A vaccine developed for the prevention of infection (POI), given prior to Mtb exposure, should control the incipient infection stage. With much higher rates of infection than evident TB disease in endemic settings, POI trials are shorter and less costly than prevention of disease (POD) trials (7, 8). Therefore, the POI trial can be used as a viable opportunity to understand the mechanisms of vaccine efficacy in humans, providing a platform to select lead candidates for further testing. A major challenge is that there is no available standardized test to measure directly the acquisition, persistence, and clearance of asymptomatic Mtb infection. Currently, assessment of Mtb infection mainly relies on alterations in specific T-cell responses induced after Mtb infection. One of the commercial interferon (IFN)-γ release assay (IGRA), QuantiFERON-TB Gold In-Tube (QFT), measures immunological sensitization to Mtb as a biomarker for Mtb infection. Compared to persistent QFT negatives, recent negative-to-positive QFT tests are associated with higher rates of Mtb infection. Therefore, it may be ideal for conducting clinical trials of prevention of Mtb infection (POI) by novel vaccines using QFT transformation as an efficacy endpoint. A positivity cutoff IFN-γ value (0.35 IU/ml) for QFT conversion is recommended by manufacturers and CDC (9), but the immunological and analytical variability of QFT tests potentially confounds the interpretation of QFT conversion as a clinical trial endpoint (10, 11). Although the tuberculin skin test (TST) can be used as an alternative method for detecting Mtb infection, since specificity is reduced by BCG vaccination or nontuberculous mycobacteria (NTM) infection, novel diagnostic methods for successful clinical results must be developed.

A POD vaccine can be administered either pre- or post-exposure to protect against disease progression after actual Mtb infection. Knight et al. reported epidemiological modeling suggesting that adolescents or young adults are the most effective targets for POD vaccination (12). According to this model, due to children having lower rates of TB notifications, lower proportions of smear-positive pulmonary TB, and making a smaller contribution to TB transmission, a novel TB vaccine targeted at infants shows a reduced immediate impact compared to one targeted at adolescents/adults. Vaccines targeting infants prevent a relatively small number of active cases, resulting in fewer secondary cases being prevented. In contrast, vaccines targeting adolescents/adults directly affect the population with the greatest burden of active TB, such as 10-year-olds vaccinated in schools and those individuals reached in mass campaigns, which leads to a reduction in transmission. Although most vaccine candidates in clinical phases aim to prevent TB disease, POD trials require more time and higher costs than POI trials because of the much lower rate of TB disease than Mtb infection (8). Nevertheless, POD trials can directly reveal Mtb infection because the evaluation is performed by measuring clinical symptoms, chest X-ray, and direct Mtb culture from clinical samples. A recent POD trial with the candidate M72:AS01_E TB vaccine (phase 2b) was conducted in Kenya, South Africa, and Zambia. Efficacy analysis was conducted on a total of 3,283 subjects, and after a period of approximately 2.3 years, the incidence of pulmonary TB was significantly lower in the M72:AS01E group than in the placebo group (13). In this trial, M72:AS01E group showed 54% vaccine efficacy among persons already infected with Mtb, but due to the inclusion of predominantly BCG-vaccinated Mtb-infected adults, it was not possible to determine the extent to which infection-generated or childhood BCG vaccination-elicited responses influenced vaccine efficacy. Similarly, a 3-year extended follow-up study demonstrated 49.7% protection by M72:AS01_E among people already infected with Mtb (14), indicating that vaccine-induced protective immune responses were maintained for at least 3 years. With these promising findings, broader applications to diverse ethnic populations in different geographic settings will be required to conduct reliable clinical trials for POD.

Vaccines aimed at the prevention of recurrence (POR) are administered during antibiotic therapy to prevent the recurrence of TB. TB recurrence generally occurs in approximately 2 to 8% of TB patients even after treatment completion, and the recurrence rate depends on the absence or presence of cavities, bacterial burden, treatment frequency per week, type of antibiotics used and transmission rate. As most cases of recurrent TB disease develop within one year after treatment completion, the targeted populations of POR trials can usually be designated (8), but trial design is complicated due to the long-term treatment period and intervention timing. Multiple promising candidates currently under evaluation for POR include the H56:IC31 and ID93:GLA-SE subunit vaccine candidates, which were noted to prevent reactivation or restrict progression to severe disease in nonhuman primates (NHPs) (15, 16), and the recombinant BCG vaccine candidate VPM1002.

Due to the characteristics of TB, a large number of subjects for trials are needed because approximately 10% of infected individuals are at the onset of the disease. In addition, long-term monitoring is required because the timing of onset is different for each individual. These characteristics make the rapidly increasing economic problem more difficult as the number of clinical trial subjects and the test period increases. Therefore, to overcome these problems, it is important to recruit a reasonably sized experimental group and set endpoints according to the purpose of the experiment, and it is important to discover a correlate of protection (COP) that can predict vaccine efficacy, which will be addressed later in this review.

According to reports, the infection dose that causes TB disease is 1-200 colony-forming units (CFU). TB is transmitted via droplets containing Mtb generated through coughing or sneezing, and the number of droplets generated through a single cough is approximately 1-400 CFU (18, 19). According to another study of TB patients, the average number of aerosolized CFU generated by coughing for 5 minutes was 16 (20). When an individual is infected with Mtb, the actual infection dose may be much lower than the number of bacteria released by coughing because not all aerosols generated by the infected person are inhaled. In addition, it has been reported that symptomatic TB patients release Mtb aerosols not only by cough but also by exhalation, with an aerosol size of 0.5-5 μm, showing that actual infection can be achieved in the context of a sustained low-dose of bacteria (21, 22). However, animal models for vaccine research in the preclinical stage can be established through a single, sufficient infection dose and used to evaluate vaccine efficacy. It is unclear whether the reduction in the bacterial burden by vaccination in the context of single-dose infection is a good predictor of actual clinical vaccine performance. This single-dose challenge could overwhelm or bypass the relevant immunological cascade and mask the full potential of candidate vaccines (23). In a mouse model, ultra-low-dose aerosol infection with 1-3 CFU resulted in characteristics more similar characteristics to human TB, such as highly heterogeneous bacterial burdens and well-circumscribed granulomas, than conventional-dose infection with 50-100 CFU (24). Recently, Dijkman et al. tested the efficacy of pulmonary BCG vaccination in a rhesus macaque model with a 1 CFU Mtb infection every day for 11 days and noted the importance of Th17 cells and IL-10 (25). This study does not represent all situations in which infection occurs, but it does provide a model that accounts for specific persistent and practical infection situations, such as household contacts. These studies suggest that it is necessary to reconsider the infection dose used in preclinical vaccine studies.

Most individuals who become infected can remain asymptomatic for a long time. Although environmental, cultural, geographical, and contextual characteristics can affect whether infection occurs, TB susceptibility due to host genetic differences has been reported as one of the determinants of TB disease (26). In the case of TB studies, more than 60% of preclinical studies have used mouse models, susceptibility to Mtb differs depending on the mouse strain. The most widely used C57BL/6 mice or BALB/c mice have relatively low susceptibility, whereas DBA/2, CBA/J, I/St, C3H, and A/J strains have relatively high susceptibility (27). In addition, it has been reported that necrotic granulomas found in patients with active TB are not formed in BALB/c or C57BL/6 mice, whereas they are formed in the TB-susceptible mouse strains I/St and C3H (28, 29). Recently, it was confirmed that human-like necrotic granulomas were formed by Mtb infection in a humanized mouse model, and caseous necrotic granulomas showed an immune phenotype and spatial organization similar to those observed in TB patients (30). Arrey et al. presented the utility of this model for the evaluation of a preclinical model of anti-TB drugs in an in vivo environment. Zelmer et al. immunized different strains of inbred mice, such as A/J, DBA/2, C57BL/6, and 129S2, displaying different susceptibilities to Mtb with BCG (31). Smith et al. used a “collaborative cross” model system created by crossing inbred and outbred mice to understand the broader host genetic susceptibility spectrum and for use in vaccine efficacy testing (32). These models can confirm the genetic immunological correlation associated with TB vaccine efficacy and can simultaneously be used to identify potential improvements and key defense factors for the development of a robust TB vaccine.

To date, most studies with NHPs have been used to model only acute TB, which is much less prevalent than latent TB in humans (33). Rhesus macaques develop active TB in approximately 90% of the infected population, whereas cynomolgus macaques develop active TB in only 60% of the infected population. It has also been reported that BCG showed a higher protective efficacy in cynomolgus macaques than in rhesus macaques (34). Rhesus macaques and cynomolgus macaques can develop acute, chronic, or latent TB depending on the route of infection, dose, and Mtb strain used for inoculation.

Vaccine trials employing the NHP model are expensive, but they can serve as a checkpoint for clinical trials, resulting in significant cost savings. Areas-402 induced a strong T-cell response but did not protect rhesus macaques against infection with 200 CFU of Mtb Erdman (35). Conversely, a clinical study of MVA85A without efficacy evaluation was performed with the NHP model, and although it was expensive, the efficacy was not proven in the clinical stage (36, 37). These results suggest that validation of efficacy for TB vaccines via the NHP model to enter the clinical stage may accelerate TB vaccine development.

In an evaluation of vaccine efficacy in animal models, the bacterial burden between vaccinated and non-vaccinated groups is one of the key factors. Previous studies have demonstrated the presence of non-replicating bacterial populations in sputum samples obtained from TB patients (38, 39). These non-replicating subpopulations have been attributed to the existence of persister-like bacilli in a non-replicating state (39). Even in preclinical TB vaccine models, nonculturable or persistent mycobacterial subpopulations may arise due to immunological pressures resulting from the characteristics of the vaccine candidate or vaccine model, which can hinder the accurate evaluation of vaccine efficacy. Resuscitation-promoting factors (Rpfs) are bacterial proteins which are primarily identified by their ability to resuscitate nonreplicating cells in vitro and in vivo (40, 41). It has been reported that non-replicating bacteria in a patient’s sputum can be revived by Rpfs and the culture time can be shortened (42, 43). The application of Rpfs to bacterial culture in conventional media has the potential to reduce errors in vaccine efficacy evaluation that can be caused by nonculturable or persistent subpopulations.

NHPs show anatomical and physiological similarities with humans as well as a wide range of clinical symptoms of TB, including pulmonary and extrapulmonary signs and symptoms. The NHP model enables the analysis of infected tissue, which is difficult in clinical stages, and at the same time, the disease course can be monitored by measuring parameters on radiographic images and examining body fluid samples, which can also be performed in humans. In addition, the NHP model allows the use of computed tomography and positron emission tomography to observe the progression of Mtb infection to disease in an individual (44). In the clinical phase, blood samples are used to measure immunogenicity. Currently, the Ag-specific T-cell response, multifunctionality of the Ag-specific T-cell response, and Ag-specific IgG antibody titers are commonly evaluated to demonstrate immunogenicity after vaccination ( Table 1 ). However, in the case of vaccine candidates, when the efficacy is evaluated through animal experiments, the analysis is not based on blood but rather on tissues, such as lung, spleen, and lymph nodes. Therefore, it is difficult to apply COPs from tissue-based analysis in preclinical studies to clinical studies. The NHP model enables analysis of indicators applicable to human clinical studies such as blood, urine, and PET-CT results, and analysis of indicators that can be measured only after sacrificing animals, which is possible only in preclinical models. Exploration and verification of significant indicators through this model can lead to an acceleration of vaccine development. Therefore, studies employing the NHP model before the clinical stage can provide meaning beyond simply being the gateway to the clinical stage.

Mtb contains approximately 4,000 individual proteins, and most Ags included in current subunit vaccines have been adopted mainly based on their immunodominant properties for T-cell responses in preclinical and clinical settings. Currently, approximately 100 Ags in the preclinical stage (approximately 3% of all Mtb Ags) have been studied ( Table 2 ). Most of the Ags for TB vaccine candidates are abundantly secreted and cell wall-associated proteins, including ESAT6, Ag85B, Ag85A, HSPX, and MPT64. In addition, cell wall-associated or virulence-associated Pro-Glu/Pro-Pro-Glu (PE/PPE) family proteins, a component of M72 subunit and ID93 subunit vaccine candidates, and heparin-binding hemagglutinin also produced promising vaccine-induced protection in mouse models (103, 104). Ags related to latency (DosR, resuscitation-promoting factor) and hypoxia-related proteins are being used for vaccine testing. Furthermore, hypothetical proteins are also used, for example, Rv1767, which is produced by the pathogen during the first week of infection of human cells. Aagaard et al. reported that the dimers EsxD-EsxC, EsxG-EsxH and EsxW-EsxV produced by the ESAT6 secretion system (ESX) were highly immunogenic. Integrating these in a fusion protein form called H65 resulted in a formulation that demonstrated protection efficacy equivalent to that of BCG without interfering with current ESAT6- and CFP10-based diagnostics (105). Liu et al. also purified 1,250 Mtb proteins with an E. coli expression system and evaluated cellular and humoral immune responses in human PBMCs and serum, respectively. They eventually identified four Ag candidates, Rv0232, Rv1031, Rv1198, and Rv2016, displaying high immunogenicity (106). Currently, only 11 Ags have been selected as a component, in the form of fusion proteins, in formulations eventually entered into clinical trials ( Table 3 ).

The challenge of Ag screening is complicated by the roles of Ags in multiple stages of Mtb infection, particularly chronic and latent infection stages. During Mtb infection in a mouse model, ESAT6 is consistently expressed, but Ag85B is mainly expressed at an early stage when Mtb is actively replicating (116). Mtb infection induced the accumulation of ESAT6-specific CD4⁺ T cells in the mouse lung parenchyma, but the T cells became functionally exhausted due to chronic stimulation of Ag. Whereas, Ag85B-specific CD4⁺ T cells maintain memory cell features during infection but contract in numbers by reduced Ag expression during persistent infection (116). These results have important implications for the rational design of TB vaccines tailored to optimize the protection conferred by specific CD4⁺ T cells that recognize Ag expressed at distinct stages of Mtb infection.

Although immunodominant Ags are generally accepted to induce superior vaccine efficacy, some studies suggest that Mtb can modulate host immunity through immunodominant Ags. T-cell epitopes among well-known immunodominant Mtb Ags are highly conserved, suggesting the possibility that being recognized by the host through immunodominant Ags may be beneficial for Mtb (117, 118). In addition, T-cell responses to some immunodominant Mtb Ags have been found to be notably greater in active TB patients than in individuals latently infected with Mtb (119, 120), indicating that enhanced T-cell responses may be associated with deteriorated lung inflammation, resulting in subsequent transmission. Therefore, to confirm this possibility, Orr et al. confirmed the efficacy of immune subdominant Ags as a TB vaccine candidate in a mouse model, but little correlation has been found between vaccine efficacy and the immunodominance of Ags during Mtb infection (121).

Furthermore, it is also important to characterize the vaccine potential of Ags likely to be associated with reactivation from latent Mtb infection. A well-characterized bacterial regulon can induce the dormant state of Mtb that is controlled by DosR–DosS, which is induced by immunological pressure of the host such as local hypoxia, nitric oxide, and carbon monoxide (122). These host immune responses can be induced by the vaccine with immunodominant Ag in an active state of Mtb, indicating the potential for immune evasion of Mtb against immune responses produced by vaccines targeting immunodominant Ags in an activated state. Therefore, it is also important to characterize the vaccine potential of Ags likely to be associated with reactivation of latent Mtb infection. Hence, developing novel vaccines that encode genes expressed during the reactivation of a dormant state, such as rpf, would be a strategic approach. In vitro and in vivo transcriptional profiling studies have shown that five rpf of Mtb are expressed at varying levels in a growth stage-dependent manner (123, 124). RpfB, one of the five Rpfs produced by Mtb, plays an important role in the resuscitation and growth in a dormant state. In addition, delayed reactivation induced by aminoguanidine in chronic TB was observed in mice infected with a strain lacking rpfB (125), and significantly higher T cell responses to recombinant RpfB and RpfE were detected in LTBI than in active TB patients (126), indicating that Rpfs are involved in the reactivation process in vivo. Moreover, RpfB has been studied as a promising candidate for DNA vaccines, shown to induce a modest but significant cellular immune response against TB with higher levels of IL-2 and IFN-γ (74). In addition, the RpfB domain can induce a humoral response, and monoclonal antibodies against Rpfs could inhibit TB relapse (127).

Vaccine strategies using fusion proteins can be designed to include multiple Ags, so they can induce a broader immune response than single Ag vaccines. In addition, this strategy has the advantage of inducing an effective immune response by fusing a protein with low immunogenicity with a protein or peptide with high immunogenicity. The M72 vaccine candidate was formed through a fusion of the Mtb32A and Mtb39A proteins, selected for their ability to provoke T-cell responses in TST-positive healthy adults. On the other hand, multistage subunit vaccines, such as H56 (which contains the latency-associated Ag Rv2660c fused with Ag85B and ESAT6), as well as LT70 and ID93, also incorporate multiple Mtb Ags differentially involved in bacterial growth, virulence, and metabolism (111, 128, 129). A new TB vaccine candidate, H107, that integrates eight individually protective Ags (PPE68, ESAT6, EspI, EspC, EspA, MPT64, MPT70, and MPT83) is highly immunogenic in both mice and humans (102). This fusion protein is composed of 4 ESAT6 molecules in the middle, which led to a significant increase in ESAT-6-specific immunogenicity. H107 with the BCG vaccine could increase Ag coverage to induce robust protective immune responses in a diverse human population by including as many protective/recognizable Ags as possible. While traditional vaccines containing BCG-shared Ags show in vivo cross-reactivity to BCG, H107 demonstrates no cross-reactivity and does not impede BCG colonization. Instead, co-administering H107 with BCG results in enhanced adaptive responses against both H107 and BCG (102).

Continuous research on innate immune factors related to the efficacy of TB vaccines has been conducted. Strategies that target these innate immune factors have been shown to improve vaccine efficacy. In addition, the characteristics of innate immune factors that correlate with vaccine efficacy show potential as biomarkers of vaccine efficacy ( Figure 1 ).

Protection against Mtb afforded by a TB vaccine in a mouse model appears to correlate with the T_CM phenotype, but data are limited. Tissue-resident memory T (T_RM) cells, parenchymal-resident noncirculating memory cells that have been studied only relatively recently, reside in tissues for early recognition of infected cells (160). Vaccines that elicit a rapidly accessible T-cell response to the pathogen early in Mtb infection are thought to enable more efficient protection via T_RM or T_EM cells. Furthermore, the protective role of antibodies in the pathogenesis of TB highlights the need for continuous exploration of the adaptive immune response as a biomarker for vaccine efficacy.

TB is the leading cause of death among people living with HIV, affecting the immune system and eventually waning defense systems against infections, leading to an increase in the risk of TB. It is well-reported that people living with HIV are more than 20 times more susceptible to developing active TB. Therefore, protection against these two diseases is of complementary importance. Because of HIV-related immunosuppression, the TB vaccine may be less immunogenic and less effective in people with HIV infection than in people without HIV infection (214). Because HIV-infected people are a large subpopulation at a high risk of TB infection and disease, it is important to include them in TB vaccine trials. A vaccine that is expected to have a protective effect against HIV is being developed based on a promising vaccine candidate in the TB vaccine clinical stage or BCG vaccine (215).

Diabetes prevalence affects TB incidence and TB mortality, resulting in two to three times the probability of developing TB, two times the risk of death during TB treatment, four times the risk of TB recurrence after completion of treatment, and two times the risk of infection with multidrug-resistant TB (MDR-TB). A cohort study reported that the longer the period of diabetes was, the more associated with TB disease, and TB was more commonly identified in patients with a fasting plasma glucose level over 202 mg/dL (216). In addition, it was confirmed that the higher the glucose concentration in the blood of diabetic patients was, the weaker the adaptive immune response to Mtb (217). Verma et al. established a latent TB infection mouse model and induced diabetes in Mtb-infected mice by administering streptozotocin to investigate the relationship between latent TB and diabetes. These hyperglycemic conditions led to a decrease in MCP-1 and MMP9 levels and increased MMP1 levels in latent TB infection, which may lead to reactivation of latent TB infection by disrupting granulomas (218). Clement et al. reported that metabolic stress caused by hyperglycemia decreases Ag presentation ability and inhibits the proliferation of CD4⁺ T cells (219). These reports suggest the possibility that diabetes can affect the formation of TB vaccine-induced protective immunity.

In old age, lung structural degeneration as well as changes in immune cell functions make people vulnerable to respiratory diseases, and these age-related immunological changes may also affect vaccine efficacy. The incidence of TB is common in elderly individuals and increases progressively with age, and mortality from TB is also higher in older patients (220). This phenomenon is related to the reactivation of lesions from a dormant state, which is affected by changes in the immune system with aging. In addition, chronic inflammation in aging individuals disrupts T-cell responses, followed by decreased vaccine efficacy. For example, the application of a delayed-type hypersensitivity model of BCG vaccination and TST of aged NHPs showed that the immune response to antigenic challenges between the tissue site and the periphery is compromised, restricting the optimal immune memory response (221). A follow-up study showed reduced or delayed T-cell recall responses to lung infection sites in aged BCG-vaccinated rhesus macaques (222).

Recently, nonspecific protective efficacy of the BCG vaccine was confirmed against respiratory diseases such as COVID-19 through immunological changes favorable to respiratory infections in elderly people (223, 224). Many findings reveal that this nonspecific protection is generated from innate immune memory via metabolomic and epigenetic reprogramming, also known as trained immunity. Blood samples before and 1 month after BCG vaccination were compared in 82 subjects between the ages of 60 and 80 years (225). It was confirmed that BCG vaccination induced reductions in the levels of pro-inflammatory cytokines (TNF-α, IL-6, IL-1β) and chemokines (CCL2 and CXCL10), acute phase proteins such as C-reactive protein, and matrix metalloproteinases (225, 226). Considering the immune activation by BCG vaccination in elderly people and the positive results of BCG revaccination, BCG revaccination in elderly people may be a beneficial strategy to reduce elderly mortality due to TB.

TB vaccine candidates currently in the clinical stage are vaccinated through the intramuscular (IM) route or ID route. In addition, studies with noteworthy results in preclinical stages through the aerosol route or the intravenous route have recently been reported. However, studies on oral route vaccination are still limited. In the case of BCG, since the safety of the ID route of BCG for mass vaccination was confirmed by Scandinavian researchers in the 1930s, it has been used until now (227). However, the BCG vaccine was initially developed as an oral vaccine and was used in that form until an incident in Germany in 1930, when the oral BCG vaccine was contaminated with Mtb. In Brazil, oral BCG vaccination was administered to newborns until 1976. Recently, Hoft et al. demonstrated the safety of oral BCG vaccination through a comparative analysis of 68 healthy adults who received BCG via the intradermal (ID) route and the oral route (228). ID-BCG vaccination induced a higher systemic Th1 response than the oral route. In contrast, oral route BCG vaccination produced more elevated Mtb-specific secretory IgA and Mtb-specific bronchoalveolar lavage T cell responses than ID-BCG vaccination (228). A lipid-based formulation has been developed for oral BCG vaccination, and the results of vaccination in BALB/c mice showed increased vaccine efficacy compared to conventional BCG vaccination (229). In addition, to increase the efficacy of oral vaccination of lipid-formulated BCG vaccination, improved vaccine efficacy was confirmed through aerosol infection after oral vaccination with recombinant BCG expressing Ag85B-ESAT6 fusion protein in a guinea pig model (230). The effectiveness of oral vaccination was also confirmed with a subunit vaccine model. Although oral immunization was less effective as a priming vaccination of fusion protein ESAT-6-Ag85B with detoxified monophosphoryl lipid A (MPL), heterogeneous priming and boosting vaccination strategies combined with oral boost induced significant systemic Th1 response, providing protection similar to or exceeding vaccination via the SC route against Mtb infection (231).

Oral vaccination is an appealing route due to the absence of needles, which eliminates the risk of cross-infection, and the ability to administer vaccines without the need for specialized healthcare professionals. Exploring the properties of this unconventional vaccine route presents an additional potential strategy for TB vaccination.