Serologic and Clinical Outcomes of 1536 Alaska Natives Chronically Infected with Hepatitis B Virus FREE

An estimated 350 000 000 persons worldwide are chronically infected with hepatitis B virus (HBV) (1). Carriers of HBV are at increased risk for hepatocellular carcinoma and end-stage liver disease (2).

Most persons infected with HBV recover from the acute infection, as manifested by clearance of hepatitis B surface antigen (HBsAg) and development of antibody to hepatitis B surface antigen (anti-HBs). Initially, chronic carriers of HBV have high levels of HBV DNA in their sera and usually remain positive for hepatitis B e antigen (HBeAg) for several years (3). Most carriers eventually clear HBeAg and develop antibody to hepatitis B e antigen (anti-HBe) (4 - 7). Hepatitis B surface antigen is sometimes cleared after many years of continuous positivity (8). Because population-based studies that have followed carriers for prolonged periods are lacking, the proportion of carriers who clear HBeAg and HBsAg over time is not known. Antiviral therapy in chronic hepatitis B can result in accelerated clearance of HBsAg and HBeAg in some carriers. However, the long-term benefit of antiviral therapy is unclear (9). Information on the clearance of these markers over a longer period in untreated patients would be helpful and could be used to compare with results observed in persons who have received antiviral therapy and are followed over a long period.

Infection with HBV is hyperendemic in Alaska Natives. In this group, HBV infection is primarily transmitted horizontally in early childhood, with some acquisition in adulthood (10). Serologic surveys beginning in the early 1970s identified carriers of HBV, and a registry of carriers was established in 1978 (11 - 12). Between 1983 and 1987, 52 000 persons—approximately 70% of the entire Alaska Native population—were screened for seromarkers of HBV, and 3.1% were found to be HBsAg positive; rates were more than 8% in southwest Alaska. Seronegative persons and all newborns were offered hepatitis B vaccine in an effort to control the spread of the infection (13 - 14). Since 1987, screening and vaccination have continued; at this time, more than 90% of the population has participated.

This program has identified 1536 persons chronically infected with HBV who have now been observed for an average of 12.6 years (19 430 carrier-years). We report the outcome of chronic HBV infection in this cohort, including clearance rates of HBeAg and HBsAg, and the incidence of and risk factors associated with hepatocellular carcinoma and end-stage liver disease.

The Alaska Native population includes persons of Eskimo, Indian, and Aleut descent. In 1998, an estimated 104 305 Alaska Natives were living in Alaska. Nearly all Alaska Natives receive their health care through a three-tiered integrated health care system, with primary care given in villages, secondary care in regional hospitals, and tertiary care at the Alaska Native Medical Center in Anchorage. The Hepatitis B Carrier Registry contains records of all Alaska Natives who have had at least one positive test result for HBsAg. All persons who were HBsAg positive on consecutive specimens obtained at least 12 months apart, determined either retrospectively or prospectively, were defined as hepatitis B carriers and were included in the analysis. When newly HBsAg-positive persons were identified, the serum bank located at the Arctic Investigations Program of the Centers for Disease Control and Prevention in Anchorage, which contains several hundred thousand serum specimens stored since the early 1970s, was searched, and if serum specimens from these persons were found, they were tested for HBV seromarkers to attempt to establish time of infection and duration of carriage.

Permission to store carriers' blood and test for HBV markers was obtained from the Institutional Review Boards of the Alaska Native Medical Center and the Indian Health Service. Approval was also obtained from Alaska Native Regional Health Boards. Individual patients signed consent forms agreeing to have their serum specimens stored and tested.

Since 1982, all carriers and their health care providers have been sent reminders to have blood drawn for HBsAg and α-fetoprotein testing every 6 months as part of an ongoing program to attempt to detect hepatocellular carcinoma at an early resectable stage (15). Serum specimens from these carriers are also tested annually for HBeAg and anti-HBe. Testing for HBsAg, anti-HBs, antibody to hepatitis B core antigen (anti-HBc), HBeAg, and anti-HBe was performed by radioimmunoassay or enzyme-linked immunoassay, using commercial assays (Abbott Laboratories, Abbott Park, Ilinois). For purposes of analysis, persons simultaneously positive for both HBeAg and anti-HBe were counted as HBeAg-positive. Carriers who cleared HBsAg continued to be followed in the same manner as the rest of the cohort. Hepatitis B virus DNA was not measured.

The approximate date of HBV infection was known in 95 carriers because they were enrolled in a prospective incidence study of HBV conducted in the mid-1970s (10), before hepatitis B vaccine was available, or because they had a previously stored serum specimen available that subsequently was found to be negative for all seromarkers of HBV. We used the midpoint between the last HBsAg-negative specimen and the first HBsAg-positive specimen to estimate date of infection.

Alaska Natives reside in 189 communities, most of which are isolated and inaccessible by road. Most communities have a clinic in which primary care is provided by village community health aides who are trained in phlebotomy. Because it can take up to 1 week for serum samples drawn in a village to arrive at the Alaska Native Medical Center or the nearest hospital laboratory and because hepatic enzymes are unstable over time when clotted tubes are used to transport blood specimens, liver aminotransferase levels are not routinely measured.

Statewide computerized medical records for Alaska Natives and death certificates issued by the state of Alaska were reviewed to determine cause of death in carriers who died and to identify carriers who developed hepatocellular carcinoma or end-stage liver disease. In our study, an adverse outcome of HBsAg carriage was defined as hepatocellular carcinoma; end-stage liver disease (defined as the presence of ascites, hepatic encephalopathy, esophageal varices, or coagulopathy); or death due directly to liver disease or death in which a manifestation of end-stage liver disease was listed as a contributing cause, such as pneumonia in a carrier with decompensated cirrhosis. Medical records of carriers with end-stage liver disease as a primary or possible contributing cause of death were reviewed.

Analysis was performed to examine rates and describe characteristics of three outcomes: carrier clearance of HBsAg, loss of HBeAg, and hepatocellular carcinoma and liver disease–related deaths. Both simple bivariate comparisons and multivariable models were used to examine relations between risk factors and outcomes. Times to clearance of HBsAg and HBeAg were calculated on the basis of time from initial HBsAg-positive test result (rather than time of infection, which was unknown) and were analyzed by using Kaplan–Meier survival curves. Risk factors for carrier clearance of HBsAg and loss of HBeAg were examined by using tables of observed frequencies of clearance rates by risk factor and were analyzed by using Cox proportional-hazards models. Two time-dependent covariates were considered during model fitting: presence or absence of HBeAg and the number of times that the measured value had changed. Hepatocellular carcinoma and liver disease–related death outcomes were modeled by using Cox proportional-hazards models after initial bivariate examination of outcomes with frequency distributions and contingency tables. For much of the analysis, change in HBeAg status was categorized as one of five groups based on initial HBeAg status and the number of switches: initial HBeAg-positive result with no switches, initial HBeAg-positive result with one switch, initial HBeAg-positive result with more than one switch, initial HBeAg-negative result with no switches, and initial HBeAg-negative result with one or more switches. This categorization permitted more accurate description of the natural course of HBeAg clearance than that afforded by specification of the number of switches only.

We used two methods to explore factors contributing to clearance of HBeAg. By defining an outcome as clearance of HBeAg within 5 years of initial diagnosis, we could cross-classify outcomes by suspected influencing factors and note the variation in observed frequency of clearance as levels of these factors change. In addition, with time until conversion of HBeAg as an outcome, we used a Cox proportional-hazards model to examine factors influencing conversion of HBeAg.

We defined a favorable serologic outcome over time as conversion from HBeAg positivity to HBeAg negativity on the last two determinations or as being HBeAg negative/anti-HBe positive throughout follow-up (Table 1, groups B and D). An unfavorable outcome was defined as being persistently HBeAg positive or as having one or more reversions to HBeAg (groups A, C, and E).

The risk factors for these outcomes include age, sex, ethnicity (Yupik Eskimo or other Alaska Native), HBeAg status at first HBsAg detection, and pattern of HBeAg switching (number of switches and HBeAg switching category). Variables in the multivariate models presented were chosen by using forced-entry regression to control for confounders. Two-way interactions and nonlinear terms were added when their addition substantially improved the model.

A separate analysis was performed for the subgroup of 95 people for whom an approximate date of hepatitis B infection could be determined. A known approximate date of infection permits a more accurate estimate of years infected and time to clearance.

Two-sided statistical tests were used; P values of 0.05 or less were considered significant. S-Plus software, version 4.0, or S-PLUS 2000 software (MathSoft, Inc., Seattle, Washington) was used for all analyses.

We identified 1536 hepatitis B carriers, of whom 908 (59.1%) were male and 628 were female. Ninety-five persons were negative for all HBV seromarkers at the time of first determination, and their serum specimens were found to be HBsAg positive on subsequent testing. The remaining 1441 carriers were HBsAg positive at first serologic testing. Of the 1536 carriers, the median age at time of first HBsAg-positive test result was 19.9 years (range, 1 to 87 years). The ethnic distribution of the cohort was 865 (56.3%) Yupik Eskimo, 327 (21.3%) other Eskimo groups, and 344 (22.4%) other Alaska Native groups (Athabaskan, Southeast Indian, or Aleut). Median duration of follow-up for the cohort was 12.6 years (range, 1 to 33.4 years; 25th and 75th percentiles, 10.5 and 14.6); total duration of follow-up was 19 430 person-years. The median number of HBeAg and anti-HBe determinations was 15 (range, 0 to 47).

Only 9 carriers had ever received antiviral therapy; 5, all of whom were HBeAg positive, received interferon-α, and 4, 3 of whom were anti-HBe-positive, received lamivudine. All 5 carriers who received interferon cleared HBeAg within 1 year of treatment; the serologic status of the patients receiving lamivudine did not change.

At the first HBsAg-positive serologic measurement, 641 carriers (41.7%) were HBeAg positive and 893 (58.1%) had anti-HBe. Two carriers (0.1%) were not tested for HBeAg and anti-HBe. Serologic test results for the 1534 carriers are shown in Table 1. Of the 641 persons who were initially HBeAg positive, 100 (15.6%) remained HBeAg positive throughout follow-up (group A), 432 (67.4%) seroconverted from HBeAg to anti-HBe and remained HBeAg negative but anti-HBe positive thereafter (group B), and 109 (17%) had one or more reactivations (that is, reversion back to HBeAg) (median, 3 reactivations [range, 2 to 7 reactivations]) after initial seroconversion to anti-HBe (group C). Of this last group, 70 persons were HBeAg negative on the last two determinations and remained HBeAg negative and anti-HBe positive for a median of 6.2 years (range, 0.9 to 10.5 years), whereas 39 were HBeAg positive on the last serologic determination.

Of the 893 participants who were anti-HBe positive and HBeAg negative on their first serologic test, 802 (90%) remained anti-HBe positive for the entire follow-up period (group D) and 91 (10%) had one or more reversions to HBeAg (median, 2 reactivations [range, 1 to 7 reactivations]) (group E). Of the latter group, 76 persons were HBeAg negative and anti-HBe positive on the last two measurements (median time from first HBeAg-negative test result to last HBeAg-negative result, 7.4 years [range, 0 to 23.7 years]) and 15 were HBeAg positive on the last test date. Figure 1 shows the time until clearance of HBeAg in groups A through E. A log-rank test indicated that the groups differed significantly (P < 0.01).

Among the 532 persons whose first test was HBeAg positive and who were followed for at least 5 years and had at least two HBeAg measurements, the observed probability of clearing HBeAg within 5 years was 0.39 (95% CI, 0.34 to 0.43) in carriers 0 to 18 years of age, 0.56 (CI, 0.43 to 0.68) in those 19 to 30 years of age, and 0.45 (CI, 0.32 to 0.58) in those 31 to 78 years of age; the overall probability of clearing HBeAg within 5 years was 0.41 (CI, 0.37 to 0.45). In a multivariable Cox proportional-hazards model, predicted clearance of HBeAg over 5 years was 0.33, 0.52, and 0.76 for persons 0 to 18 years of age, 19 to 30 years of age, and 31 to 78 years of age, respectively; these values are in reasonable agreement with the observed results.

Figure 2 shows predicted survival curves (time to first HBeAg-positive result) at 10, 20, and 50 years of age. Significant covariates included in the model are age at diagnosis of HBsAg (including a nonlinear effect), an interaction of initial recorded status of HBeAg (positive or negative) and age, and the number of HBeAg measurements recorded. The curve for a 20-year-old carrier most closely approximates the predicted clearance of HBeAg in the cohort as a whole. The observed probability of clearing HBeAg within 10 years was 72.5%. In those who cleared HBeAg, the median time to clearance was 5.6 years (range, <1 to 29.4 years). Older carriers were significantly more likely than younger ones to clear HBeAg (P < 0.001).

A favorable outcome occurred in 1208 (80.1%) of the carriers: 798 of the carriers in group D (4 persons had only one determination) and 410 carriers in group B (22 persons had just converted to anti-HBe on last determination). Three hundred carriers had an unfavorable outcome.

Carriers who had stable serologic status (groups A and D) and were HBeAg positive or anti-HBe positive for the entire duration of follow-up had fewer determinations (P < 0.01) than carriers who had one or more changes in HBeAg/anti-HBe status during follow-up (groups B, C, and E) (Table 1).

Of the 95 persons for whom the approximate time of infection was known, 67 (70.5%) were HBeAg positive and 28 (29.5%) were HBeAg negative and anti-HBe positive on first determination of their HBe status (Table 2). Those who were HBeAg positive on first determination were significantly younger than those who were HBeAg negative (P < 0.01). Of the 67 persons who were HBeAg positive at the time of their first HBsAg-positive result, 9 remained HBeAg positive throughout follow-up, 47 cleared HBeAg during follow-up and remained anti-HBe positive, and 11 reverted to HBeAg positivity at least once after clearing HBeAg. Four of the 28 persons who were initially anti-HBe positive reverted to HBeAg positivity at least once.

During 19 430 person-years of follow-up, 106 (6.9%) of 1536 HBsAg-positive carriers cleared HBsAg, yielding an average clearance rate of 0.5% per year. Hepatitis B surface antigen was cleared in 9 (9.5%) of the 95 carriers whose approximate date of infection was known and in 97 (6.7%) of the 1441 carriers who were HBsAg positive on initial determination. Time to clearance of HBsAg was shorter among carriers who were HBeAg negative on first determination than among those who were HBeAg positive (Figure 3). Multivariable analysis that adjusted for the number of measurements in an individual participant suggested that older carriers, carriers of Yupik ethnicity, and carriers who are initially HBeAg negative clear HBsAg most quickly. Figure 4 shows survival curves for HBsAg in carriers 20 years of age or older or younger than 20 years of age at first determination. After 5 years of follow-up, the 95% CIs of these two groups do not overlap. The median follow-up period for carriers who lost HBsAg (n = 106) was 12.5 years (mean, 14.3 years) compared with 12.4 years (mean, 12.5 years) for carriers who remained HBsAg positive. Two patients developed hepatocellular carcinoma 3 years and 6.5 years after they cleared HBsAg; neither had cirrhosis in the uninvolved liver.

One hundred eighty-four persons identified as carriers died during follow-up. Liver-related deaths occurred in 25 persons; 8 had end-stage liver disease and 17 had hepatocellular carcinoma. An additional 19 carriers developed hepatocellular carcinoma and are currently alive. The 44 adverse events (36 hepatocellular carcinoma and 8 end-stage liver disease) yielded an incidence of 2.3 adverse events per 1000 carrier-years of follow-up (2.5 in men and 1.9 in women). The incidence of hepatocellular carcinoma was 1.9 per 1000 carrier-years (2.3 in men and 1.2 in women). In a Cox proportional-hazards model, Yupik ethnicity (vs. other ethnicity) (hazard ratio, 2.6 [CI. 1.3 to 5.3]), older age (each 1-year increase, tempered by interaction with the number of measurements obtained in a participant) (hazard ratio, 1.04 [CI, 1.0 to 1.07]), and reversion to HBeAg positivity or multiple switches in HBeAg status (hazard ratio, 2.6 [CI, 1.3 to 5.4]) were associated with increased risk for hepatocellular carcinoma, after adjustment for potential confounders. Similar analysis for onset of end-stage liver disease was not done owing to the small number of cases.

Inclusion of eight persons with end-stage liver disease and modeling of the probability of hepatocellular carcinoma or end-stage liver disease by using a multivariable logistic model indicated similar risk factors (all P < 0.01; adjustments were made for duration of follow-up and number of measurements obtained on an individual). Results of the Hosmer–Lemeshow test were nonsignificant (P > 0.2), indicating a plausible fit of the logistic model. The area under the receiver-operator characteristic curve was 0.83.

Previous studies of the natural history of HBsAg carriers have primarily been clinic-based, consisting mainly of carriers with significant liver disease (6,16 - 19). A few population-based studies have been reported that have examined the incidence of hepatocellular carcinoma in HBsAg carriers but not the relationship between serologic events and the incidence of adverse events (20 - 24). We estimate that more than 90% of all HBsAg-positive carriers in our sample have been identified; this is the largest sample and the longest follow-up reported to date.

Two thirds of carriers who were initially HBeAg positive cleared HBeAg and developed anti-HBe during the first 12 years of follow-up, with no evidence of reversion to HBeAg positivity state (Table 1, group B). The observed clearance of HBeAg, as well as the results of two statistical models, generally agree in predicting that approximately 50% of HBeAg-positive carriers will lose HBeAg within 5 years and more than 70% will clear HBeAg within 10 years. Since only nine carriers had received antiviral therapy, five of whom were HBeAg positive at the time, our results suggest that most carriers clear HBeAg without antiviral therapy.

Of carriers who were initially HBeAg negative and anti-HBeAg positive, 90% remained so throughout follow-up. In addition, 80% of carriers who cleared HBeAg remained HBeAg negative, suggesting that once HBeAg is cleared, most carriers remain HBeAg negative and anti-HBe positive. A study conducted in Italy demonstrated that most carriers who remained anti-HBe positive had normal liver enzyme levels and little or no hepatitis on liver biopsy (25). Other studies have shown that carriers who clear HBeAg, either with or without antiviral therapy, have a significant reduction in risk for end-stage liver disease compared with carriers followed for a similar period who do not clear HBeAg (26 - 27).

Approximately 20% of carriers had a less favorable course, characterized by persistence of HBeAg positivity or reversion to HBeAg positivity after previous loss of this marker. Reversion from HBeAg-negative/anti-HBe–positive to HBeAg-positive status in untreated carriers has been associated with elevated aminotransferase levels and histologic exacerbation of hepatitis (28 - 29). In our study, 14% of carriers (200 of 1433) who cleared HBeAg or were initially anti-HBe positive had reversion to HBeAg, similar to the 13% rate of reversions in a study of carriers who cleared HBeAg after interferon therapy (30). These events could conceivably cause significant liver damage. Our data also suggest that carriers experiencing reversions are at a significantly higher risk for hepatocellular carcinoma. Carriers who have reversions or fail to clear HBeAg over a prolonged period may benefit from antiviral therapy.

Hepatitis B surface antigen was lost in 106 (6.9%) of the carriers during follow-up, or 0.5% per year. This rate is consistent with the loss of HBsAg reported in two other studies from Japan and Taiwan, in which the annual clearance of HBsAg ranged between 0.5% and 0.8% (8,31). Carriers who were older and initially anti-HBe positive in our study had a higher clearance rate of HBsAg, similar to the findings of the study from Taiwan. Chronic HBV carriers who clear HBsAg may still have HBV DNA present in both the serum and liver (8,31 - 33). In a previous study, we found HBV DNA in 17 of 33 (52%) carriers who were tested a median of 5 years after clearance of HBsAg (34). Hepatocellular carcinoma developed in 2 of our carriers who cleared HBsAg, and hepatocellular carcinoma has been reported to occur in some carriers who clear HBsAg (31 - 32).

Two patterns of chronic HBV infection have been observed. One pattern is seen in persons who acquire HBV through perinatal infection, 90% of whom become chronically infected. This pattern is frequently seen in Asia and Oceania and in Asian persons living in the United States (35 - 36). These carriers develop immune tolerance to HBV; most remain HBeAg positive throughout childhood, with normal aminotransferase levels and little or no active hepatitis in their livers. The other pattern, seen predominately in Africa but also in Asia and Oceania, occurs in children who are infected after birth and adults who become chronic carriers (37). These carriers usually have elevated aminotransferase levels during the time they are HBeAg positive but apparently convert sooner to anti-HBe than do persons infected at birth (38). In our study group, perinatal infection is infrequent and the presence of HBeAg has been significantly associated with elevation of liver aminotransferase levels (39). Thus, the course of HBV in our sample is similar to the second pattern described above, and our findings probably apply to other groups with similar patterns of transmission.

Treatment of HBeAg-positive carriers with interferon-α or lamivudine has been shown to result in significantly higher clearance of HBeAg, HBV DNA, and HBsAg during the 6 to 12 months after initiation of therapy compared with untreated controls (9,40). Over longer follow-up, it is unclear whether HBeAg is cleared in a higher proportion of carriers who received antiviral therapy than in carriers who remain untreated, or whether antiviral therapy merely accelerates clearance in the short term. This uncertainty is primarily due to lack of studies with long-term follow-up comparing treated participants with untreated controls. Long-term controlled trials are needed to answer this question.

In our study, the most frequent complication of chronic HBV infection was the development of hepatocellular carcinoma. Seroconversion from HBeAg to anti-HBe, and even loss of HBsAg, did not protect carriers from development of hepatocellular carcinoma. In addition, no evidence indicates that antiviral treatment significantly reduces the risk for hepatocellular carcinoma (41). We found older age to be a risk factor for hepatocellular carcinoma, similar to findings of a population-based study in men from Taiwan (20) and findings of a study from Toronto (24).

Several caveats in the interpretation of our findings are warranted. Whereas our cohort appears to be similar to other groups of carriers who have not acquired HBV through perinatal infection, the mean age of our cohort was 20 years at the start of the study; the risk for complications, including hepatocellular carcinoma, may change as the cohort ages. The rates of clearance of HBeAg and HBsAg may not pertain to carriers who are infected during the perinatal period (35 - 36). The frequency of testing may have differed among participants and may have affected serologic outcome. Exclusion of the nine treated carriers would inconsequentially affect the results of clearance of HBeAg and HBsAg, since these participants account for only 1% of the persons who cleared those markers. However, had we excluded them, it would have influenced the incidence of adverse events, since two of these carriers developed hepatocellular carcinoma and one developed liver failure before treatment. Carriers with reversion to HBeAg positivity had more HBe and anti-HBe determinations than did those without reversions, suggesting that we may have underestimated the number of carriers who experienced this phenomenon. The frequency of testing did not differ between carriers living in urban areas and those living in rural areas. Since we found a difference in the incidence of hepatocellular carcinoma between Yupik Eskimos and other Native groups in our sample, differences in the incidence of hepatocellular carcinoma and other complications of chronic HBV in our population could differ from that of other populations. Persons who developed hepatocellular carcinoma were more aggressively followed after diagnosis. Differences among individuals in the frequency of testing could have resulted in earlier detection of hepatocellular carcinoma at a small and asymptomatic stage, thus influencing the incidence rates. Finally, certain assumptions that we made before performing statistical analysis might have affected the findings, such as defining clearance of HBeAg as two consecutive HBeAg negative results instead of one and defining HBsAg carrier status as HBsAg positive for 12 months rather than 6 months.