https://orcid.org/0000-0003-4080-2414
Figures
Abstract
Chronic wasting disease (CWD) has become a major concern among those involved in managing wild and captive cervid populations. CWD is a fatal, highly transmissible spongiform encephalopathy caused by an abnormally folded protein, called a prion. Prions are present in a number of tissues, including feces and urine in CWD infected animals, suggesting multiple modes of transmission, including animal-to-animal, environmental, and by fomite. CWD management is complicated by the lack of practical, non-invasive, live-animal screening tests. Recently, there has been a focus on how the volatile odors of feces and urine can be used to discriminate between infected and noninfected animals in several different species. Such a tool may prove useful in identifying potentially infected live animals, carcasses, urine, feces, and contaminated environments. Toward this goal, dogs were trained to detect and discriminate CWD infected individuals from non-infected deer in a laboratory setting. Dogs were tested with novel panels of fecal samples demonstrating the dogs’ ability to generalize a learned odor profile to novel odor samples based on infection status. Additionally, dogs were transitioned from alerting to fecal samples to an odor profile that consisted of CWD infection status with a different odor background using different sections of gastrointestinal tracts. These results indicated that canine biodetectors can discriminate the specific odors emitted from the feces of non-infected versus CWD infected white-tailed deer as well as generalizing the learned response to other tissues collected from infected individuals. These findings suggest that the health status of wild and farmed cervids can be evaluated non-invasively for CWD infection via monitoring of volatile metabolites thereby providing an effective tool for rapid CWD surveillance.
Citation: Golden GJ, Ramirez EA, Stevens HN, Bourbois J, Grove DM, Bowen RA, et al. (2024) Biodetection of an odor signature in white-tailed deer associated with infection by chronic wasting disease prions. PLoS ONE 19(8): e0303225. https://doi.org/10.1371/journal.pone.0303225
Editor: Rodrigo Morales, The University of Texas Health Science Center at Houston, UNITED STATES OF AMERICA
Received: April 22, 2024; Accepted: June 24, 2024; Published: August 7, 2024
This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Data Availability: Golden, Glen J.; Ramirez, Elizabeth A.; Stevens, Hayley N.; Bourbois, Jennifer; Grove, Daniel; Bowen, Richard A.; DeLiberto, Thomas J.; Kimball, Bruce A. 2024. Biodetection data: Using odor signature in white-tailed deer associated with infection by Chronic Wasting Disease prions. Research Dataset Series. USDA, APHIS, WS National Wildlife Research Center. Ft. Collins, Colorado. https://doi.org/10.2737/NWRC-RDS-2024-001.
Funding: TWRA AP-14839 Animal and Plant Health Inspection Service and WILDLIFE RESOURCES AGENCY, TENNESSEE https://www.aphis.usda.gov/aphis/ourfocus/business-services/financial-management-division/financial_services_branch/agreements_service_center/terms-conditions-for-aphis-awards Dan Grove helped with study design, decision to publish and manuscript preparation TWRA APP-17383 Animal and Plant Health Inspection Service and WILDLIFE RESOURCES AGENCY, TENNESSEE https://www.aphis.usda.gov/aphis/ourfocus/business-services/financial-management-division/financial_services_branch/agreements_service_center/terms-conditions-for-aphis-awards Dan Grove helped with study design, decision to publish and manuscript preparation GJG APP-15390 National Wildlife Research Center https://www.aphis.usda.gov/national-wildlife-programs/nwrc No role played by funders GJG 22-7400-1646-CA National Wildlife Research Center https://www.aphis.usda.gov/national-wildlife-programs/nwrc No role played by funders.
Competing interests: The authors have declared that no competing interests exist.
Introduction
It has long been speculated that diseases might be diagnosed based on odor. In the past decade, examination of fecal volatiles has been featured for detection of chronic wasting disease (CWD) in white-tailed deer (Odocoileus virginianus) [1], avian influenza virus (AIV) in mallards (Anas platyrhynchos) [2, 3]; and bovine tuberculosis in goats (Capra hircus) [4, 5], white-tailed deer (WTD) [6], and cattle (Bos taurus) [7]. Diagnosis of disease or infection using volatile metabolites found in breath [4, 7–10] or other emanations [11–13] have also been demonstrated. Early research in a mouse model [14] established that MHC genotypes can be discriminated based on whole-body odor. Importantly, this same ability to discriminate genotypes was then demonstrated using only the urine collected from live mice (Mus musculus) [15]. More recently, studies using both analytical (GC/MS) and biodetector techniques have demonstrated that urine volatile metabolites are altered by immunization [16], injury [17], and inflammation [18]. These and other studies have led researchers to conclude that excreted metabolites can serve as indicators of changes in an animal’s physiological state.
CWD is a fatal, highly transmissible spongiform encephalopathy caused by an abnormally folded protein called a prion that affects members of the deer family. CWD is a critical management concern in both wild and captive cervid populations. The prion gene, Prnp, and the normal prion protein (PrPC) are both highly conserved among mammals [19]. Studies using knockout mice and cattle and the naturally occurring absence of the gene in some breeds of goats indicated that the gene is not required for survival and to date the precise function of the prion protein is unknown [19].
Prions are present in a number of tissues, including feces and urine in CWD infected animals, suggesting multiple modes of transmission, including animal-to-animal, environmental, and fomite. CWD is expanding its geographic range through two distinct populations: 1) a more rapid spread through captive herds although this had been slowed by import restrictions if cervids across state lines, and 2) slower through natural movements in free-ranging herds. The clinical syndrome was first recognized in 1967 at the northeastern Colorado/southeastern Wyoming border. CWD has now been detected in free-ranging and captive cervids in 32 states and 5 Canadian provinces. It has also been found outside the North American Continent in Asia and Europe with multiple, separate foci identified [19, 20]
CWD management is complicated by the lack of a practical, non-invasive, live-animal screening test. With the onset of the recent COVID-19 pandemic, there has been an expansion in the use of canine biodetectors in the detection of disease, often without a clear understanding of the fundamentals of mammalian olfaction. Through our work with mice, ferrets (Mustela furo), and now dogs (Canis lupus familiaris), we have established an understanding of the variables contributing to successful and unsuccessful mammalian biodetection of disease.
A recent study showed that volatile metabolites identified by gas chromatography/mass spectrometry (GC/MS) in feces can be used to discriminate between CWD-positive and -exposed cervids and disease-free cervids [21]. This suggests that volatile metabolites existing in many fluids or tissues could be used to identify CWD infected herds, individuals, carcasses, and environments by trained biodetectors via odor stimuli. The ability for a biodetector to generalize an odor profile from one or a few tissue types and expand it to a variety of tissue/emanations would be extraordinarily valuable. Antemortem methods currently being developed for the detection of CWD are generally not rapid and/or have varying degrees of applicability to different species. Most methods currently used to detect CWD in both farmed and free-ranging cervid populations are conducted postmortem and developing antemortem methods require sedation or restraint for tissue collection [19]. Trained canine biodetectors would provide a tool for collecting information regarding infection status of both farmed and wild populations without the need for either sedation or restraint. Additionally, canine biodetectors would provide a supplemental postmortem testing method that would augment methods already in use. Canine biodetectors that can learn the odor profile of a single tissue type from an infected individual, may also provide a method for detecting subclinical infections earlier than can be achieved by currently accepted methods and may reliably detect CWD prion contaminated environments.
Animal biodetectors have convincingly demonstrated not only that disease diagnoses may be made based on odor alteration [22–24], but that the underlying volatile signals can be detected in the face of environmental variation [25]. Based on the success of mice in detecting feces collected from avian influenza virus (AIV) infected mallards, we recently trained ferrets to detect AIV infection using fecal volatiles [2]. Not only did trained ferret biodetectors demonstrate that fecal odor was altered in waterfowl by AIV infection, but also that alteration of fecal volatiles was evident to the ferrets before viral shedding occurred with very high sensitivity and specificity.
Dogs were trained to detect and discriminate CWD-infected from non-infected WTD fecal samples in a laboratory setting and were tested in their ability to generalize what was learned from familiar fecal material to novel fecal samples, colon, and novel small intestine samples.
Materials and methods
Ethics statement
The experimental protocols for dog research were approved by the National Wildlife Research Center Institutional Animal Care and Use Committee (QA-2870, QA-3155, and QA-3532).
Biodetectors
Criteria for selecting dogs to serve as potential biodetectors included physical examination, cephalic morphology, and behavioral testing. Once a candidate was identified, other criteria were considered prior to acquisition. This included gender, age, and individual behavioral testing. Dogs were expected to make a discrimination in a complex odor environment that consisted largely of odor cues that will potentially consist of older, decaying sources, as well as fresh fecal samples. Due to these stringent demands, individuals with a strong, enduring motivation to work were considered as having greater potential for success. Six dogs were chosen from shelter and rescue populations from several states and tested for motivation using a modified version of the ball test developed by Working Dogs for Conservation [26].
White-tailed deer feces stimuli
Fecal samples were obtained from WTD. CWD infection status was identified based on real-time quaking-induced conversion (RT-QuIC), immunohistochemistry (IHC), or enzyme-linked immunosorbent assay (ELISA) from brain tissue, lymph node tissue, or both for confirmation of CWD infection. Samples that were confirmed positive for CWD were categorized as “CWD-positive” and samples in which CWD was not detected were categorized as “CWD-negative”.
Cohort I
The National Cervid Health Center provided 352 farmed WTD fecal samples collected during targeted removal events conducted throughout the U.S. Information for samples included regional origin, sex, age, codon 96 genotype, and RT-QuIC or IHC results from brain tissue, lymph node tissue, or both for confirmation of CWD infection. Fecal samples (1 g) were placed in1 ml glass vials with screwcap lids (Qorpak, Bridgeville, PA, USA). Sufficient sample material was provided to yield one or more 1 g aliquots.
Cohort 2
The Tennessee Wildlife Resources Agency (TWRA) provided 166 wild WTD fecal samples collected from hunter harvested animals and animals removed during targeted removal efforts. Samples originated from several counties in western Tennessee. Some of these counties had not yet had a confirmed case of CWD and the others came from areas with reported CWD prevalence ranging from 0.1% to 50%. For example, the number of CWD infected WTD on Ames Plantation brought into the hunter check station was approximately 46%. Sample information included sample origin county and GPS coordinates, sex, age, and CWD infection status determined with CWD ELISA. Fecal samples (1 g) were transferred to 1-mL vials.
Cohort 3
TWRA provided 81 wild WTD gastrointestinal tract (GI) samples collected from hunter harvested and targeted removal deer. Thirty additional were subsequently provided after experiments had begun. Samples originated from several counties in western Tennessee. Sample information included origin (including GPS coordinates), sex, age, and infection status determined by CWD ELISA. Three or more 1 g aliquots of colon or small intestine were place in 1-mL vials following dissection.
Presentation of samples and the dogs’ behavioral response
Laboratory training occurred at the USDA National Wildlife Research Center in Fort Collins, CO using samples from both the Cervid Health Archive (farmed WTD) and samples collected at Ames Plantation in Tennessee (wild WTD). Additionally, field training has been conducted in the CWD-endemic region in southwest Tennessee in which dogs were trained in the busy center of a working farm on Ames Plantation. As with the ferrets [27], dogs were trained using 1 in 5 bioassays. Important procedural changes were made to the sample presentation in comparison to the ferrets. Each of the five scratch boxes were attached to the top of one of two stacked 1 ft 3 scent boxes (Ray Allen Manufacturing, Colorado Springs, CO, USA) so the scratch box (described below) would be close to nose level for all six dogs. The scent box stacks were spaced 1 m apart and dogs were trained to use a passive alert (i.e., sit) when detecting CWD-positive fecal samples (Fig 1).
A session consisted of 20 trials for each of the six dogs with the position of each box being pseudo-randomized for each trial. During initial training for the passive “sit” response, it took 8 sessions to solidify the desired behavioral response.
For a detailed description of the presentation of fecal samples, please refer to Golden et. al., (2021) [27]. Briefly, the scratch boxes were configured to hold a 1 ml glass vial fitted with filter paper that allowed for the escape of volatiles but not the material placed in the vial (~1 g of feces per vial). One randomly positioned box of the five held a vial containing fecal material collected from a CWD infected WTD. The remaining four boxes held vials containing fecal material collected from CWD-negative WTD. Feces from an CWD infected WTD were considered the conditioned stimulus positive (CS+) as the dogs were rewarded for alerting to it and feces from CWD-negative WTD were considered the conditioned stimulus negative (CS-).
For the presentation of GI tract samples, the base compartment of each plastic box (16.48 cm x 12.34 cm x 7.6 cm; Glad Products Company, Oakland, CA, USA) was customized with 1” thick foam cut to allow for the retention of a 4 oz glass mason jar (6.00 cm H x 6.35 cm diameter; Homrove, Amazon, USA). Mason jar collar caps (metal septum-type screw caps with a 6.35 cam diameter opening) were fitted with 70 mm, Whatman qualitative filter paper, grade 1 (Sigma-Aldrich, USA) that allowed for the escape of volatiles but not the material placed in the vial (~1 g of GI colon or small intestine per mason jar). One randomly positioned box of the five held a mason jar containing either a ~1 g section of colon or small intestine collected from a CWD infected WTD. The remaining four boxes held mason jars containing GI tract tissue collected from CWD-negative WTD. Sections of colon or small intestine from a CWD infected WTD were considered the conditioned stimulus positive (CS+) as the dogs were rewarded for alerting to it and sections of colon or small intestine from CWD-negative WTD were considered the conditioned stimulus negative (CS-).
During training and subsequent testing, a “correct” selection by the dog is the alert given to a 0.01 anethole sample (Table 1; Training 1). When an individual dog correctly alerted to the box containing the CS+ sample, a clicker was activated, and the dog was rewarded with either toys (e.g., tennis or lacrosse balls) or various food treats (e.g., home-made tuna treats, Happy Howie’s) depending on which motivated it best. There was no food restriction and water was available ad libitum during training and off-training hours. It is important to note that 20% was the level chance of detecting the CS+ box as only 1 of 5 boxes contained a rewarded sample.
Single-blind procedure
Once dogs proved to be reliable in their response to the CS+, the dogs and the handler were blind to the location of the CS+ to avoid the possibility of the handler inadvertently communicating the position of the CS+ to the dog. An impartial coordinator positioned the CS+ and CS- scratch boxes, signaled the start of a trial, confirmed or rejected the dog handler’s call (described in the next sentence), and picked up the boxes from the previous trial to position the boxes for the next trial. The dog handler controlled when the dogs were to start a trial, called out when a dog alerted to one of the boxes, rewarded the dog if the coordinator confirmed the choice (i.e., clicked the clicker and provided a small food treat or 15 seconds play with a ball or other toy), or without providing verbal or visual cues, turned and walked away if the coordinator called the choice incorrect. The handler and dog were positioned behind a visual barrier prior to the start of each trial as the coordinator prepared the location of the scratch boxes. This procedure was used for all trials that included the shaping of behavior, training, and experimental testing, specifically during rewarded trials but not during unrewarded trials.
Odor alert response and odor discrimination training
For the methods in next three stages of training (Table 1; Training 2–4) please refer to Table 1 and Golden et. al., (2021) [27]. While shaping behavior in these early stages, achieving the target of 80% accuracy average for all dogs was considered sufficient and the next training level was initiated.
We define an extinction trial as a trial consisting of CS+ and CS- samples that are familiar to the dog (i.e., in terms of the quality, intensity, or donor identity) but is not rewarded or acknowledged in any way.
Demonstration of learned response
For detailed explanation of methods in the preparation for Experiments 1 and 2 (Table 2; Demonstration 1, Steps 1–3) please refer to Table 2 and Golden et. al., (2021) [27]. While shaping behavior in these early stages, achieving the target of 80% accuracy average for all dogs was considered sufficient and the next training level was initiated.
Experiment 1—Discrimination of feces from CWD infected WTD
The dogs’ responses to samples collected from novel WTD was examined in unrewarded generalization trials interspersed among unrewarded extinction and rewarded trials in four testing sessions on consecutive days (Table 2, Experiment 1). We defined non-rewarded generalization trials here as a trial consisting of CS+ and CS- samples that are novel to the dog (i.e., in terms of the quality, intensity, or donor identity) but is not rewarded or acknowledged in any way. In each session, there were 14 rewarded trials, 3 unrewarded extinction trials and 3 unrewarded generalization trials. There had to be at least 2 rewarded trials prior to either type of unrewarded trial to begin a session, at least 1 rewarded trial between unrewarded trials, and at least one rewarded trial to end the session. Rewarded trial CS+ samples consisted of randomly chosen familiar individual CWD-positive WTD from Cohorts 1 and 2. Rewarded trial CS- samples consisted of randomly chosen familiar individual CWD-negative WTD from Cohort 1.
Extinction trial CS+ samples (3 per session) consisted of randomly chosen familiar individual CWD-positive WTD from Cohort 1. Extinction trial CS- samples consisted of randomly chosen familiar individual CWD-negative WTD from Cohort 2. Generalization trial CS+ samples (3 per session) consisted of randomly chosen novel individual CWD-positive WTD from Cohort 1. Generalization trial CS- samples consisted of randomly chosen novel individual CWD-negative WTD from Cohort 1.
Experiment 2—Transition to detection of CWD-positive volatile odors in tissue samples
To determine if dogs can be trained to generalize responses from the odor profile of feces to the odor profile of a GI tissue sample from the same animal, dogs previously trained with defecated feces were also trained with samples of excised colon (approximately 1g and containing feces) in all rewarded generalization trials consisting of a CS+ and four CS- samples and similarly tested with unfamiliar samples not offered during training.
Six days of testing sessions were conducted over two weeks. In each session, there were 14 rewarded trials, 3 unrewarded extinction trials and 3 unrewarded generalization trials. There were at least 2 rewarded trials prior to either type of unrewarded trial to begin a session, at least 1 rewarded trial between unrewarded trials, and at least on rewarded trial to end the session. The reward trials were a mixture of colons and small intestines tissues collected from familiar donors. The extinction trials were familiar colon and small intestine samples. The generalization trials consisted of an entire panel of novel small intestine samples.
Field training—Discrimination of feces in the field
Following two days of blind, all reward training sessions, the dogs’ responses to samples collected from WTD were examined in unrewarded extinction trials interspersed among rewarded trials in a single outdoor single-blind training session (Table 2, Field Training). In this session, there were 14 rewarded trials and 6 unrewarded extinction trials. There had to be at least 2 rewarded trials prior to either type of unrewarded trial to begin a session, at least 1 rewarded trial between unrewarded trials, and at least one rewarded trial to end the session.
Rewarded trial CS+ samples consisted of randomly chosen familiar individual CWD-positive WTD from Cohort 2. Rewarded trial CS- samples consisted of randomly chosen familiar individual CWD-negative WTD from Cohorts 1 and 2. Extinction trial CS+ samples (6 per session) consisted of randomly chosen familiar individual CWD-positive WTD from Cohort 2. Extinction trial CS- samples consisted of randomly chosen familiar individual CWD-negative WTD from Cohorts 1 and 2.
Data analysis
Cumulative responses across all trained dog trials were calculated for each set of experimental generalization trials. Success rates (number of correct trials divided by the total number of generalization trials) were statistically evaluated using binomial proportion tests with a continuity correction for small numbers of observations [28]. The data were tested for independence between donor identity, testing day, and correct responses using the Cochran-Mantel-Haenszel test [29]. For all experiments, a success rate of 20% would be expected by chance. However, as the goal of these trials was to demonstrate the high specificity of trained biodetectors, dog responses were also compared to 50% and 80% success rates.
Results
Demonstration of learned response
After 11 practice days of all rewarded trials with a mix of both Cohorts, extinction trials were added to prepare for experiment 1. For two days, dogs were presented with a mix of both Cohorts in the rewarded trials. For the extinction trials, dogs were presented with panels that consisted of Cohort 2 samples for the four CS- samples and a CS+ from Cohort 1. Dogs performed at 85% accuracy with an average sensitivity of 87% and an average specificity of 96% for the rewarded trials. Dogs performed at 92% accuracy with an average sensitivity of 92% and an average specificity of 98% for the extinction trials (Fig 2).
During training, dogs correctly (black bars) identified CWD infection in a single fecal sample (n = 240; Cohort 1 and 2) from a CWD infected WTD presented among four CWD-negative samples with greater than 20% accuracy (chance). During rewarded testing trials, dogs correctly identified CWD infection with 85% accuracy. During unrewarded extinction trials, dogs correctly identified infected samples from novel WTD from uninfected samples with 92% accuracy. White bars represent incorrect choices.
Experiment 1—Discrimination of feces from CWD infected WTD
Trained dogs were highly accurate (85% correct choices in unrewarded generalization trials over 4 testing days) at discriminating between fecal samples collected from CWD-negative WTD and from CWD-positive WTD (Fig 3).
During rewarded testing trials, dogs correctly (black bars) identified CWD infection with 90% accuracy. During unrewarded generalization trials, dogs correctly identified CWD-positive samples from novel WTD from CWD-negative samples with 85% accuracy, which was greater than null hypotheses of 20% (*p < 0.0001) and 50% (†p < 0.0001), but not different from 80% (p = 0.1500). White bars represent incorrect choices.
All six trained dogs correctly identified CWD infection with 85% accuracy (Fig 3, middle bars) across four days of rewarded testing trials. During unrewarded generalization trials, dogs correctly identified CWD infection with 85% accuracy (Fig 3, right two bars). This result is different from chance (20%; p < 0.0001) and a 50% success rate (p < 0.0001) and is not different from an 80% success rate (p = 0.15). Non-parametric analysis indicated that detection ability did not vary among dogs, and test day was not associated with the selection made by the dogs in either rewarded testing (p = 0.3562) or unrewarded generalization trials (p = 0.4407).
Experiment 2—Transition to detection of CWD-positive volatile odors in tissue samples
All six trained dogs correctly identified CWD infection in colon samples with 96% accuracy and in small intestine samples with 76% accuracy (Fig 4a) over six days of rewarded trials. During unrewarded extinction trials (familiar samples) trained dogs correctly identified CWD infection in colon samples with 94% accuracy and in small intestine samples with 88% accuracy (Fig 4b). During unrewarded generalization trials, dogs correctly identified CWD infection in small intestine samples with 79% accuracy (Fig 4c). This result is different from 20% (p < 0.0001), and 50% (p < 0.0001), but not 80% success rates (p = 0.8597). Non-parametric analysis indicated that detection ability did not vary among dogs, and test day was not associated with the selection made by the dogs (p = 0.8452) in these unrewarded generalization trials (p = 0.4031).
A) During 419 rewarded testing trials with a colon sample CS+, dogs correctly (black bars) identified CWD infection with 96% accuracy. During 85 rewarded testing trials with a small intestine sample CS+, dogs correctly (black bars) identified CWD infection with 76% accuracy. B) During 52 extinction testing trials with a colon sample CS+, dogs correctly (black bars) identified CWD infection with 94% accuracy. During 92 extinction testing trials with a small intestine sample CS+, dogs correctly (black bars) identified CWD infection with 88% accuracy. C) During the 72 unrewarded generalization trials, dogs correctly identified CWD infection in small intestine samples with 79% accuracy (*p < 0.0001) in comparison to CWD-negative samples. These classification rates are greater than the null hypotheses of 20% and 50% (†p < 0.0001) success rates, but not for the 80% success rate (p = 0.8597). White bars represent incorrect choices.
Field training—Discrimination of feces in the field
Following two days of all rewarded, blind trials, all six trained dogs correctly identified CWD infection in fecal samples with 78% accuracy (Fig 5; left bars) over two days of rewarded trials. During a single day of rewarded trials and unrewarded extinction trials (familiar samples) trained dogs correctly identified CWD infection in fecal samples with 87% accuracy during rewarded trials (Fig 5; center bars) and 81% accuracy during unrewarded extinction trials (Fig 5; right bars). This final result is different from 20% (p < 0.0001), and 50% (p < 0.0001), but not 80% success rates (p = 0.4668). Non-parametric analysis indicated that detection ability did not vary among dogs (p = 0.9162) and test day was not associated with the selection made by the dogs (p = 0.1516) in these unrewarded extinction trials.
During field training, dogs correctly (black bars) identified CWD infection in a single fecal sample (n = 240; Cohort 2; left bars) from a CWD infected WTD presented among four CWD-negative samples with greater than 20% accuracy (chance). During rewarded trials in the session containing extinction trials, dogs correctly identified a fecal sample from a CWD infected WTD presented among four CWD-negative samples with 87% accuracy (n = 84; Cohort 2; center bars). During unrewarded extinction field trials, dogs correctly identified CWD-positive fecal samples from WTD from CWD-negative samples with 81% accuracy (center bars), which was greater than null hypotheses of 20% (*p < 0.0001) and 50% (†p < 0.0001), but not different from 80% (p = 0.4668). White bars represent incorrect choices.
Discussion
This study builds upon previous experiments demonstrating the ability of a mammalian biodetectors to identify disease infection in an animal with high rates of accuracy: e.g. trained mice demonstrated better than 80% accuracy in identifying AIV infection with fecal odors from ducks and ferrets demonstrated better than 90% accuracy for detecting AIV infected ducks [2]. The results of these two experiments and field training provide clear-cut evidence that dogs, are not only capable of performing an olfactory learning discrimination task but can generalize the odor profile learned from fecal samples to other tissues that are not related to the training material. And while comorbidities are a possibility, or most likely a probability, in wildlife, based on the outcome and high accuracy displayed by mice [3], ferrets [2] and dogs, we are convinced that the mammals being trained can be forced to focus on the infection status of the particular pathogen in question and to ignore the possibly confounding odors of comorbidities that may be also infecting an individual subject.
Importantly, our results suggest that dogs could potentially transition from fecal odor profiles to odor profiles by training them with gastrointestinal tracts dissected into colon and small intestine sections. The colon sections contained fecal material. The small intestine sections did not although they did contain the digested precursor material to feces. Based on the dogs’ response to novel small intestine sections we suggest that dogs can generalize what they learned from the odor profile of fecal samples to other tissues collected from a CWD-positive WTD.
This study echoes previous work demonstrating that dogs can be trained to identify CWD from feces of infected WTD [20]. In that study, Mallikarjun et al. (2023) presented samples on an eight-port scent wheel that became popular with dog cancer detection studies. As in cancer detection studies, the dogs were permitted to explore the wheel without the handler being present allowing for double-blind studies. The current study employed a 1-in-5 bioassay that replicated methods used by the Belgian Police in human scent line-ups [30] such that the same panel could be replicated multiple times with the samples in different positions [31]. In addition, more than three times the number of samples (n = 518; 120 CWD positive and 405 non-infected samples) were presented in the current study. This is important because it has been shown that dogs can easily maintain a working memory of up to 76 odors during an odor memory task [32]. But the most significant aspects of the current study were the testing of GI tract samples following training with fecal samples and the superior CWD sensitivity reported here.
CWD management is complicated by the lack of a practical, rapid, non-invasive, live-animal screening tests. This study adds to our understanding of the fundamentals of mammalian olfaction and how it can be leveraged to be used as an important tool in the surveillance and management of CWD. We have established an understanding of what variables contribute to successful and unsuccessful mammalian biodetection of CWD. Another critical tool would be to determine if dogs can detect the urine and feces of CWD infected deer at historical bait stations and long vacated salt licks. Or the detection of prions at scrapes in high prevalence locations. Another interesting question is do inert prions, free from a host, interact with the natural microbiome of the soil itself producing an odor specific change in the volatile profile of the soil making it possible to detect large deposits of prions?
In the previous ferret study, ferrets could detect the volatile odor profile associated with avian influenza virus infection before the virus was detectable by qPCR [2]. We have no evidence to present that the dog trained to detect the results of CWD infection can do something similar, but it would be very interesting to investigate this question. The samples from farmed WTD donated by the National Cervid Health Archive were tested by RT-QuIC or IHC using either lymph node or brain tissue or both. The dogs did not show a difference in their capability to detect any of the lymph node positive/brain negative versus lymph node positive/brain positive samples from Cohort 1, but it would be beneficial to have fecal samples collected from a staged infection study so the dogs could be challenged with samples from the earliest stages of CWD infection.
The current study provides evidence that biodetectors can be used to detect fecal and gastrointestinal tract samples from WTD infected with CWD. This evidence shows that dogs can identify the signature odor that is a result of infection from CWD whether it be in a fecal sample or in a small intestine sample. Regardless, these experiments raised some new questions that need to be addressed. For example, analysis of the data shows that the accuracy of the dogs was diminished by 10% when the wild WTD samples were introduced. This suggests that the volatile odor profile of CWD infected wild WTD may be sufficiently different from farmed CWD infected WTD; or at least different enough to challenge the perception of the volatile odor profile the dogs learned in training on the samples collected from farmed WTD. Most importantly, it will be interesting to determine whether a biodetector trained to identify CWD in WTD fecal samples could identify the odor identity of CWD in another species. This study clearly demonstrates the feasibility of deploying trained biodetectors for CWD surveillance in WTD.
Acknowledgments
The authors thank Jeremy Dennison and his TWRA staff for assistance in sample collection from wild WTD. Special thanks to Stephanie Karns and Sonia Mongold for being the TWRA liaison and facilitating our unique needs. Special thanks to Amelia Johnston and Jacki Young for some data analysis and canine husbandry. Mention of specific products does not constitute endorsement by Colorado State University, the United States Department of Agriculture or Monell Chemical Senses Center.
References
- 1. Ellis CK, Rice S, Maurer D, Stahl R, Waters WR, Palmer MV, et al. Use of fecal volatile organic compound analysis to discriminate between non-vaccinated and BCG-Vaccinated cattle prior to and after Mycobacterium bovis challenge. PLoS One. 2017;12(7):e0179914. Epub 2017/07/08. pmid:28686691.
- 2. Golden GJ, Grady MJ, McLean HE, Shriner SA, Hartwig A, Bowen RA, et al. Biodetection of a specific odor signature in mallard feces associated with infection by low pathogenic avian influenza A virus. PLoS One. 2021;16(5):e0251841. Epub 2021/05/27. pmid:34038460.
- 3. Kimball BA, Yamazaki K, Kohler D, Bowen RA, Muth JP, Opiekun M, et al. Avian influenza infection alters fecal odor in mallards. PLoS One. 2013;8(10):e75411. pmid:24146753.
- 4. Bergmann A, Trefz P, Fischer S, Klepik K, Walter G, Steffens M, et al. In Vivo Volatile Organic Compound Signatures of Mycobacterium avium subsp. paratuberculosis. PLoS One. 2015;10(4):e0123980. Epub 2015/04/29. pmid:25915653.
- 5. Purkhart R, Kohler H, Liebler-Tenorio E, Meyer M, Becher G, Kikowatz A, et al. Chronic intestinal Mycobacteria infection: discrimination via VOC analysis in exhaled breath and headspace of feces using differential ion mobility spectrometry. J Breath Res. 2011;5(2):027103. Epub 2011/04/23. pmid:21512209.
- 6. Stahl RS, Ellis CK, Nol P, Waters WR, Palmer M, VerCauteren KC. Fecal Volatile Organic Ccompound Profiles from White-Tailed Deer (Odocoileus virginianus) as Indicators of Mycobacterium bovis Exposure or Mycobacterium bovis Bacille Calmette-Guerin (BCG) Vaccination. PLoS One. 2015;10(6):e0129740. Epub 2015/06/11. pmid:26060998.
- 7. Ellis CK, Stahl RS, Nol P, Waters WR, Palmer MV, Rhyan JC, et al. A pilot study exploring the use of breath analysis to differentiate healthy cattle from cattle experimentally infected with Mycobacterium bovis. PLoS One. 2014;9(2):e89280. Epub 2014/03/04. pmid:24586655.
- 8. Rudnicka J, Walczak M, Kowalkowski T, Jezierski T, Buszewski B. Determination of volatile organic compounds as potential markers of lung cancer by gas chromatography-mass spectrometry versus trained dogs. Sensor Actuat B-Chem. 2014;202:615–21.
- 9. Sakumura Y, Koyama Y, Tokutake H, Hida T, Sato K, Itoh T, et al. Diagnosis by Volatile Organic Compounds in Exhaled Breath from Lung Cancer Patients Using Support Vector Machine Algorithm. Sensors-Basel. 2017;17(2). ARTN 287, pmid:28165388
- 10. Zhu JJ, Bean HD, Wargo MJ, Leclair LW, Hill JE. Detecting bacterial lung infections: in vivo evaluation of in vitro volatile fingerprints. Journal of Breath Research. 2013;7(1). Artn 016003, pmid:23307645
- 11. De Moraes CM, Wanjiku C, Stanczyk NM, Pulido H, Sims JW, Betz HS, et al. Volatile biomarkers of symptomatic and asymptomatic malaria infection in humans. P Natl Acad Sci USA. 2018;115(22):5780–5. pmid:29760095
- 12. Kimball BA, Wilson DA, Wesson DW. Alterations of the volatile metabolome in mouse models of Alzheimer’s disease. Sci Rep-Uk. 2016;6. ARTN 19495 pmid:26762470
- 13. Maurer DL, Koziel JA, Engelken TJ, Cooper VL, Funk JL. Detection of Volatile Compounds Emitted from Nasal Secretions and Serum: Towards Non-Invasive Identification of Diseased Cattle Biomarkers. Separations. 2018;5(1). ARTN 18,
- 14. Yamazaki K, Yamaguchi M, Baranoski L, Bard J, Boyse EA, Thomas L. Recognition among mice. Evidence from the use of a Y-maze differentially scented by congenic mice of different major histocompatibility types. J Exp Med. 1979;150(4):755–60. Epub 1979/10/01. pmid:512584.
- 15. Yamaguchi M, Yamazaki K, Beauchamp GK, Bard J, Thomas L, Boyse EA. Distinctive urinary odors governed by the major histocompatibility locus of the mouse. Proc Natl Acad Sci U S A. 1981;78(9):5817–20. Epub 1981/09/01. pmid:6946517.
- 16. Kimball BA, Opiekun M, Yamazaki K, Beauchamp GK. Immunization alters body odor. Physiol Behav. 2014;128:80–5. pmid:24524972.
- 17. Kimball BA, Cohen AS, Gordon AR, Opiekun M, Martin T, Elkind J, et al. Brain Injury Alters Volatile Metabolome. Chemical Senses. 2016;41(5):407–14. pmid:26926034
- 18. Millet P, Opiekun M, Martin T, Beauchamp GK, Kimball BA. Cytokine contributions to alterations of the volatile metabolome induced by inflammation. Brain Behav Immun. 2018;69:312–20. pmid:29241669
- 19. Haley NJ, Hoover EA. Chronic wasting disease of cervids: current knowledge and future perspectives. Annu Rev Anim Biosci. 2015;3:305–25. pmid:25387112.
- 20. Mallikarjun A, Swartz B, Kane SA, Gibison M, Wilson I, Collins A, et al. Canine detection of chronic wasting disease (CWD) in laboratory and field settings. Prion. 2023;17(1):16–28. pmid:36740856.
- 21. Ellis CK, Volker SF, Griffin DL, VerCauteren KC, Nichols TA. Use of faecal volatile organic compound analysis for ante-mortem discrimination between CWD-positive, -negative exposed, and -known negative white-tailed deer (Odocoileus virginianus). Prion. 2019;13(1):94–105. Epub 2019/04/30. pmid:31032718.
- 22. Bijland LR, Bomers MK, Smulders YM. Smelling the diagnosis A review on the use of scent in diagnosing disease. Neth J Med. 2013;71(6):300–7. pmid:23956311
- 23. Brooks SW, Moore DR, Marzouk EB, Glenn FR, Hallock RM. Canine Olfaction and Electronic Nose Detection of Volatile Organic Compounds in the Detection of Cancer: A Review. Cancer Invest. 2015;33(9):411–9. pmid:26114998
- 24. Oh Y, Lee Y, Heath J, Kim M. Applications of Animal Biosensors: A Review. IEEE Sens J. 2015;15(2):637–45.
- 25. Kwak J, Willse A, Matsumura K, Curran Opiekun M, Yi W, Preti G, et al. Genetically-based olfactory signatures persist despite dietary variation. PLoS ONE. 2008;3(10):e3591. Epub 2008/11/01. pmid:18974891.
- 26.
Conservation WDf. Rescues2theRescue 2024 [cited 2024 January 3]. https://www.rescues2therescue.org/.
- 27. Golden GJ, Opiekun M, Martin-Taylor T, Kimball BA. Training the domestic ferret to discriminate odors associated with wildlife disease. PLoS One. 2021;16(11):e0259415. Epub 20211101. pmid:34723972.
- 28. Carriere KC. How good is a normal approximation for rates and proportions of low incidence events? Communications in Statistics-Simulation and Computation. 2001;30(2):327–37.
- 29. Landis JR, Heyman ER, Koch GG. AVERAGE PARTIAL ASSOCIATION IN 3-WAY CONTINGENCY-TABLES—REVIEW AND DISCUSSION OF ALTERNATIVE TESTS. International Statistical Review. 1978;46(3):237–54.
- 30.
Schoon AH, Rudd. K9 Suspect Discrimination—Training and practicing scent identification line-ups. Canada: Detselig Enterprises Ltd.; 2002.
- 31. Beyene N, Sitotaw AL, Telila EG, Gebre HA, Alemu NB, Burny R, et al. The use of rats to detect drug-resistant TB. Public Health Action. 2023;13(1):1–3. pmid:37152211.
- 32. Krichbaum S, Rogers B, Cox E, Waggoner LP, Katz JS. Odor span task in dogs (Canis familiaris). Anim Cogn. 2020;23(3):571–80. Epub 20200225. pmid:32100208.