Potential conflict of interest: Dr. Asahina is on the speakers' bureau of Chugai Pharma and MSD, and received research funding from Daiichi-Sankyo and Chugai Pharma. Dr. Asahina and Dr. Kakinuma belong to a donation-funded department funded by Chugai Pharma, Toray, Bristol-Myers Squibb, Dainippon-Sumitomo Pharma, and MSD.
Supported by grants from the Japanese Ministry of Education, Culture, Sports, Science, and Technology, and the Japanese Ministry of Welfare, Health, and Labor.
Address reprint requests to: Yasuhiro Asahina, M.D., Ph.D., Professor, Department of Liver Disease Control, Department of Gastroenterology and Hepatology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8519, Japan. E-mail: email@example.com; fax: +81-3-5803-0268.
The effects of interferon (IFN) treatment and the post-IFN treatment α-fetoprotein (AFP) levels on risk of hepatocellular carcinoma (HCC) in patients with chronic hepatitis C (CHC) are unknown. To determine the relationship between AFP and alanine transaminase (ALT) levels and HCC risk, a cohort consisting of 1,818 patients histologically proven to have CHC treated with IFN were studied. Cumulative incidence and HCC risk were analyzed over a mean follow-up period of 6.1 years using the Kaplan-Meier method and Cox proportional hazard analysis. HCC developed in 179 study subjects. According to multivariate analysis, older age, male gender, advanced fibrosis, severe steatosis, lower serum albumin levels, non sustained virological response (non-SVR), and higher post-IFN treatment ALT or AFP levels were identified as independent factors significantly associated with HCC development. Cutoff values for ALT and AFP for prediction of future HCC were determined as 40 IU/L and 6.0 ng/mL, respectively, and negative predictive values of these cutoffs were high at 0.960 in each value. The cumulative incidence of HCC was significantly lower in patients whose post-IFN treatment ALT and AFP levels were suppressed to less than the cutoff values even in non-SVR patients. This suppressive effect was also found in patients whose post-IFN treatment ALT and AFP levels were reduced to less than the cutoff values despite abnormal pretreatment levels. Conclusion: Post-IFN treatment ALT and AFP levels are significantly associated with hepatocarcinogenesis. Measurement of these values is useful for predicting future HCC risk after IFN treatment. Suppression of these values after IFN therapy reduces HCC risk even in patients without HCV eradication. (Hepatology 2013;58:1253–1262)
aspartate aminotransferase to platelet ratio index
chronic hepatitis C
CT computed tomography; γ-GTP
Hepatitis C virus
magnetic resonance imaging
receiver operator characteristic
sustained virological response.
Hepatocellular carcinoma (HCC), one of the most frequent primary liver cancers,[1, 2] is the third most common cause of cancer mortality worldwide. Hepatitis C virus (HCV) infection is a common cause of chronic hepatitis, which progresses to HCC in many patients. In the last two decades, interferon (IFN) therapy has been used to treat chronic hepatitis C (CHC) with the goal of altering the natural history of this disease. Although HCV eradication with IFN therapy for CHC has been shown to prevent HCC,[5-9] HCC sometimes develops even after achieving viral eradication. Because the number of sustained virological responders (SVRs) is increasing along with recent advances in the development of effective anti-HCV therapy, it is very important to determine factors responsible for HCC development among IFN-treated patients. However, this information is difficult to determine because of the paucity of large-scale, long-term cohort studies.
The 70-kDa glycoprotein α-fetoprotein (AFP), encoded by a gene located on chromosome 4, is the major serum protein during fetal life. Shortly before birth, AFP is replaced by albumin as the major serum protein,[11, 12] and thereafter, serum AFP levels remain extremely low throughout life (<10 ng/mL). Because serum AFP levels are frequently elevated in patients with HCC and germ-cell tumors, measurement of AFP is widely used as a serological marker for these tumors.[8, 13] However, AFP levels are sometimes elevated in patients with chronic viral hepatitis and cirrhosis who do not have HCC.[3, 19] While one possible explanation for this elevation is liver inflammation, in patients with CHC, the relationship between AFP and markers of liver inflammation such as alanine aminotransferase (ALT) is unclear. Moreover, although several reports suggest that pre-IFN treatment ALT and AFP levels in patients or those in patients who did not undergo subsequent treatment are associated with the development of HCC, it is unclear whether post-IFN treatment ALT and AFP levels are associated with hepatocarcinogenesis in patients with CHC. Hence, to clarify these associations we conducted a large-scale, long-term cohort study of patients with CHC to analyze the influence of ALT and AFP levels before and after IFN therapy on hepatocarcinogenesis in addition to other host and virological factors.
Patients and Methods
Patients chronically infected with HCV who had histologically proven chronic hepatitis or cirrhosis and had undergone IFN treatment between 1992 and 2010 were enrolled in the cohort. HCC was definitively ruled out by ultrasonography, dynamic computed tomography (CT), and/or magnetic resonance imaging (MRI) on enrollment. Patients were excluded if they had a history of HCC at the time of liver biopsy, autoimmune hepatitis, primary biliary cirrhosis, excessive alcohol consumption (≥50 g/day), hepatitis B surface antigen, or antihuman immunodeficiency virus antibody. Based on these criteria, a total of 2,689 patients were initially enrolled. Of these, 223 (8.3%) patients were excluded from the cohort because of loss to follow-up. In the remaining 2,466 patients, 133 and 515 patients were excluded from this analysis because of short follow-up and retreatment with IFN-based therapy during the follow-up period, respectively. Thus, the cohort comprising 1,818 patients was analyzed in the present study. Written informed consent was obtained from all patients and the Ethical Committee of Musashino Red Cross Hospital approved this study, which was conducted in accordance with the Declaration of Helsinki.
To obtain liver specimens, laparoscopic or ultrasound-guided liver biopsies were performed with 13G or 15G needles, respectively. The median length of specimen was 18 mm (range, 11-41 mm), and the mean number of portal tracts was 17 (range, 9-35). The stage of fibrosis and the grade of inflammatory activity were scored by two pathologists according to the classification of Desmet et al. The percentage of steatosis was quantified by determining the average proportion of hepatocytes affected.
IFN Therapy and Definitions of Response to IFN Therapy
All patients had chronic HCV infection at liver biopsy, which was confirmed by the presence of HCV-RNA in serum. All IFN therapies were initiated within 48 weeks after liver biopsy. Among the 1,818 patients, 535 received IFNα or IFNβ monotherapy for 24 weeks, 244 patients received IFNα ribavirin (RBV) combination therapy for 24 weeks, 299 patients received pegylated (PEG-) IFNα monotherapy for 48 weeks, and 760 patients received PEG-IFNα RBV combination therapy for 48-72 weeks.
Patients negative for serum HCV-RNA 24 weeks after IFN therapy completion were defined as SVRs. Patients who remained positive for HCV-RNA 24 weeks after therapy completion were defined as non-SVRs. HCV-RNA was determined by the qualitative Amplicor or TaqMan HCV assay (Roche Molecular Diagnostics, Tokyo, Japan).
Data Collection and Patient Follow-up
At enrollment, patient characteristics, biochemical, hematological, virological, and histological data were collected. Age was determined at the time of primary liver biopsy. Patients were examined for HCC by abdominal ultrasonography, dynamic CT, and/or MRI every 3-6 months. Serum ALT and AFP levels were measured every 1-6 months. The surveillance protocols were in accordance with the standard of care in Japan. If HCC was suspected on the basis of the screening examination, additional procedures (e.g., dynamic CT, dynamic MRI, CT during hepatic arteriography, CT during arterial portography, contrast-enhanced ultrasonography, and tumor biopsy) were used to confirm the diagnosis. HCC diagnosis was confirmed by needle biopsy, histology of surgically resected specimens, or characteristic radiological findings. To evaluate the effects of changes in serum ALT and AFP levels during IFN therapy on hepatocarcinogenesis, the average integration values of ALT and AFP in each patient were calculated before and after IFN therapy. Data obtained more than 1 year prior to HCC development were used to exclude AFP elevation caused by HCC itself.
Follow-up was between the date of primary liver biopsy and HCC development or the last medical attendance until June 2011. The mean follow-up period was 6.1 years (range, 1.0-20.8 years).
Categorical data were compared by the chi-square test or Fisher's exact test. Distributions of continuous variables were analyzed with Student t test for two groups. All tests of significance were two-tailed and P < 0.05 was considered statistically significant. The cumulative incidence curve was determined by the Kaplan-Meier method, and differences among groups were assessed using the log-rank test. Factors associated with HCC risk were determined by the Cox proportional hazard model. As covariates in the multivariate stepwise Cox model, age, sex, stage of liver fibrosis, grade of histological activity, presence of hepatic steatosis, serum albumin levels, γ-glutamyl transpeptidase (γ-GTP) level, fasting blood sugar levels, platelet counts, pre-IFN ALT levels, pre-IFN AFP levels, post-IFN ALT levels, post-IFN AFP levels, and virological response were included. HCC development was the dependent variable. Time zero was defined as the time of primary liver biopsy. The proportional assumption was supported by log[-log(survival)] versus log(time) plots that showed parallel lines. Statistical analyses were performed using the Statistical Package for the Social Sciences software v. 18.0 (SPSS, Chicago, IL).
Patient Characteristics and Factors Associated With Risk of HCC
Table 1 shows patient characteristics at the time of enrollment. During follow-up, HCC developed in 179 patients. The cumulative incidence of HCC for 5 and 10 years was 6.5% and 15.0%, respectively. The final virological response to IFN therapy was determined in all patients. The overall rate of SVRs was 50.2% (913/1818). The cumulative incidence in SVRs was 2.3% and 5.5%, respectively, which was significantly lower than that in non-SVRs (6.9% and 21.9%, respectively; log-rank test, P < 0.0001).
Table 1. Characteristics of Patients Enrolled in the Present Study
Unless otherwise indicated, data are given as mean (SD).
Abbreviations: BMI, body mass index; ALT, alanine aminotransferase; γ-GTP, γ-glutamyl transpeptidase; LDL, low-density lipoprotein; AFP, α-fetoprotein; WBC, white blood cell; Hb, hemoglobin; HCV, hepatitis C virus; ISDR, interferon sensitivity determining region; IFN, interferon; RBV, ribavirin; PEG, pegylated.
HCV genotype was determined in 1755 patients.
HCV core mutation was determined in 409 patients with genotype 1b.
ISDR was determined in 1264 patients with genotype 1b.
Univariate analysis demonstrated factors that increase the risk for HCC development (Table 2). According to multivariate stepwise Cox analysis, older age, male gender, advanced fibrosis, severe steatosis, lower serum albumin levels, non-SVR, and higher post-IFN treatment ALT and AFP levels, but not pre-IFN treatment ALT and AFP levels, were identified as independent factors that were significantly associated with HCC development (Table 2).
Table 2. Factors Associated With Hepatocellular Carcinoma
Hazard Ratio (95% CI)
Hazard ratios for development of hepatocellular carcinoma were calculated by the Cox proportional hazards analysis. ALT, alanine aminotransferase; γ-GTP, gamma-glutamyl transpeptidase; AFP, alpha-fetoprotein; ISDR, interferon sensitivity determining region; SVR, sustained virological responder. As covariates in the multivariate stepwise Cox model, age, sex, stage of liver fibrosis, grade of histological activity, presence of hepatic steatosis, serum albumin levels, γ-GTP level, fasting blood sugar levels, platelet counts, pre-IFN ALT levels, pre-IFN AFP levels, post-IFN ALT levels, post-IFN AFP levels, and virological response were included.
Age (by every 10 year)
Degree of steatosis
Albumin (by every 1 g/dL)
Pre-ALT (by every 40 IU/L)
Post-ALT (by every 40 IU/L)
γ-GTP (by every 40 IU/L)
Fasting blood sugar (by every 100 mg/dL)
Pre-AFP (by every 10 ng/mL)
Post-AFP (by every 10 ng/mL)
Platelet counts (by every 104/μL)
Core 70 mutation
More than 1 mutation
Wild or 1 mutation
Age (by every 10 year)
Degree of steatosis
Albumin (by every 1 g/dL)
Post-ALT (by every 40 IU/L)
Post-AFP (by every 10 ng/mL)
Association of Post-IFN Treatment ALT and AFP Levels with HCC Development in SVRs and Non-SVRs
Because our multivariate analysis identified post-IFN treatment ALT and AFP levels as independent factors associated with HCC risk, we determined the cutoff values of these factors for predicting the development of HCC by receiver operator characteristics (ROC) analysis. The area under the ROC curve for post-IFN treatment ALT and AFP levels were higher than that for pre-IFN treatment ALT and AFP levels, suggesting that quantification of post-IFN treatment ALT and AFP levels rather than pre-IFN treatment levels of these values is useful for predicting HCC (Fig. 1A). From this ROC analysis, ALT <40 IU/L and AFP <6.0 ng/mL were identified as cutoff values. Negative predictive values were extremely high at 0.960 in each value, suggesting patients with ALT and/or AFP levels below these cutoff values are at a lower risk for HCC.
As shown in Fig. 1B, the hazard ratio determined by Cox proportional hazard analysis after adjustment for age, sex, stage of liver fibrosis, degree of liver steatosis, serum albumin levels, and virological response to therapy demonstrated that the hazard ratio for HCC was dependent on post-IFN treatment ALT and AFP levels. These hazard ratios increased predominantly when post-IFN treatment ALT and AFP levels were more than the cutoff values.
As shown in Fig. 2, the cumulative incidence of HCC was closely related to post-IFN treatment ALT and AST levels and was significantly lower in patients whose post-IFN treatment ALT and AFP levels was suppressed to <40 IU/L and 6.0 ng/mL, respectively. This suppressive effect was also notable in non-SVRs (Fig. 2C,D). Moreover, the cumulative incidence of HCC was significantly higher even in SVRs whose post-IFN treatment ALT and AFP levels were not <40 IU/L and 10 ng/mL, respectively (Fig. 2E,F).
Changes in ALT and AFP Levels With IFN Therapy and HCC Development
In the entire cohort, the mean ALT and AFP levels significantly decreased after IFN therapy (ALT = 78.4 to 36.6 IU/L, 95% confidence interval [CI] = 38.6-45.0, P < 0.0001; AFP = 11.3 to 6.9 ng/mL, 95% CI = 3.25-5.69, P < 0.0001; paired Student t test), and this significant decrease was found not only in SVRs, but also non-SVRs (Fig. 3A). Because post-IFN treatment ALT and AFP levels rather than pre-IFN treatment levels were significantly associated with the development of HCC in non-SVRs, we determined the effects of changes in ALT and AFP levels by IFN therapy on hepatocarcinogenesis. Even in non-SVRs with equal or higher pre-IFN treatment ALT or AFP levels than the cutoff values, the cumulative incidence of HCC was significantly suppressed in patients whose post-IFN treatment ALT or AFP level was reduced to less than the cutoff values (Fig. 3B). In contrast, persistence of post-IFN treatment ALT or AFP levels to more than the cutoff values after IFN therapy was associated with a significantly higher incidence of HCC (Fig. 3B).
Relation Between AFP and ALT or Histological Findings
Univariate analysis using logistic regression determined factors that were associated with post-IFN treatment ALT or AFP levels (Supporting Table). Although many clinical factors were associated with post-IFN ALT and/or AFP levels, post-IFN ALT and AFP levels were not correlated with each other (r2 = 0.050). Therefore, the cumulative incidence of HCC was significantly higher in patients with higher post-IFN treatment AFP levels, even when patients were stratified by post-IFN treatment ALT levels (Fig. 4A,B).
As shown in Fig. 4C-F, the cumulative incidence of HCC development was significantly lower in patients whose post-IFN treatment AFP level was <6.0 ng/mL in all subgroups stratified by stage of fibrosis and grade of activity. Therefore, reduction in post-IFN treatment AFP levels reduces HCC risk even in patients with advanced fibrosis. Although pre-IFN treatment AFP levels correlated with the advance of histological fibrosis and grade of activity, such correlations became less significant with post-IFN treatment AFP levels (data not shown).
Platelet Counts and Aspartate Aminotransferase-to-Platelet Ratio Index (APRI) in Patients Without Advanced Fibrosis
Because a substantial amount of HCC cases developed in the patients without histologically advanced fibrosis (Fig. 4C), we characterized these individuals using platelet counts and APRI, which are the readily available surrogate markers for liver fibrosis. We first determined the cutoff values of platelet counts and APRI for predicting HCC development by ROC analyses. Accordingly, platelet counts <150 × 103/μL and APRI >0.96 were identified as cutoff values, and the areas under the ROC curve for platelet counts and APRI were 0.715 (95% CI: 0.675-0.755) and 0.740 (95% CI: 0.701-0.779), respectively (Supporting Figure). Even in individuals without advanced fibrosis (F1 and F2 patients), the proportion of patients with platelet counts <150 × 103/μL or APRI >0.96 was significantly higher in patients with HCC than in those without HCC (platelet counts, 53.0% [35/66] versus 31.3% [387/1238], P = 0.0002; APRI, 53.0% [35/66] versus 26.4% [325/1229], P < 0.0001). Moreover, the cumulative incidence of HCC development was significantly higher in patients with platelet counts <150 × 103/μL or APRI >0.96 in the subgroups without advanced fibrosis (Supporting Figure). Therefore, patients with low platelet counts or high APRI still have a substantial risk for HCC development even though they were diagnosed with mild fibrosis by liver biopsy.
Hepatic Steatosis and Post-IFN ALT and AFP Levels in SVRs
To characterize SVRs without ALT and AFP normalization after IFN therapy, we evaluated the percentage of severe hepatic steatosis in these patients. Accordingly, the percentages of severe hepatic steatosis were significantly higher in SVRs without ALT and AFP normalization than in those with normal ALT and AFP (ALT, 37.9% [36/95] versus 13.8% [77/557], P < 0.0001; AFP, 31.6% [31/98] versus 14.8% [82/554], P < 0.0001). Therefore, it is likely that presence of hepatic steatosis is associated with ALT and/or AFP elevation, and it is one of the risks for HCC development even after achieving SVR.
This large-scale, long-term cohort study establishes important findings, which demonstrate a strict association between hepatocarcinogenesis and post-IFN treatment ALT and AFP levels in patients with CHC. This association was notable in both SVR and non-SVR subgroups, and suppression of these values by IFN therapy reduced the hepatocarcinogenesis risk despite failure of HCV eradication. These data, which demonstrate the efficacy of IFN against HCC development associated with suppression of AFP, have clinically important implications for physicians.
Although there have been reports on the association between baseline pretreatment AFP levels and HCC risk,[26-35] little is known regarding the effects of IFN therapy on change in post-IFN treatment AFP and its relation to HCC risk. Although a previous report demonstrated that a decrease in AFP levels in patients receiving IFN therapy reduced the incidence of HCC, this study was performed in a small number of patients (n = 382), and cutoff values, relation to ALT, or histological findings were not determined. Our study, performed in a large well-characterized cohort, had a greater advantage in that it allowed determination of cutoff values for post-IFN treatment ALT and AFP levels useful for predicting HCC development. Although a higher cutoff value of 20 ng/mL was used to determine the incidence of HCC in the previous study, we propose a lower value for negatively predicting HCC. From our results, those with AFP levels ≥6.0 ng/mL have a substantial HCC risk, even if it is <20 ng/mL. Therefore, post-IFN treatment AFP levels should be <6.0 ng/mL to suppress HCC risk in patients with CHC.
It should be noted that AFP produced by HCC itself was carefully excluded in our study. Serum AFP elevation is frequently observed in patients with advanced CHC in the absence of HCC.[19-23] Although the precise mechanisms accounting for this observation are unknown, Hu et al. found a correlation between AFP and measures of liver disease activity, suggesting that AFP production is enhanced in the presence of necroinflammatory injury of the liver. However, in our study post-IFN treatment ALT and AFP levels were not correlated, and the cumulative incidence of HCC was significantly higher in patients with higher post-IFN treatment AFP levels, even when patients were stratified by post-IFN treatment ALT levels. Moreover, multivariate analysis confirmed that AFP and ALT are independently associated with HCC risk. Therefore, observed elevation in AFP levels in patients with subsequent HCC development is not necessarily caused by necroinflammation of the liver. Alternatively, increased AFP levels have been reported during liver regeneration following hepatic resection and during recovery from massive hepatic necrosis,[39-41] suggesting that elevated AFP levels are a surrogate for proliferative activity of liver cells, which may cause hepatocarcinogenesis in patients with CHC.
Other possible reasons accounting for HCC risk related to AFP are the close association between AFP levels and the stage of liver fibrosis, which is consistent with a previous report. However, we further clarified the fact that correlation between post-IFN treatment AFP levels and liver fibrosis was less notable in patients without subsequent development of HCC (data not shown). Cumulative incidence of HCC was significantly higher in patients with higher post-IFN treatment AFP levels at each stage when patients were stratified by the histological stage of fibrosis (Fig. 4). Therefore, post-IFN treatment AFP is not just a surrogate marker for liver fibrosis, and elevation of post-IFN treatment AFP as a potential risk for hepatocarcinogenesis is not only the result of advanced liver fibrosis. Conversely, suppression of post-IFN treatment AFP levels may reduce HCC risk even in patients with advanced fibrosis.
This study has a few limitations, the first being the heterogeneity of our cohort, which included various treatment regimens with different treatment responses. However, we obtained similar results in a more uniform subgroup of HCV genotype 1b patients treated with PEG-IFNα/RBV (data not shown). The second limitation is the ethnic homogeneity of the Japanese population. Because the baseline incidence of HCC development differs among population groups, longer-term longitudinal studies in larger cohorts with various population subgroups are required to verify the generality of our results.
With the development of potent direct-acting antiviral agents combinations, IFN-free therapy is likely to be approved in the near future. This raises the question of whether posttreatment ALT and/or AFP levels will remain a significant predictor of HCC risk. Moreover, it is uncertain whether the suppressive effect of viral eradication by IFN-free regimens on hepatocarcinogenesis will be identical to that obtained by IFN-based regimens. Therefore, it is extremely interesting to prove these issues in future studies.
In conclusion, post-IFN treatment ALT and AFP levels are strictly associated with hepatocarcinogenesis risk in patients with CHC. Measurement of these values is useful for predicting future HCC risk in IFN-treated patients. Suppression of these values after IFN therapy reduces HCC risk even in patients without HCV eradication, while SVRs with increased ALT and/or AFP levels are at risk for HCC development. The present results have potentially important clinical implications for physicians and may influence their decisions regarding treatment strategy and HCC surveillance for individual patients.