Novel vaccination approaches to prevent tuberculosis in children

Pediatric tuberculosis (TB) is an underappreciated problem and accounts for 10 % of all TB deaths worldwide. Children are highly susceptible to infection with Mycobacterium tuberculosis and interrupting TB spread would require the development of effective strategies to control TB transmission in pediatric populations. The current vaccine for TB, M. bovis Bacille Calmette-Guérin (BCG), can afford some level of protection against TB meningitis and severe forms of disseminated TB in children; however, its efficacy against pulmonary TB is variable and the vaccine does not afford life-long protective immunity. For these reasons there is considerable interest in the development of new vaccines to control TB in children. Multiple vaccine strategies are being assessed and include recombinant forms of the existing BCG vaccine, protein or viral candidates designed to boost BCG-induced immunity, or live attenuated forms of M. tuberculosis. A number of these candidates have entered clinical trials; however, no vaccine has shown improved protective efficacy compared to BCG in humans. The current challenge is to identify the most suitable candidates to progress from early to late stage clinical trials, in order to deliver a vaccine that can control and hopefully eliminate the global threat of TB.


Background
Tuberculosis (TB), caused by the intracellular bacterial pathogen Mycobacterium tuberculosis, remains a major cause of mortality and morbidity worldwide. Annually there are an estimated 9.6 million new cases of clinical TB and 1.5 million deaths, the majority occurring in South-East Asia [1]. The spread of TB is fueled by the human immunodeficiency virus (HIV)/acquired immune deficiency syndrome (AIDS) pandemic, the emergence of multi-drug resistant strains and socio-political disruption to health services. TB is a chronic infection and is generally considered a diseases of adulthood, however pediatric TB is an underappreciated problem. Infants under the age of two have the highest risk of contracting TB [2]. One million children contracted TB in 2014 and approximately 10 % of all TB deaths were in children [1]. The current TB vaccine, M. bovis Bacille Calmette-Guérin (BCG) can provide some protection against severe forms of pediatric TB, but its efficacy is variable, particularly against pulmonary disease in infants and adults. This review discusses the novel strategies being used to develop new TB vaccines, provides an overview of the candidates in clinical trials and outlines the challenges of introducing a new TB vaccine into existing childhood vaccination schedules.
Vaccination against TB: the current state of play BCG, an attenuated live form of M. bovis, has been in use since the early 1920s and is the only approved vaccine for the control of TB in humans. BCG can afford some level of protection, particularly against TB meningitis and severe forms of disseminated TB in children [3]. In a case-controlled trial in Argentina to determine the efficacy of BCG vaccination against TB in children under the age of 6, the vaccine afforded 98 % protection against TB meningitis and miliary TB [4]. Meta-analysis of trial data showed that BCG could prevent 73 % of childhood TB meningitis and 77 % of miliary TB cases disease in children [5]. However, BCG affords variable efficacy against pulmonary TB, the predominant form of the disease, with a median of 50 % protective efficacy [6]. Recent analysis suggests that BCG can, to some extent, protect against initial infection with M. tuberculosis, however this only occurs in a small proportion of exposed individuals [7]. Compounding this is the evidence from field trials that protective immunity afforded by BCG is not lifelong and wanes 10-15 years after vaccination [3]. Considering that TB is a chronic disease where more than 90 % of cases are in adolescents and adults, the limited longevity of the vaccine's protective effect is a significant barrier to TB control.
TB is a complex disease and M. tuberculosis has exquisitely adapted to life within the infected host. The disease is characterized by a latent phase, where the organism exists in a form that evades immune clearance, yet is unable to cause active disease [8]. An estimated 2 billion individuals are latently infected with M. tuberculosis, representing an enormous reservoir who may reactivate TB later in life. In humans and in animal models, BCG is unable to reduce latent infection and/or prevent reactivation, and as such the development of vaccines that can target latent bacteria would be a significant advance. In addition, new vaccines should induce the type of immunity proposed to protect against M. tuberculosis infection. As an obligate intracellular pathogen, a cell-mediated T-cell response is required to contain and ideally eliminate bacteria within infected host cells ( [9], Fig. 1). The generation of 'Th1-like' CD4 + T-cell secreting multiple cytokines are considered to be those required for optimal protective immunity, and most vaccines aim to generate these T-cells at high frequency.
In order to address the deficiencies of BCG, multiple strategies are being used to develop improved TB vaccine candidates. These include boosting BCG-induced immunity, modifying the existing BCG vaccine to improve its effectiveness, or replacing BCG with improved live vaccine candidates. The remainder of this review will describe the current TB vaccine candidates that have advanced to clinical trials, including those developed to prevent TB in children.
Boosting BCG: subunit vaccines M. tuberculosis antigens It is likely that BCG will remain part of current vaccine schedules, considering its good safety profile and its effectiveness against severe childhood forms of TB. For this reason there is particular interest in the development of subunit vaccines composed of protective M. tuberculosis antigens, as these may be used to 'boost' the protective response developed by BCG. A handful of vaccines have now entered clinical trials [10], however most of these vaccines make use of a small subset of related secreted antigens. This is despite the fact that M. tuberculosis expresses literally hundreds of antigenic targets, including non-secreted proteins [11] and no single mycobacterial antigen is recognised by all M. tuberculosis-infected individuals [12][13][14]. For this reason, all protein vaccines in clinical trials encode multiple antigens to expand the immune recognition of the vaccine (Table 1). In addition, the method of subunit protein delivery is critical for vaccine efficacy and considerable effort has been expended on the identification of potent adjuvants that are nevertheless sufficiently safe and well tolerated for human use [15].
The most advanced fusion protein candidate in terms of clinical development is M72/AS01. The vaccine is a  [16,17]. In a Phase I/ II trial the vaccine was well tolerated and induced high levels of polyfunctional M72-specific CD4 + T-cell and persistent antibody responses [17]. Numerous Phase II trials in adolescents [18], adults [19] or BCG-vaccinated infants [20] demonstrated the generation of polyfunctional CD4 + T-cells, augmented humoral responses and no adverse events associated with the vaccine. Importantly, immune responses and vaccine safety were not altered when the vaccine was delivered to infants together with Expanded-Programme-on-Immunization (EPI) vaccines [20]. The vaccine was shown to retain immunogenicity in HIV + subjects on anti-retroviral therapy [21] and has recently entered a Phase IIb proof of concept efficacy study in latently-infected adults (ClinicalTrials.gov Identifier: NCT01755598). A number of fusion protein vaccines, based on the secreted Ag85B protein of M. tuberculosis, are currently under evaluation in humans. Hybrid 1/IC31 comprises Ag85B fused with the early secretory antigenic target 6 (ESAT6), an immunogenic antigen absent from BCG [22]. The fusion protein is adjuvanted with IC31, a 2component adjuvant comprising an 11-mer antibacterial peptide (KLK) and a synthetic oligodeoxynucleotide (ODN1a), a Toll-like receptor 9 agonist [23]. In Phase I and II trials the vaccine has demonstrated strong, sustained generation of T-cell responses in naïve volunteers [24], individuals previously BCG vaccinated/M. tuberculosis infected [25] and HIV + individuals [26]. Hybrid 4/ IC31 (AERAS-404) includes the TB10.4 antigen instead of ESAT-6, due to the latter being a component of the Quantiferon Gold diagnostic test for M. tuberculosis infection. Hybrid 4/IC31 induced persistent polyfunctional CD4 + T-cell responses in adults [27] and the vaccine's ability to prevent infection with M. tuberculosis in adolescents is currently under evaluation (NCT02075203). Hybrid 56/IC31 (AERAS-456) is a modified version of Hybrid 1 that includes Rv2660c, an antigen strongly recognized by the immune response of latent TB patients [28]. The vaccine was shown to protect non-human primates against reactivation of latent M. tuberculosis infection [29] and was immunogenic when delivered to healthy adults with or without previous M. tuberculosis infection [30]. An ongoing Phase I trial will assess the safety and immunogenicity of Hybrid 56/IC31 in HIVnegative subjects recently treated for drug-susceptible pulmonary TB (NCT02375698) and in BCG-primed infants (NCT01861730).
ID93/GLA-SE is a fusion of four unique antigens (Rv2608, Rv3619, Rv1813, Rv3620) identified during a screen for human anti-mycobacterial T-cell antigens [31]. The vaccine is formulated with the TLR adjuvant glucopyranosyl lipid adjuvant-stable emulsion (GLA-SE) and has shown good protective efficacy in a number of pre-clinical animal models, including mice and guinea pigs, with the generation of polyfunctional T-cell subsets [32]. No clinical data on ID93/GLA-SE has been reported; however, the vaccine is undergoing safety/immunogenicity testing in healthy volunteers (Phase I, NCT01599897, NCT01927159) and in patients following successful completion of TB treatment (NCT02465216).

Boosting BCG: recombinant viral vectors
Viral vectors have been extensively studied as vaccines to control many pathogens, due to their ability to induce robust cellular and humoral immune responses [33]. The eradication of smallpox by the vaccinia virus has focused attention on the use of poxviruses as vaccine vectors, in particular modified vaccinia virus Ankara (MVA) [33]. MVA85A (AERAS-485), in which the M. tuberculosis Ag85A protein is expressed by MVA, was the first TB vaccine to enter human trials [34]. The vaccine, which has been tested in multiple Phase I/IIa studies in adults, adolescents, children and infants, was shown to be well tolerated and induce diverse vaccine-specific T-cell responses [35]. However, in a Phase IIb efficacy trial to test a BCG-prime, MVA85A-boost regimen in BCG-vaccinated South African infants, the vaccine did not provide improved protective efficacy against M. tuberculosis infection or disease [36]. An additional Phase IIb trial assessing the efficacy in healthy adults infected with HIV in South Africa and Senegal demonstrated significant T-cell responses induced by the vaccine but there was no improved efficacy against M. tuberculosis infection or disease in the MVA85A group compared to placebo [37]. Although the consistency of the pre-clinical animal data supporting this vaccine has been questioned [38], the vaccine has been instrumental in the development of large-scale clinical efficacy trials of vaccines against TB and developing protocols for defining correlates of TB vaccine-induced protection in humans [35]. Replication-deficient adenoviral vectors are an additional class of vaccine vectors that are being utilised for recombinant antigen delivery [33]. AdAg85A is an adenoviral serotype 5 vector expressing the M. tuberculosis Ag85A protein. In preclinical animal models the vaccine provided optimal protective efficacy when delivered mucosally, in particular the boosting of prior BCG immunization [39]. The vaccine has been tested in humans after intramuscular delivery, however the study was terminated for undefined reasons (NCT00800670). One important issue with adenoviral vectors is the observation that pre-existing Ad5 antibodies have been shown to correlate with Ad5-based HIV vaccine failure [40]. However, infants appear to have reduced levels of neutralizing Ad5 antibodies, which suggests these vaccines may be more suitable for pediatric populations [41]. Pre-existing vector-specific immunity has been overcome by use of chimpanzee adenoviruses (ChAds), with a number of ChAds in clinical trials [33]. This includes a ChAd vector expressing M. tuberculosis Ag85A (ChA-dOx185A), which is being assessed together with a MVA85A boost in adults (NCT01829490). AD35.TB-S (AERAS-402) is a replication-deficient serotype 35 adenovirus that encodes a fusion of 3 M. tuberculosis (Ag85A, Ag85B, TB10.4) and has been assessed in BCG-vaccinated infants and adults, as well as HIV + individuals [42][43][44]. The vaccine induced polyfunctional CD4 + /CD8 + T-cell and antibody responses to the encoded vaccine antigens, yet induced a modest level of Ad35 antibodies [42][43][44].

Improving BCG: recombinant forms of BCG
Despite the limitations of BCG described above, the vaccine does induce some level of protection against childhood forms of TB, and BCG vaccination can reduce mortality in children due to other diseases [45]. Therefore, a major focus of TB vaccine development programs is the development of modified forms of BCG to improve the protective efficacy of the vaccine (reviewed in [46]). Of the numerous recombinant BCG strains developed, only three have undergone human trials. rBCG30, which overexpresses the immunodominant Ag85B protein, demonstrated improved efficacy in M. tuberculosisinfected guinea pigs, in terms of reduced bacterial loads and improved survival compared to BCG-only vaccinated animals [47]. Although the vaccine was demonstrated to be safe and immunogenic in a Phase I clinical trial in adults, the vaccine is not being pursued further [48]. An alternative strategy was taken in the development of VPM1002, a recombinant BCG expressing the Listeria monocytogenes enzyme listeriolysin O (LLO) and deleted of the ureC gene, in order to facilitate LLO function. Listeriolysin perforates the phagosomal membrane, allowing leakage of enzymes and bacterial components into the cytoplasm and increased apoptosis of the infected cell, resulting in enhanced CD4 + and CD8 + T-cell responses [49]. The vaccine candidate has completed a Phase I trial in infants (NCT01479972), and is currently being assessed for safety and immunogenicity in HIV-exposed newborns (ClinicalTrials.gov Identifier: NCT02391415).
A dual strategy of antigen overexpression and immune modulation was employed in the development of AERAS-422, a BCG strain expressing the pore-forming perfringolysin of Clostridium perfringens and selected immunodominant antigens expressed by M. tuberculosis during active infection (Ag85A and Ag85B) and reactivation of latent infection (Rv3407) [50]. In a Phase I trial in adults, the vaccine induced strong antigen-specific T-cell responses; however, two of eight vaccinees developed varicella zoster virus (VZV) reactivation, resulting in the discontinuation of the vaccine's development [51].

Replacing BCG: live attenuated mycobacterial strains
An alternative approach to TB vaccine design is to use live mycobacterial strains to replace BCG in the childhood vaccine schedule. When compared to virulent M. tuberculosis isolates, more than one hundred genes are absent in the BCG genome [52], and a subset of these genes may be important protective antigens. Therefore, a rationally attenuated form of M. tuberculosis may more closely 'mimic' the immune repertoire generated by natural infection. MTBVAC is the first live-attenuated M. tuberculosis-based vaccine to undergo testing in humans. MTBVAC contains two independent stable deletion mutations in the virulence genes phoP and fadD26, without the inclusion of antibiotic resistance markers, thus fulfilling the second Geneva Consensus requirements for progression of live mycobacterial vaccines to human trials [53,54]. The vaccine afforded significant protection against M. tuberculosis in pre-clinical models [55] and induced markedly enhanced T-cell immunity compared to the BCG vaccine in mice [56]. In a Phase I clinical trial MTBVAC did not induce any serious adverse events and elicited the generation of polyfunctional CD4 + central memory T-cells in vaccinees [57]. Encouragingly, vaccine safety and immunogenicity is currently being assessed in newborns in a Phase I trial (NCT02729571).
Inactivated whole-cell mycobacterial strains are also being assessed, although mainly as post-infection/immunotherapeutic vaccines, with the aim of preventing reactivation and/or shortening the course of drugtreatment for TB. The immunotherapeutic potential of these vaccines has been reviewed elswehere [58] and will not be discussed here. One inactivated vaccine, termed Dar-901, is under evaluation as a preventative TB vaccine. Dar-901 consists of the heat-inactivated non-tuberculous M. obuense and is part of a Phase I clinical trial for safety and immunogenicity healthy adults (NCT02063555), and is currently recruiting for a Phase II trial as a booster to prevent TB in adoloscents (NCT02712424).

Challenges of TB vaccine development
A number of challenges remain before a new vaccine can be introduced to either complement or replace the existing BCG vaccine. Although there is an appreciation that multiple antigens should be included into subunit vaccine to broaden the immune response generated, many candidates use single antigens-in particular Ag85A, which in a recent study was not strongly recognized by the immune response of TB patients [14]. Unlike viruses, which tend express a limited number of antigenic targets, the antigen repertoire of mycobacteria is broad and includes poorly expressed and cryptic epitopes that may contribute to protection [59]. Therefore new vaccine candidates should ideally express a selection of antigens that are strongly recognised by the human immune response during all stages of the TB life cycle. In addition, assessment of antigen immune recognition has typically been performed in adult TB patients, rather than children [14] and it is possible that the infant immune system may display differential recognition of antigens. Compounding this is the fact that most clinical trials of TB vaccine have been performed in adolescent or adult populations (Table 1), thus limiting our knowledge of vaccine-induced immunity in children.
An additional consideration is the lack of defined immune correlates of protection against M. tuberculosis infection. Certain immune responses appear to correlate with protection against M. tuberculosis in animal models, such as CD4 + T-cell secreting mutiple cytokines, termed 'polyfunctional T' cells [60]. However, the presence of polyfunctional T-cells in either MVA85Avaccinated adults or BCG-vaccinated infants did not correlate with protection against TB in humans [36,61]. While this may be a function of the particular vaccine tested, it does suggest that a greater breadth of immune paramaters should be examined as potential 'biomarkers' of protection, including non-conventional T-cells and components of the innate immune response [9]. Importantly, while the mouse is the major animal model used for testing of TB vaccines, the relative contribution of immune cell subsets may differ between the mouse, humans and other models such as non-human primates [62]. This makes it difficult to extrapolate findings from animal models to humans. While a human challenge model for TB has been proposed [63], the curent model relies on use of the attenuated BCG vaccine as the challenge organism, which lacks important TB vaccine antigens, and further optimisation is required to overcome the low level of recovery of the challenge strain [64]. However, such a model would provide an important tool for TB vaccine research, considering the expense of large phase IIb and phase III clinical trials to determine vaccine efficacy for a chronic infection such as TB, together with the lack of markers of protective immunity in humans.

Conclusions
The past decade has seen major advances in the development of TB vaccines, with a number of vaccines now in clinical trials (Table 1) and one vaccine having completed Phase IIb assessment of efficacy [36]. However, no vaccine has demonstrated improved protective efficacy in humans compared to the existing BCG vaccine, and the immunological parameters required for effective protective efficacy in humans are not known. This suggests that innovative and novel TB vaccine approaches are required and these vaccines should elicit immune responses that differ from those candidates that have already been evaluated in humans. Effective control of TB transmission will require the delivery of a vaccine that can block/limit M. tuberculosis infection during initial exposure to the pathogen, which in endemic areas would be during early childhood, and thus the testing of new candidates in infant populations should be a priority.