High incidence of Trypanosoma cruzi infections in dogs directly detected through longitudinal tracking at 10 multi-dog kennels, Texas, USA

Canine Chagas disease, caused by the protozoan parasite Trypanosoma cruzi, is increasingly recognized as a health concern for dogs in the USA, and infected dogs may signal geographic regions of risk for human disease. Dogs living in multi-dog kennel environments (kennels with more than one dog) where triatomine vectors are endemic may be at high risk for infection. We monitored a cohort of 64 T. cruzi-infected and uninfected dogs across 10 kennels in Texas, USA, to characterize changes in infection status over one year. We used robust diagnostic criteria in which reactivity on multiple independent platforms was required to be considered positive. Among the 30 dogs enrolled as serologically- and/or PCR-positive, all but one dog showed sustained positive T. cruzi diagnostic results over time. Among the 34 dogs enrolled as serologically- and PCR-negative, 10 new T. cruzi infections were recorded over a 12-month period. The resulting incidence rate for dogs initially enrolled as T. cruzi-negative was 30.7 T. cruzi infections per 100 dogs per year. This study highlights the risk of T. cruzi infection to dogs in kennel environments. To protect both dog and human health, there is an urgent need to develop more integrated vector control methods as well as prophylactic and curative antiparasitic treatment options for T. cruzi infection in dogs.


Introduction
The protozoan parasite Trypanosoma cruzi, agent of Chagas disease, is a vector-borne zoonotic pathogen endemic to the Americas. According to estimates based on 2010 data, over 5.7 million people were infected with T. cruzi in 21 Latin American countries in addition to hundreds of mammal species [1,2]. T. cruzi is predominantly transmitted in the feces of infected triatomines ('kissing bugs') through contact with wounds or mucous membranes or ingestion of infected insects or fecal material [1]. Oral transmission is thought to be the most important route in domestic dogs and wild mammals and is a highly efficient mode of transmission [3][4][5][6].
Enzootic cycles of T. cruzi transmission have been documented in the southern USA, where opossums (Didelphis virginiana), woodrats (Neotoma spp.), coyotes (Canis latrans), and other mammals naturally infected with T. cruzi come into contact with endemic triatomine species [7][8][9] and maintain parasite transmission [1,[10][11][12][13]. The wildlife species involved in the transmission of T. cruzi often inhabit peridomestic environments and share spaces with domestic dogs outdoors, increasing dogs' risk of exposure to infected vectors [11,14]. In settings where dogs share spaces with humans in and around homes, dogs serve as sentinels for human infection, maintaining T. cruzi transmission and amplifying disease in domestic and peridomestic environments [15,16]. Thus, areas of high T. cruzi seroprevalence in domestic dogs may indicate areas of heightened risk for human infection [17].
In the USA, T. cruzi infection has been reported in dogs across the southern states [18][19][20]. Studies from Texas, Oklahoma, and Louisiana report dog infection prevalence from 3.6-22.1%, and up to 57.6% in some multi-dog kennels, or kennels with more than one dog [20][21][22][23][24][25][26][27][28]. Infection with T. cruzi shows no strong breed predilection, and shelter and stray dogs are likely exposed to T. cruzi often due to their high level of exposure to vectors, resulting in high prevalence of infection [20,23,29,30]. Additionally, government working dogs along the USA-Mexico border are exposed to T. cruzi with a seroprevalence of up to 18.9% [27], in which infections have led to fatal cardiac disease [31]. Canine Chagas diagnostics are challenged by the limited number of approved tests validated for use in dogs in the USA, and often, in accordance with WHO guidelines for Chagas testing in humans, more than one serologic test is required for canine Chagas diagnoses. In both humans and dogs, Chagas disease causes a range of clinical symptoms, progressing through acute and chronic stages of disease, which can include severe heart disease and death [4,22,32]. While insect vectors are endemic to the southern USA, canine travel introduces a veterinary health concern as infected dogs move outside of endemic areas to areas where clinicians are less familiar with the symptoms of Chagas disease [33].
While measures of infection prevalence from cross-sectional studies are useful in quantifying the burden of disease in populations, the incidence rate (i.e. the number of new infections per population per unit time) can provide a direct measurement of risk. Direct measurements of incidence in natural animal populations are rare, as they require tracking of healthy, uninfected individuals over time to monitor for new infections. Alternatively, incidence can be inferred indirectly by comparing seroprevalence across different age cohorts of animals. Using this indirect approach based on serology, we are aware of two studies investigating the T. cruzi incidence rate in dog populations in the USA. The first estimated a serologic incidence of 3.8% in juvenile dogs along the Texas-Mexico border [34]; the second estimated an incidence rate of 2.3 new cases per year in dogs of all ages in Louisiana shelters [20]. We used a longitudinal study design to directly measure incidence of T. cruzi infection in dogs of Central/South Texas, an area with a high risk of infection based on suitable triatomine habitat and autochthonous human cases [35]. Uncertainty about transmission risk and the lack of an effective treatment or vaccine pose significant problems to veterinarians and dog owners following diagnoses of Chagas disease. In this study, we tracked the serostatus and PCR-status of a matched cohort of T. cruzi-positive and negative dogs at three intervals over 12-months, allowing for the direct detection of T. cruzi infection in dogs in kennel environments.

Ethics statement
All samples were obtained from privately-owned animals in accordance with guidelines approved by the Texas A&M University's Institutional Animal Care and Use Committee and Clinical Research Review Committee (2018-0460 CA).

Locations and sample collection
Using a prospective longitudinal study design, we enrolled a network of 10 multi-dog kennels throughout Central and South Texas with a prior history of T. cruzi infection in at least one of their dogs. Kennel locations were categorized by Texas ecoregions [29]. Dogs at these kennels were generally year-round residents of the kennel, typically leaving the kennel for hunting or training during the day; dogs at one kennel spent some months in the summer in a northern state to train. They were primarily bred and trained to aid hunting parties or compete in American Kennel Club dog events and were classified according to American Kennel Club breed groups. Enrollment criteria included dogs of any breed or sex at least one year of age, in residence at one of the selected kennels, and negative for other selected vector-borne infectious diseases (Anaplasma spp., Ehrlichia spp., Borrelia burgdorferi, and Dirofilaria immitis).
We aimed to identify and enroll approximately two to four T. cruzi-infected and two to four T. cruzi-uninfected dogs at each kennel for longitudinal tracking at approximate 6-month intervals. During the initial sampling time point between May and July 2018, we tested 134 dogs (between seven and 20 dogs at each of the 10 kennels) for evidence of T. cruzi and four other vector-borne infections. Based on the enrollment criteria, 64 dogs were identified as meeting the criteria and were enrolled in the study. When possible, T. cruzi-positive and T. cruzi-negative dogs were frequency (group) matched on the basis of age, sex, and breed across all kennels. Each blood sampling event consisted of collecting approximately 5mL of blood in an EDTA (ethylenediaminetetraacetic acid) tube and 4mL in a clotting tube. The second sampling event occurred between December 2018 and March 2019 (4.7-9.2 months, average = 6.3 months; 6-month Time Point), and the third sampling event occurred between May and September 2019 (5.4-7.4 months, average = 6.4 months; 12-month Time Point). All 10 kennels were visited at each time point to collect blood samples from enrolled dogs.

Serology
At each of the three time points, serum samples were tested for T. cruzi antibodies using three serological tests: Chagas Stat-Pak (ChemBio, Medford, NY, USA), Chagas Detect Plus Rapid Test (InBios International, Inc., Seattle, WA, USA), and an indirect fluorescent antibody (IFA) test. The Chagas Stat-Pak and Chagas Detect Plus are rapid tests that, while not labeled for use in dogs, have been shown to be sensitive and specific to detect T. cruzi antibodies and are commonly used in dogs for research purposes [26,36,37]. Serum samples were tested according the manufacturer's instructions, and results were determined after 15 minutes. In all cases, the integrated 'control' bands appeared as expected on the rapid tests. Rapid tests were considered negative when the 'test' band had no color development or when a very faint band-not perceptible enough to be a clear positive-developed. The IFA was performed by the Texas A&M Veterinary Medical Diagnostic Laboratory (TVMDL) using 200 μL of serum to test for IgG antibodies against T. cruzi. Reactivity on at least two of the three serologic tests was one of the two possible criteria (in addition to PCR positivity, see below) to consider a dog T. cruzipositive for enrollment. An IDEXX 4Dx test (IDEXX Laboratories, Inc., Westbrook, ME, USA) was run at the time of enrollment to exclude any dogs that may have had other vector-borne infections, including Anaplasma spp., Ehrlichia spp., Borrelia burgdorferi, and Dirofilaria immitis.

DNA extraction, quantification, and PCR
A DNA extraction procedure from large volumes (5mL) of EDTA-treated blood was used in an effort to increase sensitivity of detection of T. cruzi using the E.Z.N.A Blood DNA Maxi kit (Omega Bio-Tek, Norcross, GA, USA) according to the manufacturer's protocol, except 650μL of elution buffer was used. T. cruzi-negative controls (phosphate buffered saline) were included in the DNA extractions. For each extracted sample, DNA was quantified using a spectrophotometer (Epoch, BioTek Instruments, Inc., Winooski, VT, USA). In the case that any errors had occurred during the extraction process, as seen by low DNA yield from the MAXI extraction (<85 ng/μL) or control wells that suggested contamination, a second extraction was performed using 250μL of clotted blood using the E.Z.N.A. Tissue DNA kit, according to the manufacturer's protocol, except 50 μL of elution buffer was used (Omega Bio-Tek).
Samples were tested using qPCR for the presence of T. cruzi satellite DNA using the Cruzi 1/2 primer set and Cruzi 3 probe in a real-time assay, which amplifies a 166-bp segment of a repetitive nuclear DNA [38,39] as previously described [27]. This PCR method is sensitive and specific for T. cruzi when compared to other PCR techniques (sensitivity = 83.3, specificity = 95; compared with kDNA and Satellite DNA PCRs as described in [40]. T. cruzi DNA extracted from isolate Sylvio X10 CL4 (ATCC 50800, American Type Culture Collection [ATCC], Manassas, VA, USA; DTU TcI) was used as the T. cruzi-positive control in all PCRs. The qPCR amplification curves were checked for quality (sigmoidal curves), and a sample was considered T. cruzi-positive if the Ct value was lower than 36 [41].

Statistical analyses
Dogs were categorized as T. cruzi-positive at a given time point if they met the criteria for serological positivity (positive on at least two of three independent serologic assays) and/or if they met the criteria for PCR positivity (Ct value lower than 36). We calculated the Kappa index to determine the agreement among all three serological tests [20,27]. To describe the risk of T. cruzi infection in the population of dogs that were initially enrolled as T. cruzi-negative, we calculated the incidence rate as the number of new T. cruzi infections per 100 dogs per year. To reflect the time T. cruzi-negative dogs in the population were at risk, 0.5 years was subtracted from the population at risk (n = 34) for each of the three T. cruzi-negative dogs lost to follow up after the 6-month time point (n = 3).

Dog enrollment
Of the 134 dogs screened, we enrolled 30 T. cruzi-positive and 34 T. cruzi-negative dogs across 10 kennels in Central and South Texas (Table 1). Kennels were located in one of four ecoregions, primarily (7/10) in the South Texas Plains. Dogs enrolled in the study were between the ages of 13 months and 12.2 years (mean = 6.1 years; median = 6.1 years). Dogs were frequency matched based on age, sex, and breed: age was matched within 8 months (average T. cruzi-positive = 6.6 years; average T. cruzi-negative = 5.8 years), sex was matched within four dogs (positive = 14 males and 16 females; negative = 19 females and 15 males), and breeds were matched between T. cruzi-positive and T. cruzi-negative dogs (various hound breeds were considered "Hound" for matching purposes; three positive and negative Belgian Malinois, two positive and negative German Shorthair Pointers, three positive and negative Hounds, one positive and seven negative Labrador Retrievers, two positive and negative Brittany Spaniels, and 19 positive and 17 negative English Pointers per group). During the initial screening of 134 dogs across all 10 kennels, one dog was positive for Dirofilaria immitis and was excluded from the study. In total, five dogs were lost to follow up after the 6-month time point: two dogs enrolled as T. cruzi-negative (including one that seroconverted to positive by 6-months) moved locations and were sold to new owners, and three dogs (one T. cruzi-positive and two negative) had died.
Of the 30 dogs enrolled as T. cruzi-positive at the initial time point, 26 (86.7%) were positive on all three serologic tests; two (6.7%) were positive on both rapid tests yet negative on IFA; and one (3.3%) was positive on a single rapid test and had a high titer (>1280) on IFA (Table 2). Additionally, one dog (D76) was enrolled as T. cruzi-positive based solely on PCR

DNA extraction
Eluted DNA concentrations from MAXI extractions used in the study ranged from 88.6 ng/μL to 348.5 ng/μL (n = 104; mean = 172.4 ng/μL; median = 164.2 ng/μL). Table 2. Serology of the 30 dogs enrolled as T. cruzi-positive in the study. The two rapid tests (Chagas Stat-Pak and Chagas Detect Plus) are scored as Positive ("+") or Negative ("-"), and the IFA titer is reported in which a result of <20 is noted as Negative ("-").
Of the 34 dogs enrolled as T. cruzi-negative, 24 (70.6%) remained negative at the 6-month and 12-month time points (Fig 1). One dog (2.9%) remained serologically-negative but was PCR-positive at the 12-month time point, and nine dogs (26.5%) were seropositive by the end of the study; therefore, these 10 dogs met the study definition of T. cruzi positive (Fig 1). Notably, two dogs (5.9%) were seropositive on both rapid serologic assays at the 6-month time point and therefore met the definition for seroconversion, but were seropositive on only one rapid serologic assay by the 12-month time point. In total, three dogs enrolled as T. cruzi-negative (8.8%) were PCR-positive at the 12-month time point (Fig 1).
The Kappa Indices comparing serological tests across all time points showed the Chagas Stat-Pak and IFA had almost perfect agreement (kappa = 0.84), while there was substantial agreement between the Chagas Detect Plus and IFA and Chagas Detect Plus and Chagas Stat-Pak (kappa = 0.65 and 0.70 respectively) ( Table 4). Direct quantification of T. cruzi incidence in dogs is rare, likely due to the challenges of collecting longitudinal data [43]. More commonly, incidence is estimated indirectly. For example, researchers conducted a serosurvey of 540 dogs in Louisiana and found a 6.9% prevalence of T. cruzi infection [20]. Assuming exposure to T. cruzi infection is similar between animals and over time, they also estimated the incidence of 2.3 new infections per 100 dogs/year. This estimate is similar to that of another study, where researchers estimated the incidence in juvenile dogs along the Texas-Mexico border [34]. In this second study, the T. cruzi prevalence in juvenile dogs under 6 months old was used to estimate a T. cruzi serologic incidence of 3.8%, assuming all infections that occurred in dogs under 6 months old were newly acquired infections. This methodology does not consider congenital infections, and further, can only be applied to the juvenile age group. In contrast to these previous studies estimating incidence to be much lower than our calculated incidence, our longitudinal study allowed us to directly detect new T. cruzi infections using serology and PCR over one year, and thus report a direct measure of incidence in a closed population of dogs.
There is no gold standard to test for T. cruzi infections in dogs, and the IFA is widely used in clinical practice across the USA for veterinary Chagas diagnostics. While the Chagas Stat-Pak and the Chagas Detect Plus are often used for research purposes, discordant test results are common in canine testing for Chagas disease [27]. As recommended by the WHO for human Chagas diagnostics [44], reactivity on multiple serologic assays was the criteria used in our study to consider a dog T. cruzi-positive. We observed substantial to high agreement across serologic tests, reflecting the strict inclusion criteria and matched study design, where T. cruzipositive dogs were enrolled as positive only when there was some agreement across serologic assays. Overall, the Chagas Stat-Pak and IFA had the best agreement with an overall kappa index of 0.84 (Table 4); whereas, there was reduced pairwise agreement between the other combinations of tests (kappa range 0.65-0.70). Cross-sectional studies of dog T. cruzi serostatus have found less agreement, even high discordance, between Chagas tests using the Kappa index, with only slight to moderate agreement between tests [20,27]. Given the absence of gold standard antemortem testing methods, our data support the WHO recommendation that a multi-test approach is still necessary in dog Chagas disease diagnostics. However, none of our selected serologic assays could identify IgM antibodies, which have been used to detect acute antibody responses against T. cruzi in dogs [16,45] and could have been correlated with PCR data. PCR was used in addition to serologic tests as a diagnostic tool for T. cruzi in this study, with the ability to detect new infections prior to antibody formation. The PCR protocol used in this study was shown to be one of the most sensitive and specific PCR methods to detect T. cruzi based on an international survey of PCR methods of T. cruzi detection [40] and has been previously used to detect T. cruzi in blood samples [39]. We expect to see fewer PCR positives than antibody positive results because PCR-positivity indicates active circulation of the parasite in the animals' blood, which may occur for only short time periods in infected dogs [4]. For example, D42, D100, and D104 were all PCR-positive at the initial time point and final (12-month) time point yet were PCR-negative at the 6-month time point (Table 3). This variation in PCR results over time may be due to transient parasitemia from an initial infection, or due to re-infections [46], as our data show T. cruzi is readily infecting dogs in these areas. By using four independent tests to categorize animals as T. cruzi-positive or negative, and repeating sampling at 3 time points to look for change in infection status, our study design was robust and sensitive to capture new infections in a way that has not before been done in veterinary Chagas studies in the United States.
In this study, one dog (D76) was PCR-positive yet antibody negative at enrollment, indicating the presence of T. cruzi DNA in the blood at the time of sampling, and thus this dog was enrolled as T. cruzi-positive. We expected this dog to develop detectable antibodies by the time of the 6-and 12-month time points. However, the dog did not seroconvert and displayed a weak PCR-positive result (high-positive Ct value) at the 6-month time point. The reasons for a lack of seroconversion in this dog are unknown but could be due to health conditions that resulted in a suppressed immune system or other health measures we did not assess as part of this study, or due to a false positive PCR result.
Two dogs (D03 and D22) met the definition of incident cases because of their initial enrollment as T. cruzi-negative (each had one very faint band on one test; see Methods; classified as negative) and seroconversion at the 6-month time point, at which time they were reactive on both rapid serologic assays. However, at the 12-month time point, both dogs were only positive on one serologic assay (D03 on the Chagas Stat-Pak; D22 on the Chagas Detect Plus), and therefore failed to meet the diagnostic criteria for positivity at that time point, which required reactivity on at least two independent tests. As a new infection followed by self-cure would be unlikely [1,43], we suspect this scenario reflects the limitations of using unvalidated diagnostic tests for dogs, and future prospective sampling from these dogs would be enlightening. If these two dogs represent false conversions, this means our incidence rate would be lower, at 24.6 new infections per 100 dogs/year. Even an incidence only considering eight new infections (out of 34) is markedly higher than previously estimated dog incidence rates for T. cruzi infections. Additionally, the three dogs enrolled as negative, despite having a weakly positive band on a single rapid test, remained negative through the course of the study with either one weakly positive test or no positive test at each timepoint, providing evidence in this study that a single weak band is not likely to signal an early infection or an infection that will become more diagnostically apparent in the next year. The need for improved diagnostic tools for detection of T. cruzi infection is apparent for human medicine [47], and our study highlights the need in veterinary medicine as well [27].
Over the course of the study, five dogs were lost to follow up either due to death or to moving locations. Unfortunately, necropsies were not performed and postmortem samples were not available; thus, the causes of death (and potential role of Chagas disease) remain unknown. One of the deceased dogs was T. cruzi-positive, and the owners described a slow decline over two months with observed weight loss, but no veterinary visits or treatment were elected by the owners.
There was geographic variation in the frequency of incident cases across the 10 studied kennels. Kennels varied from no incident infections to three of the four dogs enrolled as T. cruzinegative having newly acquired T. cruzi infection. The incident cases were distributed across six of the 10 kennels. This variation may indicate natural variation in triatomine abundance or infection among kennels [42], or the importance of other factors, such as vector control around the kennels, kennel integrity, or additional exposure to kissing bugs while hunting or training outside of the kennels, such as on other areas of the owners' property.
There is no validated treatment or prevention for all stages of Chagas disease in dogs, and diagnosis often occurs during the chronic phase when experimental treatments rarely change the outcome of disease and treatment is designed to manage symptoms [4]. As no vaccines exist for dogs and cardiac symptoms can cause acute death or chronic illness [48], preventing contact with infected vectors is paramount; this requires a thorough understanding of transmission cycles and risk. Accordingly, Chagas disease prevention and control is largely focused on reducing a dog's contact with vectors. Approaches to halting contact include vector control with insecticides applied to kennel outdoor spaces, keeping dogs indoors, avoiding outdoor lights that attract bugs, and altering vegetation to reduce the density of wildlife reservoirs and kissing bugs. However, various vector control measures are challenging, as no insecticides are specifically labeled for use in the control of triatomines in the USA. In particular, it is common for adult kissing bugs to be encountered in kennel environments, posing a transmission risk to the dogs [42]. Triatomines have proven difficult to eradicate even in areas with robust vector control programs, as locating the bugs is difficult, time consuming, and sometimes unsuccessful for species that colonize the domestic environment [49,50]. Widespread vector control campaigns in the USA have not been attempted for triatomines, owing to their predominantly sylvatic occurrence [42] with occasional spillover to domestic settings.
Multi-dog kennels should be areas of targeted intervention in the USA, as the incidence of dog infection is high. We found dogs at many of these kennels have a high risk of T. cruzi infection over the course of one year, highlighting the need to develop thorough vector control methods that can be deployed in kennel environments and identify, test, and make available prevention and treatment options for T. cruzi infection in dogs in the USA. The loss of hunting and working dogs has an economic impact on breeders, trainers, and ranch owners, and developing next steps to prevent or treat canine Chagas disease is not only necessary for veterinary health but to decrease the human impact of Chagas disease as well. Further, travel of infected dogs from endemic regions of the southern USA to non-endemic regions where there is less awareness of Chagas disease is occurring [33], in which case diagnosis may be harder to attain. Similarly, in northern Mexico, canine T. cruzi infections are also common, with seroprevalence estimates of 4.4% and 9.5% in communities near the USA border [45,51]. Differences in the level of canine infection among studies may relate to differences in the local abundance of infected vectors that feed on dogs and/or may relate to different diagnostic approaches. Because dogs can achieve a level of parasitemia sufficient to infect kissing bugs during acute disease, there is concern for infected dogs serving as a local reservoir to perpetuate the parasite life cycle [23,25,52]. Accordingly, any interventions to reduce infection in dogs and improve their overall health may also have the added benefit of reducing the risk of locally acquired human disease as well.