| |
African-Americans at Risk for Kidney Disease
|
| |
| |
End-stage renal disease and chronic kidney disease in a cohort of African-American HIV-infected and at-risk HIV-seronegative participants followed between 1988 and 2004
AIDS:Volume 21(18)30 November 2007p 2435-2443
Lucas, Gregory Ma,d; Mehta, Shruti Hd; Atta, Mohamed Gb; Kirk, Gregory Da,d; Galai, Noyad,f; Vlahov, Davidd,e; Moore, Richard Dc,d
From the aDepartment of Medicine, Divisions of Infectious Diseases, USA
bNephrology, USA
cGeneral Internal Medicine, the School of Medicine, USA
dDepartment of Epidemiology, Bloomberg School of Public Health, John Hopkins University, Baltimore, Maryland, USA
eCenter for Urban Epidemiologic Studies, New York Academy of Medicine, New York, USA
fDepartment of Statistics, University of Haifa, Haifa, Israel.
"In conclusion, we found that the incidence of ESRD requiring RRT in HIV-infected African-Americans approaches 1% annually, approximately 10-fold higher than age-comparable rates among African-Americans in the general population. The incidence of RRT (renal replacement therapy) has increased nonsignificantly in the HAART era compared with the pre-HAART era. An increased risk of RRT was strongly associated with advanced HIV disease. Hepatitis C seropositivity was associated with a higher rate of RRT in HIV-seronegative participants. Hepatitis C seropositivity and injection drug use, however, were not risk factors for RRT in HIV-infected subjects. In subgroup analysis, we found that CKD incidence decreased significantly, while CKD period prevalence increased significantly in the HAART era compared with the pre-HAART era, suggesting that the increased prevalence of CKD and the stable incidence of RRT are largely a consequence of longer survival with CKD in the HAART era."
Abstract
Background: HIV-infected African-Americans are at increased risk of end-stage renal disease requiring renal replacement therapy (RRT).
Objectives: To compare the incidence of RRT in HIV-infected and HIV-seronegative African-Americans and describe temporal trends in RRT and chronic kidney disease (CKD) in HIV infection.
Design: Cohort study in Baltimore including 4509 HIV-infected and 1746 HIV-seronegative African-Americans.
Methods: Incident RRT was defined by matching participant identifiers with the US Renal Data System; CKD was defined as an estimated glomerular filtration rate < 60 ml/min per 1.73m2 for ≥ 3 months. Standardized incidence ratios (SIR) and 95% confidence intervals (CI) were calculated by indirect adjustment. Risk factors for RRT were assessed by person-time methods and Poisson regression.
Results: RRT was initiated in 24 HIV-seronegative subjects over 13 415 person-years of follow-up (SIR, 2.3; 95% CI, 1.5-3.4), in 51 HIV-infected participants without AIDS over 10 780 person-years (SIR, 6.9; 95% CI, 5.1-9.0), and in 125 participants with AIDS over 9833 person-years. SIR, 16.1; 95% CI, 13.4-19.2). In HIV-infected African-Americans, RRT incidences were 5.8 and 9.7/1000 person-years in the pre-HAART and HAART eras, respectively (adjusted rate ratio 1.2; 95% CI, 0.8-1.9). In supplementary analyses, CKD incidence declined significantly in the HAART era compared with pre-HAART, but the CKD period prevalence increased.
Conclusions: Nearly 1% of HIV-infected African-Americans initiated RRT annually, a rate that was similar in the HAART and pre-HAART eras. While new cases of CKD decreased, the prevalence of CKD increased in the HAART era, primarily because survival in those with HIV-associated CKD has improved.
Introduction
The risk of end-stage renal disease (ESRD) requiring renal replacement therapy (RRT) is approximately 4-fold higher in African-Americans than in whites in the US population [1,2]. African-Americans are also disproportionately affected by the HIV epidemic [3], and the histopathological entity of HIV-associated nephropathy occurs almost exclusively in individuals of African descent [4,5]. For these reasons, the burden of HIV-related ESRD in the United States is borne overwhelmingly by African-Americans. For example, African-American men ages 25 to 44 years constitute just 2% of the US population, but they account for 40% of HIV-infected patients receiving RRT [6].
HAART has transformed HIV infection from an almost universally fatal illness into a manageable chronic disease [7,8]. The effect of HAART on the incidence of chronic kidney disease (CKD) and ESRD in HIV-infected African-Americans has not been defined but may be complex. Prior studies suggest that HAART mitigates the clinical course of HIV-associated nephropathy [9-11] and may decrease its occurrence [12]. However, causes other than HIV-associated nephropathy generally account for at least half of the cases of HIV-associated CKD evaluated in clinical practice [4,5]. Antiretroviral medications may have direct kidney toxicity [13], and HAART has been associated with increased rates of diabetes [14] and hypertension [15], which are major risk factors for CKD. Hepatitis C [16] and injection drug use [17,18], which are common in HIV-infected individuals, also may have adverse effects on kidney function. The goals of this study are to quantify the incidence of RRT in HIV-infected and HIV-seronegative African-American participants followed in two long-running cohort studies, assess temporal trends and other risk factors for RRT among HIV-infected African-American participants, and assess how changes in the epidemiology of CKD relate to RRT incidence.
Methods
Cohorts and participants
The Johns Hopkins HIV Cohort (JHHC) and the AIDS Link to the IntraVenous Experience (ALIVE) Study are longitudinal cohorts based in Baltimore, Maryland. The JHHC includes data from over 5000 HIV-infected participants who received primary care in the Johns Hopkins HIV Clinic [19]. Participants have been enrolled on a continuing basis from 1990 onward. Information from clinical records was reviewed and abstracted by trained technicians onto structured data collection forms, then entered into a relational database. The clinic medical records, the main hospital medical record, and various institutional computerized databases (e.g., laboratory, radiology, pathology, and hospital discharge summaries) were abstracted. Comprehensive demographic, clinical, laboratory, pharmaceutical, and psychosocial data were collected at times corresponding to enrollment in the clinic and at 6-month intervals thereafter. Clinical diagnoses of AIDS-associated opportunistic conditions, other selected conditions, deaths, and dates for these events were routinely abstracted. Participants provided written informed consent and the JHHC is approved by the Johns Hopkins Medicine Institutional Review Board.
The ALIVE study is a community-based cohort of the natural history of injection drug use and HIV infection that enrolled 2946 participants between 1988 and 1989. An additional 735 participants were enrolled during recruitment periods in 1994, 1998, and 2000. All participants were 18 years of age and older, free of clinical AIDS at entry into the study, and acknowledged injection drug use within the previous 11 years. HIV-seropositive and HIV-seronegative participants were followed at 6-month intervals. Clinical, behavioral, and laboratory data were obtained according to standardized protocols and methods [20]. All AIDS diagnoses and deaths were confirmed by medical record and death certificate review and a physician-led endpoints committee. The ALIVE study was approved by the Johns Hopkins University Committee on Human Research and all subjects provided written informed consent.
Definitions
Incident RRT cases were identified by matching participant identifiers with the US Renal Data System, a publicly funded national database [21]. The date of RRT was defined as the first ESRD service date. Participants who initiated RRT prior to enrollment in the cohort were excluded, as were events that occurred ≥ 2 years after last follow-up in the cohort. Race was self-identified for each participant. In the JHHC, history of injection drug use was assessed by clinicians at enrollment. The ALIVE study enrolled only participants with a history of injection drug use. AIDS was defined according to the 1993 Centers for Disease Control and Prevention classification guidelines, which included development of specific opportunistic conditions or a CD4 cell count < 200 cells/ƒÊl [22]. The pre-HAART and HAART eras were defined as the calendar periods before and after January 1, 1996, respectively. Glomerular filtration rate was estimated using the abbreviated Modification of Diet in Renal Disease Equation [23]. CKD was defined as a glomerular filtration rate < 60 ml/min per 1.73m2 for ≥ 3 months [23].
Statistical analysis
Follow-up time for participants spanned from cohort enrollment (in either JHHC or ALIVE) to the earliest of RRT, death, last contact, or September 30, 2004. Participants with follow-up time in both cohorts were counted in the JHHC for analytical purposes. The incidence of RRT was calculated as the number of events per 1000 person-years of follow-up with 95% confidence intervals (CI). Standardized incidence ratios (SIR) and 95% CI were calculated by indirect adjustment, comparing the observed number of RRT cases with the number of cases that would be expected based on published age- and race-stratified rates in the United States [1]. Analyses were restricted to African-American participants, because the risk of ESRD was predominantly borne by this group. Analyses were conducted using Stata version 8.0 statistical software (StataCorp, College Station, Texas, USA).
Univariate and multivariate Poisson regression models were used to assess factors associated with RRT. Fixed covariates included sex, race, history of injection drug use, and hepatitis C antibody status (seropositive, seronegative, or unknown/indeterminate). Time-dependent covariates included age, calendar period (or antiretroviral treatment era), HIV serostatus, and, in HIV-infected participants, the presence or absence of AIDS. Individuals may have contributed observation time to more than one category of time-dependent covariates, if, for example, they moved into a higher age category, HIV seroconverted, or developed AIDS during follow-up. Two-tailed P values < 0.05 were considered be statistically significant. Factors that were associated with RRT in univariate analysis (P < 0.1) were considered for inclusion in multivariate models. Potential interactions between all covariates included in multivariate models and RRT were assessed by combining terms in regression models.
Additional analyses were conducted in the subset of African-American HIV-infected participants that was followed in the JHHC. Longitudinal serum creatinine values were available in this group, making the assessment of incident and prevalent CKD possible. The period prevalence of CKD was defined as the number of subjects with CKD in a defined calendar period divided by the number of subjects with follow-up time in the period. The incidences of RRT, CKD, CKD-death prior to RRT and the composite outcome of RRT or CKD-death prior to RRT were calculated as described above. CKD-death prior to RRT was defined as death occurring in an individual with CKD (not necessarily as a result of CKD) who had not received RRT. This outcome was designed to capture individuals who died with CKD before RRT was required or who may have been determined to be poor candidates for RRT because of underlying HIV disease severity. For each outcome, event rates in the pre-HAART and HAART eras were compared by unadjusted incidence rate ratios (IRR) and IRR that were adjusted for age and AIDS status using Poisson regression.
Results
RRT was initiated in 205 of 7660 participants in the complete cohort: 200 of 6255 African-Americans (3.2%) and five in other racial/ethnic groups (numbers in specific racial/ethnic groups suppressed to protect confidentiality as required by US Renal Data System). The relative risk of RRT for African-American compared with white participants was 31 (95% CI, 4-224). Baseline characteristics of 6225 African-American participants are shown in Table 1. Data from 368 subjects who participated in both cohorts are included in the JHHC. Compared with the JHHC, ALIVE participants were younger at enrollment, enrolled predominantly in 1988-1989 (original recruiting period for this cohort), had longer median follow-up, and were more likely to be positive for hepatitis C antibody. In accord with the inclusion criteria of ALIVE, 99% of participants had a history of injection drug use. All HIV-seronegative subjects were followed in ALIVE. Among HIV-positive participants, HIV disease was more advanced at baseline in participants in the JHHC than in ALIVE, in agreement with the ALIVE enrollment criterion that HIV-infected participants be free of AIDS at enrollment.
African-American participants contributed 34 028 person-years of follow-up, with an RRT incidence of 5.9/1000 person-years (95% CI, 5.1-6.8). Compared with expected rates among age-matched African-Americans in the US population [1], the RRT risk was 2.3-fold higher (95% CI, 1.5-3.4) for HIV-seronegative cohort participants, 6.7-fold higher (95% CI, 5.0-8.9) for HIV-infected participants without AIDS, and 16.2-fold higher (95% CI, 13.3-19.3) for AIDS-defined participants (Table 2). Among HIV-seronegative African-Americans, RRT was initiated in 22 of 1021 hepatitis C-seropositive subjects and in none of 160 hepatitis C-seronegative subjects (P = 0.07, log rank test). Adjusted RRT incidence rates in different calendar periods, stratified by HIV serostatus, are shown in the Figure 1.
Factors associated with RRT in HIV-infected African-Americans are shown in Table 3. In univariate analysis, increasing age, AIDS, and follow-up in the HAART era were statistically significantly associated with higher RRT incidence. Women had a slightly lower risk of RRT than men (P = 0.09). Cohort (JHHC or ALIVE), history of injection drug use, hepatitis C antibody status, and attained education were not associated with RRT. In multivariate analysis, only AIDS status remained statistically significantly associated with RRT (IRR 2.7; 95% CI, 1.9-4.0). No statistically significant interactions were detected.
Among HIV-infected African-American participants, characteristics were compared for those who received RRT in the pre-HAART (36) and HAART (140) eras. The proportions of women and injection drug users were similar in subjects who received RRT in the two treatment eras (data not shown). However, participants who initiated RRT in the pre-HAART era were significantly younger than those in the HAART era: median age 36 years [interquartile range (IQR), 33-41) and 42 years (IQR 37-48), respectively (P < 0.001). Compared with subjects who received RRT in the pre-HAART era, those who received RRT in the HAART era had lower current CD4 cell counts [median 303 cells/ƒÊl (IQR, 64-460) versus 162 cells/ƒÊl (IQR, 50-311); P = 0.03], lower nadir CD4 cell counts [median 212 cells/ƒÊl (IQR, 64-335) versus 99 cells/ƒÊl (IQR, 19-193); P = 0.002], and were more likely to be defined as having AIDS (47% versus 77%; P = 0.001).
Because the primary enrollment for the ALIVE study was in 1988-1989, this cohort's participants accounted for a larger percentage of those who received RRT in the pre-HAART era (47%) than in the HAART era (14%) (P < 0.001). As shown in Table 1, ALIVE participants were younger and had less-advanced HIV disease than JHHC participants at enrollment. However, the observations that RRT recipients in the pre-HAART era were younger and had less-advanced HIV disease than those in the HAART era were not affected when analyses were stratified by cohort. For example, among participants who initiated RRT in ALIVE, the median ages were 38 and 42 years in the pre-HAART and HAART eras, respectively (P = 0.007). Similarly, among participants who initiated RRT in JHHC, the median ages were 35 and 42 years in the pre-HAART and HAART eras, respectively (P = 0.003). Similar trends were observed in cohort-stratified analyses of current CD4 cell count, nadir CD4 cell count, and AIDS status (data not shown).
To explore the association between CKD epidemiology and RRT incidence in the two HIV treatment eras, additional analyses were conducted in the subset of HIV-infected African-American participants followed in the JHHC (3735) for whom longitudinal serum creatinine data were available (Table 4). Adjusting for age and AIDS status, the period prevalence CKD was 37% higher in the HAART era than in the pre-HAART era (P = 0.021). In contrast, the adjusted incidence of CKD was 36% lower (P = 0.002) in the HAART era than the pre-HAART era. The incidence CKD-death prior to RRT was 46% lower in the HAART era than in the pre-HAART era (P = 0.006). Consistent with data from the composite cohort, the adjusted incidence of RRT was 46% higher in the HAART era than the pre-HAART era in the JHHC subgroup (not statistically significant). The risk of the composite outcome, RRT or CKD-death prior to RRT, was similar in the pre-HAART and HAART eras (P = 0.56).
Discussion
In this 15-year cohort study, the incidence of RRT in HIV-infected African-Americans approached 1% per year, a risk approximately 10-fold higher than in age-matched African-Americans in the general population. Among HIV-infected participants, the risk of RRT was approximately 30-fold higher in African-Americans than in whites, as has been reported previously [6]. Our study is the first to provide data on the relative risk of RRT in the HAART era compared with the pre-HAART era in a well-characterized cohort. We found that the incidence of RRT in HIV-infected African-Americans has not changed significantly with the introduction and widespread implementation of HAART. The continued high risk of ESRD in this group during the HAART era is of particular public health relevance because African-Americans are disproportionately affected by the HIV epidemic. In 2005, African-Americans, who constitute approximately 12% of the general population, accounted for 49% of newly diagnosed HIV infections [3].
Our supplementary analyses of CKD in the JHHC shed light on the dynamic epidemiology underlying the stable incidence of RRT that we observed during follow-up. First, we found that age- and AIDS-adjusted incidence of CKD was approximately 40% lower in the HAART era compared with the pre-HAART era. This suggests that the availability of HAART, or potentially other temporal factors, have reduced the risk of CKD. Conversely, the period prevalence of CKD was approximately 40% higher in the later era. As prevalence is determined by both incidence and duration of disease, it is likely that longer survival among individuals with CKD in the HAART era accounts for this finding. Second, we found that the incidence of the composite outcome, RRT or CKD-death prior to RRT, which make up the competing endpoints of CKD, was similar in the pre-HAART and HAART eras, at approximately 1.5% per year. The individual components of this composite had opposing trends, with RRT incidence increasing and CKD-death prior to RRT decreasing between the pre-HAART and HAART eras.
We found substantial differences in age and HIV disease severity among participants who initiated RRT in the two HIV treatment eras. Subjects who initiated RRT in the pre-HAART era were younger, had higher current CD4 cell counts, higher nadir CD4 cell counts, and were less likely to be defined with AIDS than subjects who initiated RRT in the HAART era. Three phenomena may account for these differences. First, individuals with CKD may have been more likely to survive until RRT was required in the HAART era than in the pre-HAART era because HIV-related competing risks were lower. Second, nephrologists may have been more willing to initiate RRT in HIV-infected patients with poor prognostic factors in the HAART era than in the pre-HAART era. A third possibility is that the age and HIV disease severity differences in RRT recipients in the two treatment eras reflect temporal changes in the underlying pathophysiology of renal disease in this population.
Prior studies have documented a high prevalence and incidence of renal abnormalities in HIV-infected subjects, including proteinuria and hypercreatinemia [24-28]. In longitudinal studies, renal abnormalities have been found to be independent predictors of opportunistic conditions [29] and all-cause mortality [29,30] in HIV-infected individuals. In an analysis of data from the US Renal Data System, Eggers and Kimmel [6] found that the number of new HIV-associated cases entering the RRT registry annually remained stable between 1995 and 2000, a period corresponding to the early HAART era. However, the prevalence of HIV in the ESRD program increased 2.6-fold over this same time period because 1-year survival among HIV-infected individuals receiving RRT increased by approximately 40%. Schwartz and colleagues modeled data from the US Renal Data System and predicted continuing rise in the prevalence of HIV-associated ESRD.
Renal diseases associated with hepatitis C infection are well described [16]. Injection drug use, which is a major risk factor for hepatitis C, has also been linked to kidney damage [18,32]. We were unable to assess the association between injection drug use and RRT in HIV-seronegative individuals, because all HIV-seronegative participants in our analysis were from the ALIVE study, in which injection drug use was an enrollment criterion. We did find that hepatitis C-seropositive status was associated with an increased risk of RRT in HIV-seronegative African-Americans (borderline statistical significance); although this analysis was limited by the small number of participants seronegative for both HIV and hepatitis C. The HIV-seronegative group had a risk of RRT that was 2.3-fold higher than expected from age-matched rates in the general African-American population. It is possible that hepatitis C, injection drug use, or other associated factors accounted for the higher than expected incidence of RRT in these participants. Interestingly, we found no association between either hepatitis C antibody status or history of injection drug use and RRT incidence among HIV-infected African-Americans, despite adequate numbers of subjects for comparison.
Our study has limitations. First, while conducted in two well-characterized cohorts, the study population resided in a single metropolitan area and the results may not generalize to HIV-infected populations in other areas, particularly those where the racial makeup is different from ours. Our findings should not be extrapolated to HIV-infected African-Americans in the United States as a whole. Future studies should combine cohorts from geographically distinct areas. Second, cohort studies are potentially susceptible to selection or follow-up biases, where either inclusion in the cohort or loss-to-follow-up is not independent of the outcome of interest. The JHHC is a treatment cohort and may have been susceptible to enrolling participants at higher risk of RRT; however, individuals would be unlikely to drop out or become lost to follow-up as they became ill. In contrast, the community-based recruitment design of ALIVE would be unlikely to enroll individuals at higher risk of RRT, although drop-out or loss to follow-up might be more likely as individuals became ill. Reassuringly, we observed a similar trend for higher RRT risk in the later calendar period compared with the earlier period in both the JHHC and the ALIVE study (data not shown), arguing against cohort design as a source of bias.
This study has important implications for resource planning and future research. Although HIV-infected individuals currently account for only 1-2% of the prevalent ESRD population in the United States [6,33], the disproportionately high rates of HIV infection among African-Americans and the increased longevity and comorbidity of those living with HIV infection suggest that substantial increases will be needed for CKD and RRT resources for this population in the future [31]. Prospective, longitudinal studies are needed to characterize the precursors, natural history, and underlying etiologies of CKD in HIV-infected African-Americans. While clinical practice guidelines have been formulated for screening and management of CKD in HIV-infected individuals [34], clinical trials that assess specific screening strategies and early interventions are needed.
In conclusion, we found that the incidence of ESRD requiring RRT in HIV-infected African-Americans approaches 1% annually, approximately 10-fold higher than age-comparable rates among African-Americans in the general population. The incidence of RRT has increased nonsignificantly in the HAART era compared with the pre-HAART era. An increased risk of RRT was strongly associated with advanced HIV disease. Hepatitis C seropositivity was associated with a higher rate of RRT in HIV-seronegative participants. Hepatitis C seropositivity and injection drug use, however, were not risk factors for RRT in HIV-infected subjects. In subgroup analysis, we found that CKD incidence decreased significantly, while CKD period prevalence increased significantly in the HAART era compared with the pre-HAART era, suggesting that the increased prevalence of CKD and the stable incidence of RRT are largely a consequence of longer survival with CKD in the HAART era.
|
|
| |
| |
|
|
|
