Systematic review: outcome of compensated cirrhosis due to chronic hepatitis C infection

Authors


  • This uncommissioned systematic review was subject to full peer-review.

Prof. G. R. Foster, Blizard Institute of Cellular and Molecular Science, 4 Newark Street, London E1 4AT, UK.
E-mail: g.r.foster@qmul.ac.uk

Abstract

Aliment Pharmacol Ther 2010; 32: 344–355

Summary

Background  Most studies evaluating chronic hepatitis C virus (HCV) natural history have taken the development of cirrhosis as an end-point.

Aim  To perform a systematic review of the literature to establish the outcome of compensated HCV cirrhosis.

Methods  A systematic literature review was performed. Only data regarding HCV mono-infected patients were included. Weighted mean annual percentage rates for death/transplantation, decompensation of cirrhosis and development of HCC were calculated.

Results  Thirteen papers were included. Despite some heterogeneity, we extracted data relating to 2386 patients. In compensated HCV cirrhosis, the estimated annual rate of death/transplantation is 4.58%, that of decompensation is 6.37% per and that of HCC, 3.36%. When compared with studies of untreated patients, studies that included treated patients reported significantly lower mean annual percentage rates of HCC (2.52% vs. 4.79%, = 0.02), but not decompensation (5.34% vs. 7.88%, = 0.026) and death/transplantation (3.79% vs. 4.62%, = 0.25).

Conclusions  These rates highlight the need for continued vigilance for the occurrence of HCC, while confirming the relatively slow progress of compensated HCV cirrhosis. Heterogeneity in reporting means that these data may underestimate the rate of disease progression, particularly HCC development. It will be important to ensure clearer distinction between treatment responses in future studies.

Introduction

Chronic infection with hepatitis C virus (HCV) is a leading cause of liver disease with over 170 million infected individuals worldwide.1 A recent meta-analysis indicates that approximately 20–30%2 of patients will progress to cirrhosis over a period of 20 years, although the proportion of patients who will develop cirrhosis over longer periods of follow-up remains unclear. Once cirrhosis has developed, complications including gastrointestinal bleeding, porto-systemic encephalopathy, ascites and hepatocellular carcinoma (HCC) are common. Most studies of the natural history of chronic HCV infection have examined the development of cirrhosis as an end-point and have investigated risk factors for the development of cirrhosis.3, 4 In addition, a number of reports have charted the natural history of compensated cirrhosis due to chronic HCV infection and have documented the higher frequency of HCC in patients with HCV-related cirrhosis compared with cirrhosis associated with other aetiologies, in particular, alcohol.5, 6 However, the rate at which complications of cirrhosis from chronic HCV infection develop remains unclear and here we report the results of a systematic review of studies published over the last 14 years that have examined the development of decompensation.

Methods

On-line searches and manuscript selection

Studies were retrieved from pubmed (http://www.ncbi.nlm.nih.gov/pubmed) using the following search terms. [Search ((natural history of Hepatitis C Cirrhosis) NOT (HIV) NOT (review)) NOT (transplant) Limits: Humans, Clinical Trial, Meta-Analysis, Randomized Controlled Trial, Comparative Study, Multicenter Study, English, All Adult: 19+ years.]. Manuscripts were then reviewed and evaluated for inclusion in the analysis. The bibliographies of these manuscripts were also examined for relevant papers that had not been captured by the initial search strategy. The last search was performed on 10th April, 2010. Papers were excluded on the bases of relevance to this study (in particular clear discrimination between decompensated and compensated cirrhosis) and the inclusion of patients with other causes of chronic liver disease without clear reporting of outcome in patients with HCV alone.

Data analyses

Published data were entered into a Microsoft Excel spreadsheet and analysed. Where studies reported data on patients with co-infection or with other aetiologies for cirrhosis, only data from HCV mono-infected patients were included for further analysis.

Unless reported by the authors, annual percentage rates were calculated as the percentage of patients who reached an endpoint (for example, death or development of a complication) in each study divided by the average (whether mean or median) duration of follow-up in years. Where not otherwise stated, mean percentage rates for all studies were calculated by first multiplying each study’s annual percentage rate by the number of patients in that study and then dividing by the total number of patients.

Student’s t-test or the Mann–Whitney test was used for comparison of continuous variables. All reported P-values are two-sided, and a P-value of <0.05 was considered significant.

Results

Searches and study characteristics

Using the search strategy described above, approximately 30 papers were considered. Following review, 13 papers met the entry criteria and were included in analyses (summarized in Table 1). Two groups (Fattovich et al. and Bruno et al.) have each published more than one study which satisfied the entry criteria; however, the degree of patient overlap between these studies could not be clearly ascertained. The most relevant study from each group was therefore included,7, 8 while the remaining papers were excluded from further analysis.9–11

Table 1.   Characteristics of eligible studies identified for inclusion
StudyCountryStudy designNo. HCV (+) participantsDuration (years)Outcomes assessed
DeathTransplantHCCDecompensationIncreased CP score
  1. HCV, hepatitis C virus; HCC, hepatocellular carcinoma; CP, Child–Pugh. Duration is given as median years, with the exception of Mazzella et al.14, Okanoue et al.12 and Hu and Tong18, where it is given as mean years.

Gramenzi et al.13ItalyNonrandom controlled724.8 
Mazzella et al.14ItalyProspective cohort922.8    
Okanoue et al.12JapanRetrospective cohort555.6    
Shiratori et al.15JapanProspective cohort746.8   
Kobayashi et al.16JapanProspective cohort4908.2  
Sangiovanni et al.20ItalyProspective cohort2149.5
Benvegnu et al.17ItalyProspective cohort3127.6
Hu and Tong18USAProspective cohort1124.5 
Serfaty et al.21FranceProspective cohort1033.3 
Fattovich et al.8Western EuropeRetrospective cohort1366.8 
Toshikuni et al.6JapanProspective cohort1525.4  
Degos et al.22FranceProspective cohort4165.6   
Bruno et al.7ItalyProspective cohort15814.4 

The most frequent study design was prospective cohort study (10 reports), although two retrospective cohort studies8, 12 and one nonrandomized controlled trial13 were also identified. Two of the prospective cohort studies,14, 15 one retrospective cohort study12 and the controlled trial13 were designed to investigate the impact of interferon (IFN) therapy on the outcome of HCV cirrhosis; only data from the untreated or control groups who did not receive IFN therapy are included in the present analysis. Two prospective16, 17 and one retrospective study8 compared the outcome of patients with HBV and HCV cirrhosis, and one prospective study compared outcome of patients with HCV and alcohol-related cirrhosis.6 Only the data pertaining to patients with HCV cirrhosis are included in Table 1.

A majority (eight) of investigations were conducted in Western Europe, one was performed in the USA18 and four in Japan.6, 12, 15, 16 The number of participants was variable (median 144 patients), ranging from 55 (in the untreated cohort of one of the retrospective studies12) to 490.16 Follow-up periods varied from mean 2.8 years 14 to median 14.47 years.

The outcomes assessed were reasonably uniform, with most studies evaluating rates of death, development of HCC and decompensation of cirrhosis (including development of jaundice, ascites, variceal haemorrhage or hepatic encephalopathy). Two studies were designed to investigate the effect of interferon on development of HCC and so did not report data for other outcomes.12, 14 Transplantation was considered as an endpoint in seven studies, and was combined with death for the purposes of survival analysis.

Inclusion and exclusion criteria from the studies are summarized in Table 2. All studies required a positive diagnosis based on the presence of HCV antibody, HCV RNA detection or both for inclusion. Cirrhosis was confirmed by biopsy in the majority of patients, or clinically diagnosed in a small minority based on a variety of criteria (varices or thrombocytopaenia with ultrasound markers of portal hypertension;13 presence of irregular liver margin on ultrasound, portal hypertension with laboratory evidence of chronic liver disease;17 or according to criteria described by Bonacini et al.).18, 19 Formal staging of cirrhosis as Child–Pugh A or B was required by eight studies. All of the studies excluded patients with signs of decompensation, but in four papers, this was not specifically stated.6, 14, 15, 21 However, none of these studies had different outcomes suggesting that they too had in fact excluded patients with decompensated cirrhosis. Pre-existing HCC was generally an exclusion criterion, although this was not explicitly stated by Mazzella et al.14

Table 2.   Inclusion and exclusion criteria utilized by eligible studies
StudyInclusion criteriaExclusion criteria
HCV (+)Biopsy proven cirrhosis [n (%)]Alcohol (max g/day)Abnormal ALT/ASTHCCHIVOther liver diseaseAscitesPSEVariceal bleedingJaundiceUse of diuretics
  1. The proportion of patients where cirrhosis was diagnosed by liver biopsy was not given by Toshikuni et al.6, and was given for all patients, but not for the subgroup of HCV mono-infected, untreated patients whose outcomes were considered in this analysis by Benvegnu et al.17 and Bruno et al.7. HCV, hepatitis C virus, AST, aspartate aminotransferase; ALT, alanine aminotransferase; HCC, hepatocellular carcinoma; HIV, human immunodeficiency virus; PSE, porto-systemic encephalopathy.

Gramenzi et al.1348 (67)60
Mazzella et al.1492 (100)          
Okanoue et al.1255 (100)60     
Shiratori et al.1574 (100)        
Kobayashi et al.16490 (100)    
Sangiovanni et al.20214 (100)     
Benvegnu et al.17Not given     
Hu and Tong18106 (94.5)80  
Serfaty et al.21103 (100)80 
Fattovich et al.8136 (100)80   
Toshikuni et al.6Not given        
Degos et al.22416 (100)       
Bruno et al.7Not given    

All studies except three14, 17, 20 excluded patients with other known causes for liver disease (including coexistent HBV infection). Four studies screened for HIV, and excluded co-infected patients. No study required participants to be abstinent from alcohol, and only six set a limit on alcohol consumption permitted for inclusion in the study.

Patient characteristics

Data pertaining to 2328 patients were available from the 13 studies meeting the inclusion criteria. The characteristics of the patients at entry to each study are summarized in Table 3. The mean age of participants was 58.2 years. The average ages of the patients in 10 studies were relatively homogenous (54–61 years), but two studies described notably older (mean age 69)6 and younger patients (mean age 52.1).18 Gender distribution was more variable, ranging from 38%6 to 75%21 males. Baseline laboratory tests did not vary greatly between studies, presumably reflecting the requirement for compensated cirrhosis (Child–Pugh A or B) at enrolment. The presence of serum anti-HCV antibodies was used to diagnose HCV infection in eleven studies. The presence of serum HCV RNA diagnosed infection in the remaining two and was used to further confirm infection in seven studies. Three studies identified anti-HCV antibodies in the serum of participants, but did not confirm infection with RNA analysis.13, 14, 22 While all patients enrolled in the studies conducted by Benvegnu et al.17 and Serfaty et al.21 were seropositive for HCV, not all patients included were HCV RNA positive (89% and 94% respectively). Stored serum samples were available for only 64% of the patients enrolled by Fattovich et al.8 for HCV RNA testing, although HCV RNA was detected in all of these samples.

Table 3.   Baseline characteristics of participants
StudyGramenzi et al.13Mazzella et al.14Okanoue et al.12Shiratori et al.15Kobayashi et al.16Sangiovanni et al.20Benvegnu et al.17Hu and Tong18Serfaty et al.21Fattovich et al.8Toshikuni et al.6Degos et al.22Bruno et al.7
  1. Where not otherwise specified, figures are given as mean (s.d.) or median (range), depending on the measure used by the source study. Descriptive data given are for the untreated (control arm) of the trial reported by Gramenzi et al.13 and for patients who did not receive treatment in the studies by Mazzella et al.,14 Okanoue et al.,12 Shiratori et al.,15 Serfaty et al.21 and Bruno et al.,7 The data are for HCV monoinfected patients in the studies reported by Kobayashi et al.,16 Benvegnu et al.17 and Fattovich et al.,8 for patients without additional/alternative liver disease reported by Sangiovanni et al.20 and Toshikuni et al.,6 and for all participants reported by Hu and Tong18 and Degos et al.22. ‘Treated’ refers to the number of study participants included in the outcomes analysis who received treatment during the follow-up period. HCV, Hepatitis C virus; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; GGT, gamma glutaryltransferase; AFP, alphafetoprotein; s, seconds.

Age (years)58.1 (7.8)54 (1.22)57.6 (5.2)61 (no range)59 (25–82)55 (7)61 (36–78)52.1 (no range)56 (14)58 (22–79)69 (35–83)57 (46–64)Not given
Males [n (%)]33 (46)52 (57)28 (51)35 (47)289 (59)106 (50)142 (56)56 (50)33 (75)81 (60)57 (38)240 (58)74 (46.8)
HCV diagnosis
 Antibody + [n(%)]72 (100)92 (100)55 (100) 490 (100)214 (100)254 (100) 103 (100)136 (100)152 (100)416 (100)158 (100)
 RNA + [n (%)]  55 (100)74 (100)490 (100)214 (100)225 (89)112 (100)97 (94) 152 (100) 158 (100)
Bilirubin (mg/dL)1.1 (0.5)0.92 (0.04)  1.1 (0.4–3.0) 0.81 (0.12–2.57)0.9 (0.6)0.99 (0.53) 0.9 (0.3–2.0)0.88 (0.64–1.23)0.9 (0.7–1.1)
ALT (IU/L)89 (64)148 (11.2)118 (38)75 (no range)58 (9–315) 130 (18–973)139.6 (111.6)96 (68) 48 (12–230)  
Platelets (×1000 mm3)140.3 (74) 97 (25)105 (no range)96 (17–398) 131 (31–294)171.6 (71.9)124 (49) 103 (21–277)123 (88–170)117 (100–155)
Albumin (g/dL)3.8 (0.5)4.0 (0.03) 4.0 (no range)3.8 (3.0–5.1) 4.2 (3.2–5.3)4.1 (0.4)3.9 (0.5) 3.8 (3.0–4.8)4.1 (3.8–4.5)4.2 (3.9–4.5)
Treated [n (%)]0000023 (11)115 (45)49 (44)59 (57) 0223 (54)0

In six studies, a variable proportion of patients received treatment for HCV during the follow-up period (11–59% of participants). All patients were treated with interferon; however, the dose, regimen and treatment duration varied considerably between studies.

Outcomes

Table 4 summarizes the available data on outcomes in patients with compensated HCV cirrhosis. Losses to follow-up ranged from 3% to 57% of patients recruited and, broadly, the percentage of patients lost to follow-up related to the length of the follow-up period. Eight reports provide data on outcomes in HCV mono-infected, untreated patients.6–8, 12–16 Benvegnu et al.17 reported data on decompensation and HCC for HCV mono-infected patients (n = 254); however, figures for death/transplantation were only given for all patients, including those with HBV and HBV/HCV co-infection (a further 58 patients, total n = 312). Similarly, the outcome data presented by Sangiovanni et al.20 included patients with other liver disease in addition to HCV (alcohol abuse, hereditary haemochromatosis and HBV, affecting 46 patients, or 21% of all those for whom data are presented in that study). Three further studies combined outcomes for both treated patients and untreated patients,18, 21, 22 but detailed data on outcomes for those who did or did not receive therapy were not provided separately. Outcome data for patients according to response to therapy are not provided.

Table 4.   Outcomes for participants in each eligible study, where the data were provided
 Gramenzi et al.13Mazzella et al.14Okanoue et al.12Shiratori et al.15Kobayashi et al.16Sangiovanni et al.20Benvegnu et al.17Hu and Tong18Serfaty et al.21Fattovich et al.8Toshikuni et al.6Degos et al.22Bruno et al.7
  1. N, number of study participants for whom these data have been grouped, and may include some treated patients or patients with additional liver disease (discussed further in the text). Outcome data are given as number of participants (percentage of N). PSE, portosystemic encephalopathy; HCC, hepatocellular carcinoma.

N72925574490214312112103136152416158
Average follow-up (years)4.82.85.66.88.29.57.64.53.36.85.45.614.4
Lost to follow-up [n (%)]3 (4)  16 (22) 66 (31)35 (14) 3 (3)15 (11)87 (57) 31 (20)
       (= 312)      
Death/transplantation [n (%)]11 (15)  24 (44)270 (55)75 (35)65 (21)22 (20)19 (18)35 (26)44 (29)83 (20)89 (56)
 Liver failure7 (10)  8 (33)25 (9)15 (7)15 (5)14 (13)5 (5)18 (13)15 (10)  
 Varices2 (3)   6 (1)6 (3)2 (1) 1 (1) 3 (2)  
 HCC1 (1)  11 (46)200 (41)33 (15)41 (13)7 (6)9 (9)17 (13)17 (11)  
 Sepsis    17 (4)     3 (2)  
 Nonliver1 (1)  5 (21)22 (5)20 (9)13 (4)1 (1)1 (1)17 (13)6 (4)  
Complications      (= 254)24 (21)26 (25)72 (53)96 (63) 63 (40)
 HCC32 (44)9 (10)22 (40)35 (47) 68 (32)78 (31)9 (8)11 (11)23 (17)46 (30)60 (14)53 (34)
 Ascites19 (26)    50 (23)52 (21)10 (9)  32 (21)  
 GI bleed11 (15)    13 (6)45 (18)5 (5)  2 (1)  
 PSE1 (1)    2 (1)12 (5)4 (4)  9 (6)  
 Jaundice1 (1)    36 (17)5 (2)9 (8)  7 (5)  

Annual percentage rates of death/transplantation and decompensation of HCV cirrhosis are given in Table 5. Across the studies, the risk of the combined outcome of death/liver transplantation ranged from 2.74% to 6.72% per annum. The risk of developing any complication of cirrhosis was 2.77–11.70% per annum, with risk of HCC in particular ranging from 1.51% to 7.14% per annum. An estimate of mean annual percentage rates from all studies has been calculated by averaging the annual percentage rate data derived from each study. The results are shown in Table 5, final column. The overall estimated rate of death/transplantation was 4.58% per annum, while the estimated rate of complications was 6.37% per annum.

Table 5.   Outcome data calculated as annual percentage rates, derived from the raw data provided by each study
StudyGramenzi et al.13Mazzella et al.14Okanoue et al.12Shiratori et al.15Kobayashi et al.16Sangiovanni et al.20Benvegnu et al.17Hu and Tong18Serfaty et al.21Fattovich et al.8Toshikuni et al.6Degos et al.22Bruno et al.7Weighted mean
  1. HCC, hepatocellular carcinoma; PSE, porto-systemic encephalopathy.

Death/transplantation (%)3.18  4.776.723.692.744.375.593.785.363.523.914.58
 Liver failure2.03  1.590.620.740.632.781.471.951.83  1.16
 Varices0.58   0.150.300.08 0.29 0.36  0.22
 HCC0.29  2.194.981.621.731.392.651.842.07  2.70
 Sepsis    0.42     0.36  0.41
 Nonliver0.29  0.990.550.980.550.200.291.840.73  0.70
Complications (%)9.26     4.044.767.657.7911.70 2.776.37
 HCC5.503.497.146.96 3.342.691.793.242.495.602.542.333.36
 Ascites3.18    2.462.331.98  3.90  2.69
 Variceal bleed0.29    0.640.620.99  0.24  0.58
 PSE0.29    0.100.260.79  1.10  0.45
 Jaundice     1.77 1.79  0.85  1.48

To assess the impact of interferon therapy on the estimated annual percentage rates of death/transplantation, HCC or decompensation, we compared these rates in studies which did include patients treated with interferon vs. those which did not. The mean annual percentage rate of HCC, but not that of death/transplantation, was significantly lower in the studies where some participants received interferon (Figure 1). Although there was a trend towards a reduction in rate of decompensation amongst the studies including patients who had received interferon, this did not achieve statistical significance (mean 5.34 ± 0.79% vs. 7.88 ± 1.88% per annum, = 0.26). These patients did not differ significantly in terms of age or gender from those who had not received interferon (mean ages 56 ± 1.4 years and 59.5 ± 1.8 years respectively, = 0.16; mean percentage of males 57.8 ± 4.6 and 50.6 ± 2.69 respectively, = 0.14).

Figure 1.

 Annual percentage rates for outcomes according to inclusion of patients who had received interferon in the reported results. (a) Mean annual percentage rates of death/transplantation were 3.79 ± 0.44 and 4.62 ± 0.52 in studies reporting outcomes where some patients received interferon and those which did not include interferon treated patients in outcomes respectively (mean ± S.E.M., = 0.25). (b) Mean annual percentage rates of hepatocellular carcinoma were 2.52 ± 0.34 and 4.79 ± 0.76 in studies reporting outcomes where some patients received interferon and those which did not include interferon treated patients in outcomes respectively (mean ± S.E.M., = 0.02).

To assess the impact of ethnicity on outcomes in compensated HCC cirrhosis, we compared the average annual percentage rates of death/transplantation or HCC development between the Japanese studies and the Western European studies included in this analysis. Only one Japanese study reported decompensation rates and therefore an average could not be calculated for this outcome. As none of the patients included in the Japanese studies had received interferon therapy, their average complication rates were compared with those of the four Western studies where the outcome data were available for untreated patients only. While there was a trend for higher complication rates amongst the Japanese studies, this did not reach statistical significance (mean annual percentage rate of death/transplantation 5.62 ± 0.41 in Japanese vs. 3.63 ± 0.16 in European studies, = 0.10; mean annual percentage rate of HCC 6.57 ± 0.49 in Japanese vs. 4.27 ± 0.79 in European studies, = 0.07). Again, the patients included in these Japanese and European studies did not differ significantly in age or gender (data not shown).

Risk factors for disease progression

The majority of the studies included in the present analysis performed univariate and multivariate analyses to identify factors which independently increase risk of death/transplantation, HCC or decompensation in HCV cirrhosis. Unfortunately, direct comparison between reports was not possible due to differences in categorization of the variables studied.

Considering all studies, multivariate analyses identified a total of 8, 9 and 14 different independent variables that increase the risk of HCC, decompensation and death/transplantation respectively. Those identified by more than one study include alpha-fetoprotein,6, 7, 13, 20 male gender6, 7, 20, 22 and lack of IFN treatment13–15, 21 as independent risk factors for HCC; albumin6, 8, 13, 18 and total bilirubin7, 13, 20 as independent risk factors for decompensation; and platelets,8, 11, 16, 22 albumin,6, 8, 16, 18, 20–22 increasing age7, 8, 11, 15, 16, 22 and presence of oesophageal varices7, 22 as independent risk factors for death/transplantation.

Discussion

Here, we present a summary of the data from 13 studies published during the last 14 years on the outcome of compensated HCV cirrhosis. These data indicate that the estimated annual rate of developing any complication of HCV cirrhosis (including an episode of decompensation or development of HCC) is 6.37%; of developing HCC is 3.36% per annum, and of death/transplantation is 4.58% per annum. Although the total number of patients included in the analysis is large (2386), heterogeneity in reporting reduced the number of patients where the impact of chronic HCV infection could be distinguished from other causes of cirrhosis.

Variability in patient characteristics may account for some of the differences seen between the studies. Gender distribution was markedly different and this may have influenced outcomes as male gender is thought to be a risk factor for disease progression in HCV.4 Consistent with this, the study with the greatest proportion of male participants also showed one of the highest annual rates of death/transplantation.21 The average age of participants in most studies did not vary greatly, but there were two notable outliers and it is of interest that the study that reported the highest rate of complications (11.7% per annum6) was also the study with the oldest average age of participants, and that the study with the youngest average age18 was among the lowest (4.76% per annum). While Far Eastern origin has been suggested as a risk factor for progression in HCV cirrhosis, particularly with regard to HCC development,16 the mean annual percentage rates of death/transplantation and HCC development reported by the Japanese studies were not significantly higher than those reported by the most comparable Western European studies included in this analysis. The publication dates of studies included here span 13 years (1996–2009). However, the recruitment periods of these studies (1982–2007) overlap by a considerable degree and therefore further analysis of the results according the date of the study was not pursued.

Two studies included patients who were HCV RNA negative,17, 21 although the proportion of such patients was small. These patients may have had ongoing HCV infection with RNA levels below the lower limit of detection of the study laboratory, or alternatively, it is possible that these patients had an alternative cause for cirrhosis. Sangiovanni et al.20 excluded HCV antibody-positive, RNA-negative patients from their analysis, but acknowledged that the small number of such patients meant that it was not possible to assess whether their course of disease was significantly different from that of patients who were RNA seropositive.

Three studies16, 17, 20 presented combined outcome data from patients with cirrhosis due to HCV infection alone, and from patients with HCV infection plus additional causes for chronic liver disease (including HBV co-infection, alcohol abuse and hereditary haemochromatosis). HCV RNA seropositivity was confirmed in all patients considered HCV-infected in two of these studies,16, 20 although in only 89% of those considered to have HCV infection in the third study.17 Separate analysis showed shortened survival amongst those with HCV and coexistent causes of liver disease,16, 20 and the incidence of HCC appears increased in HCV/HBV co-infected patients, compared with patients with HCV cirrhosis alone.17 Although the numbers of such patients included in the overall outcome data were relatively small, it is possible that their inclusion may have increased the observed complication rates.

While these studies recognized alcohol abuse (>80 g/day) as a contributor to disease progression in a proportion of patients, alcohol consumption at lower levels was not an exclusion criterion in any study, and may have been an unrecognized cofactor in the progression of cirrhosis. Self-reporting of alcohol intake amongst participants may have underestimated total intake.23 The interactions between alcohol and HCV in chronic liver disease are incompletely understood, but appear to involve earlier onset and more rapidly progressive fibrosis even with levels of alcohol consumption as low as 20 g/day, and a synergistic effect of HCV and alcohol on HCC development at higher levels of consumption (>80 g/day) (reviewed in24). The effect of alcohol consumption on disease progression in these HCV-infected cohorts may therefore have been significant. Similarly, metabolic syndrome, thought to be another cofactor in progression of HCV cirrhosis,25 is not featured or controlled for in-patient selection in these studies.

The effect of IFN treatment on disease progression in HCV cirrhosis remains controversial, with previous studies variously showing no effect of IFN on outcomes in HCV cirrhosis,26 a reduction in rates of HCC,14, 27, 28 or prevention of disease progression29 with improved survival.30 Differences in results seen have been attributed to variations in patient selection, IFN treatment doses and regimens, and bias associated with retrospective cohort studies.13 Due to the clear benefits of interferon and ribavirin-based treatment regimens for HCV, ethical considerations have limited the ability to conduct randomized controlled trials in this area. The present analysis included a nonrandomized controlled trial evaluating the effect of IFN on HCV cirrhosis, which found no effect on overall or event-free survival, although there did seem to be a beneficial effect of IFN therapy on development of HCC.13 An independent protective effect of IFN therapy on HCC development and rates of decompensation in HCV cirrhosis has been demonstrated at multivariate analysis by some,14, 15, 18, 20 but not all,8 of the other studies included in this analysis. Bruno et al.7 found a beneficial effect of IFN therapy on rate of decompensation and liver-related mortality, but only in those who achieved a sustained virological response.

In an attempt to establish whether the inclusion of patients who had undergone anti-viral therapy in some studies had influenced the reported mean annual percentage rates for death/transplantation, decompensation or HCC, these outcomes were compared between studies in which some participants had received therapy vs. those in which all participants were untreated. The difference was significant only for occurrence of HCC, with a higher mean annual rate of HCC reported by studies in which no participants received anti-viral therapy. While the inclusion of outcomes for patients who had received IFN may have led to an underestimation of HCC rates in chronic HCV cirrhosis, it is also possible that selection bias may have led to an overestimation of complication rates in the studies reporting outcomes exclusively in untreated patients. The untreated patients whose outcomes were reported by Bruno et al.7 were older, had higher Model for End Stage Liver Disease (MELD) scores and were more likely to have oesophageal varices than the patients from this study who received treatment. However, other reports emphasize that IFN therapy was withheld on the basis of patient choice,14, 15 concomitant nonliver disease,14 or withheld from matched patients as part of a controlled trial.13 Unfortunately, the studies which reported outcomes for treated patients did not elaborate on the reasons for providing or withholding anti-viral treatment. Therefore, it is difficult to predict how the comparison of treated vs. untreated may be skewed.

In view of recent reports indicating that maintenance therapy with low dose pegylated IFN does not reduce the incidence of complications of chronic HCV in those who do not eradicate the virus,30 the value of viral eradication in patients with cirrhosis remains to be determined and further studies should address this important issue. Similarly, it will be important to ensure that future studies clearly differentiate between treatment responders and failures. It is to be hoped that the heterogeneity of reporting that we have documented will lead to better descriptions of the patient populations in future studies of the outcome of cirrhosis.

A number of independent variables have been postulated as independent markers for development of complications of cirrhosis, and while further analysis was not possible here, the variables identified most frequently are in accord with scoring systems used internationally to evaluate severity of liver disease (such as MELD). Furthermore, two studies evaluated the significance of oesophageal varices and both found their presence to be independently associated with death in compensated HCV cirrhosis.7, 22

Overall, the summary data presented here provide estimated rates of progression of compensated HCV cirrhosis derived from patient groups comprising treated and untreated patients, from a range of geographical locations and with varied alcohol consumption. These data confirm the relatively slow progress of HCV cirrhosis. However, they highlight the need for increased vigilance for development of HCC, particularly in those with cofactors for progressive liver disease.

Acknowledgements

Declaration of personal interests: We are grateful to Xiang He, Queen Mary’s Innovation Centre, for expert statistical advice. Declaration of funding interests: Professor Foster has performed consultancy work, received grants from and spoken on behalf of companies (including Roche, Novartis, Human Genome Science, Tibotec, GSK and Chughai) which market and/or are developing drugs to treat chronic Hepatitis C. This study was funded in part by a grant from Novartis and Human Genome Science.

Ancillary