Treatment Outcomes of Treatment-Naïve Hepatitis C Patients Co-Infected with HIV: A Systematic Review and Meta-Analysis of Observational Cohorts

Introduction Co-infection with Hepatitis C (HCV) and HIV is common and HIV accelerates hepatic disease progression due to HCV. However, access to HCV treatment is limited and success rates are generally poor. Methods We conducted a systematic review and meta-analysis to assess HCV treatment outcomes in observational cohorts. Two databases (Medline and EMBASE) were searched using a compound search strategy for cohort studies reporting HCV treatment outcomes (as determined by a sustained virological response, SVR) in HIV-positive patients initiating HCV treatment for the first time. Results 40 studies were included for review, providing outcomes on 5339 patients from 17 countries. The pooled proportion of patients achieving SVR was 38%. Significantly poorer outcomes were observed for patients infected with HCV genotypes 1 or 4 (pooled SVR 24.5%), compared to genotypes 2 or 3 (pooled SVR 59.8%). The pooled proportion of patients who discontinued treatment due to drug toxicities (reported by 33 studies) was low, at 4.3% (3.3–5.3%). Defaulting from treatment, reported by 33 studies, was also low (5.1%, 3.5–6.6%), as was on-treatment mortality (35 studies, 0.1% (0–0.2%)). Conclusions These results, reported under programmatic conditions, are comparable to those reported in randomised clinical trials, and show that although HCV treatment outcomes are generally poor in HIV co-infected patients, those infected with HCV genotypes 2 or 3 have outcomes comparable to HIV-negative patients.


Introduction
Co-infection with Hepatitis C (HCV) and HIV is common, and HIV accelerates hepatic disease progression due to HCV [1].As a result, HCV has become a leading cause of death of people living with HIV in Western settings [2].Successful treatment of HCV can improve hepatic fibrosis, reduce incidence of hepatocellular carcinoma, reduce mortality [3,4], and has the potential to reduce disease transmission [5].However, a number of factors contribute to the limited access to treatment for most of those infected globally: a long duration of therapy with a relatively complex system of treatment delivery, high drug costs, high toxicity of treatment and, perhaps most importantly, relatively poor success rates for HCV treatment in HIV/HCV co-infection.
A recent systematic review of clinical trials assessing HCV treatment outcomes in HIV co-infected patients reported that around 37% of patients achieve a sustained virological response (SVR) with pegylated interferon and ribavarin, with a lower success rate observed in patients infected with HCV genotypes 1 and 4 [6].These outcomes are poorer than those seen in HIV negative patients [7].Although clinical trials are appropriate for determining drug efficacy, outcomes may differ under programmatic conditions where adherence to treatment, patient and provider motivation and available resources may be limited [8].We conducted a systematic review to assess the outcomes of HCV treatment in HIV co-infected patients in programmatic settings.

Search Strategy and Study Selection
Our systematic review was conducted in accordance with the criteria of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses group [9].Using a pre-defined protocol (File S1) Medline and EMBASE were systematically searched from inception to 05 May, 2012 using a compound search strategy.The initial title screen was conducted by one of us (AD) with full text articles reviewed in duplicate (AD, NF).The bibliographies of relevant articles were also hand searched for potentially relevant articles.Agreement on inclusion of final articles was made through consensus by the same reviewers.No language or geographical restriction was applied during the search, but only English language publications were included in the final review.
All cohort studies that reported treatment outcomes for in HIVpositive patients chronically infected with HCV and initiating HCV treatment for the first time were reviewed.Studies were excluded if they reported outcomes among patients with selected co-morbidities other than HIV, such as haemophilic or transplant patients, and if treatment outcomes involved acute HCV infection.Randomised trials were excluded in keeping with the aim of assessing outcomes in programmatic settings (defined as cohort reports in health care settings where there was no randomisation or control group comparison).In cases of potential duplication of studies, the largest report covering the longest time period was included and authors were contacted for clarification.
Patient and study characteristics were extracted in duplicate (AD, KS), with third party arbitration in case of disagreement (NF).The primary outcome was the proportion of patients achieving a SVR, calculated on an 'intent-to-treat' basis with all patients starting treatment contributing to the denominator.Secondary outcomes included the proportion of patients achieving a rapid virological response (RVR), defined as an undetectable (,50 copies/mL) serum level of HCV RNA at week 4 of treatment; discontinuation of treatment due to adverse drug reactions; loss to care (default); and death.

Data Analysis
Point estimates and 95% confidence intervals (95% CI) were calculated for all primary and secondary outcomes.The variance of raw proportions was stabilised using a Freeman-Tukey type arcsine square-root transformation [10] and proportions were then pooled using a DerSimonian and Laird random effects model [11].We calculated the t 2 statistic using DerSimonian and Laird's method of moments estimator [11] to assess between-study heterogeneity [12].Sources of heterogeneity were explored through univariate subgroup analyses to assess the potential influence of baseline liver damage, genotype, type of HCV treatment and co-treatment with highly-active antiretroviral therapy (HAART).All analyses were conducted using Stata version 12 (StataCorp LP, College Station, Texas, USA), with a Pvalue #0.05 considered as significant.

Results
887 articles were screened, and 103 of these were reviewed in full (figure 1).After identification of further papers which did not meet the inclusion criteria (e.g.studies that included retreated patients or studies that did not report treatment outcomes in full), we retained 77 studies for detailed review.Over half of these studies (37) were from Spain, and after correspondence with authors, 37 studies were excluded as partial or complete duplicate cohorts .The final analysis included data on 5339 patients from 40 studies in 17 countries (Table 1).
The proportion of patients with liver damage at baseline ranged from 12.5% to 74%.The majority of studies (36) included a mix of HCV genotypes.Three studies (from Argentina, Spain and the USA) were exclusively comprised of patients infected with genotypes 1 and 4 and two studies (from Sweden and Spain) were exclusively comprised of patients infected with genotypes 2 and 3.
HCV treatment comprised pegylated interferon and weightbased ribavarin in most cases, and the majority of patients (84%) received concomitant antiretroviral therapy.Liver damage was assessed by biopsy in over half (25) of studies.One study used fibroscan to assess liver damage, and 3 studies used a combination of the 2 techniques.Nine studies did not assess liver damage while the remainder of the studies (3) did not state the method used.

Discussion
Currently, access to effective HCV treatment is limited, particularly for those with HCV/HIV co-infection in resourcelimited settings.This is reflected in this study by the paucity of data reoprted from such settings.Among the 40 studies assessed, only three were from resource-limited settings (two from Brazil and one from Argentina), and no reports were found from African countries, including Egypt where the burden of HCV is the highest in the world, or sub-Saharan Africa where the burden of HIV is the highest in the world.Limited access to treatment in resource-limited settings is in part due to the high cost of treatment, a perception of poorer outcomes of HCV treatment in HIV co-infected patients, and the potential difficulties associated with adherence and drug interactions under programmatic conditions.
Concern has recently been expressed that the relatively high efficacy of hepatitis treatment reported in clinical trials is not attained in subsequent effectiveness studies carried out in the general population under programmatic conditions [54].In comparison to routine programmes, patients in clinical trials tend to be more adherent to treatment, and will usually receive treatment free of charge provided by highly motivated clinical staff working in optimal clinical settings [55].Nevertheless, this review found that programmatic outcomes were in very close alignment to a systematic review of outcomes in clinical trials, which found that HCV treatment responses in HIV co-infected patients is lower than those observed in HIV-negative individuals [56].Nevertheless, for HIV-positive patients infected with HCV genotypes 2 or 3, treatment outcomes are very similar (SVR 60%) to those reported for HIV-negative HCV patients infected with the same genotypes in programme settings (SVR 59%) [7].
Treatment completion was generally high, with few patients discontinuing treatment due to adverse events or defaulting from care.The use of HAART was not associated with better outcomes, which is consistent with other studies [57,58].
We used a broad search strategy that allowed the inclusion of a large number of studies.We restricted studies to observational cohorts so that the expected outcomes would better reflect those observed in programmatic settings, but this can result in confounding.Concomitant use of medications, unreported mental or physical problems, or ancillary health service support could all influence treatment outcomes, but these factors were not reported and so could not be assessed.We attempted to use multivariate meta-regression to explore the potential influence of patient and programme level variables to explain differences in results between studies.However, this was restricted by inconsistent reporting between studies, so our exploration of associations was limited to univariate subgroup comparisons.In addition, bias may result from studies that pre-selected patients on the basis of characteristics that may influence treatment success, or excluded patients with risk factors for poor adherence.Furthermore, the final analysis only included studies published in English, which may lead to publication bias.Only five studies, however, were excluded  on the basis of language and their influence would likely be small.Nevertheless, this review should be taken as an indication of outcomes and not as an exhaustive summary.The treatment of HCV infection is likely to evolve rapidly as a result of a dynamic drug pipeline.For example, the first HCV protease inhibitors have just been recently approved.In the short to medium-term, however, the majority of HIV-positive patients living in resource-limited settings are unlikely to benefit from these newer treatments, just as they continue to lack access to many of the newer antiretroviral drugs for HIV that have been marketed in the West for many years.The results of this systematic review support the current practice of treatment in well resourced settings, whilst serving as a reminder for the need for better treatments.This review also highlights the need to encourage treatment of HCV/HIV co-infected patients in resource-limited settings to start programmes in parallel to efforts aimed at reducing costs of current treatment and gaining access to newer, interferon free regimens so that new advances in treatment can be rapidly accessed by all those that need them.

Table 1 .
Characteristics of included studies.