Trained dogs can accurately discriminate between scents of saliva samples from dogs with cancer versus healthy controls

Laurie A. Malone School of Public Health, The University of Alabama at Birmingham, Birmingham, AL

Search for other papers by Laurie A. Malone in
Current site
Google Scholar
PubMed
Close
 PhD, MPH
,
MacKenzie A. Pellin Department of Medical Sciences, School of Veterinary Medicine, University of Wisconsin-Madison, Madison, WI

Search for other papers by MacKenzie A. Pellin in
Current site
Google Scholar
PubMed
Close
 DVM, DACVIM, DACVR
, and
Kendal M. Valentine Department of Medical Sciences, School of Veterinary Medicine, University of Wisconsin-Madison, Madison, WI

Search for other papers by Kendal M. Valentine in
Current site
Google Scholar
PubMed
Close

Abstract

OBJECTIVE

To determine whether dogs can be trained to utilize olfaction to differentiate between saliva samples from dogs with cancer and those from healthy control dogs.

SAMPLE

Canine patient saliva samples were collected (October 2020 to July 2022) from 139 dogs diagnosed with malignant tumors and from 161 healthy dogs (control samples) for use during training and testing of the dog detection team. Samples from canine patients were collected prior to treatment with radiation therapy or chemotherapy.

PROCEDURES

Six pet dogs (mean ± SD age, 5.4 ± 1.9 years) were trained for odor discrimination between healthy control and malignant tumor samples. Training of the dogs, using a reward-based positive reinforcement method, took place 1 to 3 times per week for a period of 6 months (January to June 2022). After training was complete, a subset of samples not utilized during the training sessions were selected for use during odor discrimination testing of the dog team.

RESULTS

The trained dogs could accurately distinguish between samples from cancer patients versus control dogs with a mean sensitivity of 90% and mean specificity of 98%, and with mean positive and negative predictive values of 95%.

CLINICAL RELEVANCE

This study serves as preliminary evidence that dogs can be trained to detect differences in scent between saliva samples from cancer and normal patients. Further studies should expand upon these results with a larger sample, varied tumor types, use of non-cancer diseases as controls, and exploration of this technique in feline patients.

Abstract

OBJECTIVE

To determine whether dogs can be trained to utilize olfaction to differentiate between saliva samples from dogs with cancer and those from healthy control dogs.

SAMPLE

Canine patient saliva samples were collected (October 2020 to July 2022) from 139 dogs diagnosed with malignant tumors and from 161 healthy dogs (control samples) for use during training and testing of the dog detection team. Samples from canine patients were collected prior to treatment with radiation therapy or chemotherapy.

PROCEDURES

Six pet dogs (mean ± SD age, 5.4 ± 1.9 years) were trained for odor discrimination between healthy control and malignant tumor samples. Training of the dogs, using a reward-based positive reinforcement method, took place 1 to 3 times per week for a period of 6 months (January to June 2022). After training was complete, a subset of samples not utilized during the training sessions were selected for use during odor discrimination testing of the dog team.

RESULTS

The trained dogs could accurately distinguish between samples from cancer patients versus control dogs with a mean sensitivity of 90% and mean specificity of 98%, and with mean positive and negative predictive values of 95%.

CLINICAL RELEVANCE

This study serves as preliminary evidence that dogs can be trained to detect differences in scent between saliva samples from cancer and normal patients. Further studies should expand upon these results with a larger sample, varied tumor types, use of non-cancer diseases as controls, and exploration of this technique in feline patients.

Introduction

Cancer is one of the leading causes of death in canines. It is generally estimated that 1 in 4 dogs will be afflicted by cancer over the course of their lifespan, increasing to estimates of 1 in 2 dogs over the age of 10 years.1 More recent data derived from the Dog Aging Project with owner-reported data showed a lifetime prevalence of 29.7 malignant tumors per 1,000 dogs, and 14.9 benign tumors per 1,000 dogs.2 In this study,2 as well as others,3 risk of cancer increased with increasing age and body size. The most frequently diagnosed cancers worldwide are mammary cancer, lymphoma, and skin cancer.4 Certain breeds are believed to be at risk for specific neoplasia, which is thought to arise due to inadvertent selection of genetic mutations predisposing to cancer, along with the selection of specific genetic traits.47

Early detection of cancer is desirable because it may reduce the need for aggressive treatment and the incidence of metastasis. In addition, it may improve survival time in both human and veterinary patients. Furthermore, early detection can be the key to effective treatment, and the chance to improve the length and quality of life of cancer patients. However, widespread neoplasia screening programs do not currently exist for our canine patients. Yearly wellness examinations, imaging studies such as radiography and ultrasonography, bloodwork and urinalysis, and diagnostic tests such as fine needle aspiration (FNA) and surgical biopsy are useful for diagnosing cancer, but do not typically allow for detection of canine cancer at the preclinical stage.8 These tests also come with additional costs, and with 45% of households in the US owning dogs,9 may be too expensive for some owners to perform routinely. Additionally, FNA and surgical biopsy are more invasive tests, requiring sedation if not general anesthesia. Even imaging tests, while considered noninvasive overall, may require sedation in some patients to facilitate positioning. Therefore, the importance of developing novel diagnostic approaches, especially low cost, and noninvasive options, for early neoplasia detection in both cats and dogs is crucial as cancer burden seems to be on the rise.

To our knowledge, case reports of owner-pet dog interactions leading to diagnosis of human cancer via scent detection were first documented in 1989.10,11 Following these case reports, programs were developed to determine if dogs could be trained to detect cancer. An early study trained dogs to detect differences in breath samples from human breast and lung cancer patients compared to healthy controls with high accuracy.12 There now exists numerous studies that have evaluated the ability of dogs to detect samples from human patients with various types of cancer, although with mixed results. These include detecting melanoma, ovarian, lung, prostate, and breast cancer by perceiving a characteristic “odor signature” in excretions including urine, sweat, breath, and blood serum.1220

These unique odor patterns are believed to be due to volatile organic compounds (VOCs). VOCs include alkane and aromatic compounds in exhaled breath, urine, blood, and colon contents.8 They have been identified in human cancer patients utilizing gas chromatography/mass spectroscopy.21,22 Empirical evidence suggests that dogs can detect human cancer in the early stages of development as the canine nose has scent detection concentration thresholds as low as 1 to 2 parts per trillion, roughly 10,000 to 100,000 times that of humans.23 Further evidence of the utility of canine scent detection exists with police and civilian services that have utilized trained dogs to detect explosives, drugs, missing persons and endangered wildlife with the help of their olfactory system.24

Studies thus far1220 highlight the extraordinary ability of dogs to distinguish complex odors related to human cancers, however, further investigation is needed to determine if this applies to dogs detecting canine cancers. Most cancers seen in dogs are almost identical to the cancers that affect people with similar biological behavior, patterns of metastasis, and speed of cancer growth.25 However, to our knowledge, current literature on canine scent detection of cancer in canines is limited to a single study.26 In 2017, scientists utilized scent detection dogs to identify urothelial carcinoma in other dogs.26 While the study did provide important new information about the feasibility of utilizing scent detection dogs to detect canine cancer, the results did not support that dogs could detect urothelial carcinoma in urine.26 One challenge that may have hindered the experiment was the possibility that trained dogs remembered individual dog’s urine odor rather than detecting the specific neoplasia odor; differences in sample handing may have also affected the “odor signature”; additionally, controls included dogs with noncancer diseases of the urinary tract as well as clinically normal dogs, which likely lead to additional challenges in the identification of urothelial carcinoma.

Canine scent detection is a noninvasive technique which needs to be explored more thoroughly as a promising screening tool for diagnosis of cancer in dogs, with the potential to improve the overall quality of life, and lifespan of our cancer patients. The objective of this preliminary study was to measure the ability of 6 privately owned (aka “pet”) dogs trained to utilize olfaction to differentiate between saliva samples from dogs with cancer and those from healthy control dogs. Our hypothesis was that dogs can be trained to distinguish between saliva samples from dogs diagnosed with cancer and healthy controls through scent with high accuracy.

Materials and Methods

After obtaining informed consent from owners, saliva samples were collected from patients visiting University of Wisconsin Veterinary Care, the clinical teaching hospital associated with the University of Wisconsin–Madison School of Veterinary Medicine. Canine patient saliva samples were collected from individuals of any age diagnosed with malignant tumors and from healthy controls. Clinical data collected for all participants included breed, medical record number, age, sex, spay/neuter status, body condition score, and current medications. Healthy dogs were defined as being 6 years old or less with no current illness based upon clinical history or physical examination by a veterinarian. Bloodwork or imaging studies were not routinely performed for healthy dogs.

For all dogs, saliva samples were collected using 6 cotton-tipped applicators swabbed on the buccal mucosa for at least 30 seconds after food and water had been withheld for at least 1 hour. After collection, each swab sample was placed in a sterile plastic sleeve, given an identification number, and stored at 2 to 8 °C until shipped on ice to the dog training facility for storage at –24 °C until training use. Additional samples to meet training number requirements were collected from two local veterinary clinics in the Birmingham, Alabama region, and friends/family affiliated with Sprout and Penny Canine Foundation’s “Sniff 4 Life” program.

Additional information recorded for canine patients with cancer included tumor type, current therapy, and other diseases. Samples were collected prior to treatment with radiation therapy or chemotherapy, as it is unknown how these treatment modalities would affect scent detection. Samples were collected if neoplastic disease was measurable based upon palpation or imaging tests; patients with microscopic disease were excluded as it is not known how this would affect concentrations of VOCs and scent detection. Samples were collected from patients with both solitary tumors, as well as metastatic disease. Samples from patients with recurrent tumors were permissible as long as samples were collected prior to radiation and chemotherapy. Surgical biopsy and/or fine-needle aspirate was collected and submitted for histo- or cytopathology, as appropriate, to confirm the cancer diagnosis. This was most often performed on the same day as saliva swab collection, but for some dogs, confirmatory histopathology was performed at a future visit. The date of diagnosis, stage, and tumor pathology findings were recorded once received. Saliva samples from patients diagnosed with various tumor types were collected in this study as it is desirable for the scent detection dogs to demonstrate the ability to detect a wide variety of cancer types. Samples were collected from dogs diagnosed with cancers from the following categories: round-cell tumors (large cell lymphoma, mast cell tumors), sarcomas (osteosarcoma, fibrosarcoma, hemangiosarcoma), carcinomas (squamous cell carcinoma, pulmonary carcinoma, thyroid tumors, transitional/urothelial cell carcinoma), and tumors that do not fit into the above categories well, such as melanoma. Samples were collected from October 2020 through July 2022.

A total of 139 samples from dogs with malignant tumors, and 161 canine healthy control samples were collected for use in training and testing of the medical scent detection (MSD) dogs. Twenty-one samples were excluded from the malignant tumor category for the following reasons: lack of histopathologic diagnosis; histopathology type outside of those being investigated; collection in the microscopic disease setting; or commencement of radiation or chemotherapy at the time of sample collection.

Training of the MSD dog team

Six dogs were trained for odor discrimination between samples from healthy controls and samples from dogs with malignant tumors. The MSD dog team included a mixed breed (spayed female, age 7 years), a Weimaraner (spayed female, age 8 years), and 4 Beagles (1 intact male [age 2.5 years], 2 neutered males [age 5 years], 1 spayed female [age 5 years]), with a mean ± SD age of 5.4 ± 1.9 years. Five of the 6 dogs included in the training program had a minimum of 2 years training and competition experience in the sport of canine scent work.27 The sixth dog had completed 6 months of canine scent work training. Training of the MSD dog team utilized a positive-reinforcement method and took place at an established dog training facility in Vestavia Hills, AL. The dogs were transported by their owners to the training facility 1 to 3 times per week for a period of 6 months (January to June 2022). Each training session lasted 2 to 3 hours with dogs running through various scent detection exercises (Figure 1). Number of training exercises per session ranged from 4 to 10, with number of samples searched each exercise ranging from 6 to 18. This variation in number of training exercises and samples searched each session was done to ensure the MSD dogs remained interested and engaged over the course of training. For each session, a randomly determined running order for the dogs was set and then the dogs rotated turns for the training exercises scheduled that day. All training was conducted on leash using reward-based methods.

Figure 1
Figure 1

Images of the 6 pet dogs comprising the medical scent detection (MSD) dog team performing various scent training activities (January to June 2022) to differentiate between saliva samples from dogs with versus without cancer. The different types of apparatuses used for training and testing in which stainless steel canisters holding the saliva samples were placed are shown: a 6-arm carousel (A, B, D), holder made from polyvinylchloride tubing (C and E), and a polyvinylchloride pole secured in a concrete base with a stainless steel utensil holder mounted sideways to hold the sample canister (F).

Citation: Journal of the American Veterinary Medical Association 261, 6; 10.2460/javma.22.11.0486

Over the course of training, a variety of scent rack apparatuses (composed of stainless steel and/or polyvinylchloride) and configurations were used (Figure 1). For each training session, pre-selected sample swabs were placed into stainless-steel shaker style canisters that had small holes on top. The individual canisters were placed in pre-determined locations on the training apparatus, with samples and sample locations changing for each training exercise.

Saliva samples from a total of 187 dogs were used for training. From the collection of samples, samples from patients with malignant tumors (n = 78) as well as healthy controls (n = 109) were utilized for the scent detection training. Samples used during training were purposefully chosen from the pool of samples to provide the MSD dogs with the most varied exposure in terms of cancer type, sex, age, and breed of dog. Before beginning the first phase of training, the dogs were conditioned to methodically search each scent apparatus using food treats as a stimulus. Next, during the initial stages of training, samples from dogs with malignant tumors were paired with a food reward. As the dogs reached key milestones, the training protocol progressed from sample paired with food, shaping an alert response (eg, sit, down) by the dog at odor (ie, cancer sample), to blind exercises (location of cancer sample[s] unknown to handler) with each dog providing a distinct alert behavior to indicate the location of the cancer sample(s). During the later stages (ie, blind exercises) correct responses were confirmed by the trainer prompting the handler to provide the dog with a food reward at source.

Testing of the MSD dog team

After training was complete, a subset of samples not utilized during the training sessions were selected for use during testing (Table 1). When selecting samples from patients with malignant tumors, care was taken to select patients with a variety of tumor types, as well as a variety of ages and genders, to offer a wide range of characteristics to challenge the MSD dogs.

Table 1

Characteristics of samples used for testing the MSD dogs.

Sample ID Sample group Sample classification Specific cancer type Dog age Dog breed BCS (1–9)
DC-048 Cancer Round-cell tumor Lymphoma 8 Golden Retriever 6
DC-049 Cancer Round-cell tumor Lymphoma 10 Miniature Schnauzer 5
DC-060 Cancer Round-cell tumor Lymphoma 7 Brittany Spaniel 7
DC-101 Cancer Round-cell tumor Mast cell tumor 3 French Bulldog 6
DC-052 Cancer Sarcoma Osteosarcoma 8 German Shepherd Dog 7
DC-081 Cancer Sarcoma Melanoma 9 German Shepherd Dog mix 5
DC-098 Cancer Sarcoma Soft tissue sarcoma 5 Siberian Husky 5
DC-072 Cancer Carcinoma Squamous cell carcinoma 14 Mixed 4
DC-095 Cancer Carcinoma Urothelial/transitional cell carcinoma 11 Pit bull–type dog 6
DC-096 Cancer Carcinoma Urothelial/transitional cell carcinoma 11 Shih Tzu mix 5
DH-065 Healthy - - 7 Labrador Retriever 5.5
DH-066 Healthy - - 5 Labrador Retriever 5.5
DH-125 Healthy - - 5 Mixed 9
DH-091 Healthy - - 2 Mixed 5.5
DH-092 Healthy - - 5 Chihuahua 4
DH-093 Healthy - - 4 Mixed 6
DH-094 Healthy - - 1.3 Labrador Retriever 5
DH-095 Healthy - - 1 Mixed 4
DH-096 Healthy - - 1 Mixed 5
DH-098 Healthy - - 1 Pit bull–type dog 6
DH-099 Healthy - - 4 Bernese Mountain Dog 4
DH-126 Healthy - - 3 Labrador Retriever 4.5
DH-127 Healthy - - 6 Labrador Retriever 4.5
DH-129 Healthy - - 2 Labrador Retriever 4
DH-130 Healthy - - 3 Goldendoodle 4
DH-131 Healthy - - 5 Black Mouth Cur 6
DH-132 Healthy - - 4 German Shepherd Dog mix 5
DH-133 Healthy - - 6 Mixed 5
DH-134 Healthy - - 1 Mixed 5
DH-135 Healthy - - 5 Labrador Retriever 5
DH-136 Healthy - - 5 Bernese Mountain Dog 7
DH-137 Healthy - - 6 Labrador Retriever 7

MSD = Medical scent detection. BCS = Body condition score.

In total, 10 samples from dogs with malignant tumors were utilized for testing. Mean ± SD age was 8.6 ± 3.2 years (range, 3 to 14 years). For the samples from dogs with malignant tumor, breeds included Golden Retriever (n = 1), Miniature Schnauzer (1), Brittany Spaniel (1), French Bulldog (1), German Shepherd Dog (2), Siberian Husky (1), pit bull–type dog (1), Shih Tzu mix (1), and mixed (1). Five of the samples were from neutered males and 5 samples from spayed females. Mean ± SD body condition score (BCS) was 5.6 ± 1.0 (range, 4 to 7). Samples came from dogs diagnosed with round-cell tumors (n = 4), sarcomas (3), and carcinomas (3). Tumor types included large cell lymphoma (n = 3), mast cell tumor (1), osteosarcoma (1), melanoma (1), soft tissue sarcoma (1), squamous cell carcinoma (1), transitional/urothelial cell carcinoma (2).

Twenty-two healthy canine samples were used for testing. Breeds included Labrador Retriever (n = 8), mixed (7), Chihuahua (1), pit bull–type dog (1), Bernese Mountain Dog (2), Goldendoodle (1), Black Mouth Cur (1), and German Shepherd Dog mix (1). Mean ± SD age was 3.7 ± 2.0 years (range, 1 to 7). Mean ± SD BCS was 5.3 ± 1.2 (range, 4 to 9). Eight samples were neutered males, 2 samples were intact males, eleven samples were spayed females, and 1 sample was an intact female.

Blinded testing of all 6 dogs was conducted on a single day and was run by a third party individual with over 25 years of professional scent dog training experience. The tester determined the number of test trials, as well as setup and sample locations for each trial. For testing, handlers were blinded to sample types and locations used for each trial. The testing configurations and samples used are summarized (Figure 2). For testing, all samples were prepped by research assistants in a backroom and placed into 10-ounce shaker-style stainless steel canisters (2.75 x 2.75 x 4 inches). Latex free vinyl gloves were used when handling samples and canisters.

Figure 2
Figure 2

Schematic depictions of the placements (Nos. 1 through 6 or 1 through 10) of the alphanumerically labeled saliva samples from dogs with cancer (red, C-xxx) versus without cancer (blue, H-xxx) used during the 4 test trials of the MSD dog team. Sample canisters were placed in a different type of apparatus (as shown in Figure 1) for each test run: trial 1, a 6-arm carousel (A); trial 2, 10 polyvinylchloride (PVC) tubing-holders arranged in 2 rows of 5 (B); trial 3, 6 pole-holders arranged in a circle (C); and trial 4, sample canisters placed in holes cut into the top of 10 small mailer boxes (7 x 6 X x inches) arranged in 2 rows of 5.

Citation: Journal of the American Veterinary Medical Association 261, 6; 10.2460/javma.22.11.0486

Running order of the dogs for each of the 4 testing trials was randomly determined. Handler and dog teams entered the testing room 1 at a time and performed their search. During testing, the handler was blinded to the number and location of healthy samples and cancer samples for each of the 4 trials. Only the tester and videographer were in the room, while the tester recorded the responses for each dog. For Trial 1, 6 samples (2 cancer, 4 healthy) were placed in stainless steel canisters and distributed in a circular pattern on the carousel (Figure 2). For Trial 2, 10 samples (3 cancer, 7 healthy) in stainless steel canisters were placed in polyvinylchloride holders and distributed in 2 rows of 5. For Trial 3, 6 samples (2 cancer, 4 healthy) in stainless steel canisters were placed in larger holders, secured to pole, and distributed in a circular pattern. For Trial 4, 10 samples (3 cancer, 7 healthy) were placed into stainless steel canisters and then into holes cut into the tops of small mailer boxes (7 x 6 x 3 inches) distributed in 2 rows of 5. Samples were in the same positions for each dog.

Results

Testing responses were tabulated and calculated by the 3rd party tester. For Trial 1, the cancer in location 5 (Figure 2) was missed by 1 dog. For Trial 2, the cancer in locations 3 and 10 was missed by 2 dogs each. For Trial 3, no misses were made by any of the dogs. For Trial 4, the cancer in location 2 was missed by 1 dog. The following formulas were used to calculate key variables:

  • Sensitivity = [a/(a+c)] × 100

  • Specificity = [d/(b+d)] × 100

  • Positive Predictive Value = [a/(a+b)] × 100

  • Negative Predictive Value = [d/(c+d)] × 100

a = True positive; dog alerted on the cancer; b = False positive; dog alerted on a healthy; c = False negative; dog did not alert on the cancer; d = True negative; dog did not alert on healthy.

Results indicate that MSD dogs had both high sensitivity (mean 90%; range, 80% to 100%) and specificity (mean 98%; range, 91% to 100%) in distinguishing between cancer and healthy samples (Supplementary Table S1). Two cancers (true positives) were missed by a single dog and 2 other cancers were missed by 2 dogs each. Additionally, 2 dogs alerted on a total of 3 noncancerous samples (3 false positives), whereas the remaining 4 dogs did not alert on any noncancerous samples. One dog had no false positives or false negatives. The positive predictive value, or the probability that canine patients with a positive screening test truly have cancer was 95%. In addition, the negative predictive value, or the probability that canine patients with a negative screening test truly do not have cancer was also 95%.

Discussion

Current diagnostic testing available for canine neoplasia does not routinely allow for identification of cancer until more advanced stage disease.8 Studies have shown that MSD dogs can detect differences in blood, urine, and breath samples from human cancer patients.1218 Thus, there is reason to believe dogs could do the same with samples from canines diagnosed with cancer. Canine scent detection has the potential to serve as a novel, noninvasive method for early cancer detection in dogs, increasing the chance for treatment and longer lives for dogs diagnosed in the early stages of the disease. However, literature on scent detection of canine cancer up to this point has been limited to a single study, whose results did not support the ability of canines to detect urothelial carcinoma in urine.26

Our study, therefore, is a vital first step in demonstrating the ability of dogs to be accurately trained to distinguish between saliva samples from dogs diagnosed with cancer and healthy controls through scent with high accuracy. Results of this study confirmed our hypothesis that dogs can be trained to serve as a screening tool for cancer in other canines with high positive and negative predictive values.

One advantage of this study was careful selection of samples from different cancer types, along with selection of samples representing various ages and genders of dogs. Tumor types tested in this study included lymphoma, mast cell tumor, osteosarcoma, melanoma, soft tissue sarcoma, squamous cell carcinoma, and transitional/urothelial cell carcinoma. While these tumors are common in canine cancer patients seen clinically, future studies should ideally include an even more diverse population of tumor types. This would allow MSD dogs to learn how to distinguish an even wider variety of canine neoplasms, which would then benefit a wider range of patients. Tumor types outside of those tested in this study were collected and will be utilized in future studies. Throughout the study, all samples were consistently stored by refrigeration (vet clinic, 2 to 8 °C) or freezing (training facility -24 °C), leading to less room for alterations in the sample’s odor signature. While it was important for us to include a range of tumor types, one disadvantage to this is that there could be differences in concentration of VOCs within the samples from various tumor types. For example, one would assume that saliva samples from oral tumors would contain a higher concentration of cancer cells, and therefore a higher concentration of VOCs, than a saliva sample from a patient with multicentric lymphoma would.

Testing of the dogs was limited to 10 malignant cancer samples of varying cancer types, and 22 healthy canine samples. Although the results of this study provide proof of concept of the feasibility of trained dogs to detect canine neoplasia, it should ultimately be seen as a preliminary study serving as a basis for future studies with a larger number of samples. Performing larger studies will also allow us to determine if there are other differences or nuances between tumor types with scent detection.

Other limitations of our study included the lack of control over medications that patients received, and the lack of staging of healthy patients. Patients that had previously received chemotherapy or radiation therapy were excluded due to reasonable expectation that these therapies would alter the VOCs and scent patterns of samples. However, other medications were allowed and it is possible that these could alter scent patterns. Care was taken to ensure animals had not received anything by mouth within an hour prior to collection of the saliva samples. However, inherent variation in oral bacteria and presence of underlying dental disease could also affect scent patterns, which was not assessed as part of our study.28 In our study, patients were determined to be healthy based upon normal clinical history provided by the owner and overseeing veterinarian, and normal physical examination as performed by a veterinarian. Samples were also collected from younger patients, mostly under the age of 6 years, where neoplastic conditions and major systemic illnesses are overall less common. For samples collected from patients at the University of Wisconsin Veterinary Care hospital, which represented a majority of the samples, review of the medical records did not reveal that any of the control healthy patients went on to develop disease after saliva sample collection; however, the possibility of a patient having occult disease cannot be entirely excluded.

It is in our best interest to pursue further research in canine scent detection of canine cancer. As this study demonstrates, it is a noninvasive technique with great promise, and potential to serve as a future screening tool for diagnosis of cancer in dogs. Future scientific studies should involve larger trials including larger sample sizes of malignant neoplasia and healthy controls, and more trained scent detection dogs. Future studies should also evaluate scent detection in varying stages and settings of disease, such as exploring utility with microscopic disease. Other potential studies worth exploring include the feasibility of MSD dogs diagnosing other noncancerous diseases in dogs and cats such as kidney, gastrointestinal, endocrine, and liver disease. Using noncancerous diseases as controls should also be explored to further challenge the scent-detection dogs and utility of this diagnostic screening test. Further, based on success demonstrated in human cancer studies, and the successful results obtained from our study, we expect MSD dogs to have a similar proficiency in detecting differences in odor of cats diagnosed with cancer and healthy cats. Knowing this, feline neoplasia can potentially also be diagnosed using scent detection. Ultimately, the goal is to isolate the VOCs/proteins that are uniquely detected via scent to learn more about the concentration and characteristics needed, and to explore these further as specific cancer diagnostic and monitoring tests. The future of canine scent detection of small animal neoplasia is favorable, and if proven feasible would provide newfound hope for pet owners who receive a cancer diagnosis for their canine companion.

Supplementary Materials

Supplementary materials are posted online at the journal website: avmajournals.avma.org

Acknowledgments

No external founding was used in this study.

The lead author, Laurie A Malone, is President and Founder of Sprout and Penny Canine Foundation

We gratefully acknowledge the support of Sprout and Penny Canine Foundation, which provided use of their training facility and lab space. We thank ScoutMD for their support and financial assistance. From University of Wisconsin Veterinary Care we thank our clinical research technician and clinical staff for help in collection of saliva samples. We appreciate Mercy Animal Hospital and North Shelby County Animal Hospital for also collecting saliva samples. We are grateful for all the clients and patients that contributed samples to the study. A very special thanks to the dog trainer, dog handlers, training consultant, and all the other volunteers for their time and dedication.

References

  • 1.

    Cancer in pets. AVMA. Accessed October 30, 2022. https://www.avma.org/resources/pet-owners/petcare/cancer-pets

  • 2.

    Schwartz SM, Urfer SR, White M, Megquier K, Shrager S, Dog Aging Project Consortium, Ruple A. Lifetime prevalence of malignant and benign tumours in companion dogs: cross-sectional analysis of Dog Aging Project baseline survey. Vet Comp Oncol. 2022;20(4):797-804. doi:10.1111/vco.12839

    • Search Google Scholar
    • Export Citation
  • 3.

    Kraus C, Snyder-Mackler N, Promislow DEL. How size and genetic diversity shape lifespan across breeds of purebred dogs. Geroscience. 2023;45(2):627-643. doi:10.1007/s11357-022-00653-w

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4.

    Ostrander EA, Dreger DL, Evans JM. Canine cancer genomics: lessons for canine and human health. Annu Rev Anim Biosci. 2019;7(1):449-472. doi:10.1146/annurev-animal-030117-014523

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5.

    Dobson JM. Breed-predispositions to cancer in pedigree dogs. ISRN Vet Sci. 2013;2013:941275. doi:10.1155/2013/941275

  • 6.

    Pinello K, Amorim I, Pires I, et al. Vet-OncoNet: malignancy analysis of neoplasms in gogs and cats. Vet Sci. 2022;9(10):535. doi:10.3390/vetsci9100535

    • Search Google Scholar
    • Export Citation
  • 7.

    Grüntzig K, Graf R, Boo G, et al. Swiss Canine Cancer Registry 1955-2008: occurrence of the most common tumour diagnoses and influence of age, breed, body size, sex and neutering status on tumour development. J Comp Pathol. 2016;155(2-3):156-170. doi:10.1016/j.jcpa.2016.05.011

    • Search Google Scholar
    • Export Citation
  • 8.

    Flory A, Kruglyak KM, Tynan JA, et al. Clinical validation of a next-generation sequencing-based multi-cancer early detection “liquid biopsy” blood test in over 1,000 dogs using an independent testing set: The CANcer Detection in Dogs (CANDiD) study. PLoS One. 2022;17(4):e0266623. doi:10.1371/journal.pone.0266623

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9.

    AVMA. AVMA 2022 Pet Ownership and Demographics Sourcebook. AVMA; 2022.

  • 10.

    Williams H, Pembroke A. Sniffer dogs in the melanoma clinic? Lancet. 1989;1(8640):734. doi:10.1016/S0140-6736(89)92257-5

  • 11.

    Church J, Williams H. Another sniffer dog for the clinic? Lancet. 2001;358(9285):930. doi:10.1016/S0140-6736(01)06065-2

  • 12.

    McCulloch M, Jezierski T, Broffman M, Hubbard A, Turner K, Janecki T. Diagnostic accuracy of canine scent detection in early- and late-stage lung and breast cancers. Integr Cancer Ther. 2006;5(1):30-39. doi:10.1177/1534735405285096

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13.

    Lippi G, Cervellin G. Canine olfactory detection of cancer versus laboratory testing: myth or opportunity? Clin Chem Lab Med. 2012;50(3):435-439. doi:10.1515/cclm.2011.672

    • Search Google Scholar
    • Export Citation
  • 14.

    Horvath G, Andersson H, Paulsson G. Characteristic odour in the blood reveals ovarian carcinoma. BMC Cancer. 2010;10(1):643. doi:10.1186/1471-2407-10-643

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 15.

    Sonoda H, Kohnoe S, Yamazato T, et al. Colorectal cancer screening with odour material by canine scent detection. Gut. 2011;60(6):814-819. doi:10.1136/gut.2010.218305

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16.

    Ehmann R, Boedeker E, Friedrich U, et al. Canine scent detection in the diagnosis of lung cancer: revisiting a puzzling phenomenon. Eur Respir J. 2012;39(3):669-676. doi:10.1183/09031936.00051711

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17.

    Guerrero-Flores H, Apresa-García T, Garay-Villar Ó, et al. A non-invasive tool for detecting cervical cancer odor by trained scent dogs. BMC Cancer. 2017;17(1):79. doi:10.1186/s12885-016-2996-4

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18.

    Junqueira H, Quinn TA, Biringer R, et al. Accuracy of canine scent detection of non-small cell lung cancer in blood serum. J Am Osteopath Assoc. 2019;119(7):413-418. doi:10.7556/jaoa.2019.077

    • Search Google Scholar
    • Export Citation
  • 19.

    Elliker KR, Sommerville BA, Broom DM, Neal DE, Armstrong S, Williams HC. Key considerations for the experimental training and evaluation of cancer odour detection dogs: lessons learnt from a double-blind, controlled trial of prostate cancer detection. BMC Urol. 2014;14(1):22. doi:10.1186/1471-2490-14-22

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 20.

    Hackner K, Errhalt P, Mueller MR, et al. Canine scent detection for the diagnosis of lung cancer in a screening-like situation. J Breath Res. 2016;10(4):046003. doi:10.1088/1752-7155/10/4/046003

    • Search Google Scholar
    • Export Citation
  • 21.

    Amann A, Costello Bde L, Miekisch W, et al. The human volatilome: volatile organic compounds (VOCs) in exhaled breath, skin emanations, urine, feces and saliva. J Breath Res. 2014;8(3):034001. doi:10.1088/1752-7155/8/3/034001.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 22.

    Alwis KU, Blount BC, Britt AS, Patel D, Ashley DL. Simultaneous analysis of 28 urinary VOC metabolites using ultra high performance liquid chromatography coupled with electrospray ionization tandem mass spectrometry (UPLC-ESI/MSMS). Anal Chim Acta. 2012;750:152-160. doi:10.1016/j.aca.2012.04.009

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 23.

    Walker DB, Walker JC, Cavnar PJ, et al. Naturalistic quantification of canine olfactory sensitivity. Appl Anim Behav Sci. 2006;97(2-4):241-254. doi:10.1016/j.applanim.2005.07.009

    • Search Google Scholar
    • Export Citation
  • 24.

    Browne C, Stafford K, Fordham R. The use of scent-detection dogs. Ir Vet J. 2006;59(2):97-104.

  • 25.

    Rowell JL, McCarthy DO, Alvarez CE. Dog models of naturally occurring cancer. Trends Mol Med. 2011;17(7):380-388. doi:10.1016/j.molmed.2011.02.004

  • 26.

    Dorman DC, Foster ML, Fernhoff KE, Hess PR. Canine scent detection of canine cancer: a feasibility study. Vet Med (Auckl). 2017;8:69-76. doi:10.2147/VMRR.S148594

    • Search Google Scholar
    • Export Citation
  • 27.

    About us. National Association of Canine Scent Work. Accessed January 24, 2023. https://nacsw.net

  • 28.

    Roslund K, Lehto M, Pussinen P, Metsälä M. Volatile composition of the morning breath. J Breath Res. 2022;16(4):046010. doi:10.1088/1752-7163/ac8ec8

    • Search Google Scholar
    • Export Citation

Supplementary Materials

All Time Past Year Past 30 Days
Abstract Views 458 0 0
Full Text Views 4593 1852 324
PDF Downloads 3215 816 57
Advertisement