Clearance of hepatitis B surface antigen and risk of hepatocellular carcinoma in a cohort chronically infected with hepatitis B virus†
This article is corrected by:
- Errata: Correction Volume 62, Issue 4, 1330, Article first published online: 7 August 2015
Potential conflict of interest: Nothing to report.
Some individuals who are chronically infected with hepatitis B virus (HBV) eventually lose hepatitis B surface antigen (HBsAg). Hepatocellular carcinoma (HCC) has been demonstrated to occur in a few patients after loss of HBsAg. Neither factors associated with loss of HBsAg nor the incidence of HCC thereafter have been clearly elucidated. We performed a prospective population-based cohort study in 1,271 Alaska Native persons with chronic HBV infection followed for an average of 19.6 years to determine factors associated with loss of HBsAg and risk of developing HCC thereafter. HBsAg loss occurred in 158 persons for a rate of HBsAg clearance of 0.7%/year. Older age, but not sex, was associated with clearance of HBsAg, and loss of HBsAg was not associated with any particular HBV genotypes (A, B, C, D, and F) found in this population. Participants were followed for an average of 108.9 months after HBsAg loss. Six patients, two with cirrhosis and four without, developed HCC a mean of 7.3 years after HBsAg clearance (range, 2.0–15.5 years). The incidence of HCC after clearance of HBsAg was 36.8 per 100,000 per year (95% CI 13.5–80.0) which was significantly lower than the rate in those who remained HBsAg-positive (195.7 cases per 100,000 person-years of follow-up [95% CI 141.1–264.5; P < 0.001]). After loss of HBsAg, HBV DNA was detected in the sera of 28 (18%) of those who cleared a median of 3.6 years after clearance. Conclusion: HCC can occur in persons with chronic hepatitis B who have lost HBsAg, even in the absence of cirrhosis. These persons should still be followed with periodic liver ultrasound to detect HCC early. (HEPATOLOGY 2010.)
Hepatitis B (HBV) infection is a major worldwide health problem, and it is estimated that 350 to 400 million persons are chronically infected with this virus.1 Perinatal or horizontal infection early in childhood is the main route of HBV transmission. After acute hepatitis B, most persons achieve complete immune clearance of the virus, yielding lifelong immunity. However, up to 90% of infants, 30% of children under 5 years of age, and between 5% and 10% of adults develop chronic infection.2 Chronic HBV infection is confirmed by the presence of hepatitis B surface antigen (HBsAg) in serum for at least 6 months and is characterized by several clinical phases: the immune tolerant, immune active, or clearance and inactive phases.3
The immune tolerant phase usually follows perinatal infection and is characterized by the presence of hepatitis B e antigen (HBeAg), high levels of HBV DNA, normal levels of alanine aminotransferase (ALT), and minimal or no liver inflammation. The immune clearance phase is characterized by high levels of HBV DNA, elevated ALT levels, and active liver inflammation. Initially, those in the immune clearance phase will be HBeAg-positive and most will eventually clear HBeAg and develop antibody to hepatitis B e antigen (anti-HBe). A minority of patients who clear HBeAg will continue to have elevated ALT and HBV DNA levels either continuously or intermittently and have chronic hepatitis B.
The majority of those who clear HBeAg will evolve into the inactive HBsAg phase. This phase is characterized by the absence of HBeAg, presence of anti-HBe, normalized ALT levels, low or undetectable HBV DNA, and improvement or resolution of hepatic necrosis and fibrosis.3, 4
Some patients who have achieved the inactive phase will lose HBsAg and may develop antibody to hepatitis B surface antigen (anti-HBs) after many years of being HBsAg-positive. This spontaneous loss of HBsAg has been reported at a rate of 0.5%–1% per year in patients with chronic HBV infection.5, 6 Although this phenomenon has been well described, the factors associated with this occurrence (such as age, sex, and HBV genotype) have not been elucidated, and little is known about how these factors relate to clearance of HBsAg.
Sequence variability within the HBV genome is recognized as a factor that modulates the course and outcome of chronic HBV infection.7 Extensive analysis of HBV isolates from different regions of the world has identified eight genotypes (A-H) with distinct geographic distributions, separated by 8% sequence difference between genotypes.8
Significant differences in disease progression have been shown between some genotypes. For example, genotype C is associated with the development of hepatocellular carcinoma (HCC) in Taiwan, China, and Japan.9–11 Infection with genotype Bj or B1 in Japan and China predicts a lower risk of serious sequelae, while in Taiwan genotype Ba is significantly associated with the development of HCC in young patients.9 Outside of East Asia, there is a paucity of information on the clinical contribution of the remaining HBV genotypes to the progression of disease, because the majority of studies originate from Asia, where genotypes B and C predominate.12
Alaska Native persons have a high prevalence of HBV infection. The introduction of routine infant immunization with hepatitis B vaccine and mass vaccination in the mid-1980s has virtually stopped transmission of this virus in this population. However, serologic surveys for HBV and a catch-up screening campaign have identified approximately 1,500 chronically infected persons who have been followed for the past 20 years.13 A genotype study in this population found five (A, B, C, D, F) of the eight genotypes of HBV in this cohort. Furthermore, differences in the outcome of HBV infection by HBV genotype have been recently elucidated in this population, such as the association between infection by genotypes F and C and HCC and delayed clearance of HBeAg in persons infected with genotype C.14 We herein report the results of 20 years of follow-up of this HBV-infected population-based cohort to determine if the clearance of HBsAg is associated with factors such as HBV genotype, age, and sex.
Patients and Methods
We identified 1,536 Alaska Native persons chronically infected with HBV from a mass population screening for HBV markers during a hepatitis B vaccination campaign conducted between 1983 and 1987. Since then, 1,271 persons have consented to participate in this study and have been followed for an average of 19.6 years.
Since 1982, HBsAg-positive carriers have been contacted biannually and reminded to go to their village clinic and have their blood drawn to be sent to the Alaska Native Medical Center in Anchorage for testing. Between 1982 and 2000, sera were tested every 6 months for HBsAg to confirm carrier status. In addition, in 2001, after centrifuges were placed in all village clinics, ALT and aspartate aminotransferase were added to the blood tests performed. Individuals were considered to have cleared HBsAg when two successive specimens tested negative for the antigen.
This study was approved by the Institutional Review Boards of the Alaska Area Indian Health Services in Anchorage and the Centers for Disease Control and Prevention in Atlanta, GA, and the following Native Health boards: the Alaska Native Tribal Health Consortium, the Southcentral Foundation, and the Yukon-Kuskokwim Health Corporation. All living patients consented for laboratory testing and storage of sera for future testing of HBV seromarkers and HBV DNA. Information from deceased patients was used with Institutional Review Board permission.
Laboratory testing for HBsAg, anti-HBs, and antibody to hepatitis B core antigen (anti-HBc) was performed at the Alaska Native Medical Center clinical laboratory by commercial assay using enzyme-linked immunoassay (Abbott Laboratories, Irving, TX). HBeAg and anti-HBe were performed from 1982 to 2000 by way of enzyme-linked immunoassay (Abbott Laboratories). Between 2001 and 2006, HBeAg determinations were performed using the One-Step Hepatitis B ‘e’ Antigen Test Strip, and the anti-HBe status of each specimen was determined using the Maxi: Test Anti-HBe Rapid Test (IND Diagnostic Inc., Delta, British Columbia, Canada) according to the manufacturer's instructions. Both tests were validated using sera previously tested with the Abbott Laboratories enzyme-linked immunoassay kits, which are no longer commercially available.
HBV Viral Load Determination.
HBV viral loads from stored sera were determined using a real-time quantitative polymerase chain reaction (PCR) assay with a lower limit of detection of 100 copies/mL (20 IU/mL). Briefly, HBV DNA was extracted from stored serum specimens using the MagNA Pure Compact System and Total Nucleic Acid Extraction Kit I according to the manufacturer's instructions (Roche Applied Science, Indianapolis, IN). Subsequently, a 118-bp fragment of the core gene was amplified, and HBV viral loads were determined by comparison to a set of HBV DNA standards amplified in parallel with the samples being tested.15
HBV Genotype Testing.
Using stored sera, we were able to genotype 1,157 of 1,271 (91.0%) consented persons tested in this cohort by PCR and direct sequencing of the S gene. DNA extraction, nested PCR, DNA sequencing, and genotyping were performed as described.12
Proportions of persons who lost HBsAg by HBV genotype were compared using a chi-squared test or randomization test for contingency tables, as appropriate. In addition, logistic regression was used to adjust for multiple factors. The time of seroconversion was calculated as the mid-point between the date of last HBsAg-positive specimen and first HBsAg-negative specimen. Forty-nine persons were treated for hepatitis B. The data from these persons were censored at the time treatment began. One of these treated persons later lost HBsAg, but this individual was not considered in the analysis. A log-rank test and Cox regression were used to assess the time to HBsAg loss by genotype.
HBsAg Loss by Genotype.
We followed 1,271 persons with chronic HBV infection for an average of 19.6 years (24,913 person-years). The median age at initial HBsAg-positive result was 20.9 and the median age as of January 1, 2007, was 45.4 years. Of the 1,271 HBV carriers consented for this study, 158 (12.5%) became HBsAg-negative during the 20-year follow-up period. Of those who cleared HBsAg, 126 (80%) developed anti-HBs. Stored sera with sufficient DNA for genotype testing were available for 129 (81.6%) of the persons who cleared HBsAg. The genotype distribution of persistently HBsAg-positive persons and those who cleared HBsAg is shown in Table 1. HBsAg clearance by HBV genotype ranged from 6.7% for genotype B to 12.6% for genotype F. After removing the one person with HBV genotype H, the overall proportion of carriers who became HBsAg-negative did not vary significantly by genotype (P = 0.464).
Table 1. Loss of HBsAg by Genotype in 1,271 Persons Chronically Infected with Hepatitis B Virus Who Were Followed for 20 Years
|Total with genotype||129/1,157 (11.1%)||0.59%|
|No genotype available||29/114 (25.4%)||1.86%|
Overall, 0.67% of carriers cleared HBsAg per year (Table 1). This rate was significantly greater among carriers for whom a genotype could not be obtained (1.86%) than for carriers who could be genotyped (0.59% [P < 0.001]). Among carriers who had been genotyped, the annual rate of HBsAg loss did not vary significantly by genotype (P = 0.526).
Loss of HBsAg by Genotype, Adjusting for Age and Sex.
The analysis results for genotype association with age of HBsAg loss are shown in Table 2. Loss of HBsAg was significantly more common among persons ≥20 years of age at first HBsAg-positive specimen (P = 0.002). After adjusting for age at first HBsAg-positive specimen, the difference in the proportion of persons who lost HBsAg by HBV genotype remained nonsignificant (P = 0.209).
Table 2. Loss of HBsAg by Genotype by Age and Sex at Time of First HBsAg-Positive Specimen in 1,271 Persons Chronically Infected with HBV*
|A||3/66 (4.5%)||11/85 (12.9%)||5/58 (8.6%)||9/93 (9.7%)|
|B||0/5 (0%)||3/40 (7.5%)||1/22 (4.5%)||2/23 (8.7%)|
|C||0/32 (0%)||5/42 (11.9%)||4/42 (9.5%)||1/32 (3.1%)|
|D||29/329 (8.8%)||48/326 (14.7%)||39/276 (14.1%)||38/379 (10.0%)|
|F||16/153 (10.5%)||13/78 (16.7%)||12/91 (13.2%)||17/140 (12.1%)|
|H||0||1/1 (100%)||0||1/1 (100%)|
|Total||48/585 (8.2%)||81/572 (14.3%)||61/499 (12.2%)||69/668 (10.3%)|
There was no significant difference in the proportion of persons who lost HBsAg by sex (P = 0.262) (Table 2). After adjusting for sex, the difference in proportion of those who lost HBsAg by genotype was not significant (P = 0.435).
Detectable HBV DNA in HBsAg Seroconverters, by Genotype.
HBV DNA was tested a median of 3.6 years after first HBsAg-negative test (minimum-maximum, 0–29 years). HBV DNA was detected in 28 (17.7%) patients. The HBV DNA levels ranged from 28 to 2,340 IU/mL. The proportion of persons who had HBV DNA in their sera after loss of HBsAg did not vary significantly by genotype (P = 0.906) (Table 3).
Table 3. HBV DNA in HBsAg Seroconverters, by Genotype
HBsAg Loss by Initial HBeAg Status and Genotype.
Persons who were HBeAg-positive on their initial specimen were significantly less likely to lose HBsAg than those who were HBeAg-negative (P < 0.001). This difference persisted after adjustment for genotype (P < 0.001). Paradoxically, there was a statistically significant interaction between HBeAg status and genotype (P = 0.038), primarily due to the differences with genotype A. Persons infected with genotype A who were HBeAg-positive on initial test date were more likely to clear HBsAg subsequently than those with genotype A who were initially positive for anti-HBe (Table 4).
Table 4. HBsAg Loss by Initial HBeAg Status in 1,271 Persons Chronically Infected with HBV*
|A||5/34 (15%)||9/116 (8%)|
|B||0/6 (0%)||3/38 (8%)|
|C||0/36 (0%)||5/38 (13%)|
|D||21/305 (7%)||56/350 (16%)|
|F||12/126 (10%)||17/105 (16%)|
|Total||38/507 (7%)||90/648 (14%)|
Development of HCC After HBsAg Loss.
The 158 persons who cleared HBsAg were followed for an average of 8.6 years. HCC occurred in six patients a mean of 7.3 years (range, 2.0–15.5 years) after HBsAg clearance (Table 5). This rate of HCC after loss of HBsAg in these persons was 36.8 per 100,000 per year (95% confidence interval 13.5–80.0). The rate of HCC in persons from our cohort who remained HBsAg-positive was 195.7 cases per 100,000 person-years of follow-up (95% CI 141.1–264.5), which was significantly higher than the rate of HCC in those who cleared HBsAg (P < 0.001). One person was also positive for hepatitis C antibody but negative for hepatitis C virus RNA; the other five were negative for hepatitis C antibody. Family history was available for five patients, and none had any relatives diagnosed with HCC. Sera were available to test for HBV DNA in four of the patients with HCC, and all tested negative. Cirrhosis was found in two patients, and the other four did not have cirrhosis; rather, they had mild inflammation and portal fibrosis only. We decided after the first case of HCC occurred to screen all persons who had lost HBsAg with alpha-fetoprotein levels and to perform liver ultrasound every 6 months on those with alpha-fetoprotein levels above 10 ng/mL. As a result, five of the six patients were found to have small tumors that could be surgically resected, four of which were <5 cm in diameter. In three patients, HCC recurred and they died 1 to 2 years after resection. Two patients were tumor-free long-term survivors. One patient expired 10 years after resection at age 79 from chronic lung disease and had no evidence of HCC by ultrasound or alpha-fetoprotein at time of death. The other was still alive with no evidence of recurrence on liver ultrasound 13 years after resection at age 73.
Table 5. Alaska Native Patients Who Developed HCC After Clearance of HBsAg
|1||A||F||2.9||34.2||43.8||Positive/Negative||26.4||Not done||3.5 cm/one lesion||Cirrhotic nodules||Yes|
|2||A||M||18.0||74.3||76.7||Positive/Positive||202.0||Not detected||4 cm/one lesion||Portal inflammation and fibrosis||No|
|3||F||F||2.6||51.6||60.7||Positive/Positive||3,227||Not detected||2.2 cm/one lesion||Portal and lobular inflammation||No|
|4||F||M||1.2||18.3||20.3||Positive/Negative||213,720||Not done||Entire right lobe/multiple lesions||Mild portal inflammation, no fibrosis||No|
|5||F||M||10.3||49.3||64.8||Positive/Positive||599||Not detected||2.5 cm/one lesion||Cirrhosis||Yes|
|6||F||M||7.0||64.3||69.8||Positive/Positive||4,565||Not detected||2.2 cm/one lesion||Portal inflammation and fibrosis||No|
We prospectively followed a population-based cohort of 1,271 Alaska Native persons chronically infected with HBV for 20 years and determined the incidence of HBsAg clearance. For those who cleared HBsAg, we were able to follow them for an average of 8.6 years after HBsAg clearance. This cohort, we propose, was representative of persons infected with HBV because both sex and all ages were included. In addition, five of the eight HBV genotypes found in the world were represented in this population cohort, allowing us to determine if HBV genotype influenced the clearance of HBsAg. We found that age but not sex was associated with clearance of HBsAg, because older individuals were significantly more likely to clear HBsAg. This makes sense because it is more likely that older individuals were in the inactive HBV stage for a longer period than younger ones. However, we did not find any relationship to HBV genotype in HBsAg clearance. Because this cohort is relatively young (median age, 45 years) and we have shown that HBsAg clearance is higher in older individuals, it is possible that as this cohort ages, an association with HBV genotype may be found.
Some previous prospective studies have been reported following untreated cohorts of patients with chronic hepatitis B who cleared HBsAg5, 6, 16–20 (Table 6). All of these studies have been clinic-based, and ours is the only population-based study. One study, not included in Table 6, followed both treated and untreated European HBsAg-positive patients and found that the annual incidence of HBsAg clearance in 196 untreated patients was 0.8%. In that study, only one patient developed HCC. That patient was also positive for anti-HCV, and it was unclear whether this patient had received antiviral therapy or not.20 Rates of HBsAg clearance have been similar to the rate we found, ranging from 0.5% to 0.8% per year.6, 16 Similar to our study, older age but not sex was associated with HBsAg seroclearance.5 One previous case–control study found that persons infected with genotype B had a significantly higher rate of HBsAg clearance than those infected with genotype C.19 However, based on the geographic distribution of HBV genotypes, genotype B in this study was likely to be B Asia (B2-5), whereas the genotype B found in Alaska Native carriers is a new subtype (B6) that is more closely related to B Japan (B1).21
Table 6. Studies Reporting Spontaneous HBsAg Clearance in Persons with Chronic HBV Infection
|Huo et al.18/Taiwan||1,355||33 months||55||NA||23 months||11|
|Arase et al.17/Japan||5,055||48 months||156||NA||78 months||2|
|Chen et al.22/Taiwan||NA||NA||218||NA||61.7 months||2|
|Ahn et al.16/South Korea||432 inactive carriers||223 months||49||0.5%/year||19.6 months||5|
|Chu and Liaw5/Taiwan||1,965 carriers||128 months||245||0.8%/year||NA||4|
|Yuen et al.19/Hong Kong||NA||NA||285||NA||108.9 months||7|
|Present study||1,271||235.2 months||158||0.7%/year||103.2 months||6|
Persons who cleared HBsAg have been tested for HBV DNA in three studies. Presence of HBV DNA varied widely and was found in 1.3% and 1.7% at 12 months,17, 22 1.1% at 49 months,23 and 21.4% at 5 years and 14.3% at 10 years after clearance.24 In our study, 18% were found to have HBV DNA a median of 3.6 years after HBsAg clearance and we found no differences in the presence or absence of HBV DNA by genotype. Two studies found HBV DNA and even covalently closed circular DNA more frequently in liver biopsies prior to clearance than in sera.16, 23 It is likely that differences in the presence or absence of HBV DNA may be related to the sensitivity of the PCR test used and may even fluctuate over time.
We were able to determine genotype on 129 persons who cleared HBsAg. We found evidence of HBV DNA in 28 (18%) of these a median of 3.6 years after clearance. For this testing, we used a sensitive Taq Man assay that can detect ≤30 IU/mL HBV DNA. The levels of HBV DNA were always low and ranged from 28 to 2340 IU/mL. Individuals with such low levels would be less likely to transmit HBV by casual or sexual contact than persons with higher levels of HBV DNA. Recently, it has been shown that in persons who recover from acute hepatitis B and do not become chronically infected, HBV DNA can be found in as many as 50% years after resolution of the infection, suggesting that HBV, like other DNA viruses, may persist at low levels indefinitely.25, 26
The clinical outcome after clearance of HBsAg is generally much better than in persons who continue to be HBsAg-positive. Liver inflammation and fibrosis has been found to improve over time.16, 17, 23 In one study, none of 186 persons who were noncirrhotic and not infected with hepatitis C or D at the time of HBsAg clearance subsequently developed cirrhosis, and all had normal ALT levels an average of 62 months after clearance.22 However, the risk of developing HCC after HBsAg clearance has not been determined, though others have shown that HCC can still occur years after clearance of HBsAg.16, 17, 22, 23 In the study with the highest rate of liver-related complications after loss of HBsAg, 29.8% at 5 years, including 11 cases of HCC, participants were only followed for a mean of 23 months after HBsAg clearance.18 It is conceivable that many of these patients may have been incubating HCC tumors at the time of HBsAg clearance. In our study, HCC occurred in 6 patients a mean of 7.3 years after HBsAg clearance (range, 2.0–15.5 years). Thus, it is apparent that development of HCC can occur in persons who clear HBsAg. However, our study shows that a significant reduction in the risk of developing HCC occurs in chronic HBV-infected persons after clearance of HBsAg in comparison with those who remain HBsAg-positive.
In persons who cleared HBsAg, HCC can occur in the absence of pre-existing cirrhosis. This may be due to the fact that HBV DNA is able to integrate into the hepatocyte genome. Thus, even if all circulating HBV DNA were to be eliminated, the presence of integrated HBV DNA in hepatocytes could lead to mutations in the human hepatocyte genome that could occur over time during cell division and thus conceivably lead to HCC development. Whereas older carriers are significantly more likely to clear HBsAg than younger carriers, older age at clearance of HBsAg is associated with a higher risk of subsequently developing HCC.23 This may be due to the fact that persons who clear HBsAg at an older age have had a longer exposure to the damaging effects of HBV DNA integration and thus increased risk of HCC. In studies from areas of endemic HBV infection where seromarkers of HCC were examined, it has been found that a minority of patients were HBsAg-negative but positive for anti-HBc and/or anti-HBs.27, 28 It is conceivable that these patients may have been persons who cleared HBsAg prior to the development of HCC. Thus, many cases of so-called occult HBV-associated HCC may actually be chronic carriers who developed HCC after clearing HBsAg.
In conclusion, we found HBsAg clearance to occur in a large population of HBsAg carriers at a rate of 0.7%/year. Older age was the only predictive factor for clearance of HBsAg, and HBV genotype did not predict clearance. Our study is the first to show a significant reduction in the risk of developing HCC after HBsAg clearance. However, because HCC can develop in persons with a history of chronic HBV infection years after clearance of HBsAg, it seems reasonable to follow established practice in persons who cleared HBsAg after chronic infection for early detection of HCC, using periodic alpha-fetoprotein and/or liver ultrasound, as is already recommended for HBsAg-positive patients in established practice guidelines.29