Potential conflict of interest: Yao-Chun Hsu reports having received lecture fees from Merck Sharp & Dohme (Taiwan), Roche (Taiwan), and Bristol-Myers Squibb (Taiwan). There is no other conflict of interest to declare.
This work is supported by Taiwan's National Health Research Institutes (PH-100-PP-54, PH-101-PP-23) and based on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, Department of Health and managed by the National Health Research Institutes. The interpretations and conclusions contained herein do not represent those of the Bureau of National Health Insurance, Department of Health or the National Health Research Institutes. URL: http://www.nhri.org.tw/nhird/en/index.htm.
Preliminary results of this study were presented at the 63rd annual meeting of the American Association for the Study of Liver Diseases (the Liver Meeting ® 2012) on November 12, 2012; Boston, MA.
Hepatocellular carcinoma (HCC) frequently recurs after surgical resection. This population-based research aimed to investigate the association between postoperative antiviral treatment and risk of recurrent HCC in patients with hepatitis C virus (HCV) infection. By analyzing the Taiwan National Health Insurance Research Database, we initially screened a total of 100,938 patients diagnosed with HCC for the first time between October 2003 and December 2010. Among 2,237 antiviral-naïve HCV-infected patients with curatively resected HCC, there were 213 patients receiving antiviral treatment with pegylated interferon plus ribavirin for 16 weeks or more after surgery (treated cohort). These treated patients were matched 1:4 with 852 controls who were never treated for HCV infection (untreated cohort) by age, gender, cirrhosis, and the elapsed time between surgery and antiviral therapy. Cumulative incidences of and hazard ratios for recurrent HCC were calculated after adjusting for competing mortality. The recurrence rate of HCC was significantly lower in the treated than untreated cohort, with 52.1% (95% confidence interval [CI], 42.0-62.2%) and 63.9% (95% CI, 58.9-68.8%) after 5 years of follow-up, respectively (P = 0.001). The number needed to treat for one fewer recurrent HCC at 5 years was 8. The association between postoperative antiviral treatment and risk of recurrent HCC was independent of adjustment for multiple covariates, with an adjusted hazard ratio of 0.64 (95% CI, 0.50-0.83). Stratified analyses revealed that the attenuation in recurrence risk was greater in patients younger than 60 years and those without cirrhosis or diabetes. Conclusion: Postoperative pegylated interferon plus ribavirin is associated with reduced recurrence of HCC in patients with HCV infection. Age, liver cirrhosis, and diabetes mellitus appear to modify this association. (HEPATOLOGY 2013)
Hepatocellular carcinoma (HCC) is the third most lethal cancer worldwide, causing ∼600,000 deaths every year. The incidence is highest in Eastern Asia and sub-Saharan Africa, but appears to be on the rise in North America.1, 2 Almost all HCCs occur in the background of chronic liver diseases that include viral hepatitis, alcoholic liver disease, and steatohepatitis.3 Chronic infection with hepatitis B virus (HBV) or hepatitis C virus (HCV) accounts for most HCCs; chronic hepatitis C (CHC) is the leading etiology in countries where the prevalence of HBV infection is low.4, 5
Surgical resection is potentially curative for HCC and has been recommended as the treatment of choice if the hepatic reserve permits complete resection.6, 7 Nevertheless, recurrence is very common and strikes 50%-60% of patients 3 years after operation.8, 9 In addition to insidious intrahepatic spreading prior to surgery, a large proportion of recurrent HCCs originate from de novo tumor clones distinct to the resected ones.10, 11 This may result from the underlying liver disease that continuously promotes hepatocellular carcinogenesis despite removal of the primary tumor. There remains a huge unmet need for effective therapy to prevent postoperative recurrence.6, 7
Antiviral therapy may reduce the risk of HCC in patients with chronic viral hepatitis through elimination of viral oncoprotein, resolution of hepatic inflammation, and amelioration of the carcinogenic microenvironment.12-14 A growing body of evidence has indicated that interferon-based antiviral regimen may decrease development of HCC in CHC patients, particularly in those achieving sustained virological response.15, 16 Nonetheless, viral clearance cannot prevent all HCCs, especially in those of old age or with severe liver fibrosis,17 indicating that antiviral therapy may be too late to halt hepatocarcinogenesis in patients with advanced disease. The recurrence rate after HCC resection remained unknown in CHC patients receiving postoperative pegylated interferon (peg-interferon) plus ribavirin, the standard anti-HCV regimen for a decade.18 Moreover, it has not been clarified whether this antiviral regimen administered postoperatively was associated with fewer HCC recurrences. Therefore, we aimed in this population-based study to determine the recurrence rate of surgically resected HCC after postoperative administration of peg-interferon plus ribavirin, and to elucidate whether this antiviral therapy was associated with reduced recurrence of HCC in CHC patients.
CHC, chronic hepatitis C; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; NHIRD, Taiwan National Health Insurance Research Database; NSAID, nonsteroidal antiinflammatory drug; RCIPD, Registry for Catastrophic Illness Patient Database.
Patients and Methods
Study Design and Patient Population.
This open-cohort research utilized population-based data from the Taiwan National Health Insurance Research Database (NHIRD). Since National Health Insurance is a compulsory universal program for all residents in Taiwan, NHIRD is a comprehensive healthcare database that nearly covers the entire 23.7 million population of this country. Details regarding NHIRD have been reported in our previous investigations.19-21 The present study was approved by the Research Ethics Committee of the National Health Research Institutes, Taiwan (EC1010303-E).
We first screened all patients who had a first-time diagnosis of HCC from October 1, 2003, to December 31, 2010, and then identified the study population as those with CHC who underwent curative surgery. This research defined disease status principally on the basis of admission diagnoses, which were coded according to the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). Apart from the specific ICD-9-CM code (155.0), the diagnosis of HCC had to be certified in the Registry for Catastrophic Illness Patient Database (RCIPD), a subpart of NHIRD. Given that all enrolled patients had their HCC resected, histopathological confirmation was required for registry in the RCIPD. All enrolled patients received liver resection as the sole HCC treatment. Those who underwent liver transplantation, local ablation (ethanol injection, radiofrequency ablation, or microwave coagulation), or transarterial chemoembolization before or during the index admission were excluded.
Patients with metastasis or any other malignant disease were excluded. We enrolled exclusively patients coded with CHC at admission (ICD-9-CM codes: 070.41, 070.44, 070.51, 070.54, V02.62) to ascertain validity of the diagnosis. Those with HBV infection or other viral hepatitis were excluded (ICD-9-CM codes: 070.2, 070.3, V02.61, V02.69).
Antiviral Therapy and Definition of Study Cohorts.
The antiviral regimen consisted of peg-interferon alpha (either 2a or 2b) plus ribavirin, which has been reimbursed for HCV infection in Taiwan since October 1, 2003. Generally, treatment was initiated at 180 μg per week irrespective of body weight for peg-interferon alpha 2a, 1.5 μg/kg per week for 2b, and 800 to 1,200 mg per day for ribavirin, but it was individualized at the treating physician's discretion and frequently adjusted along the course. The reimbursed duration ranged from 16 weeks to 48 weeks, depending on the date of administration, viral genotype, serum viral load, on-treatment virological response, and patient tolerability.18
The treated cohort comprised antiviral-naïve patients who received peg-interferon and ribavirin for a minimum of 16 weeks after surgery. Each treated patient was matched in age, gender, and cirrhosis with four untreated counterparts randomly selected from those who never used interferon or ribavirin. Furthermore, the untreated controls were deliberately matched for the time period between surgery and administration of antiviral therapy in treated patients in order to eliminate the immortal time bias.22, 23
Postoperative Follow-up and Definition of HCC Recurrence.
The treated and untreated cohorts were followed up after initiation of antiviral regimen and matched postoperative duration, respectively, until recurrence of HCC, death, or December 31, 2010, whichever occurred first. Recurrence of HCC was defined as repeated cancer treatment for HCC during the follow-up period. Treatment modalities for HCC recurrence included liver transplantation, surgical resection, focal ablation, transarterial chemoembolization, radiotherapy, and chemotherapy. HCC that recurred within 3 months of the index surgery was not included because it might arise from incomplete primary resection.
Assessment of and Adjustment for Confounding Factors.
All comorbidities listed in the Charlson's index were considered as important covariates that might confound outcomes.24 The age-unadjusted Charlson scores were computed for both the treated and untreated cohorts. Certain medications including statin, nonsteroidal antiinflammatory drug (NSAID), aspirin, and metformin were also assessed as potential confounders because they might modify the risk of cancer.19-21 Users of these drugs were defined as those who took them on a regular basis with frequency of more than one tablet per month during the study period. The extent of hepatic surgery, namely, major (at least three segments of hepatic parenchyma) or minor resection (two or fewer segments of liver), was also analyzed.
Data Analysis and Statistical Test.
The primary and secondary outcomes were HCC recurrence and mortality, respectively. Death occurring prior to HCC recurrence, which could lead to informative censoring, was regarded as a competing risk event in estimating the incidence of recurrent HCC. The number needed to treat (NNT) represented the number of patients needed to be treated in association with one fewer recurrent HCC or death. NNT was calculated by the inverse of the absolute risk reduction. The modified Kaplan-Meier method and Gray's method were used to calculate and to compare the cumulative incidences in data with competing risks.25 After confirming the assumption of proportional hazards by plotting the graph of the survival function versus the survival time and the graph of the log (-log(survival)) versus the log of survival time, we applied the modified multivariate-adjusted Cox proportional hazard model in the presence of competing risks to examine the independent risk factors for HCC recurrence.26 The influence of antiviral therapy on HCC recurrence was further explored in stratified analyses according to age, gender, cirrhosis, comorbidity, medications, and extent of resection. All data were managed with SAS software 9.2 v. (SAS Institute, Cary, NC). The cumulative incidence and hazard ratio (HR) in the competing risk analysis were calculated using the R software with the “cmprsk_2.1-4” package (by Gray; http://biowww.dfci.harvard.edu/∼gray/). Calculated results were expressed with the estimated numbers alongside their 95% confidence intervals (CIs). All statistical tests were two-sided with significance set at P < 0.05.
Baseline Characteristics of the Study Population.
We screened a total of 100,938 patients diagnosed with HCC for the first time during the study period and finally identified 2,237 CHC patients who underwent curative resection for HCC (Fig. 1). Among the 239 patients who ever received peg-interferon plus ribavirin after surgery, 213 patients (89.1%) were treated for a minimum of 16 weeks and formed the treated cohort, whose mean (± standard deviation) duration of antiviral regimen was 25.99 ± 8.13 weeks and that of follow-up was 2.01 ± 1.67 years. The matched controls accordingly comprised 852 untreated patients randomly selected from those not receiving antiviral therapy. The untreated cohort was followed up for 1.51 ± 1.28 years. These two cohorts were generally comparable in baseline characteristics (Table 1).
Table 1. Baseline Characteristics of the Study Cohorts
Untreated Cohort (n = 852)
Treated Cohort (n = 213)
Data are expressed as median (interquartile range) or number (percentage). NSAID, nonsteroidal antiinflammatory drug.
20-29 years, n (%)
30-39 years, n (%)
40-49 years, n (%)
50-59 years, n (%)
60-69 years, n (%)
≧70 years, n (%)
Male gender, n (%)
Extent of surgery
Major coexisting diseases, n (%)
Coronary artery disease
Cerebral vascular accident
Chronic obstructive lung disease
Peptic ulcer disease
NSAIDs or aspirin
HCC Recurrence and Mortality Between the Study Cohorts.
HCC recurred cumulatively in 16.2% (95% CI, 10.9-21.4%), 41.8% (95% CI, 33.2-50.4%), and 52.1% (95% CI, 42.0-62.2%) of the treated cohort after 1, 3, and 5 years of follow-up, respectively (Fig. 2). The corresponding 1-, 3-, and 5-year cumulative incidences in the untreated cohort were 24.5% (95% CI, 21.4-27.5%), 54.3% (95% CI, 50.0-58.6%), and 63.9% (95% CI, 58.9-68.8%), respectively. Therefore, patients receiving a postoperative anti-HCV regimen had a significantly lower recurrence rate (P = 0.001). The unadjusted NNT associated with one fewer HCC recurrences after 1, 3, and 5 years were 12, 8, and 8, respectively (Table 2).
Table 2. Cumulative Incidences of Recurrent HCC and Overall Mortality Between Chronic Hepatitis C Patients With and Those Without Postoperative Antiviral Treatment
Abbreviations: HCC, hepatocellular carcinoma; CI, confidence interval; NNT, number needed to treat.
At 1 year
At 3 years
At 5 years
At 1 year
At 3 years
At 5 years
The treated cohort also had a significantly lower mortality rate as compared with the untreated counterpart (P < 0.001). The 1-, 3-, and 5-year cumulative incidences of mortality were 2.8% (95% CI, 0.4-5.2%), 10.8% (95% CI, 4.9-16.6%), and 15.4% (95% CI, 7.7-23.1%) in the treated patients, and 6.9% (95% CI, 5.1-8.7%), 24.8 (95% CI, 20.9-28.6%), and 47.0% (95% CI, 40.7-53.2%) in the untreated controls (Fig. 3). The unadjusted NNT associated with one less mortality at 1, 3, and 5 years after antiviral treatment were 24, 7, and 3, respectively (Table 2).
Multivariate-Adjusted Association of Antiviral Therapy with HCC Recurrence.
The modified Cox proportional hazard model demonstrated that postoperative antiviral therapy was independently associated with a 36% reduction in hazard of HCC recurrence (adjusted HR, 0.64; 95% CI, 0.50-0.83; P = 0.001) (Table 3). In addition, major surgical resection as compared with a minor one (adjusted HR, 0.76; 95% CI, 0.59-0.97; P = 0.027) and regular NSAID or aspirin use (adjusted HR, 0.80; 95% CI, 0.66-0.97; P = 0.026) were also linked to reduction of recurrent HCC.
Table 3. Results of Multivariate-Adjusted Cox Proportional Hazard Model for Recurrence of Hepatocellular Carcinoma
Adjusted Hazard Ratio
95% Confidence Interval
Age, per year
NSAID or aspirin
Stratified Analyses for Recurrent HCC in Association with Antiviral Therapy.
The association between postoperative antiviral therapy and reduced HCC recurrence was generally consistent across different patient subgroups, since the estimated HRs favored antiviral treatment in all strata (Fig. 4). The reduction of hazard, however, differed in magnitude among patients according to age, liver cirrhosis, diabetes mellitus, and use of metformin. The association was more pronounced in patients younger (adjusted HR, 0.47; 95% CI, 0.30-0.73) versus older than 60 years (adjusted HR, 0.80; 95% CI, 0.59-1.09), without (adjusted HR, 0.56; 95% CI, 0.40-0.80) versus with cirrhosis (adjusted HR, 0.82; 95% CI, 0.56-1.20), and without (adjusted HR, 0.60; 95% CI, 0.45-0.81) versus with diabetes (adjusted HR, 0.86; 95% CI, 0.51-1.44).
This nationwide population-based study revealed a significantly lower risk of recurrent HCC in CHC patients who were treated for their HCV infection postoperatively with peg-interferon plus ribavirin, as compared with those whose CHC was left untreated. The hazard was reduced by 36% (adjusted HR 0.64; 95% CI, 0.50-0.83; P = 0.001) after adjustment for possible confounding. The unadjusted NNT in association with one patient free of recurrent HCC at 1, 3, and 5 years after antiviral treatment were 12, 8, and 8, respectively. Nevertheless, the magnitude of association appeared to differ among patient subgroups, in that the attenuated risk of HCC recurrence was more apparent in younger patients without cirrhosis or diabetes. These findings not only imply that antiviral treatment may still ameliorate hepatocellular carcinogenesis even when HCV infection has progressed to the stage of HCC, but also characterize those who are more likely to benefit from this management.
To date, there has been no adjuvant therapy approved for HCC after curative resection.6, 7 Conventional interferon alpha has been tested for this indication, but the results from randomized trials involving CHC patients were conflicting.27-30 Kubo et al.27 reported in a small trial (N = 30) that postoperative administration of interferon for 2 years decreased recurrence of resected HCC, whereas Mazzaferro et al.28 concluded that adjuvant interferon for 48 weeks could not prevent HCC recurrence in 150 HCV-infected patients with cirrhosis. The other two randomized trials, which recruited predominantly HBV-infected patients along with some CHC patients, also conflicted in the efficacy for secondary prevention of HCC.29, 30 In addition, the therapeutic agent, dosing protocol, patient characteristic, and study endpoint also varied remarkably across these trials. Therefore, conventional interferon cannot be accepted as the standard care following HCC resection in CHC patients,7 despite a positive result from meta-analyses.31
Peg-interferon alpha plus ribavirin has become the standard anti-HCV regimen for a decade,32, 33 but its efficacy in preventing recurrence of curatively treated HCC remains undetermined. Two previous studies addressing this issue did not find peg-interferon-based therapy was associated with fewer recurrences.34, 35 In a cohort study consisting of 182 patients predominantly receiving radiofrequency ablation, Hagihara et al.34 reported HCC recurred similarly between 37 treated and 145 untreated patients (58% versus 70% at 5 years; P = 0.17). By taking a propensity score approach, Tanimoto et al.35 showed that recurrence did not differ between patients with and without postoperative peg-interferon-based treatment (55.3% versus 44.7%; P = 0.36; n = 38 in both groups). Both studies were probably underpowered because of the small number of participants. Besides, differences in demographics, HCC treatment, antiviral medication, outcome definition, and follow-up duration might also be factors in the discrepancy between their results and ours. Based on our data, it needs a large sample comprising representative subgroups to uncover the association between postoperative antiviral treatment and HCC recurrence, in that the recurrence rate among treated patients may be lower but remain substantial and that certain patient characteristics can modify the association.
Peg-interferon plus ribavirin is highly effective in achieving HCV eradication in Taiwan,36, 37 where a favorable genetic variation in IL28B is prevalent,38 and has been validated among Taiwanese patients with HCC in a multicenter trial.39 However, this study in and of itself could not show how virological response might have influenced the association. Because linking the NHIRD to individual patients' laboratory results was forbidden for privacy protection, we were unable to determine whether viral elimination mediated this association. Nevertheless, a large body of evidence has indicated that sustained virological response to antiviral treatment appears essential to reduce risk of developing HCV-related HCC.15, 16 The large-scale randomized and placebo-controlled HALT-C trial also refuted the antitumor efficacy of peg-interferon in CHC patients who failed to eradicate HCV.40 In our opinion, antiviral efficacy was more likely than an antiproliferative property to account for the observed association in this study, although further research is clearly required to clarify the underlying mechanism. Furthermore, different viral clearance rates could plausibly explain why age, cirrhosis, and diabetes modified the association with recurrent HCC, since these were all validated host features predictive of therapeutic response.41-43 Consistent with our previous study which focused on recurrent HCC in patients with HBV infection,21 this study also uncovered an inverse association between the use of aspirin or NSAID and risk of HCC recurrence. The mechanism of this intriguing finding may involve induction of cell cycle arrest and apoptosis in HCC cells,44, 45 and should inspire more investigation.
This study has applied a number of methodological procedures to avoid a biased or confounded result, in addition to adjusting for multiple parameters in the multivariate analyses. First, enrollment was explicitly restricted to patients who could tolerate and recover from liver resection, whose performance status as well as hepatic reserve was therefore unlikely to contraindicate use of interferon and ribavirin. Second, in order to ensure comparability of the study cohorts, enrolled patients were matched by age, gender, and cirrhosis. The treated and untreated cohorts were consequently similar in their baseline characteristics including the comorbidity index, i.e., the Charlson's score. Moreover, matching in the time period from surgery to administration of antiviral therapy prevented the immortal time bias.23 Third, the universal coverage of Taiwan National Health Insurance, which fully reimbursed peg-interferon and ribavirin for treating HCV infection, precluded healthcare accessibility or financial disparity as a determinant for receiving treatment or not. Last, but not least, we recognized how mortality might have confounded the estimation of the association with HCC recurrence.46 Since antiviral treatment could have affected survival by ameliorating the background liver disease, without relation to any effect on HCC, the higher mortality in the untreated patients would have overestimated their HCC recurrence rate and spuriously exaggerated the difference between the study cohorts, had the censoring caused by death been simply dismissed as noninformative.47, 48 All in all, information from these careful analyses should be valuable for physicians and surgeons who need to weigh the pros and cons of using peg-interferon and ribavirin after resection of HCC, even though the observational nature of this study prevented definite ascertainment of the causal relationship.
Several limitations warrant discussion. First, a lack of direct laboratory results in the Taiwan NHIRD prohibited exploration in terms of virological response, viral genotype, baseline viral load, size and number of tumors, and histological differentiation. Second, we were unable to determine the adverse reactions related to peg-interferon and ribavirin. Nevertheless, nearly 90% of those who ever started postoperative antiviral therapy eventually completed a minimum of a 16-week course, indicating tolerability of this regimen in these patients. Besides, a multicenter trial from Taiwan has confirmed its applicability in HCC patients.39 Finally, for the purpose of internal validity, we deliberately matched the postoperative period prior to antiviral treatment and restricted enrollment in patients free of recurrence within 3 months of surgery. Given that the time pattern of recurrence has been shown to correlate with its pathogenesis,49 recurrent HCC in this study might more likely result from de novo carcinogenesis instead of preexistent micrometastasis. We accordingly suggest caution be exercised before extrapolating our findings in the setting of immediate recurrence following resection.
In summary, recurrence of HCV-related HCC after surgical resection is reduced in patients who receive postoperative antiviral therapy with peg-interferon plus ribavirin, as compared with those who never treat their HCV infection. Moreover, greater risk reduction of recurrent HCC is observed in younger patients (<60 years) and those without cirrhosis or diabetes. These results imply that antiviral therapy appears better late than never in CHC patients with curable HCC. How to improve outcomes when the current therapy is either intolerable or ineffective warrants further research.