Maternofetal transmission of hepatitis B virus genotype E in Ghana, west Africa

Transcription

1 Journal of General Virology (2007), 88, DOI /vir Maternofetal transmission of hepatitis B virus genotype E in Ghana, west Africa Daniel Candotti, 1 Kwabena Danso 2 and Jean-Pierre Allain 3 Correspondence Daniel Candotti dc241@cam.ac.uk 1 National Health Service Blood and Transplant, Cambridge Blood Centre, Cambridge, UK 2 Department of Obstetrics and Gynaecology, Komfo Anokye Teaching Hospital, Kumasi, Ghana 3 Division of Transfusion Medicine, Department of Haematology, University of Cambridge, Cambridge, UK Received 19 April 2007 Accepted 21 June 2007 To determine whether maternofetal transmission of hepatitis B virus (HBV) is a common route of infection leading to chronic infection in west Africa, plasma samples, obtained at delivery from 1368 pregnant Ghanaian women and paired umbilical cord blood or newborn whole blood samples, were tested for HBV surface antigen (HBsAg) and DNA. A 16 % prevalence of HBV chronic carriers, defined as detectable HBsAg and/or HBV DNA, was found,.80 % contained less than 1¾10 4 IU ml 1 HBV DNA and 99 % were infected with genotype E strains. HBV maternofetal transmission was documented in 17 out of 204 (8.3 %) paired HBV carrier women cord blood/newborn samples. The rate of transmission was 55 % and 3.3 % when maternal viral load was above or below 1¾10 4 IU ml 1, respectively (P ). Maternofetal transmission of HBV genotype E was estimated to account for 8 % of the cases of chronic HBsAg carriers. Six women with low viral load at delivery (five,20 IU ml 1 ) and anti-hbe (hepatitis B e antigen) transmitted HBV. Surprisingly, while non-transmitted low viral load strains had 79 % mutations at position 1896 of HBV genome, transmitted strains were all wild-type despite anti-hbe presence (P ), suggesting the possible role of HBeAg as risk factor for HBV maternofetal transmission. The relative risk of maternofetal transmission was 2.4 when pregnant women carried high viral load and 11.5 when carrying wild-type strains at position 1896, irrespective of viral load. We conclude that viral load and pre-core wild-type at position 1896 are two independent risk factors for HBV genotype E maternofetal transmission, which remains a minor contributor to high prevalence of chronic infection. INTRODUCTION The prevalence of chronic hepatitis B virus (HBV) infection varies widely according to geographical area, and is closely interlinked with the predominant routes of HBV transmission (Maddrey, 2000). The age at which HBV infection occurs is one of the main factors that predispose to the acquisition and frequency of the chronic carriage status. Almost 90 % of infants born to HBV surface antigen- (HBsAg) and hepatitis B e antigen (HBeAg)-positive mothers and approximately 30 % of children infected before 6 years of age become chronic carriers, compared with less than 10 % of older children or adults (Hyams, 1995; Ip et al., 1989). Children may be infected horizontally in early childhood or perinatally from carrier mothers. Three mechanisms of HBV transmission from HBsAg-positive The GenBank/ENBL/DDBJ accession numbers for the sequences determined in this study are EF EF and EF EF mothers to infants were suggested: (i) transplacental intrauterine transmission; (ii) transmission during delivery by contact with maternal infected fluids in the birth canal and (iii) postnatal transmission from mothers to infants during childcare or through breast feeding (Ghendon, 1987; Shepard et al., 2006). Sub-Saharan Africa is an area of high endemicity in which more than 75 % of adults have been exposed to HBV, with an estimated 5 25 % being chronic carriers. The predominant route of HBV transmission is horizontal, most children being already infected by the age of five (Dumpis et al., 2001; Kew, 1996; Martinson et al., 1998). Maternofetal transmission seems to play a minor role in Africa, in contrast to what is observed in other highprevalence endemic areas, such as south-east Asia, where maternofetal transmission is the dominant route of transmission (Zhang et al., 1998). Here, a cross-sectional analysis of HBV serological and molecular parameters was conducted in Ghanaian pregnant G 2007 SGM Printed in Great Britain

2 HBV genotype E maternofetal transmission women at delivery and in paired cord blood/newborn whole blood samples to investigate the part played by maternofetal transmission in the overall epidemiology and natural history of HBV genotype E infection in west Africa. METHODS Patient samples. Plasma samples from pregnant women (coded M) and corresponding cord blood (coded C) were collected at delivery in the Department of Obstetrics & Gynaecology, Komfo Anokye Teaching Hospital, Kumasi, Ghana. When possible, a newborn whole blood sample (coded B) was collected two weeks after birth. This study was approved by the University of Science and Technology School of Medical Sciences committee on human research publication and ethics, Kumasi, Ghana. Informed consent was obtained from the women included in the study. No clinical outcome suggesting HBVrelated illness was reported. Serological testing. HBsAg was screened using a rapid test (1 2IUml 21 sensitivity) (Determine, Abbott Laboratories) and, in some cases, the Bioelisa HBsAg Colour kit (Biokit). HBeAg and anti- HBe were detected with Murex HBeAg/anti-HBe enzyme immunoassay (EIA) (Abbott-Murex). Antibodies against hepatitis B surface antigen (anti-hbs) were detected with Murex anti-hbs EIA and IgM anti-hbc antibodies with Murex anti-hbc IgM EIA (Murex Biotech). HBV DNA detection. Viral DNA was screened using a multiplex real-time quantitative PCR (qpcr) assay that simultaneously detected and identified HBV, human herpesvirus 8 and parvovirus B19 DNA (Candotti et al., 2006a). The 95 % and 50 % detection limits of this assay were 25 and 10 IU ml 21, respectively. HBV DNA was quantified by using a single-virus real-time qpcr assay (Allain et al., 2003). HBV DNA-positive samples were confirmed using a hemi-nested PCR in the basic core promoter/precore region (BCP/PC) (Candotti et al., 2006b). A second nested-pcr was used to amplify a 1434 bp fragment including the whole pres/s gene. PSS2 (59-ACATACTCTT- TGGAAGGCKG-39; positions ) and PSS8 (59-CGTCA- GCAAACACTTGGC-39; positions ) were used in the first amplification round, and PSS9 (59-GCCTCATTTTGYGGGTCA-39; positions ) and PSS5 (59-AGCAAARCCCAAAAGACC-39; positions ) in the second round. RESULTS Serological and molecular detection of HBV chronic infection in pregnant women Individual plasmas from a cohort of 1368 pregnant women were initially screened for HBsAg at delivery using a rapid serological test (Determine HBsAg) (Fig. 1). HBsAg was detected in 174 plasmas and confirmed by EIA in 173 (12.6 % of total). Age ranged between 15 and 48 years (median: 27 years) and no difference in age distribution was observed between HBV carriers and non-carriers. All mothers were negative for anti-hiv and none had been vaccinated against HBV or had received antiviral treatment. HBV DNA was detected by multiplex real-time qpcr in 207 samples (161/173 HBsAg reactive and 46/1195 HBsAgnon-reactive plasmas). All 207 plasmas were confirmed HBV DNA-positive with a nested-pcr assay and the prevalence of confirmed HBV viraemia was 15.1 % (Fig. 1, Table 1). The 46 HBsAg-negative/HBV DNA-positive plasmas were retested for HBsAg by EIA with a 0.2 IU ml 21 sensitive test (Biokit). Twenty-six of these plasmas were EIA-reactive, increasing the number of HBsAg-positive samples to 199 and the prevalence to 14.5 %. Of these 46 samples, 20 were HBV DNA-positive/ HBsAg-negative, indicating a 1.5 % prevalence of occult HBV infection (Table 1). Surface antigen and viral DNA were associated in 187 samples (13.7 %). Twelve samples (0.9 %) were confirmed HBsAg-positive and HBV DNAnegative. The overall prevalence of markers of HBV chronic infection was 16.0 %. Plasma viral load (median 51.8 IU ml 21 ) followed a biphasic distribution, with 183 samples (88.4 %) having HBV DNA levels below IU ml 21 (Fig. 2). In 70 randomly selected pregnant women, viraemic samples were distributed between 47 with low viral load HBV cloning, sequencing and genotyping. PreS/S amplicons were cloned using the TOPO TA cloning kit (Invitrogen). Sequences of BCP/ PC and pres/s regions were obtained by direct sequencing of PCR products or by sequencing clones. The Ghanaian sequences were aligned with reference HBV genotypes A H sequences using the CLUSTAL W software implemented in Mac Vector version 7.2 software (Accelrys), and the alignment was confirmed by visual inspection. Phylogenetic analysis was performed using the PAUP 4.0b10 software (Sinauer Associates) after exclusion of positions containing an alignment gap from pairwise sequence comparisons (Candotti et al., 2006b). Nucleotide distances were analysed by a neighbour-joining algorithm based on Kimura two-parameter distance estimation. To confirm the reliability of the phylogenetic trees, bootstrap resampling was performed for each analysis (1000 replicates). The GenBank accession numbers of the nucleotide sequences analysed in this study are EF EF (BCP/PC region) and EF EF (PreS/S region). Statistical analysis. Analyses were carried out using the SPSS software. Categorical variables were compared using Fisher s exact test and, for continuous variables, the non-parametric Mann Whitney test. All values shown were derived from the results of a two-tailed test. P,0.05 was considered statistically significant. Fig. 1. Screening of Ghanaian pregnant women and paired cord blood plasmas/newborn whole blood for HBV infection markers. Pos., positive; neg., negative

3 D. Candotti, K. Danso and J.-P. Allain Table 1. Distribution of HBsAg and HBV DNA in 1368 pregnant women at delivery HBsAg HBV DNA Number of pregnant women (%) Median viral load in IU ml 1 (range) (13.7) 63 (, ) (0.9) NA (1.4) 31 (, ) (84) NA NA, not applicable (ranging between,10 and IU ml 21 ) and 23 with high viral load (ranging between and IU ml 21 ), genotype E was identified in 69 samples (98.6 %) and one strain was genotype A1 (1.4 %). Identification and frequency of HBV maternofetal transmission In order to assess transmission by pregnant women carrying HBsAg and/or HBV DNA, the presence of HBV DNA was investigated in paired plasmas from cord blood collected at delivery and/or newborn whole blood samples collected approximately 2 weeks after birth (Fig. 1, Table 2). Confirmed HBV DNA was repeatedly detected in 17 paired samples from 204 available cord blood and/or newborn samples. The prevalence of HBV DNA in cord blood/ newborn samples was therefore 17/1353 or 1.3 %. In only one case, both cord blood and newborn samples were available and both carried HBV DNA (pair 1449). In the others, either cord blood (n512) or newborn (n54) samples were available. Fifteen cases of apparent maternofetal transmission were found in women carrying both HBsAg and HBV DNA. In two cases, transmission from HBsAg-positive/HBV DNA-negative women was found in the newborn sample (1237B and 1350B) (Table 2). To verify the apparent absence of detectable viral DNA in these two women (1237M and 1350M), nucleic acids were extracted from 500 ml (instead of 200 ml) of plasma and nested-pcr confirmed that HBV DNA was present with a viral load,10 IU ml 21. No viral DNA was detected in cord blood or newborn samples from 20 women carrying occult HBV infection (Table 2). HBV DNA level in cord blood plasma and newborn whole blood was consistently low, ranging between,10 and IU ml 21 (median: 30 IU ml 21 ) and between 15 and IU ml 21 (median: 43 IU ml 21 ), respectively (Table 3). None of the cord blood plasmas contained IgM anti-hbc. HBsAg and HBeAg were present in 7 and 10 of the 13 cord blood plasmas, respectively. In order to confirm transmission at the molecular level, paired HBV-containing maternal and cord blood/newborn samples were further investigated by phylogenetic analysis of pres/s region sequences (1293 nucleotides) obtained from all samples except 1975C, 623B and 1237B, which contained very low HBV DNA loads (10, 15 and 43 IU ml 21, respectively). Fig. 3 demonstrates close genetic relationship, but not identity, between maternal and cord blood/newborn sequences from 11 of the 14 pairs. Identical consensus sequences were found in two pairs (1648M/ 1648C and 1660M/1660C). Relatively high genetic distance between pregnant woman and cord blood was observed in pair Within the group of sequences studied, the genetic diversity observed within each pair or triplet ranged between 0 and 0.15 %. In contrast, a 1.0 % genetic diversity was observed between samples 2075M and 2075C. The overall inter-sample genetic diversity was 1.1 % (range: %). In order to determine whether the features of pair 2075 were due to external contamination or to fetus infection with a maternal variant distant from the consensus sequence obtained, seven clones from 2075M and 2075C amplicons were sequenced and phylogenetically analysed 2688 Journal of General Virology 88 Fig. 2. Distribution of HBV DNA load in HBsAg-positive mothers (black bars) and frequency of HBV transmission in each group (grey bars). The right part of the graph shows that high viral load carries % risk of maternofetal HBV transmission while the left side of the graph indicates that maternal low viral load still carries 5 20 % risk of transmission.

4 HBV genotype E maternofetal transmission Table 2. Detection of HBV DNA in cord blood plasma and newborn whole blood samples paired with chronically infected women Samples tested HBsAg+/HBV DNA+ HBsAg+/HBV DNA HBsAg /HBV DNA+ Tested Reactive Tested Reactive Tested Reactive Cord blood plasmas only Newborn whole blood only Cord blood and newborn samples Total (8.8 %) 12 2 (16.7 %) 20 0 Paired cord blood/newborn samples were not available from 15 HBV-infected women. together with the 2075M and 2075C consensus sequences (Fig. 4). The maternal and cord blood quasispecies were not distinguishable and the intra-quasispecies diversity was 0.88 % (range: %). Two clusters of sequences were apparent, representing 71 % (5 and 6 clones from maternal and cord blood samples, respectively) and 29 % (clones 2075M-3, 2075M-9 and 2075C-2) of the clone sequences, respectively. The apparent divergence between the consensus sequences 2075M and 2075C seemed to result from a complex population of variants, present in both maternal and cord blood plasmas, rather than from a minor variant having infected the fetus. On the basis of repeat HBV DNA detection by qpcr and nested-pcr assays, of obtaining at least one sequence from all maternal samples and both BCP/PC (see below) and pre-s/s sequences from 14/17 pairs and of observing clear clustering between maternal and offspring sequences, and minor sequence differences between most paired sequences, the 17 (8.3 %) cases of transmission out of 204 women with detectable HBsAg or/and viral DNA were considered well established, and none appeared to be the result of external or maternal blood contamination. When stratified according to the type of sample tested, the apparent frequency of transmission from cord blood was 9.5 % (13/137 samples), while it was 4.7 % from newborn samples (5/107) (P50.23). Characterization of viral and host factors of transmitting women The main factors associated with HBV vertical transmission previously described were HBeAg and viral load. The Table 3. HBV virological and serological parameters in paired pregnant women, cord blood, and newborn samples with maternofetal transmission Sample ID Pregnant women Cord blood Newborn blood Anti-HBc IgM HBsAg Anti-HBs IgG HBeAg Anti-Hbe IgG Viral load (IU ml 1 ) Anti-HBc IgM HBsAg HBeAg Viral load (IU ml 1 ) Viral load (IU ml 1 ) 1310 Neg. Pos. Neg. Pos. Neg NA NA NA NA Pos. Pos. Neg. Pos. Neg Neg. Neg. Pos. 10 NA 2222 Neg. Pos. Neg. Pos. Neg Neg. Pos. Pos NA 1660 Neg. Pos. Neg. Pos. Neg Neg. Pos. Pos NA 1884 Neg. Pos. Neg. Pos. Neg Neg. Neg. Pos NA 1928 Neg. Pos. Neg. Pos. Neg Neg. Neg. Pos. 10 NA 1648 Neg. Pos. Neg. Pos. Neg Neg. Pos. Pos NA 1773 Neg. Pos. Neg. Pos. Neg Neg. Neg. Pos NA 1449 Neg. Pos. Neg. Pos. Neg Neg. Pos. Pos Neg. Pos. Neg. Pos. Neg Neg. Pos. Pos NA 2203 Neg. Pos. Neg. Pos. Neg Neg. Pos. Pos NA 1975 Pos. Pos. Neg. Neg. Pos Neg. Pos. Neg. 10 NA 2881 Neg. Pos. Neg. Neg. Pos Neg. Neg. Neg.,10 NA 623 Neg. Pos. Neg. Neg. Pos NA NA NA NA Neg. Pos. Neg. Neg. Pos.,10 Neg. Neg. Neg NA 1350 Neg. Pos. Neg. Neg. Neg.,10* NA NA NA NA Neg. Pos. Neg. Neg. Pos.,10* NA NA NA NA *Samples tested HBV DNA-positive with nested-pcr assays after nucleic acid purification from 500 ml of plasma. NA, not available; Pos., positive; Neg., negative

5 D. Candotti, K. Danso and J.-P. Allain Fig. 3. Phylogenetic relationship between the HBV pre-s/s sequences from 14 carrier woman-cord blood/newborn pairs and 2 unpaired women. The 31 Ghanaian sequences were aligned with 4 HBV reference sequences from the GenBank database (genotype A: M57663; genotype E: AB091255, AB091256, and AB106564). Fig. 4. Phylogenetic tree analysis of clones of the HBV pre-s/s gene from a woman cord blood pair (2075M/2075C). Clones from the woman are identified as 2075M-3 to 5 and 2075M-7 to 10. Clones from the cord blood are identified as 2075C-1 to 3 and 2075C-6 to 8 and Sequences obtained by direct sequencing of PCR products are indicated in bold characters. median HBV DNA load in the transmitting women s plasmas was IU ml 21, with values ranging between,10 and IU ml 21. As shown in Fig. 1, a significant correlation was observed between maternal viral load IU ml 21 (high) and detectable HBV DNA in cord blood and/or newborn blood (P ). However, 6 out of 17 transmitting women carried a viral load lower than IU ml 21 (low) and three below 10 IU ml 21. All maternal samples from transmitting women contained IgG anti-hbc and two samples (2769M and 1975M) were reactive for IgM anti-hbc. Anti-HBs was not detected. In contrast, two plasmas from 74 unselected HBsAg-positive, non-transmitting women (viral loads ranging between,10 and IU ml 21 ) were weakly anti-hbs reactive (2.7 %). Eleven (65 %) transmitting women samples were reactive for HBeAg, five (29 %) were anti-hbe reactive, and one (6 %) contained neither HBeAg nor anti-hbe (Table 3). In contrast, in 71 HBsAg-positive/non-transmitting women, the prevalence of HBeAg and anti-hbe was 14 % (10/71) and 86 % (61/71), respectively. HBeAg was significantly associated with high viral load (P,0.0001) in transmitting and non-transmitting groups and as well as with HBV transmission (P,0.0001). In order to examine whether transmission was associated with known mutations affecting viral replication in the maternal strains, the nucleotide sequences of the BCP and PC regions were analysed and compared with sequences obtained from 27 non-transmitting women (8 with high and 19 with low viral load). Within the transmitter group, 2690 Journal of General Virology 88

6 HBV genotype E maternofetal transmission paired mutations 1762 T and 1764 A were associated in samples 1928M (viral load IU ml 21 ) and 2075M (viral load,10 IU ml 21 ) (Table 4). Sequence ambiguities in sample 2075M indicated the simultaneous presence of wild-type and variants, while all cord blood clones carried the 1762 T /1764 A double mutation. At the BCP and pre-core initiation site, no distinctive pattern of mutation was observed. However, paired 1762 T /1764 A mutations and 1764 A/T mutations alone were observed in three high viral load samples ( IU ml 21 ) and four low viral load samples (, IU ml 21 ), respectively. Mutant 1809 T was identified in 10 samples ( Table 4. Distribution of BCP/PC mutations in high and low viral load transmitted and non-transmitted strains Sample Transmission status Viral load (IU ml 1 ) HBeAg/anti-HBe Nucleotide at BCP/PC positions M1310 Transmitted Pos./Neg. A G T C G M2769 Transmitted Pos./Neg. A G G C G M2222 Transmitted Pos./Neg. A G T C G M1660 Transmitted Pos./Neg. A G G C G M1884 Transmitted Pos./Neg. A G G C G M1928 Transmitted Pos./Neg. T A T C G M1648 Transmitted Pos./Neg. A G G C G M1773 Transmitted Pos./Neg. A G G C G M1449 Transmitted Pos./Neg. A G G C G M2186 Transmitted Pos./Neg. A G G C G M2203 Transmitted Pos./Neg. A G T C A M1673 Non-transmitted Pos./Neg. A G G C G M1989 Non-transmitted Pos./Neg. A G G C G M2246 Non-transmitted Pos./Neg. A G T C G M1652 Non-transmitted Pos./Neg. A G T C G M1935 Non-transmitted Pos./Neg. A G G C G M719 Non-transmitted Pos./Neg. T A T C A M886 Non-transmitted NA T A T C G M1178 Non-transmitted Neg./Neg. T A G C A M1975 Transmitted Neg./Pos. A G T C G M2881 Transmitted Neg./Pos. A G G C G M623 Transmitted,10 Neg./Pos. A G G C G M2075 Transmitted,10 Neg./Pos. W R T C G M1350 Transmitted,10 Neg./Neg. A G G C G M1237 Transmitted,10 Neg./Pos. A G G C G M2250 Non-transmitted Neg./Pos. A G G C A M2102 Non-transmitted Neg./Pos. A G G C A M2092 Non-transmitted Neg./Pos. A T A C A M1912 Non-transmitted Pos./Neg. A A G C A M2557 Non-transmitted Neg./Pos. A G G C A M1998 Non-transmitted Neg./Pos. A G T C A M2086 Non-transmitted Neg./Pos. A G G C A M1629 Non-transmitted Pos./Neg. A G G C G M2458 Non-transmitted Neg./Pos. A G T C A M1451 Non-transmitted Neg./Pos. A G G C G M2406 Non-transmitted Neg./Pos. T A G C A M2356 Non-transmitted,10 Neg./Pos. A G G C A M2274 Non-transmitted,10 Neg./Pos. A A T C G M2188 Non-transmitted,10 Neg./Pos. A G T C A M2241 Non-transmitted,10 Neg./Pos. A G G C R M2214 Non-transmitted,10 Neg./Pos. T A G C A M2682 Non-transmitted,10 NA A G G C G M2421 Non-transmitted,10 Neg./Pos. T A T C A M2415 Non-transmitted,10 Neg./Pos. T A T C A NA, not available. W, A+T; R, A+G

7 D. Candotti, K. Danso and J.-P. Allain IU ml 21 ) and mutant 1809 A in one sample ( IU ml 21 ). The PC region was wild-type in all transmitted strains except for an 1896 A stop codon mutation detected in sample 2203M that carried a viral load of IU ml 21.In contrast, a significantly higher prevalence of 1896 A variant was detected in the non-transmitting group (63 % [17/27] vs 6% [1/17]; P ) (Table 4). Among the low viral load groups, none of the six transmitted strains from HBeAgnegative/anti-HBe-positive women and 15/19 (79 %) of the non-transmitted strains had the 1896 A stop codon mutation (P ). In summary, the relative risk of maternofetal transmission associated with high viral load ( 10 4 IU ml 21 ) compared with low viral load in pregnant women was 2.41 (95 % confidence interval: ; P50.048). A relative risk of transmission of 11.5 (95 % confidence interval: ; P ) was associated with the presence of wild-type precore sequences at position 1896, irrespective of viral load and HBeAg/anti-HBe status. Since no transmission was observed from 15 women infected with 1896 mutant virus with viral load,10 4 IU ml 21, the relative risk of transmission was very low. The amino acid sequences deduced from pres/s nucleotide sequences were analysed and the mean divergence rates within the pre-s1, pre-s2 and S regions were 1.5 % (range: %), 3.4 % (range: %) and 0.85 % (range: %) in the transmitting group (n516), compared with 3.1 % (range: %), 5.3 % (range: %) and 1.8 % (range: %) in the non-transmitting group (n517), respectively. The intra-group amino acid divergence observed in transmitted strains was significantly lower than that in non-transmitted strains over the whole S gene (P,0.0001). In addition, one non-transmitting sample ( IU ml 21 ) showed a deletion including the major part of the PreS1 region, and three others (viral load , and IU ml 21 ) had 4 7 amino acid deletions in the PreS2. DISCUSSION In the present study, the 16 % prevalence of HBV chronic carriers characterized by the presence of HBsAg and/or HBV DNA among 1368 unselected pregnant women from Kumasi, Ghana (Fig. 1, Table 1), the bimodal distribution of viral load, the predominance of genotype E and the high frequency of occult HBV were consistent with those previously reported among the mostly male blood-donor population from the same area (male-to-female ratio.3) (Allain et al., 2003; Candotti et al., 2006b; Owusu-Ofori et al., 2005). Equally consistent with previously reported data was the presence of rare cases (0.9 %) of HBsAg with undetectable levels of HBV DNA (Kuhns et al., 2004). Overall, the characteristics of chronic HBV infection in Ghanaian pregnant women or blood donor (80 % males) adult populations were similar, irrespective of age or gender. The predominant route of HBV transmission in sub- Saharan Africa has been reported to be horizontal, and most children are already infected by age 5 (Dumpis et al., 2001; Kew, 1996; Martinson et al., 1998). This is consistent with the repeated observation that,20 % of HBV-infected young adults in Ghana are carrying a high viral load (this study and Allain et al., 2003). However, the part of maternofetal transmission in the overall epidemiology of HBV genotype E has not been evaluated. This study provided evidence of HBV DNA in cord blood or newborn samples or both (Table 2, Fig. 2), strongly suggesting maternofetal HBV transmission. The low nucleotide sequence divergence observed in 10 out of 12 pregnant women/cord blood pairs excluded cross-contamination from maternal blood (Fig. 3). The likelihood of crosscontamination was even lower in pregnant women with low viral load, reinforcing the fact that they transmitted HBV. Identical sequences, however, were observed in 1648M/1648C and 1660M/1660C pairs in association with high maternal viral load (.10 7 IU ml 21 ), but HBsAg, which does not usually cross the placental barrier (Roingeard et al., 1993; Vranckx et al., 1999), was found, while it was undetectable in cord blood or newborn samples from non-transmitting women (Table 3). In samples 2075M and 2075C, the apparent discrepancy in the consensus sequences between maternal and fetal samples was subsequently explained by the heterogeneity of the maternal quasispecies, which was reflected in the cord blood when clones were sequenced (Fig. 4). The overall 8.3 % (17/204) HBV maternofetal transmission rate observed in Ghana was similar to the 7 % rate reported from Senegal in infants at birth, but was higher than previously reported in other studies from sub-saharan Africa investigating HBV infection in 6 9-months-old infants (Menendez et al., 1999; Roingeard et al., 1993). The Senegalese study also showed that when HBsAgpositive newborns were followed up at 6 7 months, approximately half of them became HBsAg-negative and did not develop antibodies against HBV antigens, reducing the estimated rate of transmission from 7 to 3.2 % (Roingeard et al., 1993). In the present study, the HBV DNA prevalence in cord blood at delivery (8.8 %) was higher than in 2 week-old newborns (4.7 %), but not significantly. Several potential explanations can be offered to account for this discrepancy: (i) the sensitivity of HBV DNA detection in whole blood might be lower than in plasma, affecting detection in samples generally containing low viral load; (ii) a proportion of HBV DNA detected in cord blood might be non-infectious and rapidly cleared after birth and (iii) although we have not found evidence that contamination from maternal blood occurred at sampling, it is certainly more likely to occur in cord blood than in newborn samples. According to the data collected in this study, the findings that (i) the distribution of high/low viral load in pregnant women is 12/88 %, (ii) the vertical transmission rate is 3 % with low maternal viral load and 46 % with high viral load 2692 Journal of General Virology 88

8 HBV genotype E maternofetal transmission (Fig. 1 and see below), (iii) all cases of maternofetal transmission lead to chronic infection and (iv) the prevalence of HBsAg by age is 15 %, indicate that the proportion of chronic HBV infections related to maternofetal transmission would be approximately 8 % or accounting for 1.2 % of the overall 15 % prevalence of HBsAg. This is consistent with the estimation derived from this study and the previously reported horizontal transmission as preferred route of HBV infection in Africa (Dumpis et al., 2001; Kew, 1996; Martinson et al., 1998). Multiple risk factors potentially influencing HBV genotype E maternofetal transmission have been identified in chronically infected pregnant women: viral load, HBV serological profile, duration of chronic infection, genotype and mutations in the BCP/PC or S regions (Cacciola et al., 2002; Xu et al., 2006). Three groups of HBV carriers were initially identified among the Ghanaian pregnant women according to their HBsAg/HBV DNA status (Table 2): HBsAg-positive/viral DNA 10 IU ml 21, HBsAg-positive/ viral DNA,10 IU ml 21, and HBsAg-negative/viral DNA 10 IU ml 21. HBV transmission rates of 8.7 % (15/172) and 16.7 % (2/12) were observed in the first and second group, respectively, while no evidence of transmission was found within the third group. The lack of infection in the third group might be in part related to the presence of lowlevel anti-hbs in 37 % of these women (data not shown). When HBV-carrier pregnant women were stratified according to high or low plasma viral load, transmission rates of 45.8 % (11/24) and 3.3 % (6/180) were found, respectively (Fig. 2). These results confirmed that the risk of maternofetal transmission of HBV is positively related to maternal viral load, as suggested by others in different epidemiological settings, but also clearly indicate that transmission can occur in pregnant women carrying a low or very low viral load (Burk et al., 1994; Greenfield et al., 1986; Ip et al., 1989; Wang et al., 2003; Xu et al., 2002; Zhang et al., 1998). HBe status and genotype are two additional factors associated with viral load and frequency of vertical transmission. Despite similar high prevalence of HBV chronic carriers (Maddrey, 2000), the rate of maternofetal infection in east Asia, particularly China, was estimated to range between 10 and 88 % (Ip et al., 1989; Zhang et al., 1998, 2004), compared with 8 % or less in the present and other studies conducted in sub-saharan Africa (Kew et al., 1987; Menendez et al., 1999; Roingeard et al., 1993). This difference can be largely attributed to the natural history of HBV infection of genotype B and C, as in south-east Asia where infected individuals carry HBeAg and high viral load in age groups that include most women of gestational age (Chu et al., 2002; Kao et al., 2002). In contrast, in sub- Saharan Africa, whether infected with HBV genotype A1 or E, seroconversion to anti-hbe occurs before age 15 or 16, with the consequence that most women of gestational age carry anti-hbe (Candotti et al., 2006b). Using highly sensitive molecular assays, HBV transmission was detected in three cord blood and three newborn samples from six women with low HBV DNA load, five of them,20 IU ml 21 (Table 3). It is possible that certain viral genetic features might influence the ability of genotype E HBV to be vertically transmitted. While the presence of the e antigen and of strains with wild-type nucleotide at position 1896 in transmitting women with high viral load was expected, the fact that all strains from transmitting low-viral-load women were wild-type at codon 28 despite the presence of anti-hbe was intriguing (Table 4). This feature is unlikely to affect infectivity of the strain per se, but permits the production of HBeAg, a protein related to HBV tolerance, that might play a role either in the migration of the virus through the placenta or in establishing infection in the fetus (Chen et al., 2005). Should the capacity to produce the e antigen play a significant role in establishing HBV infection, low viral load strains wild-type at the 1896 nucleotide might be given an infectivity advantage that constitutes a significant risk factor. This feature is reinforced by the comparison of genetic differences between sequences of transmitting and non-transmitting women infected with HBV genotype E irrespective of viral load. The frequency of amino acid substitutions across pres1, pres2 and S regions of viruses associated with transmission was significantly lower than that of viruses from the non-transmitting group. In addition, it is to a large extent the absence of a mutation at nucleotide 1896 in strains from transmitting women with low viral load that made significant the difference between transmitted and non-transmitted strains at this location (P ) (Table 4). As a single risk factor, wildtype at nucleotide 1896 was considerably more predictive of transmission (relative risk: 11.5) than viral load (relative risk: 2.4). One proposed route of maternofetal HBV transmission is transplacental transfusion or leakage of maternal blood into the fetal circulation (Lin et al., 1987; Ohto et al., 1987; Xu et al., 2002). Mother-to-fetus transfer of nucleated cells and DNA may occurs during the second trimester, but is mostly seen during the third trimester of gestation, and is precipitated by trauma associated with labour, particularly in threatened preterm labour (Ohto et al., 1987; Petit et al., 1997; Xu et al., 2002). Maternal nucleated cells and DNA have been detected in the fetal circulation at a fractional concentration ranging between and of nucleated fetal blood cells and cord plasma DNA (Lo et al., 2000; Petit et al., 1997). Variable amounts of maternal blood, containing low levels of infectious virus, might be sufficient to determine fetal infection. Reported cases of transfusiontransmitted HBV infections have been examined as an analogous situation. Occult HBV infection, characterized by a viral load below 500 IU ml 21 viral DNA in anti-hbcpositive donations, were shown to be infectious in approximately 50 % of the few cases recently reported (Gerlich, 2006; Inaba et al., 2005), whereas it was found to be non-infectious below 100 IU ml 21 (Dow et al., 2001). Maternofetal infection occurring close to delivery might also explain the low level of HBV DNA

9 D. Candotti, K. Danso and J.-P. Allain consistently detected in cord blood and newborn whole blood samples, and the lack of detectable anti-hbc IgM in the cord blood. This study has provided evidence that maternofetal transmission of HBV is relatively rare in a genotype E prevalent area of west Africa. Although maternal viral load remains the main risk factor for transmission, pregnant women with low viral load (, IU ml 21 ), even below 20 IU ml 21, might transmit, particularly if the strains remain wild-type in the BCP/PC as well as the pre-s/s regions. Irrespective of viral load, the absence of the position 1896 stop codon was the highest predictor of transmission, suggesting a role of HBe antigen in the maternofetal route of HBV genotype E infection. ACKNOWLEDGEMENTS The authors are indebted to the staff of the Department of Obstetrics and Gynaecology, Komfo Anokye Teaching Hospital, Kumasi, Ghana, for collecting the samples used in this study, as well as to Mr Albert Dompreh who processed the samples and Mr Francis Sarkodie who performed some of the initial HBsAg screening. This work was supported in part by grant B503/2 from the National Blood Service, England. REFERENCES Allain, J.-P., Candotti, D., Soldan, K., Sarkodie, F., Phelps, B., Giachetti, C., Shyamala, V., Yeboah, F., Anokwa, M. & other authors (2003). The risk of hepatitis B virus infection by transfusion in Kumasi, Ghana. Blood 101, Burk, R. D., Hwang, L. Y., Ho, G. Y., Shafritz, D. A. & Beasley, R. P. (1994). Outcome of perinatal hepatitis B virus exposure is dependent on maternal virus load. J Infect Dis 170, Cacciola, I., Cerenzia, G., Pollicino, T., Squadrito, G., Castellaneta, S., Zanetti, A. R., Mieli-Vergani, G. & Raimondo, G. (2002). Genomic heterogeneity of hepatitis B virus (HBV) and outcome of perinatal HBV infection. J Hepatol 36, Candotti, D., Danso, K., Parsyan, A., Dompreh, A. & Allain, J.-P. (2006a). Maternal-fetal transmission of human parvovirus B19 genotype 3. J Infect Dis 194, Candotti, D., Opare-Sem, O., Rezvan, H., Sarkodie, F. & Allain, J.-P. (2006b). Molecular and serological characterization of hepatitis B virus in deferred Ghanaian blood donors with and without elevated alanine aminotransferase. J Viral Hepat 13, Chen, M., Sällberg, M., Hughes, J., Jones, J., Guidotti, L. G., Chisari, F. V., Billaud, J.-N. & Milich, D. R. (2005). Immune tolerance split between hepatitis B virus precore and core proteins. JVirol79, Chu, C. J., Hussain, M. & Lok, A. S. F. (2002). Hepatitis B virus genotype B is associated with earlier HBeAg seroconversion compared with hepatitis B virus genotype C. Gastroenterology 122, Dow, B. C., Peterkin, M. A., Green, R. H. A. & Cameron, S. O. (2001). Hepatitis B virus transmission by blood donation negative for hepatitis B surface antigen, antibody to HBsAg, antibody to hepatitis B core antigen and HBV DNA. Vox Sang 81, 140. Dumpis, U., Holmes, E. C., Mendy, M., Hill, A., Thursz, M., Hall, A., Whittle, H. & Karayiannis, P. (2001). Transmission of hepatitis B virus infection in Gambian families revealed by phylogenetic analysis. J Hepatol 35, Gerlich, W. H. (2006). Breakthrough of hepatitis B virus escape mutants after vaccination and virus reactivation. J Clin Virol 36 (Suppl. 1), S18 S22. Ghendon, Y. (1987). Perinatal transmission of hepatitis B virus in high-incidence countries. J Virol Methods 17, Greenfield, C., Osidiana, V., Karayiannis, P., Galpin, S., Musoke, R., Jowett, T. P., Mati, P., Tukei, P. M. & Thomas, H. C. (1986). Perinatal transmission of hepatitis B virus in Kenya: its relation to the presence of serum HBV-DNA and anti-hbe in the mother. J Med Virol 19, Hyams, K. C. (1995). Risks of chronicity following acute hepatitis B virus infection: a review. Clin Infect Dis 20, Inaba, S., Miyata, Y., Ishii, H., Kajimoto, S., Yugi, H., Hino, M. & Tadokoro, K. (2005). Hepatitis B transmission from an occult HB virus carrier (abstract SP217). Transfusion 45 (Suppl. S3), 94A. Ip, H. M. H., Lelie, P. N., Wong, V. C. W., Kuhns, M. C. & Reesink, H. W. (1989). Prevention of hepatitis B virus carrier state in infants according to maternal serum levels of HBV DNA. Lancet 1, Kao, J. H., Chen, P. J., Lai, M. Y. & Chen, D. S. (2002). Clinical and virological aspects of blood donors infected with hepatitis B virus genotypes B and C. J Clin Microbiol 40, Kew, M. C. (1996). Progress towards the comprehensive control of hepatitis B virus in Africa: a view from South Africa. Gut 38 (Suppl. 2), S31 S36. Kew, M. C., Kassianides, C., Berger, E. L., Song, E. & Dusheiko, G. M. (1987). Prevalence of chronic hepatitis B virus infection in pregnant black women living in Soweto. J Med Virol 22, Kuhns, M. C., Kleinman, S. H., McNamara, A. L., Rawal, B., Glynn, S. & Busch, M. P. The REDS Study Group (2004). Lack of correlation between HBsAg and HBV DNA levels in blood donors who test positive for HBsAg and anti-hbc: implications for future HBV screening policy. Transfusion 44, Lin, H.-H., Lee, T.-Y., Chen, D.-S., Sung, J.-L., Ohto, H., Etoh, T., Kawana, T. & Mizuno, M. (1987). Transplacental leakage of HBeAgpositive maternal blood as the most likely route in causing intrauterine infection with hepatitis B virus. J Pediatr 111, Lo, Y. M. D., Lau, T. K., Chan, L. Y. S., Leung, T. N. & Chang, A. M. Z. (2000). Quantitative analysis of the bidirectional fetomaternal transfer of nucleated cells and plasma DNA. Clin Chem 46, Maddrey, W. C. (2000). Hepatitis B: an important public health issue. J Med Virol 61, Martinson, F. E., Weigle, K. A., Royce, R. A., Weber, D. J., Suchindran, C. M. & Lemon, S. M. (1998). Risk factors for horizontal transmission of hepatitis B virus in a rural district in Ghana. Am J Epidemiol 147, Menendez, C., Sanchez-Tapias, J. M., Kahigwa, E., Mshinda, H., Costa, J., Vidal, J., Acosta, C., Lopez-Labrador, X., Olmedo, E. & other authors (1999). Prevalence and mother-to-infant transmission of hepatitis viruses B, C, and E in Southern Tanzania. J Med Virol 58, Ohto, H., Lin, H. H., Kawana, T., Etoh, T. & Tohyama, H. (1987). Intrauterine transmission of hepatitis B virus is closely related to placental leakage. J Med Virol 21, 1 6. Owusu-Ofori, S., Temple, J., Sarkodie, F., Anokwa, M., Candotti, D. & Allain, J.-P. (2005). Predonation screening of blood donors with rapid tests: implementation and efficacy of a novel approach to blood safety in resource-poor settings. Transfusion 45, Petit, T., Dommergues, M., Socié, G., Dumez, Y., Gluckman, E. & Brison, O. (1997). Detection of maternal cells in human fetal blood during the third trimester of pregnancy using allele-specific PCR amplification. Br J Haematol 98, Journal of General Virology 88

10 HBV genotype E maternofetal transmission Roingeard, P., Diouf, A., Sankale, J. L., Boyce, C., Mboup, S., Diadhiou, F. & Essex, M. (1993). Perinatal transmission of hepatitis B virus in Senegal, West Africa. Viral Immunol 6, Shepard, C. W., Simard, E. P., Finelli, L., Fiore, A. E. & Bell, B. P. (2006). Hepatitis B virus infection: epidemiology and vaccination. Epidemiol Rev 28, Vranckx, R., Alisjahbana, A. & Meheus, A. (1999). Hepatitis B virus vaccination and antenatal transmission of HBV markers to neonates. J Viral Hepat 6, Wang, Z., Zhang, J., Yang, H., Li, X., Wen, S., Guo, Y., Sun, J. & Hou, J. (2003). Quantitative analysis of HBV DNA level and HBeAg titer in hepatitis B surface antigen positive mothers and their babies: HBeAg passage through the placenta and the rate of decay in babies. J Med Virol 71, Xu, D.-Z., Yan, Y.-P., Choi, B. C. K., Xu, J.-Q., Men, K., Zhang, J.-X., Liu, Z.-H. & Wang, F.-S. (2002). Risk factors and mechanism of transplacental transmission of hepatitis B virus: a case-control study. J Med Virol 67, Xu, H., Peng, M., Qing, Y., Ling, N., Lan, Y., Liang, Z., Cai, D., Li, Y. & Ren, H. (2006). A quasi species of the pre-s/s gene and mutations of enhancer II/core promoter/pre-c in mothers and their children infected with hepatitis B virus via mother-to-infant transmission. J Infect Dis 193, Zhang, S.-L., Han, X.-B. & Yue, Y.-F. (1998). Relationship between HBV viremia level of pregnant women and intrauterine infection: nested PCR for detection of HBV DNA. World J Gastroenterol 4, Zhang, S.-L., Yue, Y.-F., Bai, G.-Q., Shi, L. & Jiang, H. (2004). Mechanism of intrauterine infection of hepatitis B virus. World J Gastroenterol 10,