Mycobacterium bovis infection and the resulting disease, commonly referred to as bovine tuberculosis (bTB), affects a broad range of species including humans, domestic animals, and wildlife (1, 2). Although M. bovis, as the name suggests, mainly affects bovids including cattle (Bos taurus), bison (Bison bison), African and Asian buffaloes (Syncerus caffer and Bubalus bubalis), it has been isolated from numerous other mammals, compromising animal and human health worldwide with reports of this species on all continents except Antarctica (3–5). Infection of wildlife further impacts livestock health due to the development of maintenance and spillover hosts within wildlife populations, including badgers (Meles meles) in the United Kingdom (UK), African buffaloes in southern Africa, and farmed and wild cervids in the United States (US) (6–8). In the UK and Ireland, M. bovis in badgers complicates bTB control efforts and similarly in the US, New Zealand, and Spain where deer, possum, and wild boar, respectively, are also recognized M. bovis reservoirs (4, 9). In the Americas, there are 15 wild species reported as infected with M. bovis (10). In South Africa, bTB is endemic in two major national parks, namely Kruger National and Hluhluwe-iMfolozi Parks (7). Moreover, M. bovis has been identified in more than 21 wildlife species in private and public sectors (11, 12).

In addition to the control challenges posed by infected wildlife species, bTB negatively affects wildlife-related industries, resulting in consequences for conservation, tourism, and game sales. Infection and disease in livestock and wildlife may lead to decreased productivity, trade restrictions, impacts on food security and zoonotic transmission, resulting in significant economic losses (5, 12). Bovine tuberculosis costs the global cattle industry alone 3 billion US dollars annually (13). In addition, ~5 million deer are farmed worldwide and bTB is the primary health threat across multiple species in this growing economic venture (14, 15). Despite effective bTB control measures in the US, states such as Michigan have endemic bTB in white-tailed deer (Odocoileus virginianus) populations that negatively affects the hunting and wildlife industry, in addition to the regular spillback affecting cattle (16, 17).

For these reasons, it is crucial to improve the detection, diagnosis and understanding of bTB across affected species to enable development of more effective and comprehensive control strategies. Early and accurate diagnosis of subclinical infection can inform more efficient management of affected animals. Alternatively, if associations can be accurately established between M. bovis infection or bTB disease and detectable host responses, limited available resources can be focused toward removing animals or populations that pose transmission risks to preserve those with higher economic, conservation or genetic value. However, the immune responses that arise from interactions between animal hosts and M. bovis are typically species-specific and particularly in the case of wildlife species, less well-characterized. Subsequently, validated diagnostic tests based on specific host responses to M. bovis infection and bTB are also scarce in wildlife.

The diagnostic standard for M. bovis detection, still in use despite being developed over a century ago, is the TST. The TST measures a host cell-mediated immune (CMI) response to mycobacterial purified protein derivative (PPD) antigens, either M. bovis-derived (PPD_b) alone or together with M. avium (PPD_a) for the single intradermal comparative tuberculin test (SICTT) (18, 19). In cattle, a meta-analysis of TST use in the UK and Ireland reported 100% specificity (Sp), in agreement with a Great Britain study displaying almost 100% Sp for standard and severe cut-off interpretations (20, 21). However, the sensitivity (Se) can range between 50 and 80%, with variations in Sp and Se between species and a lack of host-specific cut-off validation for all affected hosts (18, 21). A major confounding factor is the exposure to environmental mycobacteria in addition to vaccine strain antigens (for cattle in limited areas) that cause cross-reactions due to homology between antigenic peptides, demonstrated at gene and protein levels (19, 22, 23). In New Zealand, the TST displays reduced Se in deer exposed to environmental mycobacteria and is not recommended in herds with a high likelihood of M. bovis infection (14). When applied to fallow deer (Dama dama) in Texas (USA), the test had overall low Se and Sp, indicating minimal diagnostic value for this cervid species (14). There are also multiple cases of M. bovis crossing international borders in imported, infected deer due to negative TST results (14). In addition to variable test performance, the TST is subject to several limitations. The test is labor-intensive and logistically challenging in high income countries; in developing countries, this is exacerbated by restricted access to reagents and animals, veterinary capacity, and handling facilities (11, 24). An additional complication is management of cattle considered inconclusive reactors (IRs); in England and Wales, there were 3,755 IRs in 2015 alone (18). Animals in these herds may be infected, as suggested by the report that 21% of these herds had positive reactors when re-tested. In Ireland, between 11.8 and 21.4% of IRs slaughtered before a re-test were reported to be M. bovis-infected (18). However, movement of such herds is not typically restricted unless there is a recent history of TB, exacerbating M. bovis transmission risks when using the TST alone (18).

In attempting to address drawbacks associated with the TST in bovids, at least one ancillary test is recommended (25). One of the key focus areas of bTB research that has emerged is the discovery of new CMI biomarkers and consequently the development of CMI-based tests for the early detection of infection by pathogenic mycobacteria (26). Biomarker is a broad term defined as any indicator of pathogenic biological processes, either pathogen- or host-based (27, 28). The most common adjunct to the TST is the CMI biomarker-based interferon-gamma (IFN-γ) release assay (IGRA) (19, 29). However, despite improvements in Se for detecting M. bovis-infected animals, ancillary tests can introduce extra costs and logistical challenges. For example, IGRAs require that whole blood samples be processed within 8–24 h of collection, depending on the stimulation method, which is impractical if herds are located far from available laboratories (24). Moreover, IGRAs rely on a single measurement that is unable to discriminate M. bovis infection from bTB disease and further, may fail to diagnose early infection when blood is stimulated with more specific mycobacterial peptides instead of PPDs (29, 30). To overcome this, a combination of biomarkers may be used to improve diagnostic Se. For example, progress has been made in enhancing the detection of bTB infection in cattle by combining IGRAs with the simultaneous measurement of antigen-specific interleukin-1β (IL-1β) or IFN γ-induced protein 10 (IP-10) (29–31). A systematic review by Domingos et al. (10) of M. bovis diagnostic methods in wildlife species also suggested the application of at least two tests to aid the detection of infection and potentially different disease stages.

A primary goal for accurate M. bovis detection is the early identification of a maximal number of infected animals for more effective control measures. The investigation of candidate biomarkers for M. bovis infection and bTB is ongoing, as discussed below. However, if current biomarker research could be further advanced to differentiate infection from bTB disease, this would enable a focus on detecting stages in which the animal is potentially infectious, improving the elimination of major transmission risks or preventing infected individuals from transitioning to a diseased state. The primary ante mortem diagnostic tools in livestock and wildlife rely on the development of antigen-specific CMI responses to M. bovis (30, 32, 33). Therefore, this review will focus on current understanding and knowledge gaps regarding the host response to M. bovis infection and the use of host CMI biomarkers for the improved understanding and diagnosis of M. bovis infection and bTB disease.

With the availability of bovid host and M. bovis genomes, the understanding of host responses to M. bovis has continued to develop (24). Immunity, in brief, comprises the innate and adaptive immune systems. The adaptive immune response has two distinct components, namely humoral and cell-mediated immunity, of which the latter is vital for host protection against M. bovis infection. All the major T lymphocyte subsets have demonstrated involvement in the anti-mycobacterial immune response in cattle (34). In M. bovis-infected cattle, CD4⁺ T cells produce IFN-γ for activation of macrophage anti-mycobacterial functions whereas CD8⁺ T cells are involved in the lysis of infected cells (34). The development of a T helper type 1 (Th1) -based CMI response, resulting in production of cytokines and chemokines such as tumor necrosis factor-α (TNF-α), IL-12, IL-6, and IFN-γ by dendritic cells and macrophages, is considered essential for the control of mycobacterial infection (35). Cytokine production is activated by pathogen antigen presentation to immune cells, and disease progresses when these pathways are disrupted (36). Cytokines play a critical role in determining the characteristic immune responses to infection; the balance between Th1 and T helper type 2 (Th2) functions is often studied from the perspective of their cytokine profiles, each of which fluctuate throughout disease progression (35). Cytokine function can vary from pro-inflammatory, which promotes activation of cells to kill mycobacteria, to anti-inflammatory, which reduces the pro-inflammatory response to control and prevent tissue damage by necrosis (37). In the primary stages of mycobacterial infections, Th1-dominant cytokine responses develop (36). The Th1 immune response is critical for defense against intracellular pathogens, most notably through the production of IFN-γ (34). However, a shift from Th1- to Th2-based responses is typically observed as disease develops, with a shift in CMI responses along with increasing humoral (B cell) responses (38). The Th2 response is associated with the production of cytokines including IL-4, IL-5, and IL-10 with studies suggesting that the Th1/Th2 balance is critical for determining bTB disease progression and outcomes (39). Therefore, any changes in immune responses during the course of infection are likely to affect the diagnostic performance of immunological assays.

Although early detection of M. bovis infection is primarily dependent upon the measurement of CMI responses in many species, humoral responses have also been used to diagnose bTB (35, 40–42). In certain species, for example elephants (Loxodonta africana and Elephas maximus) and suids, anti-mycobacterial antibodies can be detected during early infection (43–46). In species such as cattle and lions (Panthera leo), on the other hand, antibody detection usually occurs only after bTB disease progresses and the immune responses shift toward a Th2 profile (41, 47). However, the application of serological tests requires insight into the humoral immune profile of specific species to determine the differences in immunodominant antigen recognition and response kinetics (48, 49). Due to a paradigm that early immune responses are cell-mediated, this has been the primary focus of bTB research in animals to date although further studies on comparative immunobiological responses in different species are warranted.

In cattle and some other species, a CMI-dominated early and specific response to M. bovis infection is observed with proliferation of antigen-specific T lymphocytes, secretion of the regulatory cytokine IL-2, and release of pro-inflammatory cytokines including IFN-γ. These responses have been reported, for example, in two studies by Corner et al. (50) and Gormley and Corner (51) that investigated CMI responses to M. bovis in experimentally infected badgers, one of the few relatively well-studied non-bovid species. The earliest CMI immune responses were observed 3 weeks post infection (WPI) in the animals infected with the highest M. bovis dose; these animals also demonstrated the most consistent CMI responses of peripheral blood mononuclear cells (PBMCs) to M. bovis PPD (PPD_b) stimulation. The overall CMI responses were also positively associated with pathological changes, i.e., the presence of gross lesions, detected post mortem. On the other hand, the antibody responses in badgers were only sporadically detected.

Functional genomics and proteomics have enabled a better understanding of CMI responses to M. bovis infection (52). Studies typically focus on the in vitro assessment of gene expression profiles in different cells or tissues, isolated after infection with M. bovis or related strains, and ex vivo analyses aimed at identifying cytokine gene expression signatures for the detection of M. bovis-infected animals. Wedlock et al. (53) compared gene expression in primary bovine alveolar macrophages (AMs) infected with a virulent M. bovis strain and an attenuated version. Although both strains grew at comparable rates, the results suggested a 45% difference in gene expression between the two infection groups. Of the 10 most differentially expressed genes, with the virulent strain inducing higher levels of expression, seven of these were chemokines with IL-8 and monocyte chemoattractant protein (MCP)-1 (CCL2) having the greatest expression. The conclusion was that AMs infected with virulent M. bovis displayed a more dominant pro-inflammatory gene expression profile than those infected with the attenuated strain. Another study using a similar method observed lower expression of chemokines from AMs of M. bovis-infected cattle than M. tuberculosis-infected cells, highlighting a possible mechanism employed by M. bovis to circumvent activation of the host's chemotactic response, thereby evading killing (54).

Although the host immune responses during M. bovis infection and disease remain less understood than for M. tuberculosis, recent studies have begun to shed more light on this complex pathogen-host interplay. However, susceptibility to infection, routes of infection and disease progression can vary significantly between the wide range of host species that are affected. Hence, it would be expected that immune responses and therefore, specific diagnostic biomarkers may also differ, impacting advances in host biomarker discovery and application (36).

A suitable diagnostic biomarker or multiple marker biosignature, for infection or disease, is a host- (or pathogen-) specific molecule/protein that is associated with the underlying pathological process (28, 55). Identification of biomarker-based assays that can enable more accessible, affordable, and efficient diagnosis has been a high priority for TB and bTB diagnostic research in recent years (28, 56). Considering the World Health Organization's target product profiles, that define the performance and operational characteristics of suitable tests, it is expected that non-DNA markers are more likely to meet practical and cost targets (28, 57). Non-DNA methods can be performed without advanced instrumentation, utilize more easily accessible samples such as blood and serum, and are generally more affordable than DNA-based tests (27, 28). Hence, the focus of current bTB diagnostic research is host biomarkers with an emphasis on cytokine/chemokine (chemotactic cytokine) proteins and antibodies, in addition to the detection of cytokine RNA expression (5, 24, 58).

Critical barriers to progress regarding biomarker discovery include the lack of standardization in specimen collection methods and a reliance on convenient or opportunistic samples, in addition to inadequate handling and storage of samples which can drastically alter detectable biomarker levels (51, 59). Another common limitation is suboptimal statistical power due to low sample numbers, hence, associations between biomarkers and potential disease states are challenging to investigate and prove (59). A biomarker should address a definitive question related to a particular state, or the prognosis, of pathogenic mycobacterial infection in individuals or populations (55). However, the stages of TB that have been elucidated from human research are complex and include infection, incipient or subclinical or active disease, and latency; moreover, these states are not regarded as fixed but instead constitute a dynamic spectrum that may be affected by multiple factors, many of which remain poorly understood (55, 60).

For M. bovis infections in animals, insight regarding the mycobacterial disease spectrum is even more limited (36, 51, 60). In general, M. bovis infection in animals is often not detectable until disease is advanced. In addition, clinical signs are often non-specific such as weight loss, decreased milk production (in dairy cattle), and occasionally coughing or other respiratory signs (11, 61). Chronic disease can progress to mortality over the course of months to years. The hallmark lesion of bTB in cattle is the granuloma that can be found in the lung, lymph nodes and potentially other organs (62). Granuloma stages have been well-described in cattle and are based on morphological characteristics, i.e., degree of necrosis and mineralization, and presence of a connective tissue capsule (63). However, M. bovis disease development and pathological lesions can vary widely between species, with cavitary pulmonary lesions instead of granulomas in goats and gastrointestinal infection observed in predators, for example lions (64–66).

To date, the only method of determining an animal's true infection status is comprehensive post mortem examination and a battery of mycobacteriological, immunological, histopathologic, and molecular tests (19). Accurate ante mortem, in-field reference tests for infection and/or active bTB disease are lacking (56). Current understanding of the pathogenesis of M. bovis infection is also confounded by the multitude of susceptible animal hosts and the complexity of the host-pathogen interactions (36, 51). Although naturally-infected animals are useful in the investigation of differences between infection and disease, experimental models illuminate disease progression kinetics and immune responses initiated from a fixed point in time with a set dose, generating the insight necessary for diagnostic test development (67, 68). A major limitation for M. bovis and bTB diagnostic research is the lack of clarity regarding case definitions for infection vs. active disease and, whether latent infections exist (51, 60, 69, 70). Considering the various limitations outlined above, the fitness for purpose criteria [as defined by the World Organization of Animal Health (OIE)] are critical for the validation and application of diagnostic bTB biomarkers in animals and should be applied on an individual species basis.

The ante mortem diagnosis of bTB in animals primarily relies on the detection of host CMI responses, most commonly with the use of protein and gene expression assays (5). Although the most widely used ancillary test, IGRA, in addition to the TST, can produce false negative results as the CMI responses required may diminish in advanced disease stages (anergy), while the humoral response increases (35, 69). Furthermore, IGRAs demonstrate limited Se in early infection stages and in general, bTB diagnostic assays are unable to distinguish infection from disease (13, 21, 25, 58, 71). On the other hand, a recent study by Bernitz et al. (72) observed that levels of IFN-γ and IP-10 in incubated unstimulated whole blood were elevated in infected buffaloes with observable pathological changes consistent with bTB in comparison to uninfected controls. Furthermore, increased IFN-γ significantly correlated with increasing severity of pathological changes in the infected buffaloes, consistent with observations of associations between antigen-stimulated IFN-γ and bTB pathology in cattle and badgers, demonstrating the potential for these cytokines to be used as indicators of bTB disease (72–74).

A limitation of cytokine/chemokine protein assays can be the inability to detect high levels of the biomarker target due to, for example, binding saturation limits; therefore, mRNA responses and their detection may provide a more robust alternative (5). There are several studies that investigate cytokine/chemokine RNA expression to achieve diagnostic test objectives. However, another purpose of biomarkers is to improve the fundamental understanding of disease pathogenesis, which is often the focus of bTB biomarker RNA expression research, for example studies that characterize the immunological profile of M. bovis-infected cattle (experimental or natural) by measuring cytokine/chemokine mRNA expression at various time points during infection and/or disease (35, 39, 53, 55, 75–77). Immune response changes during bTB disease progression, particularly in terms of anti- and pro-inflammatory cytokine/chemokine profiles, can signify disease outcomes and hence play a role in the diagnosis and control of infection (35). The CMI response is the dominant immunological response to M. bovis (in most studied species), responsible for the killing and elimination mechanisms in addition to formation of characteristic granulomas against this intracellular pathogen (69). The following section will further describe some of the CMI biomarkers of infection and bTB disease in animals.

Several promising cytokine/chemokine biomarkers of M. bovis infection and disease in domestic and wildlife species have been revealed in recent years, in addition to the widely recognized host biomarker IFN-γ. These include, although are not limited to, IP-10, IL-1β, IL-4, IL-8, IL-17A, CXCL9, IL-10, and IL-22 (37, 40, 78–84). Both protein and gene expression assays have been used to assess biomarker levels in different species and have facilitated identification of additional targets for diagnostic development (5). This section will describe select current and candidate CMI biomarkers of M. bovis infection in domestic and wild animal species, as summarized in Figure 1 and Table 1.

This review describes host CMI biomarkers with diagnostic potential for the detection of M. bovis infection and bTB, with a focus on more recent research and knowledge gaps, especially as these pertain to wildlife species. The CMI response is a vital component of host adaptive immunity to M. bovis (34, 35). Cattle and badger studies have demonstrated the involvement of the major T lymphocyte types and production of CMI-based cytokines and chemokines during M. bovis immune response development, accompanied by a shift from the predominant pro-inflammatory, Th1-biased response to more anti-inflammatory, Th2 functions as infection develops (36, 37, 51, 94). With a focus on the observation and measurement of CMI responses, detected through changes in gene expression, protein release or both, promising diagnostic biomarker targets and their associated limitations for multi-species application have emerged.

The choice of CMI-based methods for diagnostic purposes is in part motivated by the early and specific response that is elicited in most M. bovis host species studied to date (51). The delayed type hypersensitivity reaction measured in vivo by the TST is mimicked in stimulated blood cultures and can therefore result in a higher Se and Sp, due to the controlled in vitro conditions and parameters, for M. bovis detection (43, 58). Additionally, the specific cytokines and chemokines, produced during the adaptive immune response, that are detected by these assays is a feature applicable to most (if not all) mammalian species, thus allowing for translational use across species. Further advantages of in vitro CMI assays include even earlier detection times of an immune response than the TST (as early as 1-week post infection), only a single immobilization or handling event to collect blood samples, and more potential for the standardization of tests and reagents without in-field variation and operator bias (69). Moreover, the cost differences between in vivo (TST) and in vitro CMI assays may be over-estimated and CMI tools could prove more cost-efficient than assumed, with further cost reductions anticipated by increased use allowing the scale-up of production, and the automation of assays (such as ELISA kits) that lower laboratory costs (13).

The most used cytokine biomarker for TB diagnostic assays is IFN-γ, a critical Th1 cytokine produced upon lymphocyte activation in defense against M. bovis (43). This cytokine is also vital for the formation and function of granulomas, a hallmark of the response to M. bovis in host species (94, 120). Compared with other cytokine assays to date, IGRAs are robust and comparatively easy to standardize, and the development of new IGRAs for multiple non-bovine species has been suggested for bTB screening and control (13, 120). The increasingly widespread use of IGRAs (such as Bovigam®) in cattle, due to US- and EU-approval as a TST adjunct for parallel testing in bTB eradication programs to increase Se, may also explain interest in its use for wildlife species. Although IGRAs have been successfully applied to several domestic and wildlife species, including but not limited to cattle, goats, cervids, buffalo, white rhinoceros, and wild dogs, there have been limitations encountered when using them in other M. bovis host species (43, 111, 113, 114, 137). Aside from technical challenges related to the lack of available diagnostic tests and reagents for most wildlife, this may also be due to species heterogeneity in predominant immune response pathways to M. bovis (45, 51). Although M. bovis was first strongly associated with cattle, this pathogen has adapted and evolved to infect a broad range of animal host species, which may present with shared or unique characteristics in their immune responses (4, 9, 120). Although less studied, there may be differences in the IFN protein structure or amount produced between species, resulting in decreased expression and thus detection of this cytokine as observed in lions and warthogs (83, 98). Another difference between species is disease susceptibility together with infection pressure in the specific environment, hence, some animals could be exposed to high pathogen doses leading to rapid disease progression, or lower levels resulting in sub-clinical infection (137). The disease state and dose-dependent immune response may also influence the patterns of cytokines detected, as observed in experimentally infected badgers that showed consistent CMI responses with a high dose of M. bovis yet those with subclinical presentation had weak CMI responses, although with no effects on the humoral response (51). This could explain the more efficient use of alternative biomarkers or methods observed in species such as lions and warthogs, in which IFN-γ detection appeared less optimal (45, 47, 83, 98). However, biomarker signatures associated with a subclinical bTB state in animals have not yet been investigated, in contrast to human studies (138). A final factor to consider is the influence of M. bovis strains on host immune responses. Although it is acknowledged that the virulence of pathogenic mycobacteria is linked to genetic and phenotypic characteristics, this has only been confirmed for M. tuberculosis, aided by in-depth research using animal models (120, 139). Hosts of M. bovis are typically outbred species, thus, discerning the influence of M. bovis lineage from host heterogeneity is challenging. However, virulence variations between strains have been reported that suggest potential correlations between strain genetic differences and bTB development (139, 140).

In addition to a dominant early response, changes in IFN-γ responses have been correlated to pathological changes, although an IGRA to distinguish between M. bovis infection and bTB disease has not yet been developed (84, 89). Considering the correlations observed between IFNG expression and granulomatous lesion development, the quantitative measurement of IFN-γ in different cohorts of defined bTB states from early infection to active disease could enable the development of an IFN-γ cytokine assay to differentiate M. bovis-infected from diseased animals. However, species specific validation would be critical due to observed differences in IFNG expression profiles during bTB progression, as observed between fallow deer and cattle (93, 94, 123). In light of the potential drawbacks to IFN-γ detection, including sample handling that necessitates a short period between blood collection and processing and variable test performance parameters depending on species and disease stage, alternative cytokines, and chemokines to IFN-γ have emerged as additional tools for M. bovis infection and bTB diagnosis.

Pro-inflammatory cytokines are released early after M. bovis infection and are thus expected to be promising candidates for biomarkers of infection due to the predominant Th1 response observed in several host species. In addition to IFN-γ, two major pro-inflammatory cytokines are IL-1β and IL-6, although not much more than a role in general mycobacterial infection is known for the latter. The former, IL-1β, has demonstrated high Se for M. bovis detection in cattle, particularly when used in parallel with an IGRA (5, 31). Although not explicitly pro-inflammatory, two biomarkers (IP-10 and IL-8) have been detected at higher concentrations than IFN-γ and have shown possible correlation with bTB progression in cattle and African buffaloes (5, 72, 84). Considering their robust response early after M. bovis infection, these biomarkers could be well-suited for detecting M. bovis infection in less studied species. Another cytokine that demonstrates robust levels from an early infection stage is CXCL9, responsible for CD4⁺ lymphocyte recruitment (5, 100, 101).

There are limitations associated with CMI-based diagnostics that include reduced Se due to anergy (reduced detectable Th1-biased CMI responses as bTB progresses), and interference from co-infection and vaccination (120). However, the application of a parallel testing scheme, whereby at least two different tests able to detect slightly different sub-populations of infected animals (i.e., animals at different stages of infection) are used together, has shown promise in combatting these drawbacks, particularly in M. bovis-endemic settings (69, 111). The demonstration of two dominant pro-inflammatory cytokines, IFN-γ and IL-1β, that increased Se without compromising Sp when diagnosing M. bovis infection in cattle, indicates that even slight differences in cytokine pathways and functions can provide a testing scheme to improve detection of the maximum number of infected animals. Moreover, for species in which IFN-γ detection is problematic, the detection of additional or alternative pro-inflammatory cytokines could aid the early detection of M. bovis. Therefore, considering the robust levels of IL-8 and its good agreement with both the TST and IGRAs, this marker could also prove a promising option for enhancing M. bovis detection. Another example of parallel application is the measurement of IP-10 with IFN-γ that has enhanced Se in buffaloes and cattle, with high IP-10 production having a particular advantage in very early stages of infection (29, 30). Moreover, Klepp et al. (25) demonstrated the utility of IL-22 for the diagnosis of M. bovis infections that both the TST and IGRA failed to detect, not unexpected from a cytokine marker that stems from a separate and recently described T cell lineage, Th17.

Infection with M. bovis typically results in an early Th1 bias that tends to decline as the Th2 response increases, with shifts from Th1 to Th2 responses showing correlations to increased pathology (94). Hence, whether the objective is optimum detection of all infected animals, the detection of highly infectious individuals (i.e., animals that are shedding M. bovis bacilli) or the distinction between infection and active disease, it is expected that a diagnostic algorithm would benefit from using a combination of pro- and anti-inflammatory biomarkers. Moreover, the immunological response to M. bovis, in terms of bTB disease progression, is a dynamic process and non-linear (77). Thus, different biomarker signatures could be used to identify bTB progression, as suggested by Blanco et al. (96), Kelley et al. (56), and Palmer et al. (5). One example, shown by immunohistochemical analysis of granulomatous lesions of lymph nodes and lung in cattle, is the combination of pro-inflammatory IFN-γ and IL-1β with IL-10 (94). The two pro-inflammatory cytokines demonstrated contrasting yet equally significant associations with granuloma development and lesioned vs. non-lesioned tissue, highlighting the diversity of individual cytokine functions. The study also observed a lack of IL-10 expression in advanced granulomas that correlates with previous findings of progressively decreasing IL-10 in cattle with severe pathology (76, 94). The anti-inflammatory cytokines IL-4 and IL-10, which typically present with inverse correlations to IFN-γ over the course of infection, have shown good potential in distinguishing disease states; thus, their inclusion is suggested for biomarker signatures that cover both early and late stages of infection and bTB, with the potential to enhance diagnostic Se or Sp particularly for enhanced detection of infected animals (37, 75, 78, 94, 95). A panel of biomarkers could also provide a promising method for wildlife species in particular, due to the measurement of varied immune responses from a single sample that would provide greater confidence in the animals' disease state without the need for repeat testing, a significant challenge for most wildlife testing. Alternative potential biomarkers of bTB disease progression include IL-17A, also of the Th17 subset, and CXCL9 due to the observed increase as infection progresses with additional correlations to lesion severity (25, 81). This highlights the utility of adding biomarkers from different T cell subsets in a bTB testing scheme and warrants their investigation in wild animal species, especially considering the infeasibility and lack of standardization of the TST outside of common domestic species.

Due to the recognized role of wildlife in the maintenance of M. bovis, research on the development of assays for wild species (partially or fully validated) is increasing. A significant challenge in the development of CMI biomarkers for M. bovis in wildlife is the difficulty in obtaining sufficient reference samples to investigate and validate diagnostic assays. The lack of validated ante mortem tests makes diagnostic tests even more challenging to perform under field conditions, particularly in endemic settings (19, 56). Furthermore, the procurement of gold standard reference cohorts remains a trade-off between obtaining reliable samples and the loss of valuable animals, particularly for negative cohorts, that are required to obtain data from post mortem examination and mycobacterial culture (43). There are statistical tools that provide alternatives to the use of reference standards, such as latent-class or Bayesian approaches; however, statistical methods require large sample sizes, among other limitations (19, 21, 71). In acknowledgment of these challenges, recent policy (in the form of additional chapters for the OIE Terrestrial Manual) was adopted between 2014 and 2016, including statistical approaches to validation, that extends to the validation standards for diagnostic tests to wildlife (Chapter 2.2.7., OIE Terrestrial Manual 2018). Two validation pathways are provided that allow provisional recognition of a test if the complete validation process is hindered. This agrees with the chapter's emphasis on fit-for-purpose assays, particularly as diagnostic testing objectives may differ between domestic animal and wildlife species. The chapter also recognizes the difficulties in obtaining wildlife reference standards (particularly negative cohorts), suggesting latent-class models for performance estimates in lieu of a perfect (gold standard) reference (141). On the other hand, the imperfections of the gold standard requirement for mycobacterial culture have been acknowledged more recently; hence, alternative definitions for reference cohorts that improve test performance parameters may also be applicable (13, 51, 69).

The successful use of IGRAs and other cytokine assays in wildlife depends on the availability (and costs) of suitable tests or reagents. However, commercial CMI biomarker detection platforms are being developed and optimized using common domestic species with unknown cross reactivity to wildlife species. Therefore, researchers may need to develop novel species-specific reagents and assays, although these would have a very limited market. However, if cross reactivity can be determined using commercial reagents, this has the advantage of ready access by other researchers without having to exchange reagents. Indirect ELISAs with cross-reactive reagents have facilitated their use in related species, such as the application of bovine ELISAs to African buffaloes (32, 111) or equine ELISAs in white rhinoceros (113). However, another viable alternative is cytokine gene expression assays that can be adapted or developed for use in closely related species and even extended to non-related species if the sequences are conserved, which has been demonstrated for cytokine sequences in wildlife species (26, 142). Gene expression assays have been used in less common species such as warthogs, hyenas, and lions (83, 98, 102). The development of cytokine and chemokine multiplex assays that detect several targets simultaneously in a single sample (of small volume), with options for customization, has also aided the development of biomarker signatures (13, 143). Recently, the MILLIPLEX® bovine cytokine/chemokine platform demonstrated detection of several novel protein markers in buffaloes, with the aim to enhance M. bovis detection and potentially differentiate bTB disease states with the use of multiple diagnostic markers in this species (112). Transcriptome analysis has also been conducted using in vitro murine bTB models and PBMCs from cattle to identify more comprehensive gene panels involved in responses to M. bovis, although correlations between signatures and a potential incipient vs. advanced bTB state remain an avenue for future research (80, 144). Considering the multiple challenges faced when validating a single new CMI-based diagnostic assay, including the limited funding and research on candidate biomarkers in wildlife, the lack of progress regarding multiple marker signatures is not surprising. Moreover, the costs and feasibility of performing these assays for a range of species should be considered when prioritizing the development of biosignatures for wildlife bTB diagnosis.

The success of management and control strategies for bTB is only as effective as the diagnostic assays it relies upon. The use of CMI cytokine and chemokine biomarkers has already improved insight on the comparative immunology of M. bovis-infected hosts, with experimental and natural infection studies conducted in cattle, badgers and cervids, among other species. Moreover, understanding the contribution and role of dominant profiles, such as the Th1/Th2 responses, to the development of pathology aids the identification and development of diagnostic biomarkers and biomarker panels. However, despite progress in understanding M. bovis-induced immune responses, the research and diagnostic biomarkers described here are still primarily restricted to cattle. In comparison to domestic species, limited resources are allocated to studies of bTB in wildlife, resulting in a paucity of information on M. bovis infection and disease development in other naturally infected hosts (43, 51). Hence, more species-specific research is required, together with the development of standardized, multi-species tests and reagents, and investigation of candidate biomarkers in alternative samples types, i.e., serum, due to the practical, in-field and diagnostic potential of circulating responses. CMI-based assays may also be further improved with the addition of enhanced, immunodominant antigens for stimulation (in addition to ESAT-6 and CFP-10) to increase test Se and Sp. Additionally, advances in techniques from multiplex biomarker detection platforms to powerful statistical approaches that estimate population characteristics when true disease status is unknown, or when logistical challenges prevent acquisition of gold standard reference cohorts, will further enable the validation of enhanced immunological tools in both domestic animals and wildlife.

KS, MM, and WG conceived the idea for the review. MM and RW provided funding. KS, MM, LK, WG, and RW contributed to the drafting and editing of the manuscript. All authors reviewed the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.