Abstract
Objective. Apolipoprotein L1 gene (APOL1) G1 and G2 renal risk alleles (RRA) are associated with endstage renal disease in blacks with lupus nephritis (LN). The present study determined frequencies of APOL1 RRA in nonwhite Brazilian patients with LN and controls to assess association with renal outcomes.
Methods. APOL1 RRA were genotyped in 222 healthy blood donors (controls) and 201 cases with LN from 3 outpatient clinics. Two single-nucleotide polymorphisms in the G1 (rs73885319 and rs60910145) and an indel for the G2 (rs71785313) variant were genotyped.
Results. The frequency of APOL1 RRA in nonwhite Brazilian LN cases did not differ significantly from healthy controls, and few participants had 2 RRA. In the sample, 84.6% of LN cases and 84.2% of controls had 0 RRA, 13.4% and 15.3% had 1 RRA, and 2.0% and 0.4% had 2 RRA, respectively. LN cases with ≥ 1 APOL1 RRA had similar baseline characteristics and renal responses to treatment, yet faced higher risk for progressive chronic kidney disease (CKD) to an estimated glomerular filtration rate < 30 ml/min/1.73 m2 compared to those with 0 RRA (11.2% with 0, 29.6% with 1; 50% with 2 RRA, p = 0.005). Although glomerular lesions and activity scores on initial kidney biopsy did not differ significantly between individuals based on APOL1 genotype, chronicity scores, tubular atrophy, and interstitial fibrosis were more severe in those with ≥ 1 RRA (p = 0.011, p = 0.002, p = 0.018, respectively).
Conclusion. Although initial kidney lesions and treatment responses were similar, a single APOL1 RRA in nonwhite Brazilians with LN was associated with increased risk of advanced CKD and possibly more tubulointerstitial damage.
Nondiabetic chronic kidney disease (CKD) is significantly more prevalent in those who possess recent African ancestry, a finding related in part to the presence of apolipoprotein L1 gene (APOL1) renal risk alleles (RRA). Two coding nephropathy variants in APOL1, G1 (rs73885319; rs60910145) and G2 (rs71785313), appear to have been selected in sub-Saharan Africa because their circulating proteins provide resistance to Trypanosoma brucei rhodesiense and development of African sleeping sickness1,2. Although 13% of African Americans possess APOL1 high-risk genotypes, defined as having 2 copies of the G1 and/or G2 allele, only a minority develops CKD. It appears likely that modifying factors are required to initiate APOL1 nephropathy.
High interferon (IFN) states, including human immunodeficiency virus (HIV) infection [producing HIV-associated nephropathy (HIVAN)]3, exogenously administered IFN4, and systemic lupus erythematosus (SLE) are linked with collapsing glomerulopathy in carriers of 2 APOL1 RRA (autosomal recessive inheritance)5. In addition, severe lupus nephritis (LN), LN–endstage renal disease (ESRD), is associated with APOL1 in an autosomal recessive inheritance pattern6,7. Effects of APOL1 on nondiabetic ESRD reveal OR for association of 3 in patients with LN-ESRD and 29–89 in those with HIVAN3,6,8. A recent large genome-wide association study searching for modifying genes in APOL1 nephropathy failed to identify second genes or additional variants meeting genome-wide significance for association with LN-ESRD, suggesting environmental modifiers often trigger APOL1 nephropathy7.
Relative to whites, African Americans and Hispanics develop more aggressive LN with earlier onset and poorer longterm renal outcomes9. European ancestry is reportedly protective from LN in patients with SLE10. Moreover, familial clustering of LN and CKD suggests a role for genetic factors11, and African Americans with ≥ 1 APOL1 RRA were reported to initiate renal replacement therapy earlier than those lacking APOL1 RRA12,13. South American populations have variable contributions of West African ancestry because of the slave trade that occurred 500 years ago14,15. This should result in a range of APOL1 RRA frequencies in this relatively understudied population16. As in other areas of Latin America, the repeated forced migration of individuals of West African ancestry during the slave trade resulted in significant genetic admixture in Brazil (i.e., interbreeding of 2 previously separated and distinct populations)17. Brazilians are an admixed population, with differing proportions of Amerindian, African, and European ancestry14,15,18. Frequencies of APOL1 RRA have been variable, depending on the region of Brazil. One study of Brazilians with LN revealed that about 30% of their genome was African; however, only 10% of cases had 2 APOL1 RRA without significant association with CKD19. Another report genotyped black and mixed Brazilian populations with ESRD; it found 10-fold higher frequencies of APOL1 renal-risk genotypes (2 RRA) compared to related controls20. The latter study reveals that APOL1 is associated with non-diabetic ESRD in Brazilians in autosomal recessive fashion; however, cases lacked LN.
The primary hypothesis of our study was to determine whether there was an association between APOL1 RRA and development of progressive CKD defined as a sustained estimated glomerular filtration rate < 30 ml/min/1.73 m2 in this nonwhite (mixed) Brazilian population. Secondary analyses assessed the effect of APOL1 RRA on additional kidney outcomes in LN, including kidney histology and longterm kidney function.
MATERIALS AND METHODS
Cases with LN were enrolled from 3 outpatient clinics in Brazil specializing in treatment of glomerulonephritis (GN): Federal University of Pernambuco, Prof. Fernando Figueira Integrative Medicine Institute — IMIP (Recife, Northeastern Brazil), and Federal University of São Paulo — EPM/UNIFESP (São Paulo, Southeastern Brazil). All participants provided written informed consent. The study was approved by the Brazilian National Committee for Ethics in Research (report number: 2.568.450) and performed in accordance with the Declaration of Helsinki.
Overall, 309 patients with a previous diagnosis of LN were recruited between August 2015 and July 2018. All were > 18 years of age, unrelated, met Systemic Lupus International Collaborating Clinics Classification Criteria, and had negative serologies for hepatitis B, hepatitis C, HIV, and syphilis. All patients had a renal biopsy. Biopsies were analyzed by 2 renal pathologists, 1 from IMIP and 1 from EPM/UNIFESP. The classification and characteristics of LN were described according to the International Society of Nephrology/Renal Pathology Society guidelines. We excluded 30 patients with non-LN histologic patterns [including IgA nephropathy, vasculitis, postinfectious GN, idiopathic membranous GN, focal segmental glomerulosclerosis (FSGS), or collapsing GN] and those with < 6 months of followup after diagnosis of LN. In addition, 9 patients with inadequate DNA and 72 self-reporting their ancestry as white were excluded. The remaining 201 cases had LN on initial kidney biopsy. None had Class I or Class VI (> 90% of glomeruli globally sclerosed) LN. We analyzed cases with Class II mesangial proliferative LN (pure mesangial hypercellularity and/or matrix expansion); Class III focal proliferative LN (involving < 50% of the total number of glomeruli); Class IV diffuse proliferative or global LN (involving > 50% of the total number of glomeruli), and Class V membranous LN21. Eight of 201 cases (4%) did not have enough kidney tissue to classify LN, but were retained in the analyses based on appropriate clinical presentations with followup similar to the other LN cases (1 had Class IV LN on a subsequent renal biopsy during a second SLE flare several months later). Those with a history of essential hypertension or with blood pressure readings ≥ 140 mmHg systolic and/or ≥ 90 mmHg diastolic on at least 2 occasions were considered to have hypertension.
Historical data regarding initial laboratory tests, first induction/maintenance therapy, and treatment response were recorded from chart review. Thereafter, participants were followed prospectively during routine care through February 2019. During acute flares of nephritis, cases with LN underwent induction therapy with intravenous methylprednisolone, followed by oral prednisone and 6 boluses of intravenous cyclophosphamide 0.5–1 g or mycophenolate mofetil (MMF) 2–3 g/day. Postinduction, they received maintenance azathioprine or MMF, based on established protocols. At baseline, hydroxychloroquine was prescribed to more than 80% of LN cases. Changes in proteinuria and serum creatinine concentration (SCr) were recorded from chart reviews at 6, 12, and 24 months and/or latest followup, according to Kidney Disease Improving Outcomes guidelines22. Renal responses to therapy were classified as complete, partial, or non-responsive22. LN cases who developed CKD stage 3 or stage 4 (defined as a sustained eGFR < 60 or < 30 ml/min/1.73 m2 using the CKD-Epidemiology Collaboration equation, respectively) and ESRD defined as the need for renal replacement therapy or eGFR < 10 ml/min/1.73 m2 were recorded. Refractory LN was defined as lack of a complete or partial response after 2 different induction treatments, including at least 1 course of cyclophosphamide (some may have received cyclosporine with MMF or rituximab).
A total of 222 unrelated, nonwhite, adult healthy blood donors from 2 Brazilian blood centers (Recife–PE and Ribeirão Preto–SP) were genotyped and served as non-SLE controls.
Genomic DNA was isolated from anticoagulated whole blood collected in EDTA blood tubes using the PureGene system, based on manufacturer instructions. Samples were shipped on ice to Wake Forest School of Medicine for APOL1 genotyping. Two single-nucleotide polymorphisms in the G1 nephropathy risk variant (rs73885319; rs60910145) and an indel for the G2 nephropathy risk variant (rs71785313) were genotyped using Taqman assays on the ViiA 7 platform (Applied Biosystems for Life Tech). APOL1 high-risk genotypes were present if participants had 2 RRA (G1G2, G1G1, or G2G2).
Participant characteristics were compared using a Student t test or Mann-Whitney U test (i.e., Wilcoxon rank-sum test) as distributionally appropriate or Fisher’s exact test for categorical variables. Given the low frequency of 2 APOL1 RRA, Kaplan-Meier survival curves were computed separately for APOL1 RRA = 0 and RRA ≥ 1, and differences were computed using the log-rank test. Cox proportional hazards models were computed to estimate an HR for RRA = 0 versus RRA ≥ 1. The comparison of RRA = 0 versus RRA ≥ 1 on development of progressive CKD (defined based on sustained eGFR < 30 ml/min/1.73 m2) was the primary a priori inference. Significance was set at p < 0.05. Additional outcomes, including renal histologic changes and longterm clinical variables, were considered secondary outcomes.
We computed 3 power analyses to quantify the effect size detectable with 0.80 power and a type 1 error rate of α = 0.05. For binary outcomes (e.g., ESRD) between LN cases and controls, the study has 0.80 power to detect effects with OR of 1.78. For continuous outcomes, the study has 0.80 power to detect differences between cases and controls that explain 1.9% of the variation (i.e., r2 = 0.019), and case-only continuous traits that explain 3.9% of the variation.
RESULTS
APOL1 genotypes and demographic characteristics in self-reported nonwhite LN cases and non-SLE controls are displayed in Table 1. As expected, cases with LN had more females than did non-SLE controls (89% vs 36%). APOL1 allele frequencies did not differ significantly between LN cases and controls. Among the 72 self-described white LN cases excluded from the analyses, 3 had 1 APOL1 RRA (4%) and none had 2 RRA. Thus, white non-SLE controls were not genotyped.
Race was categorized as self-reported white and nonwhite (including mixed or black); Asians and Amerindians were not present. Household income was not analyzed because > 90% of the national public health system (Sistema Único de Saúde) users are paid < US$100 monthly, and immunosuppressive medications are provided by the government23. Table 2 displays demographic characteristics, baseline laboratory results, kidney biopsy findings, and longterm outcomes in nonwhite LN cases, stratified by APOL1 genotype. Because only 4 LN cases (2%) possessed 2 APOL1 RRA, groups were analyzed based on the presence of ≥ 1 APOL1 RRA. Although not statistically significant, cases with 1 or 2 APOL1 RRA tended to be younger and have shorter LN durations than cases with 0 RRA (p = 0.09 and 0.36, respectively). However, higher frequencies of CKD stage 4 and 5 (ESRD) were present in LN cases with ≥ 1 APOL1 RRA (p = 0.005 and 0.007, respectively). This occurred despite similar baseline demographic characteristics, CKD risk factor profiles, eGFR, proteinuria, and histologic class of LN. In addition, prescribed treatments were similar in LN cases regardless of APOL1 genotype. Although no differences were observed in the initial clinical response between genotype groups, LN cases with ≥ 1 APOL1 RRA more often developed sustained eGFR < 60 ml/min/1.73 m2 six months after induction therapy, compared to those with 0 RRA (21.7% vs 4.4%, p = 0.018; OR 5.12, 95% CI 1.6–17.6; Table 3).
In secondary analyses, a trend toward higher percentages of glomeruli with global glomerulosclerosis and crescents were seen in LN cases with ≥ 1 APOL1 RRA, although types of LN-related glomerular lesions were not different between genotypes (Table 2). Interstitial damage, measured by the percentage of tubular atrophy and interstitial fibrosis, was more severe in LN cases with ≥ 1 APOL1 RRA (p = 0.002 and p = 0.018, respectively). The activity index was similar between genotype groups (p = 0.92), but the chronicity index on the initial kidney biopsy was significantly higher in LN cases with ≥ 1 APOL1 RRA (4.1 ± 2.3), versus 0 RRA (2.8 ± 1.6; p = 0.011). Fifty of 201 LN cases (43 with 0 APOL1 RRA and 7 with > 1 RRA) received a second kidney biopsy (Table 4). There was no statistically significant difference in renal histology between genotype groups, except that median percentage of crescents (not presence) was higher on the second biopsy in LN patients with ≥ 1 APOL1 RRA (p = 0.03). It is difficult to estimate the value of the second biopsy done during relapses from only a quarter of participants.
Figure 1 displays Kaplan-Meier renal survival curves for CKD, eGFR < 30 ml/min/1.73 m2 (p = 0.003, HR 2.97, 95% CI 1.1–8.2), and ESRD (p = 0.006, HR 3.49, 95% CI 1.0–12.5).
The time from initial diagnosis of LN to ESRD was significantly shorter in LN cases with ≥ 1 APOL1 RRA compared to those with 0 RRA [14 (25–75th = 9–22) vs 114 (25–75th = 36–220) mos, p = 0.0023]. Thus, faster progression to ESRD was present in those with ≥ 1 RRA (Table 3).
Table 5 displays the outcomes in the 4 LN cases with 2 APOL1 RRA. Despite the small sample, half progressed to CKD stage 4 (eGFR < 30 ml/min/1.73 m2) and 1 had persistent proteinuria after 3 rounds of induction therapy.
DISCUSSION
The results of our study of Brazilians with LN demonstrate that participants with ≥ 1 APOL1 RRA had more severe kidney disease at initial diagnosis and higher stages of CKD after 6 months of therapy compared to those with 0 APOL1 RRA. Populations with mixed ancestry are not typically screened for APOL1 RRA; frequencies are expected to vary based on extent of recent African ancestry16. The Brazilian population is heterogeneous as a result of interethnic mating of peoples from 3 continents: European colonizers (mainly Portuguese), African slaves, and local Amerindians14,15. Our study genotyped self-reported nonwhite healthy controls and cases with LN. Cases and controls had similar and low frequencies of APOL1 high-risk genotypes (2 RRA), 0.4% and 2.0%, respectively. A study in the Brazilian city of Salvador genotyped 45 ESRD cases and identified only 1 (2.0%) with 2 APOL1 RRA24. In contrast, Riella, et al reported a higher prevalence of APOL1 2 RRA (12.4%) and 1 RRA carriers (17.5%) among 274 self-declared Brazilian mixed-race and black patients with ESRD; those with auto-immune kidney disease were excluded20. They also analyzed 106 matched first-degree relatives of cases and found lower frequencies of APOL1 2 RRA carriers (0.9%) and similar frequencies with 1 RRA (13.2%)20. The APOL1 frequencies in their controls appear similar to those in healthy blood donor controls from our present study.
A study from São Paulo genotyped APOL1 in 196 female outpatients with LN; participants had 30% African ancestry based on ancestry informative markers (AIM)19. Of these, 10% possessed 2 APOL1 RRA and there was no significant association of APOL1 with doubling of the baseline SCr in a recessive genetic model19. In the present cohort of LN cases and controls, AIM were lacking because of a paucity of DNA. Although skin color is not an accurate predictor of AIM in such an admixed population, those self-described as black or mixed Brazilians reportedly have a higher African ancestry index (AAI)14. We detected no significant difference in genetic ancestry based on skin pigmentation in Brazilians; participants from Recife had 59.7% European ancestry, 23.0% African ancestry, and 17.3% Amerindian ancestry18. Other studies using AIM from different Brazilian regions revealed similar patterns of European dominance, followed by African, and to a lesser extent Amerindian genetic ancestry15,25.
A study comparing the AAI among black and white Brazilians from each region of the country found similar AAI between individuals from the Northeast and Southeast regions of Brazil, but lower AAI in original Africans (and higher than in the founding Portuguese)14. The prevailing hypothesis is that APOL1 G1 and G2 RRA arose in the past 10,000 years in sub-Saharan Africa, likely in West Africa where they were subjected to intense positive selection since circulating APOL1 RRA proteins provide resistance to T. brucei rhodesiense1,26. South America was likely colonized around 15,000 years ago, likely by a single wave of migration27 and before positive selection for APOL1. This suggests that APOL1 RRA came from the trans-Atlantic slave trade during the 16th to 19th centuries.
Asian, Native American, and white populations with CKD generally have very low frequencies of APOL1 RRA28,29,30,31. Among Native Americans, African-derived risk alleles in the DNA sequence of APOL1 coding regions were absent, providing additional evidence that these risk variants are present only in those with recent African ancestry32. However, among admixed (with African ancestry) Hispanic and Latin Americans, APOL1 two RRA genotypes were present in 2% of individuals30. This is similar to our present study, with low rates of CKD.
The low frequency of APOL1 2 RRA carriers in our Brazilian LN cohort did not permit performance of outcome analyses using the traditional autosomal recessive model. However, presence of even 1 APOL1 RRA demonstrated significant association with advanced CKD during followup. Presence of ≥ 1 APOL1 RRA confers immunity against T. brucei rhodesiense33. APOL1 cellular toxicity may arise from the same trypanolytic factors that produce chloride channels in lysosomes, producing damage to cell membranes, mitochondria, and cell death34,35.
Genetic risk for APOL1-associated CKD in humans is autosomal recessive; animal models are complicated by the lack of APOL1. Few animal models have tested the heterozygous state, typically a disease-free condition in humans36. Zebrafish embryos with APOL1 CRISPR/Cas9 genome editing revealed podocyte loss and glomerular filtration defects that could be rescued by expression of wildtype APOL1 mRNA37. However, the APOL1 G1 RRA did not ameliorate defects caused by suppression of APOL1, nor did G2, which was deleterious to protein function37. African Americans with 1 or 2 APOL1 RRA are known to require dialysis an average of 5 years and 9 years earlier than those with 0 RRA13. Moreover, as the number of APOL1 RRA increased in the present study, duration from SLE onset to ESRD decreased6.
Untreated patients with HIV who carry 2 APOL1 RRA have among the highest OR for CKD (29–89); however, even 1 RRA was associated with HIVAN in Africans (OR 5.49)8. A single APOL1 RRA also confers a 1.7–fold increased risk for FSGS, although 2 RRA confer a 10-fold higher risk3. These findings support the influence of a single APOL1 RRA in kidney injury. Chromosome 22q is also enriched for gene duplications in the APOL1-4 gene cluster and copy number variation may change gene dosage and expression. Additional copies of APOL1 were observed more frequently in CKD cases than in controls, possibly increasing susceptibility to CKD in heterozygotes38. Association between null variants in APOL3 and ESRD has been reported39, irrespective of APOL1 genotype status and percentage of African ancestry. This supports the concept that other APOL proteins (besides APOL1) may influence risk for nondiabetic CKD.
The spectrum of APOL1 nephropathy has known mediating factors in those with 2 APOL1 RRA, including HIV infection and IFN in collapsing glomerulopathy3,4,5. IFN are upregulated in patients with active SLE. Thus, this milieu might trigger APOL1 nephropathy even in cases with 1 RRA. LN reflects a chronic type I IFN-induced state and α-IFN increases APOL1 mRNA expression in endothelial cells4. Our present study identified a higher chronicity index and more frequent moderate to severe tubular atrophy and interstitial fibrosis on initial kidney biopsies in cases with LN with ≥ 1 APOL1 RRA, versus 0 RRA. However, significant differences in the type of glomerular lesion were not seen between genotypic groups, except a trend toward more global glomerulosclerosis and crescent formation in those with ≥ 1 APOL1 RRA. As in Larsen, et al, we did not detect differences among histologic classes of LN based on APOL1 genotypes, but saw a trend toward higher chronicity index in the ≥ 1 RRA group40, with an increased risk for progression to ESRD in cases with at least 1 RRA.
This study has strengths and limitations. Strengths include longitudinal followup in a relatively large sample of Brazilians with LN. A weakness included the lack of AIM in self-described nonwhite cases and controls due to insufficient DNA; instead, we relied on self-reported ancestry. We note that the “nonwhite” cases and controls were from the same geographic region, self-reported ancestry was obtained in the same fashion in each group, and APOL1 RRA frequencies were generally consistent with those expected. We note that Parra, et al also found that Brazilians self-reporting as black or mixed had higher proportions of African ancestry14. Therefore, we restricted our sample to those self-reporting as nonwhite. Another limitation was absence of SLE controls without LN. However, when comparing LN cases with SLE controls lacking LN, it is possible that some “non-nephropathy controls” may develop LN given longer followup. A large number of the LN cases in our cohort first developed kidney disease 5 (or more) years after their diagnosis of SLE. The infrequent presence of 2 APOL1 RRA in this cohort and few cases with LN-ESRD did not permit evaluation of APOL1 associations in an autosomal recessive model. However, among the 4 Brazilian LN cases with 2 APOL1 RRA (Table 5), the only case that had a complete response initially presented with Class V (non-proliferative) membranous LN on kidney biopsy, a less aggressive lesion known to have lower Th1 lymphocytes response41.
Frequencies of APOL1 RRA in nonwhite Brazilians with LN are not significantly different from those in healthy nonwhite Brazilians; but participants with ≥ 1 APOL1 RRA had more severe kidney disease at presentation and higher stages of CKD after therapy compared to those with 0 APOL1 RRA. However, results do not preclude a recessive model. Our sample lacked sufficient numbers of individuals with 2 APOL1 RRA needed to detect such an effect. Regardless of treatment for LN, presence of ≥ 1 APOL1 RRA is associated with higher rates of chronic tubulointerstitial injury and increased risk for advanced stage 4 CKD and ESRD; there was no difference in the type of renal glomerular lesion. APOL1 genotyping in this admixed South American population sheds new light on the role of precision medicine in LN. Treatment approaches may need to be more aggressive or directly target the APOL1 gene to reduce rates of ESRD due to LN in the nonwhite Brazilian population.
Acknowledgment
The authors thank all the patients enrolled in our study, Michelle Tiveron for processing the samples at EPM/UNIFESP, and the administrative staff from the Division of Nephrology/Federal University of Pernambuco: Poliana Cassia and Ivanize Souza.
Footnotes
Supported by US National Institutes of Health grants R01 DK084149 and R01 DK070941 (both to Dr. Freedman). Wake Forest University Health Sciences and Dr. Freedman have rights to an issued US patent related to APOL1 genetic testing (www.apol1genetest.com). Dr. Freedman is a consultant for AstraZeneca Pharmaceuticals and Renalytix AI.
- Accepted for publication November 4, 2019.