Dogs have been using their superior sense of smell to help humans for centuries—leading hunters to their quarry, locating missing persons, detecting explosives and narcotics, even detecting falling glucose levels in their owners. But new research from the UW School of Veterinary Medicine suggests that their exquisite olfactory systems might be able to help their own species as well. Mackenzie Pellin DVM ’11, oncologist and clinical associate professor, published a study this Spring in the Journal of the American Veterinary Medical Association (JAVMA) suggesting that dogs can be trained to detect cancer biomarkers in other dogs’ saliva. It’s the first study of its kind to demonstrate that dogs can accurately distinguish between samples taken from healthy dogs and samples taken from dogs with known malignant tumors.
Pellin says the research is particularly important because cancer has long been one of the leading causes of death in dogs, in part because veterinarians don’t have a non-invasive screening tool that could detect cancer in dogs sooner. “Dogs can’t tell us when something feels different in their bodies, so we usually don’t know something is really wrong until they start exhibiting more obvious and sometimes distressing clinical signs,” Pellin says. “This study brings us one small step closer to having a reliable non-invasive screening tool that could help us choose more effective treatments earlier and possibly provide a cure.”
Research that examines dogs’ ability to detect cancer in other dogs is relatively new, but case reports of dogs detecting cancer in humans via scent detection were first documented in 1989. Since then, multiple studies have shown that dogs are capable of being trained to detect some cancers in humans by perceiving specific “odor signatures” in samples of urine, sweat, breath, and blood serum. Because most cancers seen in dogs are almost identical to cancers that affect people, Pellin and other researchers hypothesized that dogs would have similar abilities to detect cancer in dogs as well.
To test this theory, Pellin and colleagues from the University of Alabama collected 139 saliva samples from dogs with malignant tumors and 161 samples from healthy dogs and sent them to a canine training facility in Alabama. Using positive reinforcement methods, trainers exposed a team of six scent detection dogs of various breeds to a select set of samples and then rewarded them with food treats when they accurately identified known cancer samples. During the second round of training, these same dogs were exposed to the samples without reward to determine if they could still correctly identify the samples that came from dogs with cancer.
For the final testing, researchers carefully selected 10 samples from canine patients with a variety of tumor types, ages, and genders to best challenge the dogs. Twenty-two samples from healthy dogs were used as controls. Each dog participated in four different testing trials, with varying number of healthy and cancer samples. During the testing, the dog handlers were blinded to the number and location of healthy and cancer samples.
The results of the test show the scent detection dogs had both a high sensitivity rate and high specificity rate in distinguishing between cancer samples and the healthy samples – two important statistical metrics of reliability when detecting disease. High scores on each of these metrics is not that common and indicate a very high level of accuracy. In total, two cancers (true positives) were missed by a single dog and two other cancers were missed by two dogs each. Additionally, 2 dogs alerted on a total of 3 non-cancerous samples (false positives), while the other four dogs did not alert on any non-cancerous samples. One dog scored perfectly during all four trials.
Although the results of the study provide proof of concept that trained dogs can detect canine cancer with high levels of accuracy, Pellin said this preliminary study should serve as a foundation for future studies that use larger number of samples and more trained scent detection dogs. “Performing larger studies will help us determine if there are other differences or nuances between tumor types with scent detection, but it will also give us the opportunity to study if dogs can distinguish between healthy and non-cancerous diseases, and between cancer and non-cancerous diseases.” said Pellin. She also hopes to repeat all these studies with samples from feline patients.
Other members of the research team include Laurie Malone, researcher at the University of Alabama at Birmingham, School of Public Health; and Kendal Valentine, clinical research assistant at the UW School of Veterinary Medicine.
– Gian Galassi
This response was featured in the Winter 2023-24 issue of On Call.