Cirrhosis regression in hepatitis C patients with sustained virological response after antiviral therapy: a meta-analysis

Authors


Abstract

Background & Aims

Chronic hepatitis C may be associated with cirrhosis, liver failure and hepatocellular carcinoma. Studies have demonstrated improved clinical outcome in patients who achieved a sustained viral response (SVR).

Methods

A systematic literature search was performed to identify studies that assessed the association between SVR and cirrhosis regression. The main outcome studied was cirrhosis regression in patients with a SVR as compared with patients without a SVR. Six studies totalling 443 patients were included. Dichotomous outcomes were reported as risk ratios (RR) with 95% confidence intervals (CI).

Results

Of the 443 patients with cirrhosis, 137 achieved a SVR. Of these 137 patients who achieved an SVR, 73 (53%) patients had regression of cirrhosis. The risk ratio of cirrhosis regression was 2.69 [Confidence Interval (CI) 1.45–4.97, P < 0.01] in patients who achieved a SVR. The risk of cirrhosis regression was consistently in favour of patients who achieved a SVR regardless of the length of the biopsy or whether the biopsy was reviewed by a single or multiple pathologists. The risk ratio of cirrhosis regression was related to the duration of follow-up between biopsies. The relative risk for regression of cirrhosis in studies in which the mean or median time for the follow-up liver biopsy was greater than 36-month was 4.33 (CI 1.1–17.0, P = 0.04) as compared to a relative risk of 1.79 (CI 1.26–2.29, P < 0.01) in studies with a mean or median time between the follow-up biopsy of less than 36-month.

Conclusions

Our results suggest that the majority of patients with cirrhosis who achieve a SVR develop cirrhosis regression. Time between biopsies appears to be an important determinant of the likelihood of cirrhosis regression.

Hepatitis C is one of the leading causes of cirrhosis in the United States [1]. Estimates of the number of Americans infected with hepatitis C range from 3 to 7 million people [2, 3]. The public burden is increasingly realized as the percentage of patients with hepatitis C today with cirrhosis is between 15% and 20% [4]. Patients with cirrhosis are at risk of liver failure and hepatocellular carcinoma. Indeed, HCV is currently the most common indication for transplantation in the United States [5].

The end point of successful antiviral therapy is achieving Sustained Virological Response (SVR) [6]. Sustained virological response has been associated with arrest of disease progression; and improvements in quality of life and reduction of liver-related complications and hepatocellular carcinoma risk [7-9]. A number of studies have also documented improvement in liver histology after sustained virological response [10-17]. Histological improvement has been noted even in the context of cirrhosis [14]. While a large meta-analysis has previously demonstrated reduction of cirrhosis-related complications in patients achieving SVR [18], no previous meta-analysis has examined the likelihood of cirrhosis regression in patients who achieved a SVR.

Cirrhosis has been regarded as the final common pathway of liver disease [19]. Multiple studies have successfully challenged the premise that cirrhosis is irreversible particularly when the liver disease culprit is eliminated [20-23]. In this meta-analysis, we sought to better understand the relationship between SVR and cirrhosis regression in patients with HCV treated with antiviral therapy.

Methods

Objective

To perform a systematic review of the literature and meta-analysis to determine whether cirrhosis is reversible in hepatitis C patients with sustained virological response to antiviral therapy.

Selection of studies

Trials that met the following criteria were included: (a) prospective or retrospective cohort studies as well as randomized, controlled, open or blinded trials pertinent to the subject matter and published as an article or abstract, (b) studies that reported follow-up data on patients greater than or equal to 6 months, (c) studies including subjects with serological confirmation of chronic HCV infection, (d) SVR defined as no detectable levels of HCV RNA by PCR at least 24 weeks after antiviral treatment, (e) studies that included paired liver biopsies and data regarding histological preparation of biopsies, such as biopsy length, time between biopsies, presence of central pathologist and a validated method of staging cirrhosis, (f) studies that included at least 10 patients. Articles excluded were (a) studies looking specifically at causes of cirrhosis other than Hepatitis C, including Wilson's disease, PSC, PBC, haemochromatosis, alpha-1 antitrypsin and alcoholic cirrhosis, (b) studies including patients with immunosuppression secondary to chronic steroid use, HIV, or any other aetiology, (c) studies in which data could not be extracted for a subset of cirrhotic patients with and without SVR.

Search strategy

A comprehensive search of the MEDLINE database and the Cochrane Database of Systematic Reviews was performed to find studies published in the English language up to October 2013 that investigated cirrhosis regression in Hepatitis C end-stage liver patients treated with antiviral therapy. We used combinations of the keywords Hepatitis C, antiviral agents, liver cirrhosis, SVR, viral suppression, histology, revers*, regression and improvement. We also manually searched manuscript references to identify additional studies that may have been missed with a MEDLINE-assisted strategy. Medical science liaisons for the appropriate antiviral therapies were contacted to assess for additional studies to review.

Data extraction

Studies were subjected to inclusion and exclusion criteria. Two reviewers (EA and VM) independently and in duplicate assessed the eligibility and quality of trials. A formal scoring system to rate the study quality of each individual study was not used. Reviewers noted patient liver biopsy length, duration between biopsies, fibrosis scoring system, baseline biopsy score, antiviral therapy, length of treatment and regression of cirrhosis. Three of the six studies examined cohorts of patients with any amount of liver fibrosis on histology [26-28]. In these studies, only data regarding cirrhotic patients were extracted from the text. Efforts were made to contact the authors for demographic and clinical characteristics of cirrhotic patients. Collaboration was established with one author. Regression of fibrosis was defined individually for each scoring modality and is noted in the results section below.

Statistical analysis

We used the statistical package RevMan (Review Manager, Version 5.2. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration,v2012). RevMan software was developed by the Cochrane Collaboration to facilitate development of systematic reviews and meta-analyses. The Mantel–Haenszel procedure for binary data was used to determine clinical significance of effect. Sensitivity analysis was two-tailed and set at P ≤ 0.05. A random-effects model was employed because of the anticipated variability between trials in terms of patient populations, interventions and concomitant interventions. Heterogeneity between trials was assessed by the chi-squared test with significance set at P ≤ 0.10. The approximate proportion of total variability in point estimates attributed to heterogeneity was calculated by use of the I2 statistic [25].

Results

Number of studies

A total of 172 relevant articles were identified using the search criteria detailed above. Twenty-one manuscripts were reviewed in full. After applying the inclusion criteria, six studies [12, 14, 24, 26-28] were used in the final analysis (Fig. 1). The six studies included a total of 443 cirrhotic patients. The median number of patients across each study was 62.5 (range 15–153) [12, 14, 24, 26-28].

Figure 1.

Study selection – Algorithm depicting the literature search flow chart and why studies were included or excluded.

Diagnosis of cirrhosis

Cirrhosis was diagnosed by liver biopsy in all studies. Biopsies were evaluated by Metavir score in five studies [12, 14, 24, 27, 28]. The remaining study used the Ishak scoring method to evaluate cirrhosis. A Metavir score of F4 or an Ishak fibrosis score of ≥5 was used to define cirrhosis. All studies reported results for paired biopsies. One biopsy was taken prior to antiviral therapy and one biopsy was taken after antiviral therapy. Biopsy length was reported in five of the six studies [12, 14, 24, 26, 27]. The time between each biopsy was listed for all studies (Table 1). A central pathologist was used for diagnosis in two of the six studies [26, 27]. These data were not reported in one study [14]. The remaining studies used 2–3 independent reviewers [24, 27, 28].

Table 1. Study characteristics of patients with hepatitis C
StudyCirrhotic patientsCohort: Cirrhosis regressionFibrosis scoring systemBiopsy length (mm)Time between biopsies (months)Treatment regimenTreatment duration (weeks)Regression of cirrhosis (%)
  1. a

    All biopsies in Poynard et al. were required to have minimum length of 15 mm.

  2. IFN, interferon alpha; IQR, Interquartile range; RBV, ribavirin; Rx, treatment; Non-SVR, patients without sustained virological response; NR, not reported; SVR, patients with sustained virological response.

Shiratori [28]54

SVR

Non-SVR

MetavirNR

Median 44.4 (12–20)

Range

IFN, IFN-beta8–24

11/24 (46%)

9/30 (30%)

Poynard [27]153

SVR

Non-SVR

Metavir>15aMean 21 ± 4PEG ± RBV, IFN ± RBV24–48

25/37 (68%)

50/116 (43%)

Arif [26]15

SVR

Non-SVR

IshakMean 12 ± 2Mean 27 ± 9IFN + RBV12–48

5/6 (83%)

4/9 (44%)

Abergel [24]61

SVR

Non-SVR

Metavir

Median 13 (4–22)

IRQ

12 pre-rx

6 post-rx

IFN + RBV24–48

6/18 (33%)

4/43 (9%)

Pol [12]64

SVR

Non-SVR

Metavir

F4 → F2

Mean 16.8 ± 4.4Mean 55.2 ± 26.4IFN + RBV24–48

4/17 (24%)

1/47 (2%)

Mallet [14]96

SVR

Non-SVR

Metavir

Median 15(10–20)

IQR

Median 38 (23–69)

IQR

IFN + RBV, PEG + RBV48

22/35 (63%)

5/61 (8%)

Antiviral therapy

All studies used interferon-based regimens. Five of the six studies treated patients using interferon or pegylated interferon with or without the addition of ribavirin [12, 14, 24, 26, 27]. In Shiratori et al., patients were treated solely with interferon [28]. Duration of therapy across studies varied from 8 to 48 weeks. Specific details regarding antiviral therapy and treatment duration for each study are noted in the table (Table 1). Many of the studies were retrospective analyses of prior randomized controlled trials comparing antiviral dosages and length of therapy (Table 1) [12, 14, 27, 28]. The median number of patients achieving sustained virological response across studies was 34% (range 24–44%) [12, 14, 24, 26-28].

Regression of cirrhosis

Regression of cirrhosis was defined as reduction in Metavir stage to ≤F3 or Ishak fibrosis score to ≤4. Median regression of cirrhosis in patients with SVR across studies was 55% (range 24–83%). Overall, a total of 73 patients with SVR had regression of cirrhosis of a total of 137 patients with SVR (53%). Of note, regression of cirrhosis in patients who did not achieve SVR was also observed. Median regression of cirrhosis in these patients was 19.5% (range 2–44). Risk ratios were calculated to compare regression of cirrhosis in patients with SVR against regression of cirrhosis in patients without SVR. Relative risk for regression of cirrhosis across all studies was 2.69 (95% CI 1.45–4.97, P < 0.01) (Fig. 2).

Figure 2.

Meta-analysis overall result – Cirrhosis regression in patients with and without a sustained viral response.

Subgroup analysis

A number of subgroup analyses were also performed. The relative risk ratio for studies that studied only cirrhotic or advanced fibrosis patients was 6.15 (95% CI: 3.18–11.91, P < 0.01) [12, 14, 24]. The relative risk ratio for studies utilizing a central pathologist to review liver biopsy slides as compared with studies without a central pathologist was 1.60 (95% CI: 1.20–2.13, P < 0.01) [26, 27] and 3.97 (95% CI: 1.51–10.45, P = 0.005) [12, 14, 24, 28] respectively. Studies in which the mean biopsy length or median biopsy length of liver samples was less than 15 mm had a relative risk ratio of 1.91 (95% CI: 1.18–3.09, P = 0.008) [24, 26, 28], whereas studies that had mean biopsy length or median biopsy length greater than 15 mm had a relative risk ratio of 4.38 (95% CI: 0.95–20.25, P = 0.06) [12, 14, 27] for regression of cirrhosis. Subgroup analyses were also performed comparing studies in which the mean time between liver biopsies or median time between liver biopsies was less than 36-month or greater than 36-month. The relative risk ratio for cirrhosis regression in these studies was 1.79 (95% CI: 1.26–2.29, P < 0.01) (Fig. 3) [24, 26, 27] and 4.33 (95% CI: 1.1–17.0, P < 0.05) (Fig. 4) [12, 14, 28] respectively.

Figure 3.

Follow-up biopsy time mean or median <36-month subgroup analysis – Cirrhosis regression in patients with and without a sustained viral response in trials in which the follow-up biopsy had mean or median time of <36-month.

Figure 4.

Follow-up biopsy time mean or median >36-month subgroup analysis – Cirrhosis regression in patients with and without a sustained viral response in trials in which the follow-up biopsy had a mean or median time of >36-month.

Discussion

Overall, there was considerable variability in the amount of cirrhosis regression across studies (Median 55%, range: 24–83%). When the data were pooled and analysed, we obtained a relative risk of 2.96 indicating that achieving sustained virological response (SVR) led to an almost three-fold increase in chance of cirrhosis regression than not achieving SVR. Half the trials in this analysis studied not just regression of cirrhosis but also improvement in histological score for non-cirrhotics [26-28]. When our analysis was performed only in manuscripts that specifically studied antiviral therapy in patients with severe fibrosis or cirrhosis, a greater likelihood of cirrhosis regression was noted – 6.15.

The severity of liver disease is a well-known predictor of antiviral response, even with newer direct-acting agents [29]. Potentially, regression may be less likely in patients with more advanced or established cirrhosis. Differences in patient selection may help explain the greater likelihood of cirrhosis regression in studies focusing on patients with advanced liver disease. For instance, it is possible that more stringent entry criteria were utilized in studies focused specifically on patients with severe fibrosis or cirrhosis. We were unable to cumulatively analyse steatosis, comorbidities, alcohol among studies to determine their association with cirrhosis regression.

An important finding in our analysis is that the likelihood of cirrhosis regression may increase over time after SVR is achieved. This is highlighted by the fact that the risk ratio for regression is 4.13 when the follow-up biopsy is taken ≥36-month as compared to a risk ratio of 1.79. These results are consistent with studies identifying SVR as a statistically significant predictor of histological response to antiviral therapy [13, 30]. Individual studies have demonstrated reduced liver-related morbidity and mortality in patients with advanced hepatic fibrosis or cirrhosis and SVR [31, 32]. These results were further supported by a recent meta-analysis demonstrating a significant risk reduction in hepatic decompensation, hepatocellular carcinoma and liver-related mortality in patients with SVR [18]. The histological outcomes observed in this meta-analysis may provide an explanation for the clinical outcomes observed in previous studies.

A noteworthy finding in our analysis is that cirrhosis regression was seen even among treated patients who did not achieve a SVR [12, 14, 24, 26-28]. There are several potential explanations for this. First, the ‘Non-SVR’ comparative group in our analysis did receive interferon [12, 14, 24, 26-28]. Poynard et al. noted identified factors other than SVR that were associated with decreased fibrosis after treatment, including age <40, lower BMI, and mild or no activity of the virus in a multivariate analysis [27]. In addition, it is possible that other factors promoting cirrhosis such as alcohol consumption may have improved in the ‘Non-SVR’ group. Indeed, 25% of the patients in the study by Shiratori et al. reported alcohol prior to the study initation [28]. But, the majority of the patients had completely stopped drinking at study completion.

Optimal liver biopsy length has been a subject of intense controversy because of the risk of sampling variation and interpretation [34]. Biopsy sampling variation can limit the interpretation of results. Studies have shown that this sampling variation can lead to ≥1 stage change in fibrosis score when biopsies are taken from different lobes of the liver or even when taken through the same skin puncture site [35-38]. Within our analysis, we found that the cirrhosis regression was better documented when a minimal biopsy length was utilized in the analysis. Although there was a trend, the difference was not statistically significant. To reduce further discrepancies, AASLD has issued a Class IC recommendation that liver biopsy length be at least 2–3 cm [34]. Analysis by an experienced pathologist or consensus reading between pathologists has been associated with increased agreement [39]. Interestingly, studies that used more than one pathologist had a higher regression of cirrhosis benefit as compared with those studies using a central pathologist (RR 3.96 vs. 1.71).

There are a number of limitations to this study. Heterogeneity between studies is an important limitation of this meta-analysis. Possible factors accounting for the heterogeneity may include relatively small study size, antiviral therapy, patient population, prevalence in confounding factors or duration of follow-up. It is not possible to extrapolate our results to patients with decompensated liver disease. All studies limited clinical trial entry to patients with compensated cirrhosis. This may be secondary to the risk of hepatic decompensation and worsening cytopenias with interferon-based therapy. Now with the introduction of non-interferon-based therapy for hepatitis C, the pool of patients with advanced liver disease who are eligible for treatment will expand [40, 41]. A third limitation is the utilization of only two literature databases – Cochrane and Medline. We may have missed studies not found in those indices. However, we searched through the references of all identified manuscripts to be as complete as possible. Our results may also be subject to publication bias in that authors were more likely to publish if there was a cirrhosis regression than not.

With the emergence of new therapies and better therapies for hepatitis C upcoming, further study into whether these new therapies may lead to different cirrhosis regression rates must be evaluated [41]. Patients with advanced liver disease are more likely to undergo therapy with non-interferon based therapy. For instance, moderate thrombocytopenia is not necessarily an absolute contraindication with the newer therapies ([40]–[42]). We believe that with the advent of the newer therapies, previously interferon ineligible patients may be candidates for antiviral therapy but the rate of cirrhosis regression remains to be seen when treating patients with more advanced liver disease.

Acknowledgements

Roles: study concept and design (All); acquisition of data (VM, EA); analysis and interpretation of data (All); drafting of the manuscript (All); critical revision of the manuscript for important intellectual content (All); study supervision (SS).

 Financial support: Sue and Martin Solomon Patient Care Fund.

 Conflicts of interest: VM and EA have no conflicts of interest. SS is on the Speaker Bureau and is a Consultant to Merck, Gilead, Janssen, Bristol Myers Squibb and Vertex. SS has grant support from Gilead, Merck and Vertex.

Ancillary