Tuberculosis (TB) is an increasing threat to wildlife, yet tracking its spread is challenging because infections often appear to be asymptomatic, and diagnostic tools such as blood tests can be invasive and resource intensive. Our understanding of TB biology in wildlife is therefore limited to a small number of well-studied species. Testing of fecal samples using PCR is a noninvasive method that has been used to detect Mycobacterium bovis shedding amongst badgers, yet its utility more broadly for TB monitoring in wildlife is unclear. We combined observation data of clinical signs with PCR testing of 388 fecal samples to characterize longitudinal dynamics of TB progression in 66 wild meerkats (Suricata suricatta) socially exposed to Mycobacterium suricattae between 2000 and 2018. Our specific objectives were 1) to test whether meerkat fecal samples can be used to monitor TB; 2) to characterize TB progression between three infection states (PCR-negative exposed, PCR-positive asymptomatic, and PCR positive with clinical signs); and 3) estimate individual heterogeneity in TB susceptibility, defined here as the time between TB exposure and detection, and survival after TB detection. We found that the TB detection probability once meerkats developed clinical signs was 13% (95% confidence interval 3–46%). Nevertheless, with an adapted test protocol of 10 PCR replicates per sample we detected hidden TB infections in 59% of meerkats before the onset of clinical signs. Meerkats became PCR positive approximately 14 mo after initial exposure, developed clinical signs approximately 1 yr after becoming PCR positive, and died within 5 mo of developing clinical signs. Individual variation in disease progression was high, with meerkats developing clinical signs from immediately after exposure to 3.4 yr later. Overall, our study generates novel insights into wildlife TB progression, and may help guide adapted management strategies for TB-susceptible wildlife populations.
Emerging infectious diseases in wildlife are increasing because of human encroachment into wildlife habitats, thereby facilitating transmission of pathogens between wildlife, livestock, and humans (Jones et al. 2008; Neiderud 2015). Monitoring wildlife disease and investigating pathogen epidemiology in wildlife is challenging, because it often requires long-term, high-frequency, and invasive sampling of animals. Cryptic pathogens that do not manifest in clinical signs immediately, or are undetectable during latent stages, present particularly complex and difficult cases, and these subclinical infections often play a disproportionately large role in pathogen transmission (Ten Bosch et al. 2018). For such pathogens, the absence of noninvasive diagnostic tests (e.g., from fecal samples) to detect infections in asymptomatic hosts without capture limits our understanding of variation in disease susceptibility and survival after infection across wildlife populations. This in turn hinders the development of epidemiological models and the identification of super spreaders (Woolhouse et al. 1997; Lloyd-Smith et al. 2005), and restricts management policies that may help prevent pathogen transmission within wildlife populations and potential spillover into livestock and humans (Wood et al. 2012).
Wildlife tuberculosis (TB) is a cryptic multihost infection caused by a pathogen that can remain undetected for prolonged periods in infected hosts, and is becoming one of the major emerging challenges for conservation globally (Alexander et al. 2002). The Mycobacterium tuberculosis complex consists of a genetically related group of multihost Mycobacterium species that cause TB in a large number of mammalian species. For example, TB frequently occurs in African buffalos (Syncerus caffer), Asian elephants (Elephas maximus), banded mongoose (Mungos mun-go), meerkats (Suricata suricatta), and European badgers (Meles meles), among others (Wirth et al. 2008). However, our knowledge of basic TB progression, such as infection duration in hosts, is limited to a small number of long-term and longitudinally monitored model systems (Thomas et al. 2021). In wild European badgers, four relevant infection states have been recognized: negative (all cultures and antibody tests negative), exposed (antibody positive), excretor (positive culture), and superexcretor (multiple positive cultures), with the assumption that the progression to each stage is an irreversible process and that excretors and superexcretors are infectious (Graham et al. 2013). Negative effects of mycobacterial infections on survival and energetics appear to be minimal to moderate in some mammals (Cross et al. 2009; Graham et al. 2013; Barbour et al. 2019); European badgers may live for up to 8 yr after testing positive for TB (Tomlinson et al. 2013), potentially acting as super spreaders (McDonald et al. 2018). The generality of these findings across other host systems is unclear.
A major constraint to increasing our understanding of wildlife TB is the limited availability and application of noninvasive diagnostic tools, such as testing fecal samples. Diagnosis of TB in live wild animals is notoriously challenging, and no gold standard test exists (Drewe et al. 2010; McDonald and Hodgson 2018); even postmortem examinations may be inconclusive (Jenkins et al. 2008; Murphy et al. 2010). Instead, the recommended approach is to model the probability of infection based on multiple tests, including Mycobacterium spp. cultures and immunoassays in combination (Drewe et al. 2010; McDonald and Hodgson 2018). Unfortunately, these tests are invasive and are logistically difficult to obtain for many wildlife species, and in these cases TB diagnosis often relies on external clinical signs (e.g., Patterson et al. 2017), characteristic of late-stage and contagious TB, which underestimates TB prevalence.
A potentially valuable method to detect and monitor TB in wildlife noninvasively could be to test fecal samples for TB, because fecal material can contain Mycobacterium when infected saliva is swallowed (Bassessar et al. 2014), or if the pathogen is able to migrate from infected organs such as mesenteric lymph nodes to the digestive tract. Crucially, some level of Mycobacterium shedding is required for fecal detection, yet, because shedding tends to be episodic, only a proportion of infected animals may be shedding the bacterium, for example 25–50% in badgers (Delahay et al. 2000; Gallagher and Clifton-Hadley 2000; Corner et al. 2011); therefore negative samples do not indicate an absence of infection. Moreover, shedding patterns are likely to vary between species, with badgers tending to be most infectious during late-stage TB (Corner et al. 2008; McDonald et al. 2019), whereas red deer may be more likely to shed intermittently from early stages of infection (Palmer et al. 2001; Santos et al. 2015). The detectability of Mycobacterium DNA in badger feces correlates with the frequency of shedding (Courtenay et al. 2006), and fecal testing generally only detects cases where Mycobacterium concentration in feces is relatively high (Stewart et al. 2017); therefore multiple replicates are recommended to maximize detection probability (King et al. 2015b; Murphy et al. 2020). In contrast to other tissue and secretion samples, PCR-based methods rather than culture-based methods are recommended for detecting Mycobacterium in fecal samples, which contain a high density of bacteria, because they are better able to detect specific taxa in high-density communities (King et al. 2015a). Together, these studies suggest that fecal samples can be a useful tool for monitoring TB in field conditions, in particular detection of actively shedding animals, yet the use of fecal samples in TB detection remains uncommon and the utility of this method for wildlife disease surveillance is unclear.
Our study combined fecal Mycobacterium detection with clinical sign data to examine longitudinal TB progression in 66 wild, socially exposed meerkats over an 18-yr period. Meerkats are highly social mammals inhabiting arid regions of southern Africa, and periodically experience TB outbreaks caused by Mycobacterium suricattae (Parsons et al. 2013). Meerkats are cooperative breeders living in groups of 2–50 individuals, with a dominant pair largely monopolizing reproduction, and a variable number of subordinate helpers of both sexes that support pup rearing (Clutton-Brock and Manser 2016). Individuals roving between groups provide TB transmission opportunities (via grooming or biting) between social groups (Drewe 2010), and TB outbreaks within social groups may last many years, with pups born during outbreaks being socially exposed from birth. Although variation in behavior and environmental conditions contributes to the probability of individuals becoming infected after exposure (Drewe 2010; Patterson et al. 2017), group members co-habiting with an infected individual are likely to be consistently socially exposed to the pathogen via grooming, fighting, and co-sleeping in shared burrows. Typical clinical signs of TB in meerkats are submandibular swellings, inguinal and cervical lumps, emaciation and lethargy, and eventual death (Drewe et al. 2009; Patterson et al. 2017), which match the clinical signs recorded in some other wildlife species (de Lisle et al. 2002). A previous study estimated a latent period (defined there as the time between a meerkat developing antibodies and becoming infectious to others) of 385 d, based on 39 TB-infected meerkats over a 2-yr study period (Drewe et al. 2011). However, how long meerkats may be TB positive before developing the characteristic external clinical signs, or how TB progresses across an individual lifetime, is still unclear.
Our study aimed to characterize the longitudinal dynamics of TB progression during meerkat lifetimes by combining PCR-based detection of M. suricattae shedding in fecal samples with external clinical signs. We focus on three distinct states of TB progression in meerkats and aim to quantify how individuals vary in the time taken to switch between these states. Although a multitude of biological and environmental factors can influence progression of TB at each stage, such as behaviors linked to higher or lower TB exposure, or physiological correlates of higher or lower susceptibility (Drewe 2010; Hawley and Altizer 2011), understanding the determinants of this variation was not the focus of this study. We categorized TB progression into three distinct states: PCR-negative exposed, PCR-positive asymptomatic, and PCR positive with clinical signs (Figure 1; see Table 1 for definitions and interpretations). Our specific aims were 1) to test whether fecal samples can facilitate the monitoring of TB prevalence and progression; 2) to characterize TB progression between the three infection states, estimating the time periods individuals spend in each infection stage; and 3) to estimate individual heterogeneity in measures of TB susceptibility and survival (Table 1).
MATERIALS AND METHODS
Study system and sample collection
Our study population consisting of habituated social groups is located in the Southern Kalahari in South Africa (Clutton-Brock et al. 1998). Available social groups have been monitored 3–5 d per week since 1993, providing extensive life history data, birth and death dates as well as detailed information on signs of disease, including typical clinical signs of TB, such as submandibular or inguinal lumps (Clutton-Brock and Manser 2016). Fecal samples from individually identified animals were collected upon defecation and either freeze-dried and stored at room temperature, or frozen immediately and stored at –80 C.
We focused on meerkats belonging to three well-monitored social groups (named Vivian, Van Helsing, and Baobab) that experienced extended TB outbreaks. We selected a total of 66 adult individuals (Vivian: n=21 [13 males; eight females]; Van Helsing: n=20 [12, eight]; Baobab: n=18 [11, seven]; other groups: n=7 [one, six]) for which fecal samples had been consistently collected up until the point of death. Seven individuals not part of the three focus groups were selected for a pilot study to optimize the PCR protocol, because they were euthanized for TB (detailed in Supplementary Methods S1). Of the others, 37 individuals showed clinical signs of TB at the end of their life, whereas 22 never exhibited clinical signs, but were good candidates for subclinical infections as they cohabited with infectious individuals. Overall, we analyzed 388 samples (1–17 samples per individual, mean±SD 7±4) collected over the course of the study individuals' lives (2.4±1.5 yr, maximum 6.7 yr).
Extraction of DNA and PCR amplification
We extracted fecal DNA using the Nucleo-Spin® 96 Soil kit (Macherey and Nagel, Düren, Germany), following the manufacturer's protocol. Approximately 80 mg of feces were used per extraction, and DNA was eluted to 50 µL. To amplify TB-causing Mycobacterium agents, we designed a primer pair primer TB 421F (5′-CCCCGATGGTTTGCGGTGG-3′) and TB 574R (5′-GCGGCTGATGTGCTCCTTGA-3′), targeting the highly conserved TB insertion element IS6110 region (Thierry et al. 1990). The primers were chosen because they performed the most reliably in the pilot study when compared with three other sets of candidate primers (Kox et al. 1994; Kent et al. 1995; Supplementary Table S1). Target DNA fragments were then PCR amplified using conditions described in Supplementary Methods S2. Each sample was PCR tested repeatedly up to 10 times or until it was recorded positive; samples that remained negative after 10 replications were categorized as TB negative. We used a one-way classification system of no return commonly applied to TB in wildlife (Tomlinson et al. 2013), assuming that individuals that test TB positive once remain positive until death, because there is no evidence for natural TB infection clearance without treatment (Houben and Dodd 2016). The first TB-positive PCR product of an individual was sequenced using Sanger sequencing to ensure the amplified product aligned (>99.1% of base pairs) to the reference sequence M. tuberculosis strain H37Rv (Cole et al. 1998).
Performance of test protocol
To assess the performance of our diagnostic test, we estimated the probability of a positive result given the meerkat showing clinical signs at the time of sampling and therefore assumed to be truly infected, thereby providing a benchmark with which to assess our diagnostic test performance. Because most of the 388 fecal samples included in this study were collected before the development of clinical signs, our performance analysis included 65 fecal samples from 33 individuals with clinical signs (range 1–6, mean of 3±2 samples per individual), that were PCR tested in total 374 times. We modeled the probability of a positive result using a binomial generalized linear model implemented with the lme4 package (Bates et al. 2015) and including sample number and individual ID as random effects, and no fixed effects. This measure provides an indication of the detection probability (i.e., probability of the test outcome being positive given true infection). We acknowledge that this approach has its limitations because 1) M. suricattae concentration was not measured and probably varied across samples; 2) the TB infection status of all meerkats with clinical signs was not verified using additional diagnostic tests; and 3) the approach is only representative for late-stage TB infections accompanied by clinical signs, with performance potentially being lower for infected animals without clinical signs.
Characterizing TB progression, susceptibility, and survival
For each individual, we created a timeline categorizing the individual's status into three conceptually and epidemiologically meaningful infection states based on PCR results and TB clinical sign status at the time of sampling: PCR-negative exposed, PCR-positive asymptomatic, and PCR positive with clinical signs (Fig. 1). All TB-related terms and their definitions are summarized in Table 1. We defined the exposure date as the first date an individual was seen cohabiting with an individual assumed to be infectious, either because of exhibiting clinical signs or shown to be PCR positive in this study. Exposure is required for infection, yet we do not assume that exposed individuals always become infected. Rather, we aim to quantify the variation in disease progression from exposure to becoming PCR positive (denoting shedding) or developing clinical signs. We defined clinical signs as animals exhibiting submandibular, inguinal or cervical lumps, or TB being recorded in the health data set, defining the first date of whichever case as the first instance of the development of clinical signs. If an animal showed a subcutaneous lump or swelling that was not explained by an injury, snake bite, or similar, and developed clear TB clinical signs within 1 yr, the date of the subcutaneous lump was used as the date of onset of clinical signs. We defined death date as the date that an individual with advanced clinical signs of TB disappeared from the study population (n=36), was found dead (n=12), or was euthanized because it showed very advanced stages of TB and was deemed close to death (n=18). To reduce bias when estimating the length of the relevant infection states, we excluded two individuals whose exact exposure data was unclear, and the nine individuals that developed clinical signs before becoming PCR positive.
Lastly, we aimed to understand variation in TB progression patterns across the tested individuals. Therefore, we estimated individual TB susceptibility, defined here as the time span between exposure to TB and becoming PCR positive. Although variations in both the level of exposure and host propensity to become infected upon exposure (i.e., susceptibility) contribute to the time between exposure and contracting the infection (Hawley and Altizer 2011), we chose the term susceptibility here to capture this between-individuals variation. We also measured individual survival after TB detection, defined as the time span between becoming PCR positive and death. We acknowledge the fact that we cannot rule out a TB infection before a meerkat was tested PCR positive; therefore our estimate represents the minimum time span of infection. All analyses were performed in R version 3.6.2 (R Core Team 2019).
Performance of test protocol for M. suricattae detection
We retrospectively assessed the performance of the test protocol on 66 fecal samples taken from 33 individuals with clinical signs at the time of sampling, which were therefore assumed to be truly infected. The probability of a PCR-positive result in feces from meerkats with clinical signs of TB was 13% (95% confidence interval 3–46%; n=374 PCRs; Supplementary Table S2). Our model indicated that variation in PCR outcome between samples was much higher than variation within samples (i.e., some samples were always likely to be positive and other samples were always likely to be negative), but that variation between individuals was minimal (i.e., samples from different individuals had the same probabilities of being PCR positive).
Alignment of PCR results and clinical signs
We identified the earliest detectable time point of Mycobacterium shedding using PCR of fecal samples for 66 TB-exposed meerkats, and aligned this information with field observations of clinical signs to examine TB progression within individuals over their life (Fig. 2). For 78.8% of meerkats tested, PCR results and observed clinical signs matched by the end of life: PCR-negative individuals stayed asymptomatic and PCR-positive individuals developed clinical signs. Of these, 53% of meerkats were PCR positive and showed clinical signs over time, and 25.8% of meerkats were never PCR positive and never developed clinical signs despite being exposed to TB. Of the 21.2% of meerkats for which PCR and clinical sign data did not align, 7.6% were tested PCR positive yet remained in an asymptomatic state until death, whilst 13.6% displayed clinical signs of TB but feces were never detected as PCR positive (Supplementary Table S3).
Progression of TB
Most meerkats included in our study were born into a group infected with TB and thus exposed to TB at birth, although approximately 30% of individuals were exposed later (Fig. 3). After first TB exposure, animals became PCR positive on average 427±328 d (∼1.2 yr, median 363 d) later, although interindividual variation was high (Table 2). The asymptomatic period between an individual becoming PCR positive and developing clinical signs was mean 349±211 d (∼1 yr, median 330 d), although the maximum asymptomatic period was 834 d (∼2.3 yr; Table 2). Once an animal showed clinical signs, and thus was probably infectious, it survived mean 160±173 d further (∼0.4 yr, median 98 d), although one meerkat exhibited clinical signs for 595 d (∼1.6 yr). We did not detect sex-specific differences in our data set (Supplementary Table S4).
Individual variation in susceptibility and survival
We found high variation in TB susceptibility across individuals, measured here as the time span between TB exposure and becoming PCR positive (427±328 d; Fig. 4A). As outlined earlier, 25.8% of exposed meerkats (n=17) never became PCR positive nor showed clinical signs, suggesting a significant proportion of the population may not be susceptible to TB, although we cannot rule out these individuals may be infected. Out of meerkats that did detectably contract TB, there was one case where TB was detected 1,518 d (∼4.2 yr) after exposure, whilst other meerkats appeared to shed M. suricattae immediately upon exposure. Variation in survival, which we define as the period between first positive PCR sample and death, was also very high (377±283 d (∼1 yr; Table 2), with some meerkats surviving for up to 1,040 d (∼2.8 yr) after becoming PCR positive (Fig. 4B).
Monitoring TB in wildlife is challenging because infections often appear asymptomatic, and the most reliable diagnostic tools (such as cultures from tissue and respiratory tracts in combination with immunoassays) require capturing the animals (Thomas et al. 2021). Our study, which pairs observed external clinical sign data with PCR testing of fecal samples collected noninvasively across most of the life span of 66 naturally TB-exposed wild meerkats, suggest that fecal samples can be used to detect TB in wildlife and to estimate the parameters associated with TB progression, providing a better estimate of infection duration and progression than clinical signs alone. Notably, the use of PCR testing on fecal samples revealed hidden TB infections on average 1 yr prior to the onset of external clinical signs, more than doubling the estimated infection period compared with clinical sign data.
Our results add to our current understanding of TB epidemiology in wild meerkats in several ways. First, a full quarter of consistently TB-exposed meerkats never became PCR positive or developed clinical signs, suggesting a significant pool of individuals within the population that have very low susceptibility. Second, our average time between exposure and a meerkat becoming PCR positive of 427 d is consistent with a latent period of 385 d proposed by a study using antibody tests and culture of clinical samples on 39 longitudinally monitored meerkats (Drewe et al. 2011). This provides further evidence that results from fecal samples are comparable to those from more sophisticated detection methods. Thirdly, we show that TB progression is highly variable across individuals at every stage of TB progression. The long infection period, and particularly the long asymptomatic period, is likely to have consequences for TB maintenance, social structures, survival, and reproduction, as with other pathogens (Jolles et al. 2005; Perez-Heydrich et al. 2012; Edmunds et al. 2016; Lopes et al. 2016). During this time, asymptomatic but TB-infectious meerkats could function as undetected super spreaders (Martin et al. 2019), being seemingly unaffected themselves, but having a major negative impact on the population in terms of TB. Lastly, we found high heterogeneity in our estimates of susceptibility and survival after infection in TB in meerkats. The underlying factors generating such variation require further research, but may be linked to variation in genetics and maternal effects (Marjamäki et al. 2021), as well as factors such as social standing and behavior (VanderWaal and Ezenwa 2016).
Detection of TB from feces does have limitations. Mycobacterium detection in feces definitely indicates shedding, but testing fecal samples cannot identify latent infections before the onset of shedding. Moreover, the detection probability for animals with clinical signs was low (13%), and in 13.6% of our study individuals clinical signs were observed yet fecal samples were never detected as PCR positive, highlighting the possibility of false-negative results even in presumably infectious and shedding animals. This may be because of natural fluctuations in shedding patterns, but may also be explained by differences in the form of TB infection: most individuals displayed lesions in the head lymph nodes and lungs at postmortem examinations (Drewe et al. 2009), which may lead to swallowing of mycobacteria, later detectable in the feces; however, if the main site of lesions is in the spleen, liver, or mesenteric or mediastinal lymph nodes (Drewe et al. 2009), this might not lead to Mycobacterium shedding via the feces.
Our study was additionally limited by the fact that we could not distinguish between true and false negatives, and therefore sensitivity and specificity of tests could not be measured. Previous studies using fecal samples for TB detection found high sensitivity (96.7%) and specificity (99%) when using samples spiked with M. bovis as positive controls (Murphy et al. 2020), suggesting that in theory a PCR-based approach for TB detection is useful to detect Mycobacterium shedding via feces. However, sensitivity is low if the aim is to detect infected yet not necessarily shedding states (14–25%; King et al. 2015b), given that Mycobacterium shedding is episodic, and low levels of shedding may not be detectable via feces (Stewart et al. 2017). The latter issue may be in part controlled for by applying quantitative PCR to estimate Mycobacterium concentration and thereby identify detection thresholds (King et al. 2015b; Murphy et al. 2020). Another potential issue is false positives, which may severely bias results given that one positive sample may have a large effect on the estimation of progression. The probability of a false positive was estimated at 2.6% in a previous study (King et al. 2015b), and test specificity is generally high (King et al. 2015b; Murphy et al. 2020); thus false positives are much less likely than false negatives. We reduced the likelihood of false positives by sequencing the first positive PCR result per individual to verify strain identity, and therefore consider false positives unlikely unless via contamination, which was always assessed via PCR-negative controls. Despite these limitations, PCR testing of feces still provides a useful diagnostic tool to assess TB shedding in systems where study animals can be easily observed, enabling some noninvasive monitoring of disease presence and progression before the onset of clinical signs.
Our study is one of the first to investigate TB progression over the entire life of wild animals using only noninvasive methods. We not only demonstrate the potential of PCR-based TB detection from fecal samples as a diagnostic tool in wildlife, but also provide valuable insights into the course of TB following social exposure. Our results indicate that TB progression in meerkats is characterized by an extended asymptomatic period followed by a shorter period with exhibited clinical signs, with high heterogeneity, potentially promoting TB persistence within even asymptomatic populations. Establishing routine noninvasive TB screening, particularly in longitudinal studies of known individuals, may thus be a suitable approach to investigate which mechanisms and factors determine individuals' susceptibility, survival, and potential differences in fitness related to TB in much greater detail than using purely observational data of clinical signs. Being able to detect asymptomatic infections noninvasively increases our ability to estimate the relative contributions of individuals with and without clinical signs to disease transmission, assisting epidemiological modeling and more effective TB management of endangered species or populations.
We are grateful to the Kalahari Research Trust and the Kalahari Meerkat Project for access to facilities and habituated animals in the Kuruman River Reserve, South Africa. This article relied on records of individual identities and/or life histories maintained by the Kalahari Meerkat Project, supported by the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (research grants 294494 and 742808 to T.H.C.-B. since 7 January 2012), the Human Frontier Science Program (funding reference RGP0051/2017), the University of Zurich, the Swiss National Science Foundation and the Mammal Research Institute at the University of Pretoria, South Africa. Research for this study was conducted with permission of the ethical committee of Pretoria University and the Northern Cape Conservation Service, South Africa (permit EC031-13, FAUNA 1020-2016). The study was funded by the German Research Foundation to S. Sommer (DFG SO 428/15-1).
Supplementary material for this article is online at http://dx.doi.org/10.7589/JWD-D-21-000063.