There’s been a lot of news lately about cancer-sniffing dogs after a new study by Japanese researcher Hideto Sonoda and his colleagues was published in the medical journal Gut. So we couldn’t help but wonder, is that possible?
If you haven’t heard about it, the recent study suggests that specially-trained dogs can identify the scent of volatile chemicals (those that evaporate into the air at room temperature) present in colon cancer. [more]
In the study, a Labrador sniffed breath samples and stool samples from patients who were known to have colon cancer, which had been identified by colonoscopy and biopsy. The dog correctly recognized cancer-related scents in 91% of the breath samples and 97% of the stool samples. And amazingly, when the dog was allowed to sniff breath and stool samples from patients who didn’t have colon cancer, the dog incorrectly “diagnosed” only 1% as malignant. This study adds to a small number of other published articles showing similar results for bladder, lung, and breast cancers, and to a recent conference presentation regarding prostate cancer.
In addition to these experimental studies, there have been several anecdotal reports of patients whose pet dogs seemed attracted to or upset by skin and breast cancers.
Although the idea of dogs recognizing cancer might initially seem hard for some people to believe, it does seem biologically plausible. We know that trained dogs can distinguish the scent of one person from another person, and that some dogs have great ability to follow the trail a person has taken. Dogs can also be trained to recognize very low concentrations of explosives or illegal drugs in the air. Scientists have already identified some of the chemical differences between normal and malignant tissues, so it is not surprising that some dogs can also recognize these differences.
The researchers say their next step is using these trained dogs to help them discover the specific chemicals associated with colon cancer, or with other types of cancers. Once that is done, scientists will try to develop laboratory tests that can be used to find these chemicals, and therefore the cancer.
Some readers might wonder whether it is time to use dogs in the clinic for cancer screening. Much more research will be needed before we can seriously think about dogs taking on roles in finding cancer similar to their current ones in law enforcement. In this study, each round of testing included a few benign samples and one known to have come from a person with cancer, and the dog was rewarded shortly after a correct “diagnosis.” In a real clinical screening situation, there might be several samples that contain cancer, or none at all – and it could take weeks or months to know for sure. This is because when a trained dog performs correctly, you want to reinforce that behavior promptly. Reinforcement doesn’t need to happen every time, but not knowing which samples are malignant would complicate the reinforcement. The trainers would have to run some test rounds with known samples occasionally to reinforce the behavior.
Also, in the protocol used in this study, the dogs were trained to recognize the single malignant case in every round. If real testing included some rounds that had no malignant samples or that had several malignant samples, that would probably make things more difficult for the dogs (and the humans). That doesn’t mean it couldn’t work, but as far as I know dogs have not yet been tested in a situation that reflects what they would encounter in actual clinical practice, and the dog’s accuracy under real clinical conditions might not be as good as it was under ideal experimental conditions.
One more limitation is that the dog in this study did not reliably recognize any abnormal scent from precancerous polyps (although it was not specifically trained to do so). In contrast, some colon cancer tests such as colonoscopy can accurately recognize these polyps so they can be removed by the doctor to prevent colon cancer before it is fully developed. Could a dog be trained to sniff out polyps? This important question would need to be tested.
Perhaps one day, dogs will be able to find many different types of cancer early enough and consistently enough to earn a role in a real-life situation. Or, researchers may be able to build an “artificial nose” that recognizes the same cancer-related scents that dogs do. In the meantime, continuing to get cancer screening tests recommended by ACS guidelines is still the best way we have to catch cancer before it spreads.
Dr. Gansler is director of medical content for the American Cancer Society and editor of CA: A Cancer Journal for Clinicians.