Hepatitis B Virus Infection in Human Immunodeficiency Virus Infected Southern African Adults: Occult or Overt – That Is the Question

Hepatitis B virus (HBV) and human immunodeficiency virus (HIV) share transmission routes and are endemic in sub-Saharan Africa. The objective of the present study was to use the Taormina definition of occult HBV infection, together with stringent amplification conditions, to determine the prevalence and characteristics of HBV infection in antiretroviral treatment (ART)-naïve HIV+ve adults in a rural cohort in South Africa. The presence of HBV serological markers was determined by enzyme linked immunoassay (ELISA) tests. HBV DNA-positivity was determined by polymerase chain reaction (PCR) of at least two of three different regions of the HBV genome. HBV viral loads were determined by real-time PCR. Liver fibrosis was determined using the aspartate aminotransferase-to-platelet ratio index. Of the 298 participants, 231 (77.5%) showed at least one HBV marker, with 53.7% HBV DNA−ve (resolved) and 23.8% HBV DNA+ve (current) [8.7% HBsAg+ve: 15.1% HBsAg−ve]. Only the total number of sexual partners distinguished HBV DNA+ve and HBV DNA−ve participants, implicating sexual transmission of HBV and/or HIV. It is plausible that sexual transmission of HBV and/or HIV may result in a new HBV infection, superinfection and re-activation as a consequence of immunesuppression. Three HBsAg−ve HBV DNA+ve participants had HBV viral loads <200 IU/ml and were therefore true occult HBV infections. The majority of HBsAg−ve HBV DNA+ve participants did not differ from HBsAg+ve HBV DNA+ve (overt) participants in terms of HBV viral loads, ALT levels or frequency of liver fibrosis. Close to a quarter of HIV+ve participants were HBV DNA+ve, of which the majority were HBsAg−ve and were only detected using nucleic acid testing. Detection of HBsAg−ve HBV DNA+ve subjects is advisable considering they were clinically indistinguishable from HBsAg+ve HBV DNA+ve individuals and should not be overlooked, especially if lamivudine is included in the ART.


Introduction
Hepatitis B virus (HBV) and human immunodeficiency virus (HIV) share transmission routes and represent the two most important blood-borne pathogens in terms of prevalence, morbidity and mortality in sub-Saharan Africa, where both viruses are endemic. Of the 33.3 million adults and children living with HIV globally, 22.5 million reside in sub-Saharan Africa [1]. Moreover, it is estimated that 65% to 98% of populations in sub-Saharan Africa have been exposed to HBV and 8% to 20% are chronic carriers of HBV [2], far exceeding the 4% to 6% lifetime exposure rates and 0.2% to 0.5% carrier rates in regions of low endemicity. Thus, widespread co-infections are likely to occur, with 16% to 98% of HIV +ve individuals in sub-Saharan Africa being carriers of HBV or showing exposure to HBV [3].
The progression of chronic HBV to cirrhosis, end-stage liver disease (ESLD), and hepatocellular carcinoma (HCC) is more rapid in HIV +ve individuals than those with HBV alone [4], with a significant increase in hepatic-related mortality rates [5]. Furthermore, HBV co-infection negatively impacts on HIV outcomes [6].
Before the introduction of antiretroviral therapy (ART), the majority of HBV/HIV co-infected individuals were more likely to die from the clinical consequences of HIV than those of HBV [3]. However, since the introduction of ART, the disease profile has changed, with increases in the proportion of mortality attributed to HBV-associated ESLD [7]. Thus, HBV/HIV co-infection can potentially impact on the safety and effectiveness of ART, requiring an integrated approach for the appropriate management of co-infected individuals [8].
There is a paucity of comprehensive and standardized data describing HBV/HIV co-infection from southern African countries, where HIV prevalence is extremely high. Existing data show large discrepancies, with exposure rate to HBV in HIV +ve South Africans varying from 28% to 99.8% and HBsAg prevalence ranging from 0.4% to 23% [9][10][11][12][13][14][15][16][17][18]. Differences can be attributed to different locations, study designs, laboratory measures and/or the composition of the study populations.
HIV infection has been implicated as a risk factor for the development of occult HBV infection (OBI) [12], defined by the Taormina expert panel as the ''Presence of HBV DNA in liver (with detectable or undetectable HBV DNA in the serum) of individuals testing HBsAg negative by currently available assays. When detectable, the amount of HBV DNA in the serum is usually very low (,200 IU/ml)'' [19]. Because liver biopsies are not commonly available, especially in resource-limited environments, OBI is usually detected by the analysis of sera [19]. Furthermore, the experts differentiate between true occult (HBV viral load ,200 IU ml 21 ) and false occult where HBV DNA levels are comparable to those detected in HBsAg +ve infection (overt) and are usually as a result of infection by HBV variants with S gene escape mutants, producing HBsAg that is not recognized by detection assays [19]. The clinical implications of OBI are unclear.
The prevalence of OBI in HIV infected individuals varies depending on the definition used, the sensitivity of the assay and the HBV viral loads [11][12][13]16,17]. Furthermore, studies performed outside Africa, in areas of low HBV and HIV endemicity, cannot necessarily be extrapolated to Africa because of differences in host factors, epidemiology, transmission patterns and genotypes of the viruses between the two regions.
The objective of the present study was to use the Taormina definition of OBI [19], together with stringent amplification conditions, to determine the prevalence and characteristics of HBV infection in ART-naïve HIV +ve adults entering a rural cohort in Mpumalanga Province, which has a HIV prevalence of 15.4% [20]. No in-depth studies have been undertaken to determine the prevalence and characteristics of HBV/HIV coinfection in this province.

Subjects
A new rural cohort was established at Shongwe Hospital in Mpumalanga Province in South Africa and 298 ART-naïve, HIV +ve adults were enrolled from July to November 2009. All had qualified for ART according to the then-current South African ART guidelines (CD4 counts ,200 cells mm 23 ) [21] and were recruited while undergoing treatment-readiness counselling. Universal HBV vaccination at 6, 10, and 14 weeks of age was introduced into the South African Expanded Programme on Immunization (EPI) in 1995 and therefore none of the participants were likely to have received this vaccination and self-reported as unvaccinated. Clinical and demographic data (including ALT levels, CD4 T-cell count, age, sex, height and weight) were obtained from hospital records, the National Health Laboratory Services (NHLS) databases and the TherapyEdge-HIV (TE) TM electronic patient record. All participants signed informed consent. The study was approved by the Human Research Ethics Committee (Medical) of the University of the Witwatersrand and Mpumalanga Department of Health Research Ethics Committee.

Serology
The presence of HBsAg, anti-HBsAg and anti-HBcAg was determined for 298 sera using the Monolisa TM HBsAg ULTRA, HBsAb ULTRA and HBcAb PLUS ELISA kits (Bio-Rad, Hercules, CA), respectively. HBeAg and anti-HBe tests were performed on HBV DNA +ve sera using the Monolisa TM HBeAg-Ab PLUS kit. Anti-HBcAg IgM was determined for 17 anti-HBc +ve HBV DNA +ve samples for which serum was available using the ARCHITECTH kit (Abbott Diagnostics, Wiesbaden, Germany). The M30-ApoptosenseH ELISA (Peviva AB, Stockholm, Sweden) was used on all sera to quantify the apoptosisassociated cytokeratin 18Asp396 neo-epitope as a measure of hepatocyte apoptosis [22].

Measurement of liver fibrosis
The aspartate aminotransferase (AST)-to-platelet ratio index (APRI) = (AST[/ULN]*100)/platelet count [10 9 L 21 ], a noninvasive measure of liver fibrosis in patients with chronic HBV [23], was calculated for 163 subjects for whom AST levels and platelet counts were available. APRI indicates liver fibrosis only when liver disease has reached a severely advanced stage, with significant fibrosis defined as APRI$1.5, and no fibrosis as APRI#0.5 [24].

Polymerase chain reaction (PCR)
DNA was extracted from 200 ml blood plasma with the QIAamp DNA Blood Mini Kit (QIAGEN Gmbh, Hilden, Germany) and eluted into 75 ml of best-quality water (BQW). Known positive and negative sera and BQW were used as controls for the extraction. Three regions of the HBV genome were amplified in a MyCycler TM thermocycler (Bio-Rad, Hercules, Ca, USA) using Promega Taq DNA polymerase (Promega, Madison, WI) ( Table 1). To avoid cross-contamination and false positives, the precautions and procedures of Kwok and Higuchi [25] were strictly adhered to. DNA extraction, PCR, and electrophoresis were performed in physically separated venues.

Real-time PCR quantification of HBV DNA
PCR primers, HBV-Taq1 and HBV-Taq2 covering a region of the S gene (321 to 401 from the EcoRI site) with a FAM/TAMRA labelled TaqMan BS-1 probe [26] were used to quantify HBV DNA in an ABI 7500 Real Time PCR System (Applied Biosystems, Foster City, Ca, USA). A serial dilution of cloned plasmid DNA containing a single genome of HBV DNA, with concentrations ranging from 2610 1 to 2610 11 IU ml 21 , was used as template to generate the standard curve. The second WHO International Standard for HBV Nucleic Acid Amplification Techniques (product code 97/750 National Institute for Biological Standards and Controls (NIBSC); Hertfordshire, UK), which has a final concentration of 10 6 IU ml 21 was used as the internal standard. The standard curve, blank, positive and negative controls, and samples were all tested in duplicate. The measured IU/ml for each reaction was calculated using the Ct (cycle threshold) value of each PCR interpolated against the linear regression of the standard curve. The lower detection limit of our assay is ,20 IU ml 21 . The conversion formula of IU = copies/4.7 was used [11,27].

Statistical analysis
Clinical data were inspected visually. As all continuous variables showed a skewed distribution, the Mann-Whitney U test (Wilcoxon rank-sum test) was used to compare samples. Chisquared and Fisher's exact test were used to compare categorical variables. Exhaustive multivariate logistic regression analyses were performed. The R statistical language was used throughout [28].

Serological and nucleic acid testing for HBV
The study group consisted of 298 adults (114 men and 184 women) with median age, CD4 count and BMI of 34 years, 147 cells mm 23 and 22 kg m 22 , respectively. Men were older than women and had lower CD4 counts ( Table 2).
Screening for HBV DNA was carried out using primers targeting three non-overlapping regions of the HBV genome (Table 1) Of the entire group of 298, 26 (8.7%)/28 (9.4%) HBsAg +ve participants were HBV DNA +ve and together with the 45 (15.1%) HBsAg 2ve HBV DNA +ve participants were classified into 6 serogroups ( Figure 2). Within the HBsAg 2ve groups, the frequency of HBV DNA was significantly higher in anti-HBc +ve alone individuals (16/57; 28.1%) compared to those anti-HBc +ve anti-HBs +ve (17/123; 13.8%) (p,0.05). The relative risk of an HBsAg 2ve individual, who was anti-HBc +ve alone, being HBV DNA +ve was twice as high as that of one with anti-HBc +ve anti-HBs +ve . The frequency of HBV DNA in the serologically 2ve group was not significantly different to that in the anti-HBc +ve alone or anti-HBc +ve anti-HBs +ve . Moreover, HBV DNA was not detected in any of the 11 isolated anti-HBs +ve individuals. Sufficient serum was available to test for anti-HBc IgM in 17 of 57 anti-HBc +ve HBV DNA +ve participants and all tested negative. Only three HBsAg 2ve HBV DNA +ve participants had viral loads ,200 IU ml 21 , thus meeting the Taormina criterion for true OBI [19]. These participants had serological patterns of groups A, D and E, respectively ( Figure 2). All other HBsAg 2ve HBV DNA +ve individuals had HBV viral loads .200 IU ml 21 .
Comparison of demographic and clinical characteristics between HBV DNA +ve and HBV DNA 2ve groups Visual inspection of plots and linear regression models of each of the continuous variables in Tables 2 and 3 (age, age at sexual debut, lifetime sexual partners, BMI (body mass index), ALT, Apoptosense, CD4 cell count, HBV viral load) against each other, for HBV DNA +ve versus HBV DNA 2ve , and HBsAg +ve versus HBsAg 2ve groups, revealed no significant correlation.
A multiple logistic regression model was used to determine predictors of HBV DNA positivity. In this model, only ALT levels were significant when all variables were included (p,0.05; OR = 1.01; 95% CI: 1.002-1.020). When the data were split according to gender, number of lifetime sexual partners was the only predictor in the females (p,0.05; OR = 1.16; 95% CI: 1.01-1.36) and ALT in the males (p,0.05; OR = 1.02; 95% CI: 1.004-1.030).
As shown in Table 2, the only variable that differentiated the HBV DNA +ve and HBV DNA 2ve groups was number of lifetime sexual partners (p,0.05). Regardless of whether they were HBV DNA +ve or HBV DNA 2ve , males were older than females, had a higher ALT and lower CD4 count. In the whole cohort and the HBV DNA 2ve group, females had a higher BMI (p,0.05) and fewer sexual partners than males (p,0.05). These differences were not seen in the HBV DNA +ve group. The age of sexual debut was Table 1. PCR primers and cycling parameters used for amplification of the three regions of the HBV genome. significantly different only when comparing males and females in the whole cohort.
Comparison of demographic and clinical characteristics between HBsAg +ve HBV DNA +ve and HBsAg 2ve HBV DNA +ve groups Data from the HBV DNA +ve participants were examined by logistic regression for predictors of HBV DNA-positivity in the absence of HBsAg. Only increasing age was weakly significant. The female subset showed that age was a significant predictor (p,0.05; OR = 1.28; 95% CI: 1.06-1.72). No predictors in the male subset were significant.
In the HBsAg 2ve HBV DNA +ve group, men were older and had significantly lower CD4 cell counts compared to females (p,0.05). Although the difference in ALT levels between the HBsAg +ve HBV DNA +ve and HBsAg 2ve HBV DNA +ve groups did not reach statistical significance (Table 3), individuals who were HBsAg +ve anti-HBc +ve HBV DNA +ve [group C] had significantly higher ALT levels compared to individuals who were either serologically 2ve HBV DNA +ve [group A] (p,0.05) or anti-HBc +ve HBV DNA +ve [group E] (p,0.05) (Figure 2). There was no significant difference between the HBsAg +ve and HBsAg 2ve DNA +ve groups when ALT levels were coded into binary groups: .29 U/L for males and .19 U/L for females. HBV viral loads did not differ significantly between HBsAg +ve and HBsAg 2ve groups ( Table 3).

Discussion
In this group of 298 southern African ART-naïve HIV +ve individuals, 231 participants had at least one HBV marker, giving an overall exposure to HBV of 77.5%, comparable to that in HBV monoinfected individuals [2]. In addition, almost one quarter of the group was HBV DNA +ve (Figure 1) of whom almost two thirds were HBsAg 2ve . Direct comparison with other South African ART-naïve HIV +ve cohorts is difficult because of the different markers were used to measure exposure. In Limpopo Province, exposure to HBV, measured by anti-HBc and/or anti-HBs positivity, was 28.2% in a rural cohort [17] and 39.2% in antinatal HIV +ve women [16]. This differs from the 63% HBV exposure rate (measured by at least one marker: HBsAg, anti-HBs or anti-HBc) found in a rural-urban HIV +ve cohort in Limpopo [13] and the much higher exposure rate of 99.8% in hospitaladmitted HIV +ve patients [12]. In Gauteng Province, a 47% exposure was seen in an urban HIV +ve cohort where ,15% were HBV-positive as follows: 4.8% HBsAg +ve [10], 7.6% anti-HBc +ve HBV DNA +ve [11] and 2.4% serologically 2ve HBV DNA +ve [27].
The HBV serology of the cohort, the high frequency of HBeAgnegativity, the absence of anti-HBc IgM and the relatively low HBV viral loads (Figure 2) reflect the natural history of HBV infection in sub-Saharan Africa, where most individuals are infected at childhood by horizontal transmission [2]. This means that most individuals have been exposed to HBV, and are protected by anti-HBV antibodies, by the time they become sexually active and acquire HIV. All isolated anti-HBs +ve participants were HBV DNA 2ve and the presence of anti-HBs with anti-HBc reduced the risk of being HBV DNA +ve . None of the participants had received HBV vaccination.
The HBsAg prevalence in this HIV +ve cohort was not different to HIV 2ve cohorts [2,3]. This differs from observations in areas of low HBV and HIV endemicity, where HBV and HIV are acquired simultaneously and therefore HBsAg prevalence in HIV +ve individuals is significantly higher than in HIV 2ve individuals [3]. Only four participants in the present study were HBsAg +ve alone: two were HBV DNA +ve , whereas the other two were HBV DNA 2ve , even after repeated attempts to amplify HBV DNA, possibly indicating low viral loads undetectable by PCR. This might reflect the process of natural HBsAg clearance [32]. Although immune suppression by HIV may lead to the HBsAg +ve anti-HBc 2ve profile [33], this is unlikely in these two cases, considering that ,59% of the participants were anti-HBc +ve , with a third of these having isolated anti-HBc. Moreover, HIV +ve patients with CD4,100 cells mm 3 are more likely to have isolated anti-HBc [34].
In agreement with other studies [38,39], there were similar ALT levels in HBsAg +ve and HBsAg 2ve HBV DNA +ve participants and between HBV DNA +ve and HBV DNA 2ve participants. The absence of transaminitis is as a result of the immunosuppressed state of the HIV +ve subjects. Immunosuppression causes HBV reactivation and can lead to high viremia without clinical manifestation [40]. The APRI score was used to compare the frequency of liver fibrosis in the HBV +ve versus HBV 2ve participants. The frequency of liver fibrosis was significantly higher in HBsAg +ve HBV DNA +ve individuals compared to seronegative HBV DNA 2ve ones, but not relative to seropositive HBV DNA 2ve ones. It is intriguing that there was no difference in the frequency of liver fibrosis between HBV DNA +ve individuals, with and without HBsAg.
The reactivation of an infection, which originated in childhood, can explain why no significant difference was seen in the HBV viral loads between the HBsAg +ve and HBsAg 2ve participants (Table 2), nor between the different serological groups (Figure 2). Following HIV infection, HBV can reactivate in anti-HBs +ve only individuals, with and without the reappearance of HBsAg [41]. Group F, which had the lowest CD4 count of ,100 cells mm 23 , and by inference was the most immunosuppressed, was HBsAg +ve anti-HBc +ve anti-HBs +ve HBV DNA +ve with a viral load .10 2 IU ml 21 (Figure 2). Spontaneous reverse seroconversion, where anti-HBs disappears and HBsAg reappears can also occur in the presence of CD4 counts ,200 cells mm 23 [42]. Although HBV viral loads have been shown to be higher in HBV +ve HIV +ve individuals compared to HBV +ve ones [43], the HBV viral loads detected in the present study were comparable to those detected in HBV mono-infected individuals [44]. This is probably because the majority of individuals were infected with subgenotype A1 [45], which is characterized by relatively low viral loads in monoinfected individuals compared to other genotypes or subgenotypes [44].
Only three HBsAg 2ve HBV DNA +ve patients had HBV loads ,200 IU ml 21 , meeting the Taormina criterion for OBI. Thus the majority of HBsAg 2ve HBV DNA +ve would be classified as false ''occult'' [19]. It is possible that immunosuppression precludes true occult HBV infection. Because the majority of HBsAg 2ve HBV DNA +ve (''occult'') participants did not differ from HBsAg +ve HBV DNA +ve (overt) participants in terms of viral loads, CD4 counts, ALT levels and frequency of liver fibrosis, it may be more accurate to refer to these HBV infections as HBsAg-covert (HBsAg-cryptic overt) instead of false ''occult'' [19].
HIV infection was demonstrated to be a risk factor for HBsAg 2ve HBV infection [12], and pre-S mutations preventing HBsAg secretion [46], 'a' determinant mutations leading to detection escape and overlapping polymerase mutations affecting replication, may be responsible for this. This possibility was investigated and is presented in a follow-up paper, where 12 of 13 HBV S region sequences, from HBsAg 2ve participants, had pre-S and/or S mutations [45]. Another possible explanation for HBsAg-negativity may be that HIV co-infection prevents HBsAg secretion, as shown in co-infected hepatic cell lines [47].
Despite the possible limitations of this study, including its crosssectional nature, the absence of HIV viral loads, no HBV monoinfected patients and patients with higher CD4 counts for comparison, a number of important conclusions can be reached. The number of lifetime sexual partners was the only factor differentiating HBV DNA +ve and HBV DNA 2ve infections, suggesting sexual transmission of HBV and/or HIV. HBV +ve HIV +ve individuals were found to have significantly higher lifetime sexual partners than HBV-monoinfected individuals [18]. HBV infection in HIV +ve individuals was predominantly HBsAg 2ve , which did not differ significantly from HBsAg +ve infections in terms of viral loads, CD4 counts, ALT levels and frequency of liver fibrosis.
The detection of HBV DNA in the absence of HBsAg in this and other South African studies [11][12][13]17] has important implications for the clinical management of HIV in sub-Saharan Africa, where the burden of HBV/HIV co-infection is disproportionately high (24% in this study). Although the World Health Organization recommends that ART be initiated in HBV/HIV co-infected individuals irrespective of CD4 count, in South Africa we face a number of challenges. The most recent South African guidelines recommend initiation of treatment of patients with CD4 counts ,350 cells mm 23 and HBsAg testing if ALT levels exceed 100 U L 21 . Considering that the highest median ALT levels (IQR) of 30 (19-59) U L 21 were found in the HBsAg +ve HBV DNA +ve group (Table 3), which also had the highest frequency of advanced fibrosis, this cut-off value is inappropriate. Moreover, 65% of the 71 participants, who were HBV +ve HIV +ve , lacked HBsAg and HBV could only be detected by nucleic acid testing, which is unaffordable in resource-limited environments. Although the clinical significance of HBsAg 2ve infection is under debate [32], it is imperative that HBV/HIV co-infection is detected before ART initiation, especially because lamivudine remains in two of the three drug regimens currently provided by the South African government and HBV can develop resistance to lamivudine. To determine the clinical relevance of HBsAg-covert HBV infection in our setting, prospective studies following ART initiation are in progress.