Incidence rates of heart failure (HF) associated with quintiles of cystatin C (A), estimated glomerular filtration rate (eGFR) (B), and creatinine (C). To convert creatinine to micromoles per liter, multiply by 88.4.
Spline analyses to compare heart failure risk across the range of cystatin C values, stratified by race.
Population attributable risk of heart failure associated with an elevated cystatin C concentration, stratified by race.
Bibbins-Domingo K, Chertow GM, Fried LF, Odden MC, Newman AB, Kritchevsky SB, Harris TB, Satterfield S, Cummings SR, Shlipak MG. Renal Function and Heart Failure Risk in Older Black and White IndividualsThe Health, Aging, and Body Composition Study. Arch Intern Med. 2006;166(13):1396-1402. doi:10.1001/archinte.166.13.1396
Chronic kidney disease is a risk factor for heart failure, an association that may be particularly important in blacks who are disproportionately affected by both processes. Our objective was to determine whether the association of chronic kidney disease with incident heart failure differs between blacks and whites.
The study population comprised participants in the Health, Aging, and Body Composition Study without a diagnosis of heart failure (1124 black and 1676 white community-dwelling older persons). The main predictors were quintiles of cystatin C and creatinine concentrations and estimated glomerular filtration rate. The main outcome measure was incident heart failure.
Over a mean 5.7 years, 200 participants developed heart failure. High concentrations of cystatin C and low estimated glomerular filtration rate were each associated with heart failure, but the magnitude was greater for blacks than for whites (cystatin C concentration: adjusted hazard ratio for quintile 5 [≥1.18 mg/dL] vs quintile 1 [<0.84 mg/dL] was 3.0 [95% confidence interval 1.4-6.5] in blacks and 1.4 [95% confidence interval, 0.8-2.5] in whites; estimated glomerular filtration rate: adjusted hazard ratio for quintile 5 (<59.2 mL/min) vs quintile 1 (>86.7 mL/min) was 2.7 [95% confidence interval, 1.4-4.9] in blacks and 1.8 [95% confidence interval, 0.9-3.6] in whites). For cystatin C, this association was observed at more modest decrements in kidney function among blacks as well. The population attributable risk of heart failure was 47% for blacks with moderate or high concentrations of cystatin C (≥0.94 mg/dL) (56% prevalence) but only 5% among whites (64% prevalence).
The association of kidney dysfunction with heart failure appears stronger in blacks than for whites, particularly when cystatin C is used to measure kidney function.
Individuals with impaired kidney function are at risk for developing heart failure.1- 4 No large epidemiologic study has determined whether the risk of incident heart failure associated with impaired kidney function differs between blacks and whites. Blacks are significantly affected by both chronic kidney disease (CKD) and heart failure. Blacks are more likely to develop CKD at younger ages compared with whites,5,6 and blacks with CKD are more likely to progress to end-stage renal disease,7 differences not fully accounted for by racial differences in risk factors such as hypertension or its treatment.8- 10 Heart failure is more prevalent among blacks,11 and blacks with systolic dysfunction,12,13 diastolic dysfunction,14 or left ventricular hypertrophy15 progress more rapidly to clinical heart failure and death compared with their white counterparts.
We sought to determine the relative importance of kidney function in the development of heart failure among elderly blacks and whites. We determined the association between 3 markers of kidney function (creatinine, estimated glomerular filtration rate [eGFR], and cystatin C) and incident heart failure in the Health, Aging, and Body Composition (Health ABC) Study. Because blacks are disproportionately affected by CKD and heart failure, we hypothesized that kidney dysfunction would have a stronger association with heart failure risk in blacks than in whites. We further hypothesized that markers most sensitive to small changes in kidney function (cystatin C concentration and eGFR) would better capture the association of kidney function with heart failure risk, particularly among blacks.
The Health ABC Study was designed to explore causes of functional decline linked to weight and weight-related conditions in a cohort of highly functional older persons, nearly balanced among men and women and blacks and whites. Two study sites, Pittsburgh, Pa, and Memphis, Tenn, each recruited participants aged 70 to 79 years from a list of Medicare-eligible adults between April 1997 and June 1998. Whites were recruited from a random sample of Medicare beneficiaries; blacks were recruited from all age-eligible subjects within the regions. Race was defined by participant self-report. Participants were included in the Health ABC if they (1) were able to walk a quarter of a mile (400 meters), climb 10 steps, and perform basic mobility-related activities of daily living without difficulty and without the need of assistive devices; (2) did not have a life-threatening illness; (3) planned to remain in the geographic area for at least 3 years; and (4) were not participating in another trial of a lifestyle intervention. The cohort enrolled 3075 participants who completed baseline evaluations, with 40% of participants being black. All participants gave informed written consent. The protocol was approved by the institutional review boards of the clinical sites and the Data Coordinating Center of the University of California, San Francisco.
To identify a cohort without heart failure, we excluded the 94 participants who reported a history of heart failure on entry into the Health ABC Study and an additional 148 participants who were using loop diuretics on entry into the study (because loop diuretics are common in the treatment of heart failure). We excluded 31 participants who did not have adequate plasma samples for cystatin C measurement and 2 other participants who did not have complete records for this analysis, leaving 2800 participants for this analysis. Because loop diuretics can be used for the treatment of CKD as well as for the treatment of heart failure, we repeated the analyses with inclusion of individuals using loop diuretics.
Plasma cystatin C and serum creatinine concentrations were calculated from serum samples drawn from participants at their baseline examination. Cystatin C levels were measured at the Health ABC core laboratory (University of Vermont, Burlington) using a BNII nephelometer (Dade Behring Inc, Deerfield, Ill) that used a particle-enhanced immunonephelometric assay (N Latex Cystatin C; Dade Behring Inc).16 Among 61 healthy individuals with 3 measurements of cystatin C levels over a 6-month period, the intraindividual coefficient of variation was 7.7%, reflecting long-term stability of the measurement. The assay range is 0.195 to 7.330 mg/dL, with the reference range for young, healthy individuals reported as 0.53 to 0.95 mg/dL. The assay remained stable over 5 cycles of freezing and thawing without change in the measurement. Because cystatin C values were not normally distributed, we conducted analyses using quintiles of cystatin C concentrations based on values for the entire cohort (<0.84 mg/dL, 0.84-0.93 mg/dL, 0.94-1.03 mg/dL, 1.04-1.17 mg/dL, and ≥1.18 mg/dL).
Serum creatinine concentrations were measured by a colorimetric technique on a Johnson & Johnson Vitrols 950 analyzer (Johnson & Johnson, New Brunswick, NJ) (2% coefficient of variation). We calculated eGFR using the abbreviated (4-variable) Modification of Diet in Renal Disease equation17 and categorized participants into quintiles of eGFR based on the entire cohort and sex-specific quintiles for serum creatinine concentration.
To determine whether the association of cystatin C concentration with incident heart failure was independent, we evaluated potential confounders in multivariable regression analyses. These included sociodemographic factors (age, sex, race, clinical site, and self-reported education level); lifestyle factors (self-reported current smoking and alcohol use and baseline body mass index as a continuous measure); comorbid conditions (self-reported diabetes, coronary heart disease, and cerebrovascular disease; left ventricular hypertrophy on baseline electrocardiogram; and baseline systolic blood pressure as a continuous measure); and serum chemistry analyses (glucose, total cholesterol, high-density lipoprotein cholesterol, triglycerides, and albumin). Serum chemistry levels were measured from blood samples drawn on entry by a colorimetric technique on a Johnson & Johnson Vitrols 950 analyzer. Low-density lipoprotein cholesterol levels were calculated using the equation of Friedewald et al.18 Medications were brought in by the participant at baseline and recorded using the Iowa Drug Information System (IDIS) and then classified into major therapeutic groups; we included aspirin, β-blockers, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, statins, and non-loop diuretics in this analysis as potential confounders.
Incident heart failure was defined as heart failure requiring overnight hospitalization in an acute care hospital. This outcome was ascertained during interviews with participants that occurred every 6 months either in person or over the telephone. When an event was reported, hospital records were collected and evaluated by a Health ABC Disease adjudicator at each clinical site. A hospitalization was attributed to heart failure when at least 1 of the following conditions was present and judged to be responsible for the hospitalization: (1) physician diagnosis of congestive heart failure and treatment with a diuretic plus digitalis or a vasodilator (eg, nitroglycerin, hydralazine, or angiotensin-converting enzyme inhibitor), (2) cardiomegaly and pulmonary edema on chest radiograph, or (3) evidence of a dilated ventricle and global or segmental wall motion abnormalities with decreased systolic function either by echocardiography or contrast ventriculography.
We compared baseline characteristics of black and white participants using the 2-tailed, unpaired t test, the Wilcoxon rank sum test, and the χ2 test, as appropriate. We examined the association of quintiles of each marker of kidney function with incident heart failure using unadjusted and adjusted proportional hazards models. Age, sex, and recruitment site were retained in all models. Other covariates associated with the incident heart failure at P<.10 were also retained in the final model. This method was cross-checked using an automated backward-selection procedure.
Incident heart failure rates were calculated per quintile of each marker of kidney function in black and white participants. To explore the association of intermediate ranges of cystatin C concentration with heart failure, we used a smoothing spline procedure to model the hazard ratio (HR) over the range of cystatin C values in blacks and whites.
We adapted to methods of Kooperberg and Petitti19 to estimate the population attributable risk for each quintile of cystatin C concentration. By using this approach, population attributable risk is estimated by the average excess hazard divided by the average hazard.20 Because the fitted hazard is equal to the baseline hazard times the HR, we can substitute HRs obtained from the Cox model in this computation. Thus, the excess hazard for each participant is estimated by the difference between the fitted HR from the adjusted model and the fitted HR that would be obtained if the participant's cystatin C level were in the lowest quintile, holding other covariates constant; this quantity is easily computed using results from standard Cox model software. The primary rationale for this procedure is to ensure that the sum of the population attributable risks for different exposures never exceeds 100%.
All statistical analyses were conducted using Stata 8.0 (Stata Corporation, College Station, Tex), with the exception of the spline analyses, which were performed using R software (R Foundation for Statistical Computing, Vienna, Austria), and the population attributable risk, which was calculated using SAS version 8.2 software (SAS Inc, Cary, NC).
Among the 2800 participants in the Health ABC Study included in this analysis, 1124 were black and 1676 were white. Compared with white participants, blacks were more likely to be female, diabetic, and current smokers and more likely to have left ventricular hypertrophy and were less likely to have completed high school or to drink alcohol regularly (Table 1). Black participants had a higher mean body mass index and systolic blood pressure and higher low- and high-density lipoprotein cholesterol concentrations but had lower triglyceride levels. Black participants were less likely to be using β-blockers, statins, and antiplatelet agents. Mean serum creatinine concentrations and eGFR were higher among blacks, while concentrations of cystatin C were lower.
Over a mean 5.7 years of observation, 7.7% of blacks (87 of 1124) and 6.7% of whites (113 of 1676) developed heart failure, with a higher incidence rate for blacks than for whites (14.0 events per 1000 person-years vs 11.6 per 1000 person-years; P<.001). Decreased kidney function was associated with the risk of heart failure in the entire cohort. Individuals with the highest quintile of each marker of kidney function were at twice the risk of heart failure (compared with those in the lowest quintile), even after adjusting for demographic factors, comorbid conditions, and medications (cystatin C concentration: HR = 1.9 [95% confidence interval, 1.2-3.0]; eGFR: HR = 2.0 [95% confidence interval, 1.3-3.2]; and creatinine level: HR = 2.0 [95% confidence interval, 1.2-3.4]).
We observed that the association of the highest quintile of cystatin C concentration and eGFR with heart failure differed significantly by race (Table 2; adjusted P for interaction, .02 for both cystatin C concentration and eGFR). In other words, higher cystatin C concentration and lower eGFR were each associated with higher heart failure risk in both races, but the risk was much more pronounced among blacks. Blacks with the worst kidney function using cystatin C concentration or eGFR were at 3 times the risk of heart failure compared with those with the best kidney function, whereas the risk in whites for the worst kidney function was more modest and did not reach the level of statistical significance (Table 2). We found no significant interaction between either cystatin C concentration or eGFR and sex; conversely, the interaction between each marker of kidney function and race was present within both men and women. We also observed the same results when individuals using loop diuretics were included in the analysis.
We explored whether the association between kidney dysfunction and heart failure risk was observed at intermediate levels of kidney dysfunction as well. Although both cystatin C concentration and eGFR exhibited an association with heart failure risk at the highest quintiles that was stronger for blacks than for whites, this risk was observed at intermediate levels only for cystatin C concentration (Figure 1). Each successive quintile of cystatin C concentration was associated with a graded increased risk of heart failure in blacks, a risk that was confirmed in our multivariable models (Table 3).
This impression was reinforced by the spline curves displayed in Figure 2. Among white participants, the association between cystatin C concentration and heart failure was only observed at the highest concentrations of cystatin C, with no increased risk below a concentration of 1.2 mg/dL. By contrast, among black participants, a nearly linear relationship between cystatin C concentration and heart failure risk was confirmed across the entire range of cystatin C values.
We determined the population attributable risk of heart failure for increasing concentrations of cystatin C among blacks and whites (Figure 3). When we considered the overall risk associated with concentrations of cystatin C in quintiles 3 through 5 (≥0.94 mg/dL), we found that nearly half of the risk for incident heart failure in blacks appeared to be attributable to high cystatin C concentrations in blacks (prevalence of cystatin C concentration ≥0.94 mg/dL, 56%), whereas the corresponding population attributable risk was only 5% for whites (prevalence, 64%).
In this study, we found that the association of kidney function with heart failure risk was stronger among older black persons compared with older white persons, and this difference was particularly evident when we used cystatin C concentration to measure kidney function. The more linear association of cystatin C concentration with heart failure risk in blacks may be particularly important for assessing the presence and stage of CKD among black patients. Our results suggest that a modest increase in cystatin C concentration is strongly associated with increased risk of heart failure among blacks and had a calculated attributable risk estimate of approximately 50%.
Although prior studies have demonstrated the increased risk of heart failure associated with CKD,1- 3 ours is among the first to demonstrate a stronger association for elderly blacks than for elderly whites and a nearly linear relationship between kidney function, as measured by cystatin C concentration, and heart failure risk in blacks. Cystatin C concentration is a novel measure of kidney function that overcomes many of the limitations of using creatinine level measurements because the association between kidney function and cystatin C concentration does not appear to vary by age, sex, and body mass.21- 24 Several recent studies also suggest that cystatin C concentration is more sensitive for detecting small changes in renal function over a broad spectrum of GFR, overcoming the limitations of the Modification of Diet in Renal Disease–based estimates that fail to discriminate higher GFRs.22- 24 We hypothesize that the improved sensitivity of cystatin C concentration as a measure of kidney function allowed us to observe this relationship between modest declines in kidney function and heart failure risk that may be of particular importance in blacks and may be undetected using creatinine-based measures of kidney function.
The increased risk from impaired kidney function in blacks may reflect racial variations in the direct effects of the kidneys on the heart. Differences by race have been observed in the renin-angiotensin-aldosterone system,25 in sensitivity to salt,26,27 and in ventricular remodeling.28,29 The risk of heart failure associated with impaired kidney function may also reflect manifestations of end-organ damage from a risk factor common to both the kidney and the heart, such as hypertension, that is on average more severe among blacks compared with whites.10,30,31
Although we hypothesize that the link between elevated cystatin C concentrations and heart failure reflects the improved ability of cystatin C concentration to discriminate levels of kidney function, we cannot exclude the possibility that cystatin C concentration is itself a novel risk factor for cardiovascular events and may even be directly pathogenic.32 However, our hypothesis that elevations in cystatin C concentration predominantly represent decrements in kidney function is supported by our observation that a reduced eGFR lower than 60 mL/min was similarly associated with heart failure, and that association was also more pronounced among blacks.
Cystatin C concentration may offer an improvement over creatinine-based eGFR by detecting the risk present at an eGFR higher than 60 mL/min.4,33,34 In our analysis, elevated cystatin C concentration (≥0.94 mg/dL) was associated with a significant increased risk of heart failure among blacks with an eGFR higher than 60 mL/min. While this observation may simply reflect improved discrimination at modest declines in kidney function associated with cystatin C concentration compared with creatinine-based eGFR, it is also possible that there is an elevated risk of cardiovascular complications associated with the “preclinical” stages of CKD (before eGFR drops below 60 mL/min) that may be captured by the more sensitive cystatin C measure. With increasing attention focused on heart failure prevention and the early identification of modifiable risk factors, markers such as cystatin C concentration that signify the risk associated with kidney disease in its early stages may be particularly useful. Our observation that the risk of heart failure in this “preclinical” period is particularly prominent for black individuals may be in keeping with the epidemiologic findings of higher prevalence of heart failure in blacks but merits confirmation in other studies with sizeable numbers of black individuals.
Our findings are similar to those from the Cardiovascular Health Study (CHS),4 although we did not observe the significant association between cystatin C concentration and incident heart failure among white participants who were observed in the CHS, which included a larger elderly cohort who were mostly white. Although this difference between the Health ABC Study and the CHS in the association between cystatin C concentration and heart failure in whites is puzzling, we believe that this difference most likely is caused by some combination of the smaller number of whites in the Health ABC Study, the lower concentrations of cystatin C in whites in the Health ABC Study compared with the CHS, and chance. Interestingly, a stronger association between cystatin C concentration and heart failure in blacks compared with whites was also noted in the CHS, although a detailed analysis of these differences by race was beyond the scope of the CHS.
There are several limitations to our study. We studied a population of high-functioning older persons, most without evidence of kidney dysfunction or mild CKD. As a consequence, our results may not generalize to younger individuals or those with a more severe spectrum of kidney dysfunction. Second, we used participant self-report of heart failure, verified by use of heart failure medications, to define a cohort without preexisting disease. This approach may be subject to misclassification because medications may be used for multiple indications. We also cannot exclude the possibility that differences in educational attainment and access to care between blacks and whites may lead to systematic differences in the self-report of heart failure and other comorbid conditions. Third, because echocardiographic data were not collected as a part of the Health ABC Study, we cannot evaluate whether the association between cystatin C concentration and heart failure varied by systolic or diastolic dysfunction. In addition, other potential confounders such as hemoglobin levels were not measured in the Health ABC Study and therefore were not included in this analysis.
Our study has several implications. We have confirmed the importance of kidney dysfunction as a predictor of heart failure but demonstrated a much stronger association in blacks than in whites. As cystatin C concentration detects a more linear gradient of risk for blacks compared with creatinine-based eGFR, measurement of cystatin C may lead to better characterization of this risk in blacks. Although our study is unable to elucidate the mechanisms that underlie these differences, such exploration should be the focus of future studies, particularly in cohorts with sizable numbers of black participants. Understanding the test characteristics and risk associations of cystatin C concentration in a diverse patient population is crucial if this test is to be adopted more widely as a measure of kidney function. Exploring the link between CKD and heart failure is particularly important in blacks, a population that is significantly affected by both conditions, to better guide strategies for risk reduction and prevention.
Correspondence: Kirsten Bibbins-Domingo, PhD, MD, Division of General Internal Medicine—SFGH, University of California, San Francisco, Box 1364, San Francisco, CA 94143-1364 (firstname.lastname@example.org).
Accepted for Publication: April 7, 2006.
Financial Disclosure: None reported.
Funding/Support: The Health ABC Study was supported through the National Institute on Aging contracts N01-AG-6-2101, N01-AG-6-2103, and N01-AG-6-2106. Dr Bibbins-Domingo is funded by the Robert Wood Johnson Foundation (Amos Faculty Development Program) and by the National Heart, Lung, and Blood Institute contract N01-HC-95095 (Diversity Supplement). Dr Shlipak is funded by R01 HL073208-01, R01 DK 066488, the American Federation for Aging Research and National Institute on Aging (Paul Beeson Scholars Program), and the Robert Wood Johnson Foundation (Generalist Faculty Scholars Program).
Additional Information: A formula for the estimation of population attributable risk is available from the authors.