Chronic kidney disease (CKD) is an important risk factor for the development of heart failure (HF), myocardial infarction (MI), stroke, and peripheral arterial disease (PAD).1–5 The use of cystatin C in older adults has extended these associations across a broader range of kidney function than could be appreciated with creatinine-based estimated GFR (eGFR). Nonetheless, the mechanisms underlying the association between CKD and cardiovascular risk remain incompletely determined. Current understanding of kidney–cardiovascular disease (CVD) relationships is limited by the measurement of kidney function on a single occasion. This approach fails to discriminate people with declining kidney function from those with diminished but stable kidney function. An association between dynamic changes in kidney function with cardiovascular events independent of both other cardiovascular risk factors and the current level of kidney function would add risk information and strengthen the argument that reduced kidney function has a causal role in CVD.
In the Cardiovascular Health Study (CHS), a community-based cohort of ambulatory elderly individuals, kidney function was measured at the baseline visit and at years 3 and 7 of follow-up. In a recent article, we evaluated the associations of changes in kidney function using each measure with subsequent mortality risk.6 Using either cystatin C (eGFRcys) or creatinine (eGFRcreat), annual eGFR declines of >3 ml/min per 1.73 m2 were independently associated with elevated all-cause and cardiovascular mortality risk; however, no association was observed for milder changes in kidney function.
In this study from CHS, we evaluated the associations of rapid decline in kidney function with the incidence of specific cardiovascular end points: HF, MI, stroke, and PAD. We hypothesized that rapid kidney function decline would have independent associations with higher incidence of each outcome but with the strongest association with HF on the basis of our previous studies using baseline kidney function.7–9
Characteristics of Study Population
Among 5888 participants with baseline levels of kidney function, 4380 had repeated measurements of both creatinine and cystatin C. Of the remaining 1508, 204 died before the second measurement of kidney function; the remainder either lacked adequate blood samples or did not present for follow-up clinic visits (Supplemental Table 1). Overall, among the 4378 participants, 1083 (25%) had rapid decline (Table 1) in kidney function on the basis of eGFRcys. Patients with rapid decline were slightly older at baseline and more likely to be black. Risk factors more common in patients with rapid decline included diabetes and hypertension but not smoking or lipoprotein levels. Those with rapid decline had somewhat higher baseline eGFRcys and eGFRcreat but substantially lower follow-up measures of eGFRcys and eGFRcreat.
We initially evaluated the association of quintiles of kidney function decline by eGFRcys with each CVD outcome. For the outcomes of HF, MI, and stroke, we observed little difference across the lower four quintiles in either demographic adjusted or fully adjusted analyses, and each hazard ratio (HR) was approximately 1.0 for quintiles 2 through 4. In contrast, for PAD events, intermediate risk was observed for quintiles 2 and 3 and much higher risk for quintiles 4 and 5. Compared with the lowest quintile of decline, the highest quintile was associated with adjusted HRs of 1.33 (95% confidence interval [CI] 1.08 to 1.65) for HF, 1.35 (95% CI 1.01 to 1.81) for MI, 1.16 (95% CI 0.87 to 1.55) for stroke, and 2.93 (95% CI 1.21 to 7.09) for PAD. All results were similar using quintiles of rapid decline by eGFRcreat except for PAD events; none of the quintiles of change in eGFRcreat was significantly associated with PAD.
We next dichotomized change in kidney function at >3 ml/min per 1.73 m2 and found the incidences of HF, MI, stroke, and PAD events to be significantly higher among participants with previous rapid kidney function decline (Table 2). After adjustment for demographic characteristics, rapid decline remained associated with a higher risk for each CVD event (Table 2); however, after further adjustment for CVD risk factors, rapid decliners had an approximately 30% higher risk for HF, 40% higher risk for MI, and 60% higher risk for PAD, although the PAD finding did not reach statistical significance. In models using eGFRcreat, rapid kidney decline was significantly associated with only incident HF; HRs were 1.45 (95% CI 1.22 to 1.74) for HF, 1.22 (95% CI 0.94 to 1.58) for MI, 1.21 (95% CI 0.95 to 1.53) for stroke, and 1.36 (95% CI 0.76 to 2.46) for PAD.
We next categorized participants by their CKD status at the final kidney measurement and by whether they had previous rapid kidney function decline (Table 3). Overall, participants who had either CKD or rapid decline had higher event rates compared with participants with neither. In multivariate analysis, no interactions were observed between CKD and rapid decline for any of the four outcomes; however, for HF, rapid decline seemed associated with increased risk only among participants with CKD. In contrast, for MI and PAD, rapidly declining kidney function was independently associated with higher risk even among participants without CKD.
In this study, we evaluated the independent associations of longitudinal declines in kidney function with four CVD end points. Independent of demographic characteristics and CVD risk factors and of baseline kidney function, rapid declines in kidney function (>3 ml/min per 1.73 m2/yr) detected by eGFRcys were associated with higher risk for HF, MI, and PAD events but not stroke. Analyses using creatinine found rapid decline to be associated only with risk for HF. To our knowledge, this is the first study to determine the association between longitudinal changes in cystatin C and risk for CVD events. If confirmed in future studies, then the findings suggest that both the current level of kidney function and the trajectory are important for assessing cardiovascular risk.
The primary importance of this article is that it provides additional evidence regarding the likelihood that reduced kidney filtration has a causal association with CVD. Although the presence of CKD is generally considered to be a risk factor for CVD, proving a causal association is challenging.10 The finding that dynamic changes in kidney function are independently associated with CVD risk adds to the likelihood that reduced kidney function is causally related to CVD onset and progression. Potential pathways that could mediate this association include the kidney's effects on BP regulation, sodium handling, retention of toxic solutes, and disrupted mineral metabolism. Regardless of the underlying mechanisms, our findings raise the hypothesis that interventions that slow the decline of kidney function could have salient effects on CVD risk even in patients without overt CKD.
From the clinical perspective, our findings in Table 3 suggest that a provider would gain insight into a patient's cardiovascular risk by considering not only the current kidney function but also the recent trajectory. Although this would be obvious for estimating the risk for progression to kidney failure, to our knowledge, it has never been demonstrated for CVD risk. Because the risk for rapid decline was apparent before the onset of CKD, these results suggest that interventions to stabilize kidney function in the normal range could have substantial health benefits. The association of rapid kidney decline with CVD events was stronger using cystatin C than creatinine as the indicator of kidney function decline. Our previous studies found that cystatin C identified a larger proportion of participants as having rapid decline than did creatinine but that rapid decline by either measure was strongly associated with mortality risk. It is not clear why creatinine changes are more prognostic for mortality than CVD, but we believe cystatin C likely offers a more accurate reflection of longitudinal kidney function.
The primary limitation of this study is that we still cannot determine definitively whether the associations of changes in kidney function are causally related to the incidence of CVD. Although we adjusted for confounding factors, it is possible that another process caused both the declines in kidney function and the increase in CVD events in parallel. An additional limitation is that we used indirect measurements of kidney function that were measured at most three times during 7 yr of follow-up, and we did not have baseline measurements of albuminuria. The infrequency of kidney function measurement allowed many participants to be censored before their repeat measurements. Because these excluded participants had worse kidney function at baseline than included participants, our results may not necessarily generalize to patients with advanced kidney disease; however, we believe that rapid kidney decline would be at least as deleterious in patients with low eGFR, who have less physiologic reserve.
Previously observed evaluations of the association of kidney disease with CVD have overlooked the dynamic aspects of kidney function, which may play an integral role in CVD risk. Using longitudinal measurements of cystatin C, we found rapid declines in kidney function to be independently associated with higher risk for HF, MI, and PAD, even in patients without CKD. If replicated in future studies, then these findings suggest that interventions to stabilize kidney function over time could have the additional benefits of decreasing CVD risk. Future studies should consider not only static measures of kidney function but also ongoing changes in function to appreciate fully the cardiovascular consequences of kidney disease.
The objective of this longitudinal study was to evaluate the association of changes in kidney function during the first 7 yr of CHS follow-up with incident cardiovascular events that occurred after the final measurement of kidney function. The timeline of this study is illustrated in Figure 1. For each of the four cardiovascular outcomes, participants were excluded if they had experienced that outcome at any time before their final measurement of kidney function.
The CHS is a longitudinal study of community-dwelling older adults designed to evaluate risk factors for CVD. The design of this study has been described previously.11 In brief, participants ≥65 yr of age were recruited from Medicare eligibility lists in four US communities (Forsyth County, NC; Sacramento County, CA; Washington County, MD; and Pittsburgh). An initial 5201 participants were recruited between 1989 and 1990. Blood samples were drawn from these individuals in 1989 to 1990 (baseline), 1992 to 1993 (year 3), and 1996 to 1997 (year 7). 687 black participants were added to the study in 1992 to 1993; these individuals had blood samples drawn in 1992 to 1993 (year 3) and 1996 to 1997 (year 7). In total, 4380 of the 5888 participants in CHS had at least two blood samples from these visits available for measurement of cystatin C and thus met criteria for potential inclusion in analysis of cardiovascular outcomes.
Measurements of Kidney Function
The primary measure of kidney function for analysis was cystatin C. Frozen sera stored at −70°C from the visits at baseline, year 3, and year 7 were available for measurement of cystatin C. Cystatin C was measured using a particle-enhanced immunonephelometric assay (N Latex Cystatin C; Siemens) with a nephelometer (BNII; Siemens). Previous work showed this assay to be stable through several freeze–thaw cycles.12 For cystatin C, intra-assay coefficients of variation (CVs) range from 2.0 to 2.8% and interassay CVs range from 2.3 to 3.1%. The 1992 to 1993 blood samples for cystatin C were assayed first, in 2003. Subsequently, the 1989 to 1990 and 1996 to 1997 levels were assayed in 2006.
As a confirmatory measure, kidney function was also analyzed using creatinine. Creatinine was measured immediately after each group of samples were drawn, using a colorimetric method (Ektachem 700; Eastman Kodak). The mean CV for monthly controls was 1.94 (range 1.16 to 3.60%).
Predictor Variables: Change in eGFR Using Cystatin C and Creatinine
For measurement of (eGFRcys), we used the CKD epidemiology equation, which was derived from a pooling of cohorts that used iothalamate clearance as the criterion standard (eGFRcys = 76.7 * cysC−1.19).13 This cystatin C equation is based on the largest cohort of individuals from multiple data sources that all used nephelometric methods for cystatin C. We indirectly calibrated the measured serum creatinine in the CHS cohort to the Cleveland Clinic Laboratory using Third National Health and Nutrition Examination Survey (NHANES III) data, as described previously14,15
As a secondary measure of kidney function, we estimated GFR using indirectly calibrated serum creatinine (eGFRcreat), and the four-variable Modification of Diet in Renal Disease (MDRD) equation: eGFRcreat = 186.3 × serum creatinine−1.154 × age−0.203 × 1.212 (if black) × 0.742 (if female).16
Rates of annual change in eGFR for each marker were calculated using the two or three available measurements. We defined a “rapid decline” in kidney function by an annual loss >3 ml/min per 1.73 m2.17 This magnitude of change is three times the expected rate previously described in aging studies, which represents the highest quartile of kidney function loss in CHS.18 Although we had observed a threshold effect in our previous study,6 in this study we also evaluated quintiles of eGFR decline to determine whether smaller changes were associated with each CVD end point.
Outcome: Cardiovascular Events
CVD events were ascertained by annual examinations and interim 6-mo telephone interviews through June 30, 2005, with a median follow-up time of 9.9 yr (maximum 11.1 yr). The methods of ascertaining and adjudicating events have been described previously.19 Briefly, the definitions of each end point are described next.
For the diagnosis of incident HF, a physician's diagnosis was followed by a review of the participant's medical records; the CHS Cardiovascular Events Committee based the diagnosis on physician report, as well as consideration of symptoms, signs, chest x-ray findings, and treatment of HF.2,19 MI was defined from hospital records by the clinical history of cardiac symptoms, cardiac enzymes, and serial electrocardiogram changes. Stroke cases were adjudicated by a committee of neurologists, neuroradiologists, and internists on the basis of patient interviews, medical records, and brain imaging studies.20 Incident PAD events were identified by International Classification of Diseases, Ninth Revision, Clinical Modification procedure codes for lower extremity bypass surgery, major lower extremity amputation, and peripheral angioplasty.
We chose a broad range of candidate covariates as potential confounding factors on the basis of their biologic plausibility or on previous studies linking them to CVD. The following covariates were examined: Baseline cystatin C and creatinine; demographic variables (age, gender, and race); cardiovascular risk factors, hypertension defined by history and use of antihypertensive agents, or an average of three BP measurements >140/90 mmHg; diabetes defined by use of insulin or an oral hypoglycemic agent or a fasting blood sugar >126 mg/dl; total cholesterol, LDL cholesterol, HDL cholesterol, and triglycerides; and novel cardiovascular risk factors (C-reactive protein, fibrinogen, and hemoglobin). For the multivariate models, BP was included both using a categorical variable and as the systolic and diastolic BP reading.
Rate of change in kidney function using cystatin C was categorized by quintiles and also as a dichotomous variable, with the cut point at 3 ml/min per 1.73 m2. Unadjusted incidence rates for each cardiovascular outcome were calculated per 1000 person-years by kidney function category. Characteristics of participants were compared with and without rapid decline in kidney function.
Patients were followed from the completion of kidney function measurements until the first cardiovascular outcome of interest or until they were censored because of death, loss to follow-up, or the end of outcome ascertainment for this study in June 2005. Cox proportional hazards models were used to estimate the association of kidney function decline with each cardiovascular outcome. To address the possibility of regression to the mean, we considered several different statistical adjustments for baseline kidney function: Adjustment for baseline kidney function, final kidney function, and average kidney function.21,22 Our results were robust regardless of the method used. We present results adjusted for the average kidney function over the two or three measured values.
The initial multivariate analysis, model 1, included adjustment for kidney function, age, gender, and race. The final model, model 2, included model 1 covariates and added selected covariates from the candidates listed above in the covariate section. Variables were retained in the final model when their inclusion altered the coefficient for the primary predictor variable (rapid decline in kidney function) by at least 5%. Interactions between baseline and change in kidney function were evaluated for each outcome. Analyses were performed using S-Plus 8.0 (Insightful, Inc., Seattle, WA) and SPSS 126.96.36.199 (SPSS, Inc., Chicago, IL).
We also conducted analyses that categorized participants at their final kidney measurement by the presence or absence of CKD (eGFRcys <60) and by whether they had rapid decline in kidney function. This analysis reflects the clinician's perspective by asking whether a participant's recent trajectory of kidney function decline is a risk factor independent of the current eGFR. Both demographic-adjusted and full multivariate analyses were conducted, and interactions were tested between rapid decline and CKD status.
This work was supported by M.S.'s American Heart Association Established Investigator Award and R01 DK066488 (principal investigator M.S.). In addition, it was supported by R01 AG027002 (principal investigator M.J.S.).
Published online ahead of print. Publication date available at www.jasn.org.
See related editorial, “It's about Time: Extending our Understanding of Cardiovascular Risk from Chronic Kidney Disease,” on pages 2486–2487.
Supplemental information for this article is available online at http://www.jasn.org/.
1. Bibbins-Domingo K, Lin F, Vittinghoff E, Barrett-Connor E, Hulley SB, Grady D, Shlipak MG: Predictors of heart failure among women with coronary disease. Circulation 110: 1424–1430, 2004
2. Gottdiener JS, Arnold AM, Aurigemma GP, Polak JF, Tracy RP, Kitzman DW, Gardin JM, Rutledge JE, Boineau RC: Predictors of congestive heart failure in the elderly: The Cardiovascular Health Study. J Am Coll Cardiol 35: 1628–1637, 2000
3. Fried LF, Shlipak MG, Crump C, Bleyer AJ, Gottdiener JS, Kronmal RA, Kuller LH, Newman AB: Renal insufficiency as a predictor of cardiovascular outcomes and mortality in elderly individuals. J Am Coll Cardiol 41: 1364–1372, 2003
4. Brugts JJ, Knetsch AM, Mattace-Raso FU, Hofman A, Witteman JC: Renal function and risk of myocardial infarction in an elderly population: The Rotterdam Study. Arch Intern Med 165: 2659–2665, 2005
5. O'Hare AM, Glidden DV, Fox CS, Hsu CY: High prevalence of peripheral arterial disease in persons with renal insufficiency: Results from the National Health and Nutrition Examination Survey 1999–2000. Circulation 109: 320–323, 2004
6. Rifkin D, Shlipak MG, Katz R, Fried L, Siscovick DS, Chonchol M, Newman AB, Sarnak MJ: Rapid kidney function decline and mortality risk in older adults. Arch Intern Med 168: 2212–2218, 2008
7. Shlipak MG, Sarnak M, Katz R, Fried LF, Seliger SL, Newman AB, Siscovick DS, Stehman-Breen CO: Cystatin-C and risk for mortality and cardiovascular disease in elderly adults. N Engl J Med 352: 2049–2060, 2005
8. Sarnak MJ, Katz R, Stehman-Breen CO, Fried LF, Jenny NS, Psaty BM, Newman AB, Siscovick D, Shlipak MG: Cystatin-C as a risk factor for heart failure in older adults. Ann Intern Med 142: 497–505, 2005
9. O'Hare AM, Newman AB, Katz R, Fried LF, Stehman-Breen CO, Seliger SL, Siscovick DS, Shlipak MG: Cystatin C and incident peripheral arterial disease events in the elderly: Results from the Cardiovascular Health Study. Arch Intern Med 165: 2666–2670, 2005
10. Sarnak M, Levey AS, Schoolwerth AC, Coresh J, Culleton B, Hamm LL, McCullough PA, Kasiske BL, Kelepouris E, Klag MJ, Parfrey P, Pfeffer M, Raij L, Spinosa DJ, Wilson PW American Heart Association Councils on Kidney in Cardiovascular Disease, High Blood Pressure Research, Clinical Cardiology, and Epidemiology and Prevention: Kidney disease as a risk factor for development of cardiovascular disease: A statement from the American Heart Association Councils on Kidney in Cardiovascular Disease, High Blood Pressure Research, Clinical Cardiology, and Epidemiology and Prevention. Circulation 108: 1254–1269, 2003
11. Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller LH, Manolio TA, Mittelmark MB, Newman A, et al: The Cardiovascular Health Study: Design and rationale. Ann Epidemiol 1: 263–276, 1991
12. Finney H, Newman DJ, Gruber W, Merle P, Price CP: Initial evaluation of cystatin C measurement by particle-enhanced immunonephelometry on the Behring nephelometer systems (BNA, BN II). Clin Chem 43: 1016–1022, 1997
13. Stevens L, Coresh J, Schmid CH, Feldman HI, Froissart M, Kusek J, Rossert J, Van Lente F, Bruce RD, Zhang Y, Greene T, Levey AS: Estimating GFR using serum cystatin C alone and in combination with serum creatinine: A pooled analysis of 3418 individuals with CKD. Am J Kidney Dis 351: 395–406, 2008
14. Weiner DE, Tighiouart H, Amin MG, Stark PC, MacLeod B, Griffith JL, Salem DN, Levey AS, Sarnak MJ: Chronic kidney disease as a risk factor for cardiovascular disease and all-cause mortality: A pooled analysis of community-based studies. J Am Soc Nephrol 15: 1307–1315, 2004
15. Shlipak MG, Fried LF, Cushman M, Manolio TA, Peterson D, Stehman-Breen C, Bleyer A, Newman A, Siscovick D, Psaty B: Cardiovascular mortality risk in chronic kidney disease: Comparison of traditional and novel risk factors. JAMA 293: 1737–1745, 2005
16. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D: A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 130: 461–470, 1999
17. Shlipak M, Katz R, Kestenbaum B, Fried L, Newman A, Siscovick D, Stevens L, Sarnak M: Rate of kidney function decline in older adults: A comparison using creatinine and cystatin C. Am J Nephrol 30: 171–178, 2009
18. Lindeman RD, Tobin J, Shock NW: Longitudinal studies on the rate of decline in renal function with age. J Am Geriatr Soc 33: 278–285, 1985
19. Ives DG, Fitzpatrick AL, Bild DE, Psaty BM, Kuller LH, Crowley PM, Cruise RG, Theroux S: Surveillance and ascertainment of cardiovascular events. The Cardiovascular Health Study. Ann Epidemiol 5: 278–285, 1995
20. Price TR, Psaty B, O'Leary D, Burke G, Gardin J: Assessment of cerebrovascular disease in the Cardiovascular Health Study. Ann Epidemiol 3: 504–507, 1993
21. Rosner B: The analysis of longitudinal data in epidemiologic studies. J Chronic Dis 32: 163–173, 1979
22. Cain K, Kronmal R, Kosinski A: Analyzing the relationship between change in a risk factor and risk of disease. Stat Med 11: 783–797, 1992