Long-term follow-up of alpha-interferon treatment of patients with chronic hepatitis B



Data on the long-term effects of interferon alfa (IFN) treatment on disease progression and mortality in patients with chronic hepatitis B (CHB) are limited. To evaluate factors that influence clinical outcome and survival, we performed a follow-up study on 165 hepatitis B e antigen (HBeAg) positive CHB patients treated with IFN between 1978 and 2002. The median IFN dose was 30 megaunits (MU)/week (range, 2–70 MU/week), and the median duration of therapy was 16 weeks (range, 1–92 weeks). Response to treatment was defined as HBeAg loss within 12 months after the end of IFN therapy. Median follow-up was 8.8 years (range, 0.3–24 years). Fifty-four patients (33%) responded to IFN treatment. Relapse (HBeAg reactivation) occurred in 7 of the 54 (13%) responders. Fifty-two percent of the responders lost hepatitis B surface antigen (HBsAg) as compared with 9% of the nonresponders (P < .001). Liver histology showed a decreased necroinflammatory activity and less progression of fibrosis in responders. Twenty-six patients died during follow-up. Hepatocellular carcinoma (HCC) was found in 8 patients, 6 of whom were nonresponders. Of the two responders who developed HCC, one patient had relapsed after discontinuation of therapy. Multivariate analysis showed significantly improved survival (relative risk (RR) of death 0.28, 95% CI 0.10–0.78) and reduced risk of developing HCC (RR 0.084, 95% CI 0.09–0.75) in responders. In conclusion, response to IFN therapy results in a prolonged clinical remission with an increased rate of HBsAg seroconversion and improved liver histology. Our results indicate that after correction for baseline factors, response to IFN therapy increases survival and reduces the risk of developing HCC. (HEPATOLOGY 2004;39:804–810.)

Chronic hepatitis B (CHB) affects 400 million people worldwide and is a major cause of morbidity and mortality. Patients with CHB are at an increased risk of developing liver cirrhosis and its complications, such as hepatic decompensation and hepatocellular carcinoma (HCC), compared with matched controls from the normal population.1–5 During their lifetime, 25% to 40% of CHB patients will develop serious complications.5 Treatment of CHB aims at inactivation of liver disease as indicated by hepatitis B e antigen (HBeAg) seroconversion and disappearance of serum HBV DNA, measured by hybridization techniques. Spontaneous HBeAg seroconversion occurs in about 8–12 % per year.6–8 Interferon alfa (IFN) has been reported to increase the HBeAg seroconversion rate to 33%.8 Particularly for Caucasians, data on the long-term effects on disease progression and mortality are limited. To further evaluate factors that influence clinical outcome and survival, we performed a retrospective cohort study on all HBeAg positive CHB patients treated with IFN in our liver unit.


IFN, interferon alfa; CHB, chronic hepatitis B; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; HCC, hepatocellular carcinoma; ALT, alanine aminotransferase; MU, megaunits.

Patients and Methods


Patients were identified through a search of our trial databases and our computerized hospital-wide patient information system. All HBeAg positive CHB patients treated with interferon between 1978 and 2002 were included in the study. CHB was documented by hepatitis B surface antigen (HBsAg) positivity for more than 6 months and by a liver biopsy. We identified a total of 165 HBeAg positive patients who were treated with IFN in our center. One hundred thirty-two patients participated in one or more trials9–16; the remaining 33 patients were treated electively. Sixty-two patients (38%) received treatment with nucleoside analogs as combination therapy with IFN (zidovudine, n = 12; acyclovir, n = 10; lamivudine, n = 40); one patient received two courses of combination therapy. Forty-eight nonresponders were retreated and received two or more courses of IFN. Patients were seen every 2–4 weeks during treatment. Thereafter patients were scheduled for visits every 3–12 months. Response was defined as loss of HBeAg at two consecutive measurements at least one month apart, during therapy or within 12 months after the end of the first IFN therapy. All other patients were considered nonresponders.


Data were collected by review of patient records. We contacted all patients and asked them to return for evaluation and laboratory examination. If a patient could not be traced, relevant information was obtained from the patient's primary care physician. Data were collected on survival, occurrence of decompensated liver disease (defined as ascites, encephalopathy, variceal bleeding,or bilirubin >34 μmol/L), and HCC. For all patients who died, the exact date and cause of death were documented. Follow-up time was calculated from the start of IFN treatment until death or until the last visit. Secondary end points were occurrences of liver-disease-related complications (HCC, decompensated liver disease). For calculation of survival times, liver transplantation was combined with death. For 16 patients, no recent information was available. Only one of these patients had cirrhosis at the last biopsy, but minimal inflammatory activity. None of the patients lost to follow-up showed any signs of decompensated liver disease or HCC at the last visit.


We reviewed biochemical data (alanine aminotransferase [ALT], asparate aminotransferase, bilirubin, and albumin) and virological data from the patient records. These tests were routinely performed at each visit to our outpatient clinic. For measurement of HBV DNA, samples obtained pretreatment, one year posttreatment, and at follow-up were retested. HBV DNA was assessed by TaqMan real-time polymerase chain reaction assay17 (dynamic range, 373 – 1010 genome equivalents/mL, Eurohep standard). HBeAg, HBsAg, antibodies to HbeAg, and antibodies to HBsAg were measured using enzyme-linked immunnnosorbent assays or solid-phase radioimmunoassays (Abbott Laboratories, Abbott Park, IL). Liver biopsies were obtained at the start of therapy. For follow-up, the most recent liver biopsy was used. All liver biopsies were scored for inflammatory activity (combined score for portal inflammation and interface hepatitis) and fibrosis by two experienced pathologists using predefined criteria. Inflammatory activity was scored as none (0), minimal (1), mild (2), moderate (3), and severe (4). Fibrosis was scored as none (0), mild (1), moderate (2), severe (3), probable cirrhosis (4), and cirrhosis (5).


The chi-square test was used to compare frequencies, and the Mann-Whitney test was used to compare means between groups for various variables. The Kaplan-Meier method was used to estimate survival and time to HBeAg and HBsAg seroconversion. The following variables were considered for univariate analysis: age; sex; baseline asparate aminotransferase, ALT, albumin, bilirubin, and HBV DNA levels; necroinflammatory and fibrosis score on liver biopsy; liver cirrhosis; mean weekly dose, total dose of IFN therapy and combination therapy with other antivirals. Response was analyzed as a time-dependent factor. Patients who responded might have had longer survival times, because they lived at least until response. To avoid this bias, response was entered as a time-dependent covariate, i.e. all patients were entered in the nonresponder group at the start of therapy and, in case of response, were censored from the nonresponder group at the time of response and included in the responder group.18 Cox regression analysis was applied to determine which factors were independently associated with survival. First, all baseline factors that were related to survival in the univariate analysis with P < .10 were assessed by both the forward stepwise and backward stepwise method. Then baseline factors that predicted death were entered in a multivariate model together with the time-dependent factor response to therapy. The same analysis was performed for the secondary end points decompensated liver disease and HCC. P values are based on likelihood-ratio statistics. To exclude a confounding effect of coinfection with HCV, HDV, or HIV, survival analyses were repeated after exclusion of patients with these coinfections. All analyses were performed using SPSS version 10.1 (SPSS Inc., Chicago, IL).


Baseline Characteristics

Baseline characteristics of the patients are shown in Table 1. The patient group consisted of 165 patients, of which 118 (72%) were male and 123 (75%) were Caucasian. The median ALT level was 89 U/L (range, 12–730 U/L). The median dose of IFN was 30 megaunits (MU)/week (range, 2–70 MU/wk), and the median duration of therapy was 16 weeks (range, 1–92 wks). Forty-eight patients (45 nonresponders and 3 responders who relapsed), were retreated with IFN; 33 patients received 2 courses of IFN, 8 patients received 3 courses, 5 patients received 4 courses, and 2 patients received 5 courses. Median follow-up time was 8.8 years (range, 0.3–23.9 yrs). Coinfection with HCV, HDV, and HIV was present in 3, 2, and 7 patients, respectively.

Table 1. Baseline Characteristics
 Total n = 165Responders n = 54Nonresponders n = 111P Value
  1. NOTE. Data are presented as median (range).

  2. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; HBV, hepatitis B virus; IFN, interferon alfa; MU, megaunits.

Age (years)34 (10–67)37 (10–64)32 (16–67).081
Male (%)118 (71.5)41 (75.9)77 (69.4).38
Race (%)   .056
 Caucasian123 (74.5)45 (83.3)78 (70.3) 
 Asian32 (19.4)6 (11.1)26 (23.4) 
 Other10 (6.0)3 (5.6)7 (6.3) 
Mode of transmission (%)   .91
 Vertical13 (7.9)3 (5.6)10 (9.0) 
 Parenteral13 (7.9)4 (7.4)9 (8.1) 
 Sexual60 (36.4)21 (38.9)39 (35.1) 
 Unknown79 (47.9)26 (48.2)53 (47.7) 
ALT (U/L)89 (12–730)102 (21–730)81 (12–654).020
AST (U/L)46 (12–367)59 (15–264)41 (12–367).002
Bilirubin (μmol/L)11 (2–45)12 (5–32)11 (2–45).91
Albumin (g/L)43 (14–54)42 (14–50)44 (28–54).057
Log HBV DNA9.0 (3.5–10.1)8.5 (3.5–9.9)9.1 (4.2–10.1).080
Cirrhosis (%)30 (19)15 (29)15 (14).034
Total IFN dose (MU)510 (33–3130)504 (55–3130)510 (33–1650).864
Mean weekly IFN dose30 (2–70)30 (9–36)30 (2–70).254
Follow-up (years)8.8 (0.3–23.9)10.6 (0.8–19.5)7.9 (0.3–23.9).2756


Virological Response.

Of all 165 patients, 54 (33%) exhibited a response. Recurrence of HBeAg was seen in 7 responders (13%). Responders had a significantly higher pretreatment ALT level than nonresponders (median, 102 U/L [range, 21–730 U/L] vs. median, 81 [range, 12–654], respectively; P = .02). Patients with preexisting cirrhosis (n = 30) had a higher response rate than patients without cirrhosis (50% vs. 29%, P = .026). At the end of follow-up, HBV DNA was negative by polymerase chain reaction in 61 patients (43%). Responders exhibited loss of HBsAg (52%) and HBV DNA (70 %) significantly more often than nonresponders (9% and 30%, respectively, P < .001). The time to loss of HBeAg and HBsAg in responders and nonresponders is shown in Figure 1. Among the 111 nonresponders, 57 patients lost HBeAg during follow-up. Of these patients, 17 lost HBeAg within a year after retreatment with IFN, and 40 patients exhibited a spontaneous loss of HBeAg.

Figure 1.

Time to (A) HBeAg-loss and (B) HBsAg-loss of responders compared with nonresponders. Responders exhibited HBeAg seroconversion during or within a year after the first course of IFN.

ALT Normalization.

ALT levels at the end of follow-up were significantly lower in the group of responders (median, ALT 21 U/L vs. 35 U/L, P = .001). In the responder group, 39 of the 54 patients (72%) had a normal ALT level at the end of follow-up, compared with 63 of the 111 (57%) nonresponders. This difference tended towards significance (P = .055).

Liver Histology.

At the start of therapy, 155 patients underwent a liver biopsy. For 123 patients, paired liver biopsies were available (mean follow-up time, 3.1 years; range, 0.4–14.6 years). At baseline, responders had higher inflammatory activity (mean score, 2.4 vs. 2.1, P = .006) and more advanced fibrosis (mean score, 2.4 vs. 1.6, P = .001) than nonresponders. Inflammatory score improved in 30 of 43 responders (70%) and in 27 of 81 nonresponders (33%) (P < .001); mean score at follow-up was 1.6 in responders vs. 2.0 in nonresponders (P = .011). In the responder group, fewer patients had progression of fibrosis (8 of 42 responders [19%], vs. 35 of 81 nonresponders [43%], P = .039). Also, more responders showed an improvement in fibrosis score (12 of 42 responders [29%], vs. 13 of 81 nonresponders [16%]), but this difference did not reach statistical significance (P = .1).


In 18 patients, complications of CHB (HCC and/or decompensation of liver disease) occurred. Two patients underwent liver transplantation, one of whom died of transplantation-related complications. One patient underwent a partial liver resection for HCC but had a recurrence thereafter. Two patients died while on the waiting list for liver transplantation. The other patients had inoperable HCC (n = 6) or were not eligible for transplantation because of comorbidity, high age, or clinical condition (n = 7). Twenty-six patients died during follow-up, 16 of complications of liver disease, and 10 of unrelated causes. Overall, five-year survival was 90% (95% CI: 86–94%), and 10-year survival was 84% (95% CI: 78–91%). The survival of responders and nonresponders is shown in Figure 2. Overall, not corrected for baseline factors, there was no difference in survival between responders and nonresponders (Fig. 2A). However, the percentage of patients with cirrhosis in the responder group was twice as high as in the nonresponder group (29% and 14%, respectively), giving the responders a priori a substantially higher mortality risk. When analyzed separately for patients with and without preexisting cirrhosis, survival was improved in both groups for patients responding to IFN (Fig. 2B). In the univariate analysis of all 165 patients, the baseline factors sex, age, AST, bilirubin, albumin, cirrhosis, liver inflammation, and mean weekly IFN dose were related to survival (P < .1) (Table 2). To correct for possible changing treatment schedules and inclusion criteria over time, we stratified for 5-year cohorts, but survival was not different among these cohorts. To evaluate the effect of combination therapy with lamivudine (n = 40), zidovudine (n = 12) or acyclovir (n = 10), we stratified for these therapies. Only zidovudine combination therapy was associated with survival (worse survival with zidovudine), but this can be explained by the fact that this group included 3 patients with HIV coinfection.

Figure 2.

Survival among responders and nonresponders to IFN, (A) overall and (B) stratified for preexisting cirrhosis. All patients were entered as nonresponders at start of therapy, with response as a reason for data censoring. Survival of responders was calculated from time of response.

Table 2. Univariate Analysis of Baseline Factors. Kaplan-Meier Estimates of Survival Compared by Log-rank Testing
  n5-Year Survival (%)P Value
  1. Abbreviations: AST, aspartate aminotransferase; ULN, upper limit of normal; ALT, alanine aminotransferase; HBV, hepatitis B virus; IFN, interferon alfa; MU, megaunits.

AST<1.5x ULN9393.087
 ≥1.5x ULN7086 
ALT<2x ULN6694.17
 ≥2x ULN9588 
Log HBV DNA<96689.35
Inflammation (biopsy)None–mild10096.003
Combination therapyNo10291.66
Mean weekly IFN dose<30 MU5988.08
 ≥30 MU10592 
Total dose of IFN<510 MU8188.47
 ≥510 MU8394 

The factors related to survival in univariate analysis were then included in the multivariate analysis. In this analysis, the baseline factors sex, age, albumin, and cirrhosis were independently associated with death. Next, response was entered in a multivariate model as a time-dependent factor together with these baseline factors. Corrected for the aforementioned baseline factors, response to therapy significantly improved survival (Table 3). Response was associated with a significantly decreased risk of HCC but not of decompensated liver disease. When we repeated the analysis for patients without cirrhosis, survival was still significantly improved in responders (P =.007). Exclusion of the small number of patients with HCV (n = 3), HDV (n = 2), and HIV (n = 7) did not influence the analysis. Also excluding patients treated with low (<15 MU) and high (>45 MU) weekly doses of IFN did not influence the outcome. We then analyzed whether spontaneous HBeAg loss influenced survival in the nonresponder group. Among the nonresponders, survival was significantly improved in patients who exhibited spontaneous HBeAg loss compared with patients who remained HBeAg positive (RR 0.11, P = .04). The time to HBeAg loss was not a predictor of survival.

Table 3. Predictive Factors for Death, Complications of Liver Disease and HCC: Multivariate Cox Analysis
VariablesDeath (n = 27)Decompensated Liver Disease (n = 16)HCC (n = 8)
RR (95% CI)P ValueRR (95% CI)P ValueRR (95% CI)P Value
  1. Abbreviations: HCC, hepatocellular carcinoma; RR, relative risk.

Age1.04 (1.01–1.07)0.0221.10 (1.03–1.18)0.0031.07 (1.01–1.13).032
Sex0.17 (0.02–1.28)0.027  
Albumin0.88 (0.82–0.94)0.002 0.79 (0.67–0.92).003
Cirrhosis2.68 (1.12–6.37)0.0258.06 (1.97–33.0)0.0039.62 (1.76–52.5).009
Response0.28 (0.10–0.78)0.008 0.08 (0.01–0.75).027

The development of decompensated liver disease and HCC is shown in Figure 3. HCC occurred in 8 male patients. Among the 111 nonresponders. 6 patients (5.4 %) developed HCC after a median interval of 2.2 years (range, 0.8–14.4 yrs). In the responder group, two of the 54 patients (3.7%) developed HCC, one after 9.5 years and the other after 4.1 years of follow-up. Of the two responders who developed HCC, only one patient had a lasting response to therapy, whereas the other relapsed after discontinuation of therapy. As previously mentioned, response also significantly reduced the risk of developing HCC in a time-dependent multivariate analysis (Table 3).

Figure 3.

Development of (A) decompensated liver disease and (B) HCC among responders and nonresponders to IFN, stratified for preexisting cirrhosis. All patients were entered as nonresponders at start of therapy, with response as a reason for data censoring. Survival of responders was calculated from time of response.


Treatment with IFN has been reported to increase the HBeAg seroconversion rate and induce disease remission in patients with CHB. In our study, 33% of patients lost HBeAg within one year after the end of therapy, which is comparable to the response rates found in other studies.7, 8, 19 We also found a high rate of HBsAg loss (52%) in patients who exhibited HBeAg loss after IFN therapy, as has been described before, particularly in Caucasians.20, 21 However, the long-term effects of IFN therapy on morbidity and mortality are controversial. In our study, we found a significantly prolonged survival in patients who responded to IFN. A number of studies have reported the long-term clinical course of CHB patients treated with interferon. Most studies compared the survival of IFN treated patients with that of untreated controls.2, 18, 21–23 Among these studies, two were randomized controlled trials21, 22 and the others were cohort studies with a control group. Di Marco et al. reported a better overall survival and complication-free survival after IFN treatment in Italian patients.2 However, this study included both HBeAg-positive and -negative patients, and patients with coinfections. Also, the different groups were not fully comparable, as treated patients had significantly higher baseline ALT levels than did untreated controls. In Asian patients, Lin et al. found a prolonged survival and a decreased incidence of HCC in IFN-treated patients.22 In our study, the majority of deaths were due to complications of liver disease. We found that the risk of developing HCC was decreased in responders to IFN, when all independent predictors of complications of liver disease were included in the multivariate analysis. As we found no significant difference in occurrence of decompensated liver disease, the survival benefit of response in our patients may be explained by the decreased incidence of HCC. Because of the small number of patients developing HCC in our study (n = 8), the analysis of this end point should be confirmed in future studies. Considering the disease remission following response to IFN treatment, one might expect a prolonged survival and fewer HBV-related complications in IFN treated patients. Particularly, in Caucasian patients, even after long-term follow-up the proportion of patients with disease remission (HBeAg-negative, HBV DNA-negative, normal ALT levels) remains higher in IFN-treated patients.18, 20 In patients with preexisting cirrhosis, HBeAg clearance and biochemical remission have been reported to be predictors of survival.24, 25 Niederau et al. also reported prolonged survival and a decreased risk of complications of liver disease after HBeAg clearance.18

Nevertheless, two large studies—one in a predominantly Caucasian patient group,21 one in Asians23—found no difference in survival or liver-related complications between IFN-treated patients and controls. The difference between the results of these studies and our current study may be explained by several factors, the first being the limited follow-up of studies and the slow natural course of the disease. In patients with active disease and minimal fibrosis who remain untreated or do not respond to IFN therapy, the development of liver cirrhosis and its complications may take considerably longer than the mean follow-up found in most of the previous studies. Only the study by Yuen et al.23 had a comparable follow-up time to our study (median follow-up, 9 years); but due to the very young patient group (median age, 27 years—much lower than the median age in all other studies), the incidence of complications was low. Secondly, the design of most previous studies was different. Several studies compared IFN-treated patients with untreated controls. However, a long-term benefit of IFN therapy might occur only in patients who exhibit IFN-induced HBeAg loss. As a consequence of the limited response rate to IFN (30–40%), this would make the benefit of therapy difficult to prove when the whole treated group is compared with untreated controls. Our study aims to overcome these difficulties with a longer follow-up time than in previous studies, and by comparing IFN responders to nonresponders. In this approach, one has to take into account that responders to IFN tend to have more disease activity than do nonresponders (higher ALT, more inflammation on liver biopsy). In addition, the percentage of cirrhotic patients among responders was two times higher than among nonresponders. These patients with preexisting cirrhosis clearly exhibited a substantially increased risk of complications and mortality. Using a multivariate time-dependent analysis, correcting for cirrhosis and other baseline factors, we found a significantly better survival in responders to IFN. Also, if we analyzed patients without cirrhosis separately, we found an improved survival after response to IFN. To exclude a confounding effect of retreatment with IFN in nonresponders, survival analyses were repeated after data censoring at the start of a second IFN course. After censoring patients at the start of retreatment with IFN, response remained significantly associated with survival. Also, the survival of patients who received one course of therapy was not significantly different from that of patients who received two or more courses of IFN. Our results are supported by the findings of Lau et al.,20 who reported a better complication-free survival in patients who responded to IFN treatment compared with nonresponders, when cirrhosis was considered in the analysis. In our opinion, the durability of response to IFN, high HBsAg-seroconversion rate after response, combined with the improved long-term clinical outcome, justifies the use of IFN as first-line therapy for HBeAg-positive CHB patients with compensated liver disease.

We conclude that in CHB patients, response to IFN therapy results in a prolonged clinical remission with an increased rate of sustained HBsAg seroconversion and improved liver histology. Our results indicate that, after correction for baseline factors, response to IFN therapy increases survival and reduces the risk of developing HCC.


The authors thank Dr. RA Heijtink, from the Department of Virology at Erasmus Medical Center in Rotterdam for providing serum samples of study patients.