The only licenced vaccine against TB, Bacillus-Calmette Guérin (BCG), provides good protection against severe forms of infant TB, but highly variable efficacy against pulmonary TB and therefore limited impact on disease transmission. Ongoing challenges also exist in the accurate diagnosis of both TB infection and active disease, increasing drug resistance and treatment burden even for fully sensitive disease (10). Despite huge research efforts, developments in all of these areas are hampered by gaps in our understanding of intricate host-pathogen interactions, the complex spectrum of disease states that cannot be replicated fully in animal models and lack of defined immune correlates of protection (CoP). Judicious use of a mycobacterial CHIM could help facilitate advances in many of these domains, as a complement to animal and field studies (11). For example, a mycobacterial CHIM could enable the prioritisation of vaccine candidates that most effectively control mycobacterial growth, prior to larger, more costly field efficacy studies. Samples from such a CHIM could also be used to interrogate immune parameters that correlate with control after a defined timepoint infection, with any positive steps towards finding a validated TB immune CoP proving potentially transformative.

Intentionally infecting humans with virulent Mycobacterium tuberculosis (M.tb) would not be ethical, with the potential for significant morbidity and mortality. Even if these are avoided, long treatment duration with the risk of significant drug side effects, risk of M.tb transmission to others, inability to prove cure at the end of treatment and possibility of disease recurrence are all substantial arguments against a CHIM with wild-type M.tb. Therefore, researchers must pursue the use of alternative challenge agents (Summarised in Table 1 ), aiming to address key scientific questions with an acceptable risk profile to both volunteers and the wider community.

The tuberculin skin test (TST), where tuberculin Purified Protein Derivative (PPD) is injected intradermally, has traditionally been used as a diagnostic test for latent TB infection (LTBI). It has been employed as a challenge agent to investigate immunological responses to mycobacterial antigens at the site of skin challenge, for example identifying exaggerated Th17 responses in those with active TB disease as a potential target for host directed therapies (12, 13, 24). PPD has also been used to assess local respiratory mucosal responses following intrabronchial instillation (14, 15, 21). Whilst these methods may contribute to our knowledge of mycobacterial immunopathogenesis they cannot be utilised directly to assess efficacy of vaccines or therapeutics.

A CHIM that is to be used to evaluate vaccine efficacy requires a live replicating organism, for example an attenuated strain of mycobacteria. BCG itself is such a live attenuated mycobacteria, initially derived via passage from Mycobacterium bovis (M. Bovis), that does not cause disease or latency in healthy humans (25). The loss of key virulence genes encoded in the Region of Difference 1 (RD1) during this process confers the advantageous safety profile of BCG but means the full immunopathogenic pathways of M.tb are not entirely replicated and it could not be used to evaluate vaccines which incorporate RD1-encoded antigens, such as ESAT-6 and CFP-10. However, BCG has been shown to induce similar canonical CD4⁺ T cell-mediated immune responses to M. tb in humans (26) and assessment of vaccine efficacy using a BCG challenge in animal models are comparable to results obtained using M.tb as the challenge agent (27, 28). BCG manufactured under good manufacturing practice (GMP) conditions for human use is readily available and this therefore represents the only live replicating TB CHIM agent currently available (11).

CHIMs using intradermal (ID) BCG as a mycobacterial challenge agent have been developed and are able to detect a known BCG vaccine effect in animals and humans (17, 18, 27–29). The ID route allows straightforward quantification of mycobacteria from an easily accessible site, for example via minimally invasive punch skin biopsies (17). However, the natural route of M.tb infection is via the respiratory tract and initial pathogen interactions with the specialised host respiratory mucosal system cannot be evaluated using an ID CHIM.

Efforts are ongoing to develop pulmonary CHIMs that more closely mimic the natural route of M.tb infection. BCG delivered via aerosol (Clinicatrials.gov NCT02709278, NCT03912207, NCT04777721) or instilled directly into the lungs per bronchoscope (21) are both being evaluated and have been shown to be safe and well tolerated. A defined timepoint pulmonary mycobacterial infection would allow examination of localised mucosal immunology and the relationship to induced system responses, which are key areas of research interest. Vaccines or therapeutics tested using these CHIMs would have the advantage of accounting for the contribution of the specialised respiratory mucosa in conferring protective immunity. However, sampling of the respiratory mucosa for immunological interrogation and quantification of recoverable BCG in pulmonary models are both more complex and invasive than in skin models (30).

Following on from initial studies using BCG, live mycobacterial CHIMs could be enhanced by the use of rationally attenuated genetically modified organisms. BCG which has been modified, for example to include a fluorophore reporter gene or exhaled volatile compound detectable by mass spectrometry could reduce the need for invasive sampling for mycobacterial recovery and quantification (31, 32). Use of current live vaccine candidates such as MTBVAC, a rationally attenuated form of M.tb (33, 34) or VPM1002, a recombinant BCG (35, 36), could allow investigation of the antigens or immunological pathways missing from BCG.

Whilst it is some way off from clinical evaluation, efforts are underway to develop a conditionally replicating M.tb strain with a genetically inserted suicide mechanism. This would aim to recapitulate the initial immunological mechanisms of M.tb, whilst ensuring complete eradication at a predefined timepoint and, if successful, could hugely advance the field of human TB study (22, 31).

Finally, for a TB CHIM to be truly useful, it should be safe, acceptable and deliverable in TB endemic populations and settings. Different environmental exposures, level of nutrition, microbiome composition, prior exposure to mycobacteria and prevalence of co-infections are just some of the known factors impacting vaccine efficacy. Utilising an ethically appropriate CHIM in endemic settings would ensure vaccines are tested in relevant populations (11, 37).

Early in the COVID-19 pandemic, the World Health Organisation (WHO) acknowledged the potential benefits of a SARS-CoV-2 CHIM, for example to allow rapid prioritisation of vaccine candidates. A working group was promptly established to consider the practicalities, feasibility and ethics (38). Initial expert consensus was divided with concern about the lack of a suitable “rescue” therapy, potential for severe illness and high transmissibility, as well as the benefit and applicability of such a model over field studies (39).

Accruing data suggested that infection of young, healthy adults in whom disease was generally much milder could be justifiable. This prompted UK manufacture of a challenge virus under GMP conditions and development and rigorous ethical review of study protocols for both a UK SARS-CoV-2 naïve CHIM (NCT04865237) and one in previously infected volunteers (NCT04864548) (40). GMP manufacture of challenge viruses is a time-consuming process and enrolment did not commence in these studies until March (NCT04865237) and May, 2021 (NCT04864548) respectively, by which point several highly efficacious vaccine candidates were being deployed in the UK population (41, 42).

Despite the widespread availability of highly effective vaccines against SARS-CoV-2, there remains a justifiable role for SARS CoV-2 CHIMs. A clear advantage of a CHIM over natural infection field studies is the known-timepoint of infection; allowing the detailed characterisation of both viral kinetics and the host immune response post-exposure. The dose of virus can also be carefully controlled and adjusted, providing crucial information about how the infectious dose affects the clinical and immunological response to the virus. Importantly, CHIMs also allow the collection of pre-exposure samples. These baseline samples can be assessed against clinical outcomes to identify immune correlates of protection (CoP).

Whilst current literature clearly defines the role of neutralising antibodies (nABs) as a correlate for sterilising immunity against SARS-CoV-2 (43–46), emerging evidence, particularly with the evolution of Variants of Concern (VoC) that escape nABs, is that the immune response to SARS-CoV-2 is more complex. Cell-mediated immunity, memory B cells and non-neutralising Fc-mediated effector functions may all play a role (47–53). Local mucosal immune responses have demonstrably protected against infection from other respiratory pathogens (54, 55) but mucosal immunity against SARS-CoV-2 remains poorly described in the literature. A CHIM with infection at a controlled timepoint allows the detailed interrogation of all aspects of the protective immune response, particularly the early host mucosal responses that are often missed in natural field infection studies.

Furthermore, the ability to control confounders such as inoculum strain, route of exposure, viral load and patient heterogeneity in a CHIM allows direct comparison of vaccine and therapeutic candidates as well as dosing regimens. With the roll-out of successful vaccines, it is unfeasible and unethical to maintain an unvaccinated placebo group for the testing of new vaccine candidates. Non-inferiority trials require large sample sizes and sufficient naturally acquired infection which can be time consuming and expensive. A CHIM could be of particular use in assessing novel vaccines, including those developed to be mucosally-delivered, which may have differing end-points (such as prevention of infection or viral shedding) that would be extremely difficult to study without a defined timepoint of infection. Whilst field studies are considered gold standard for vaccine licensure, there are instances where CHIMs have been used directly as proof of efficacy (4).

To date there are three registered SARS-CoV-2 CHIMs (Summarised in Table 1 ). The wild-type (pre-Alpha) SARS-CoV-2 CHIM in healthy, seronegative, UK 18-29-year olds demonstrated infection in 53% (18/34) of volunteers using a low inoculum dose of 10TCID50 (50% tissue culture infectious dose). Challenge was safe and well-tolerated with no evidence of lower respiratory tract involvement, although smell disturbance was common and prolonged in a small number of volunteers (23). Killingley et al. were able to accurately delineate the viral kinetics of primary infection and identified differences in viral dynamics depending on swab site. Viable virus measured by focus forming assay (FFA) persisted for on average 10 (maximum of 12) days, consistent with pre-Alpha isolation guidance (23). FFA was shown to closely correlate with lateral flow antigen (LFA) tests performed on the same swab samples. This first in human SARS-CoV-2 CHIM has demonstrated the broad utility of CHIMs, strengthening confidence in the public health measures (such as isolation periods and use of LFA tests) employed in the UK. Exploration of immune correlates of protection in this seronegative cohort, such as cross-reactive responses from seasonal coronaviruses, is ongoing.

With increasing global seroprevalence to SARS-CoV-2 from vaccination and/or infection (56), a seropositive SARS-CoV-2 CHIM is needed in order to facilitate future vaccine and therapeutic development in volunteers that reflects real world immunity. Successfully establishing a re-infection model additionally allows the identification of both local and systemic immune markers attained via the infection or vaccination process that are protective against re-infection, which could inform future public health strategies as well as design of therapeutics and vaccines.

Ongoing use of a pre-Alpha strain for a seropositive CHIM has several potential issues. Field data suggests that acquired immunity (either by vaccination, natural infection or both – hybrid immunity) offers strong resistance to homologous re-infection (57, 58). Achieving consistent infection rates may therefore prove more difficult than in a study of naïve participants.

Much of knowledge of re-infection rates was obtained prior to the emergence of variants such as Delta and subsequently Omicron, which are known to escape immunity. Both variants have antigenic divergence due to mutations in the spike protein and have been shown to demonstrate reduced neutralisation titres compared to pre-Alpha strains in vaccinated and hybrid cohorts (59–62). One approach which may circumvent any difficulty achieving infection in seropositive volunteers is to use variants more likely to cause breakthrough infections as the challenge agent, such as the Delta variant (isrctn.com ISRCTN94747181). Manufacture of an Omicron challenge agent is also being pursued (63).

There are pros and cons to the use of Delta or Omicron in a CHIM. Neutralisation against the Omicron variant is more markedly reduced than delta and associated with a higher rate of breakthrough infections (47, 61, 64) making it plausible that it would be easier to achieve infection in a CHIM. Omicron may also be a safer challenge agent demonstrating milder disease severity and reduced lower respiratory tract disease (65–68). However, the shorter infection course seen with the Omicron variant may also make it difficult to assess post-infection therapeutics (69).

Studies using currently prevalent variants are arguably more relevant both for the development pipeline of vaccines and therapeutics and understanding CoP. Limitations to this approach are that manufacturing a new challenge strain under GMP conditions takes at least 6 months (70). Furthermore, any specific clinical risks of that variant need to be understood from real world data prior to use in an ethically sound CHIM. The high incidence of SARS-CoV-2 and associated viral replication globally has resulted in the relatively rapid acquisition of mutations and development of new VoCs, meaning that by the time an inoculum strain is ready for use in a CHIM it may no longer be the dominant variant in the real world. However, developing several CHIMs that use variants derived from different lineages will enable broad assessment of different therapeutics and vaccines.