Presented to the Second Annual Conference of the International Liver Cancer Association, Chicago, Illinois, USA, September 2008
Systematic review and meta-analysis of interferon after curative treatment of hepatocellular carcinoma in patients with viral hepatitis†
Version of Record online: 11 AUG 2009
Copyright © 2009 British Journal of Surgery Society Ltd. Published by John Wiley & Sons, Ltd.
British Journal of Surgery
Volume 96, Issue 9, pages 975–981, September 2009
How to Cite
Breitenstein, S., Dimitroulis, D., Petrowsky, H., Puhan, M. A., Müllhaupt, B. and Clavien, P.-A. (2009), Systematic review and meta-analysis of interferon after curative treatment of hepatocellular carcinoma in patients with viral hepatitis. Br J Surg, 96: 975–981. doi: 10.1002/bjs.6731
- Issue online: 11 AUG 2009
- Version of Record online: 11 AUG 2009
- Manuscript Accepted: 27 MAY 2009
A combined antiviral and tumoricidal effect of interferon (IFN) is assumed to occur after resection or ablation of hepatocellular carcinoma (HCC).
An electronic search of the Medline, Embase and Central databases from January 1998 to October 2007 was conducted to identify randomized controlled trials evaluating adjuvant effects of IFN after curative treatment of HCC. A meta-analysis was performed to estimate the effects of IFN on 2-year outcome.
Seven trials enrolling a total of 620 patients were included in the meta-analysis. Adjuvant treatment with IFN significantly reduced the 2-year mortality rate after curative treatment of HCC, with a pooled risk ratio of 0·65 (95 per cent confidence interval 0·52 to 0·80); P < 0·001) in absence of any significant heterogeneity (I2 = 0 per cent, P = 0·823 for χ2). The effect on reduction of tumour recurrence was less pronounced but still significant (pooled risk ratio 0·86 (95 per cent c.i. 0·76 to 0·97); P = 0·013). IFN had to be discontinued in 8–20 per cent of patients.
IFN has a significant beneficial effect after curative treatment of HCC in terms of both survival and tumour recurrence. Copyright © 2009 British Journal of Surgery Society Ltd. Published by John Wiley & Sons, Ltd.
The impact of adjuvant immunotherapy with interferon (IFN) after curative resection or ablation of hepatocellular carcinoma (HCC) is still controversial. Any benefit of IFN on tumour recurrence and survival is accompanied by a variety of side-effects.
The incidence of HCC is rising faster than those of most other cancers owing to the increasing prevalence of hepatitis B (HBV) and C (HCV) virus infections, especially in Asia1, 2. Liver transplantation can be offered to only a small proportion of patients3 because of selection criteria and high costs. Therefore, liver resection and local ablation in patients with well preserved hepatic function are the only curative treatment modalities for most patients with HCC4, 5, with a 5-year survival rate of 40–60 per cent6–8. Locoregional tumour recurrence occurs in more than 50 per cent of patients within 3 years of resection or ablation3, 9, 10 and is, in combination with the underlying liver disease, the main cause of death.
IFN suppresses the replication of HBV and HCV and, in addition, has a tumoricidal effect on a number of tumours including HCC11–13. This double action of IFN is believed to be the principal mechanism underlying the prevention of HCC recurrence after surgical resection or ablation in patients with viral hepatitis.
A number of randomized controlled trials (RCTs) investigating the effects of IFN on survival and tumour recurrence after curative resection or ablation of HCC were inconclusive, either because effects were not statistically significant or considered only for defined subpopulations14–20. The aim of the present study, therefore, was to perform a systematic review of RCTs to evaluate the effects of IFN in an adjuvant therapeutic setting of HCC.
The electronic search was carried out by an information specialist, and included searches in Medline, Embase, the Cochrane Central Register of Controlled Trials (Central) and the Database of Abstracts of Reviewers of Effectiveness (DARE). The search was limited to the period between January 1998 and October 2007 because IFN was not approved for the treatment of viral hepatitis before the late 1990s. ‘Hepatocellular carcinoma’, ‘interferon’, ‘adjuvant treatment’, ‘liver resection’ and ‘liver ablation’ were entered as keywords. The detailed search strategy is available from the authors. All included studies were also entered into the PubMed ‘related articles’ function and the science citation index to cross-search for similarly indexed studies. The reference list of included studies was also scrutinized.
RCTs evaluating the effect of adjuvant therapy with IFN-α or -β in patients after surgical resection or ablation of HCC related to HBV or HCV were considered for review. Non-randomized trials, uncontrolled studies, and surveys and case reports, were excluded from analysis.
The bibliographic details of all retrieved articles were stored in an endnote file and duplicate records resulting from the various database searches were removed. There were no restrictions regarding the language of articles. Two authors independently assessed the titles and abstracts of all identified citations. Their decisions were recorded individually in the endnote file and then compared. Any disagreement was resolved by a third author. Two authors (S.B. and D.D.) evaluated the full text of all potentially eligible papers and decided whether to include or exclude each study according to the inclusion criteria. All studies that did not fulfil all of the criteria were excluded.
One author (D.D.) extracted details about the study design, interventions, patients and outcome measures, such as survival and tumour recurrence, in a predefined Microsoft® Windows® Excel (Microsoft, Redmond, Washington, USA) form. Owing to the variability in median observation periods among the included studies outcome variables were extracted after 2-year follow-up in order to guarantee complete and comparable follow-ups across trials without having any patient dropouts. Dichotomous data were abstracted into 2 × 2 tables. Another author (S.B.) checked data extraction for correctness and a third author resolved any discrepancies if the first two reviewers disagreed. Bibliographic details such as author, journal, year of publication and language were also registered.
Two authors (S.B. and D.D.) independently evaluated all included trials using a list of selected quality items assessing components of internal validity. Method of randomization, concealment of random allocation, blinding of outcome assessors and reporting of an intention-to-treat analysis were assessed. Trials were considered to be of good quality if they reported on three or four of these quality items, of moderate quality if they reported on one or two items, and of low quality if they reported none of the items. The reporting of this systematic review is in accordance with the QUOROM statement21.
Treatment effects were expressed as risk ratios with corresponding 95 per cent confidence intervals (c.i.). Data across studies was pooled in the absence of significant heterogeneity (P > 0·1000 for χ2) using fixed-effects (inverse variance method) and random-effects (DerSimonian–Laird method) models. If the fixed- and random-effects models produced the same result, only the fixed-effects estimate was reported. Heterogeneity was assessed using the χ2 statistic and the proportion of variation due to heterogeneity was expressed as I2. In the presence of significant heterogeneity (P < 0·100 for χ2) two a priori defined sources of heterogeneity were used, which included the type of IFN (α or β) and the type of viral hepatitis (HCV or mixed HCB and HBV). All analyses were conducted with Stata® for Windows® version 8.2 (StataCorp, College Station, Texas, USA).
Identification and characteristics of studies
From 395 citations identified by database searches, seven eligible RCTs including a total of 620 patients were included in the meta-analysis (Fig.1). The main reason for exclusion of studies was lack of randomization.
The size of the included trials varied from 20 to 236 patients. The primary endpoints were survival or tumour recurrence with a median follow-up between 25 months and 7·1 years. One trial had three treatment arms, whereas all the others had two arms (Table1). The methodological quality was good in the trial by Sun et al.22 and moderate in the other six RCTs14–17, 23, 24 (Table2). The results from fixed- and random-effects models were almost identical, so only the fixed effects models were reported.
|Ikeda et al.14||2000||20 patients (13 men, 7 women; mean age 61 years) with HCV infection and curative treatment of HCC (surgical resection or percutaneous ethanol injection). Median follow-up 25·0 months. Primary endpoint: tumour recurrence||Group A: 10 patients received natural IFN-β 6 million units twice a week for 36 months|
|Group B: control group of 10 patients|
|Kubo et al.15||2002||30 men with HCV infection and curative surgical resection of a single HCC tumour. Endpoint: tumour recurrence||Group A: 15 patients (mean age 61·9 years) received 6 million units IFN-α intramuscularly every day for 2 weeks, then three times weekly for 14 weeks, and finally twice weekly for 88 weeks. Median follow-up 5 years|
|Group B: control group of 15 patients (mean age 60·0 years). Median follow-up 4 years|
|Shiratori et al.16||2003||74 patients (52 men, 22 women) with compensated cirrhosis owing to HCV after curative ablation of a maximum of 3 HCC lesions with percutaneous ethanol||Group A: 49 patients (median age 61 years) received 6 million units IFN-α intramuscularly three times weekly for 48 weeks|
|injection. Mean follow-up 7·1 years. Primary endpoint: tumour recurrence||Group B: control group of 25 patients (mean age 63 years)|
|Lin et al.17||2004||30 patients (23 men, 7 women) after non-surgical treatment (transarterial chemoembolization or percutaneous acetic acid injection) of HCV- or||Group A: 11 patients (median age 60 years) received 3 million units IFN-α2β intramuscularly three times weekly for 24 months|
|HBV-related HCC nodules. Median follow-up 27 months. Primary endpoint: tumour recurrence||Group B: 9 patients (median age 63 years) received 3 million units IFN-α2β daily for 10 days every month for 6 months, followed by 3 million units daily for 10 days every 3 months for an additional 18 months|
|Group C: control group of 10 patients|
|Sun et al.22||2006||236 patients (208 men, 28 women; mean age 50 years) after curative resection of HBV-related HCC. Median observation time 36·5 months. Primary endpoint: disease-free survival||Group A: 118 patients (mean age 52·2 years) received 3 million units IFN-α intramuscularly twice a week for 2 weeks and then 5 million units three times weekly for 18 months|
|Group B: control group of 118 patients (mean age 50·4 years)|
|Mazzaferro et al.23||2006||150 patients (112 men, 38 women) after curative resection of HCV-related (n = 80) or HCV and||Group A: 76 patients (mean age 65 years) received 3 million units IFN-α2β three times weekly for 48 weeks|
|HBV-related (n = 70) HCC. Median follow-up 45 months. Primary endpoint: recurrence-free survival||Group B: control group of 74 patients (mean age 67 years)|
|Lo et al.24||2007||80 patients (65 men, 15 women) after curative resection of predominantly HBV-related HCC. Minimum follow-up 30 months. Primary endpoint: tumour recurrence||Group A: 40 patients (mean age 49 years) received 10 million units/m2 IFN-α2β subcutaneously three times weekly for 16 weeks|
|Group B: control group of 40 patients (mean age 54 years)|
|Random allocation (description of procedure)||Concealment of random allocation||Blinding of persons who assess treatment effects||Intention-to-treat analysis|
|Ikeda et al.14||−||−||−||+|
|Kubo et al.15||+||−||−||+|
|Shiratori et al.16||+||−||−||+|
|Lin et al.17||−||−||−||+|
|Sun et al.22||+||+||−||+|
|Mazzaferro et al.23||+||−||−||+|
|Lo et al.24||+||−||−||+|
Effects of interferon treatment on survival and tumour recurrence
Six of seven studies reported on survival (Fig.2). The trials were consistently favourable for IFN, with a pooled risk ratio of 0·65 (95 per cent c.i. 0·52 to 0·80); P < 0·001) without statistical heterogeneity (I2 = 0 per cent, P = 0·823 for χ2 ) in the meta-analysis.
Recurrence rates were reported by all studies (Fig.3). IFN treatment significantly reduced the risk of tumour recurrence with a pooled risk ratio of 0·86 (95 per cent c.i. 0·76 to 0·97); P = 0·013). The moderate-to-large statistical heterogeneity (I2 = 44 per cent, P = 0·100 for χ2) could partly be explained by the type of IFN although the meta-regression did not show a significant influence of the type of IFN used (P = 0·461). The trial by Ikeda et al.14 using IFN-β showed a large effect with a wide c.i. (risk ratio 0·14 (95 per cent c.i. 0·02 to 0·96); P = 0·050). When this trial was excluded from the analysis the effect of IFN was still significant (risk ratio 0·88 (95 per cent c.i. 0·78 to 1·00); P = 0·042) with little heterogeneity (I2 = 16 per cent, P = 0·213 for χ2). IFN was neither used in a pegylated form nor in combination with other antiviral agents (for example ribavirin). A statistical assessment of the antiviral effect of IFN was not possible because of the incomplete recording of follow-up data on hepatitis virus activity.
Common side-effects after initial IFN injection included fever, chills, myalgia and headache. These affected most patients in all of the studies. A dose reduction of more than 50 per cent because of severe side-effects, such as haematological disorders, malaise, hyperthyroidism, hepatotoxicity and depression was required in up to 25 per cent of the patients (Table3). Of 328 patients treated with IFN, 37 (11·3 per cent) developed severe side-effects requiring a discontinuation of IFN treatment (Table3). The rate of treatment discontinuation ranged from 8 to 20 per cent.
|No. receiving IFN treatment||No. with > 50 per cent dose reduction*||No. who discontinued therapy|
|Ikeda et al.14||10||0 (0)||1 (10)|
|Kubo et al.15||15||0 (0)||3 (20)|
|General fatique, depression, renal abscess|
|Shiratori et al.16||49||6 (12)||7 (14)|
|Cause not reported||Depression, thrombocytopenia, intolerance|
|Lin et al.17||20||5 (25)||3 (15)|
|Thrombocytopenia, neutropenia||Malaise, depression, arthralgia|
|Sun et al.22||118||0 (0)||14 (11·9)|
|Fever, haematological disorders, hyperthyroidism|
|Mazzaferro et al.23||76||3 (4)||6 (8)|
|Mild toxicity||Malaise, neutropenia, hyperthyroidism, deterioration of Child score|
|Lo et al.24||40||0 (0)||3 (8)|
|Hepatotoxicity, convulsion, lymphopenia|
This review demonstrates a significant benefit of IFN after HCC resection or ablation, with an incidence of severe side-effects ranging between 8 per cent24 and 25 per cent17. The strength of the present analysis, based on trials of moderate-to-good methodological quality, is that the pooled risk ratios were based on low statistical heterogeneity, particularly regarding survival (0 per cent for survival, 44 per cent for recurrence). A further strength of the study is the restriction of the meta-analysis to 2-year mortality; no trial suffered from dropouts up to this point in time and the mortality data are comparable across trials.
Patients treated by both surgical resection and ablation of HCC lesions (predominantly by ethanol injection) were included within the individual trials. Although treatment strategies for HCC vary throughout the world7, partial hepatectomy remains the ‘gold standard’ approach in the absence of extrahepatic disease, with a reported 5-year survival rate of 50 per cent6, 8. However, many patients are not suitable candidates for surgery because of advanced disease or impaired liver function. Ablative procedures achieving complete necrosis of small tumours25, 26 are widely accepted as an alternative curative option in HCC.
The impact of the type of concomitant hepatitis infection on the response rate to IFN could not be assessed convincingly. The RCT of Sun and colleagues22 included only HBV infections. This study demonstrated a significant overall survival effect of IFN-α after 2 years, but disease-free survival and recurrence rates were not statistically different. Another trial by Lo and co-workers24 involving predominantly HBV infections failed to show significant benefit. It is noteworthy that four trials included predominantly HCV infections14–16, 23. Mazzaferro and colleagues23 demonstrated a significant survival effect exclusively for patients with HCV. Recurrence rates were significantly improved in the relatively small studies of Lin and co-workers17 and Ikeda and colleagues14.
IFNs possess several properties including antiviral, immunomodulatory, antiproliferative and antiangiogenic effects11–13. Such activities could explain why the beneficial effect of IFN treatment calculated in this meta-analysis is greater for survival than for tumour recurrence. In particular, the antiviral effect might delay further progression of cirrhosis and deterioration of liver function. Both features might have a more significant impact on survival than recurrence. As six of the seven trials used IFN-α, it is interesting that the only study on IFN-β14 showed the largest beneficial effect on tumour recurrence. Although this study is small, further clinical evaluation of IFN-β in the adjuvant setting of HCC treatment seems to be indicated.
Side-effects of IFN-α are dose dependent and often serious. Lo and co-workers reported that minor influenza-like side-effects would have been so frequent that double-blinding in their study was not possible24. In the present analysis, severe adverse effects of the adjuvant IFN treatment leading either to treatment disruption or dose reduction occurred in up to a quarter of the patients. This high rate of side-effects was statistically addressed by an intention-to-treat modality. The relationship between adverse effects and dosage of IFN is supported in the studies of Lo and colleagues24 in which an additional high-dose arm (30 million units, three times a week) had to be discontinued. A greater tolerance of IFN dose is noted in some Asian trials14–16, 24, and there might be a different perception of adverse events in the Asian compared with the European population27. In accordance with this, Lo and co-workers24 used the highest IFN dosage (10 million units three times weekly) of all of the included studies and produced the lowest rate of treatment discontinuation (8 per cent) among all trials.
As a limitation, the present meta-analysis included only the early follow-up phases. The differentiation into ‘early’ and ‘late’ recurrences of HCC was proposed by Llovet and colleagues28. It is assumed that early recurrences (within the first 2 years) may be the result of subclinical residual hepatic tumours after a presumed R0 resection, whereas later recurrences probably represent new tumours20, 29, 30. Owing to the variable median follow-up of the individual studies (between 25 months14 and more than 7 years16), a meta-analysis with follow-up over 2 years would have been difficult because of incomplete and inconsistent follow-up data across studies. The influence of IFN on the incidence of de novo tumours was not, therefore, assessed in the present study. Another limitation is that two of seven studies were subject to methodological weakness. However, because of their small number of patients (20 and 30 patients respectively14, 17) they had only a marginal effect on the overall analysis.
The present analysis demonstrates that IFN improves outcome after curative treatment of HCC, but severe side-effects are common. Nevertheless the data are encouraging in terms of promoting further research on adjuvant immunotherapy. In particular, work is needed to optimize the type and dosage of IFN to minimize side-effects, and to study the combination of IFN treatment with other (neo)adjuvant agents.
The authors declare no conflict of interest. S. Breitenstein and D. Dimitroulis contributed equally to this work.
- 1Global epidemiology of hepatocellular carcinoma. In Liver Cancer, KokudaET (ed). Churchill Livingstone: New York, 1997; 13–28..
- 14Interferon beta prevents recurrence of hepatocellular carcinoma after complete resection or ablation of the primary tumour—a prospective randomized study of hepatitis C virus-related liver cancer. Hepatology 2000; 32: 228–232., , , , , et al.