TB-specific considerations

TB vaccine development is facing two key challenges that currently preclude effective and efficient prioritization of vaccine candidates to advance into the clinic: 1) the lack of accepted or validated immune correlates or surrogates of protection and 2) the lack of a validated animal model(s) that predict TB disease and outcomes in humans. This has an impact on pre-clinical screening of potential candidates and on optimisation of the vaccine formulation and regimen, since even lengthy and costly protection experiments in animal models have not yet been validated to predict clinical efficacy. The lack of an established immune correlate also hampers the development of an appropriate, qualified assay to measure potency which would accelerate and harmonise product characterisation and QualityControl.

Measurement of antibody responses has provided the tool to monitor the quality and quantity of the adaptive immune response for many vaccines, but not for BCG. There is no consensus on what the protective antigen(s) against TB are, nor is the mechanism of protection against TB disease fully understood. Hence, the assessment of the immune responses to a TB vaccine candidate should be guided by the nature of the vaccine and broadened to include  measurement of innate and adaptive immune responses, both humoral and T-cell mediated responses. This provides a greater technical and financial challenges, as preclinical results will not reduce the risks for further development to the same extent as for vaccines for which a correlate of protection exists.

The TB Vaccine DevelopmentPathway tool addresses and provides guidance for such TB vaccine-specific challenges.

Vaccine technology specific considerations

In the field of vaccinology, new technological platforms have emerged, each with specific challenges for development (Pollard et al., 2021). Two classical vaccine platforms stay close to the original pathogen, either as an inactivated or a modified live version that causes no disease. There are several examples of attenuated and inactivated vaccines against viral diseases (e.g. MMR and IPV),whereas there are fewer examples for bacterial pathogens. BCG is a rare example of a marketed live attenuated vaccine and whole-cell pertussis is one of the few licensed inactivated vaccines. For TB, vaccines based on several different platforms are in advanced stages of clinical development, with an overview available in the TB vaccine pipeline. Besides live attenuated or inactivated vaccine candidates, there are TB subunit vaccine candidates that comprise selected Mtb antigens delivered by a variety of platforms.

These include classical adjuvanted protein vaccines, viral vector-based vaccines and mRNA and DNA encoded vaccines.

Protein subunit vaccines combine pathogen-specific antigens with an adjuvant as a supportive immunostimulant. Examples are the Hepatitis B subunit vaccine, bacterial toxoid vaccines e.g. Tetanus and Diphtheria, and conjugate vaccines e.g. pneumococcal vaccines.  For TB, several protein subunit vaccines are in advanced stages of clinical development, including single recombinant polypeptide antigens and multiple, fused proteins combined with liposomes and Toll-like Receptor specific adjuvants.

Recombinant viral vector vaccines use a replicating or non-replicating (abortive) viral vector as a carrier for antigens of a pathogen and production of the antigens is induced in cells infected with the viral vector. They aim to induce an adaptive response against the pathogen specific antigens, while also providing some innate immune stimulation.An example is the recently licensed adenovirus vector-based COVID-19 vaccine.For TB, several candidate vaccines based on viral vectors, including ModifiedVaccinia Ankara (MVA) adenoviruses, and influenza virus are in early stages of clinical development.

The DNA/RNA vaccines contain the genetic information of a pathogen specific antigen which is then produced by target cells inoculated with the DNA or mRNA. Progress has been made in the delivery of DNA (electroporation). However, the licensed mRNA vaccines for COVID-19 have provided a major incentive for TB vaccines based on mRNA formulated into lipid nanoparticles (LNP). The mRNA/LNP platform provides flexible vaccine design and generally induces strong T-cell and antibody responses. This includes candidates in clinical development and multiple preclinical projects aiming to enter clinical testing in the near future.

TB vaccine target population considerations

Information is given for specific TB vaccine target populations and indications for which there may be different types of vaccines or distinct approaches or requirements for the development pathway. Two main target populations have been listed below. This guidance refers to, and is aligned with, the PreferredProduct Characteristics (PPCs) for different TB vaccines prepared by the WHO.

1) Adolescent/ adult populations
Adolescents and adults with pulmonary TB are the primary sources of Mtb transmission, and modelling predicts that vaccination ofthese two populations would have greater and more rapid impact on the TB epidemic than improved neonatal vaccines (Harriset al., 2016). A recent SAGE recommendation stated that additional data be collected to inform the use of new TB vaccines for adolescents and adults in those with and without evidence of Mtb exposure atthe time of vaccination. In addition, in high-incidence sites little efficiency gain is expected in only enrolling participants with a positive IGRA for the efficacy evaluation (Cobelens et al., 2025). Further, vaccines should be safe for use in HIV-infected and other subpopulations such as pregnant individuals and individuals who are breastfeeding. Vaccination ideally aims at preventing TB disease whether it results from progression, reactivation of an old (re-)infection or new infection. The clinical evaluation of a vaccine is driven by the sponsor's chosen target indication and target population(s), as well as by considerations such as budget, pricing and implementation strategies. The relevant WHO PPC is described in section 6 of the document WHO Preferred Product Characteristics for New Tuberculosis Vaccines, and WHOEvidence Considerations for Vaccine Policy (ECVP) development for TB vaccines intended for adults and adolescents.

2) Neonates/ infant populations
New TB vaccines for neonates could either be in the form of a replacement for neonatal BCG or a booster vaccination administered to infants with the aim to improve current BCG vaccination, providing greater and longer protection (prevention of disease) and having a safety profile at least as safe as BCG. Therefore, the benchmark for the development stages and functions is BCG, in contrast to vaccines for adults and adolescents for which no vaccine and thus no benchmark is yet available.Benefits of a BCG replacement vaccine should be evaluated in Phase 3 clinical studies. Information and recommendations on the preferred product characteristics can be found in section VII of the document WHO Preferred Product Characteristics for New Tuberculosis Vaccines.

Of note:

• Other priority populations identified by the WHO ECVP that are at increased risk of TB or have specific safety concerns include: immunocompromised and immuno suppressed individuals, including people living with HIV, pregnant and breastfeeding individuals, malnutrition (underweight or morbid obesity), people deprived of liberty and  other high-risk priority populations (e.g., migrants, refugees, homeless people, people living in high-risk congregate settings, people with unhealthy alcohol use or substance misuse, miners, and people living in high-density areas). Vaccine developers may consider the inclusion of these priority populations in clinical trials testing new TB vaccine candidates. However, although these arepopulations that are important to vaccinate, they may be targeted at some point only after a recommendation has been made to vaccinate other priority populations.
• Therapeutic vaccines are deployed as an adjunct to chemotherapy in any age group to improve treatment efficacy in individuals who currently suffer from TB disease.The development pathway for therapeutic TB vaccines is substantially different from prophylactic indications, in particular the design of clinical studies.The target outcomes of therapeutic vaccines are to improve success of treatment, particularly for multi-drug resistant TB, or to decrease or prevent relapse. Shortening the duration of drug treatment and/or reducing the number of drugs necessary to cure TB disease should also be considered. Because of these multiple possible outcomes, there will be a number of aspects of the development pathway which differ compared to prophylactic vaccines. Notably, clinical trial designs will need to evaluate safety, define optimal timings and dosage in relation to drug treatment and ensure that standard chemotherapy isnot adversely affected by vaccination. More information is available in the WHOPPC for therapeutic vaccines.Therapeutic vaccines will not be discussed further in this document.

Animal Model Considerations

Because of various limitations and bottlenecks in clinical experimental medicine and trials (ethics, costs, capacity), preclinical evaluation of vaccine prototypes and/or candidates is relevant and pertinent at different stages of the vaccine development process. Pre-clinical studies in animals can help in guiding vaccine platform and antigen selection, in generating data on preclinical safety, or in estimating optimal formulation, dose, route, and/or immunisation interval. Further, animal models can help provide proof-of-concept for efficacy. Importantly, no single animal model is validated as predictive for clinical outcome and there is no legal or regulatory framework that dictates the use of certain (animal) models at a specific stage or for a specific purpose. Thus, a product developer has considerable freedom whether or not to include preclinical animal studies into a product development plan, based on a product specific risk-benefit assessment. Models are typically used to simplify 'real world' (here: clinical)complexity to leverage our knowledge base and/or relieve a resource-limited clinical development path. As all models inherently have their limitations, this guidance document aims to reflect - in a non-exhaustive manner - on somemajor species and modelling conditions that can support and facilitate the development of TB vaccines.

1)   Considerationsfor the Mouse model:

Mice represent the most widely used animal model for TB vaccine testing.Typically, inbred mouse strains (limiting host variability and reducing the number of animals needed for a study), such as C57Bl/6 or Balb/c mice, are immunized and subsequently infected with an aerosolized Mtb dose of50-100 CFU and protection is assessed by measuring the lung bacterial burdens between 28-42 days post-infection. In this model, BCG immunization typically reduces the lung bacterial burden by ~1 log, when assessed at these timepoints. However, the protection conferred is transient and usually dissipates by 3-4months post-infection and outcomes of these studies have not correlated well with results from human efficacy trials. This might be improved by adapting the mouse strain and/or challenge strain used:

Mouse strains:

Mice have fewer MHC molecules than humans; the most commonly used strain(C57Bl/6) has only a single MHCII molecule, thus some antigens may not be protective in a given mouse strain simply because it lacks epitopes capable of binding the MHC molecules of that strain. Testing immunogenicity in a variety of mouse strains with different MHC haplotypes (e.g., C57Bl/6, Balb/c, and C3H)could reduce this risk. Subsequent efficacy testing could then use strains in which the vaccine is confirmed to be immunogenic. One consideration would be to use an F1 strain (e.g., C57Bl/6 x Balb/c) between two strains that are immunogenic to increase the number of MHC molecules capable of present ingantigens and to optimize the breadth of the T cell response, more akin to the breadth that would occur in humans. Another potential benefit of F1 strains is that they are heterozygous at most loci, reducing the propensity for extremes of immunologic phenotypes in some inbred strains (e.g., C3HeB/FeJ mice). Diversity outbred mice have been used in some experimental TB vaccine studies, however, because each mouse is genetically unique, large numbers of mice are needed, andF1 mice may represent a balanced approach which introduces some genetic diversity, while requiring animal numbers which are feasible.

Challenge dose considerations for mice:

The human infectious inoculum is thought to be only 1-3 CFU and there is growing evidence that a more physiologic challenge dose may offer several advantages for TB vaccine testing in a variety of animal models. In mice, a challenge dose of 1-3 CFU enables assessment of: prevention of Mtb dissemination to the contralateral lung, prevention of detectable infection, as well as overall Mycobacterial lung and spleen burden, for example. Moreover, the immune mechanisms may be different from those that reduce bacterial burdens in response to a supraphysiologic challenge dose. As a result, the hierarchy of vaccine efficacy may differ in response to a physiologic compared to asupraphysiologic challenge dose. There are two disadvantages of using a 1-3 CFU challenge dose: 1) it requires more mice and is therefore more expensive, and2) it is currently not routinely performed by most labs and may require using aTB vaccine testing service with this capacity.

2)   Considerations for the Guinea pig model:

Overview: The outbred Dunkin–Hartley strain of guinea pig is the most commonly used, although inbred strains are available. The typical route of infection with M. tuberculosisis by the aerosol route. Guinea pigs are very susceptible to tuberculosis (TB)and disease is induced with very low doses of M. tuberculosis. The progression of the disease that follows infection of guinea pigs with M.tuberculosis displays many features of human TB including well defined granulomas with central necrosis enclosed by lymphocytes, macrophages, and multinucleate giant cells and a fibrotic capsule. Different infection and readout regimens allow the guinea pig models to test a range of vaccines and endpoints.

Even with recent efforts, there remain few immunological assays to interrogate immune responses in guinea pigs compared to other animal species. Research continues to identify antibodies which could be used in such assays.Despite this, much progress has been made in using histopathological analysis and immunohistochemical techniques to quantify host response to infection following vaccination.

Vaccine candidates that have previously demonstrated efficacy in mouse models can be further evaluated for protective efficacy in the guinea pig model if testing across different species is expected to be of value for the preclinical package.  

3)   Considerationsfor the NHP Model

NHP provide the closest proxy to humans by phylogeny, are naturally susceptible to extremely low dose Mtb infection (<10 CFU is commonlyused for vaccine testing) and display TB disease symptoms across the range thatis reflective of human infection and clinical disease phenotypes. Theexceptional cross-reactivity of primate-specific reagents for investigatinghost disease and immunity and the notable resemblance of BCG efficacy in manynot all of its aspects, including prevention of extra-pulmonary TB disseminationand variable protection depending on the cohort under investigation, addssystem validity to NHP for TB vaccine R&D, though even this model is notvalidated as predictive for clinical outcome. For harmonisation andstandardisation, many NHP vaccine researchers gathered in the NHP researchcommunity under the Collaboration for TB Vaccine Discovery (CTVD) have agreedto use a single source barcoded Mtb strain Erdman (BEI resources) for infectious challenge experiments. In rhesus macaques in particular, Mtb Erdman can be considered highly virulent, thereby posing a rather stringent test for protective vaccine performance. Though standardization can be useful, other Mtbstrains, however, could provide different insights and  are being used and reported as well.

NHP strains

Macaque species (Macacaspp), including rhesus macaques (M. mulatta) and cynomolgus macaques (M.fascicularis, or long-tailed macaques)are the most common species used in TB vaccine R&D. Of these, rhesus over cynomolgus monkeys are more prone to develop TB disease upon (low dose) Mtbinfection. However, there are genotypic (or so-called spectrotype) differences within each of the two species, attributable to limiting founder population characteristics (e.g. Mauritian M.fascicularis) or natural geographic barriers contributing to genetic drift (e.g. Indian versus Chinese-type rhesus).Moreover, environmental factors relating to NHP husbandry conditions (nurture)add to the spectrum of the outcome of TB vaccination and infection modelling.Depending on the breeding and housing conditions, exposure to non-typicalnon-tuberculous mycobacteria (NTM) can be expected. NHP are typically available from specialized breeding centres and/or licensed suppliers that are mostly well equipped to warrant the quality of outbred NHP colonies and/or their housing conditions.

Ethical restrictions

The use of NHP in biomedical research can face restrictions, depending on the cultural and/or legal environment. Legal frameworks sometimes prohibit the use of NHP unless there is no suitable alternative model to answer a specific research question. Positive indicators for vaccine efficacy from evolutionarily 'lower vertebrates' (most typically mice or guinea pigs) are often required for ethical approval to progress into NHP studies.

4)   Challenge strainsin preclinical models.

Mostly, 2 different laboratory strains are used: H37Rv and Erdman, and in addition several clinical strains are used (table 2 in Pozo-Ramos and Kupz, 2025), such as the more virulent HN878 strain (Manca et al., 2001). In addition, a triple-kill-switch (TKS) strain was developed for safe use inhuman challenge models that might also be used preclinically (Wang et al., 2025). Although Mtbis genetically highly conserved, it is important to be aware that antigen expression can differ markedly between strains (Solans et al., 2014). Moreover, as culturing strains in different labs can lead to phenotypic and genetic variation, as with the different existing BCGsub-strains (Zhang et al., 2016), it is recognised that quality control is important. For example, when Mtb (or BCG) is cultured in vitro, including propionate and omitting Tween-80 in growth medium can enhance an essential virulence lipid (PDIM) and prevent attenuation of pathogenicity. PDIM-positivity was shown to be easily screened using avancomycin assay (Mulholland etal., 2024).

Clinical trial endpoints

1)    PreventionOf Disease (POD):

Developing a POD TB vaccine targeted at adolescents and adults is a global health priority. In addition to reducing individual level TB related morbidity and mortality, modelling suggests that a Prevention Of Disease (POD) TB vaccine targeted at adolescents and adults would have the greatest public health impact by preventing infectious pulmonary TB and reducing Mtb transmission at a population level (Harris et al., 2016) Recent phase 2b/3 efficacy trials of M72/AS01E and MTBVAC have focused on powering the trial in the Interferon Gamma Release Assay (IGRA) positive participants to maximise incident TB case accrual and reduce trial size and duration. However, since risk of progression to TB disease is highest soon after Mtb infection, it might be possible to include IGRA-negative individuals in a cost-effective trial design, if it were conducted in very high Mtb transmission settings. The regulatory and policy evidence considerations forIGRA negative participants are discussed.

The benefits and risks of innovative approaches to reduce the size, duration and cost of phase 3 trials that are considered in this document, including the use of composite asymptomatic (using chest Xray and microbiological confirmations as confirmation of disease) and symptomatic TB endpoints, enrolling a mixedIGRA positive and negative study population, integrated trial phases and adaptive designs.

2)    PreventionOf Infection (POI)

The purpose of a POI TB vaccine is to prevent sustained TB infection occurring among Mtb unsensitised individuals and thereby reduce the risk of developing pulmonary TB disease. POI studies enrol healthy participants and have conversion of  IGRA responses as a surrogate clinical endpoint. Of note, infection can occur without IGRAc on version and those with a converted IGRA may have cleared the infection.

3)    PreventionOf Recurrence (POR)

The aim of a POR TB vaccine administered to patients with pulmonary TB at the end of treatment, is to reduce the risk of TB treatment failure and TB recurrence. POR studies enrol patients who recently completed TB treatment and have recurrent TB disease over a defined period as a surrogate clinical endpoint.

Of note: POI and POR TB vaccine trials were previously used as a strategy to avoid going into large, costly phase 3 TB vaccine efficacy trials. However, this approach of using POI or POR to provide a so-called ‘pre-proof of concept(‘pre-POC) for a POD licensure indication carries risks, as neither alternative endpoint is a true surrogate or correlate for POD. A pre-POC trial with a POI endpoint may fail to indicate protection for a vaccine that would be protective in a POD trial if the vaccine’s mode of action is largely through preventing disease in people already Mtb infected. Conversely, a pre-POC trial witha POI endpoint may indicate protection for a vaccine that would not be protective in a POD trial if the vaccine’s mode of action is largely through preventing against infections that would not progress to disease. To know that a vaccine will protect people from developing active TB disease and avert cases and deaths, it must be demonstrated that a vaccine candidate shown to prevent infection does so in individuals who, without vaccination, would have developed active TB. A negative result in a pre-POC POR trial could lead to termination of a candidate that, if tested, might have demonstrated POD.

The findings of recent efficacy trials using eitherPOI or POR designs have lowered expectations that the findings of POI or POR trials would accelerate entry into large POD licensure trials by providing a pre-POC efficacy signal in a smaller, shorter and less expensive efficacy study. Although a prior POI trial of BCG revaccination in IGRA-negative adolescents showed 45% vaccine efficacy against the secondary endpoint, sustained IGRA conversion through six months (Nemes et al., 2018), a larger trial intended to validate this result in a wider age range at multiple sites in South Africa showed no efficacy against sustained IGRA conversion (Schmidt et al., 2025). Another POI trial using DAR-901 conducted inTanzania also did not show efficacy (Munseri et al., 2020).

Similarly, two recent POR trials, of the protein-subunit vaccine H56:IC31 (Borgeset al., 2025) and the live attenuated, recombinant BCG vaccine VPM1002 (unpublished), failed to show that vaccination of TB patients at the end of treatment protects against recurrent TB disease. Although these negative POR results may be vaccine-specific and studies of immune correlates of risk are ongoing, it is likely that POR efficacy will be considered too high a bar to fund more trials using this design as a means to de-risk a POD trial in the short-medium term. Based on current insights, references to POI and POR ‘pre-POC’ trials in the TB vaccine development pathway have therefore been removed from the document.