Search strategy

Between October 2017 and January 2018, we searched the published literature, in MEDLINE (PubMed; https://www.ncbi.nlm.nih.gov/pubmed), SCOPUS (https://www.elsevier.com/solutions/scopus) and EMBASE (https://www.elsevier.com/en-gb/solutions/embase-biomedical-research). Our search strategy is detailed in S1 Table (documenting use of PRISMA criteria and selection of studies) and S2 Table (listing our search criteria). The earliest paper we identified on HBV drug resistance in Africa was published in 2007. We reviewed the titles and abstracts matching the search terms and only included those relating to drug or vaccine resistance in HBV infection, including only those that presented original data and had undergone peer review. All retrieved articles were in English, therefore no exclusion in relation to language was required.

For each publication we recorded reference, publication year, study design, sample size, study population, proportion of participants who tested HBsAg+ or HBV DNA+, country, year(s) of specimen collection, genotype identified, antiviral treatment, sequencing method, gene sequenced, number of sequenced samples, participant recruitment site and sequence accession number. Data were curated using MS Excel software (Microsoft, Redmond, WA).

RAMs reported in published sequences not represented in primary studies

We expanded our search for evidence of RAMs by identifying publicly available HBV sequences from Africa, that had not been included in the results of our primary literature search. We used both the Hepatitis B Virus database (https://hbvdb.ibcp.fr/ [36] and Hepatitis Virus Diversity Research Alignments database (http://hvdr.bioinf.wits.ac.za/alignments/) [37].

Analysis

In order to determine the prevalence of RAMs and VEMs, we first reported these using the denominator (total number of HBV positive patients) and numerator (total number of HBV positive patients with the specified mutation) as reported in published studies. We also pooled data by country in order to provide regional estimates. Downloaded sequences were managed using Sequence editor, database and analysis platform, SSE version 1.3, for analysis [38].

Data visualisation

We developed an R package, gene.alignment.tables, for the visualisation of the sequence data in this study; this is available on Github [39] and can be used for visualising generic gene sequence datasets. The package was developed by University of Oxford’s Interactive Data Network and a specific instance of the visualisation is hosted as a Shiny app which can be viewed here: https://livedataoxford.shinyapps.io/1510659619-3Xkoe2NKkKJ7Drg/ [40].

Study characteristics

Epidemiological data for HBV represented by the 37 studies we identified are summarised in Table 1. Studies included were from Southern Africa (Botswana, Mozambique, South Africa, Zambia and Zimbabwe), East Africa (Ethiopia, Kenya, Malawi, Sudan and Uganda), West Africa (Cote d’Ivoire, Gambia, Ghana, Guinea-Bissau and Nigeria) and Central Africa (Cameroon, Gabon). There was considerable heterogeneity in recruitment protocols and exposure to anti-viral treatment. Twenty-six studies recruited from hospitals, three studies recruited from the community [8,41,42] and eight studies did not specify where recruitment was undertaken [10,43–49]. All studies were observational.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Prevalence of HBsAg and HBeAg from 37 studies of HBV drug resistance in Africa.

https://doi.org/10.1371/journal.pntd.0006629.t001

Study populations were categorised as follows:

• HBV/HIV coinfected patients: (n = 28 studies), [8–10,12–14,20,43–48,50–56,58–62,64,67,70];
• HBV infected with and without HIV coinfection: (n = 8 studies), [41–43,57,63,65,66,68];
• Chronic HBV monoinfection: (n = 1 study), [69].

Antiviral treatment exposure varied as follows:

• Treatment-naïve: (n = 8 studies), [14,46–48,63,64,68,70];
• 3TC-based regimen only: (n = 10 studies), [8,9,20,44,45,49,53,59,61,69];
• Regimens including 3TC or TDF: (n = 6 studies), [10,13,51,52,55,65];
• Mixed regimen where some received 3TC, others TDF, while others left untreated; (n = 7 studies), [12,50,54,56,60,62,66];
• Treatment regimen not specified: (n = 6 studies), [41–43,57,58,67].

HBV amino acid polymorphisms were studied from within the following proteins;

• Pol only (n = 13 studies), [8,9,12–14,20,44,53,55,56,59,63,68]; only one of these used a deep sequencing method [20];
• S only (n = 3 studies), [43,57,65];
• Pol and S (n = 12 studies),[10,41,45,48,51,52,54,58,60,62,64,67];
• Pol, S and PC/BCP (n = 4 studies), [47,50,61,69];
• S and PC/BCP region (n = 3 studies), [42,46,70];
• Whole genome (n = 2 studies), [49,66].

All studies, except for two [53,68], specified the HBV genotype (S3 Table & S2 Fig).

Prevalence of HBsAg, HBeAg and HDV coinfection

The prevalence rates of HBsAg in these study cohorts ranged from 3%-26%; however, the populations included were highly selected and therefore not necessarily representative of the general population, particularly as a result of a strong bias towards HIV-infection (Table 1). Only three studies included in this review reported on HDV prevalence: two studies did not detect any HDV antibodies [63,65], whereas the other study reported a HDV prevalence of 25% in Guinea-Bissau [56].

RAMs identified in African cohorts

The co-occurrence and distribution of HBV RAMs and VEMs are summarised according to the region where they were identified (Fig 2). This illustrates the patchy and limited data that are available, with South Africa, Ghana and Cameroon best represented, but with large areas (especially in northern and central Africa) not represented at all in the literature.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Annotated map to summarise HBV drug resistance associated mutations (RAMs) and vaccine escape mutations (VEMs).

Mutations identified from 33 studies of African cohorts published between 2007 and 2017 (inclusive). Four studies identified by our systematic literature review were not represented here as they did not report any RAMs. Full details of each citation can be found in Table 1.

https://doi.org/10.1371/journal.pntd.0006629.g002

Although 35 studies specified the HBV genotype, it was only possible to group RAMs according to genotype in fourteen studies [8,9,13,14,44,46,47,50,51,56,60–62,69] (S1 Fig; S2 Fig). The remaining 21 studies generally reported the genotypes identified, but did not specifically state the genotype of HBV within which RAMs were identified.

We have developed an interactive tool to display the genomic positions of RAMs identified through our literature review alongside relevant metadata. This can be accessed on-line here: https://livedataoxford.shinyapps.io/1510659619-3Xkoe2NKkKJ7Drg/ [40].

Overall, the most prevalent RAM was rtM204V/I in both treatment experienced and treatment naïve individuals, and occurring either alone or in combination with other polymorphisms rtL80I/V, rtV173L, rtL180M, rtA181S, rtT184S, rtA200V and/or rtS202S (Fig 3); mutations among individuals with and without exposure to HBV therapy are listed in S4 Table and S5 Table, respectively). This mutation was present in 29 studies at a highly variable prevalence of between 0.4% [12] and 76% [69]. Across all cohorts, the mutation was present in 208/2569 (8%) of all individuals represented. The mutation, by itself, was most prevalent in South Africa; on pooling data for three studies from this setting, it was present in both treatment experienced and treatment naïve patients (n = 13/17, 76% [69] and n = 16/72, 22% [67,68] respectively). In addition to South Africa, rtM204I/V was also frequent in Malawi among treatment experienced patients (n = 24/154, 16% [20,45]) (Fig 3), and in genotype non-A infection: in this setting, the mutation was detected in genotype C infection (n = 2/17, 12% [69]) (S2 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Prevalence of HBV resistance associated mutations (RAMs) in Pol/RT proteins among HBV infected patients in Africa.

These data are derived from 27 studies of HBV drug resistance in Africa published between 2007 and 2017 (inclusive). The countries represented are listed in alphabetical order. A detailed summary of RAMs identified from each study is presented (Fig 2, S4 Table, S5 Table). Prevalence of RAMs for a specific country was determined by grouping all studies from that country that reported a specific mutation. We used all individuals who tested HBsAg positive to generate a denominator in order to provide a conservative estimate of RAM prevalence, and the numerator was the total number of individuals with that specific mutation from these studies. A: treatment naïve; B: treatment experienced.

https://doi.org/10.1371/journal.pntd.0006629.g003

The rtM204I/V mutation by itself confers resistance to 3TC; in combination with A194T it may also be associated with reduced efficacy to TDF, and in combination with L180M and V173L with vaccine escape, through corresponding substitutions in the surface antigen sites targeted by neutralising antibodies. Although TDF has a high genetic barrier to resistance, and is associated with reliable suppression of HBV viraemia [7,10,21,22], mutations rtN236T and rtA194T, which have been linked with resistance to both TDF and ADV [7], have been identified in Southern Africa in both treatment naïve [14] and treatment experienced [10] patients.

WHO guidelines recommend a first-line regimen including TDF in HIV/HBV coinfected patients [6], and the South African Department of Health HIV/AIDS treatment guideline included TDF as first-line regimen from 2010 [71], however we found a minority of studies (9/37, 24%) reporting TDF-containing regimens for HIV/HBV coinfected individuals. As anticipated, most of the studies that did use TDF were carried out after 2010, whereas those that used 3TC were generally earlier (S3 Table).

From this dataset, it is difficult to ascertain whether RAMs are genuinely more prevalent in genotype A infection, or this simply reflects enrichment of genotype A in sub-Saharan African populations (S2 Fig). Interpreting RAMs according to sub-genotypes was difficult since most studies did not specify sub-genotype and others did not indicate which RAMs were identified in which genotype. Of concern is the detection of RAMs even in reportedly treatment naïve individuals (Fig 3 & S4 Table), suggesting that RAMs are being transmitted. A study in South Africa that recruited 3TC-naïve HBV infected adults with or without HIV, reported rtM204I in 13/35 (37%) individuals [68].

HBV RAMs in published sequences from Africa

We searched the Hepatitis B Virus database and GenBank to identify HBV sequences derived from Africa, from studies not already included in our review. We identified an additional 69 isolates: 23 had undergone full length genome sequencing whereas 46 isolates represented either the polymerase (n = 3) or S region (n = 43) of the HBV genome Table 2. To avoid duplication of results, we excluded fourteen studies already identified by our literature review that had submitted their sequences to GenBank (S3 Table). RAMs in the additional 69 isolates were as follows:

• rtM204V in genotype A (2/69, 2.9% of sequences), this occurred in combination with rtL180M;
• rtM204V + rtL180M in genotype E (1/69, 1.5%);
• rt180M + rtA181V in genotype E (1/69 (1.5%);
• rtQ215S identified in genotype D (4/69, 5.8%).

All these mutations are associated with 3TC resistance; rtA181V has also been associated with reduced susceptibility to TDF [7,15].

In the S gene, the most prevalent mutations were:

• sD144A/E/G occurring in genotype A (6/69, 8.7%), D (10/69, 14.5%) and E (7/69, 10.1%) associated with VEM;
• sI110L occurring in genotype A (3/69, 4.3%), D (4/69, 5.8%) and E (11/69, 15.9%) associated with immunoglobulin resistance.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. HBV drug resistant mutations (RAMs) identified from HBV genome sequences from Africa downloaded from the Hepatitis B Virus database (https://hbvdb.ibcp.fr/) [36] and GenBank database (http://hvdr.bioinf.wits.ac.za/alignments/) [37].

https://doi.org/10.1371/journal.pntd.0006629.t002

VEMs

VEMs were identified in Central, East, West and Southern Africa (Fig 2). However, it was not possible to ascertain whether individuals harbouring these mutations had been vaccinated against HBV infection. The most common VEM was the triple mutation rtV173L + rtL180M + rtM204I/V, found in the Pol gene. This suite of mutations was identified in 14 studies [12,13,20,44,50–55,58–60,62], at a pooled prevalence of 4% (57/1462). Another significant VEM, sG145K/R [16], was identified in six studies [12,42,47,57,60,62] and sM133L/T, associated with VEM, immunoglobulin and diagnostic escape mutation [12,48], was identified in seven studies [12,41,47,48,57,62,70] (Fig 2).

Summary

To our knowledge, this is the first systematic review that assesses RAMs and VEMs for HBV in Africa. The high rates of HBV infection among HIV infected individuals in some locations including Cameroon [60] and South Africa [10] could be an indication that HBV infection has been previously under-reported, possibly due to lack of routine screening, poor awareness, stigma, high costs and limited clinical and laboratory infrastructure [4,8–10,45,53]. The literature suggests a widespread exposure of the HIV-infected population to 3TC-based treatment. This may be changing over time in line with current ART treatment recommendations (regimens for Africa summarised in S6 Table), but the introduction of TDF-based regimens for HIV treatment has been inconsistent, and TDF monotherapy is not consistently available for HBV infection in the absence of HIV.

In keeping with other settings, the most common RAM identified here was rtM204I/V, either alone or in combination with compensatory mutations rtL180M ± rtV173L. Of concern, rtM204V/I was seen in 76% of treatment experienced patients [69] and 22% of treatment naïve patients [67,68] in South Africa. A review of worldwide incidence of RAMs among treatment naïve patients also described rtM204V/I as the most frequent, but with a much lower prevalence of 5% [72]. The contribution of unreported or undocumented 3TC exposure in the reportedly treatment naïve populations remains to be determined. A European study demonstrated that the most frequent primary mutation was rtM204V/I, found in 49% of treatment experienced patients [73], while in China rtM204I, rtN236T and rtL180M+rtM204V+rtV173L/rtS202G were also the most prevalent RAMs [74].

The triple mutation rtM204V + rtL180M + rtV173L has been identified in East, West and Central Africa [20,44,51–54,59]. This combination of polymorphisms is associated with both vaccine escape and resistance to 3TC and other L-nucleoside analogues [20,44,51,54,59,60]. Interestingly, this triple mutation has not been reported in the Southern African region to date, which is likely to reflect the composition of the study populations.

Clinical and public health significance of RAMs

Apart from the nature of drugs being used for HBV treatment, other reported predictors of HBV drug resistance include HBV viral load, HBV intra host heterogeneity, HBeAg status, host body mass index and serum alanine aminotransferase (ALT) activity [20,75,76]. Individuals with rtM204V/I plus compensatory mutations typically exhibit high HBV DNA levels [20] and are therefore highly infectious to others. The spread of RAMs may lead to a rise in drug resistance in treatment naïve chronic HBV infection, representing a substantial challenge for Africa and highlighting an imperative to ensure routine use of TDF in preference to, or in combination with, 3TC-based therapy.

Although these data provide a preliminary picture of the prevalence of RAMs in some settings, there are no recommendations to stipulate any specific prevalence threshold above which HBV drug resistance mutations represent a significant barrier to successful treatment at a population level, and/or RAM prevalence thresholds that should trigger a switch to alternative first-line therapy. For HIV, surveillance for transmission of RAMs is based on screening recently infected, treatment naive individuals, and classifies drug resistance using thresholds of <5%, 5–15%, and >15% to stratify the risk to public health [77]. Similar thresholds and recommendations for HBV could help to underpin the assimilation of epidemiological data and to unify treatment approaches.

TDF resistance

The identification of mutations associated with reduced TDF susceptibility are of concern, as they suggest the potential for increasing prevalence of polymorphisms that confer partial or complete viral escape from a drug that to date has not been widely associated with resistance. There is now potential for increasing selection of TDF resistance as this drug becomes more widely used. However, as a new first line single tablet option incorporating 3TC, TDF and Dolutegravir (DTG) (triple therapy abbreviated to ‘LTD’) emerges as a recommended option for HIV treatment in Africa, surveillance is needed to determine the clinical outcomes for HBV [78].

If clinically significant, TDF resistance mutations may still represent a particular problem for many African settings, as resource constraints make it unrealistic to provide baseline screening for RAMs, or to monitor patients on treatment with serial viral load measurements. Despite these potential concerns, it has been shown that TDF is effective even in the presence of RAMs and that there is comparable efficacy among 3TC-experienced and NA-naïve patients [79].

VEM

VEM were identified in 16 different countries in East, West, Central and Southern Africa. Information on vaccine exposure was not available, but there are two strands of evidence to support significant population exposure to HBV vaccination. First, vaccination has been progressively rolled out in most countries in sSA since the mid-1990’s; second, most HBsAg mutations reported by these studies are located within the common immunodominant B cell epitope (aa 124–147) in which selection of polymorphisms is associated with HBV vaccination [80,81].

VEM have been more robustly reported from Asia, in settings where the HBV infant vaccination programme is well established; for example, in Taiwan, VEM prevalence among vaccinated children increased from 7.8% to 23.1% within 15 years of the launch of the universal vaccine program, although the decline in VEM prevalence thereafter may be partly related to a smaller HBV carrier pool [80]. HBV infection despite immunoprophylaxis can occur either as a consequence of MTCT of pre-existing VEM, or as a result of de novo selection of escape mutations from vaccine- induced immune responses, particularly in the setting of delayed vaccination [80,81].The HBV genotype sequence used for vaccines may potentially have an influence on immunogenicity against non-vaccine genotypes, but there are limited data to support this [82]. Only 11 African countries recommend the first HBV vaccine dose at birth, in contrast to the majority of African countries in which HBV vaccination is delayed until 6 weeks of age [33]. It is likely that this delay not only provides a window of infection but also increases the possibility of transmitted VEM and/or emergence of new escape mutations.

High maternal HBV viral load and immunosuppression are other risk factors associated with VEM among infants [80]; both of these are pertinent for emergence of VEMs in Africa given that HBV viral load testing is not routinely available, and HIV is highly prevalent in some populations. Effective PMTCT strategies in Africa, including screening and treating antenatal women, increasing access to viral load monitoring, and introducing HBV birth dose vaccine will help to decrease the prevalence of VEM [4,33,83].

HDV/HBV coinfection

One study from our literature review reported a high HDV prevalence of 25%; however, in this cohort, RAMs occurred in individuals with HBV monoinfection [56]. Given that HDV is characteristically associated with decreased HBV replication [84], it is possible that emergence of HBV RAMs is altered in this setting. However, as the true prevalence and impact of HDV in sSA is not known [85], further studies are needed to determine the impact of HDV coinfection on HBV RAMs.

Limitations of current data

Screening for HBV infection is not routinely performed in many African settings and therefore the true prevalence and characteristics of HBV infection are not known [4,7–10]. We identified very few published studies; only a minority of patients had HBV sequencing undertaken, and there were no data from certain regions of Africa. This highlights the substantial problem of HBV neglect in Africa, and a specific blind-spot relating to sequence data [4]. Identifying the true prevalence of resistance mutations, and characterising the populations in which these are selected and enriched, is currently not possible due to sparse data and lack of clear descriptions of the denominator population. Most such studies do not perform a truly systematic assessment, but focus on high risk groups–particularly including those with HIV/HBV coinfection: of the 37 studies included here, only one exclusively reported on participants who were HBV mono-infected [69]. Although we have made every effort to assimilate the relevant data to build up a regional picture for Africa, the heterogeneity between studies makes it difficult to draw robust conclusions from pooled data. These findings are a reflection of the little attention paid towards the burden of this disease in Africa and the neglect in robust epidemiological data.

Only nine studies undertook a longitudinal approach to detection of drug resistance [8–10,20,44,45,49,52,53]. The results of the other 28 studies that undertook a cross-sectional approach could be skewed by the timing of recruitment of study participants, with a risk of under-representation of drug resistance if screening is undertaken only at baseline, and potentially an over-representation if screening is undertaken in patients with HIV coinfection, who are more at risk of advanced disease and prolonged drug exposure. As most of these studies recruited individuals from hospital settings, this raises the latter possibility.

Mutations across the whole genome might be relevant in determining resistance [86]. However, most of the included studies analysed only defined genes from within the HBV genome; only two sequenced the whole genome, and these determined consensus sequence. This potentially results in an under-representation of RAMs and VEMs that may be present as low numbers of quasispecies, but could become significant if selected out by exposure to drug or vaccine.

In studies that reported RAMs among treatment naïve individuals, the literature suggests that sequence analysis was performed prior to ART initiation. However, we cannot exclude the possibility that that some of these participants had prior ART exposure. Due to the nature of the cohorts that have been studied, most of the RAMs identified were from HIV/HBV coinfected individuals. It is possible that HIV increases the risk of HBV RAMs both in terms of drug exposure, and also as a function of increased HBV viral loads. A study from Malawi demonstrated the rapid emergence of 3TC resistance in HIV coinfection, with virtually all treatment naïve HBeAg positive individuals starting antiviral treatment showing emergence of rtM204I by six months. Likewise, a study carried out in Italy revealed that patients with HIV coinfection were more likely to harbour the rtM204V mutation and to show multiple mutations compared to HBV monoinfected patients [87]. It would be worth further exploration of this observation in Africa, as there are currently very limited data.

Challenges and opportunities for Africa

A major challenge for Africa is to improve coverage rates of infant vaccination, deploy catch-up vaccination programmes for older children and adults, adopt widespread screening and develop treatment programmes for HBV. While HBV vaccine is effective, gaps in vaccine coverage in Africa can be demonstrated by the high perinatal transmission rate of HBV in sSA (estimated at 38% among women with a high HBV viral load) and the observation that up to 1% of newborns in sSA are still infected with HBV [88]. Sustained efforts are required to build robust PMTCT programmes that deliver screening and treatment for antenatal women, and timely administration of HBV birth vaccine for their babies [33,83].

Although the WHO recommends monitoring for the development of drug resistance once on therapy [6], implementation remains challenging as viral load monitoring and sequencing are both rarely available [7]; despite the advancement and availability of HIV testing and monitoring, in many settings it remains uncommon to monitor HIV viral load after ART initiation [89,90]. Affordable, accessible and sustainable platforms for quantifying both HIV and HBV viral loads remain an important priority for many settings in Africa, given the lack of on-treatment monitoring in many settings. Given the simplicity and relative ease of collection, preparation and transport of dried-blood-spot (DBS) samples [91], adopting DBS testing could improve access to HBV diagnosis, viral load monitoring and linkage to care, especially in areas with limited access to laboratory facilities.

Development of a cheap, rapid test for the detection of the most frequently observed RAMs and VEMs should be considered as a potentially cost-effective strategy for Africa. Proof of principle for a rapid test for diagnosis and detection of resistance has been demonstrated by the GeneXpert MTB/RIF assay for Mycobacterium tuberculosis (MTB) [92]. A similar approach has been applied for HBV through use of a multiplex ligation-dependent probe real time PCR (MLP-RT-PCR) [93]. Although this assay is able to detect RAMs quickly and cheaply, there are still limitations as the test requires high viral load samples, is based on detection of known RAMs from within discrete regions of the genome, and may not identify RAMs that are present as minor quasispecies.

New metagenomic sequencing platforms, such as Illumina and Nanopore, provide the opportunity for whole deep genome sequencing, which can reveal the full landscape of HBV mutations in individual patients, quantify the prevalence of drug resistance mutations among HBV quasi-species, and determine the relationship between these polymorphisms and treatment outcomes [87]. Nanopore technology also has the potential to develop into an efficient point of care test that could detect viral infection and coinfection, as well as determining the presence of VEMs and RAMs [94], but is currently limited by cost and concerns about high error rates.

There have been few studies looking at the correlation between genotype, clinical outcomes of disease, response to antiviral therapy and RAMs/VEMs, but none from Africa. Studies outside Africa have shown that genotype A is more prone to immune/vaccine escape mutants, pre-S mutants associated with immune suppression, drug associated mutations and HCC in HIV/HBV coinfected participants [46,87,95]. Studies investigating the role of genotypes in predicting response to antiviral therapy and their association with various types of mutations are urgently needed in Africa, particularly in light of the high frequency of genotype A infection and high population exposure to antiviral agents that have been rolled out over the past two decades as a component of first-line ART.

Existing infrastructure for diagnosis, clinical monitoring and drug therapy for HIV represents an opportunity for linkage with HBV care. Particularly in settings of limited resource, joining up services for screening and management of blood-borne virus infection could be a cost-effective pathway to service improvements.

Conclusions

This review highlights the very limited data for HBV RAMs and VEMs that are available from Africa. Scarce resources resulting in lack of diagnostic screening, inconsistent supply of HBV drugs and vaccines, and poor access to clinical monitoring contribute to drug and vaccine resistance, potentially amplifying the risk of ongoing transmission and adding to the long-term burden of HBV morbidity and mortality in Africa. We call for urgent action to gather and analyse better data, particularly representing the HBV monoinfected population, and for improved access to TDF.

HBV RAMs and VEMs have been identified in several African countries among HIV/HBV coinfected and HBV monoinfected patients, before and during treatment with NAs but the data are currently insufficient to allow us to form a clear picture of the prevalence, distribution or clinical significance of these mutations. Overall, the data we describe suggest a significantly higher prevalence of drug resistance in some African populations than has been described elsewhere, and that is not confined only to drug-exposed populations, highlighting an urgent need for better population screening, assessment of HBV infection before and during therapy, and increasing roll out of TDF in preference to 3TC. At present, TDF accessibility is largely confined to HIV/HBV coinfected individuals; we now need to advocate to make monotherapy available for HBV monoinfected individuals. However, there are uncertainties as to whether its long-term use might result in nephrotoxicity, and potentially in an increase in selection of TDF RAMs.

We should ideally aim for the goals of a combined HBV test that includes diagnosis of infection, genotype and presence of RAMs/VEMs; new sequencing platforms such as Nanopore make this technically possible, although cost remains a significant barrier at present. Sustainable long-term investment is required to expand consistent drug and vaccine supply, to provide screening infection and for drug resistance, and to provide appropriate targeted clinical monitoring for treated patients.