A Sustained Viral Response Is Associated With Reduced Liver-Related Morbidity and Mortality in Patients With Hepatitis C Virus:

A Sustained Viral Response Is Associated With Reduced Liver-Related Morbidity and Mortality in Patients With Hepatitis C Virus: risk of these outcomes is reduced by 70%–80% for patients who achieve SVR- mortality, hepatic decompensation, HCC - pdf attached


Clinical Gastroenterology & Hepatology March 2010

Amit G. Singal, Michael L. Volk, Donald Jensen‡, Adrian M. Di Bisceglie§, Philip S. Schoenfeld∥

Division of Gastroenterology, University of Michigan Medical Center, Ann Arbor, Michigan

‡ Division of Gastroenterology-Hepatology Section, University of Chicago School of Medicine, Chicago, Illinois

§ Division of Gastroenterology and Hepatology, St Louis University School of Medicine, St Louis, Missouri

∥ VA Center of Excellence in Health Services Research, Ann Arbor, Michigan

published online 30 November 2009.


Also at AASLD similar finding from HALT-C was presented in an oral talk:
SVR Improves Survival, Risk for Liver Cancer, Decompensated Liver Disease and Liver Transplant/Death - Also, Transient Viral Suppression (breakthroughs/relapsers) Improves Clinical Outcomes Too - (11/09/09)

"On the basis of our meta-analyses, SVR is associated with a significant risk reduction in liver-related mortality, HCC, and hepatic decompensation in patients with advanced fibrosis or cirrhosis.........Our meta-analysis also demonstrated that relapsers were significantly less likely to develop HCC compared with nonresponders (patients who never cleared viremia during treatment) (RR, 2.30; 95% CI, 1.26–4.19.....These data will facilitate decision-making by HCV-infected treatment failure patients and physicians when considering whether to re-treat with currently available therapies or to delay treatment until newer therapies are available.....Physicians and HCV-infected patients should understand the likelihood of developing decompensated cirrhosis, HCC, and liver-related mortality after failing therapy to make appropriate decisions about whether to undergo retreatment.....more advanced fibrosis is associated with higher rates of liver-related morbidity and mortality.36 Results should also be reported separately for patients who achieve SVR because achieving SVR is associated with a large decrease in liver-related morbidity and mortality. Our data demonstrate that rates of liver-related mortality (2.73%/year; 95% CI, 1.38–4.08), HCC (3.22%/year, 95% CI, 2.02–4.42), and hepatic decompensation (2.92%/year; 95% CI, 1.61–4.22) are substantial among HCV treatment failures with advanced fibrosis, but the risk of these outcomes is reduced by 70%–80% for patients who achieve SVR....Chronic hepatitis C patients with SVR are much less likely to suffer liver-related mortality compared with chronic hepatitis C patients with treatment failure who do not achieve SVR in studies with HCV patients with all stages of fibrosis (RR, 0.23; 95% CI, 0.10–0.52) (Figure 1C) or studies that only enrolled patients with advanced fibrosis (RR, 0.19; 95% CI, 0.10–0.37)....lso, chronic hepatitis C patients with SVR are much less likely to suffer HCC compared with chronic hepatitis C patients who do not achieve SVR among studies that enrolled patients with any stage of hepatic fibrosis (RR, 0.21; 95% CI, 0.16–0.27) (Figure 2C) and among studies that only enrolled patients with advanced fibrosis or cirrhosis (RR, 0.32; 95% CI, 0.23–0.44)....Chronic hepatitis C patients with SVR are much less likely to suffer liver-related mortality compared with chronic hepatitis C patients with treatment failure who do not achieve SVR in studies with HCV patients with all stages of fibrosis (RR, 0.23; 95% CI, 0.10–0.52) (Figure 1C) or studies that only enrolled patients with advanced fibrosis (RR, 0.19; 95% CI, 0.10–0.37)....lso, chronic hepatitis C patients with SVR are much less likely to suffer HCC compared with chronic hepatitis C patients who do not achieve SVR among studies that enrolled patients with any stage of hepatic fibrosis (RR, 0.21; 95% CI, 0.16–0.27) (Figure 2C) and among studies that only enrolled patients with advanced fibrosis or cirrhosis (RR, 0.32; 95% CI, 0.23–0.44) ....The prevalence of HCV-infected patients with advanced fibrosis in the US is likely to rise precipitously in the near future, producing a corresponding increase in liver-related morbidity and mortality.....In the US, the prevalence of HCV infection in 50- to 59-year-old individuals will be approximately 4% by 2012, and the majority of these patients will have been infected with HCV for 20–30+ years.2.....SVR stops progression of fibrosis or because it leads to regression of fibrosis. Mallet et al32 assessed 96 HCV-infected patients with cirrhosis (Metavir F4) who were treated with interferon-based regimens and had at least 1 post-treatment liver biopsy. Thirty-nine of these patients achieved SVR, and 44% (17/39) of these patients demonstrated cirrhosis regression, defined as a decrease in Metavir score of >2. Although none of the patients with SVR and cirrhosis regression suffered from liver-related morbidity or mortality during follow-up, 18% (4/22) of patients with SVR without cirrhosis regression suffered liver-related morbidity or mortality (3 cases of HCC and 1 variceal bleed). Therefore, future prospective observational studies might consider performance of post-treatment liver biopsy to enhance our understanding of the importance of regression of fibrosis on the natural history of these patients."

ABSTRACT

Background & Aims


The incidences of decompensated cirrhosis (defined by ascites, hepatic encephalopathy, or bleeding esophageal varices), hepatocellular carcinoma (HCC), and liver-related mortality among patients infected with hepatitis C virus (HCV) who achieve a sustained viral response (SVR), compared with patients who fail treatment (treatment failure), are unclear. We performed a meta-analysis to quantify the incidences of these outcomes.

Methods


This meta-analysis included observational cohort studies that followed HCV treatment failure patients; data were collected regarding the incidence of decompensated cirrhosis, HCC, or liver-related mortality and stratified by SVR status. Two investigators independently extracted data on patient populations, study methods, and results by using standardized forms. The agreement between investigators in data extraction was greater than 95%. Data analysis was performed separately for studies that enrolled only HCV patients with advanced fibrosis.

Results


We identified 26 appropriate studies for meta-analysis. Among treatment failure patients with advanced fibrosis, rates of liver-related mortality (2.73%/year; 95% confidence interval [CI], 1.38–4.080), HCC (3.22%/year, 95% CI, 2.02–4.42), and hepatic decompensation (2.92%/year; 95% CI, 1.61–4.22) were substantial. Patients with SVR are significantly less likely than patients who experienced treatment failure to develop liver-related mortality (relative risk [RR], 0.23; 95% CI, 0.10–0.52), HCC (RR, 0.21; 95% CI, 0.16–0.27), or hepatic decompensation (RR, 0.16; 95% CI, 0.04–0.59).

Conclusions


HCV patients with advanced fibrosis who do not undergo an SVR have substantial liver-related morbidity and mortality. Achieving SVR is associated with substantially lower liver-related morbidity and mortality.


Abbreviations used in this paper: CI, confidence interval, HCC, hepatocellular carcinoma, HCV, hepatitis C virus, RR, relative risk, SVR, sustained viral response.

Approximately 3% of the world's population is infected with hepatitis C virus (HCV),1 including approximately 3.2 million people in the United States.2 HCV is a leading cause of decompensated cirrhosis, hepatocellular carcinoma (HCC), liver-related mortality, and liver transplantation in the US Management of these disorders costs the US health care system more than a billion dollars per year.3 The long incubation time between HCV infection and development of advanced fibrosis suggests that liver-related morbidity and mortality rates will continue to increase despite the declining incidence of new HCV infections.2 In the US, HCV-related mortality is projected to peak in 2030 at approximately 13,000 deaths/year.4


However, many HCV-related deaths are preventable if HCV-infected patients are treated successfully. Among HCV patients with SVR, rates of liver-related mortality, HCC, and hepatic decompensation are very low, and life expectancy might be equivalent to the age-matched and sex-matched general population.5 Unfortunately, more than 50% of HCV patients are treatment failures with pegylated interferon and ribavirin antiviral therapy.6


The natural history of these “treatment failure” patients has not been studied extensively, although patients and physicians need to understand the natural history of these patients as they weigh the potential benefits of antiviral retreatment versus the risks of delaying therapy while awaiting new treatments. The natural history of liver-related morbidity and mortality might need to be assessed separately in HCV treatment failure patients versus HCV patients who have never been treated. HCV-infected patients with treatment failure might be more likely to have significant fibrosis compared with HCV patients who have never been treated because physicians are more likely to recommend antiviral therapy for HCV patients who are at the greatest risk for developing liver-related morbidity and mortality in the near future.7 Alternatively, other patients with significant fibrosis might remain untreated, given comorbid conditions or issues with noncompliance. Although not well studied to date, one cannot assume that the natural histories of these 2 groups are the same.


No previous systematic review has quantitatively assessed the natural history of HCV-infected patients with treatment failure or the reduction in decompensated cirrhosis, HCC, and liver-related mortality for HCV-infected patients who achieve sustained viral response (SVR) versus HCV-infected patients with treatment failure. The primary objective of our meta-analysis is to quantify annual rates of decompensated cirrhosis, HCC, and liver-related mortality among HCV treatment failure patients. We also quantify the risk reduction in decompensated cirrhosis, HCC, and liver-related mortality for HCV-infected patients with SVR versus HCV treatment failure patients.

Discussion



Physicians and HCV-infected patients should understand the likelihood of developing decompensated cirrhosis, HCC, and liver-related mortality after failing therapy to make appropriate decisions about whether to undergo retreatment. Prognostic data should be stratified on the basis of severity of hepatic fibrosis. Stratifying natural history data by stage of hepatic fibrosis is important because more advanced fibrosis is associated with higher rates of liver-related morbidity and mortality.36 Results should also be reported separately for patients who achieve SVR because achieving SVR is associated with a large decrease in liver-related morbidity and mortality. Our data demonstrate that rates of liver-related mortality (2.73%/year; 95% CI, 1.38–4.08), HCC (3.22%/year, 95% CI, 2.02–4.42), and hepatic decompensation (2.92%/year; 95% CI, 1.61–4.22) are substantial among HCV treatment failures with advanced fibrosis, but the risk of these outcomes is reduced by 70%–80% for patients who achieve SVR.


The prevalence of HCV-infected patients with advanced fibrosis in the US is likely to rise precipitously in the near future, producing a corresponding increase in liver-related morbidity and mortality. The incubation time between HCV infection and advanced fibrosis is probably 20–30+ years. Minimal fibrosis develops during the first 10 years of infection, but the rate of fibrosis development accelerates with each subsequent decade.37 After 30 years of infection, up to 30% of HCV-infected patients will have developed bridging fibrosis and liver-related morbidity and mortality.38 In the US, the prevalence of HCV infection in 50- to 59-year-old individuals will be approximately 4% by 2012, and the majority of these patients will have been infected with HCV for 20–30+ years.2


On the basis of our meta-analyses, SVR is associated with a significant risk reduction in liver-related mortality, HCC, and hepatic decompensation in patients with advanced fibrosis or cirrhosis. This might occur because SVR stops progression of fibrosis or because it leads to regression of fibrosis. Mallet et al32 assessed 96 HCV-infected patients with cirrhosis (Metavir F4) who were treated with interferon-based regimens and had at least 1 post-treatment liver biopsy. Thirty-nine of these patients achieved SVR, and 44% (17/39) of these patients demonstrated cirrhosis regression, defined as a decrease in Metavir score of >2. Although none of the patients with SVR and cirrhosis regression suffered from liver-related morbidity or mortality during follow-up, 18% (4/22) of patients with SVR without cirrhosis regression suffered liver-related morbidity or mortality (3 cases of HCC and 1 variceal bleed). Therefore, future prospective observational studies might consider performance of post-treatment liver biopsy to enhance our understanding of the importance of regression of fibrosis on the natural history of these patients.


Future prospective studies should also report their results by using Kaplan–Meier plots to enhance our understanding of these issues. Most studies in this meta-analysis did not report their results by using these plots, although limited data suggest that rates of decompensated cirrhosis, HCC, and liver-related mortality decreased virtually to zero for the first 5 years of follow-up after achieving SVR.30, 31, 32, 34, 35 Notably, rates of decompensated cirrhosis remained essentially zero during 5–10 years of follow-up of cirrhotic patients with SVR, although the number of patients available for follow-up was quite small.30, 31, 35


Our meta-analysis also demonstrated that relapsers were significantly less likely to develop HCC compared with nonresponders (patients who never cleared viremia during treatment) (RR, 2.30; 95% CI, 1.26–4.19) (Figure 4). The reason for this finding is unclear. Inadequate data are provided in the individual studies to determine whether there were important differences in fibrosis stage between relapsers and nonresponders. We also speculate on other reasons for this finding. Host-virus interactions resulting in relapse after treatment might be such that they also allow fewer clinical complications of chronic HCV infection. This finding could represent a state of more interferon “inducible” innate immunity or less induced suppressor genes (eg, USP-18). To clarify this finding, future prospective observational studies and randomized controlled trials should stratify their end point analyses between relapsers and nonresponders and seek to determine whether differences in liver-related morbidity and mortality are associated with differences in progression of fibrosis.


This meta-analysis is limited by several factors. First, studies in this meta-analysis did not consistently stratify outcomes by severity of hepatic fibrosis when HCV-infected patients failed treatment. Specifically, we do not have adequate data about the natural history of HCV treatment failures who only had mild-moderate fibrosis (Ishak score 1–2 or Metavir F1/F2) at initiation of HCV treatment. Second, most studies in this meta-analysis did not attempt to estimate duration of HCV infection among their study population. This reinforced our decision to use failure of HCV therapy as the defining characteristic for the inception cohort of included studies. Third, many of these studies were conducted in Asia and examined Asian populations,5, 11, 15, 16, 17, 18, 20, 21, 22, 24, 25, 26, 27, 28, 29, 33, 34, 39 and it is unclear whether these results are generalizable to populations in the US and Europe. Fourth, many investigators published multiple reports about near-identical cohorts of patients. Therefore, we carefully assessed studies for overlapping cohorts of patients and only used the single, most recently published report in this meta-analysis. Finally, studies did not consistently provide data about the initial treatment regimen among HCV-infected patients with treatment failure. It is conceivable that HCV-infected patients who failed after interferon monotherapy might be clinically different from HCV treatment failures after combination therapy (eg, lower HCV RNA levels).


Our meta-analysis is also limited because our estimates about the annual rates of decompensated cirrhosis, HCC, and liver-related mortality demonstrated heterogeneity in individual study results. Study populations might have been substantially different for various reasons, including differences in prevalence of confounding factors (eg, alcohol consumption) or differences in duration of HCV infection. To account for this heterogeneity, we used random-effects meta-analysis, which results in wider confidence intervals around our point estimates of pooled outcomes. Nevertheless, many of the studies in our meta-analysis were relatively small and used a retrospective study design, which might lead to imprecise estimates of liver-related morbidity and mortality. Notably, the HALT-C trial might overcome some of these deficiencies. This trial8 was a large prospective trial with clear criteria for the diagnosis of HCC, but we did not include it in our meta-analysis because patients received maintenance therapy. In this trial, maintenance therapy did not significantly impact outcomes, and a combined cohort of treated and untreated cirrhotic patients produced an annual HCC rate of 1.1%, which is lower than the annual rate estimated in our meta-analysis.


In conclusion, our meta-analysis suggests that HCV-infected treatment failure patients with advanced fibrosis suffer decompensated cirrhosis, HCC, and/or liver-related mortality at an annual rate of approximately 2%–3%. Among studies of patients with varying levels of fibrosis, the annual rates vary from 0.56%–1.39%. Patients who achieve SVR are substantially less likely to suffer from liver-related mortality, HCC, or hepatic decompensation (RR, 0.16–0.23). These data will facilitate decision-making by HCV-infected treatment failure patients and physicians when considering whether to re-treat with currently available therapies or to delay treatment until newer therapies are available.

Results


Results of Literature Search


The computer-assisted search yielded more than 2276 potentially relevant articles (Supplementary Figure 1). After initial review of titles and abstracts, 215 studies were potentially appropriate, and abstracts were reviewed in detail. One hundred four studies were potentially appropriate and underwent full manuscript review along with 3 additional studies identified through recursive literature searches. After application of the inclusion and exclusion criteria, 26 studies were included in our review, and descriptive characteristics can be found in Table 1.5, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35

Among studies reporting liver-related mortality rates, no evidence of publication bias was present on the basis of Begg's (P = .45) and Egger's (P = .27) tests. Furthermore, neither trimming the outliers nor filling in the missing studies changed the estimate of pooled mortality rate by more than 0.1%. However, visual inspection of the funnel plot suggests that publication bias could be present. Among the studies reporting HCC rates, little evidence of publication bias was seen on visual inspection of the funnel plot by Begg's (P = .2) or Egger's tests (P = .06). Among the studies reporting rates of hepatic decompensation, no evidence of publication bias was present on the basis of Begg's (P = .5) or Egger's tests (P = .2). However, visual inspection of the funnel plot suggested possible publication bias, and trimming the outliers and filling in the missing studies changed the estimate of pooled hepatic decompensation rate from 0.56% to 0.1%, indicating that the pooled estimate is likely to be substantially affected by publication bias. On meta-regression, differences between studies in follow-up time, mean age of the cohort, and median stage of fibrosis at cohort inception were not associated with liver-related morbidity and mortality results.

Liver-Related Mortality


The primary outcomes reported by each study are shown in Table 2. Among studies that enrolled patients with any stage of hepatic fibrosis, the pooled liver-related mortality rate among HCV treatment failures was 0.81%/year (95% confidence interval [CI], 0.55–1.07) (Figure 1A), but the pooled liver-related mortality rate among HCV treatment failures was substantially higher at 2.73%/year (95% CI, 1.38–4.08) (Figure 1B) among studies that only enrolled patients with advanced fibrosis or cirrhosis. Both analyses demonstrated moderate statistical heterogeneity between studies (P = .001 for any stage, P < .001 for advanced fibrosis). Chronic hepatitis C patients with SVR are much less likely to suffer liver-related mortality compared with chronic hepatitis C patients with treatment failure who do not achieve SVR in studies with HCV patients with all stages of fibrosis (RR, 0.23; 95% CI, 0.10–0.52) (Figure 1C) or studies that only enrolled patients with advanced fibrosis (RR, 0.19; 95% CI, 0.10–0.37) (Figure 1D). Neither analysis demonstrated statistical heterogeneity (P = .09 for any stage, I2 = 0%, P = .75 for advanced fibrosis).

Hepatocellular Carcinoma


Among studies that enrolled patients with any stage of hepatic fibrosis, the pooled HCC rate among HCV treatment failures was 1.84%/year (95% CI, 1.36–2.32) (Figure 2A), but the pooled HCC rate among HCV treatment failures was numerically higher at 3.22%/year (95% CI, 2.02–4.42) (Figure 2B) among studies that only enrolled patients with advanced fibrosis. Again, both analyses demonstrated moderate statistical heterogeneity between individual studies (P < .001 for any stage, P < .001 for advanced fibrosis). Also, chronic hepatitis C patients with SVR are much less likely to suffer HCC compared with chronic hepatitis C patients who do not achieve SVR among studies that enrolled patients with any stage of hepatic fibrosis (RR, 0.21; 95% CI, 0.16–0.27) (Figure 2C) and among studies that only enrolled patients with advanced fibrosis or cirrhosis (RR, 0.32; 95% CI, 0.23–0.44) (Figure 2D). Again, neither of these analyses demonstrated statistical heterogeneity (P = .93 for any stage, P = .61 for advanced fibrosis).

Hepatic Decompensation


Among studies that enrolled patients with any stage of hepatic fibrosis, the pooled hepatic decompensation rate among HCV treatment failures was 0.54%/year (95% CI, 0.11–0.97) (Figure 3A), but the pooled hepatic decompensation rate among HCV treatment failures was substantially higher at 2.92%/year (95% CI, 1.61–4.22) (Figure 3B) in studies that only enrolled patients with advanced fibrosis or cirrhosis. Both of these analyses demonstrated moderate statistical heterogeneity (P < .001 for any stage, P < .001 for advanced fibrosis). Chronic hepatitis C patients with SVR are much less likely to suffer hepatic decompensation compared with chronic hepatitis C patients who do not achieve SVR in studies that enrolled patients with any stage of hepatic fibrosis (RR, 0.16; 95% CI, 0.04–0.59) (Figure 3C) and studies that only enrolled patients with advanced fibrosis or cirrhosis failures (RR, 0.13; 95% CI, 0.06-0.27) (Figure 3D). No statistical heterogeneity was identified between studies (P = .69 for any stage, P = .27 for advanced fibrosis).

Natural History of Relapsers Versus Nonresponders


Four of the studies12, 18, 22, 23 reported outcomes according to whether the patients initially became virus-negative (relapsers) versus remaining persistently viremic (nonresponders). The study by Coverdale et al12 was the only one to report mortality outcomes with liver-related mortality occurring in relapsers at a rate of 0.7%/year and in nonresponders at a rate of 1.8%/year (P = .4). The study by Pradat et al23 was the only one to report rates of hepatic decompensation that occurred at 1.7%/year in relapsers versus 2.9%/year in nonresponders (P = .5). All 4 studies reported rates of HCC, and nonresponders were more likely to develop HCC compared with relapsers (RR, 2.30; 95% CI, 1.26–4.19) (Figure 4). There was no evidence of statistical heterogeneity (P = .49).

Methods


Literature Search


A computer-assisted search with the Ovid interface to Medline was conducted to identify potentially relevant published articles. A search of the Medline database from 1966 to 2008 was performed by using the exploded (exp) medical subject heading (MeSH) terms (exp Hepatitis C) AND (exp Treatment Failure OR exp Retreatment OR exp Recurrence OR exp Drug Resistance OR keywords refractory, relapse, retreatment, nonrespond, non-respond, fail, unsuccess). The results of all searches were limited to human studies published in the English language. Manual searches of reference lists from potentially relevant articles were also performed to identify any additional studies that might have been missed by using the computer-assisted strategy.

Study Selection Criteria


Two investigators (M.V., A.S.) independently reviewed the titles and abstracts of all citations identified by the literature search. Potentially relevant studies were retrieved, and the selection criteria were applied. The inclusion criteria were (1) cohort studies of treatment failure HCV patients who were treated with interferon/pegylated interferon ± ribavirin; (2) reporting of liver-related morbidity or mortality end points stratified by SVR status; and (3) published in the English language in full manuscript form. Treatment success was defined as HCV-infected patients who achieve an SVR, undetectable virus in the serum 6 months after the end of therapy. Treatment failure was defined as HCV patients who do not achieve SVR. Morbidity end points were defined as HCC or decompensated cirrhosis (defined by ascites, hepatic encephalopathy, or bleeding esophageal varices). Liver-related mortality was defined as death as a result of complications of decompensated cirrhosis, HCC, and/or liver transplantation, even if patients survived the procedure. The following types of studies were excluded: (1) cross-sectional studies with short-term follow-up (<1 year); (2) studies reporting only surrogate end points such as stage of fibrosis on liver biopsy or AST/ALT levels; (3) studies with unclear follow-up time; (4) studies that enrolled patients with evidence of decompensated cirrhosis at inception of the cohort; (5) studies in which the primary end point was response to therapy; (6) studies of patients with human immunodeficiency virus coinfection, on dialysis, or post-transplant; (7) duplicative studies reporting on the same cohort of patients; and (8) studies in which patients received maintenance treatment or studies in which patients were initially treated with agents other than interferon alfa/pegylated interferon ± ribavirin (eg, interferon beta or glycyrrhizin). Maintenance therapy was defined as any continuous interferon-based treatment beyond a standard course (ie, 48 weeks for genotype 1 patients) with the intent of suppressing hepatic inflammation and fibrosis.8 Studies that only enrolled patients with advanced fibrosis or cirrhosis were included, although these studies were analyzed separately from studies enrolling HCV-infected patients with treatment failure with all stages of fibrosis.

Data Extraction


The eligible articles were reviewed in a duplicate, independent manner by 2 investigators (M.V., A.S.). Data extraction on standardized forms was also performed in a duplicate, independent manner by the 2 investigators.9 Data were collected on annual rates of liver-related mortality, HCC, and decompensated cirrhosis among patients failing to achieve SVR and patients who achieved SVR. Data were also collected on the design of each study (retrospective vs prospective), the number of patients in each group, demographic characteristics of the patient population, median stage of fibrosis at cohort inception, proportion of patients with advanced fibrosis (Metavir stages F2–F4 or Ishak stages 3–6) at cohort inception, and the duration of follow-up.

Statistical Analysis


Pooled rates of liver-related mortality, HCC, and decompensated cirrhosis in HCV treatment failure patients were calculated by using random-effects meta-analysis according to the method of DerSimonian and Laird.10 Separate pooled rates are provided for HCV treatment failure patients from studies that only enrolled patients with advanced fibrosis or cirrhosis (Metavir stages F3-F4 or Ishak stages 3-6) at cohort inception. Also, the relative risk (RR) of each end point is compared between HCV treatment failure patients versus patients who achieved SVR. Finally, the RR of each end point is also compared between patients who became virus-negative during treatment (relapsers) versus those remaining persistently virus-positive (nonresponders) to determine whether natural history differs between these groups of HCV treatment failure patients.


Heterogeneity was assessed by using the I2 index or the variation in effect size attributable to heterogeneity. Meta-regression was then used to assess the following prespecified reasons for heterogeneity: differences between studies in follow-up time, mean age of the cohort, and median stage of fibrosis at cohort inception. Visual inspection of funnel plots, as well as statistical testing according to the methods of Begg and Egger, were used to assess for publication bias. These analyses were limited to studies including all stages of fibrosis to prevent heterogeneity from obscuring evidence of publication bias. The impact of publication bias on the pooled rates was then estimated by the trim and fill method.