The Impact of Hepatitis C Virus Coinfection on HIV Progression Before and After Highly Active Antiretroviral Therapy
*Marina B. Klein; *Richard G. Lalonde; †Samy Suissa
Divisions of *Infectious Diseases/Immunodeficiency and †Clinical Epidemiology, Department of Medicine, Royal Victoria Hospital, McGill University Health Centre, Montreal, Quebec, Canada
J Acquir Immune Defic Syndr 33(3):365-372, 2003.
".....In the post-HAART era, HCV-seropositive individuals were approximately three-times more likely to die and 2.5-times more likely to be hospitalized compared with HCV- individuals...."
To compare the impact of hepatitis C virus (HCV) coinfection on progression of HIV infection in the eras before and after the introduction of highly active antiretroviral therapy (HAART), the authors conducted a retrospective cohort
study. One hundred twenty-five HCV+ patients and 1076 HCV- patients were studied; 83% of HCV+ patients were injection drug users. HCV+ subjects experienced no clear benefit from HAART. The adjusted hazard ratios (HRs) of
opportunistic infection, death, and hospitalization were 0.74 (95% CI: 0.31-1.78), 1.78 (95% CI: 0.59-5.37), and 2.1 (95% CI: 0.90-4.90), respectively, comparing the post-HAART era with the pre-HAART era. In contrast, HCV-
subjects experienced rate reductions for all outcomes. Comparable HRs for opportunistic infection, death, and hospitalization were 0.49 (95% CI: 0.37-0.64), 0.28 (95% CI: 0.19-0.41), and 0.51 (95% CI: 0.38-0.67), respectively. HCV+ subjects remained at increased risk for death and hospitalization post-HAART even after additional adjustment for antiretroviral use and time-updated CD4 cell and viral load measures. Deaths and hospitalizations in
HCV+ patients were primarily for non-AIDS-defining infections and complications of injection drug use. HCV coinfection and comorbidity associated with injection drug use are preventing the realization of substantial health benefits associated with HAART.
As the nature of the HIV epidemic changes, there is increasingly an overlap between the populations at risk for acquiring HIV and hepatitis C virus (HCV) infections because of shared routes of transmission. Injection drug use is the fastest growing risk for both infections, accounting for 26% of new HIV infections in Canada in 1999. The impact of HCV on the progression of HIV infection is debated. Most studies, conducted in the era before highly active antiretroviral therapy (HAART), suggest that HCV has a limited effect on the progression or severity of HIV disease. The two studies conducted in the post-HAART era have conflicting results. The Swiss Cohort found that HCV coinfection was associated with HIV progression, whereas the Johns Hopkins Cohort found no such association.
The introduction of HAART has led to a dramatic reduction in the morbidity and mortality associated with HIV infection overall. Therapy for HIV in the context of HCV coinfection, however, can be problematic. High rates of active
substance abuse and resulting unstable social conditions among the coinfected may be preventing such individuals from ever accessing antiretroviral therapy. For those who do take treatment, these factors predispose to poor adherence,
which, in turn, may lead to treatment failure. Furthermore, several recent reports have demonstrated hepatic decompensation and liver injury following the introduction of HAART. Thus, the medications that could provide the
most benefit may not be well tolerated in this group of patients. For these reasons, it is quite likely that HCV-coinfected patients may not have derived equal benefit from the revolution brought about by HAART.
We undertook this study to further investigate the influence of HCV coinfection on morbidity and mortality associated with HIV infection and to examine the impact of the availability of HAART on these outcomes for each group.
Nearly 4 million persons in the United States and an estimated 170 million worldwide have been infected with hepatitis C virus (HCV). Transmission occurs primarily via the parenteral route, and injection drug users (IDUs) are at
particularly high risk. It has been reported that 40-85% of IDUs are HCV seropositive, and injection drug use (IDU) accounts for approximately 60% of HCV transmission in the United States. Although acute HCV infection is usually
mild, 70-80% of those infected develop chronic hepatitis, with substantial long-term morbidity and mortality due to HCV-related chronic liver disease.
Among HIV-infected IDUs, rates of HCV co-infection range from 52-93%. As potent antiretroviral therapy for HIV has decreased AIDS-associated morbidity and mortality, chronic HCV has become a significant cause of morbidity and
mortality in this population. End-stage liver disease has become a leading cause of death in HIV-infected persons. HIV co-infection may accelerate progression of HCV infection, with more severe liver fibrosis12 and a higher frequency of decompensated liver disease and cirrhosis. HIV infection is also associated with a higher level of HCV RNA. However, HCV RNA levels have not been closely correlated with progression of liver disease, unlike HIV RNA levels
and immunodeficiency. In HCV-antibody-positive persons, detectable HCV RNA has been associated with HIV seropositivity, and HCV RNA levels have been found to be higher in HIV-positive subjects, those with advanced age, and those with high HIV viral loads. Some studies have found a relationship in co-infected persons between CD4+ lymphocyte count and HCV RNA level, while others have not. Additionally, in some studies HCV RNA levels have been found to be higher and histologic severity of liver disease worse in persons with HCV genotype 1 infections. However, >90% of the person-to-person HCV RNA level variability remains unexplained by factors so far examined.
The primary outcomes studied were progression to a new AIDS-defining illness, death, and hospitalizations. Opportunistic infections and malignancies were considered to be AIDS defining (category C) according to the Centers
for Disease Control and Prevention classification of 1993. Medical records and death certificates were reviewed to verify database information.
PATIENTS AND METHODS
Data Collection and Patients
The Immunodeficiency Clinic at the Montreal Chest Institute, a large university-based HIV clinic, has maintained a database of all patients since 1989 into which clinical, laboratory, and prescription information has been prospectively
collected. As of 1998, risk factors for HIV acquisition have been prospectively collected. For subjects entered into the cohort prior to 1998, data on risk factors were obtained through chart review. Subjects were included for study if they were HIV-seropositive and had more than one documented clinic visit.
The pre-HAART era was defined as between January 1, 1989 and December 31, 1995, the period before routine use of protease inhibitor-containing regimens. Entry into the pre-HAART cohort began on the date of the first clinic visit (time
zero). Subjects were followed-up until January 1, 1998 or were right censored if they began HAART. The post-HAART era comprised patients entering the clinic between January 1, 1996 and June 30, 1999. These subjects were followed through December 31, 1999. Patients who were lost to follow-up were censored on the date of their last documented clinic visit.
Of 1462 HIV-seropositive patients in the database, 1201 had more than one clinic visit recorded. The pre-HAART cohort comprised 662 individuals, whereas 539 subjects were in the post-HAART cohort.
Forty-two patients (6%) were seropositive for HCV. Median follow-up time was 557 days (range: 2-2673 days). One hundred seventy-six subjects were censored because they began HAART, and 226 were lost to follow-up. As the majority (64%) of HCV+ patients entered the cohort in 1994 and 1995, HCV+ subjects had less advanced HIV infection at cohort entry, with a shorter duration of HIV infection (p = 0.04), higher baseline CD4 cell counts (p = 0.01), and
fewer prior AIDS-defining illnesses (p = 0.01) compared with HCV- patients. HCV+ subjects were also more likely to have injection drug use as a risk factor for acquiring HIV.
HCV infection was documented in 83 patients (15%). Median follow-up time was 532 days (range: 2-1405 days). One hundred seventy-six of 539 subjects were lost to follow-up. The only significant difference between HCV+ and HCV-
subjects was the proportion with injection drug use as a risk factor (83% of HCV+ subjects vs. 5% HCV- subjects; p <.001). HCV+ subjects were somewhat less likely to have had a prior AIDS-defining illness (p = .08).
Comparison of Hepatitis C Virus+ Subjects With Hepatitis C Virus- Subjects in the Two Eras
In the pre-HAART era, the unadjusted rates were consistently lower for HCV+ subjects compared with HCV- subjects for all outcomes. The rates were 11.0 versus 22.9 per 100 person-years for opportunistic infection, 6.9 versus 14.9 per 100 person-years for death, and 9.7 versus 21.0 per 100 person-years for hospitalization, respectively, and are illustrated in crude survival analyses. After adjustment, rate ratios remained below 1, but confidence bounds
In the post-HAART era, there was no difference in the rate of opportunistic infection observed between HCV+ and HCV- subjects (unadjusted rate: 12.8 vs. 13.1 per 100 person-years).
A reversal occurred with respect to the other outcomes, however. HCV+ subjects in the post-HAART era experienced greater rates of both death (unadjusted rate: 9.5 vs. 5.3 per 100 person-years) and hospitalization (unadjusted rate: 25.9 vs. 12.6 per 100 person-years). The adjusted risks of death and hospitalization remained significantly higher among
In both the pre- and post-HAART eras, baseline CD4 cell count was an independent protective factor against opportunistic infection, death, and hospitalization (i.e., in the post-HAART era, for each additional 100 cells, the HR of opportunistic infection was 0.50, 95% CI: 0.39-0.65, p = .001). History of prior AIDS-defining illness increased risks of death and hospitalization, and longer duration of HIV infection increased risk of death. Calendar year of cohort entry was strongly associated with primary outcomes. For example, after HAART, the risk of death diminished substantially in each additional year (i.e., HR = 0.046, 95% CI: 0.013-0.165 for 1999 compared with 1996). Adjustment for calendar year did not change risk estimates associated with HCV status, however. The following factors were not associated with primary outcomes: age, gender, birthplace, and baseline CD8+ cell count.
Time-updated CD4 values were stronger predictors than baseline CD4 measures for all outcomes, particularly opportunistic infection and death. To a lesser extent, time-updated viral load measures were better predictors than
baseline viral load measures of outcomes. Adjustment for time-dependent CD4 and viral load measures, however, had little impact on the risk estimates obtained for the effect of HCV serostatus on the risk of primary outcomes.
In the pre-HAART era, both HCV+ subjects and HCV- subjects died predominantly from AIDS-related complications. Despite chart review, the exact cause of death could not be determined for a number of subjects who died at home or in a hospice (likely AIDS related). In the post-HAART era,
deaths and hospitalizations among HCV+ subjects were primarily attributable to non-AIDS-defining infections (i.e., bacteremia, pneumonia) and complications of injection drug use.
Distinguishing Hepatitis C Virus Serostatus From Injection Drug Use
In the post-HAART era, greater than 80% of HCV+ subjects had injection drug use as a risk factor compared with fewer than 5% of HCV- individuals, thus precluding inclusion of both variables in a single analysis. When injection drug use history was modeled as the primary exposure, very similar hazards of the primary outcomes to those observed with HCV status were obtained (i.e., adjusted HR of death = 3.03, 95% CI: 1.42-6.45). In a stratified analyses, HCV+
status remained associated with death independent of the history of injection drug use (adjusted HR = 1.97, 95% CI: 0.25-15.82), but because of the small number of outcomes among HCV+ non-IDUs, the CIs obtained were very large,
rendering interpretation of the risk estimates difficult.
Hepatitis C virus infection is fast becoming one of the greatest challenges facing an increasing number of HIV-infected individuals and their care providers. Understanding the impact of coinfection on morbidity and mortality in HIV infection is essential to optimizing the management of the epidemic in developed countries. In this large clinical cohort, we observed a 60% increase in the seroprevalence of HCV over a period of 10 years, likely reflecting a shift in the HIV epidemic to involve more IDUs. As expected, the introduction of HAART led to a significant reduction in the rates of opportunistic infection (40%), death (64%), and hospitalization (14%) among HCV- individuals. HCV-seropositive subjects did not derive benefit from this reduction in morbidity and mortality, however. In fact, among HCV+ subjects, the rates of opportunistic infection and death remained similar to those observed in the pre-HAART era, and hospitalization rates actually increased. In the post-HAART era, HCV-seropositive individuals were approximately three-times more likely to die and 2.5-times more likely to be hospitalized compared with HCV- individuals.
The impact of HCV infection on the progression of HIV infection in the era prior to widespread use of HAART has been debated. The majority of older studies have shown no association between HCV infection and survival from HIV disease.[1-3, 14] In contrast, more recent studies have reported that immunologic progression, development of AIDS, and shorter survival after AIDS were associated with HCV coinfection.
In designing this study, we attempted to overcome many limitations present in earlier studies. Notably, care was taken to ensure that the impact of time in terms of the duration of HIV infection and calendar time was accounted for in the analysis of outcomes. In addition, we adjusted for potential confounders, namely, CD4 cell count, viral load, and prior treatment history. Finally, we examined the role of exposure to antiretroviral medications and changes in immune and virologic parameters during the course of follow-up using time-dependent proportional hazards modeling. After such adjustment, we found no increased risk of disease progression or death among HCV+ individuals in the pre-HAART era. Death was attributable to HIV-related conditions in both groups, suggesting that without optimal control of HIV replication, HCV coinfection has little impact on morbidity and mortality as AIDS rapidly ensues.
Although the association between HCV status and outcomes was not found to be significant in the pre-HAART era, it is notable that the risk estimates obtained for all outcomes were less than 1. The changing epidemiology of HIV infection may have affected our ability to assess the role of HCV in the pre-HAART era. Injection drug use began to account for a significant proportion of HIV infections only after 1995. This shift has two potential implications for this study. First, fewer HCV+ patients had injection drug use as a risk factor in the pre-HAART era; thus, any contribution of injection drug use to morbidity and mortality may have been less pronounced. Second, it meant that the majority of coinfected patients entered into the study late in the pre-HAART era and had a shorter duration of HIV infection. Thus, HCV+ subjects were clinically less advanced with respect to their HIV infection and had shorter follow-up than the controls; therefore, there may have been residual confounding not completely corrected for in the adjusted analyses. It is also possible that a number of the untested individuals may have been HCV+, resulting in misclassification of their exposure status, which would have tended to diminish any effect of HCV status on outcomes. Nevertheless, given that more than 98% of untested individuals who subsequently underwent HCV testing after study completion were found to be HCV-, such misclassification bias was likely negligible.
Fewer studies have examined the impact of HCV on HIV progression in the post-HAART era, and findings have also been contradictory. Recently, the Swiss HIV Cohort Study reported that HCV infection was independently associated with the combined outcome development of a new AIDS-defining condition or death (HR = 1.7, 95% CI: 1.26-2.30) in individuals beginning HAART. In contrast, the Johns Hopkins Cohort, which included both treated and untreated individuals, detected no increased risk in HCV-coinfected persons of a first opportunistic infection, CD4 cell decline, or death once adjustments were made for use of HAART and failure to suppress HIV replication. The reason for these discrepant findings is unclear but may be related to demographic differences in the populations studied and the inclusion criteria employed, primarily the receipt of HAART. By examining both the pre-HAART and post-HAART eras, we attempted to assess the impact of HCV on HIV progression directly while accounting for HAART exposure.
In the post-HAART era, we found that HCV status was clearly associated with an increased risk of both death and hospitalization but not of opportunistic infection. The lack of an observed effect of HCV status on the risk of opportunistic infection raises the possibility that HCV status does not influence HIV-related outcomes so much as other health events (i.e., other infections, liver disease). The use of prophylactic medications by the majority of subjects at risk in both the HCV+ and referent groups may also have reduced any difference between the cohorts. The reasons for increased rates of death and hospitalization observed among HCV-seropositive subjects are likely multifactorial. Although unexpected, and in contrast to the Johns Hopkins Cohort, we found that HCV+ subjects were prescribed HAART as often as HCV- subjects, suggesting that HCV infection itself did not appear to hinder physicians from using these therapies. Immune responses experienced by HCV+ subjects to therapy may have been weaker once such therapy was initiated, however. HCV+ subjects experienced a 50% reduction in the likelihood of achieving a CD4 response of at least 50 cells/\0xB5L despite having similar HIV suppression. A similar finding was observed in the Swiss Cohort study. Reduced immune efficacy of HAART among HCV+ subjects could be explained by poorer adherence. We did not have data on adherence to address this possibility directly. Arguing against it, however, was the lack of difference in virologic control on HAART between HCV+ and HCV- subjects. Furthermore, there was no evidence of a greater rate of virologic failure among HCV+ subjects. Finally, adjustment for time-updated CD4 cell counts and viral load measures did not alter the risk estimates obtained, supporting that factors other than treatment failure were responsible for the differences in outcomes observed. HAART does not result in diminished HCV replication.[17, 18] Cytokine dysregulation associated with ongoing HCV infection may thus impair immune recovery despite adequate HIV control. Direct infection of CD4+ T cells by the HCV virus is another potential mechanism whereby CD4 cell expansion may be prevented in the setting of coinfection.
Undoubtedly, one of the major factors contributing additional increased risk of morbidity and mortality associated with HCV infection is the high prevalence of active injection drug use in this population. This lifestyle has been shown to be associated with violent and accidental deaths, suicide, and overdose, which were also seen in our cohort. The Swiss Cohort Study found that active injection drug use was independently associated with opportunistic infection and death combined. Because we did not have reliable data on the number of HCV+ subjects actively using drugs, we could not control for this important covariate. In a stratified analysis, however, HCV+ status remained associated with death independent of the history of injection drug use, suggesting that factors other that active drug use were also responsible for the association of HCV with death.
This is the only study to have examined the morbidity and mortality associated with HCV coinfection in both the pre-HAART and post-HAART eras and regardless of antiretroviral treatment. Our findings suggest that HCV infection, and comorbidity associated with injection drug use, are preventing the realization of improved health outcomes in this population.