The effect of antiretroviral therapy on liver disease among adults with HIV and hepatitis C coinfection

Authors


  • Conflict of interest: Nothing to report.

Abstract

In the era of antiretroviral therapy (ART), liver disease has emerged as an important cause of death among persons with human immunodeficiency virus (HIV)/hepatitis C virus (HCV) coinfection. The objective of this study was to estimate the burden of liver disease and evaluate determinants of liver fibrosis and necroinflammatory activity among HIV/HCV coinfected patients receiving ART. We studied 112 randomly selected and 98 referred HCV-infected patients undergoing care in the Johns Hopkins University HIV clinic. Liver disease was characterized clinically and histologically. Of the 210 individuals studied—64% of whom had received ART within 2 years of liver disease assessment—33% had no fibrosis (F0), and 26% had bridging fibrosis or cirrhosis (≥F3). The median necroinflammatory activity score was 3 (range, 0-9 of 18). ART was not associated with fibrosis; however, significantly less hepatic necroinflammatory activity was observed among persons who had received highly active antiretroviral therapy longer (P = .02) and more effectively (defined by HIV RNA suppression; P < .01). Twelve percent of individuals had previous ART-associated liver enzyme elevations (grades 3-4), but liver fibrosis was not more severe if the liver enzyme elevation resolved. On the other hand, liver fibrosis was more severe in persons with persistent liver enzyme elevations (grades 1-4). In conclusion, despite widespread exposure to ART and documented instances of ART-related hepatitis, we found no evidence that ART caused serious histological liver disease. Recognition of bridging fibrosis and cirrhosis in some but not most patients underscores the importance of identifying and treating liver disease in HIV/HCV coinfected persons. (HEPATOLOGY 2005;41:123–131.)

As a result of shared routes of transmission, hepatitis C virus (HCV) and human immunodeficiency virus (HIV) coinfection is common in the United States, affecting 15% to 30% of HIV-infected persons.1, 2 In most studies, coinfection with HIV has been associated with accelerated progression of HCV-related liver disease.3 Moreover, antiretroviral therapy (ART) has led to a marked decline in most opportunistic illnesses, and HCV infection has emerged as an important cause of morbidity and mortality in HIV-infected persons.4–6 Indeed, data from nationwide surveillance of death certificates in the United States indicate that the proportion of liver-related deaths increased three- to four-fold in HIV-infected persons in the last decade. In some settings, complications of HCV-related liver disease have been reported to be the leading cause of hospitalization and death among HIV-infected persons.6, 7

Although the impact of ART on HIV disease has been remarkable, the impact on liver disease in coinfected patients remains unknown. Although it is possible that ART can attenuate liver disease progression through the reversal or prevention of HIV-related immunosuppression, it is also plausible that antiretroviral use may exacerbate liver disease. In fact, significant liver enzyme elevations (grade 3 or 4 hepatotoxicity) are observed in approximately 5% to 10% of persons taking a new antiretroviral regimen. Furthermore, the incidence of ART-associated hepatoxicity is approximately threefold greater in HIV/HCV-coinfected persons than in persons without hepatitis C.8, 9 Although many liver enzyme elevations resolve even when ART is continued,10 the effect of ART use on the progression of HCV-related liver disease is uncertain.

Hepatologists and HIV care providers are increasingly faced with managing liver disease in HIV/HCV-coinfected patients receiving ART. However, data regarding the prevalence of significant liver disease and its relationship to ART are needed. The objectives of this study were to estimate the burden of liver disease among HIV/HCV-coinfected persons receiving ART and to evaluate if liver disease is associated with factors that we could identify (and possibly modify), such as ART-related liver enzyme elevations (LEEs).

Abbreviations

ART, antiretroviral therapy; HIV, human immunodeficiency virus; HCV, hepatitis C virus; LEE, liver enzyme elevation; ALT, alanine aminotransferase; AST, aspartate aminotransferase; MHAI, modified histological activity index; ULN, upper limit of normal; IQR, interquartile range; HAART, highly active antiretroviral therapy; AOR, adjusted odds ratio.

Patients and Methods

Study Population and Sample Selection.

The population for this study derives from HIV/HCV coinfected members of the Johns Hopkins University HIV clinic cohort.11 To estimate the burden of liver disease in this HIV cohort, we evaluated persons who had liver biopsies performed between April 2001 and July 2002. To ensure that our estimate of the burden of disease was not biased, we randomly selected 137 of the 630 HIV/HCV cohort members who had received ART for 2 or more years and had not yet received treatment for HCV infection. Of the 137 individuals screened, 25 were excluded because of the following reasons: undetectable HCV RNA (n = 5), end-stage liver disease (n = 3), medical contraindications to liver biopsy (n = 8), died before biopsy (n = 1), lost to follow-up (n = 5), and other (n = 3), leaving 112 individuals in what we refer to as the random cohort. These patients were not substantially different from other eligible individuals who were not randomly selected with respect to age, sex, alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels, CD4 cell count, and HIV RNA level (P > .05; data not shown). However, patients in the random sample were slightly more likely than the other eligible individuals to be African American and to have a history of alcohol abuse (P < .05).

In addition to the 112 persons in the random cohort, another 98 HIV cohort members (clinical care cohort) had liver biopsies performed during the same time interval. As with the random cohort, these 98 were coinfected with HIV and HCV and had not received treatment for HCV before the biopsy. Compared with the 112 persons in the random sample, those in the clinical care cohort were similar with regard to age, sex, liver fibrosis, and liver activity but were less likely to have abused alcohol, had less exposure to ART, and had higher ALT levels at the time of biopsy (data not shown). Although analyses were all performed separately in each group, no differences were detected, and the data are presented in aggregate as described in Results.

For all 210 HIV/HCV-coinfected persons who had undergone liver biopsy, information on prescribed medications and laboratory parameters preceding the liver biopsy was obtained from the Johns Hopkins University clinical and laboratory database. As described previously, data on patient demographics, social practices, clinical and laboratory parameters, and prescribed antiretroviral and other medications are collected and abstracted from patient charts by trained personnel and transferred electronically from the Johns Hopkins University laboratory database at enrollment and subsequent 6-month intervals.11 Validity checks are performed periodically on a 5% sample of abstracted fields in the database; to date, errors have only been found in 0.2% of variable fields. Designation of current (or past) injection drug use and alcohol abuse was based on physician diagnosis; in addition, patients in the random cohort underwent an audio computer-assisted self interview to obtain information on alcohol use via the validated Alcohol Use Disorders Identification Test scale. This 10-item screening scale—which includes subscales on dependence symptoms and problems caused by harmful alcohol use and hazardous alcohol use—was developed by the World Health Organization for cross-cultural use in primary health settings to identify a broad spectrum of hazardous and harmful alcohol users and to distinguish them from hospitalized alcoholics and normal drinkers.12

The study was approved by the Johns Hopkins University Joint Committee on Clinical Investigation, and written informed consent was obtained for all participants.

Laboratory Evaluations.

Patients had standard laboratory assessments at each clinic visit performed by licensed clinical laboratories, including a complete blood cell count, serum chemistry panels, ALT level, AST level, CD4 cell count, and plasma HIV RNA level (reverse-transcriptase polymerase chain reaction). HCV testing was routinely performed on all cohort participants by a licensed clinical laboratory using a second- or third-generation enzyme immunoassay (EIA 2.0, Abbott Laboratories, Abbott Park, IL; EIA 3.0, Ortho Diagnostics, Raritan, NJ) and confirmatory HCV RNA testing (COBAS AMPLICOR MONITOR assay, Roche Diagnostic Systems). HCV genotype testing was performed using reverse-transcriptase polymerase chain reaction. Individuals in the random sample also had quantitative testing for hepatitis B virus DNA via polymerase chain reaction assay (lower limit of detection 200 copies/mL).

A transcutaneous liver biopsy was performed using an 18-gauge needle. Liver tissue was then fixed in 10% formalin, and paraffin-embedded sections were stained with hematoxylin-eosin and trichrome stains. Slides were evaluated by a single pathologist (M.T.), deemed adequate based on specimen size and number of portal tracts, and scored according to the Ishak modified histological activity index (MHAI) scoring system.13 For fibrosis stage, biopsies were also evaluated according to the METAVIR system (0 [no fibrosis] to 4 [cirrhosis]).14 All individuals with a MHAI grade of F3 or higher also had a METAVIR score of 2 or higher, so METAVIR scores were not analyzed further.

Statistical Analysis.

Univariate and multivariate logistic regression analysis were used to assess nonhistological and histological determinants (in separate models) of bridging fibrosis or cirrhosis (defined as MHAI ≥ F3). Similarly, we also analyzed determinants of significant necroinflammatory activity (MHAI ≥ 5; i.e., the 75th percentile). Initially, correlates of F3 fibrosis or higher and activity were examined separately in the random (n = 112) and clinical care cohort samples (n = 98). The Breslow-Day test for homogeneity of odds ratios across strata was used to determine whether odds of F3 fibrosis or higher and activity in the random sample were equal to those in the clinical care sample. There was no difference in the odds of fibrosis and activity for any of the variables used in the analysis, so final models for fibrosis and necroinflammatory activity included a combined sample of individuals from both groups. For multivariate models, variables that were associated with fibrosis and necroinflammatory activity in the univariate analysis with a P value less than .15 were considered in multivariate models after assessment of multicollinearity by variance inflation factors and tolerance. Variables were entered into multivariate models in a backward stepwise fashion. Those that were significant in the multivariate models with a P value less than .05 were retained in the final model.

Factors considered in analysis included demographics, alcohol abuse (clinician-diagnosed and self-reported via audio computer-assisted self interview), duration of HCV infection (estimated as the amount of time since first injection drug use), past and current ART, liver enzymes, HIV RNA level, CD4 cell count, necroinflammatory activity, and liver steatosis. Furthermore, to capture the longitudinal experience of each patient, we analyzed ALT and AST levels, HIV RNA levels and CD4 cell count both at the time of biopsy and also as cumulative variables of the longitudinal exposure data. HIV RNA and CD4 cell count were analyzed as the proportion of HIV RNA measurements that were undetectable and the proportion of CD4 cell counts that were less than 200 cells/μL. ALT and AST were examined as the cumulative proportion of ALT and AST levels more than 2.5 times the upper limit of normal (ULN) (ALT > 93 U/L and AST > 100 U/L). This was further dichotomized as 33% (1 in 3) of ALT and AST measurements more than 2.5 times the ULN (75th percentile of the data) versus those with 33% or less of their values being more than 2.5 times the ULN.

We also examined the prevalence of prior episodes of ART-associated LEEs and their association with histological evidence of bridging fibrosis or cirrhosis and necroinflammatory activity on subsequent liver biopsy. ART-associated LEEs were defined as a grade 3 or 4 serum ALT or AST level elevation (5-10 and >10 times the ULN) within 180 days of starting a new antiretroviral regimen as described previously.8 Analyses were performed using SAS version 8.12 software (SAS Institute, Cary, NC).

Results

Description of Study Population.

The demographic and clinical characteristics of the study population (n = 210) are described in Table 1. The median age was 44.5 years; 67% were male, 85% were black; and 77% reported ongoing or prior injection drug use. Although few reported current alcohol use on computer-assisted interview, nearly 40% had a clinical diagnosis of antecedent alcohol abuse. The majority (64%) were receiving ART at the time of liver biopsy, and few were completely naive to ART at the time of biopsy (19%). Twelve percent had a previous episode of grade 3 or 4 ART-associated LEE. The median follow-up time since entry into HIV care was 5 years (interquartile range [IQR], 2.9-7.5 years). For the 91 individuals in the random cohort who acknowledged injection drug use, the estimated median duration of HCV infection (estimated as the amount of time since the first injection) was 23.5 years (IQR, 20.6-29.0). At the time of biopsy, hepatitis B virus DNA was assessed in all 112 members of the random cohort but was detected in only 1 individual. For 93 of the 112 individuals with adequate sera, median HCV RNA levels were 3.7 × 106 IU/mL (IQR, 1.1-5.9 × 106).

Table 1. Characteristics of 210 HIV/HCV-Coinfected Members of the Johns Hopkins HIV Clinic Who Received Liver Biopsies Between April 2001 and July 2002
Characteristic*Study Population (N = 210)
  • NOTE. Values are expressed as the median (interquartile range) for continuous variables and n (%) for categorical variables.

  • Abbreviations: NRTI, nucleoside reverse-transcriptase inhibitor; NNRTI, nonnucleoside reverse-transcriptase inhibitor; PI, protease inhibitor; HAART, highly active antiretroviral therapy.

  • *

    At time of biopsy unless indicated otherwise.

Median age (yr)44.5 (40.0–47.8)
Female sex69 (32.9)
African American177 (84.7)
Injection drug use161 (76.7)
History of alcohol abuse (physician diagnosis)83 (39.5)
Median ALT (U/L)47 (29–82)
Median AST (U/L)52 (33–79)
Median CD4 cell count (cells/μL)366 (219–528)
Median HIV RNA (copies/mL)263 (42–126)
CD4 nadir (cells/μL)139 (29–294)
Current ART (%) 
 None75 (35.7)
 NRTIs only21 (10.0)
 NNRTI/PI + NRTIs114 (54.3)
Cumulative PI exposure (yr)2.0 (0–4.4)
Cumulative NNRTI exposure (yr)0.3 (0–1.7)
Cumulative NRTI exposure (yr)4.8 (1.9–6.9)
Cumulative HAART exposure (yr)4.0 (0.8–5.3)

Prevalence of Hepatic Fibrosis and Activity.

No hepatic fibrosis (F0) was detected in 33% (n = 69) of individuals, and 41% (n = 86) had fibrosis restricted to the portal tracts (F1-F2) (Fig. 1A). Bridging fibrosis (F3-F4) or cirrhosis (F5-F6) was noted in 9% (n = 19) and 17% (n = 36) of persons, respectively. Of the 210 individuals, 208 had complete activity scoring. Overall, necroinflammatory activity was relatively mild; 39% had no periportal or periseptal interface hepatitis, none had confluent necrosis, and 26% had no apoptosis and only focal inflammation (Fig. 1B). Sixty-three percent had a portal inflammation score of 1. The median composite necroinflammatory activity score was 3 (IQR, 2-5), and 58 (28%) had activity scores of 5 or higher.

Figure 1.

(A) MHAI fibrosis staging and (B) necroinflammatory activity grading for 210 HIV/HCV-coinfected patients undergoing HIV care. *Actual observations that are outside the interquartile range of the variable.

Determinants of Liver Fibrosis.

In univariate analysis, presence of bridging fibrosis or cirrhosis (≥F3) was significantly associated with alcohol abuse, female sex, elevated ALT and/or AST levels at the time of biopsy, persistently elevated ALT and/or AST levels prior to the biopsy (>33% of AST or ALT measurements more than 2.5 times the ULN), histological evidence of necroinflammatory activity (MHAI ≥ 5), and steatosis (P < .05) and was marginally associated with age of more than 50 years (P = .10) (Table 2). Notably, the presence of F3 fibrosis or higher was not associated with race, current or cumulative highly active antiretroviral therapy (HAART) exposure, current or cumulative HIV RNA level, or current or cumulative CD4 cell count. In multiple logistic regression analysis, independent nonhistological determinants of bridging fibrosis or cirrhosis were age of more than 50 years (adjusted odds ratio [AOR], 3.0; 95% CI, 1.1-8.1), female sex (AOR, 2.8; 95% CI, 1.4-5.8), alcohol abuse (AOR, 3.1; 95% CI, 1.5-6.3), and persistently elevated ALT levels before biopsy (AOR, 5.0 for >33% of prebiopsy ALT values more than 2.5 times the ULN or 93 U/L; 95% CI, 2.3-10.5) (Table 2). In addition, a hepatic necroinflammatory activity score of 5 or more (AOR, 4.3; 95% CI, 2.2-8.7) and any hepatic steatosis (AOR, 2.5; 95% CI, 1.2-4.9) were independent histological determinants of liver fibrosis.

Table 2. Determinants of Bridging Fibrosis or Cirrhosis (≥F3) Among 210 Study Subjects
Determinants*Odds Ratio (95% CI)Adjusted Odds Ratio (95% CI)
  • NOTE. Results are from logistic regressions. Separate models were used for nonhistological and histological determinants. Odds ratios and 95% CI are reported for the unadjusted logistic regression; adjusted odds ratios and 95% CI are reported for the multiple logistic regression.

  • Abbreviations: NRTI, nucleoside reverse-transcriptase inhibitor; PI, protease inhibitor; NNRTI, nonnucleoside reverse-transcriptase inhibitor.

  • *

    At biopsy unless indicated otherwise.

Nonhistological  
 Random cohort versus clinical care cohort1.3 (0.68–2.3) 
 Age >50 years1.9 (0.82–4.6)3.0 (1.1–8.1)
 Female sex2.3 (1.2–4.4)2.8 (1.4–5.8)
 African American versus Caucasian0.61 (0.27–1.4) 
 Injection drug use0.66 (0.33–1.3) 
 Alcohol abuse2.6 (1.4–4.8)3.1 (1.5–6.3)
 ALT (odds ratio per 50 U/L)1.4 (1.1–1.8) 
 AST (odds ratio per 50 U/L)1.5 (1.2–1.9) 
 AST/ALT ratio1.7 (0.9–3.3) 
 More than 33% of prebiopsy ALT values >2.5 × ULN3.8 (1.9–7.6)5.0 (2.3–10.5)
 More than 33% of prebiopsy AST values >2.5 × ULN5.5 (2.7–11.4) 
 CD4 cell count (cells/μL)  
  <2001.0 
  200–5000.96 (0.44–2.1) 
  >5001.0 (0.43–2.3) 
 CD4 nadir (cells/μL)  
  <501.0 
  50–2001.2 (0.53–2.6) 
  >2000.83 (0.40–1.7) 
 Median % CD4 <200 cells/μL (IQR)0.93 (0.38–2.3) 
 Detectable HIV RNA (>400 copies/mL)1.5 (0.81–2.8) 
 More than 75% of prebiopsy HIV RNA levels undetectable (≤400 copies/mL)1.6 (0.74–3.5) 
 ART  
  No ART1.0 
  NRTIs only0.61 (0.18–2.0) 
  PI/NNRTI + NRTIs0.92 (0.48–1.8) 
 Cumulative PI exposure (yr)1.04 (0.9–1.2) 
 Cumulative NNRTI exposure (yr)1.03 (0.84–1.3) 
 Cumulative NRTI exposure (yr)1.03 (0.94–1.1) 
 Cumulative HAART exposure (yr)1.01 (0.9–1.2) 
Histological  
 Necroinflammatory activity: MHAI ≥5 vs. MHAI <54.8 (2.5–9.3)4.3 (2.2–8.7)
 Steatosis versus none3.1 (1.6–5.9)2.5 (1.2–4.9)

Determinants of Necroinflammatory Activity.

In univariate analysis, elevated necroinflammatory activity score (>5) was associated with female sex, alcohol abuse, elevated ALT and/or AST levels at the time of biopsy, ratio of AST to ALT levels at biopsy, persistently elevated ALT and AST levels prior to biopsy, nadir CD4 cell count, detectable HIV RNA at biopsy, cumulative HIV suppression prior to biopsy, cumulative exposure to HAART, and receiving HAART at the time of biopsy (Table 3). Notably, neither age, race, current CD4 cell count, nor recent initiation of HAART (within the past 3 months) was associated with activity. In multiple logistic regression analysis, female sex (AOR, 3.5; 95% CI, 1.8-7.5), alcohol abuse (AOR, 2.6; 95% CI, 1.4-5.6), ALT level at the time of biopsy (odds ratio, 1.4 per 50 U/L; 95% CI, 1.1-1.9), lack of suppression of HIV replication at biopsy (AOR, 3.4; 95% CI, 1.6–7.1), and shorter cumulative HAART exposure (AOR per year of exposure, 0.8; 95% CI, 0.70-0.96) were independently associated with greater odds of activity. Because cumulative exposure to HAART and HIV replication were associated with decreased necroinflammatory activity, we further investigated these relationships. Compared with individuals who had active HIV replication at the time of biopsy, those with undetectable HIV RNA at the time of biopsy had significantly lower activity scores (P < .01) (Fig. 2). We failed to detect any association of specific antiretroviral drug or drug class (e.g., protease inhibitors) with hepatic activity grade.

Table 3. Determinants of Necroinflammatory Activity (MHAI ≥5) Among 208 Study Subjects
Determinants*Odds Ratio (95% CI)Adjusted Odds Ratio (95% CI)
  • NOTE. Results are from logistic regressions. Odds ratios and 95% CI are reported for the unadjusted logistic regression; adjusted odds ratios and 95% CI are reported for the multiple logistic regression.

  • Abbreviations: NRTI, nucleoside reverse-transcriptase inhibitor; PI, protease inhibitor; NNRTI, nonnucleoside reverse-transcriptase inhibitor.

  • *

    At biopsy unless indicated otherwise.

Clinical care cohort vs. random cohort1.0 (0.56–1.9) 
Age > 50 years0.75 (0.29–2.0) 
Female sex3.0 (1.6–5.7)3.5 (1.8–7.5)
African American versus Caucasian0.59 (0.27–1.3) 
Injection drug use1.3 (0.6–2.6) 
Alcohol abuse2.7 (1.5–5.1)2.6 (1.4–5.6)
ALT (odds ratio per 50 U/L)1.4 (1.1–1.7)1.4 (1.1–1.9)
AST (odds ratio per 50 U/L)1.4 (1.1–1.8) 
AST/ALT ratio2.2 (1.2–4.3) 
More than 33% of prebiopsy ALT values >2.5 × ULN2.8 (1.4–5.5) 
More than 33% of prebiopsy AST values >2.5 × ULN3.3 (1.6–6.6) 
CD4 cell count (cells/μL)  
 <2001.0 
 200–5000.79 (0.36–1.7) 
 >5001.3 (0.56–2.8) 
CD4 nadir (cells/μL)  
 <501.0 
 50–2002.8 (1.2–6.6) 
 >2002.3 (1.1–5.0) 
Median % CD4 <200 cells/μL (IQR)0.35 (0.13–0.9) 
Detectable HIV RNA (>400 copies/mL)3.8 (2.0–7.3)3.4 (1.6–7.1)
More than 75% of prebiopsy HIV RNA levels undetectable (≤400 copies/mL)0.3 (0.1–0.8) 
ART  
  No ART1.0 
  NRTIs only0.39 (0.1–1.2) 
  PI/NNRTI + NRTIs0.27 (0.14–0.5) 
Cumulative PI exposure (yr)0.84 (0.7–1.0) 
Cumulative NNRTI exposure (yr)0.85 (0.7–1.1) 
Cumulative NRTI exposure (yr)0.92 (0.84–1.0) 
Cumulative HAART exposure (yr)0.8 (0.7–0.9)0.8 (0.7–0.96)
Figure 2.

Necroinflammatory activity score by HIV viral load at the time of biopsy (undetectable [≤400 copies/mL] vs. detectable). *Actual observations that are outside the interquartile range of the variable.

Antiretroviral-Associated Hepatotoxicity.

Because we observed that persistent liver enzyme levels were associated with more severe liver disease, we evaluated the relationship of prior episodes of antiretroviral-associated LEE and subsequent liver histology. Prior to liver biopsy, grade 3 or 4 liver enzyme elevations were observed in 26 (12%) of 210 persons. Following the LEE, persons appeared to have one of two outcomes: for 12 persons (46%), liver enzymes remained elevated (defined as > 33% of post-LEE measurements more than 2.5 times the ULN) and liver enzyme levels substantially diminished for 14 (54%) (resolved LEE, defined as ≤ 33% of post-LEE ALT or AST measurements more than 2.5 times the ULN) (Fig. 3). Among those with resolved episodes, the median ALT or AST level before, during, and 6 months after the hepatotoxic event was 43 U/L, 291 U/L, and 72 U/L, respectively. In contrast, among those with persistent liver enzyme elevations after LEE, the median ALT or AST level, before, during, and 6 months after the initial elevations was 58 U/L, 318 U/L, and 122 U/L, respectively. The median time between LEE and liver biopsy was 2.5 years (IQR, 1.0-3.2) and the median ALT at the time of biopsy was 140 U/L for individuals with prolonged elevated enzymes; these values were 0.8 years (IQR, 0.6-2.3) and 36 U/L, respectively, for individuals with a self-limited event. Among the 26 individuals who had a LEE, the change from the pre-ART level to the LEE was 250 U/L (IQR, 201-425) for those who had persistent elevation after the LEE and 252 U/L (IQR, 164-312) for those who had resolved elevations. Of the individuals who had a LEE, 14 (54%) remained on the same regimen, 6 (23%) changed regimens, and 6 (23%) stopped their regimen without starting a new one. No subject underwent liver biopsy at the time of LEE to evaluate the cause of the ALT or AST elevation.

Figure 3.

Prebiopsy ALT levels for persons with minimal or no fibrosis (<F3) and persons with bridging fibrosis or cirrhosis (≥F3). ALT, alanine aminotransferase.

We further explored the relationship between ART-associated LEE and fibrosis and activity. Compared with those who had neither LEE nor persistently elevated ALT or AST levels before the biopsy (defined as > 33% of prebiopsy ALT or AST measurements more than 2.5 times the ULN), those who experienced a resolved LEE were no more likely to have bridging fibrosis or cirrhosis (OR, 0.9; 95% CI, 0.2-4.6). In addition, while the odds of bridging fibrosis or cirrhosis were significantly greater among persons with LEE followed by persistent liver enzyme elevations (OR, 5.7; 95% CI, 1.7-19.5) compared with those with no persistent elevations, these odds were similar to that of persons with persistent liver enzyme elevations without a history of LEE (OR, 5.5; 95% CI, 2.7-11.1). Additionally, prior ART-associated LEE was not associated with greater necroinflammatory activity on subsequent liver biopsy in univariate or multivariate analysis (data not shown).

Discussion

Through comprehensive evaluation of a large sample of antiretroviral-experienced HIV/HCV-coinfected patients, this study provides an important estimate of the overall burden of liver disease in this setting. The findings are important in the HAART era not only because liver disease causes an increasing proportion of morbidity and mortality in HIV-infected persons,4–6 but also because there have been widely divergent expectations as to how HAART itself would impact liver disease. Although HAART may cause liver damage (i.e., hepatotoxicity),8, 9, 15 it is also tenable that HAART-induced immune restoration might attenuate the increased risk of cirrhosis associated with HIV coinfection16, 17; thus several of our findings are particularly noteworthy.

First, despite a relatively long duration of hepatitis C infection, one third of patients had no evidence of hepatic fibrosis, which is a higher proportion than has been observed in some other studies.16, 18, 19 Although there could be several explanations for this result, the lack of selection bias in studying patients in an HIV clinic (rather than a liver clinic) probably contributes, because many of these individuals had not been referred for management of hepatitis C. However, the remarkable range of severity of liver fibrosis among the patients is also notable. Although minimal liver disease was detected in a significant proportion of patients, more than 25% already had advanced disease with histological evidence of bridging fibrosis or cirrhosis. Likewise, significant hepatic inflammation was found in some (but not most) patients. Thus our data indicate that the range of HCV-related liver disease among HIV-infected persons treated with ART is broad, supporting recent recommendations that the management of hepatitis C in HIV-infected patients must be individualized.20

It is important to identify the HIV/HCV-coinfected patients who have significant liver disease. As expected, compared with those persons with less fibrosis, persons with bridging fibrosis or cirrhosis were older and more likely to have abused alcohol. We found that the predictive value of a single ALT or AST around the time of biopsy was limited, suggesting that a single measurement is not sufficient to predict disease stage.16, 18, 21, 22 However, persons with long-term patterns of elevated liver enzymes (having more than 1 in every 3 ALT or AST measurements > 100 U/L) had a more than five-fold greater risk of bridging fibrosis or cirrhosis compared with those persons with lower liver enzyme levels. Thus, although HIV/HCV-coinfected patients often have elevated liver enzyme levels, physicians caring for coinfected patients should not underestimate the significance of persistent ALT and/or AST elevations.

We failed to detect less or more severe hepatic fibrosis among persons treated with effective antiretroviral therapy. However, this finding is not unexpected. Although we prospectively assessed ART exposure for a period more than 7 years prior to the biopsy, our patients were exposed to HCV for an average of 23 years and to HAART for an average of only 3 years. Because liver fibrosis typically represents the net accumulation and breakdown of collagen over relatively long periods, the duration of HAART exposure (relative to HCV infection) may have been too brief to have had a substantial impact on liver fibrosis in our cohort.23 Nonetheless, this finding is different than some published cross-sectional studies. For example, Benhamou et al.16 reported that the prevalence of cirrhosis was lower in coinfected subjects treated with protease inhibitor-based HAART for an average of 14 months.16 Similarly, Feldman et al.24 found that the use of protease inhibitor–based HAART (mean exposure, 851 days) was associated with a lower risk of bridging fibrosis or cirrhosis, whereas the use of a specific antiretroviral drug, nevirapine (mean exposure, 527 days), was associated with a greater risk of disease. In addition to a relatively brief duration of exposure to HAART, there are a number of other factors like alcohol use that affect both liver fibrosis (increases) and HAART use (decreases), thereby complicating these inferences.25

Although longitudinal follow-up is needed to address the question of ART and fibrosis progression, we did find that greater effectiveness and longer duration of ART exposure were independently associated with decreased necroinflammatory activity, an index that probably is more temporally correlated with exposures than liver fibrosis. Too little is known about how HIV infection influences the pathogenesis of HCV-related liver disease to speculate on the basis for this finding. However, it is possible that the link between ART effectiveness and inflammation will eventually manifest as an association with fibrosis, because inflammation is believed to contribute to fibrosis and, in our study, persons with higher necroinflammatory scores were themselves 5-fold more likely to have bridging fibrosis or cirrhosis.26 If validated by prospective research, the finding that effective HAART attenuates the progression of liver disease would support expansion of the use of ART to prevent liver disease in HIV/HCV-coinfected persons, even those with relatively high CD4 cell counts.

We found no evidence for the alternate hypothesis that ART causes chronic histological liver disease. Rather, we found the long-term liver enzyme pattern to be a much stronger predictor of eventual liver fibrosis. These data support the finding of Qurishi et al.17 that ART significantly reduced long-term liver-related mortality among HCV/HIV-coinfected patients, despite the development of severe drug-related hepatotoxicity in approximately 14% of treated patients.17 Thus, although persistent ALT or AST elevations are an important marker of liver disease, our data suggest that ART-associated LEE does not increase the risk of advanced fibrosis, and that the increased risk of LEE among HCV-coinfected patients should not make physicians reluctant to prescribe ART.

Interestingly, after adjusting for potential confounders, we observed that women were more likely to have significant hepatic fibrosis and necroinflammatory activity compared with men. While male sex has been consistently associated with advanced fibrosis in HIV-uninfected patients, other studies of HIV/HCV-coinfected patients have also failed to identify male sex as an independent determinant of HCV-related liver disease.16, 18, 24 Although the reasons for the association of female sex with advanced fibrosis in the present study are not known, our data emphasize the need to evaluate hepatitis C in coinfected women.

In addition to the inherent limitations of assessing liver disease at a single point in time, our data are subject to several other limitations. First, because our sample was limited to patients who actually attended clinic visits, our findings may not be generalizable to HIV/HCV-coinfected persons not actively engaged in medical care. However, we have previously studied HIV/HCV-coinfected injection drug users in our community and observed similar prevalence of liver disease.27 Second, although we longitudinally measured exposures such as ART, liver enzymes, immunosuppression, and alcohol over a period of approximately 5 years before liver biopsy, we were unable to assess these exposures over most of the course of the HCV infection. Third, because the ART was prescribed according to national standards for the treatment of HIV disease, we were unable to evaluate liver histology among large numbers of antiretroviral naive patients. Moreover, we did not have sufficient power to examine the associations between individual antiretroviral drugs and fibrosis. Fourth, because the date of HIV seroconversion was unknown for most persons, we did not have information on the duration of HIV infection. Finally, a central weakness of all studies of this nature is in the assessment of alcohol use. Although we evaluated alcohol using the validated Alcohol Use Disorders Identification Test scale through a computer-assisted interview technique in a sample of our population, we found that the physician diagnosis of alcohol abuse was actually a stronger predictor of fibrosis than self-reported data.

In conclusion, in this setting bridging fibrosis or cirrhosis is found in some, but not most HIV/HCV-coinfected persons. Although ART use was not protective against liver fibrosis, individuals with longer cumulative exposure to HAART and HIV RNA suppression had significantly less necroinflammatory activity. These findings are consistent with the hypothesis that ART may attenuate HCV-related liver disease; however, prospective research is needed evaluate the effect of ART on liver fibrosis progression in HIV/HCV-coinfected patients and to identify noninvasive methods to identify individuals who are at greatest risk of cirrhosis.

Ancillary