (G0/G1, G0/G2) versus those with no risk variant (G0/G0). The odds of having CKD was higher in participants with G0/G1 or G0/G2 compared to those with G0/G0 genotype, (adjusted OR 1.18 (95% CI 1.04–1.33) (Table 2, Figure 1 Panel B). We observed a dose-response relationship between dose of APOL1 risk variants and odds of CKD. Four major histologic patterns were identified in 866 participants who had kidney biopsies, MCD (300/866, 34.6%), FSGS (214/866, 24.7%), LN (101/866, 11.7%), and MN (88/866, 10.2%) (Table S2). High-risk APOL1 carriers had 84% (95% CI 1.30–2.61) higher adjusted odds compared to those with low-risk haplotypes of having FSGS (Table 4, Figure 1 Panel A). The odds for MCD, LN, and MN, were not increased (Table 4, Figure S3). There was increased odds of having FSGS with one risk compared with no variant (adjusted OR 1.61; 95% CI 1.04–2.48; Figure 1 Panel B).

Overall, 43.0% and 29.7% participants carried 1 and 2 APOL1 risk variants, respectively (Table S4). The G1 haplotype was more frequent (40.7%) than the G2 haplotype (13.9%). The prevalence of APOL1 high-risk carriers was higher among persons with CKD compared to those without CKD (31.6% vs. 25.7%). The frequency of G1 or G2 risk haplotypes in Ghana and Nigeria, and in the major ethnic groups in the two countries are shown in Tables S5 and S6, and Figure S1.

Among participants carrying 2 APOL1 risk alleles, odds ratio of CKD was 1.25 (95% CI 1.11–1.40) compared to low-risk carriers after adjusting for age, sex, mean arterial pressure, HIV status, diabetes mellitus, clinical sites, tobacco use, and ethnic group (Table 2, Figure 1). Odds of having CKD were 1.37 higher (95% CI 1.16–1.61), 2.05 higher (95% CI 1.35–3.13), 1.34 higher (95% CI 1.12–1.61) for G1/G1, G2/G2, and G1/G2 genotype, respectively. In APOL1 high-risk carriers, there is a graded increased odds by CKD stage; adjusted OR (95% CI) of 1.20 (1.04–1.38), 1.32 (1.12–1.56), 1.37 (1.18–1.59) for CKD stages 2, 3, and 4/5 respectively (Table 3). There were no differences by sex, age, and hypertension, the number of participants with history of diabetes and individuals positive for HIV were too small for meaningful interpretation of the association between CKD and high risk APOL1 genotype in these subgroups. (Figure S2). Secondary analysis excluding participants with CKD stage 2 or history of diabetes had minimal impact on the effect size of association between CKD and 2 APOL1 variants (Tables S7 and S8).

To determine the impact of one APOL1 variant with disease risk, we compared the odds of CKD in participants with a single disease risk variant (G0/G1, G0/G2) versus those with no risk variant (G0/G0). The odds of having CKD was higher in participants with G0/G1 or G0/G2 compared to those with G0/G0 genotype, (adjusted OR 1.18 (95% CI 1.04–1.33) (Table 2, Figure 1 Panel B). We observed a dose-response relationship between dose of APOL1 risk variants and odds of CKD.

Four major histologic patterns were identified in 866 participants who had kidney biopsies, MCD (300/866, 34.6%), FSGS (214/866, 24.7%), LN (101/866, 11.7%), and MN (88/866, 10.2%) (Table S2). High-risk APOL1 carriers had 84% (95% CI 1.30–2.61) higher adjusted odds compared to those with low-risk haplotypes of having FSGS (Table 4, Figure 1 Panel A). The odds for MCD, LN, and MN, were not increased (Table 4, Figure S3). There was increased odds of having FSGS with one risk compared with no variant (adjusted OR 1.61; 95% CI 1.04–2.48; Figure 1 Panel B).

The genetic locus containing APOL1 was initially identified as an important risk factor for CKD among AAs.6 However, definitive estimates of the genetic risk of APOL1 on CKD obtained from large studies in West Africa, the ancestral origin of most AAs have been lacking. In the present study, among 8,355 participants from Ghana and Nigeria, 29.7% of the population were identified as high-risk APOL1 carriers. High-risk APOL1 carriers in the present study had 25% higher odds of having CKD than low-risk carriers. The association was even stronger in patients with biopsy-proven FSGS who had 84% higher odds of FSGS compared with those with low-risk genotypes. There were also increasing odds of disease with advanced CKD. An important and novel finding is that we observed a strong association of a single risk variant with CKD (18% higher odds) and FSGS (61% higher odds) compared to those with no risk variant. This implies that the dose of APOL1 genetic risk variants may also be relevant for kidney disease.

of disease with advanced CKD. An important and novel finding is that we observed a strong association of a single risk variant with CKD (18% higher odds) and FSGS (61% higher odds) compared to those with no risk variant. This implies that the dose of APOL1 genetic risk variants may also be relevant for kidney disease. We found the frequency of high-risk APOL1 alleles in Ghana and Nigeria as 29.7%, which is higher than the 11.0–16.5% reported among AAs.18–20 This is expected given that AAs have an approximately 20% European admixture and derive their origins predominantly from West Africa and also from other parts of Africa. The prevalence of APOL1 high-risk variants in the present study were also higher than the 7–11% reported in East Africa but similar to prior small single center studies from West Africa.11–12 Prior to the present study, the largest study of APOL1 and kidney disease in sub-Saharan Africa included 10,769 participants.21 However, APOL1 variants were imputed (not genotyped) in that study, which lacked data on biopsy-proven kidney disease.21 A recent large study comprising ~2000 participants provided new estimates of APOL1 risk alleles and associations with CKD in rural South Africa.22 The addition of our study to growing data may reveal clearer picture of how APOL1 high-risk variants’ geographic distribution, haplotype frequencies, associations and effect sizes vary across Africa as a result of population and migration patterns over millenia interacting with infections and other selection pressures.23 We observed a wide variation in high-risk APOL1 haplotypes among ethnolinguistic groups in our study, suggesting that West African people are not monolithic regarding the distribution of APOL1 risk variants and that ethnic origin should be considered when evaluating genetic risk factors for CKD in sub-Saharan African persons.

rved a wide variation in high-risk APOL1 haplotypes among ethnolinguistic groups in our study, suggesting that West African people are not monolithic regarding the distribution of APOL1 risk variants and that ethnic origin should be considered when evaluating genetic risk factors for CKD in sub-Saharan African persons. APOL1 risk variants for CKD are a major contributor to the disparity in prevalence of CKD among AAs compared with European Americans. The strength of association of APOL1 high-risk variants varies depending on the cause of CKD, stage of CKD, or study design, and reported estimates of 1.5 – 10.5 increased risk among AAs.24–25 In the present study, we observed an association with CKD similar to that already reported among AAs.24,26 The 27% higher odds of CKD and 84% higher odds of FSGS in the present study represent a substantial public health burden of kidney disease among the millions of West Africans, especially in light of the high and premature mortality associated with CKD in West Africa.1

KD similar to that already reported among AAs.24,26 The 27% higher odds of CKD and 84% higher odds of FSGS in the present study represent a substantial public health burden of kidney disease among the millions of West Africans, especially in light of the high and premature mortality associated with CKD in West Africa.1 We also found a stronger association with CKD progression, suggesting that not only is APOL1 high-risk haplotype a risk factor for CKD in this population, it may also be a risk factor for disease progression as previously reported for AAs.26 The findings are also consistent with recent reports in patients with FSGS and MN.27–28 APOL1 variants were first discovered in association with ESKD and advanced FSGS, and subsequently with higher risk of disease progression.24, 26, 29–30 The evidence to date points to APOL1 CKD risk variants acting as a genetic risk factor for rapid deterioration in kidney function among West African and African ancestry populations globally regardless of the initial kidney insult.26, 30

anced FSGS, and subsequently with higher risk of disease progression.24, 26, 29–30 The evidence to date points to APOL1 CKD risk variants acting as a genetic risk factor for rapid deterioration in kidney function among West African and African ancestry populations globally regardless of the initial kidney insult.26, 30 In the present study, we found an almost two-fold increased odds of developing biopsy-proven FSGS in participants with APOL1 high-risk genotype. Since FSGS is a primary podocytopathy, this finding is not surprising because podocytes are the major target of cytotoxic activity of APOL1 high-risk variant protein within the kidney.31 The odds ratio (OR) of ~2 associated with 2 APOL1 risk variants in patients with FSGS observed in this study is modest compared with OR of ~10 reported in AA.6 The reason for this is unclear, but possible explanations include survival bias in the AA population, and the possibility that European admixture in AAs may have changed the genetic background against which the APOL1 risk variant phenotype is expressed. Further, studies reporting high OR are discovery studies that tend to focus on extreme phenotypes and therefore would be expected to have large effect size compared with population and longitudinal studies. Nonetheless, we would still recommend routine genotyping for APOL1 in patients with FSGS from Nigeria and Ghana. Adopting this recommendation would, in our opinion, help in design and recruitment of participants for future clinical trials.32–33 Consistent with other reports, we did not find any association with other biopsy-proven glomerulopathies such as MCD, MN, and LN.28–30

r APOL1 in patients with FSGS from Nigeria and Ghana. Adopting this recommendation would, in our opinion, help in design and recruitment of participants for future clinical trials.32–33 Consistent with other reports, we did not find any association with other biopsy-proven glomerulopathies such as MCD, MN, and LN.28–30 We also found modest increased odds of CKD in participants with one APOL1 risk variant compared with those with no risk. In previous studies, carrying one APOL1 variant was associated with progression to lupus nephritis in patients with systemic lupus erythematosus (SLE), but not with risk of sickle cell disease nephropathy.34–35 While there is a need for large studies to evaluate the role of APOL1 risk variant dose in CKD of diverse etiology, one might speculate that one APOL1 risk variant in the presence of genetic or environmental modifiers may predispose to the development or progression of CKD. In the present study, high-risk genotype was not associated with CKD in those with a history of diabetes, consistent with the findings that the APOL1 high-risk genotype is not a risk factor for diabetic kidney disease, but may be a risk factor for disease progression.26 There are likely to be other genetic and environmental factors associated with CKD in Africans, which would need to be tested in large longitudinal studies that account for gene-environment interactions and avoid the limitations of previous underpowered studies.36–39

but may be a risk factor for disease progression.26 There are likely to be other genetic and environmental factors associated with CKD in Africans, which would need to be tested in large longitudinal studies that account for gene-environment interactions and avoid the limitations of previous underpowered studies.36–39 Our study has several strengths. It is a large study of APOL1-kidney disease associations on the African continent in which the APOL1 variants were directly genotyped and has provided much needed estimates of effect sizes of APOL1 high-risk variants in West Africa. Additionally, we obtained kidney biopsies in >800 study participants. The availability of a tissue diagnosis on a large sample size facilitated tests of association between APOL1 variants by histological classification. The finding of a monotonal increase in odds of APOL1-associated kidney disease has an important clinical implication because it changes the risk classification of carriers of one copy of an APOL1 high-risk variant.

large sample size facilitated tests of association between APOL1 variants by histological classification. The finding of a monotonal increase in odds of APOL1-associated kidney disease has an important clinical implication because it changes the risk classification of carriers of one copy of an APOL1 high-risk variant. This study also has certain limitations. The observed moderate association between APOL1 high-risk variants and CKD may be due to the heterogeneity of CKD. However, in exploratory analyses, using CKD stratifications, and sub-analysis of participants with biopsy-proven FSGS, we detected a stronger association. There is a need for longitudinal studies in Africa to evaluate the association between APOL1 haplotype and progression of CKD. We were not able to genotype for the recently reported APOL1 N264K (rs73885316) G2 disease-associated modifier.39–40 Therefore, the potential impact of this variant on our findings is unclear. Furthermore, we did not screen for monogenic kidney diseases by whole genome sequencing (WGS). Future studies utilizing WGS in CKD patients from Africa will define the impact of the N264K variant and other modifiers on risk of CKD in Africans, and also remove the confounder effects of participants with monogenic kidney diseases.39–40 Study participants were recruited from two West African countries, and our results may not be generalizable for other regions of Africa.

will define the impact of the N264K variant and other modifiers on risk of CKD in Africans, and also remove the confounder effects of participants with monogenic kidney diseases.39–40 Study participants were recruited from two West African countries, and our results may not be generalizable for other regions of Africa. In this large study of the prevalence and association of APOL1 high-risk variants with CKD in persons in West Africa, a region that contributes substantially to the ancestry of African Americans; almost one-third of those tested carried high-risk variants of APOL1. In conclusion, both monoallelic (G1/G0, G2/G0) and biallelic (G1/G1, G2/G2, G1/G2) risk variants had 18% and 25% higher odds of CKD, and 61% and 84% higher odds of FSGS, respectively.