Translational Mini-Review Series on Vaccines:
Development and evaluation of improved vaccines against tuberculosis


  • Guest Editor: Danny Douek

Clare Sander, Centre for Vaccinology and Tropical Medicine, Churchill Hospital, Old Road, Headington, Oxford OX3 7LJ, UK. E-mail:


Tuberculosis (TB) continues to be a major global health disaster, despite the widespread use of BCG and effective drug therapies. The development of an efficacious new TB vaccine would be an important component of disease control in the future. Many approaches are being utilised to enhance understanding of the requirements of a successful vaccine. Numerous vaccines are being designed and assessed in a series of animal models, with a few progressing to clinical trials. Here, the steps involved in the development and evaluation of TB vaccines will be discussed, including description of the most frequently used animal models and the processes involved in advancing vaccines to phase III trials.

Global need

It is estimated that 2 billion people, one-third of the world's population, are infected with Mycobacterium tuberculosis and it causes 2 million deaths each year [1]. Of these deaths, 98% occur in the developing world, where the disease predominates. The burden of disease falls on young adults, consequently TB is having a devastating effect on the economies of countries overwhelmed by this pathogen [2]. Although South-East Asia remains the region with the highest number of cases, the area with the highest rate of increase is sub-Saharan Africa. Much of this is fuelled by the HIV pandemic [3]. Being dually infected with human immunodeficiency virus (HIV) and TB worsens outcome from both diseases and TB is the leading cause of death among HIV-positive individuals [4]. Rates of TB are also rising in Eastern Europe, a region which has one of the highest incidences of drug resistance. This is largely the consequence of the breakdown of the infrastructure for TB control [3].

The World Health Organization (WHO) launched a strategy for gaining control over TB in 1995, known as Directly Observed Therapy Short course (DOTS), which aimed to reduce the spread of infection by diagnosing and treating infectious cases. Since its conception, the strategy has evolved. It now utilizes different providers for drug administration and encompasses issues of multi-drug resistance and HIV co-infection into its policies. Between 1995 and 2003, the estimated number of smear positive cases treated under DOTS increased fourfold, with an estimated treatment efficacy of 82%. However, this was thought to account for only 45% of the total smear-positive cases [5]. Consequently there is increasing recognition that a broader approach needs to be adopted to combat the scale of the problem. The STOP TB partnership announced a new global plan to control TB at the beginning of 2006, emphasizing the additional importance of researching new diagnostic, vaccination and therapeutic tools for the future [6]. There are clearly many advantages to disease prevention, and vaccination is known to be a cost-effective strategy [7]. Fortunately, with an increasing awareness of the global nature of the disease and the increasing concern over multi-drug resistance, increased funding is becoming available for TB vaccine research and progress is being made. However, the challenge of working with TB means that answers do not come quickly.

Can a vaccine work, and at what stage of the disease process would it intervene?

Among immunocompetent people, only 30% of individuals exposed to M. tuberculosis become infected with bacteria and of these only 10% go on to develop disease in their lifetime [8,9]. Although the innate immune response can account for much of the early resistance, later protection is the consequence of an adaptive response. This strongly suggests that immune control is feasible and is the premise on which TB vaccinology is based. Vaccination could intervene at a number of time points in the TB disease process. An effective pre-exposure vaccine could reduce the number of individuals infected with bacteria or could modify the immune response to further reduce the chance of later reactivation. In areas of high TB endemicity such a vaccine would have to be administered early in life to be maximally effective. A post-exposure vaccine, to be used in healthy individuals infected with M. tuberculosis, could reduce the probability of going on to develop TB disease. It could work by sterilizing bacteria that are residing in a dormant state, by preventing reactivation and/or by reducing the chance of reinfection by exogenous M. tuberculosis. Finally, a therapeutic vaccine could function alone, or more probably alongside antibiotic regimens, for individuals with active TB disease and could potentially shorten the treatment period. Although it is known that mycobacteria change their transcription patterns during different clinical conditions [10–12], it remains unknown whether vaccines designed primarily as pre-exposure vaccines could have an effect in the infected or active disease setting or vice versa. The majority of research to date has been in the development of vaccines designed primarily as pre-exposure vaccines, and work in this area is the subject of this review.

What do we want a vaccine to do?

Understanding natural immune correlates, or biomarkers, of protection is important both in guiding vaccine design but additionally in determining potential early markers of efficacy for vaccines that reach clinical trials.

M. tuberculosis is an intracellular pathogen; consequently the cellular arm of the immune system, particularly T helper 1 (Th1) cells, is pivotal to host defence. The increased rate of TB reactivation seen in HIV-positive individuals illustrates the important role played by CD4 T cells in protective immunity against mycobacteria. This is emphasized by the fact that vulnerability to TB disease increases with falling CD4 counts [13]. Other human models of M. tuberculosis infection have suggested that interferon (IFN)-γ is a central player. Increased IFN-γ production is seen in pleural lymphocytes from patients with TB pleuritis [14,15] and from peripheral blood mononuclear cells (PBMCs) from latently infected individuals [15,16]. Furthermore, increased susceptibility to recurrent mycobacterial infections has been seen in individuals with rare Mendelian abnormalities, such as complete or partial IFN-GR1 [17], interleukin (IL)-12 R deficiency [18,19] or complete IFN-GR2 deficiency [20] again pointing to the role of the IFN-γ/IL12 pathway. Finally, extensive animal data support the central roles of CD4 T cells, IFN-γ and IL-12 [21–27].

Other essential components of the host's natural response to M. tuberculosis infection have been, or are being, deciphered by a number of different experimental approaches.

Animal models

Knock-out and transgenic mice have demonstrated the importance of major histocompatibility complex (MHC) class I and CD8 T cells [27–30], in addition to cytokines such as tumour necrosis factor (TNF)-α[31,32], IL-6 [33], IL-7 and IL-15 [34]. New data suggest that the IL-23/IL-17 pathway may also play a role [35]. Excess immune regulation can prevent eradication of infection, although it may be important in preventing collateral pathology. Th2 cells, T regulatory cells (Treg) and/or immunosuppressive cytokines such as IL-10 or transforming growth factor (TGF)-β can all act as immune regulators. However, there is no animal data supporting the theory that increased levels of IL-4 [36,37] or IL-10 enhance the likelihood of active infection [36,38], although IL-10 may play a role in reactivation [39]. In contrast, TGF-β has been shown to suppress mycobacterial immune responses in a guinea pig model [40] and forkhead box P3 (FOXP3)+ CD25+ Treg have been implicated in impairing protective immunity in mice [35].

Immune responses in TB patients

Using blood and bronchoalveolar lavage samples from TB patients has confirmed the importance of immune regulation in the human TB disease process. TB patients have been shown to have an expansion of Treg among PBMCs which can suppress Th1 responses [41]. Furthermore increases in IL-10 and TGF-β, have been detected in bronchoalveolar lavage of TB patients, suggesting that down-regulation of the effector immune response can result in TB reactivation [42]. The importance of other cytokines in natural resistance against M. tuberculosis has been revealed, with the introduction of new therapeutic agents. For example, anti-TNF-α therapy, used in the treatment of rheumatoid arthritis and other autoimmune disorders, has been associated with reactivation of TB disease, highlighting the importance of TNF-α in host defence [43].

Cross-sectional studies

The development of new tools to diagnose TB infection has facilitated the design of large studies to determine natural immune correlates of protection. Until recently, only the tuberculin skin test was available, with limitations insensitivity and specificity [44]. However, following the sequencing of the M. tuberculosis genome [45], antigens from the region of difference (RD1), such as early secretory antigenic target 6 kDa protein (ESAT-6) and culture filtrate protein-10 (CFP-10), which are absent from bacille Calmette–Guérin (BCG) and most environmental mycobacteria, have been identified. Tools based on the in vitro measurement of IFN-γ to these antigens have been developed [45–47]. Two European Union-funded cross-sectional studies among different African populations have compared the host immune responses in TB patients with household controls or in those with latent infection. Reverse transcription–polymerase chain reaction (RT-PCR) of PBMCs has been used to study Th1, Th2 and other biomarkers of disease susceptibility. These studies have suggested that a shift from Th1 to Th2 is associated with TB disease and that suppression of Th2 is important for TB control [48–50]. Further intensive investigation of other correlates of protection is being planned through a Gates-funded Grand Challenge project, co-ordinated by the Max Planck Institute. This multi-centre study aims to identify molecular and cellular responses of pathogens and patients which correlate with protection [51].

We already have a vaccine, why do we need another?

The BCG vaccine, developed from an M. bovis strain that had been subcultured 230 times, was first administered in 1921. It is now used extensively worldwide, with recent estimates of 76% of children being vaccinated [7]. A recent meta-analysis revealed an efficacy of 73% and 76% efficacy against childhood TB meningitis and miliary TB, respectively, in the first 5 years of life and showed BCG vaccination to be highly cost-effective [7]. BCG is also known to provide protection against leprosy [52] and possibly other causes of childhood mortality [53,54]. However, the greatest burden of TB is an adult pulmonary disease and the efficacy of BCG in this group is much more questionable. A meta-analysis revealed efficacies ranging from 0 to 80% in this population [55]. Importantly, the trials that took place nearer the tropics, such as Chingleput and Malawi, revealed efficacies of 0%. Many explanations have been proposed to explain this variability, including issues of genetics, cold-chain and vaccine strain. However, the hypothesis with the greatest support is that increased exposure to environmental mycobacteria can interfere with the apparent efficacy of BCG. There is evidence for two mechanisms by which this inhibition could occur. Environmental mycobacteria themselves can induce some protective immunity and thereby mask the effect of BCG at a population level [56], but they can also induce an immune response that prevents BCG replicating and inducing a protective immune response [57,58].

How does BCG protect?

In addition to enhancing knowledge on natural host immunity, increasing the understanding of mechanisms of protection induced by BCG and reasons for its failure may give further clues for potential mechanisms for new vaccine candidates to target.

Current understanding of the mechanisms of BCG-induced protection in humans and in animal models is poor.

Animal models

In mice, BCG vaccination has been shown to induce CD4 IFN-γ producing T cells, which results in an accelerated response to TB infection and the more rapid development of granulomas [59]. In this model, BCG does not prevent infection; instead it contains infection, preventing bacillaemic spread and reducing the overall bacterial burden. In the guinea pig, BCG prolongs survival but the mechanism by which this is achieved remains to be elucidated.

Household contact study

Studies investigating the effect of BCG on the immune response in humans are increasing. Until recently, the consensus had been that BCG had no effect on preventing M. tuberculosis infection, but did prevent dissemination of infection from the lung, accounting for its efficacy in preventing severe childhood forms of TB, TB meningitis and miliary TB. However, recent evidence from a household contact study in Turkey, which identified infected cases by positive responses in a T cell assay to ESAT-6 and CFP-10, suggested that BCG vaccination can prevent infection [60].

Cross-sectional studies

Comparative studies have investigated the induction of IFN-γ responses to purified protein derivative (PPD) following BCG vaccination in Malawi and in East London. These have revealed a greater enhancement in IFN-γ response in UK teenagers than in Malawi, with baseline PPD levels being higher in Malawi. It has been suggested that this reflects enhanced exposure to mycobacteria in Malawi, whether that be environmental mycobacteria or M. tuberculosis[61]. The reduced IFN-γ levels induced by BCG in Malawi may account directly for the 0% efficacy of BCG in this area, or alternatively could reflect a reduced induction of other protective immune mechanisms. It is known that BCG can induce both antigen-specific CD4 T cells and CD8 T cells, the latter in much lower frequency [62]. In vitro work has demonstrated that BCG-vaccinated T cells can inhibit mycobacterial growth killing compared to non-vaccinated controls [63].

Case–control study

An extremely important study, led by the University of Cape Town and sponsored by Aeras Global TB Foundation, is studying immune correlates of protection in infants after BCG vaccination. A total of 5675 infants have been vaccinated with BCG. Blood samples were taken 10 weeks after vaccination and are being stored for wide-ranging immune and molecular immune assays. Community follow-up has enabled cases and matched controls to be identified, and will hopefully allow the identification of potential immunological correlates of protection. The range of tools and optimized assays that are being used make this a very powerful study and the study design is likely to be used as a model for the investigation of correlates of protection for newly designed TB vaccines [64,65].

Improving BCG

Knowing that BCG prevents deaths in young children and is cost-effective in areas of high TB endemicity means that it would be unwise and unethical to replace it completely as a TB vaccine. However, there is undoubtedly enormous scope to improve on BCG-induced protection, both in terms of lengthening the period of protection and in providing higher levels of protection against pulmonary disease.

Ways of improving BCG?

Until recently there was little information to support or refute the use of repeated BCG vaccination, but a large cluster–control study in Brazil, published recently, provides no evidence to support this practice [66]. However, the heterologous prime-boost strategy, whereby two different vaccines containing the same antigen administered a time-period apart, has been shown to be highly effective at inducing both CD4 and CD8 T cells in animals and humans [67,68]. Furthermore, this strategy has been shown to enhance protective efficacy against M. tuberculosis challenge in animal models [69]. BCG can be incorporated into this strategy, if the boost contains an antigen also present in BCG [70,71]. Using a heterologous boost means that there is less antibody and T cell immunity against the vector, thus optimizing the boosting potential of the second vaccine. Different approaches in vaccine design are being pursued, although most are focusing on enhancing the Th1 response, together with CD8 T cell response. Broadly, groups are either working on improving the mycobacterial prime or on developing effective subunit boosting vaccines. Both recombinant BCG vaccines and boosting vaccines have reached clinical trials. There are suggestions that combining these two approaches may also be effective and preclinical studies have begun to investigate some of these combinations. The most advanced vaccine candidates within each group are listed below.

Improving ‘prime’-recombinant BCG and auxotrophic M. tuberculosis

A number of novel strategies are being pursued to either improve BCG or to develop attenuated M. tuberculosis strains to use as an alternative to BCG. There are supportive preclinical data for each of these strategies.

Endosome escape BCG

A BCG which has had the urease gene deleted and a lysin gene from Listeria monocytogenes inserted, rBCGΔUreC:Hly+, showed greater protective efficacy than parental strain BCG against aerosol M. tuberculosis challenge for > 200 days in a murine model [72]. Engulfed mycobacteria, including BCG, inhibit the acidification of phagosomes through the secretion of urease. This immune evasion strategy, which cannot occur in BCG lacking urease, prevents mycobacterial death and antigen presentation to T cells via MHC class I. The lysin gene product enables the phagosome membrane to be punctured, which allows passage of the BCG into the cytoplasm and access to MHC class I. In addition, rBCGΔUreC:Hly+ is pro-apoptotic, enabling enhanced cross-priming. The improved protection seen with this vaccine has been accounted for by the greater CD8 response [72]. There are plans for this vaccine to enter phase I clinical trials soon.

Over-expressing antigens

BCG has also been developed, as a vector for other immunodominant antigens. For example, rBCG30 is a Tice strain of BCG, engineered to over-express antigen 85B, a mycolyl transferase involved in mycobacterial cell wall synthesis. This vaccine has shown enhanced protection against TB challenge in two animal models [73,74] and has completed one phase I clinical trial but with disappointing immunogenicity [75].

Combined endosome escape and antigen over-expresssion

A recombinant BCG that is designed to escape from the phagosome using the pH-independent perfringolysin and additionally over-expressing mycobacterial antigens Ag85A, Ag85B and TB10·4 is being developed by Aeras and is scheduled for phase I clinical trials in 2007 [75].

Introduction of deleted genes

Genes that were lost from M. bovis in the development of BCG and account for the attenuation of BCG have been reinserted. For example, the RD1 region has been reinserted into BCG, producing a vaccine that is highly immunogenic [76], although there are safety concerns for its use in humans [77]. A further limitation of such a vaccine is that it would interfere with the new immunodiagnostic tools.

Attenuated TB

There are, understandably, many concerns about the use of attenuated M. tuberculosis vaccines. These include issues of safety, public perception, regulatory authority hurdles and the potential to release antibiotic resistance gene markers into the environment. A WHO committee announced recently that such vaccines would require two non-reverting independent mutations, demonstration of safety in SCID mice and that use would be limited to HIV-negative individuals [78]. The most advanced attenuated TB vaccines in development are vaccines with disruptions in lipid metabolism pathways, mc2 6020 (pan C, pan D, lysA) and mc2 6030 (pan C, pan D, ΔRD1), which also has the RD1 region deleted [79–81]. Both of these vaccines are scheduled for phase I studies in 2006.

Improving boost

In designing subunit boosting vaccines, one has choice in which mycobacterial antigen or antigens to use and in which vehicle to deliver them.


All groups to date have focused on antigens that are immunodominant in natural host immunity in animals and humans. Those antigens that are secreted by mycobacteria into their culture filtrate have aroused the most interest. These include antigens from antigen 85 complex (A and B) [82,83], antigens of RD1 region including ESAT-6, TB10·4 and Mtb 9·9 [84,85] and antigens belonging to the PPE family of proteins (containing conserved proline and glutamate residues) Mtb39a-e and Mtb41 [86–88]. The concern over compromising the specificity of new immunodiagnostics means that subunit vaccines containing ESAT-6 are no longer being developed. The use of antigens that are conserved among environmental mycobacteria may be particularly beneficial, as individuals could theoretically be repeatedly boosted to such antigens.

Delivery vehicle

DNA, fusion proteins plus adjuvants and viral vectors have all been considered as vaccine delivery vehicles. DNA vaccines have been shown to be highly immunogenic in animal models, but have consistently given disappointing results in humans. In fact, there is no prophylactic TB DNA vaccine currently in development. The adjuvants that are being trialled with fusion proteins include ASO2A and IC-31. These are all known to be powerful adjuvants for CD4 T cells. Strong CD8 T cell inducing adjuvants are yet to be identified. Both poxviruses and adenoviruses are being pursued currently as possible viral vectors, the latter inducing strong CD8 T cell responses. An adenovirus (human serotype 5), AdHu5, containing antigen 85A has shown efficacy superior to BCG [89] and an adenovirus (human serotype 35) AdHu35 containing antigen 85A, antigen 85B and TB10·4 has also shown protective efficacy comparable to BCG [90]. However, there are issues surrounding the high level of pre-existing immunity to ubiquitous viruses such as AdHu5 and AdHu35 [91]. Consequently, work is being conducted to circumvent this through the development of either Simian vectors or chimeric vectors, using proteins from rare human serotypes [92]. The boosting vaccines which have reached the clinical trial stage of assessment are as follows.

Fusion proteins

Mtb72F, a fusion protein of Mtb 39 and Mtb 32 administered with the adjuvant ASO2A, has shown protective efficacy against M. tuberculosis challenge comparable to BCG in both mouse and guinea pig models [93,94]. This vaccine has completed clinical trials in the United States and Europe. Results from the European trial showed the induction of predominantly IL-2-producing CD4 T cells [95]. Hybrid-1, a fusion protein of Ag85B and ESAT-6 (Hybrid-1), also showed efficacy in a guinea pig model [96] and clinical trials began in Europe in November 2005. This particular vaccine is unlikely to be developed further because of the inclusion of ESAT-6 for the reason highlighted above.

Viral vectors

Modified vaccinia virus Ankara (MVA) has been shown to be a strongly immunogenic vector for both CD4 and CD8 T cells both in animal models and in human trials [67,68].

MVA85A, a recombinant MVA expressing antigen 85A, has been shown to enhance the protective efficacy of BCG in a murine [70] and macaque model (Verreck, unpublished data). This vaccine entered phase I clinical trials in September 2002 and has been assessed subsequently in series of trials in the United Kingdom and The Gambia and has recently entered phase II trials in South Africa. It has been shown to be safe and immunogenic in naive, BCG-vaccinated [97] and latently infected volunteers (Sander, unpublished data). Importantly, the boosting of IFN-γ cells, determined by the ex-vivo enzyme-linked immunospot (ELISPOT) assay, remained significantly elevated at 6 months [97] and 1 year after vaccination (McShane, unpublished data).

Steps in evaluation of new vaccines

The candidate vaccines described above are only a fraction of those that have been developed. Selecting the vaccines that are most likely to be effective in humans from the hundreds that are being developed is challenging, particularly as there is no clearly defined correlate of protection. Consequently, the use of animal models and the development of immunoassays for use both in animal models and in human trials is essential. In time, validation of models and immunoassays will be possible as results of human efficacy trials become available.

Animal models

Once a vaccine has been developed, protective efficacy is usually assessed first in the murine TB challenge model. Classically, 4 weeks after final vaccination mice are infected with M. tuberculosis ideally by the aerosol route. Bacterial load in lungs and spleens, together with pathology, are assessed 4 weeks later. This model is cheap and relatively short (for TB experiments). It is also possible to investigate the mechanisms by which vaccines protect, given the wide availability of immunological reagents. The major disadvantage of this model is that the pathology induced is distinct from humans. Granulomas form a collection of macrophages and lymphocytes but are much less organized than those seen in humans and do not develop necrosis. Additionally, there is a limit to the sensitivity of the model, as there is a threshold of protection. This makes discrimination between vaccines difficult, particularly vaccines designed to boost BCG.

Vaccines that have demonstrated a protective efficacy at least as good as BCG in the mouse may be assessed further in the guinea pig model. These animals are highly susceptible to TB. However, BCG induces protection which enables guinea pigs to survive 1–2 years post-low-dose challenge, which makes for lengthy and costly experiments. A more unphysiological high-dose TB challenge has been used as an alternative, to enable earlier discrimination from BCG [98]. The outcome measures in this model have traditionally been survival and pathology. Granulomas in guinea pigs are much more representative of those in humans, although they mineralize rather than cavitate. Bacterial load is used occasionally as a secondary outcome measure. The major limitation of the guinea pig model, other than the excellent protective efficacy of BCG, has been the lack of immunological reagents. However, advances in the production of guinea pig antibodies and the use of RT-PCR will facilitate immunological investigation in the near future.

Some vaccines are assessed further in the macaque model, in either cynomolagous or rhesus macaques. However, the expense and low availability of these animals limits their use. As one may imagine, these outbred animals behave in similar ways to humans, producing a similar spectrum of disease and pathology. Following low-dose aerosol challenge only 40% of animals develop active disease, with the remaining developing the equivalent of latent infection [99]. This can add difficulties to experimental set-up since larger group sizes are necessary to account for this variability in disease type. Extensive immunological reagents, including some cross-reactive human reagents, allow immunological assessment to be performed.

Many other animal models have been used to investigate TB immunopathogenesis, including the zebra fish, rat, rabbit and cow. In terms of vaccine studies, the cow is being increasingly used. It has the advantage of producing pathology similar to humans, but also that infection with M. bovis produces a natural host–pathogen pair [100].

Limitations of animal models

There are some consistent limitations to all these models for the assessment of vaccine efficacy. First, the interval between vaccination and TB challenge tends to be short, an average of 4–6 weeks. This means that the induced immune response has not had the opportunity to reach its true memory state. Assessment of protection at later time-points is more likely to truly represent vaccine efficacy in humans. Most murine studies are also curtailed at about 4 weeks following challenge and consequently provide only a snapshot of protection. There is no true consensus about which animal model is more representative of the human situation. None of the models described have been validated completely and this will not be possible until some vaccines have completed human efficacy trials. However, collaboration between teams working on vaccines in phases I and II clinical trials and those working with animal models means that comparative immunology is being performed. For the moment the different animal models are complementary, adding different information about mechanisms of actions of vaccines. Given that TB is a complex disease, with pathology being consequent upon bacterial burden and immune responses, use of different outcome measures within a model is important. For example, in the guinea pig model it has been shown that some vaccines do not reduce bacterial load but result in a lymphocytic pathology that prolongs survival [83]. Developments and application of new technology, for example radiology such as magnetic resonance imaging (MRI) to assess pathology in lungs, will improve understanding [101]. Investigating the immunological mechanism of action of vaccines in different animal models is important, as protective efficacy could be derived in different ways by different vaccines.

Comparing different vaccines in the same animal model

With so many new vaccines being developed it is important that some comparative animal work can be performed to provide some discernment about the probable efficacy of different products. Two centres of excellence have obtained funding for comparative trials of new vaccines. These are Colorado State University, funded by the National Institutes of Health (NIH) and the Health Protection Agency, Centre for Applied Microbiological Research (CAMR) funded by the European Union. The former centre has developed expertise in murine and guinea pig models and the latter centre in guinea pig models. Both centres have modified their systems in an attempt to obtain as discriminatory models as possible. Additionally, all researchers are being encouraged to use a WHO supply of Erdman strain M. tuberculosis and BCG from the Statens Serum Institut (SSI) in their animal models to try to produce more uniformity in published work.

Animal models in different settings

TB exposure in the field is clearly more complex than that which is being modelled in animals. Given that an estimated one-third of the world is infected with M. tuberculosis, models of latent infection are required to predict efficacy of vaccines in this setting. Many attempts have been made to establish a model of latent TB, including adapting two basic murine models, but this is extremely lengthy and challenging and there are major limitations to each of these models [30,102–106]. The macaque appears to be a promising option, except for the expense involved, particularly since large numbers need to be infected in order to obtain a decent group size of latently infected animals. Other field scenarios that are important and for which animal models are being developed include adjusting for environmental mycobacteria exposure [56,57], helminth infection [107] and HIV, the latter requiring a Simian immunodeficiency virus (SIV) macaque model.

Progress from animals to humans

Once vaccines have shown efficacy in animal models, consideration can be given to entering them into early clinical trials. This is a costly exercise and funding is the limiting factor dictating which vaccines progress to the essential next stage. In order for vaccines to enter clinical trials, they have to be manufactured under good manufacturing procedures (GMP) [108], undergo toxicology studies and pass through the appropriate regulatory authorities and ethical review.


Phase I clinical trials are designed primarily to investigate safety. This is particularly important for TB vaccines, given the history of Koch's experiments in the 1890s. He claimed that he had found a cure for tuberculosis, culture filtrate protein. Following some preliminary experiments in guinea pigs he administered it publicly to patients with active TB. This had devastating consequences, several of his patients dying from tissue necrosis thought to be the result of immunopathology induced by the ‘remedy’[109]. The likelihood of this phenomenon occurring is thought to relate to bacterial load, as it has been seen following DNA vaccination in a murine model of active TB but not latent TB [110]. Concern about the possibility of inducing such a phenomenon meant that trials of the first new subunit vaccine began in those estimated to be as mycobacterially naive as possible (PPD-negative on skin test, BCG-naive), progressed to individuals vaccinated with BCG and subsequently to individuals known to be latently infected with M. tuberculosis (defined by positivity to ESAT-6 and CFP-10 in the ELISPOT assay) [111]. No vaccine other than M. vaccae has yet been trialled in active TB disease.


Establishing the immunogenicity of vaccines in phase I trials is also essential. DNA vaccines have classically produced very strong IFN-γ responses in animal models, but this does not translate to the human setting. There is no doubt that IFN-γ is essential to host defence. Although it is recognized that induction of IFN-γ alone does not correlate with protection, it is the best surrogate marker of protection currently available and remains the primary immune assay for phase I TB vaccine trials. Clearly, a number of immune assays can determine IFN-γ production and different research groups with vaccines in clinical trials are using different assays as their primary read-outs. These include the ex-vivo IFN-γ ELISPOT assay [97], the whole-blood IFN-γ enzyme-linked immunosorbent assay (ELISA) [61,112] and intracellular cytokine staining [113]. Additionally, for each of these assays there is also variation in methodology between laboratories; for example, different mycobacterial products or stimulants are used. A group within the TB vaccine cluster (TBVAC; EU sixth framework) have investigated the reproducibility of each of these assays and have found the ex-vivo IFN-γ ELISPOT to provide the most consistent results (R. Audran and F. Spertini, unpublished data). The whole-blood IFN-γ ELISA is also considered to have an important role in the field, because of its simplicity. It seems sensible that there is some harmonization of methodology so that some comparison can be made of the relative immunogenicity of new vaccine candidates. However, it is important that this does not inhibit investigative research. Other important aspects of cellular immunity that are being determined in phase I trials include production of other important cytokines such as IL-2 and TNF-α, induction of CD8 T cells and regulatory T cell responses, whether that be Th2 or Treg cells themselves. Investigation of T cell phenotype will also be important in addition to other measures of T cell function, such as proliferation. Mycobacterial inhibition assays have been very challenging to perform, but methodology has developed and these assays provide a very important read-out [63]. With the development of new molecular techniques, global gene expression using microarrays will also provide new insight into the broad effects of new vaccines. Investigation of lung immune responses induced by vaccines will also be important, although currently requires the use of invasive techniques to obtain bronchoalveolar lavage fluid. One hopes that knowledge gained from trials such as the South African Tuberculosis Vaccine Initiative (SATVI) BCG trial with efficacy end-points will feed back into the design of immune assays for other vaccine trials. If correlates of protection are discovered, this will enable trials to produce speedier end points.

Phases I–III

Many steps are involved to progress a vaccine from a phase I trial to a phase III efficacy trial, funding being the major hurdle. There is a requirement for further phase I trials in all appropriate patient groups, for example latently infected and HIV positive individuals and in populations of the appropriate age range for the final efficacy trial. Phase II trials need to take place among the population that will be vaccinated for the efficacy trial, to ensure immunogenicity is maintained and that there are no further safety issues. Because only one in 10 infected individuals develop TB in their lifetime, large numbers will be required to demonstrate efficacy of any new vaccines in phase III trials. Of interest, > 200 000 children were entered into the recently published BCG revaccination study in Brazil [66]. Consequently, trial sites with high incidences of M. tuberculosis, such as in Cape Town, South Africa, become very attractive. In addition to the issues of funding, consideration needs to be given to which population to use: for example, infants or adolescents, inclusive or exclusive of HIV positives, the choice of clinical end-point and the ethics of using one trial site.


This is a very exciting time in TB vaccinology, with several vaccines reaching clinical trials. It is essential that endeavour continues to enhance understanding of the complex host–pathogen interaction, which results in most exposed individuals not developing persistent infection and most infected individuals not developing disease. Additionally, deciphering the mechanism of protection induced by BCG will aid vaccine design rationale. However, there must be awareness that different vaccine regimens may have different mechanisms of action. BCG appears to induce its protective efficacy by the prevention of dissemination of infection but new vaccines may work, for example, by enhancing mycobacterial clearance or by reducing the development of pathology. Therefore, alongside basic microbiological and immunological research and immuno-epidemiological research, vaccines need to progress from the preclinical stage to clinical trials. There needs to be a concerted effort to use the latest immunological and molecular techniques to decipher correlates of protection for the vaccines that do advance into phase III studies and there needs to be an iterative approach with those working on both preclinical vaccine design and with animal TB challenge models. The vaccines that have reached the most advanced stage of assessment can establish pathways for future vaccine development. There is a further hope that TB vaccine trials will enhance capacity building in different countries, with increased training opportunities and the development of an improved infrastructure in the different trial sites.