| |
Immunosuppression, Hepatitis C Infection, and Acute Renal Failure in HIV-Infected Patients
|
| |
| |
[Epidemiology and Social Science]
JAIDS Journal of Acquired Immune Deficiency Syndromes: Volume 42(3) July 2006 pp 368-372
Franceschini, Nora MD, MPH*; Napravnik, Sonia PhD ; Finn, William F. MD ; Szczech, Lynda A. MD ; Eron, Joseph J. Jr MD
From the *Department of Epidemiology, School of Public Health, Division of Infectious Disease, School of Medicine and Center for AIDS Research, and Division of Nephrology and Hypertension, School of Medicine, University of North Carolina, Chapel Hill, NC; and Division of Nephrology, Duke University, Durham, NC.
Abstract
Background: Low CD4 cell counts predict HIV-related morbidity and mortality and may be associated with acute renal failure (ARF).
Objective: To estimate the effect of CD4 cell count on the incidence rate (IR) of ARF in ambulatory HIV-infected patients with access to highly active antiretroviral therapy.
Methods: Observational clinical cohort of HIV-infected patients recruited from a university-based infectious diseases clinic, between 2000 and 2002, and followed up until December 31, 2002. Poisson log-linear regression models were used to calculate ARF IRs, IR differences, and IR ratios.
Results: The mean age of the 705 study participants was 40 years, two thirds were male, and 61% were African American. Incidence rates of ARF were higher at lower CD4 cell counts and among patients who were coinfected with hepatitis C. Patients with hepatitis C coinfection who also had low CD4 cell counts had the highest adjusted IR of ARF.
Conclusion: Immunosuppression and hepatitis C virus coinfection are associated with increased IRs of ARF in ambulatory HIV-1-infected patients.
Additional results
Acute renal failure IRs increased with increasing HIV RNA levels with ARF IRs of 2.3, 4.2, and 11.0 events/100 person-years among patients with less than 2.6, 2.6 to 5.0, and more than 5.0 log10 HIV RNA copies/mL, respectively.
Patients during their first year of ART had the greatest ARF IR, in contrast to patients with more than 1 year of ART exposure (19.2 and 3.3 events/100 person-years, respectively), and patients on their first ART regimen had a higher ARF IR than those who were on at least their second regimen (5.7 and 3.8 events/100 person-years, respectively).
DISCUSSION
A low CD4 cell count is the most important predictor of HIV-1-related morbidity and mortality,11,12 and based on our results, a low CD4 cell count is also a strong predictor for experiencing an incident ARF event. We observed a dose-response pattern where decreasing levels of CD4 cell count were associated with increasing ARF IRs. A CD4 count lower than 200 cells/mL, independent of ART exposure and other factors, remained an important predictor of experiencing ARF.
Hepatitis C virus coinfection is common among HIV-1-infected patients and may be associated with increased morbidity and mortality.13,14 Among HIV/HCV-coinfected patients, an estimated 30% of ARF events are caused by underlying liver disease.8 In this study, we observed that increased ARF IRs due to HCV coinfection may be a more substantial concern among patients with relatively preserved immune function. These findings have important clinical implications for the care of HIV-1-infected patients, because ARF by itself is associated with increased morbidity and mortality, and liver failure has been associated with increased in-hospital death in patients with ARF in the intensive care unit (odds ratio 3.1; 95% CI, 1.9-4.9).15
Multiple factors may account for the increased ARF IRs among HIV/HCV-coinfected patients, although the precise mechanism of this observed association is unclear. Underlying chronic liver disease and its complications and susceptibility to infections due to IDU or other high-risk behaviors may predispose these patients to developing ARF. Some studies have suggested faster progression of HCV liver disease and an increased HCV viral replication in HIV-1-infected patients.16,17 It is also possible that HIV/HCV-coinfected patients have an increased susceptibility to infection due to impairment of the immune system, independent of CD4 cell counts. Because, in the United States, IDU is strongly correlated with HCV infection, it is possible that some of the effects of HCV are in fact due to IDU.
Acute renal failure IRs were higher in patients during their first year of ART use, suggesting that increased monitoring of patients for ARF is indicated during ART initiation. The reason for this association is unclear but may be related to other conditions and medications patients are exposed to around the time of ART initiation. Before the availability of HAART, morbidity and mortality in both HCV-infected and uninfected patients were predominantly HIV related,18 possibly obscuring effects of chronic conditions such as liver disease on ARF incidence. However, since the introduction of HAART, deaths and hospitalizations among HIV/HCV-coinfected patients are mainly due to HIV-unrelated conditions.18,19 In addition, previous studies assessing ARF in HIV-infected patients have only considered ARF identified through hospital records or biopsy databases,1-5 biasing these findings toward severe clinical conditions associated with acute tubular necrosis. Hence prerenal ARF was underrepresented in these studies because of the transitory nature of the event and the unlikelihood of obtaining a renal biopsy or dialysis need. However, prerenal ARF (including hepatorenal syndrome) accounts for the majority of the cases of ARF in patients with cirrhosis.20
Some HCV infection misclassification may have occurred because we relied exclusively on HCV antibody testing, as HCV viral loads or genotypes were generally unavailable. However, the positive predictive value of this test is estimated to be between 90% and 95% in a high-risk population such as ours,21 and any existing bias is likely toward underestimating an HCV effect because HCV antibody is a sensitive but not a specific marker of hepatic damage. In addition, as in any observational clinical study, we are relying on retrospectively collected data on variables such as history of IDU, hypertension, and diabetes, the accuracy of which depends on both medical provision and documentation. Because our outcome depended on creatinine measures available during follow-up, it is possible that patients with more measurements were more likely to have ARF detected. Although patients at lower CD4 cell counts did have a greater number of creatinine measurements available, this difference was not substantial (mean number of 13 vs 10 measures comparing patients with CD4 counts of less than 200 and 200 cells/mL or more, respectively). We are also limited by not being able to adjust for unmeasured confounders.
Because consensus on defining ARF does not exist, previous studies have relied on different criteria to identify ARF.22 By using an algorithm of proportional increases in serum creatinine obtained prospectively for case identification, followed by medical chart review, we were able to identify mild to severe cases of ARF, which have not been previously described in ambulatory HIV-infected patients.
In conclusion, relying on a large observational clinical cohort of HIV-infected patients receiving HAART allowed us to identify a dose-response effect of decreasing CD4 cell counts on increasing ARF IRs. In addition, we observed an important effect of HCV coinfection on the incidence of ARF events. Acute renal failure should be considered a potentially important complication among HIV-infected patients with normal renal function and access to HAART.
RESULTS
Study Patients
Sixty-nine percent of the 705 patients in this study were male, and 61% were African American, with a mean age of 40 years (Table 1). At baseline, the median CD4 count was 352 cells/mL (interquartile range [IQR], 162-554 cells/mL), and 30% had a CD4 count of less than 200 cells/mL. The median baseline creatinine was 0.7 (IQR, 0.6-0.9) and did not differ by CD4 cell count (P = 0.27). Patients with CD4 counts of less than 200 cells/mL had on average 3 additional creatinine measures performed during follow-up than patients with CD4 counts of 200 cells/mL or more (P < 0.001). Baseline (ie, the first HIV clinic visit after January 1, 2000) was the date of HIV care initiation for 230 patients (33%), with the remainder in HIV care a median of 3.3 years (IQR, 1.6-5.6 years). Patients were diagnosed with HIV infection a median of 5 years (IQR, 2-8 years). Almost 1 in 4 patients were HCV infected, 17% had chronic hypertension, and 6% had diabetes mellitus.
Patients received ART for a median of 5.5 years (IQR, 3.1-8.3 years), and 7% (n = 46) remained ART naive, through their first ARF event or the end of follow-up. Seventy-five percent of patients (n = 496) had received at least one PI, and 63% (n = 417) at least one NNRTI, with patients exposed to a median of 5 antiretroviral drugs (IQR, 4-8). Thirty-eight percent (n = 253) had ever received indinavir, and 22% (n = 145) received tenofovir. Thirty-eight percent (n = 270) of patients had HIV RNA levels of less than 400 copies/mL at baseline. As expected, HIV RNA level was strongly associated with CD4 cell count (P < 0.001). However, CD4 cell count was not associated with baseline renal function, age, or HCV infection.
Incidence Rate of ARF
A total of 109 ARF events occurred in 69 patients between baseline and December 31, 2002, for an overall ARF IR of 6.4 (95% confidence interval [CI], 5.2-7.7). The median time at risk when patients were censored at their first ARF event was 34.3 months (IQR, 21.0-36.5 months), giving a first event ARF IR of 4.3 events per 100 person-years (95% CI, 3.3-5.4). Time at risk was shorter among patients with ARF (n = 69) (median, 15.4; IQR, 2.6-23.7) than those without (n = 636) (median, 36.0; IQR, 23.2-36.5; P < 0.001).
Acute renal failure IRs decreased with increasing CD4 cell counts with highest rates occurring at CD4 counts less than 100 cells/mL decreasing thereafter (Fig. 1). This dose-response relationship was observed in both HCV-infected and HCV-uninfected patients, although IRs were consistently higher among HCV-infected patients especially at lower CD4 cell counts.
We found a statistically significant interaction between CD4 cell count and HCV serostatus in both unadjusted and adjusted analyses (test of homogeneity P = 0.09 and 0.09, respectively); therefore, we chose to stratify our results by CD4 cell count and HCV infection (Table 2). After adjustment for age, sex, race, diabetes mellitus, hypertension, and HBV infection, ARF IRs decreased in all strata of CD4 cell count and HCV infection, with sex accounting for the majority of the confounding observed (Table 2).
In adjusted analyses, the relative effect of HCV coinfection on ARF IRs was greater among patients with CD4 counts of ≥200 cells/mL than those with <200 cells/mL (IRR = 3.2 [95% CI, 1.5-6.9] and IRR = 1.2 [(95% CI, 0.6-2.6], respectively), corresponding to an absolute difference of 1.9 ARF events per 100 person-years (95% CI, 0.5-6.4) and 0.7 (95% CI, 0.2-2.3), respectively. The relative effect of having a CD4 count of less than 200 cells/mL versus 200 cells/mL or more on ARF IRs was greater among patients without HCV coinfection in comparison to patients with HCV coinfection (IRR = 4.7 [95% CI, 2.5-8.8] and IRR = 1.6 [95% CI, 0.7-3.9], respectively), corresponding to an absolute difference of 3.7 ARF events/100 person-years (95% CI, 1.2-11.4) and 2.5 (95% CI, 0.6-9.7), respectively.
Acute renal failure IRs increased with increasing HIV RNA levels with ARF IRs of 2.3, 4.2, and 11.0 events/100 person-years among patients with less than 2.6, 2.6 to 5.0, and more than 5.0 log10 HIV RNA copies/mL, respectively. Patients during their first year of ART had the greatest ARF IR, in contrast to patients with more than 1 year of ART exposure (19.2 and 3.3 events/100 person-years, respectively), and patients on their first ART regimen had a higher ARF IR than those who were on at least their second regimen (5.7 and 3.8 events/100 person-years, respectively). In a multivariable model including HIV RNA level, years of previous ART, and ART regimen number, both HIV RNA level and years of prior ART remained independently predictive of ARF incidence (P < 0.001, <0.001, and 0.27, respectively). Once we added CD4 cell count and HCV infection to this model, only years of previous ART, CD4 count, and HCV infection remained to be statistically significant predictors of ARF IR (P < 0.001, <0.001, and 0.02, respectively).
INTRODUCTION
Since the introduction of highly active antiretroviral therapy (HAART), few studies have addressed the incidence and risk factors of acute renal failure (ARF) despite being the most common cause of renal dysfunction in HIV-1-infected patients. Before HAART availability, severe immunodeficiency and opportunistic infections were the most common conditions associated with ARF.1-5 However, HAART use has significantly reduced HIV-1-related morbidity and mortality.6,7 Although ARF usually occurs in the setting of acute illness and multiple organ failure, it can also result from insults affecting only the kidneys, such as with drug exposure, or during transitory medical illnesses which lead to dehydration. Therefore HIV-1-infected patients may continue to experience an elevated risk of ARF because of the high prevalence of comorbid conditions, extensive provided pharmacotherapy, and increased susceptibility to infections.
We have previously described the overall incidence and underlying clinical causes of ARF in a large cohort of HIV-infected ambulatory patients.8 In brief, most ARF episodes were due to infections or antibiotic and antifungal drug toxicity, with fewer than 10% of events directly related to antiretroviral nephrotoxicity (including indinavir, tenofovir, and nevirapine). In this study, our primary aim was to estimate the effect of immunodeficiency, measured by CD4 cell counts, on the incidence rate (IR) of ARF in HIV-infected patients.
METHODS
Study Population
We relied on the University of North Carolina (UNC), Center for AIDS Research, HIV Cohort Study, which has been previously described.9 In short, all patients receiving primary HIV care at the UNC Infectious Diseases outpatient clinic are eligible to participate in this ongoing observational clinical cohort, which has enrolled more than 1900 participants since January 2000. Study data include all institutionally available electronic patient records and periodic standardized medical record reviews. For this study, we included all UNC, Center for AIDS Research, HIV Cohort Study participants receiving HIV care at UNC between January 1, 2000, and December 31, 2002 (N = 813) and defined baseline as the first visit in this interval. Patients were excluded if they did not have at least 2 measures of serum creatinine between 2000 and 2002 (n = 45), were on chronic dialysis at baseline (n = 14), or had unavailable baseline CD4 cell counts (n = 25) or hepatitis C virus (HCV) antibody results (n = 24).
Measurements
Acute renal failure was defined as an increase in serum creatinine (during at least 2 days) of (1) 0.5 mg/dL for patients with baseline serum creatinine level of less than 2.0 mg/dL, (2) 1.0 mg/dL with baseline 2.0 to 5.0 mg/dL, and (3) 1.5 mg/dL with baseline 5.0 mg/dL or higher.10 The first author (N.F., a nephrologist) reviewed all ARF events. Cases of glomerulonephritis due to the subacute nature of kidney loss were not considered ARF events.
CD4 cell counts and HIV-1 RNA levels were measured within 3 months before baseline. Hepatitis C virus serostatus was based on hepatitis C antibody testing (Ortho-Clinical Diagnostic, Raritan, NJ). Hepatitis B virus (HBV) infection was defined as a positive serum hepatitis B surface antigen (HBsAg) and/or hepatitis Be antigen (HBeAg). Covariates included age, race, sex, previous AIDS-defining clinical condition, systemic chronic hypertension, diabetes mellitus, and history of injection drug use (IDU) and were all measured at baseline. Antiretroviral therapy (ART) variables included years on therapy, number of antiretroviral agents received, and protease inhibitor (PI) and nonnucleoside reverse transcriptase inhibitor (NNRTI) use. We only included therapy received before the first ARF event or end of follow-up, whichever occurred earlier.
Statistical Analysis
Person-time at risk was defined as the time between baseline and the first ARF event, last date of follow-up, or December 31, 2002, whichever occurred earlier. Therefore, patients only contributed person-time until their first ARF episode, irrespective if they experienced a second ARF event. Incidence rates for ARF were calculated by dividing the number of events by person-time at risk. To estimate incidence rate differences (IRDs) and incidence rate ratios (IRRs) for the number of ARF events per person-time at risk, we relied on Poisson log-linear regression models. To assess model fit, we used the deviance X 2 test and calculated likelihood ratio tests of fitting alternate distributions (ie, the negative binomial).
We considered a number of possible confounding factors of the effect of CD4 cell count on ARF IR, including age, sex, race, hypertension, diabetes mellitus, and HBV coinfection. The full multivariable model included all candidate confounders and interaction terms. We first assessed for interaction between CD4 cell count and each covariate relying on a likelihood ratio test P value less than 0.1. We then used a backward elimination strategy based on a change in point estimate, where a covariate was retained if it changed our main effect estimate by 10% or more. In estimating the effect of CD4 cell count on ARF IRs, we chose not to control for possible confounding by ART exposure or HIV RNA level. This was done because the effect of CD4 cell count on ARF IRs may in part be mediated through ART exposure. CD4 cell count is an indicator of therapy use and successful therapy increases CD4 cell counts. Moreover, because HIV RNA level is closely correlated with ART exposure, we also chose not to adjust for HIV RNA level. Instead, we fit separate models for the effects of ART exposure and HIV RNA level on ARF IRs. Finally, we did not include IDU in any model that included HCV because these 2 factors are highly correlated. All analyses were performed using the SAS statistical package version 8.2 (SAS Institute, Cary, NC).
|
|
| |
| |
|
|
|
