|Home | About | Journals | Submit | Contact Us | Français|
Cardiovascular disease (CVD) is now a leading cause of death in HIV-infected persons; however, risk markers for CVD are ill-defined in this population. We examined the association between longitudinal measures of kidney function and albuminuria with risk of atherosclerotic CVD and heart failure in a contemporary cohort of HIV-infected individuals.
We followed a national sample of 17,264 HIV-infected persons receiving care in the Veterans Health Administration for: (1) incident CVD, defined as coronary, cerebrovascular, or peripheral arterial disease; and (2) incident heart failure. Rates of CVD and heart failure were at least 6-fold greater in the highest risk patients with an estimated glomerular filtration rate (eGFR) <30 mL/min/1.73m2 and albuminuria ≥300 mg/dL versus those with no evidence of kidney disease (eGFR ≥60 mL/min/1.73m2 and no albuminuria). After multivariable adjustment, eGFR levels 45–59, 30–44, and <30 mL/min/1.73m2 were associated with hazard ratios (HR) for incident CVD of 1.46 (95% confidence interval 1.15–1.86), 2.03 (1.47–2.82), and 1.99 (1.46–2.70), compared with eGFR ≥60 mL/min/1.73m2. Similarly, albuminuria levels 30, 100, and ≥300 mg/dL had HR’s for CVD of 1.28 (1.09–1.51), 1.48 (1.15–1.90), and 1.71 (1.30–2.27), compared with absent albuminuria. The associations between eGFR and albuminuria with heart failure were larger in magnitude and followed the same trends.
In this national sample of HIV-infected persons, eGFR and albuminuria levels were strongly associated with risk of CVD and heart failure. Kidney function and albuminuria provide complementary prognostic information which may aid CVD risk stratification in HIV-infected persons.
With the widespread use of antiretroviral therapy (ART), non-AIDS-related conditions such as cardiovascular disease (CVD) and chronic kidney disease (CKD) have emerged as important contributors to mortality in the HIV-infected population. The majority of studies indicate that HIV infection is a risk factor for premature atherosclerotic vascular disease, leading to increased rates of CVD events compared with HIV-uninfected controls.1–5 Cardiovascular disease is now the third leading cause of death among HIV-infected persons in the US.6, 7 Similarly, kidney disease is an increasingly frequent complication of HIV infection.8, 9 Between 1995 and 1999, the prevalence of end stage renal disease (ESRD) caused by HIV-associated nephropathy doubled and the proportion of deaths attributed to kidney disease increased from 6% to 9% among HIV-infected patients nationally.10, 11
In recognition of the need to improve the quality of CVD care in HIV infection, the American Heart Association recently convened a panel of experts to identify urgent clinical issues and research challenges facing this population.12, 13 During this conference, the need to identify characteristics that will aid in CVD risk stratification was designated as a research priority. Although CKD is an established risk factor for CVD in the general population, to our knowledge, this association has not been studied in HIV-infected persons.14, 15 Previously, we and others have found that CKD is a risk marker for death in HIV-infected individuals, a relationship thought to be mediated by CVD in the HIV-uninfected population.16–19 Therefore, we conducted this study to describe the association between markers of kidney disease - estimated glomerular filtration rate (eGFR) and albuminuria - with cardiovascular events in HIV-infected persons. To address this question, we use a national registry of HIV-infected persons receiving care in the Veterans Health Administration, which is the largest provider of HIV care in the United States. We hypothesized that reduced kidney function and albuminuria would have independent and complementary associations with risk of CVD and heart failure events in HIV-infected persons.
The analytic dataset was assembled from a variety of component data sources from the Department of Veterans Affairs (VA) and Centers for Medicare and Medicaid Services. The VA monitors health care utilization for all HIV-infected veterans using the VA HIV Clinical Case Registry (CCR), which contains all demographic, clinical, laboratory, pharmacy, utilization, and death information entered in the VA electronic medical record.20 This data source was linked to the VA National Patient Care Database, Medicare claims, and the VA Beneficiary Identification and Records Locator Subsystem (BIRLS) Death File to supplement demographic, clinical, and vital status data, and to capture hospitalizations for CVD and heart failure events outside of the VA system.21–25 In addition, we used the United States Renal Data System, a comprehensive national ESRD registry, to ascertain prevalent ESRD cases at the time of cohort entry and incident cases of ESRD during follow-up.26
We studied the association between kidney disease and CVD in a national sample of HIV-infected US veterans. Geographically, the Veterans Health Administration is national in scope and offers low-cost, comprehensive clinical services to US veterans.27 We identified 34,733 HIV-infected persons receiving ambulatory care in the VA since 1999. Among these individuals, we excluded 15,415 patients who did not have both outpatient serum creatinine and albuminuria measurements available, and 2,054 with prevalent CVD or heart failure, leaving 17,264 persons who were included in our analysis. Patients entered the cohort at the time of their first creatinine measurement.
The primary outcome was the time from study entry to incident atherosclerotic CVD, defined as hospitalization for coronary, cerebrovascular, or peripheral arterial disease. The secondary outcome was time from study entry to hospitalization for heart failure. We also analyzed each specific CVD event separately. Outcomes were defined by primary discharge diagnoses and procedural codes in VA and non-VA data sources using validated algorithms described previously (Appendix A).18
The primary predictors were measures of kidney function and albuminuria updated over time using the last value carried forward method.18, 28 This method has been used in prior studies of CVD in the general population, and is considered to more accurately characterize kidney function and albuminuria, because it avoids misclassification due to reliance on single measurement at a fixed point in time.18, 28–30 Kidney function was defined as eGFR, calculated using the abbreviated Modification of Diet in Renal Disease (MDRD) formula based on age, sex, race, and serum creatinine.31 Patients were categorized according to eGFR as follows: ≥60 (normal or mildly reduced kidney function), 45–59 ml/min/1.73m2 and 30–44 ml/min/1.73m2 (moderately reduced kidney function), and <30 ml/min/1.73m2 (severely reduced kidney function or dialysis dependent kidney failure). Albuminuria was defined by urine dipstick measurements and categorized as 0, 30, 100, or ≥300 mg/dL.
Demographic characteristics of the subjects were determined from VA or Medicare health-plan databases.32 We identified coexisting illnesses using a combination of hospitalization discharge diagnoses, ambulatory diagnoses, physician problem lists, procedures, laboratory results, and medication prescriptions (Appendix A). With this information, we applied validated algorithms to define the following conditions: diabetes, hypertension, hyperlipidemia, hepatitis C or B virus co-infection, chronic obstructive lung disease, liver disease, and smoking.18, 33, 34 Since 1998, the VA has achieved 95% smoking and tobacco use screening rates because of mandatory screening guidelines tied to performance measures.35, 36 Clinical information such as blood pressure, body mass index, CD4 T-cell counts, HIV RNA level, and random plasma glucose levels, as well as ART exposure, were also included in statistical models. Baseline characteristics including comorbid illnesses, medication exposures, and clinical variables were defined in the 2 year period prior to entering the study, and then all covariates were updated over time during the period of follow-up.
We compared incidence rates of CVD and heart failure stratified by level of kidney function and albuminuria. We obtained estimates of relative risk using a time to event analysis with Cox proportional hazards regression models, including both eGFR and albuminuria category, and adjusted for demographic characteristics, CVD risk factors [diabetes, hypertension, hyperlipidemia, smoking, blood pressure, body mass index, random glucose] and HIV-specific factors [CD4 count, viral load, AIDS, ART, duration of HIV infection]. We built parsimonious models using a backward stepwise procedure with P<0.05 used as the criterion for inclusion. Patients were censored at the time of death or the last day of follow-up through December 31, 2007. A missing indicator variable was included to retain observations for the less than 5% of data that were missing in our final models. Assumptions of the Cox regression models were checked by comparing plots of log (−log(survival)) versus log of survival time and the Schoenfeld test.
A number of supplemental analyses were conducted to test the robustness of the primary results. To determine whether medication exposure to abacavir, tenofovir, or protease inhibitors influenced the association between kidney disease and CVD, we additionally adjusted for receipt of these medications. We also tested for interactions with antiretroviral treatment status, CD4 count, and viral load, to assess whether these factors modified the association between kidney disease and CVD. Finally, we compared risk markers for kidney disease and CVD, and CVD rates in the HIV-infected population excluded from our analysis, including those with eGFR, but without albuminuria measured, and those with neither kidney function nor albuminuria measured. Analyses were conducted using Stata version 10.1 (Stata Corp, College Station, TX). This study was approved by the Committee on Human Research at the San Francisco VA Medical Center and the VA Public Health Strategic Health Care Group on July 18, 2008.
At the time of study entry (Table 1), 7% of persons had reduced kidney function (eGFR<60 mL/min/1.73m2). Individuals with impaired kidney function at baseline were more likely to be older and have traditional risk factors for kidney disease such as hypertension, diabetes, and dyslipidemia. Whereas HIV viral load did not differ between those with and without reduced kidney function, CD4 counts and receipt of ART were lower in those with eGFR<60 mL/min/1.73m2. In addition, compared to included patients with no evidence of CKD, excluded individuals without prevalent CVD, who did not have eGFR or albuminuria measured, had a lower prevalence of comorbidities, abnormal clinical and HIV-related characteristics, and lower rates of CVD - suggesting that these individuals were not screened for CKD because they were at low risk for this condition.
We analyzed a total of 219,913 outpatient eGFR measurements [9 (inter-quartile range [IQR] 3, 19) per subject] and 100,931 outpatient albuminuria measurements [4 (IQR 2, 8) per subject] over a median of 7 (IQR 3, 8) years. There were a total of 370 heart failure and 833 atherosclerotic CVD outcomes, comprised of 503 coronary, 219 cerebrovascular, and 118 peripheral arterial disease events, over 119,051 person-years of observation.
We observed a stepwise increase in the incidence of CVD with increasing severity of time-updated albuminuria and reduced eGFR (Figure 1A). There was approximately a six-fold difference in rates between those with no signs of kidney disease (eGFR ≥60 mL/min/1.73m2 and no albuminuria) and the highest risk category (eGFR <30 mL/min/1.73m2 and albuminuria ≥100 mg/dL). We observed a similar pattern for the incidence of heart failure by the presence of albuminuria or reduced eGFR (Figure 1B). Heart failure events were rare in those with eGFR ≥60 mL/min/1.73m2 and no albuminuria (1.6 events per 1000 person-years), and increased 30-fold to 51 events per 1000 person-years in the highest risk group.
In the multivariable models - adjusted for demographic characteristics and time-updated hypertension, diabetes, chronic obstructive lung disease, dyslipidemia, smoking, CD4 count, viral load, ART and albuminuria - level of eGFR was independently associated with increased risk of the composite outcome of atherosclerotic CVD (Table 2). Compared to those with normal or mildly reduced kidney function (eGFR <60 mL/min/1.73m2), those with an eGFR 45–59 mL/min/1.73m2 had a 46% increase in the adjusted risk of a vascular event, while those with an eGFR 30–44 mL/min/1.73m2 or <30 mL/min/1.73m2 had approximately a 100% increase in risk. The magnitude and strength of the associations between level of eGFR and heart failure were stronger than for CVD, with a greater than two-fold adjusted risk for eGFR 45–59 mL/min/1.73m2 or 30–44 mL/min/1.73m2, and a more than three-fold adjusted risk for <30 mL/min/1.73m2. For individual CVD outcomes, associations appeared somewhat stronger for coronary disease than cerebrovascular and peripheral arterial disease, but confidence intervals were too wide to detect heterogeneity across these outcomes.
As with eGFR categories, albuminuria levels were independently associated with both CVD events and heart failure, but with much stronger associations with heart failure (Table 3). Among the individual CVD outcomes, the association with albuminuria appeared strongest with peripheral arterial disease.
In order to determine whether reduced eGFR and albuminuria provided complementary information about CVD risk, we categorized individuals into mutually exclusive groups based on eGFR<60 mL/min/1.73m2 and/or albuminuria ≥30 mg/dL (Table 4). In this analysis, both reduced eGFR and albuminuria were independently associated with CVD and heart failure, and the effects appeared to be complementary. Increased risks were observed for all outcomes, but were particularly strong for heart failure and peripheral arterial disease.
In supplemental analyses, hazard ratios for eGFR and albuminuria changed minimally with additional adjustment for abacavir, tenofovir, or protease inhibitor exposure (data not shown). There was no evidence of effect modification by ART, CD4 count, and HIV viral load (p >0.1). The fully-adjusted associations between eGFR and CVD were larger or similar in magnitude when all individuals with eGFR measurements, irrespective of albuminuria measurement, were analyzed.
In this national sample of HIV-infected persons receiving care in the Veterans Health Administration, we observed a graded, independent association between level of kidney function and albuminuria with atherosclerotic CVD and heart failure. Importantly, eGFR and albuminuria level provided complementary prognostic information, with severity of kidney disease correlating with an incremental level of CVD and heart failure risk. To our knowledge, this is the first study to describe impaired kidney function and albuminuria as risk markers for adverse cardiovascular events in an HIV-infected population. Our results are clinically relevant, since they may help providers to identify HIV-infected persons at high risk for CVD events, and they provide rationale for future intervention studies aimed at reducing albuminuria and improving kidney function in this population.
Our study addresses several of the research objectives established by the American Heart Association to address the problem of CVD in people living with HIV/AIDS, including the need to describe the effect of metabolic complications on the risk of CVD and to identify characteristics to aid in risk stratification.13 Our findings identify reduced eGFR and albuminuria as independent risk markers for adverse cardiovascular outcomes and raise the possibility that kidney disease is a mediator of CVD in HIV-infected persons. These findings are supported by multiple studies conducted in the general population, which have found that kidney disease is associated with dyslipidemia, anemia, left ventricular hypertrophy, arterial stiffness, inflammation, and endothelial dysfunction.37 Our findings merit further study to investigate the specific mechanisms by which kidney disease may lead to CVD and heart failure in persons with HIV.
Regardless of the mechanisms underlying the association of kidney disease with CVD and heart failure, we consider several clinical applications for our findings that could be tested further in future clinical studies. First, measurement of kidney function and albuminuria could be useful for identifying HIV-infected individuals at elevated risk for adverse cardiovascular events. In contrast to many other novel risk factors for CVD, both creatinine-based eGFR and albuminuria are clinically available, and both are potentially modifiable markers of risk. Our findings may have immediate clinical application, since testing for kidney function and albuminuria are already recommended on a regular basis in the routine care of HIV-infected individuals to screen for kidney disease.12 Second, further study is needed to determine whether interventions directed at the treatment of kidney disease will reduce cardiovascular risk. For example, in the HIV-uninfected population, studies evaluating treatment with angiotensin converting enzyme inhibitors and angiotensin receptor blockers have suggested that cardiovascular risk is lowered in proportion to the extent of albuminuria reduction, independent of the effect of traditional cardiovascular risk factors.38–40 Prospective evaluation of these treatments in HIV-individuals with reduced eGFR or albuminuria may be warranted to determine their efficacy in this unique population. Third, our study may have implications for the “when to start” antiretroviral therapy question. Although some commonly used antiretroviral drugs can be nephrotoxic, several reports suggest that the overall effect of antiretroviral therapy is beneficial to the kidneys compared with the potential harm associated with uncontrolled viral replication. 41, 42 Therefore, early initiation of antiretroviral therapy to preserve kidney function may be protective against premature cardiovascular disease.
A major strength of this study is the national scope and large size of the study cohort which was representative of an HIV-infected population receiving relatively uniform health care. In addition, the associations of eGFR and albuminuria with adverse cardiovascular events were robust despite adjustment for a large number of clinical measures, including comorbid conditions, medications, vital signs, and laboratory measurements. We also expand on previous reports of CVD in the VA through the use of non-VA data sources to capture key outcomes.43, 44 The use of multiple measures of kidney function and albuminuria to characterize these variables more accurately is an additional strength.18, 30 Finally, we were able to evaluate both eGFR and albuminuria as CVD risk markers; joint analysis of these risk markers has rarely been performed in the general population and never in HIV-infected persons.
Our study also has several limitations. Estimating equations for kidney function based on serum creatinine have not been independently validated for the measurement of GFR in HIV-infected persons. Serum creatinine levels were not calibrated in a single laboratory in this national sample of patients. Kidney function, estimated from serum creatinine, may also miss substantial loss of kidney function that could be detected by newer filtration markers such as cystatin C;9 however, neither cystatin C nor directly measured GFR was available in the VA for clinical use. Urine dipstick testing is also considered a relatively insensitive method to detect microalbuminuria. However, both of these measures are widely used in routine clinical care and are specifically recommended by Infectious Disease Society of America guidelines for assessment of kidney disease in the HIV-infected population.12, 45 Second, our results may not be generalizeable to populations poorly represented in our analysis, such as women, non-veterans, or HIV-infected persons without access to medical care. Third, despite adjustment for a large number of clinical factors, residual confounding may still be present. In particular, individuals with CKD had a greater number of comorbid conditions and eGFR may be a general marker of age-related processes. Therefore, the magnitude of our observed risk estimates may be inflated. Similarly, although our study provides evidence that CKD may be a useful risk marker for incident CVD, it may also simply be a sign of disease severity (of traditional CVD risk factors such as hypertension). Studies devoted to elucidating causal mechanisms linking CKD and CVD in HIV-infected persons are needed.
In this large, national sample of HIV-infected persons, we observed an independent association between level of eGFR and albuminuria with atherosclerotic CVD and heart failure. With the increasing importance of non-AIDS related morbidity in the aging HIV-infected population, improving cardiovascular risk profiles have emerged as an urgent clinical objective among these patients. These results may aid clinicians in risk stratifying HIV-infected persons for future risk of cardiovascular events and provide rationale for interventional studies aimed at treatment of kidney disease or reduction of albuminuria.
Cardiovascular disease (CVD) is now a leading cause of death in HIV-infected persons. In recognition of the need to improve the quality of CVD care in HIV infection, the American Heart Association recently convened a panel of experts to identify urgent clinical issues and research challenges facing this population. During this conference, the need to identify characteristics that will aid in CVD risk stratification was designated as a research priority. We conducted this study to describe the association between markers of kidney disease - estimated glomerular filtration rate (eGFR) and albuminuria - with cardiovascular events in HIV-infected persons. To address this question, we use a national registry of HIV-infected persons receiving care in the Veterans Health Administration, which is the largest provider of HIV care in the United States. Both eGFR and albuminuria were strongly associated with the risk of incident CVD (defined as coronary, cerebrovascular, and peripheral arterial disease) and heart failure in a graded, independent manner. In addition, both markers provided complementary prognostic information. In contrast to many novel risk factors, eGFR and albuminuria are already recommended for CKD screening in all patients upon diagnosis of HIV and annually in high risk groups. Therefore, our findings may have immediate clinical application in aiding clinicians to risk stratify HIV-infected persons for future cardiovascular events and provide rationale for interventional studies aimed at treatment of kidney disease or reduction of albuminuria.
Funding Sources: This study was supported by the National Institutes of Health (K23DK080645-01A1, 1R03AG034871-01, K24AI069994, R01 DK066488-01), the National Center for Research Resources (KL2 RR024130), the American Heart Association Early Investigator Award, and the Veteran Affairs Public Health Strategic Health Care Group. These funding sources had no involvement in the design or execution of this study.
Disclosures: Paul A. Volberding serves on a data safety and monitoring board for Merck and TaiMed, an endpoints adjudication committee for Schering, and scientific advisory boards for Pfizer, Bristol-Myers Squibb (BMS), Gilead, and Schering.