| |
HCV coinfection is independently associated with decreased adherence to antiretroviral therapy in a population-based HIV cohort
|
| |
| |
AIDS: Volume 20(3) 14 February 2006 p 323-331
..... In their first year of ART, HCV/HIV-coinfected patients were significantly less likely to be at least 95% adherent to their treatment (42% versus 72%; P < 0.001).....Our data support those of others who have found that HCV infection is associated with treatment discontinuations and treatment interruptions (see Author Discussion below)..... coinfected with HCV and HIV were significantly more likely to have had ALT ≥ 2.5X ULN in their first year of treatment (26% versus 8%; P < 0.001), ALT ≥ 5X ULN (4% versus 1%; P = 0.049), AST ≥ 2.5X ULN (15% versus 2%; P < 0.001) or albumin ≥ 30 g/l (20% versus 11%; P = 0.034).....
Braitstein, Paulaa,b,e; Justice, Amyf; Bangsberg, David Rg; Yip, Benitaa;
Alfonso, Victoriaa,c; Schechter, Martin Ta,b; Hogg, Robert Sa,b; Montaner, Julio SGa,d
From the aBritish Columbia Center for Excellence in HIV/AIDS
bDepartment of Health Care and Epidemiology
cDepartment of Counseling Psychology
dDepartment of Medicine, University of British Columbia, Vancouver, Canada
eInstitute of Social and Preventive Medicine, University of Bern, Bern, Switzerland
fYale University School of Medicine and New Haven VA Connecticut Healthcare System, West Haven, Connecticut
gUniversity of California-San Francisco AIDS Research Institute, USA.
The primary objective of the present study was to describe the effect of HCV coinfection, controlling for biochemical markers of liver injury and IDU, on adherence to ART during the first year of therapy in a population-based HIV/AIDS drug treatment programme. It was hypothesized that HCV serostatus would have an independent effect on adherence, and that this would be the result of more liver injury, as evidenced by an interaction between HCV serostatus and any biochemical marker of liver injury. A secondary hypothesis was that the effect of HCV infection on adherence would be independent of the effect of IDU on adherence.
Abstract
Objective: To characterize the impact of hepatitis C (HCV) serostatus on adherence to antiretroviral treatment (ART) among HIV-infected adults initiating ART.
Methods: The British Columbia HIV/AIDS Drug Treatment Program distributes, at no cost, all ART in this Canadian province. Eligible individuals used triple combination ART as their first HIV therapy and had documented HCV serology. Statistical analyses used parametric and non-parametric methods, including multivariate logistic regression. The primary outcome was ≥ 95% adherence, defined as receiving ≥ 95% of prescription refills during the first year of antiretroviral therapy.
Results: There were 1186 patients eligible for analysis, including 606 (51%) positive for HCV antibody and 580 (49%) who were negative. In adjusted analyses, adherence was independently associated with HCV seropositivity [adjusted odds ratio (AOR), 0.48; 95% confidence interval (CI), 0.23-0.97; P = 0.003], higher plasma albumin levels (AOR, 1.07; 95% CI, 1.01-1.12; P = 0.002) and male gender (AOR, 2.53; 95% CI, 1.04-6.15; P = 0.017), but not with injection drug use (IDU), age or other markers of liver injury. There was no evidence of an interaction between HCV and liver injury in adjusted analyses; comparing different strata of HCV and IDU confirmed that HCV was associated with poor adherence independent of IDU.
Conclusions: HCV-coinfected individuals and those with lower albumin are less likely to be adherent to their ART.
Results
Of the 1388 individuals in the HOMER cohort, there were 1186 patients available for this analysis, including 606 (51%) who were HCV antibody positive, and 580 (49%) who were negative. Those with HCV serological data available (i.e., baseline sample available and with sufficient quantity, and HCV serology results unambiguous) were more likely to be male (86% versus 75%; P < 0.001), but there was no different in their history of IDU (27% versus 27%; P = 0.921), having an AIDS diagnosis at baseline (13% versus 14%; P = 0.683), median age at baseline (37 versus 37 years), median baseline absolute CD4 cell count (260 versus 270 cels/μl; P = 0.142) and baseline HIV RNA (5.1 versus 5.1 log10 copies/ml; P = 0.364).
HCV antibody-positive individuals were less likely to be male (78% versus 93%; P < 0.001) or to have had an AIDS diagnosis at baseline (11% versus 15%; P = 0.028), but they were more likely to have any history of IDU (47% versus 6%; P < 0.001) (Table 1). There were no statistical differences (P > 0.05) between HCV-positive and HCV-negative patients at baseline in terms of age, absolute CD4 cell count, log plasma HIV RNA, or type of therapy initiated.
In their first year of ART (Table 2), HCV/HIV-coinfected patients were significantly less likely to be at least 95% adherent to their treatment (42% versus 72%; P < 0.001). They had a higher median ALT (52 versus 35 IU/l; P < 0.001), AST (46 versus 29 IU/l; P < 0.001), and INR (1.1 versus 1.0; P = 0.037), and lower albumin (39 versus 41 g/l; P < 0.001). Those coinfected with HCV and HIV were significantly more likely to have had ALT ≥ 2.5X ULN in their first year of treatment (26% versus 8%; P < 0.001), ALT ≥ 5X ULN (4% versus 1%; P = 0.049), AST ≥ 2.5X ULN (15% versus 2%; P < 0.001) or albumin ≥ 30 g/l (20% versus 11%; P = 0.034). There was a trend of a difference in the proportions of having INR ≥ 1.3 (15% versus 8%; P = 0.136). Overall, those coinfected with HCV/HIV were significantly more likely to have any biochemical marker of liver injury (26% versus 8%; P < 0.001).
As summarized in Table 3, adherent patients were more likely to be male (91% versus 78%; P < 0.001), to have had an AIDS diagnosis at baseline (16% versus 8%; P < 0.001) and to be slightly older at baseline (median 37.6 versus 36.7 years; P = 0.001). They were less likely to have a history of IDU (21% versus 35%; P < 0.001), to be HCV antibody positive (38% versus 69%; P < 0.001) and to have any biochemical marker of liver injury (12% versus 25%; P < 0.001), including ALT ≥ 5X ULN (1% versus 4%; P = 0.019) or albumin ≥ 30 g/l (8% versus 24%; P < 0.001). Adherent patients had significantly lower median AST (31 versus 37 IU/l; P < 0.001) and higher albumin (41 versus 38 g/l; P < 0.001), but there was no difference in their median ALT or INR.
Tables 4 and 5 summarize the results of the logistic regression analysis of factors independently associated with being at least 95% adherent to ART. As seen in Table 4, which presents the results using continuous measures of liver injury, the adjusted odds ratio (AOR) and 95% confidence interval (CI) for HCV infection remained significant in the final model (AOR, 0.39; 95% CI, 0.22-0.73; P = 0.003), as did higher albumin levels (AOR, 1.07; 95% CI, 1.03-1.11; P = 0.002) and male sex (AOR, 2.42; 95% CI, 1.18-4.96; P = 0.016). Even when only adjusting for biochemical markers of liver injury but not HCV coinfection, IDU became non-significant (AOR, 0.67; 95% CI, 0.41-1.14).
As seen in Table 5, which contains the final model, both HCV coinfection (AOR, 0.35; 95% CI, 0.24-0.51; P < 0.001) and liver injury (AOR, 0.47; 95% CI, 0.23-0.94) were negatively associated with adherence, while male sex remained positively associated (AOR, 2.58; 95% CI, 1.59-4.18; P < 0.001). The interaction term of HCV plus liver injury fell out of the final model (AOR, 1.42; 95% CI, 0.62-3.25; P = 0.412). Interestingly, although IDU became non-significant in the final model (AOR, 0.80; 95% CI, 0.55-1.16; P = 0.243), it did remain significant after adjusting only for liver injury, but not other factors such as HCV.
Of the 544 individuals with no reported history of IDU and negative HCV serology, 400 (74%) were at least 95% adherent, and 144 (26%) were not. Of the 36 individuals who had a reported history of IDU but negative HCV serology, 20 (55%) were adherent and 16 (45%) were not (P = 0.019). Of the 321 individuals with no history of IDU but who had positive HCV antibodies, 134 (42%) were adherent while 187 (58%) were not (P < 0.001). Finally, of the 285 individuals with both a reported history of IDU and HCV infection, 121 (42%) were adherent, while the remaining 164 (58%) were not (P < 0.001). Pearson's correlation coefficient between any IDU and hepatitis C positive antibodies was 0.46.
Tables 6 and 7 summarize the results of the analyses intended to distinguish the effects of HCV and IDU. As seen in Table 6, without exception, the individuals who had a history of IDU but no HCV infection had the most favourable results regarding biochemical markers of liver injury, while those with positive HCV antibodies had the worst. In logistic regression analysis only controlling for baseline sociodemographic and clinical factors (Table 7), the IDU-positive/HCV-negative group was less likely than those negative for both IDU and HCV to be at least 95% adherent to medication (AOR, 0.47; 95% CI, 0.23-0.93; P = 0.030); however, those with HCV infection with or without IDU history were even less likely to be adherent (IDU-negative/HCV-positive group: AOR, 0.27; 95% CI, 0.20-0.36; P < 0.001; IDU-positive/HCV-positive group: AOR, 0.30; 95% CI, 0.22-0.42; P < 0.001). When also controlling for biochemical markers of liver injury as continuous measures, those with a history of IDU but no HCV were not independently less likely to adhere to treatment (AOR, 0.19; 95% CI, 0.03-1.32; P = 0.126), while those with HCV infection (with or without IDU) remained significantly less likely to be adherent (IDU-negative/HCV positive group: AOR, 0.34; 95% CI, 0.18-0.64; P < 0.001; IDU-positive/HCV-positive group: AOR, 0.44; 95% CI, 0.23-0.87; P = 0.018).
Discussion
The data presented here suggest that biochemical markers of liver injury (notably albumin), male sex and to a lesser extent age are those factors most strongly associated with incomplete adherence to ART in HIV/HCV coinfection. While our findings do not suggest an interaction between HCV and liver injury, they do support the hypothesis that HCV is more strongly associated with adherence than a history of IDU.
Our data support those of others who have found that HCV infection is associated with treatment discontinuations and treatment interruptions. Melvin et al. [24] reported ART discontinuation rates owing to hepatic toxicity of more than two-fold in HCV/HIV-coinfected individuals compared with those with only HIV infection. Among a population of 465 previously ART-naive individuals, HCV seropositivity was associated with an adjusted 40% increased risk of discontinuing or changing initial HAART regimens within the first year of treatment [23]. The authors could not differentiate whether this was a consequence of histological damage, reduced adherence or increased hepatotoxicity, because hepatic cirrhosis was also independently associated with HAART discontinuation (AOR, 2.1; 95% CI, 1.1-3.8) [23]. Aceti et al. [34]. reported that 83% of those who discontinued ART because of hepatotoxicity were coinfected with HCV. Of note, d'Arminio Monforte et al. [25] found no impact of HCV on treatment interruptions owing to toxicity. As our adherence measure was based on prescription refills, treatment interruptions and discontinuations were encompassed in our analysis.
There are a number of important strengths to our analysis. First, the data were drawn from a population-based programme and, therefore, are more broadly generalizable than clinic-based populations. Second, we endeavoured to untangle the rather complex relationship between HCV infection, IDU and hepatic injury, enabling a much deeper understanding of adherence as it relates particularly to HCV infection and IDU. Third, to our knowledge, we are the first to examine explicitly the question of HCV/HIV coinfection and adherence to ART, and to ask the question about the combined effect of HCV/HIV coinfection and hepatic injury. Finally, more effective treatment for HCV (i.e., pegylated interferon combined with ribavirin) only became available in British Columbia in June 2003, and as this analysis was restricted to patients' first year on HIV treatment, the effect of HCV treatment on our results was negligible.
There may also be potential limitations to our analysis. One, the markers of liver injury were only available for 72% of the cohort. Two, the measures of liver injury were biochemical, and ALT and AST levels in particular may not be reliable measures of liver inflammation or injury, as others have shown that patients with normal ALT levels may have hepatic scarring [35]. Furthermore, there are many symptoms of HCV disease and ART hepatotoxicity that could impact on adherence, without being captured by these measures, including nausea, malaise, etc. This may have particularly affected individuals concurrently receiving treatment for HCV infection, although prior to June 200 very few HIV-infected individuals received any treatment for their HCV infection because of the very poor response rates [2]. Increased morbidity from HCV disease or ART toxicity may be the reason why HCV infection itself remains associated with adherence in adjusted models. A third limitation to our study is that our measure of IDU was a composite one, drawing upon both physician and patient self-report data. The fact that only 47% of those with HCV infection have any reported history of IDU suggests that there may be underreporting of this variable. However, as HCV is highly infectious and individuals at risk for HIV infection may also have multiple risk factors for HCV (e.g., certain sexual practices, tattooing, intranasal cocaine use), this underestimation may be minimal. Further, the IDU measure was based on any history of IDU, rather than active or current use, and may have included those who used injection drugs recreationally many years ago but who were otherwise not typical of those IDU generally considered at high risk of non-adherence (i.e., actively street involved and actively using injection drugs). A final limitation of this analysis is that we were not able to account for patients' educational backgrounds, or the effects of alcohol use or depression/depressive symptoms on adherence; alcohol in particular may have affected both adherence and progression of HCV disease. Indeed, as we were not able to delineate reasons for non-adherence, or to distinguish between non-adherence linked to lifestyle issues and non-adherence or treatment interruptions linked to increased toxicity or comorbidity, these results should be viewed as preliminary and indicative that further research is warranted.
Our data have important implications for the conduct of clinical research regarding HIV/HCV coinfection and IDU. Accurate estimates of effects owing to IDU may be difficult to obtain without adjusting for the effects of HCV infection, and vice versa. Further research is urgently needed to elucidate the mechanisms involved in precluding or preventing HIV/HCV-coinfected individuals and those with a history of IDU from achieving maximal adherence to their ART. Further, these data may explain why some authors have found a negative association between HCV infection and survival in the setting of ART [11]. As non-adherence is clearly associated with mortality, and HCV infection is associated with reduced adherence, it may be that the effect of HCV on survival is mediated through adherence to treatment rather than a biological interaction between HIV and HCV.
In conclusion, our data suggest that HCV infection, albumin (as a marker of hepatic injury) and male sex are independently associated with adherence; after adjusting for these other factors, use of injection drugs is not. Further, although HCV/HIV-coinfected individuals are more likely to have any biochemical evidence of liver injury, the combined effect of HCV and liver injury in this analysis was negligible, although both HCV and liver injury were both independent factors.
Introduction
It is estimated that, on average, 30% of HIV-positive people in North America and Europe are coinfected with hepatitis C virus (HCV) [1,2]. However, because HCV infection is so widespread in some populations, including injection drug users (IDU), prison inmates and haemophiliacs, the prevalence of HCV coinfection among HIV-positive individuals in these populations ranges from 50 to 99% [3-6]. As survival has been increasing for patients with HIV infection through the use of HAART, liver disease has emerged as a leading cause of morbidity and mortality in HIV-infected populations [7-9]. Coinfection with HCV, in particular, because of its prevalence and increased pathogenicity in the setting of HIV [2,10], is now a leading cause of death for HIV-positive individuals [9,11-13].
Adherence to antiretroviral therapy (ART) is paramount to its effectiveness [14-16]. Factors previously found to be associated with adherence are physician experience [17], age [18], alcohol use and incarceration [18], toxicity [19] and IDU [18,20]. The last has also been identified as an important predictor of treatment discontinuation [21,22]. A majority of HIV-positive IDU are also coinfected with HCV; however, only a few studies have examined HIV/HCV coinfection as a factor in switching or interrupting HAART. While most have found that there is a strong relationship between treatment interruptions and HCV coinfection [23-25], none have looked explicitly at the question of adherence. There may be several reasons why HCV coinfection could independently lead to non-adherence, particularly increased toxicity or comorbidity [11,26].
Methods
Data source and study population
The British Columbia HIV/AIDS Drug Treatment Program has been distributing ART centrally at no cost to eligible HIV-infected individuals since 1986. This programme remains the only free source of ART in this Canadian province (and is a unique programme in Canada). It has received ethical approval from the University of British Columbia Ethics Review Committee at its St Paul's Hospital site, and the programme conforms with the province's Freedom of Information and Protection of Privacy Act.
The Drug Treatment Program distributes ART based on specific guidelines generated by the Therapeutic Guidelines Committee [27]. These guidelines have been and continue to be consistent with those treatment guidelines published by the International AIDS Society [28-31].
Individuals are automatically entered into the Drug Treatment Program when they are first prescribed any antiretroviral agent. At programme entry, and with each subsequent physician visit, the participant's complete history (if any) of ART use, CD4 cell count, plasma viral load and disease stage are recorded. Typically, patients are followed-up at 3-month intervals, at which time prescriptions are renewed or altered as necessary.
The HAART Observational Medical Evaluation and Research (HOMER) study is a nested cohort within the BC Centre for Excellence's HIV/AIDS Drug Treatment Program. It includes all previously ART-naive individuals who initiated ART between July 1996 and August 2000 with a triple-drug regimen consisting of two nucleoside reverse transcriptase inhibitors plus either a protease inhibitor or a non-nucleoside reverse transcriptase inhibitor.
The data used in these analyses were from individuals in the HOMER Cohort, for whom there was HCV serological data. This data was obtained through retrospective testing of stored plasma samples taken 0 to 6 months prior to treatment initiation. Plasma was separated within 6 h of blood collection by centrifuging at 800-1600 X g for 15 min at room temperature and was stored frozen at -20 C until time of processing. Samples were tested for HCV antibody using the Axsym System version 3.0 (Abbott, Abbott Park, Illinois, USA). This assay has a sensitivity greater than 99% and a specificity of 99% in immune-competent patients [32]. All testing was completed at the University of British Columbia Diagnostic Virology and Reference Laboratory at St Paul's Hospital, and ethics approval was obtained prior to testing.
The BC Centre for Excellence in HIV/AIDS is located within Providence Health Hospital in Vancouver. The markers of hepatic injury were obtained through a linkage with the Providence Laboratory at St Paul's Hospital in Vancouver, where a majority of HIV-infected patients in the province receive their HIV care.
Explanatory and outcome measures
The primary outcome measure for this analysis was adherence to ART. The definition of adherence was based on the time that the medication dispensed would last compared with the follow-up time. This calculation was restricted to each patient's first year on therapy to avoid the reverse causation that could occur among patients who cease ART after they have become too sick to take medication. It has been previously demonstrated how this estimate strongly predicts virological response [15] and survival [14,33]. For the purposes of these analyses, adherence was a binary variable where 1 was ≥ 95% adherent (meaning having refilled at least 95% of their prescriptions during the year), and 0 was < 95% adherent.
Baseline characteristics examined were gender (male versus female), age (continuous), absolute baseline CD4 cell count (continuous) and plasma HIV RNA on a log10 basis (continuous), whether patients had an AIDS diagnosis (yes versus no) or any history of IDU (yes versus no), and type of therapy initiated (based on protease inhibitor or a non-nucleoside reverse transcriptase inhibitor). The IDU variable was based on both physician report from the initial prescription when individuals commenced treatment and patient self-report in the participant survey that the BC Centre for Excellence in HIV/AIDS conducts annually.
Markers of liver injury were alanine aminotransferase (ALT), aspartate aminotransferase (AST), albumin, and international normalized ratio (INR). The analysis was restricted to the patient's first year on ART, and if a patient had more than one test result in that year, the latest result was taken. These markers were analysed as continuous variables to allow for the possibility that patients could be experiencing symptoms of liver injury without necessarily meeting a particular threshold of 'significant' liver injury [e.g. ALT ≥ 2.5X the upper limit of normal (ULN)]. Markers of liver injury were also defined dichotomously accordingly to the following thresholds: ALT ≥ 2.5X ULN, AST ≥ 2.5X ULN, albumin ≥ 30 g/l and INR ≥ 1.3. A global measure of hepatic injury was created by defining a patient as having any liver injury if they had at least one of the four dichotomous outcome measures, thus enabling an interaction to be created between HCV serostatus and 'any liver injury'.
Statistical analysis
HCV-positive and HCV-negative participants were compared using both parametric and distribution-free methods. Categorical data were analysed using Pearson's X2 test. Fisher's exact test was used for contingency tables in which 25% or more of the expected cell frequencies were < 5. Continuous variables were analysed using the Wilcoxon rank sum test. Adherent versus non-adherent individuals were analysed in a similar fashion.
In unadjusted logistic regression, HCV antibody status (positive versus negative), any history of IDU (yes versus no), gender (male versus female), baseline CD4 cell count (per 100 cells/μl), age at baseline (per 10 years) and whether the patient had an AIDS diagnosis at baseline (yes versus no) were separately tested for their association with adherence, as were ALT, AST, albumin, and INR, as both continuous and dichotomous variables. Separate models were constructed to assess the relative effects of HCV-positive biochemical markers of liver injury, IDU-positive biochemical markers of liver injury or HCV/IDU-positive biochemical markers of liver injury. The final adjusted model of factors independently associated with ≥ 95% adherence used a forward stepwise multivariate logistic regression model fit by including HCV/IDU-positive biochemical (continuous) markers of liver injury plus baseline sociodemographic and clinical characteristics considered potential sources of confounding (gender, age, baseline CD4 cell count and whether they had an AIDS diagnosis at baseline).
Separate modelling was conducted using the global measure of hepatic injury in order to create an interaction term between HCV coinfection and hepatic injury based on the hypothesis that, while those with HCV infection would be less likely to adhere to treatment, this would be most strongly associated with the combined effect of HCV infection and liver injury.
To elucidate the differential effects of IDU and HCV, a four-level variable was constructed: IDU and HCV negative, 0; IDU positive but HCV negative, 1; IDU negative but HCV positive, 2; and both IDU and HCV positive, 3. This variable was described using Pearson's X2 statistic. These categories were also used to create separate variables, each with no IDU and no HCV as the reference, and entered into multivariate logistic regression models. Pearson's correlation coefficient was used to assess correlation between the HCV and IDU variables.
|
|
| |
| |
|
|
|
