Changes of Hepatitis B Surface Antigen Variants in Carrier Children Before and After Universal Vaccination in Taiwan

Transcription

1 Changes of Hepatitis B Surface Antigen Variants in Carrier Children Before and After Universal Vaccination in Taiwan HONG-YUAN HSU, 1,2 MEI-HWEI CHANG, 2 SHWU-HUEY LIAW, 3 YEN-HSUAN NI, 2 AND HUEY-LING CHEN 2 Mutants of a determinant of hepatitis B surface antigen (HBsAg) identified in vaccinated children pose a potential threat to long-term success of vaccination programs. We examined the mutants of a determinant (residues ) of HBsAg in hepatitis B virus (HBV) DNA positive children identified during previous serosurveys in Taipei undertaken just before (1984), 5 years after (1989), and 10 years after (1994) universal vaccination began. In HBV DNA positive children from 3 surveys, the prevalence of a determinant mutants increased from 8 of 103 (7.8%) in 1984 to 10 of 51(19.6%) in 1989 and 9 of 32 (28.1%) in 1994 and was higher in those fully-vaccinated than unvaccinated (12/33 vs. 15/153, P.0003). Most amino acid changes of the variants clustered in residues and In all 27 children with detectable mutants, the mean age of those vaccinated was younger than those unvaccinated ( vs yrs, P F.05); and mutations occurred in a region with greatest local hydrophilicity (residues ) more frequently in those vaccinated than in those unvaccinated (10/12 vs. 6/15, P.0253). More mutated residues and more mutations at neutralizing epitopes, such as N146, C147, T148, and C149, were found in the 1994 survey. Vaccinated children may contract variant infections through vertical or horizontal transmission. Universal vaccination has accelerated an accumulation of HBsAg a determinant mutants with amino acid changes critical for immune escape in vaccinated children who became carriers, suggesting that new vaccination strategies should be considered. (HEPATOLOGY 1999;30: ) Hepatitis B virus (HBV) infection is a global health problem and is the major cause of hepatitis, liver cirrhosis, and hepatocellular carcinoma. 1 Hepatitis B virus surface antigen (HBsAg) contains the dominant neutralizing epitopes of HBV and is thus used in current vaccines. Nine different subtypes of HBsAg have been identified, and each consists of a common group-specific (a) determinant and 2 sets of Abbreviations: HBV, hepatitis B virus; HBsAg, hepatitis B surface antigen; PCR, polymerase chain reaction; ALT, alanine transaminase. From the Departments of 1 Emergency Medicine and 2 Pediatrics and the 3 Institute of Molecular Medicine, College of Medicine, National Taiwan University, Taipei, Taiwan. Received February 8, 1999; accepted August 16, This study was supported by a grant (DOH87-DC-1006) from the Department of Health, Executive Yuan, Taiwan, R.O.C. Address reprint requests to: Mei-Hwei Chang, M.D., Department of Pediatrics, National Taiwan University Hospital, No.7, Chung-Shan South Road, Taipei, Taiwan. mhchang@ha.mc.ntu.edu.tw; fax: Copyright 1999 by the American Association for the Study of Liver Diseases / $3.00/0 subtype-specific (d/y, w/r) determinants. 2 The a determinant is the major target for the neutralizing antibody produced during natural infection or following vaccination. 3 Production of antibodies to this a determinant mediates crossprotection against all subtypes. The a determinant is generally regarded as a conformational cluster of epitopes located within residues This notion has been expanded to include residues because of the discovery of 2 new potential epitope clusters. 4 HBsAg variant studies, synthetic peptide studies, studies using the pepscan approach, and studies using site-directed mutagenesis have all suggested that the major HBsAg epitopes are located in the highly hydrophilic conserved residues , 5 particularly the cysteines 6 and residues (KPTDG). 7 Other regions are more diverse and can serve to determine the subtype specificity. 8 Some vaccine-selected escape mutants, in which one amino acid is replaced by another within the a determinant, have been identified in children who acquired HBV infection despite the presence of surface antibody (anti-hbs) These a determinant variants may go undetected by conventional HBsAg screening tests or cause breakthrough infections in immunized persons. This has raised concerns about the long-term success of hepatitis B immunization and suggests the possible need for modification of current vaccines. 11,12 The influence of widespread vaccination on the variety and frequency of these mutants within a vaccinated population has not been well addressed, mainly because of the lack of detailed analyses involving HBV isolates from both prevaccinated and postvaccinated individuals living in the same geographic area where HBV infection is highly prevalent. A mass vaccination program began in Taiwan in We performed serial HBV seroepidemiologic studies in 1984 (just before vaccination), 1989 (5 years after mass vaccination began), and 1994 (10 years after mass vaccination began) and found a significant decrease in the HBsAg carrier rate from 9.8% in 1984 to 1.3% in 1994 in Taiwanese children Taking advantage of the data and stored sera from these studies, we compared the prevalence of a determinant mutants in Taiwanese carrier children before and after the mass hepatitis B vaccination program began to assess the impact of universal vaccination on the emergence of HBsAg a epitope mutants PATIENTS AND METHODS A mass hepatitis B vaccination program was launched in Taiwan in July, The details of the program have previously been described. 16 A baseline seroepidemiologic study of children just before the program launch was conducted in In 1989 and 1994, we conducted postvaccination studies in the same district of Taipei City. 14,15 In 1994, this district had a population of 174,985, of

2 HEPATOLOGY Vol. 30, No. 5, 1999 HSU ET AL which 41,282 were children younger than 15 years old. The population is quite stable, with a migration rate of 10% per year. The subject recruitment and participation rates were similar in all 3 studies. Children under 3 years were recruited from the well-baby clinic at the National Taiwan University Hospital in all 3 studies. They were volunteers recruited by posters and health staff invitation. For school-aged children, we selected 3, 2, and 6 kindergartens in 1984, 1989, and 1994, respectively, from 30 kindergartens in the district; the number selected varied because student population varied. Participation rates were 81%, 84%, and 78% in 1984, 1989, and 1994, respectively. The same elementary school was used for all 3 studies; 4 classes in each grade were randomly selected. Participation rates were 82%, 94%, and 76% in 1984, 1989, and 1994, respectively. All parents of enrolled children signed an informed consent and provided the child s vaccination history based on a booklet provided by the health administration authority in Taiwan. Thus, the number of subjects recruited for HBV marker examinations were 1,200 in 1984, 1,134 in 1989, and 1,515 in 1994 (Table 1). After assaying for serum HBV markers in each participant by radioimmunoassays (Abbott Laboratories, North Chicago, IL), the sera were stored at 70 C until later analyzed. The number of children who received 3 or more doses of hepatitis B vaccine was 356 (31.4%) among 1,134 children in 1989 and 1,289 (85%) among 1,515 children in Maternal and other family members HBsAg serological status were initially investigated in 20 HBsAg positive children identified during the 1994 survey and extended to other HBsAg negative but HBV DNA positive children identified during the same survey after serum PCR assays for HBV DNA. The subjects recruited in each of the original seroepidemiologic surveys were divided into several subgroups according to their hepatitis B serological profiles. The number of subjects with various serological profiles identified during each survey is listed in Table 1. Children who were HBsAg, anti-hbc, and anti-hbs positive represented a subgroup actually having more mutants because breakthrough infections with a determinant variants might occur in the presence of anti-hbs in vaccinated children. Children who were HBsAg and anti-hbc positive but anti-hbs negative represented a subgroup of chronic carriers with the most frequently encountered HBV marker profile, while children who were only HBsAg positive reflected an unusual HBV marker status encountered in chronic carriers. Sera from all HBsAg positive children found during each survey, if available and sufficient for HBV DNA testing, were subjected to nested PCR for amplification of HBV DNA followed by direct sequencing. To prevent possible underestimation of the variants prevalence, we also studied HBsAg negative sera from 3 surveys for the presence of diagnostic escape a determinant mutants. In all 3 surveys, a total of 31 children with anti-hbc positive only markers probably represented the window period of acute infection, carrier state with a high titer of anti-hbc and a very low level of wild type HBsAg undetectable by current assays, or carrier state with an undetectable mutant HBsAg. Of the 31, sufficient sera was available for only 19 children for HBV DNA testing by PCR. Combined anti-hbc and anti-hbs positive only marker profile in unvaccinated children is generally a result of an earlier infection of HBV, whereas the same marker profile occurring in vaccinated children could be because of a breakthrough infection caused by a vaccine escape mutant. 12 Initially, 39 of 48 children with this marker profile in the 1994 survey had sufficient sera available and all were subjected for HBV DNA tesing. Because a large number of children possessed this marker profile in the 1984 and 1989 surveys, it was difficult for us to select the same proportions of anti-hbs and anti-hbc positive subjects as in the 1994 survey. Instead, similar numbers of such subjects to that of the 1994 survey were randomly selected from the 1984 and 1989 surveys. Thus, 33 out of 210 such children in 1984, and 35 out of 101 such children in 1989 were randomly selected for serum HBV DNA testing. There was no statistical difference in sex, age, or district of residence among these 3 groups of children derived from 3 surveys who were studied for HBV DNA, suggesting that these children selected for PCR represented unbiased study samples that could reflect the status of those with combined anti-hbs and anti-hbc only marker profile in the 1984, 1989, and 1994 surveys. Thus, a total of 107 such children from 3 surveys were studied for the presence of serum HBV DNA. Those with positive HBV DNA then were included in nucleotide sequence analysis. Children with anti-hbs positive only represented individuals who have recovered from previous, remote HBV infections with anti-hbc falling to an undetectable level, or who have been vaccinated and without previous HBV infections. It is unlikely that these subjects could be seropositive for HBV DNA and therefore we did not analyze their serum HBV DNA by PCR. The exception is that only 4 children who were classified into 1,140 children with this marker profile in the 1994 survey were processed, because these 4 samples showed very weak HBsAg positivity, positive anti-hbs, and negative anti-hbc on initial HBV marker testing but retesting of the same sera showed HBsAg negativity. However, none of the children with anti-hbs positive only in the 1984 and 1989 surveys showed such an equivocal result of HBsAg status on initial HBV marker testing so that no such subjects were assessed in these 2 surveys. Serum HBV DNA was extracted by proteinase K treatment, phenol/chloroform extraction and ethanol precipitation. The region encoding the a determinant and the flanking sequence (nucleotide position 425 to position 840) of the S gene were amplified by nested TABLE 1. Prevalence of HBV With Surface Gene Mutation in the a Determinant in Serum Samples From Children With Various Serologic HBV Markers During Seroepidemiologic Surveys in the Years 1984, 1989, and Survey 1989 Survey 1994 Survey HBsAg Anti-HBs Anti-HBc (total) (total) (total) No. of children found during the survey (1,200) (1,134) , (1,515) No. of children whose sera were analyzed for HBV DNA by PCR NP NP (148) NP NP (91) NP (65) No. of children seropositive for HBV DNA (103)* (51) (32)* No. of children who harbored a determinant mutants (8)* (10) (9)* Abbreviation: NP, not performed. *P P.0593, by 2 test with Yate s correction. These 4 children were tested because they were very weakly positive for HBsAg on the initial testing but seronegative for HBsAg on retesting.

3 1314 HSU ET AL. HEPATOLOGY November 1999 PCR. The primer pairs and the optimal reaction condition for PCR have been described in our previous study. 17 The 415-base pair (bp) product of the PCR was then sequenced by the Sanger dideoxy chain termination method. The nucleotide sequences encoding the a determinant, and their deduced amino acids were derived from all studied subjects were compared with the published sequences. 18,19 New sequence variants found were confirmed by repeated amplification and sequencing. HBV strains showing mixed signals were cloned and sequenced as previously described 17 to validate that bands at one position in 2 lanes on the sequencing gel were because of 2 sequences. 2 tests with Yate s correction were used to compare the mutation rates between different groups. For quantitation of HBV DNA concentration in sera containing the variants and in sera from HBV DNA positive children with HBsAg negative breakthrough infections, a Digene Hybrid Capture System (Digene Diagnostics, Inc., Beltsville, MD) was used according to the manufacturer s instructions. Briefly, serum samples were mixed with HBV RNA probes and transfered to capture tubes, which were coated with anti-rna:dna hybrid antibody. After the capture step, the immobilized hybrid was then reacted with an anti-hybrid alkaline phosphatase conjugate and detected with a chemiluminescent substrate (LumiPhos, Digene Diagnostics, Inc., Beltsville, MD). The intensity of the light emitted was read on a luminometer (DCR-1, Digene Diagnostics, Inc.) and was proportional to the amount of target DNA in the serum sample. The data was converted to HBV DNA concentrations (pg/ml) as plotted on a standard curve. This assay had a linear range of detection of HBV DNA from 10 pg/ml to 2,000 pg/ml and was considered undetectable if the concentration was below 1 pg/ml. RESULTS a Determinant Mutants in HBsAg Positive Children. Of 174 HBsAg positive children (107 in 1984, 47 in 1989, and 20 in 1994, Table 1) from 3 surveys studied for the presence of HBV DNA by PCR, 169 sera (103 in 1984, 46 in 1989, and 20 in 1994, Table 1) were seropositive for HBV DNA. Among those seropositive for HBV DNA, 8 out of 103 (7.8%) in 1984, 9 out of 46 (19.6%) in 1989, and 5 out of 20 (25%) HBV DNA positive children in 1994 harbored the mutants (Table 1). Of the 57 HBsAg positive children with cocirculating anti-hbs found during the 3 surveys, 5 were fully-vaccinated, and 52 were unvaccinated. Three of the 5 (60%) and 3 of the 52 (5.8%) had a detectable a epitope mutant; this difference was significant ( , P.0026). Frequent exposure to HBV in children before vaccination may produce carrier status with heterotypic anti-hbs, 20 whereas breakthrough infection with a determinant variants in the presence of anti-hbs can occur in vaccinated children. a Determinant Mutants in HBsAg Negative Children. HBV DNA positive sera were also found in HBsAg negative children (Table 1). In children seropositive for both anti-hbs and anti-hbc only, none of 33 sera in 1984, 4 out of 35 (11.4%) in 1989, and 8 out of 39 (20.5%) in 1994 were HBV DNA positive. Only the 2 children with anti-hbs and anti-hbc positive only in 1994 harbored the mutants (Table 1). Eight of these 12 HBV DNA positive sera were from fully vaccinated children, suggesting the possibility of breakthrough infection and hidden HBV carrier status in vaccinated children. The a determinant mutant was detected in one (Table 2, case 94-7) of these 8 fully vaccinated children and in one (Table 2, case 94-1) of the remaining 4 unvaccinated children. In children seropositive for anti-hbc alone, none of the 8 TABLE 2. Clinical Characteristics and HBsAg a Determinant Variants Among 27 HBV-Carrier Children From Seroepidemiologic Surveys in the Years 1984, 1989, and 1994 Case No. Age Sex HBsAg Anti HBs Anti HBc HBeAg Anti HBe Vaccination Subtype Variants of a Determinant HBV DNA (pg/ml) y9m M u adw T126A NA y5m F u adw M133L 4, y5m F u adw F134L y1m M u adw C138S 1, y F u adw T140R y3m M u adw T140I y10m F u adw T143M 2, y7m M fv adw D144A y M u adw T126A y M u adw T126A y8m M u adw P127T y1m M u adw Q129H 3, y8m F u adr S143W (mix) 1, y4m M u adr S143W (mix) 2, y6m F fv adw G145R (mix) y M u adw G145R y2m M fv adw G145R y11m M fv adw W156L y F u adw T125A y6m F fv adw P120Q P127T 1, y F fv adw T126A T143M y11m M fv adw T126S (mix) D144H (mix) NA y7m F fv adw T140P (mix) 2, y4m M fv adw D144H (mix) G145R m M fv adw N146S y6m M fv adw T148I NA y10m F fv adw C147R (mix) C149R (mix) 1,979 Abbreviations: y, years; m, months; M, male; F, female; u, unvaccinated; fv, fully vaccinated; mix, mixed viral population containing both the mutant and wild type virus; NA, insufficient serum for this testing.

4 HEPATOLOGY Vol. 30, No. 5, 1999 HSU ET AL sera in 1984, one of the 9 sera in 1989, and none of the 2 sera in 1994 were HBV DNA positive. Only the one serum positive for HBV DNA found in 1989 contained the mutant (Table 2, case 89-2). In 1994 sera positive for anti-hbs only, 4 samples virtually showed very weak HBsAg positivity, positive anti-hbs and negative anti-hbc on initial testing but retesting of the same sera revealed HBsAg negativity. All 4 were HBV DNA positive on nested PCR and 2 of them (Table 2, case 94-4 and case 94-8) contained the mutants, suggesting a low level of replication that was insufficient to induce a detectable anti-hbc response. The 2 mutants associated with HBsAgnegative retesting may have been a result of the assay s failure to detect the variant antigen or the presence of exceptionally low serum HBsAg concentration. Frequency and Genotypes of a Determinant Mutants in the 3 Surveys. As shown in Table 1, the overall prevalence of a determinant mutants in HBV DNA positive children increased from 8 out of 103 (7.8%) in 1984 to 10 out of 51 (19.6%) in 1989 and 9 out of 32 (28.1%) in 1994 (7.8% in 1984 vs. 28.1% in 1994, P.0064). Table 2 shows that, except for the viral provenance of S143W variant, which was of genotype C (adr subtype). The remaining various strains of HBsAg variants we found belonged to genotype B (Asian adw2 subtype). Based on the variants found by us and others, it seems that there was no association between specific variants and various genotypes collected from various geographic areas. Novel Variants Identified That Are Different From Previous Studies. A total of 22 variants were found in our 3 surveys. Nine variants have been reported previously: P120Q in one HBV strain from France, 19 in which variation at residue 120 was found to affect the binding of the major epitope residues to antibodies; 21 T126A, Q129H, and M133L found in infants with chronic infections despite immunoprophylaxis, 22 in which T126 seems important for the w subdetermination and affects the recognition properties of the surface antigen; 8 P127T, in which residue 127 is the critical site for the serological variations between w1/w2 (Pro) in relation to w3 (Thr), and w4 (Leu); 8 F134L, which is an HBV mutant from one vaccinated Taiwanese carrier child; 23 C149R variant in Sardinia, 4 in which C149 was previously shown to be important for its immunological activity; and D144A and G145R which were proposed to be vaccine-induced escape mutants. 9,11 Three residue mutations previously observed have new changes: T126S, T143M, S143W, and D144H, whereas T143L has been seen in many sporadic isolates from Spain. 21 Nine novel variants in 7 new residues were found: T125A, C138S, T140R/I/P, N146S, C147R, T148I, and W156L. In all variants we found, the most prevalent mutated residues were residues 126 (n 5), 145 (n 4), 143 (n 4), 144 (n 3), and 140 (n 3). G145R is the most commonly found mutant all over the world. 23 To the best of our knowledge, 13 out of 22 variants found were novel ones. Changes of Frequency and Amino Acid Substitutions of the a Determinant Mutants in Relation to Vaccination. In all HBV DNA positive children from 3 surveys, the prevalence of a determinant mutants in fully-vaccinated children was much higher than that in unvaccinated children (12/33 vs. 15/153, , P.0003). We found an increasing number of mutated residues in the 1984, 1989, and 1994 surveys (8, 10, and 14 mutated residues, respectively; Table 1). Taken together, in all 27 HBV DNA positive children with detectable mutants, the age of those vaccinated was younger than those unvaccinated (mean yrs [range 8 m-12 yrs] vs. mean 7.9 yrs 2.3 yrs [range 2 yrs, 9 m-12 yrs], P.05, Student s t test); and amino acid substitutions within residues were observed in 10 of 12 (83.3%) fully vaccinated carrier children but in only 6 out of 15 (40%) unvaccinated carrier children (P.0253, Fisher s exact test). In addition, more mutations occurring at important residues, such as N146, C147, T148, and C149, were noted in the 1994 survey. The coexistence of 2 mutants or coexistence of a wild and mutant type of HBV in serum was observed in some patients, particularly those drawn from the 1994 survey (Table 1). Variants may arise when wild type viruses are subjected to immunological pressure, giving rise to a mixed viral population, as was previously observed in immunized infants born to carrier mothers infected with only the wild type virus. 17 HBeAg and Quantification of Serum HBV DNA in Children who Harbored the Mutants. Among all 27 carrier children with detectable a determinant mutants identified from 3 separate surveys, 7 out of 12 vaccinated (58.3%) and 10 out of 15 unvaccinated (66.7%) were seropositive for HBeAg, suggesting an active viral replication and possibly high infectivity of the hosts. There was sufficient available serum for quantitation of HBV DNA for 24 of the 27 children infected with the variants. Serum HBV DNA levels ranged from 0 to 3,027 pg/ml (Table 2). Sera containing Q129H, T143M, G145R, or C147RC149R had relatively high concentrations of HBV DNA as well as were positive for HBeAg, suggesting their high infectivities. In contrast, serum HBV DNA from 12 children with HBsAg negative infections was detectable by nested PCR but undetectable by the hybridization method, suggesting their low level viremia. Cloning and sequencing of the cloned PCR products confirmed that 4 children (cases 89-5, 89-6, 89-7, and 94-5) harbored the mutant and wild type viruses; and that 2 children (cases 94-4 and 94-9) had a mixed infection by 2 variants as well as wild type viruses. Maternal HBsAg Status in Relation to their Children Contracting Variant Infections. The HBV carrier status of other family members was known in only 25 out of 32 HBV DNA positive children identified in the 1994 survey. Among the 25 who were fully-vaccinated, 4 out of 9 children of carrier mothers and 5 out of 16 children of noncarrier mothers harbored the a determinant mutants (P.05) (Table 3). Because children of noncarrier mothers did not receive HBIG during immunoprophylaxis, HBIG administration is not an essential factor for the emergence of the mutants. The mothers of cases 94-3, 94-6, 94-7, and 94-9 were HBsAg carriers; the father of case 94-2, the uncle of case 94-8, and 2 siblings of case 94-4 were HBsAg carriers. None of the family members of cases 94-1 and case 94-5 were HBsAg carriers. This suggests that both perinatal and horizontal transmission are possible routes for TABLE 3. The Relationship Between the Presence of a Determinant Mutants and Maternal HBsAg Status Among 25 HBV DNA Positive Children With Known Carrier Status of Family Members Identified in the 1994 Survey a Determinant Mutant (n) Wild Type HBV (n) Maternal HBsAg () 4 5 Maternal HBsAg () 5 11 NOTE. 2 test with Yate s correction. P.8214.

5 1316 HSU ET AL. HEPATOLOGY November 1999 our vaccinated infants and children to contract the variant infections. Extents of Liver Diseases in Relation to a Determinant Mutants. Eight out of 22 patients who harbored a determinant mutants received long-term, regular follow-up for liver function tests and HBV markers. Of the 5 who were initially seropositive for HBeAg, 2 (T143M, Q129H) developed hepatitis activity at some points (peak ALT level: 840 and 33 IU/L, respectively) but remained HBeAg positive; one (D144A) developed transient and mild ALT rising (40 IU/L) with subsequent seroconversion to anti-hbe; the remaining 2 (C138S, S143W) had persistent normal ALT levels and HBeAg positivity on followup. Of the 3 who were initially seronegative for HBeAg but seropositive for anti-hbe, 2 (T140R, T140I) had persistent normal ALT levels and were positive for anti-hbe, whereas the remaining one (F134L) developed abnormal liver function and died of hepatocellular carcinoma at the age of 11 years old. DISCUSSION Taking advantage of previous serosurveys in children living in the same area before and after universal immunization, we have provided a meaningful comparison of surface antigen variants between unvaccinated and vaccinated children populations within the same age range. Convincing evidence for the influence of universal vaccination on the emergence of a determinant mutants comes from the following observations: (1) In all HBV DNA positive children from the 3 surveys, there was a significantly higher prevalence rate of the mutants in those fully-vaccinated (36.4%) than that in those unvaccinated (9.9%). (2) The prevalence of a determinant mutants in HBV DNA positive children increased from 7.8% in 1984 to 19.6% in 1989 and 28.1% in 1994 survey. If the mutants emerged independent of vaccination, a dramatic fall in the carrier rate in children with universal vaccination would be accompanied by a steady prevalence rate of a determinant mutants in the carrier children population from each survey. (3) In children with detectable mutants, the mean age of those vaccinated was younger than those unvaccinated, suggesting that the emergence of the mutants was accelerated by the universal vaccination program; and amino acid substitutions at residues , which constitute as a neutralizing epitopes cluster, were more frequently found in those vaccinated than those unvaccinated. (4) There were an increasing number of mutated residues from the 1984, 1989, and 1994 surveys and more mutations occurring at important residues, especially neutralizing epitopes within the a determinant mutants, were noted in the 1994 survey. (5) We and others have previously shown that a determinant mutants were detected in immunized infants born to HBeAg positive carrier mothers infected with only the wild type virus, 9,10,17 suggesting that the mutants had emerged following vaccination. In unvaccinated, HBsAg negative children with positive anti-hbs and anti-hbc, a detectable serum HBV DNA possibly represent a remnant viremia during loss of HBsAg when they were clearing viral infection after acute hepatitis. This study and previous reports also showed that HBV DNA can be detected by PCR in vaccinated children with HBsAg negative breakthrough infections. 24,25 An a determinant mutant, probably a diagnostic escape mutant, was found in one of 8 such infections. For the remaining 7 cases, natural wild type HBV infection in vaccinated children, perhaps through frequent household contact with infective carriers, may result in a low level viremia, evidenced by a detectable serum HBV DNA by PCR but undetectable by the hybridization method. In a birth cohort of vaccinated infants of HBeAg positive carrier mothers, we found that in some infants a detectable serum HBV DNA by PCR only with negative HBsAg coincided with the appearance of anti-hbc. Such low level viremia may be transient with rapid clearance of HBV DNA, or persist for a relatively long period (Lee PI and Hsu HY, unpublished observations, October, 1998). The other possibility is that the remaining HBsAg negative breakthrough infections may associate with surface gene mutations located outside the a determinant. It has been suggested that HBsAg changes as variants include subtype-related variation, found in natural isolates and selected over centuries; and mutants, which have been selected over a short period by antibody, such as therapy, immunization, prophylaxis with HBIG, or even natural anti-hbs seroconversion. 26 However, many subtype-related variants occured after active immunization in the present study and other reports, 22,28 and mutations, for example T143M and G145R, occur naturally in our unvaccinated children. G145R in natural isolates has also been reported in Japan. 28 It is interesting that most of the amino acid changes observed in our surveys were clustered in 2 regions, residues and residues Residues have been shown to have the greatest local hydrophilicity, 29 contain predominantly antigenic a determinants, and may constitute the major neutralizing epitopes, whereas residues have been shown to be subtype-specificity determinants. 8 Results in our 3 surveys suggested that both types of variants could be selected by natural human immunity and vaccination. Mutation in each type of variant may change its antigenicity, evade immune pressure, and then contribute to viral persistence. Nevertheless, mutations at the neutralizing epitopes theoretically may result in larger antigenic changes. More novel epitope neutralization-related mutants such as N146S, C147R, T148I, and C149R in 1994 may imply that vaccine-selected mutation may be more likely to occur at the neutralizing epitopes than at subtype-specificity determinants, because of larger immunologic changes. This consideration is consistent with reports that most previously proposed vaccine-selected mutations occur at residues ,4 In the variants we identified, G145R is a well characterized immune escape mutant. In addition, T126S and Q129H have also been shown to exhibit various degrees of altered binding of HBsAg to several monoclonal antibodies. 30 By studying the changes in hydrophilicity and conformation resulting from exchange of residues in the a determinant, amino acid substitutions in T126A, M133L, F134L, and T143M were found to have changes in the turn secondary structure. 31 The substitution of amino acid residues within the presumed second loop (residues ) of a determinant may involve a change of the side chain of amino acid residues, for example, a loss of a negative charge in D144A mutant. In N146S variant, the asparagine at amino acid position 146 is the site of N-linked glycosylation and variation at this position may reduce antibody binding. Antigenicity is lost when the cysteine residue at amino acid position 147 or 149 is substituted. 32 Failure of T156L variant to bind to anti-hbs has also been shown. 33 Based on the above information, many mutants we identified may be able to escape HBV vaccines.

6 HEPATOLOGY Vol. 30, No. 5, 1999 HSU ET AL However, altered reactivity with anti-hbs in many of the variants still needs further confirmation by expressing these mutants in an in vitro expressing system, and performing antibody binding studies with the serum of patients compared with cell lines transfected with wild type HBV. Because the absolute number of carriers with a determinant mutants had not increased throughout 3 serosurveys between 1984 and 1994, we believe that a determinant mutants have not become widespread so far. However, unless infections with the mutants are transient in nature, with ongoing universal vaccination there will be a gradual increase in the total number of carriers with the mutants in the whole population. Carriers with these escape mutants could potentially transmit HBV by blood donation with undetectable HBsAg or through community-acquired infection both of which could not be prevented by current vaccines or hepatitis B immunoglobulin. In summary, the current vaccination programs are effective in protecting the vaccinees against perinatal transmission of the wild type HBV, which still constitutes the majority of the circulating HBV in Taiwan and presumably in other parts of the world and should be continued. However, our results indicate that amino acid changes within the a determinant are different in location and type between vaccinated and unvaccinated carrier children. Universal vaccination has accelerated an accumulation of more mutated residues and more mutations at neutralizing epitopes within the a determinant of HBsAg in vaccinated children who became carriers. These immune escape mutants are potentially widespread in a vaccinated population because of their selective advantage in the presence of anti-hbs. Further characterization of the variants antigenicity and continuous monitoring of the circulating HBV strains in our vaccinated population will finally determine the necessity of incorporating wild-type and mutant S proteins in future HBV vaccines. REFERENCES 1. Beasley RP, Hwang LY, Lin CC, Chien CS. Hepatocellular carcinoma and hepatitis B virus: a prospective study of men in Taiwan. Lancet 1981;2: Howard CR, Allison LM. Hepatitis B surface antigen variation and protective immunity. Intervirology 1995;38: Waters JA, O Rourke SM, Richardson SC, Papavengelou G, Thomas HC. Quantitative analysis of the humoral immune response to the a determinant of HBs antigen after innoculation with plasma derived or recombinant vaccine. J Med Virol 1987;21: Carman WF, Van Deursen FJ, Mimms LT, Hardie D, Coppola R, Decker R, Sanders R. The prevalence of surface antigen variants of hepatitis B virus in Papua New Guinea, South Africa, and Sardinia. HEPATOLOGY 1997;26: Gerin JL, Alexander H, Shih JWK, Purcell RH, Dapolito G, Engle R, Green N, et al. Chemically synthesized peptides of hepatitis B surface antigen duplicate the d/y specificities and induce subtype-specific antibodies in chimpanzees. Proc Natl Acad Sci U S A 1983;80: Prange R, Mangold CM, Hilfrich R, Streeck RE. Mutational analysis of HBsAg assembly. Intervirology 1995;38(1-2): Neurath AR, Pride MW, Strick N, Thanavala YM. Toleration of amino acid substitutions within hepatitis B virus envelope protein epitopes established by peptide replacement set analysis. I. Region S ( ). Peptide Res 1990;3: Magnius LO, Norder H. Subtypes, genotypes and molecular epidemiology of the hepatitis B virus as reflected by sequence variability of the S-gene. Intervirology 1995;38: Carman WF, Zanetti AR, Karayannis P, Waters J, Manzillo G, Tanzi E, Zuckerman AJ, et al. Vaccine-induced escape mutant of hepatitis B virus. Lancet 1990;336: Okamoto H, Yano K, Nozaki Y, Matsui A, Miyazaki H, Yamamoto K, Tsuda F, et al. Mutations within the S gene of hepatitis B virus transmitted from mother to babies immunized with hepatitis immuneglobulin and vaccine. Pediatr Res 1992; 32: Harrison TJ, Oon CJ, Zuckerman AJ. A novel antibody escape variant (Ala144) of hepatitis B virus in an identical twin before selection in mother. In: Nishioka, Suzuki H, Mishiro S, Oda T, eds. Viral Hepatitis and Liver Diseases. Springer-Verlag, Tokyo 1994: Fortuin M, Karthigesu V, Allison L, Howard C, Hoare S, Mendy M, Whittle HC. Breakthrough infections and identification of a viral variant in Gambian children immunized with hepatitis B vaccine. J Infect Dis 1994;169: Hsu HY, Chang MH, Chen DS, Lee CY, Sung JL. Baseline seroepidemiology of hepatitis B virus infection in children in Taipei, 1984: A study just before a mass hepatitis B vaccination program in Taiwan. J Med Virol 1986;18: Tsen YJ, Chang MH, Hsu HY, Lee CY, Sung JL, Chen DS. Seroprevalence of hepatitis B infection in children in Taipei, 1989: five years after a mass hepatitis B vaccination program. J Med Virol 1991:34: Chen HL, Chen MH, Ni YH, Hsu HY, Lee PI, Lee CY, Chen DS. Seroepidemiology of hepatitis B virus infection in children: Ten years after a hepatitis B mass vaccination program in Taiwan JAMA 1996;276: Chen DS, Hsu NH, Sung JL, Hsu TC, Hsu ST, Kuo YT, Lo KJ, et al. A mass vaccination program in Taiwan against hepatitis B surface antigencarrier mothers. JAMA 1987;257: Hsu HY, Chang MH, Ni YH, Lin HH, Wang SM, Chen DS. Surface gene mutants of hepatitis B virus in infants who develop acute or chronic infections despite immunoprophylaxis. HEPATOLOGY 1997:26: Ono R, Onda H, Sasada Y, Igarashi K, Sugino Y, Nishioka K. The complete nucleotide sequences of the cloned hepatitis B virus DNA: subtype adr and adw. Nucleic Acid Res 1983;11: Norder H, Hammas B, Lofdahl S, Courouce AM, Magnius LO. Comparison of the amino acid sequences of nine different serotypes of hepatitis B surface antigen and genomic classification of the corresponding hepatitis B strains. J Gen Virol 1992;73: Chen DS, Lai MY, Sung JL. Anti-HBs reactivity in hepatitis B surface antigen positive serum samples-with special emphasis on heterotypic antibody. J Formosan Med Assoc 1982;81: Wallace LA, Echevarria JE, Echevarria JM, Carman WF. Molecular characterization of envelope antigenic variants of hepatitis B virus from Spain. J Infect Dis 1994;170: Oon CJ, Lim GK, Ye Z, Goh KT, Tan KL, Yo SL, Hopes E, et al. Molecular epidemiology of hepatitis B virus vaccine variants in Singapore. Vaccine 1995;13: Lee PI, Chang LY, Lee CY, Huang LM, Chang MH. Detection of hepatitis B surface gene mutation in carrier children with or without immunoprophylaxis at birth. J Infect Dis 1997;176: Chabaud M, Depril N, Le Cann P, et al. Detection of hepatitis B virus DNA by polymerase chain reaction in vaccinated and non-vaccinated Senegalese children. Arch Virol 1993;8(Suppl):123S-31S. 25. Karthigesu V, Allison LML, Fortuin M, Mendy M, Whittle HC, Howard CR. A novel hepatitis B virus variant in the sera of immunized children. J Gen Virol 1994;75: Carman WF. The clinical significance of surface antigen variants of hepatitis B virus. J Viral Hep 1997;4(Suppl):11S-20S. 27. Kato J, Hasegawa K, Tori N, Yamauchi K, Hayashi N. A molecular analysis of viral persistence in surface antigen-negative chronic hepatitis B. HEPATOLOGY 1996;23: Yamamoto K, Horkita M, Tsuda F, Ioh K, Akahane Y, Yotsumoto S, Okamoto H, et al. Naturally occurring escape mutants of hepatitis B virus with various mutations in the S gene in carrier children seropositive for antibody to hepatitis B surface antigen. J Virol 1994;68: Hopp TP, Wand KR. Prediction of protein antigenic determinants from amino acid sequences. Proc Natl Acad Sci U S A 1981;78: Chiou HL, Lee TS, Kuo J, Mau YL, Ho MS. Altered antigenicity of a determinants of hepatitis B virus. J Gen Virol 1997;78: Zhang YY, Nordenfelt E, Hansson BG. Increasing heterogeneity of the a determinant of HBsAg found in the presumed late phase of chronic hepatitis B virus infection. Scand J Infect Dis 1996;28: Mangold CMT, Streeck RE. Mutational analysis of the cysteine residues in the hepatitis B virus small envelope protein. J Virol 1993;67: Norder H, Hammas B, Lee SD, Bile K, Courouce AM, Mushahwar IK, Magnius LO. Genetic relatedness of hepatitis B viral strains of diverse geographical origin and natural variations in the primary structure of the surface antigen. J Gen Virol 1993;74: