Introduction

Tuberculosis (TB) is among the top ten causes of death overall and is the leading cause of death owing to infection with a single type of pathogen1. It is estimated that almost one-quarter of the global population, between 2 billion and 3 billion people, has been infected with Mycobacterium tuberculosis (Mtb) and may be at risk of progression to TB2. Up to 80% of the lifetime risk of progression to disease is thought to occur in the first 2 or 3 years after infection3. In 2017, an estimated 10 million people developed TB and 1.6 million people died from the disease1.

It is now appreciated that Mtb infection does not represent a single, uniform state and that the historical division of TB into either latent (latent TB infection (LTBI)) or active TB has gravely underappreciated the complex and dynamic nature of the host–pathogen interactions4,5,6. Progression from Mtb infection to clinical disease appears to transition via a number of continuous asymptomatic infection states that have previously been classified as LTBI, which include the asymptomatic states recently termed incipient TB and subclinical TB, before active, clinical TB disease manifests7,8. Further, within an individual host, Mtb-infected lesions in the lung or draining lymph nodes do not develop in a uniform synchronized manner but independently and therefore represent a spectrum of pathology that can span all stages from sterile, calcified granulomas through to caseous, necrotic lesions with exceedingly high bacterial burdens9. It is important to consider both the complexity of the human–Mtb interactions and the magnitude and diversity of the global epidemic as a backdrop to the challenges faced by vaccine development. Given that in some high incidence areas the majority of the adult population is infected, classic prophylactic vaccination (before any exposure to the pathogen) is applicable for only a proportion of the individuals in need. Therefore, TB vaccine development efforts are currently focused on developing vaccines for the following administration regimens: prophylactic vaccination, which is a vaccine administered to individuals in order to prevent Mtb infection or clinical disease (prophylactic vaccines can be either priming vaccines such as Mycobacterium bovis bacillus Calmette–Guérin (BCG) (see Box 1) used in neonates or booster vaccines for later administration); postexposure vaccination, which is a vaccine administered to Mtb-infected individuals to prevent the development of active disease (many priming and booster vaccines are currently developed for postexposure administration owing to their ability to boost and supplement the naturally occurring infection-promoted responses); and therapeutic vaccination, which is a vaccine administered to individuals with clinical disease in combination with or after antibiotic treatment to prevent recurrence of disease. Development of both prophylactic and postexposure vaccines is actively being pursued with several candidates in clinical trials, but therapeutic vaccines are also receiving increasing attention.

Table 1 Tuberculosis vaccine candidates that are currently in clinical trials or have recently completed clinical trials
Full size table

Five years ago, the results of a large phase IIb efficacy trial of the first TB booster vaccine candidate tested in infants were published. Vaccination of 4–6-month-old infants, who had received neonatal BCG vaccination, with MVA85A induced no additional protection against Mtb infection or active TB beyond that observed in the placebo arm of the study10. This result was a great disappointment to both the TB vaccine research community and funders and a call to action for researchers in basic and applied vaccine development. Today, we are witnessing immense progress in both preclinical and clinical TB vaccine research, including the first proof-of-concept study showing that revaccination with BCG can protect adolescents from sustained Mtb infection11 and that the subunit vaccine M72/ASO1E provides protection against the development of TB disease in Mtb-infected adults12. Here, we discuss recent breakthroughs in our understanding of the mechanism of protective immune responses, provide an overview of the vaccine candidates in clinical trials and discuss whether it is time to reconsider BCG revaccination as part of a future improved TB vaccine strategy.

Immune responses to Mtb

Mtb can establish infection in susceptible individuals after the inhalation of a single or a few bacteria that are taken up by alveolar macrophages. The pathogen has developed a refined set of evasion mechanisms that delay bacterial transport to the regional lymph node and allow it to evade host cellular immunity, giving it sufficient time to establish a productive infection13,14,15. The result is a delayed onset of the natural adaptive immune response observed both in animal models of TB16 and in human clinical TB17,18.

One of the primary roles of vaccination is to establish efficient and long-lived immune memory in order to shorten the interval between infection and the onset of an adaptive immune response at the site of infection, such that the infection can be controlled rapidly and spreading to secondary sites is avoided. In humans, the first exposure to Mtb typically occurs after the immune response has been primed by other mycobacterial encounters, either in the form of BCG vaccination or environmental mycobacteria. As a result, TB vaccine strategies should consider how prior induction of T cells (and other immune responses) by these exposures may influence the function, trafficking and survival of vaccination-induced responses and their effectiveness against Mtb. Furthermore, in settings with very high rates of Mtb infection, such responses must be able to resist the effects of repeated reinfection and long-term continuous antigen exposure from this chronic infection.

The role of CD4+ T cells

Recent data in animal models suggest that vaccine-induced CD4+ cells of the T helper 17 (TH17) cell subtype, which naturally traffic to the airways, can accelerate the recruitment of protective TH1 cells19,20,21. In fact, a recent study of a rhesus macaque model of pulmonary vaccination showed that vaccination with BCG induces pulmonary TH1 cells and/or TH17 cells, which co-express IFNγ, TNF, IL-2 and IL-17 and protect against infection upon repeat low-dose challenge with Mtb22. When CD4+ T cells arrive at the site of infection, they encounter aggregates of Mtb-containing macrophages and other immune cells and together form the tight cellular structure referred to as the granuloma. The CD4+ T cells secrete cytokines, which activate infected macrophages to control bacterial growth and attract more immune cells to the granuloma (reviewed elsewhere23).

Most vaccine research has focused on TH1 cells and the effector cytokine IFNγ as a readout for successful vaccination and a potential indicator of vaccine efficacy. However, it is clear from animal studies22,24,25 that expansion beyond the narrow focus on IFNγ is necessary to identify new biomarkers (also called immune correlates of protection (COP); see Box 2) to support vaccine evaluation and optimization. This is also supported by conflicting data on the role of IFNγ in human studies. A study of COP in the participants of the MVA85A phase IIb trial suggested that higher frequencies of BCG-reactive IFNγ-secreting cells, as quantified by ELIspot assay, were associated with a reduced risk of developing TB26. By contrast, the MVA85A booster vaccine referred to above induced long-lived CD4+ T cells that co-expressed IFNγ, TNF and IL-2 (ref.27), a functional subset of cells termed polyfunctional by many in the field, but this subset did not afford protection10. Similarly, in a study of 10-week-old infants that were vaccinated with BCG at birth, there was no association between the frequencies of BCG-reactive TH1 cells or the co-expression patterns of IFNγ, TNF and IL-2 of these cells and subsequent risk of developing TB10,26,27,28. Collectively, these studies suggest that TH1 cell responses are necessary but not sufficient to mediate protection against Mtb and that other functions and characteristics of T cells, and perhaps other arms of immunity, are involved in protection against TB.

Recently, the expression of CD153, a surface molecule of the TNF superfamily that is expressed by Mtb-specific CD4+ T cells during infection, was suggested as a promising marker of protection in animal models. CD153 was also found to be expressed by Mtb-specific CD4+ T cells in humans who successfully control Mtb infection29. This molecule is an example of the new candidate biomarkers of protective immunity that should be considered in analyses of immune COP in the context of vaccine trials.

The role of CD8+ cells

Mtb-specific CD8+ T cell responses increase during disease progression with kinetics that appear to positively correlate with the bacterial burden30,31, but their role in protective immunity to Mtb is unclear. Evidence that CD8+ cells may play a protective role comes from a non-human primate (NHP) study in which CD8+ T cells were depleted, which resulted in compromised BCG vaccine-induced immune control of Mtb32. However, recent data from murine models suggest that even very high numbers of vaccine-induced CD8+ T cells that are specific for antigens involved in protective immunity fail to recognize Mtb-infected macrophages or affect Mtb proliferation in animal infection studies33,34. Similarly, antigen-specific CD8+ T cells induced by an adenovirus-based TB vaccine in humans failed to recognize Mtb-infected dendritic cells in vitro35. Although the subject of numerous studies, the role of CD8+ T cell responses in protective immunity against TB therefore still remains unresolved.

The role of B cells

Antigen-specific antibodies, and their functional attributes in immunity against Mtb, have recently received increasing attention. Compelling evidence shows marked differences in the plasma levels of natural mycobacteria-specific IgG or IgA and their glycosylation profiles and Fc functions when comparing individuals with LTBI and patients with active TB36,37. Further, a recent study of a rhesus macaque model of pulmonary BCG vaccination showed that high levels of antigen-specific IgA in bronchoalveolar lavage fluid were associated with protection against Mtb infection and disease22.

Mtb escape mechanisms

In addition to delaying the onset of adaptive immune responses, Mtb has also evolved a set of escape mechanisms aimed at inhibiting CD4+ T cell activation during later stages of the infectious process38. Examples include the downregulation of certain target antigens to very low levels39 and the active transfer of immunodominant antigens to uninfected bystander dendritic cells and macrophages40. The outcome of this host–pathogen stand-off is bacterial survival in a state of latent infection. It is notable that established latent infection seems to provide significant protection against reinfection in classical epidemiological studies41,42 and in NHP models43, a phenomenon that provides evidence for protective natural immunity against Mtb. We note that it is possible that immune mechanisms necessary for protection against the establishment of Mtb infection may be different from those required for successful long-term containment of an established Mtb infection such that progression to disease is averted. Ongoing studies of vaccine-induced COP, described below, will provide important insights into this.

The diversity of the CD4+ T cell response to tuberculosis

For protection against TB, the CD4+ T cell subset is of major interest. CD4+ T cells differentiate into T central memory (TCM) cells that home to secondary lymphoid organs and, on the basis of their expression of adhesion molecules, most likely also to inducible bronchus-associated lymphoid tissue structures in the lung44. Upon antigen re-exposure, TCM cells differentiate into T effector memory (TEM) cells and effector cells of either the TH1 cell or TH17 cell lineage that migrate to and exert their effector functions in infected tissues (see Fig. 1). A proportion of these T cells subsequently remains in the lung as T tissue-resident memory (TRM) cells. An efficient frontline defence in the lung depends on both TRM cells localized in the lung before infection and newly recruited T effector (TEFF) cells that arrive after infection. However, for a chronic infection such as Mtb, the longevity of the immune response, that is, the ability to withstand the continuous exposure to antigen for very long periods without exhaustion, is likely of equal importance. This is where T stem cell memory (TSCM) cells and TCM cells play a central role because their proliferative potential can maintain the supply of tissue-homing T cells. Designing TB vaccine strategies therefore requires careful consideration of the distribution of different subsets of CD4+ T cells, how they respond to repeated antigen exposure during persistent infection and their ability to traffic to and be retained within the lung, both before and during ongoing Mtb infection.

Fig. 1: The balance between Mtb-specific T cell differentiation into either T effector memory or T central memory cells.
figure 1

After Mycobacterium tuberculosis (Mtb) infection of alveolar macrophages (red arrows), Mtb is transported to lung-draining lymph nodes by infected dendritic cells, initiating T cell priming or triggering the activation of pre-existing memory T cells, which appears to preferentially drive T cell differentiation towards late-stage T effector memory (TEM) cell and T effector (TEFF) cell responses. Ongoing antigen expression is likely a driver of this T cell differentiation process, which favours primarily IFNγ -expressing and/or TNF-expressing T cells and little IL-2 expression. Vaccine administration in the skin or muscle promotes antigen uptake by dendritic cells, which traffic to draining lymph nodes to prime or activate T cells (blue arrows). In the case of subunit protein-adjuvant vaccines, the resulting T cell responses appear to be dominated by less differentiated T central memory (TCM) cell responses; these cells primarily express IL-2 and/or TNF. Achieving long-lived protective immunity by vaccination may require the establishment of a careful balance between TCM cell and TEM cell responses, such that a durable pool of memory cells resides in secondary lymphoid tissues while an appropriate tissue-resident population with rapid effector function is maintained in peripheral airway tissues.

Full size image

In humans, Mtb infection promotes the development of CD4+ T cells that span a range of differentiation states, from the most early TSCM CD4+ cells45 through to fully differentiated TEFF cells that predominantly express IFNγ46,47,48. Many factors are likely to determine where in this range a given specific T cell response will lie, including the expression level of the particular Mtb antigen that is targeted, the stage of disease progression or host–pathogen interaction and the location of the T cell itself. In order to allow a conclusive interpretation of results, it is therefore essential that human studies of T cell function and differentiation clearly and carefully characterize the clinical phenotype of study participants to define infection stage. Overall, an increasing mycobacterial load correlates with progressive differentiation of Mtb-specific CD4+ T cell responses away from TCM cells that secrete IL-2 and towards TEFF cells that secrete predominantly IFNγ46,47,48,49. In animal models of Mtb infection, ongoing antigen exposure is a significant challenge for the host immune system and results in CD4+ T cell exhaustion through inhibitory receptors such as TIM3 (also known as HAVCR2)50 and the upregulation of exhaustion markers such as killer-like lectin receptor G1 (KLRG1)51. This results in the loss of self-renewing TCM cell subsets and in functional impairment, eventually resulting in uncontrolled growth of Mtb52. In TB vaccine studies in the mouse model, it has become clear that T cell responses promoted by an adjuvanted vaccine formulation typically differ from Mtb-induced T cell responses in that they preferentially induce IL-2-producing TCM cells. Compared with TEM cells and TEFF cells induced by continuous exposure to Mtb, these TCM cells are less likely to become terminally differentiated, predominantly IFNγ-secreting KLRG1+ T cells (Fig. 1).

Such vaccine-induced T cells therefore have the desirable ability to resist terminal differentiation, which would eventually result in functional impairment and depletion of the Mtb-specific T cell pool48,49,52. Animal models of TB that investigated T cell differentiation have demonstrated that BCG vaccination, similarly to Mtb infection, pushes T cell differentiation towards TEFF cells, which results in a failure to efficiently maintain long-term protection against Mtb53,54.

Initially, the main role of less differentiated CD4+ T cells (such as TCM cells and TSCM cells) was thought to relate exclusively to their ability to resist differentiation, replenish TEM cells and maintain long-lived memory both before and after infection. Recent insights into T cell migration patterns have added important new facets to this interpretation. Using an intravascular staining technique, a less differentiated memory subset of T cells, which expresses the checkpoint molecule PD1 and the chemokine receptor CXCR3, was shown to enter the Mtb-infected lung parenchyma55. By contrast, the more differentiated TEFF cell subsets, characterized by the expression of KLRG1 and the fractalkine receptor CX3CR1, lose their ability to enter the lung parenchyma and to protect against Mtb55. The negative influence of a highly differentiated and strongly TH1 cell-polarized response on protection against Mtb is further supported by recent data from experimental animal models that investigate this question from different angles56,57,58.

Choosing the best antigens

Two critical and as yet unresolved questions in TB vaccinology are how to select the best antigens and how many antigens to include in a vaccine. The retrospective analysis of the phase IIb clinical trial of the MVA85A booster vaccine, which contains a single antigen (Ag85A), highlights this question59,60. Given that earlier studies had shown that Ag85A is expressed at only low levels during chronic Mtb infection in mouse models61,62, it is possible that the disappointing results were due to poor antigen choice61,62. Some of the most vaccine-relevant T cell antigens are virulence factors such as those associated with the ESX1 protein secretion system, which is instrumental for pathogen survival and is expressed at high levels in vivo during both the acute and chronic phase of infection61,62,63. Although one might assume that this high level of expression might indicate suitability as an immune target, immune responses to highly expressed antigens may eventually become exhausted during chronic infection. An unanswered question is whether repeated exposure or reinfection in humans who live in high-transmission settings also drives T cell exhaustion. Interestingly, most of the immunodominant antigens are conserved, with minimal sequence variation among different clinical strains of Mtb, indicating that the induction of T cell exhaustion may be an integral part of the Mtb survival strategy64,65,66. In a direct comparison between T cell responses specific for the ESAT6 antigen (which shows high and continuous expression during Mtb infection) and Ag85B (which is primarily expressed early during infection), the different expression profiles were found to have a profound influence on the quality of the immune response. Ag85B-specific T cells developed into classical CCR7+KLRG1− TCM cells after primary Mtb infection in mice, whereas ESAT6-specific T cells were maintained in an effector state and gradually increased their expression of KLRG1 and lost IL-2 expression54,63. Similar differences between the differentiation state and functional capacity of human ESAT6-specific and Ag85B-specific CD4+ T cells were also observed in healthy trial participants with LTBI who were vaccinated with the H1:IC31 or H56:IC31 vaccines, which contain both antigens63.

These findings are in agreement with previous reports of the influence of antigen persistence and load on the distribution of different subsets of human memory T cells in clinical studies of other infections or vaccines67. A vaccine that provides a single exposure to an antigen followed by antigen clearance, such as the tetanus vaccine, primarily induces IL-2-producing TCM cells, whereas pathogens that cause contained infections that provide low but persistent antigen exposure, such as herpes simplex virus 1, induce TEM cells that express IFNγ, TNF and low levels of IL-2 (ref.67). TEFF cells that express only IFNγ on the other hand are induced by infections such as cytomegalovirus that provide high and persistent antigen exposure67.

Striking the balance to achieve longevity and efficacy

As discussed above, excessive induction of TEFF cells by a vaccine modality or its antigen components may lead to impaired maintenance of memory and functional ability of the immune response, and, in the most extreme case, with very high expression of effector cytokines, such as IFNγ or IL-17, the result can be immunopathology68,69,70,71. TB vaccine strategies therefore need to strike an optimal balance between the self-renewing TSCM cell and TCM cell subsets and the more differentiated TEM cell and TEFF cell subsets that provides an efficient first-line defence in the lung (Fig. 1). This is complicated by the large proportion of individuals with LTBI (in some high-endemic regions more than 50% of the population)2 and extensive reinfection in high-transmission settings. Individuals with LTBI have an already established T cell response and represent a challenging population in which vaccine modalities and doses intended for initial priming in naive hosts may be suboptimal72. A recent analysis of the literature on the pathogenesis of human TB before antibiotics were introduced furthermore suggests that immune responses required to prevent progression to reactivation TB (that is, progression of established LTBI to active disease) are likely to be different from those required to control or prevent the establishment of the primary infection9. This follows from the argument that, in the case of progression to reactivation TB, a successful immune response would need to protect against tissue damage and cavitation. A recent interim analysis of the ongoing phase IIb trial of the subunit vaccine M72/ASO1E as a postexposure vaccine, which showed a vaccine efficacy of 54% against progression to TB relative to placebo12, clearly shows that efficacious vaccine modalities in pre-sensitized populations are possible. Delineating the functional, phenotypic and differentiation characteristics of T cell responses induced by the two antigens in this vaccine (Mtb32A and Mtb39A) will be critical for our understanding of immune COP against TB.

Tuberculosis vaccines in clinical trials

The development of an efficacious TB vaccine strategy relies on a healthy pipeline of TB vaccine candidates that represent a diverse repertoire of formulations and mycobacterial antigens and that induce a broad range of immune responses with different characteristics. Eleven TB vaccine candidates (Table 1) are currently in clinical testing for prophylactic, postexposure or therapeutic indications.

Whole cell vaccines — live

Live, attenuated whole cell vaccines were initially developed as prophylactic, priming vaccines with the aim to replace BCG-prime vaccination in infants, but they are now also being assessed as postexposure vaccines in adolescents and adults. Two of these vaccines, the recombinant BCG vaccine VPM1002 and the live, attenuated Mtb vaccine MTBVAC, are currently in clinical trials. Both induce a complex and diverse immune response to many antigens, which may offer an advantage over subunit vaccines that have a response restricted to a few antigens. However, such live vaccines are likely to be subject to the same interference caused by prior immunological sensitization by non-tuberculosis mycobacteria (NTMs) as reported for BCG. VPM1002 is also being assessed as a postexposure vaccine for the prevention of recurrence of active TB.

Whole cell vaccines — inactivated

On the basis of a classical vaccine development paradigm, these products, which include RUTI, Mycobacterium vaccae-based vaccines and the Mycobacterium obuense-based DAR-901 vaccine, utilize killed whole mycobacterial cells or mycobacterial cell extracts to safely induce complex immune subsets against multiple Mtb antigens. RUTI and M. vaccae-based vaccines are primarily being pursued as therapeutic vaccines, while DAR-901 is being developed as both a prophylactic and a therapeutic vaccine.

Adjuvanted protein subunit vaccine

Subunit vaccines are based on protein antigens administered with adjuvants. These are primarily developed as prophylactic or postexposure vaccines that boost responses that were initially primed by BCG or Mtb infection for preventing the establishment of Mtb infection, active TB or recurrent disease. Subunit vaccines that are currently in clinical testing include H4:IC31, H56:IC31, ID93 + GLA-SE and M72/AS01E. Some of these vaccines are also tested as therapeutic vaccines (for example, H56:IC31 and ID93 + GLA-SE) to prevent recurrence in patients who have completed chemotherapy for active TB.

Viral vectored vaccines

Live, attenuated, non-replicating viruses can be engineered to deliver genes encoding the antigens of interest into host cells. Such vaccines allow for the intracellular production of the antigen in vivo and activate cells of the innate immune system and therefore do not need to be adjuvanted. Viral vectored vaccines are being developed as both prophylactic vaccines and postexposure vaccines that boost responses primed by BCG or Mtb infection. A potential problem with viral delivery is the induction of vector-specific immunity that can interfere with subsequent booster vaccinations. Two viral vectored TB vaccine candidates, MVA85A and Ad5Ag85A, are currently in clinical testing in prime–boost combinations, including trials of the MVA85A candidate administered by aerosol to the airways.

BCG revaccination — time to reconsider?

Revaccination with BCG at different ages, but primarily in children, was practised for decades in several countries with limited evidence for its clinical value or cost-effectiveness73. Two large cluster-randomized controlled trials conducted in Brazil and Malawi evaluated BCG revaccination for the prevention of TB disease. Neither demonstrated efficacy74,75, resulting in World Health Organization (WHO) recommendations against this policy. Subsequent follow-up studies conducted as part of the REVAC study in Brazil have added a new layer of understanding to this disappointing result and suggest that prior mycobacterial sensitization is a major factor in preventing BCG revaccination efficacy in regions with a high prevalence of environmental mycobacteria exposure76 (Fig. 2). In the recent trial with the H4:IC31 subunit vaccine or BCG revaccination in Cape Town, South Africa, BCG revaccination provided significant protection (45% efficacy) against sustained Mtb infection (measured as prevention of sustained IFNγ release assay (IGRA) conversion for a follow-up period of 24 months in healthy adolescents who received a BCG prime in infancy)11. The readout of prevention of sustained IGRA conversion may be interpreted in different ways, which are all indicative of protective immunity: the prevention of primary infection, the accelerated clearance of the bacilli after infection, the long-term containment of the primary infection below the IGRA cut-off level or even the prevention of reinfection during the observation period. Regardless of the mechanistic interpretation of this readout, BCG revaccination resulted in a surprisingly high efficacy signal in this study. A plausible explanation for this observation most likely relates to the fact that individuals with Mtb-specific responses at enrolment, as measured by IGRA, were rigorously excluded11,77 and that Cape Town is thought to be an area with low to modest NTM exposure levels78. Compared with the REVAC study in Brazil, the main source of potential sensitization in this trial population therefore most likely came from remaining immune responses to neonatal BCG vaccination 12–16 years previously. So, even though pre-existing mycobacteria-specific T cell responses were found to be common in trial participants in the H4:IC31–BCG revaccination study, the results suggest that, without continuous high-level exposure from environmental mycobacteria or LTBI, the levels of immunity that persist from neonatal BCG vaccination are modest and do not block the efficacy of BCG revaccination significantly in adolescents (Fig. 2).

Fig. 2: The effects of immunological sensitization to mycobacteria on vaccine efficacy.
figure 2

The hypothesized interaction between the magnitude of immune sensitization and vaccine efficacy by Mycobacterium bovis bacillus Calmette–Guérin (BCG) vaccination (red), adjuvanted protein subunit vaccines (orange) or a BCG-prime, subunit-boost strategy (blue) is shown. According to this model, BCG, and other live whole mycobacterial vaccines, are not efficacious in individuals with substantial prior immunological sensitization owing to latent tuberculosis infection (LTBI), recent BCG vaccination or exposure to atypical non-tuberculous mycobacteria (NTMs) from the environment. Subunit vaccination, by comparison, would be efficacious in such a pre-sensitized population, as would a BCG-revaccination, subunit-boost strategy. In persons with low or no mycobacterial sensitization, the efficacy of BCG vaccination is significantly increased, and the efficacy of subunit vaccination will be largely independent of the levels of sensitization, but the BCG-revaccination, subunit-boost strategy may provide synergistic effects that result in enhanced vaccine efficacy.

Full size image

This efficacy signal therefore opens up consideration of BCG revaccination in certain settings as part of an overall improved TB vaccination strategy as previously suggested78. Samples collected in the H4:IC31–BCG revaccination trial provide a unique opportunity to investigate whether the level of immune sensitization to NTMs at enrolment is associated with differences in the quality of the immune responses to the vaccine and/or protection against Mtb infection.

Building on recent subunit vaccine success

For the first time since the introduction of universal BCG vaccination into the WHO expanded programme for immunization in 1974, we have encouraging efficacy signals in trials of a TB vaccine11,12. In a phase IIb trial of healthy adolescents who received BCG in infancy but tested negative in the IGRA, vaccination with H4:IC31 was associated with 30.5% efficacy against sustained Mtb infection. This result was of statistical significance at the pre-defined statistical threshold of an 80% CI (3.0–52.0%), but not at a more rigorous 95% CI (–15.8% to 58.3%)11. Moreover, interim results of the ongoing phase IIb trial of M72/AS01E, conducted in 3,575 IGRA-positive adults, clearly illustrate that booster vaccination with a subunit vaccine can protect against TB disease in highly sensitized individuals with Mtb infection. This trial, which was conducted in HIV-negative adults from Kenya, Zambia and South Africa, most of whom received neonatal BCG, accrued 10 patients with microbiologically confirmed pulmonary TB into the M72/AS01E group and 22 patients into the placebo group, translating to an overall vaccine efficacy of 54.0% (95% CI 13.9–75.4%)12. An important consideration when interpreting the trial results of M72/AS01E is how the level of Mtb exposure and reinfection of trial participants may have influenced the efficacy of the subunit vaccine. A recent analysis of the incubation period of TB suggested that the vast majority of clinical TB cases occur early, within 1–2 years following Mtb infection or reinfection. This is different from reactivation of Mtb, which typically happens much later after a long period of latent infection3. The recent efficacy trial was conducted in settings with high Mtb infection rates where reinfection with Mtb is likely to be a frequent occurrence. Understanding whether the vaccine protected against reactivation disease or progression to disease following reinfection is likely to be important and will no doubt be the subject of investigation in coming years.

Whether the levels of efficacy observed in the H4:IC31–BCG and M72/AS01E trials are sufficient and persist long enough for programmatic implementation of the vaccines in their present form is an important question, especially because the reports from each trial followed trial participants for only 2 years. However, combined into a prime–boost strategy with BCG vaccination, novel subunit vaccines may promote a robust response that could significantly add to the efficacy of the BCG vaccine and compensate for its failures in sensitized populations. In accordance with the model in Fig. 2, induction of immunity by BCG is either blocked or masked by high levels of pre-existing T cell-mediated immune responses. By contrast, comparison of H56:IC31 or M72/ASO1E vaccination in naive versus LTBI individuals indicated that these subunit vaccines can markedly boost BCG-induced and Mtb-primed immune responses79,80,81. Therefore, a combined BCG revaccination and/or subunit vaccine strategy may have great potential in adult and/or adolescent populations.

Conclusion and future perspectives

The large number of different vaccine candidates and their advanced stages in clinical development denote a unique and exciting phase in TB vaccine research. There are also a large number of novel vaccine candidates in preclinical development, including more recently developed vaccine formats such as DNA vaccines, new adjuvants and delivery systems and combination vaccines. It is important that the most promising of these new candidates are advanced to efficacy studies in animals and clinical testing to augment the pipeline of TB vaccine candidates and concepts.

However, a notable limitation of the current clinical development landscape is a lack of inter-trial harmonization or standardization, which precludes a direct comparison of the immunological outcomes of different TB vaccine candidates. A recent analysis attempted to tackle this problem by comparing antigen-specific CD4+ and CD8+ T cell responses induced by BCG and six of the novel TB vaccine candidates, including MVA85A, AERAS-402, H1:IC31, H56:IC31, M72/AS01E and ID93 + GLA-SE. The investigators retrieved published data on antigen-specific T cell responses from clinical trials completed in adolescents or adults at a single trial site in South Africa82. The results show that the magnitude of vaccine-induced TH1 cell-polarized CD4+ T cell responses measured several months after vaccination was the T cell response feature that diverged the most between the different candidates. Unlike the response magnitude, co-expression profiles of IFNγ, TNF and IL-2 by CD4+ T cells suggested a relative lack of functional diversity in responses induced by the different vaccine candidates (see Table 1). Interestingly, the analyses suggested that M72/AS01E induced the highest antigen-specific memory CD4+ cell response among the candidates. Unfortunately, the study did not include results from whole cell or live vaccine candidates, which are known to induce a more diverse and broader repertoire of immune responses.

Overall, the recent positive clinical trial data referred to above represent a very important milestone in international efforts to develop a novel efficacious TB vaccine. These successes illustrate that TB vaccine research is on the right track and will be able to deliver a much-needed improved vaccine strategy that is so critical for controlling the global TB epidemic. It is critical that the field moves forward with urgency towards phase III licensure trials so that an impact on the epidemic can be achieved swiftly. These findings therefore signal the end of the ‘post-MVA85A period’ of fundamental doubts about both the usefulness of the TB vaccine research strategy and the TB animal models in vaccine discovery83. The M72/ASO1E vaccine contains only two antigens, and the next generation of vaccines may be improved by adding more antigens to increase immune coverage and avoid the risk of escape. The efficacy signal observed with M72/ASO1E will likely also establish the M72/ASO1E-induced protection as a minimum benchmark in preclinical animal models. As discussed above, many subunit vaccine candidates appear to induce a response that is typically characterized by early differentiated CD4+ TCM and TEM cells, whereas it seems that both viral and live mycobacterial vectors promote a more differentiated CD4+ TEFF cell response27,52,54,82,84,85. It will therefore be important to agree on a standard set of parameters that would allow an accurate comparison between studies and vaccines to determine whether this is a reproducible pattern in clinical trials. Recent results of preclinical studies using a recombinant human cytomegalovirus encoding several Mtb T cell antigens have shown impressive protection in an NHP model, where prophylactic vaccination prevented infection in one-third of experimentally infected rhesus macaques86. Because cytomegalovirus vectored vaccines establish a persistent lifelong infection and induce a high level of antigen-specific CD4+ and CD8+ T cell responses, these findings suggest that, in addition to T cell differentiation as discussed above, the sheer size of the pool of Mtb-specific CD4+ T cells has an impact on protection. This is also supported by the M72/ASO1E trial results, as this vaccine promotes a high frequency of antigen-specific T cells. However, more is not always better, and there is a risk of less protection and insufficient immune memory or even of inflicting immunopathology with vaccines that induce very strong T cell responses, particularly when used in the postexposure setting in individuals with LTBI. With an efficacy signal in young adolescents and/or adults from both BCG revaccination and subunit vaccine studies, it is intriguing to speculate whether the combination of both, administered sequentially or simultaneously87, may pave the way for a vaccination strategy that protects both uninfected and infected people while providing the possibility of a synergistic effect for inducing more diverse and broader-ranging immune responses. If an additive effect can be demonstrated, the combination of BCG and subunit vaccines may represent a new strategy that elevates the efficacy signal into a range that triggers clinical implementation.