HBV seroepidemiology data for Africa provides insights into transmission and prevention

International goals for elimination of hepatitis B virus (HBV) infection set ambitious targets for 2030. In many African populations, HBV prevalence remains high (≥8%) despite the roll-out of infant HBV immunisation from the mid-1990’s onwards. Enhanced efforts are now urgently required to improve an understanding of population epidemiology, in order to determine which interventions are most likely to be effective in advancing populations towards elimination goals. In populations with a high prevalence of infection, catch-up HBV vaccination of adults has sometimes been deployed as a preventive strategy. An alternative approach of ‘test and treat’ could be applied as a tool to interrupt transmission. We used a systematic approach to investigate the relationship between prevalence of HBV infection (HBsAg) and exposure (anti-HBc) in Africa, and then applied a mathematical model to investigate the impact of catch-up vaccination and a ‘test and treat’ strategy in Uganda, representing a high prevalence setting. We demonstrate a strong relationship between the prevalence of HBsAg and anti-HBc (p<0.0001), but with region-specific differences that may reflect different patterns of transmission. In high prevalence settings, catch-up vaccination may have a transient effect but this intervention does not contribute to a sustained decline in prevalence. In contrast, diagnosing and treating infection has a marked impact on reducing prevalence, equivalent to that of infant immunisation. Conclusion: We have developed a high-resolution picture of HBV epidemiology across Africa. Developing insights into regional differences provides an evidence base for the most effective interventions. In combination with robust neonatal immunisation programmes, testing and treating infection is likely to be of most impact in making advances towards elimination targets.


159
We recorded total anti-HBc prevalence (i.e. proportion of population exposed to HBV, 160 irrespective of chronic infection status, termed 'total exposure') and also calculated the 161 proportion of the population with cleared infection (i.e. anti-HBc prevalence minus 162 HBsAg prevalence, termed 'exposed and cleared'). For studies reporting prevalence 163 data from ≥2 cohorts (e.g. HIV-positive and HIV-negative populations), we recorded 164 these as a single publication but ≥2 distinct data points. Studies in a language other 165 than English were translated using Google Translate (https://translate.google.com/).

166
We considered Uganda as an exemplar setting where HBsAg seroprevalence in adults   The UN geoscheme classifies Africa into Central, Eastern, Northern, Southern and 178 Western regions; this is a standard approach for sub-dividing macro-geographical 179 areas for statistical analysis. For the regional analysis, each study was assigned equal 180 weighting when analyzing the data, regardless of the study size. We analysed 8 prevalence data for anti-HBc and HBsAg using Graphpad Prism v7 . 0. For non-182 parametric data, we sought significant differences between data sets using Mann-183 Whitney U tests, and for multiple comparisons we used 1-way ANOVA. We used linear 184 regression to derive lines of best fit, 95% confidence intervals and to interpolate HBsAg 185 prevalence from anti-HBc prevalence. We generated maps to illustrate the location of 186 the HBV cohorts and seroprevalence of relevant markers using R ( Table 1). PMTCT (combining accelerated neonatal immunisation with HBIg and 208 9 antiviral therapy in pregnant mothers) and vaccine-based interventions were modelled 209 as previously described (10), and we added a 'test and treat' strategy. The latter was 210 simplified to reducing the transmission potential of the HBV infected proportionally to 211 the control effort (e.g. 20% coverage of test and treat in a particular age-group equated 212 to a 20% reduction in that group's force of infection).

224
The distribution of these cohorts and the prevalence of HBV serological markers is

237
Western Africa has the highest population exposure and correspondingly highest rates 238 of HBsAg positivity ( Fig 2E, Fig 3). HBsAg prevalence differs significantly between

259
HBc prevalence was also higher in HIV-positive cohorts than in HIV-negative cohorts 260 for 2/3 studies (22, 24). In a third study of highly exposed cohorts in South Africa, anti-  Table 4).

272
Based on combining the mean prevalence values from Uganda cohorts to provide a 273 broad overview, 54% of adults across this country have been exposed (among these,

299
We also modelled the impact of 'test and treat', based on the premise that the whole 300 population is screened, projecting that this strategy has the fastest reduction in HBV 301 population prevalence of all interventions with 62% reduction in prevalence by 50 302 years, and 98% at 200 years ( Fig 4A, purple band). Recognising the significant barriers 303 to identifying all cases of infection, (including silent infection, lack of education, poor 304 access to laboratory facilities, and stigma) (2), we also modelled the outcome for 'test 305 and treat' strategies that reach <100% of the HBV-infected population. Diagnosis and 306 treatment for 80% of infected adults (Fig 4B, green band) or 50% of the whole infected 307 population (Fig 4B, red band) delivers a reduction in HBsAg prevalence over time that 308 is comparable to infant immunisation (Fig 4A, blue band). Even reducing the population 309 tested and treated to only 50% of adults (Fig 4B, orange band) is still substantially 310 more effective than 100% catch-up vaccination (Fig 4A, orange band).

336
there is evidence of region-specific differences in exposure and transmission.

338
In order to make progress towards HBV elimination goals, we therefore suggest that

361
Our seroepidemiology review highlighted considerable regional differences in the

386
There was no evidence from our dataset that HIV+ individuals were more likely to be 387 either HBV infected or exposed, in keeping with previous reports (36

405
which is rarely detected due to lack of availability and high cost of HBV DNA testing.

406
However, individuals with occult HBV would still generate anti-HBc; thus while we may 407 be underestimating the prevalence of active infection, these subjects are still included 408 within our exposed population.

410
We did not include data for anti-HBs prevalence (immunised population) as a limited

416
We included papers published after the EPI introduction of HBV vaccine in 1995, in 417 order to make our study applicable to current-day vaccinated populations, although in 418 practice, roll-out of the vaccine was patchy and adopted at a variable rate over the

458
We advocate significant investment in capacity building for improving HBV diagnosis

465
Factor Rationale for contribution to regional differences in HBV seroepidemiology Circulating HBV viral genotype Predominant genotype varies by region with genotype-A common in Southern Africa, genotype-D in the North and genotype-E in the West (39).

Host ethnicity and genetics
HLA-type and T-cell repertoire have been linked to the ability to control the infection (40-42).

Transmission differences
Subtle differences in the transmission patterns (vertical vs horizontal) of the HBV genotypes have been documented. Transmission route is fundamentally linked to age at exposure (43).

Age at exposure
The probability of developing chronic HBV after exposure is strongly associated with age (44). Populations with a younger age at exposure are therefore likely to have a higher HBsAg prevalence relative to the anti-HBc prevalence (Fig 5A).

Co-infection within population
Risk factors for acquisition of blood-borne viruses overlap between HIV, HBV and HCV. Egypt and the Nile Delta have some of the highest reported prevalences of HCV globally. Co-infection of HBV and HCV has been linked to spontaneous clearance of HCV although evidence of the impact on HBV remains scarce (45, 46).

Political instability
Central Africa includes several regions disrupted by recent conflict and resulting population migration, with powerful influence on increases in interpersonal violence and sexual assault, reduced access to barrier contraception, inadequate screening of blood products, and reduced access to healthcare, all of which can increase exposure rates in the adult population.

Traditional cultural practices
Exposure to blood-borne viruses is influenced by traditional healing practices, scarification, piercing, tattooing and non-sterile surgical practice (e.g. circumcision).

Uptake of HBV vaccination in the region
Countries with earlier uptake of the HBV vaccine are likely to have lower anti-HBc and HBsAg prevalence than countries that implemented the vaccine later. Prevalence of vaccine escape mutants may contribute, although data for Africa are scarce (33).

Potential role of insect vector
Biting insects capable of mechanical transmission of HBV may be prevalent in some regions, although there is a lack of firm evidence base for HBV transmission (47).

530
In adults, assuming that different populations are exposed at the same rate and the 531 risk of chronic infection is constant (estimated to be 5% in healthy adults as shown in vaccination on incidence is therefore related to S, with reduced impact in highly 537 exposed populations. In a test and treat scenario, with 50% of cases identified and 538 treated, incidence is consistently reduced, regardless of S.