MTBC is composed by a broad group of mycobacteria species characterized for its genetic proximity and its pathogenic ability of affecting humans, such as M. tuberculosis and M. africanum, and a wide variety of wild and domestic animals, such as M. bovis and M. caprae. M. bovis stands out for being the primary etiological agent responsible for bovine tuberculosis (bTB), also considered as the main cause of animal tuberculosis due to the multi-host character of this bacterium (45). Animal tuberculosis is a zoonotic disease with great impact on public health, agriculture, wildlife, and trade areas (20, 46). In this sense, although most cases reported as human tuberculosis are caused by M. tuberculosis, ~30% of these cases are related to M. bovis infection (zoonotic tuberculosis) (28), especially in developing countries (47) where prevalence of livestock bTB becomes substantial (48–50). Despite huge efforts that are currently focused on the eradication of bTB, there are many difficulties mainly associated with the performance of the different diagnostic techniques as well as with other geographical and epidemiological conditions, which make it very difficult in endemic countries (51). Therefore, zoonotic tuberculosis is often under-reported, emphasizing the importance of providing appropriate diagnostic tools in livestock to reach the eradication of M. bovis and reduce zoonotic tuberculosis cases.

MAP is the causative agent of paratuberculosis (PTB) or Johne's disease, a chronic infection that affects the small intestine of ruminants resulting in a marked reduction of animal productivity (31) and sometimes in death (1). MAP is also believed to be related to Crohn's disease, a chronic bowel disease in humans, although this fact is yet to be defined (52–54).

The main importance of PTB comes from the great economic losses in animals due to reduced milk and meat yields as well as slaughter value (32). MAP diagnosis becomes a challenge because of its pathogenesis: while the main clinical signs are only present in the late progression of the disease, when the body condition is severely affected, the animals intermittently spread bacteria during a previous subclinical phase. These features result in a low sensitivity of the current direct (fecal culture and genome detection) and indirect (specific antibodies detection) diagnostic methods (55, 56) (Table 2). Hence, reliable and complementary diagnostic methodologies are of key importance to enhance the current diagnostic repertoire of techniques focused on the identification of infected animals so as to improve the sensitivity of the diagnosis.

Field and ante-mortem surveillance tests against MTBC infection are mainly based on the detection of a delayed-type hypersensitivity response to the intradermal skin test (IST) through the inoculation of purified protein derivative from M. bovis (bPPD; tuberculin protein), and on quantifying the concentration of gamma interferon (IFN-γ) after culturing blood samples in the presence of tuberculin, in the case of IFN-γ assay test, a supplemental or confirmatory test (25). IST is the OIE prescribed test for international trade and is currently considered as the official diagnostic screening technique in many countries worldwide; it is the primary ante-mortem test to support control and eradication programs in different geographical areas (46), responsible for its effectiveness as it is compulsory in the slaughtering of those animals with a positive result (64). In Europe, the aforementioned information is regulated by the Council Directive 64/432/EEC. Although IST and IFN-γ assay have reasonable sensitivity and good specificity (Table 2), both techniques require a minimum of 48–72 h to obtain a result (21, 65) besides presenting other disadvantages and limitations. On the one hand, IST requires visiting the farm and restraint of the animals twice, and a delicate and difficult administration and interpretation of skin results, which may vary due to differences in tuberculin doses, site of application (Table 2), and interpretation schemes (25, 27, 66); on the other hand, IFN-γ assay implies a complex laboratory methodology (18), a considerably more expensive price than a skin test (65, 67), and suffers from cross-reactivity with other related mycobacteria resulting in false-positive results (2). In addition, performance of these tests can be compromised by factors associated with the immune response and health status of the animal leading to a misinterpretation of the results. Development and use of a pre-screening test before field tests would be useful to reduce work efforts and diagnostic time (27).

Although microbiological culture is considered the gold-standard approach for the diagnosis of mycobacterial infection, it is characterized by a long incubation time to confirm the presence of mycobacteria (around 8–12 weeks) (2). Furthermore, the isolation of mycobacteria sometimes requires specific compounds such as mycobactin, a siderophore which determines the viability and growth of some mycobacteria species, as is the case of MAP; and, sometimes, additional steps such as decontamination. For all these reasons, culture becomes a tedious and laborious, although necessary, option in M. bovis diagnosis.

Other in vitro assays, such as serologic assays (ELISA) or PCR, have limitations associated with accuracy and execution that restrict their use (66). While ELISA sensitivity is affected by the delayed and irregular antibodies response in bTB (68), PCR is considered a postmortem diagnostic option with promising findings but still under development, focused on the search of markers that ensure diagnostic sensitivity (61) (Table 2). Therefore, the reliability of these tests depends on the stage of infection and, in addition, these require transporting of animal samples to the laboratory, which finally increases diagnostic time too (40), highlighting the interest on the availability of portable equipment.

The intermittent and sometimes low shedding of the mycobacteria as well as the irregular seroconversion in the subclinical phase of PTB (69, 70) gives a limited sensitivity to the in vivo diagnosis (32), currently based on serological assays (ELISA) and PCR from feces. Although ELISA has a limited sensitivity (Table 2), the irregular spread of bacteria via feces has raised serology as the most common technique used for the monitoring of PTB (71) due to its cheap and easy use. In addition, fecal shedding and immune response vary individually to a large extent (72). For example, the sensitivity of PCR methods can be affected by the variable bacterial load in samples and the co-purification of PCR inhibitors during DNA extraction (73). Therefore, and although the current combination between serology, vaccination (when regulation allows this option due to its possible interference on bTB eradication campaigns), and slaughtering constitutes a strategy with remarkable effectiveness, there is a need for diagnostic tests with higher sensitivity and decreased processing time to reduce false-negative results and enable effective disease control strategies, as different authors have highlighted before (31, 32). Volatilome evaluation has been capable of discriminating MAP infection before clinical illness occurs, offering an early diagnosis and significant time savings (1). This could be considered as one of the main advantages against the current techniques in use.

In short, against the current situation, it would be of help to have an ante-mortem diagnostic methodology capable of detecting mycobacterial infection with repeatability, a good quality/price ratio, high sensitivity and specificity, and rapid detection and obtaining of results. VOC strategy could be a complementary option because it mostly fulfills these requirements, and it has been successfully used for mycobacterial diagnosis in many animal species (Table 3). In this sense, an initial approximation analyzing stable air as matrix has been proposed to evaluate MAP infection in cattle (74); however, due to the low number of infected animals included within each infected group, further studies are required to confirm the suitability of this approach.

Analysis of VOCs as a diagnostic option for mycobacterial disease has been evaluated both in vivo and in vitro. Compared with in vivo assays, the large number of existing in vitro studies, which basically consist in mycobacteria culturing, reveals the early stage of development where this research area stands (36–38, 75, 76). The reviewed literature in the present study suggests some drawbacks related to those in vitro studies.

First, mycobacteria growth, which as aforementioned requires several weeks or even months, is required to identify changes in the analysis of VOCs from microbiological culture to allow the distinction between negative and positive samples. In other words, although in vitro experiments can detect VOC changes related to different stages of the mycobacteria growth (76, 77), it still takes a long time to identify these changes, which is one of the main disadvantages linked to the current diagnostic methodology. Accordingly, researchers point to reduce the diagnostic time by avoiding the limiting step of culturing and suggesting other innovative techniques such as VOC measurement directly in vivo (32).

It is also important to highlight the low correlation existing between results obtained from cultured bacteria compared with those VOCs produced from other biological samples studied in in vivo experiments (40). For example, (32) detected two compounds only present above MAP cultures which were ranked among the top discriminating VOCs in their statistical analysis. However, in the comparison with their in vivo results, these two compounds tended to be in lower concentration in MAP-inoculated animals compared with non-inoculated animals. This situation emphasizes the caution required when adopting in vitro findings to in vivo conditions since the influence from the host, its microbiome and host–microbiome interactions (78), as well as the influence from environmental factors, such as diet, age, or drug use (40), needs to be considered. In addition, another hurdle of the in vitro settings is related to the different VOC profiles obtained depending on the substrate where the mycobacteria grow resulting in inconclusive findings (79).

The effectiveness of in vivo approach is supported by the results of many studies where VOCs from biological samples have been used to distinguish between infected animals with different mycobacteria species and non-infected animals (1, 18, 31, 80). Many different biological matrices such as serum, breath, or feces have been studied as a source of information for VOC analysis in this field, existing great differences between their nature and characteristics. This constitutes another problem in the comparison between in vitro vs. in vivo experiments, giving inconsistent results. When (31) compared in vivo results obtained from feces and breath samples with the in vitro VOC profiles obtained from different MAP strains' culture by (77), their conclusions were not very clarifying: from more than 100 substances detected in feces and breath, only 15 and 5 of them, respectively, were found in the bacterial in vitro pattern.

Another variable to consider when evaluating VOCs as an option for mycobacterial diagnosis in animals is the type of infection: natural or experimental. Although experimental infections are logically the most common and easy option for this kind of approximation, studies in naturally infected animals are of paramount importance. Experimental infections allow controlling different environmental conditions that may impact on the results, being the most studied option in the analysis of VOCs for mycobacterial diagnosis (Table 1). However, assays with natural infections are needed to validate the results obtained from any new diagnostic tool, such as volatilome analysis, in experimental settings. Along this review, only a single article has been found to include the analysis of VOCs from both experimentally and naturally infected animals (23). These researchers found that differences between negative and positive animals were more pronounced in the natural infection group than in the experimentally infected one. This fact highlights the importance of performing studies in field conditions in the future to compare with those with experimentally infected animals and to validate the results from the latter ones.

The principle of using exhaled breath lies in its capability for discerning disease-related changes and biomarkers in the organism that are reflected into the breath through exchange via the lungs (25), because of its ability to cross the alveolar membranes before being exhaled (26). The use of exhaled breath offers several advantages because it is a non-invasive sample produced in ample supply, having the potential for direct, inexpensive, and eventually real-time monitoring (25, 90). Although in the literature it is considered as a sample that is relatively easy to obtain, its sampling methodology in animals is diverse, revealing a lack of standardization: from modified equine nebulization masks or nostril samplers for cattle, specific ventilators for mice or intubation for macaques, to automated alveolar sampling devices for goats. Furthermore, some factors can affect the sampling methodology, such as eructation in ruminants, which has been shown to significantly affect exhaled VOC profile (91). VOCs from breath are normally concentrated to sorbent materials, such as Tenax or Carbopack Y, Carbopack X, and Carboxen 1000 (18, 80), which simplify its transport and storage, and later these are used to quantify and evaluate the volatile substances with different analytical techniques.

Healthy and diseased animals have been successfully distinguished in mycobacterial infections by identifying volatile molecules in exhaled breath (Table 3). (18) performed breath collection and analysis in M. bovis–inoculated cattle with two strains obtaining good sensitivity and specificity: 83.8–96.4% and 97.4–99.2%, respectively, using the microbiological culture as reference technique. In addition, (25) reported the measurement of two VOCs from breath linked with M. bovis infection and other two VOCs associated with samples from negative individuals, obtaining sensitivity and specificity values of 100 and 79%, respectively.

The studies included in this revision evaluating exhaled breath in the context of mycobacteria infection highlight some important variables to be considered (41). The use of different animal species models, the Mycobacterium species and strain used, the infection phase, the breath volume collected, and the sorbent phases used to concentrate VOCs are factors that often differ between the existing assays. Considering all the aforementioned information, a comparison between the existing results is a challenge.

Feces are regarded as the most accessible sample for research (92). Considering that feces constitute the main media for eliminating metabolic products, these are an important source of information about the internal homeostasis (17). The reason for testing changes in VOCs in feces is based on the common assumption that any abnormality in the activity or composition of the intestinal microbiota and in the whole organism may alter the odor of this matrix (1), which is supported by studies from both human (93–96) and animal medicine (97–100). Consequently, examination of volatile fecal emission could be a very useful non-invasive diagnostic approach (1).

However, as a remarkable fraction of VOCs found in feces is generated by gut commensal microbiota (101), a well-matched control group and knowledge on these bacteria are necessary to identify VOC patterns of pathogenic conditions (31). Despite this shortcoming, using feces as matrix has many advantages; besides an easier sampling, it is not necessary to restrain the animals, eliminating the stressful situation that it implies. Moreover, and in contrast with human medicine, feces offer many different possibilities in terms of sampling protocol: per rectum, after sacrifice, using laboratory animal cages, or just after defecation are some options in veterinary research. The studies using feces reviewed in the present work highlight the existing heterogeneity between the published results (Tables 1, 2). However, the obtained results have placed fecal volatilome analysis as an innovative diagnostic approach in the current research context for mycobacterial infections. In this sense, attention has been focused not only on the discrimination between infected and healthy individuals (1, 13, 31, 32) but also in the use of fecal VOC profile for other purposes, such as identification of vaccinated animals in white-tailed deer (13) and cattle (28).

Serum is the sample of choice in many studies because of its relative ease to obtain and store, and safe distribution (24). Blood or serum is the means of transport of many different substances, compounds, and markers through the organism, existing a complex exchange with the lung or the intestine, among other systems (102). Alterations in VOCs from serum can be detected when a disease, an infection, or a pathologic condition occurs (103).

Serum has been used to distinguish the infection by M. bovis or MAP in different animal species through volatilome evaluation (Table 3) obtaining very interesting results. For example (30), were able to discriminate MAP and Brucella spp. infection in cattle through VOC analysis; and (27) reported an analyzing time of only 20 min to differentiate between bTB-infected and bTB-free bovine sera. However, and although blood and serum could be the most routine samples used in diagnostic field, its collection supposes a stressful situation as it is an invasive method that requires individual immobilization.

In conclusion, three different biological samples have been discussed as source of information in mycobacterial diagnosis in animals through volatilome analysis. Although interesting and useful findings have been shown, there is still a lack of homogeneity among many different study conditions. This often leads to non-comparable and inconsistent results. For example, despite studying the same pathogen (MAP), and using the same biological samples (exhaled breath and feces) and animal species (goats), contradictory conclusions can be found in the literature: while some showed that differences in VOC profiles were less pronounced from breath than those obtained in feces (31, 32), others suggest that volatilome evaluation from exhaled breath might be superior compared with the one from feces (1). In fact, the researchers usually acknowledge that their hypotheses should be verified by future studies, considering their findings as starting points (1, 32). Hence, no reliable comparisons or conclusions can be made with the available information, being advisable to carry out studies where the biological matrices are used simultaneously with the same methodological conditions.

GC-MS has become one of the most preferred options for marker identification of bacterial origin with a very good sensitivity (40). It has a huge potential for both identification and quantification of unknown VOCs from complex matrices (106, 107). Ellis et al. (18) found that 4-hydroxy-4-methyl-2-pentanone, benzaldehyde, 1-ethyl-2-pyrrolidinone, α,α-dimethyl-benzenemethanol, and nonanal were present in significantly greater concentration in M. bovis–infected animals than in control ones. Moreover, Bergmann et al. (31) found 16 and 3 VOCs in feces and breath, respectively, which provide detectable differences at any infection time between MAP-inoculated and non-inoculated animals. GC-MS has the capacity of detecting VOCs within a range of parts per billion range, or lower, with good reproducibility and linearity (22, 25). In other words, GC-MS not only seems to be the most suitable for bacterial biomarkers search; in fact, it is the most used technique to diagnose these infections (40). The present review highlights the usefulness of GC-MS as an analytic tool to evaluate VOC changes due to mycobacteria infection employing different biological samples such as exhaled breath or feces (Table 3).

In spite of these many advantages, GC-MS has also several drawbacks: most GC-MS equipment are still not implemented as a portable tool; it requires high levels of expertise, qualified personnel, and pre-concentration techniques; and it is currently an expensive instrumentation (40). Therefore, given the aforementioned cons and the significant sampling and analysis time that it implies, GC-MS is not suitable for being used in end-user or point-of-care sites (25, 40).

It is also worth mentioning that comprehensive two-dimensional GC-MS (GCxGC-MS) stands out for the possibility of analyzing VOCs coming from complex matrices (40) and for providing a more complex and unparalleled separation as well as three-dimensional chromatograms' visualization (108).

The electronic nose (EN) is an instrument based on chemical sensors combined with a pattern recognition system (109), able to detect different VOCs, such as odors, flavors, and vapors (110, 111). The main advantages of this methodology are the ease of use, its low price, and the rapid analysis time (27). Furthermore, EN methodology avoids sample transport to laboratory, positioning itself as one of the optimal techniques for pen-side use (27). However, it has problems with background separation, it does not identify substances detected, and sometimes its detection limit is high, giving insufficient sensitivity (31, 112).

The huge variety of applications where EN has shown effectiveness could be also considered as another of its strengths: versatility. In this sense, the reviewed information reveals the applicability of many types of EN sensors for different species of mycobacteria diagnosis through VOC identification (Table 3). Despite the good and interesting results obtained, the ease of transport of this device has not been exploited in depth because most studies using EN has analyzed VOCs from serum (Table 3) and not from other types of samples, such as feces or exhaled breath. The aforementioned information enhances the importance of carrying out future studies using EN focused on non-invasive biological matrices which would permit to develop a portable tool. In this sense (25), used their GC-MS results to tailor an artificial olfactory system to detect bTB in cattle exhaled breath. Although their new system successfully identified all infected animals (100% sensitivity), it wrongly classified 21% of the non-infected individuals (79% specificity).

DMS is an IMS modality that has been occasionally used for volatilome assessment in mycobacterial infections (Table 3). This instrumentation has a lower cost, and it can be used alone or coupled with a GC column which acts as a pre-separation stage (40). Its relatively low price, robustness, reliability, and miniaturization turn IMS technology into one of the potential alternatives for portable VOC analysis in disease diagnosis (40). As with EN, one of its main drawbacks is its lack of capacity to identify specific VOCs (1). This analytical device used by (1) permitted to discriminate healthy from MAP-infected goats, noting a direct correlation among postmortem findings and in vivo measurements.

SIFT-MS is a quantitative technique for trace gas analysis based on the ionization of these volatile compounds by positive precursor ions along a flow tube. Although its main advantages are a rapid analysis time and a lower mass range, biological samples usually provide complex data which need computational assistance to be analyzed (24). Spooner et al. (24) applied multivariate analysis for the first time to SIFT-MS data to evaluate serum headspace analysis as a faster screening tool for M. bovis infection in badgers, obtaining a much faster diagnosis. However, the insufficient accuracy achieved (88% of true positive and 38% of false positive) makes this approach unsuitable as an alternative for conventional diagnostic techniques.

The existing differences between the reviewed analytical techniques suggest the importance of using methodologies, such as GC-MS, as a first-line analysis, with the objective to identify and define tentative biomarkers. Then, other approaches, such as IMS or EN, could be developed and adapted to a field or point-of-care use.