Racial disparities in end-stage renal disease in a high-risk population: the Southern Community Cohort Study

Chronic kidney disease (CKD) is one of the fastest growing chronic health conditions worldwide, particularly among minority populations, and is associated with substantially increased risks of end-stage renal disease (ESRD) and cardiovascular mortality [1, 2]. In 2015, there were 124,114 incident cases of ESRD in the United States. ESRD incidence rates for blacks are more than three times greater than for whites, at 865 per million in 2013 [3]. There is a striking disparity in the lifetime risk of developing ESRD: 8% among blacks compared to 2–3% among whites [4]. To identify specific subpopulations at risk and their risk factors, a number of cohort studies have examined racial differences in ESRD and have consistently demonstrated higher ESRD incidence among blacks compared to whites [5‐7]. Neither traditional risk factors including diabetes and hypertension, nor the presence of high risk genotypes, fully explain the racial differences in ESRD [5‐8], emphasizing the need for longitudinal cohort studies in vulnerable populations to examine novel factors underlying racial differences.

The Southern Community Cohort Study (SCCS) is a large ongoing, prospective study of black and white participants residing in the southeastern United States, where rates of ESRD are among the highest in the nation due to reasons that remain to be fully elucidated [3]. The age-adjusted ESRD incidence rate in the SCCS study population is 79/100,000 person-years among whites compared to 285/100,000 person-years among blacks. The SCCS provides unique advantages for examining racial differences in ESRD, since all participants have similar (typically low) household income and education levels regardless of race, limiting confounding by socioeconomic differences. In an earlier report we described ESRD incidence among black and white participants in the SCCS and its association with traditional risk factors such as male gender, low income, diabetes, and hypertension [9]. While the pattern of risk factors was similar for blacks and whites, stronger associations were observed for traditional ESRD risk factors in blacks. After adjustment for these known risk factors, the risk of ESRD nonetheless remained more than two-fold higher for blacks [hazard ratio (HR): 2.4, 95% confidence interval (CI): 1.9, 3.0]. However, these analyses lacked data on baseline kidney function. We hypothesized that the observed racial disparity in risk may be explained by differences in kidney function at baseline. To address this, we used a case-cohort design, measured baseline serum creatinine to characterize kidney function, and evaluated associations with ESRD risk among black and white participants of the SCCS.

Compared to the subcohort, a higher proportion of ESRD cases were black (71% vs. 87%, respectively) (Table 1). In addition, ESRD cases were slightly older at enrollment than subcohort members and were of similar socioeconomic status. In contrast to the subcohort, ESRD cases had a higher baseline prevalence of cardiovascular risk factors, notably hypertension (85.9% versus 54.9%) and diabetes (68.7% versus 22.4%). The proportions of participants with baseline eGFR ≤30, 31–60, 61–90 and ≥ 90 ml/min/1.73m² in the SCCS subcohort were 1.0, 4.8, 24.6 and 69.6%, respectively (Table 1), while among those who subsequently developed ESRD they were 19.9, 26.0, 23.5 and 30.6%. Median (25th, 75th percentile) baseline eGFR among subcohort members was 103.2 (86.0, 117.9) compared to 63.3 (36.0, 98.2) among ESRD cases. The vast majority of the subcohort (94.2%) and the majority of ESRD cases (54.1%) had an eGFR > 60 at enrollment.

Black subcohort members had a higher median eGFR at baseline compared to whites (107.7 vs. 96.9 ml/min/1.73m²) (Table 2). Despite the higher incidence rate of ESRD in blacks, the proportions of black and white ESRD cases with normal kidney function (eGFR > 60 ml/min/m²) at baseline were not meaningfully different (54.7% in blacks versus 49.8% in whites). Black and white subcohort members were similar in terms of baseline characteristics, including education and income level, obesity, and the prevalence of cardiovascular risk factors (hypertension, diabetes, stroke). Among ESRD cases, blacks were younger at cohort enrollment compared to whites (median age 52 versus 59 years, respectively), and were more likely to be overweight or obese than white cases. Also, black ESRD cases were slightly more likely to have hypertension compared to white ESRD cases (86.4% versus 82.6%, respectively), but less likely to have a history of diabetes (67.8% versus 74.5%), stroke (11.3% versus 21.3%) or myocardial infarction/bypass (12.8% versus 26.2%).

Across all baseline eGFR categories, a higher proportion of ESRD cases compared to subcohort members were black; the only apparent exception, in race-by-sex tabulations, was a lower proportion of black women among ESRD cases with eGFR > 90 compared to subcohort (33.7% vs. 45.0%) (Additional file 1: Table S1).

The median (range) time of follow up for the overall study population was 9.3 years (6.9, 11.1). In multivariable models (Table 3), after adjustment for baseline eGFR as well as clinical, lifestyle and demographic characteristics (age, sex, BMI, smoking, education, income, hypertension, diabetes, stroke/TIA and MI/bypass), the greater than two-fold elevated ESRD risk among blacks compared to whites persisted (HR: 2.58; 95% CI: 1.65, 4.03). Women were at significantly lower risk for ESRD compared to men (HR: 0.42; 95% CI: 0.31, 0.58), and diabetes duration ≥20 years was strongly associated with ESRD risk in our study population (HR: 12.17; 95% CI: 7.34, 20.16); the association between hypertension and ESRD was attenuated by adjusting for baseline eGFR but remained statistically significant (HR: 1.96; 95% CI: 1.21, 3.19). The race and sex associations with ESRD observed in the cause-specific analysis were also observed in the multi-state model (Additional file 2: Table S2 and Additional file 4: Figure S1). The addition of medication use for diabetes (including insulin) to the models did not change the results.

In a partial effect plot based on multivariable Cox models, we examined the association between baseline eGFR and incident ESRD, by race. As shown in Fig. 3 (Additional file 3: Table S3), baseline eGFR was a strong predictor of ESRD risk. While the ESRD risk in blacks was significantly higher than the risk for whites for most levels of baseline eGFR < 100 ml/min/1.73m², the curves crossed at about eGFR < 40 ml/min/1.73m², demonstrating heterogeneity of the eGFR effect across racial groups (P for race*eGFR interaction = 0.029). Furthermore, compared to blacks, the slope of the curve for whites appeared steeper; each 10 mL/min lower eGFR at baseline was associated with a greater increase in log relative hazard of ESRD.

In this large case-cohort study in a population with a high burden of ESRD, we demonstrated that blacks compared to whites were at disproportionate risk for developing ESRD, after accounting for baseline kidney function and other common confounders. For most baseline eGFR values < 100 ml/min/1.73m², the risk of ESRD was higher in blacks than whites. Remarkably, more than half of black participants with ESRD had an eGFR ≥60 ml/min/1.73m² at baseline, with a median time from study enrollment to ESRD of 6–7 years. In addition, blacks had an age-adjusted ESRD incidence rate more than three times greater than their white counterparts despite having a marginally higher median eGFR at baseline than whites. Differences in baseline risk profiles were also apparent, with blacks who subsequently developed ESRD more likely to be obese and hypertensive while diabetes and CVD were more common among whites who developed ESRD.

The persistently higher risk of ESRD among blacks compared to whites suggests an influence of additional factors beyond those already known. Diabetes and hypertension remain the leading risk factors for ESRD. However, our study and others have previously demonstrated that these do not fully explain the racial disparity in ESRD risk [8]. In light of the relatively rapid development of ESRD described herein, the lower SES of most participants, and the availability of information on baseline kidney function, the SCCS cohort is uniquely positioned to study additional risk factors, including potentially modifiable lifestyle behaviors, for development of ESRD in this vulnerable population. The SCCS has resources available that allow further examination of metabolic and genetic profiles associated with CKD progression. Furthermore, the case-cohort design has several advantages. The subcohort we created can be used to study other outcomes in the context of chronic kidney disease and unlike in nested case-control studies the design allows for obtaining measurements at any time after the cohort was originally set up [17].

Currently, the management goals in CKD focus on influencing clinical factors (such as blood pressure and diabetes) and treating complications (anemia, acidosis, among others). However, our understanding of CKD progression is limited and effective interventions to prevent progression are lacking [18]. Apoliproprotein L1 (APOL1) risk variants, common in individuals of African ancestry, are associated with an increased risk of ESRD [19]. APOL1 affects endosomal trafficking, inflammasome activation and autophagic flux and can promote podocyte loss and glomerulosclerosis [18]. Blacks with two high-risk variants have a 7.3- to 29-fold increase in the odds of developing ESRD [20]. Therapies targeting APOL1-related injuries are currently being experimentally addressed [21]. Given its importance, clinical testing for APOL1 variants is under discussion but unexplained variance remains, and evidence of kidney disease is not observed among all carriers of two high-risk variants. Moreover, it is not known what comorbid conditions or exposures in patients with these high-risk alleles are associated with progression to ESRD [22]. Our observation of a high ESRD incidence rates in this population with relatively preserved kidney function at enrollment, highlights opportunities to identify and better characterize additional ESRD risk factors, including modifiers of the effect of APOL1.

Our study confirmed roles for diabetes and hypertension in kidney disease as observed in other large cohort studies such as the Framingham Heart/Offspring Study [23], Multiple Risk Factor Intervention Trial (MRFIT) [24], and Chronic Renal Insufficiency Cohort (CRIC) [25]. Strengths of our study include the availability of data on both baseline eGFR in addition to ESRD incidence. Furthermore, our study population included sizable numbers of blacks and women. The CRIC study included a significant number of blacks (42%), had a high prevalence of baseline cardiovascular risk factors (48% with diabetes and 86% with hypertension) and well-characterized baseline kidney function, and had a high ESRD incidence rate. By design, CRIC included a significant number (~ 20%) of participants who had significant kidney function impairment (eGFR < 30 ml/min/1.73m²) at baseline. In contrast, we observed a high incidence of ESRD in our study despite the large number of participants with a relatively preserved eGFR (70% had an eGFR > 90 at enrollment), perhaps signifying rapid progression over a 5–6 year period. There remains a need to characterize patterns of progression to ESRD and risk factors that may impact ESRD progression across a wide eGFR range in high risk groups.

For most levels of baseline eGFR < 100 ml/min/1.73m² blacks had a higher ESRD risk than whites. While the CRIC study demonstrated an increased ESRD risks of blacks compared to whites (HR 1.55, 95% CI 1.29–1.86) after adjusting for baseline eGFR, we demonstrate that the difference in ESRD risk may diminish at an eGFR of less than 40. These data may indicate a need to target blacks for CKD screening, treatment, and therapeutic trials earlier in their disease course, when yet to be elucidated risk factors may play a significant role in progression. As CKD advances, common mechanisms of progression including hyperfiltration injury, elevated glomerular pressure, and fibrosis, may predominate in determining the path to ESRD. Most current clinical ESRD risk calculators do not include race as a variable. Future studies should examine whether race is helpful when estimating ESRD risk based on baseline eGFR level and time horizon. The general benefit of these approaches requires further evaluation [26, 27].

A well- known risk factor for CKD progression to ESRD is proteinuria [28]. Different levels of proteinuria have been associated with racial differences of CKD progression [29] and this may hypothetically account for some of the racial disparities in ESRD risk. However, the role of proteinuria in the disparity of ESRD incidence among blacks remains ill-defined and the unavailability of baseline urine protein levels precluded assessment of this risk factor in the SCCS. Additional limitations of our study warrant consideration. These include the self-reported nature of the baseline questionnaire, the lack of time-dependent covariates and the lack of follow-up creatinine measurements, which precluded calculation of eGFR slopes over time. Finally, residual confounding arising from inadequate control of diabetes or hypertension severity cannot be excluded. Only participants who enrolled in 2004 or later and responded affirmatively to the question of whether a doctor ever told them they have high blood pressure were asked additional follow-up questions on the use of prescription medications for blood pressure control; thus restricting our analyses to those who enrolled after 2004 yielded an insufficient sample size.

We have previously shown a survival advantage among blacks compared to whites in the SCCS, both overall [30] and among those with diabetes [31], but in sensitivity analyses taking into account competing risks, higher rates of ESRD in black males and black females do not appear to be the result of different rates of death between black and white subjects.

In conclusion, we describe a cohort of blacks and whites with a high prevalence of baseline risk factors who, despite relatively high levels of baseline renal function, progress to ESRD. Furthermore, our study provides evidence that additional unknown risk factors account for the racial disparity of increased ESRD incidence in blacks, and demonstrates the need for early interventions that are not solely based on kidney function assessment in at-risk populations.