Partner Series
If Dogs Can Smell Cancer, Why Don't They Screen People?
Fido has a powerful nose.
Credit: Shutterstock

Dogs can be trained to be cancer-sniffing wizards, using their sensitive noses to detect cancerous fumes wafting from diseased cells. This sniffing is noninvasive and could help diagnose countless people, which begs the question: If these pups are so olfactorily astute, why aren't they screening people for cancer right now?

Here's the short answer: Dogs do well in engaging situations, such as helping law enforcement track scents or guiding search-and-rescue teams in disaster areas. But sniffing thousands of samples in which only a handful may be cancerous is challenging work with little positive reinforcement.

Moreover, it takes time and energy to train these pups, who, despite extensive preparation, still might miss a diagnosis if they're having a bad day, experts told Live Science. [20 Weird Dog and Cat Behaviors Explained by Science]

But that's not to say that dogs can't be helpful in the development of manmade screening tools that "smell" cancer. It's known that cancerous cells emit unique odors, but scientists have yet to identify the specific compounds responsible for these scents.

One way dogs might be able to help pinpoint cancer-specific odors is to give the dogs certain cancerous samples to sniff, and then slowly remove compounds from the sample. If the dog stops responding to the sample after several components are removed, "then you know you've taken out that component of the mixture that is specific to the cancer," said Dr. Hilary Brodie, a professor in the Department of Otolaryngology at the University of California, Davis. Researchers could then analyze these individual components and develop biochemical tests that could reliably screen patients, he said.

"There's lots that the dogs can do, but I don't think wholesale screening of the population is where it's heading," Brodie told Live Science.

In 1989, the British journal The Lancet published the first dog-sniffing-out-cancer report. In a letter to the editor, two dermatologists described how a dog reportedly spent several minutes each day sniffing a colored lesion on its owner's thigh, and even tried to bite off the spot when she wore shorts. Concerned, the woman had doctors inspect the lesion, which turned out to be a malignant melanoma.

"This dog may have saved her owner's life by prompting her to seek treatment when the lesion was still at a thin and curable stage," the doctors wrote in the letter.

Other reports of dogs detecting malignant melanomas followed, but it wasn't until 2006 that high-quality, double-blinded studies were published, said Dr. Klaus Hackner, a pulmonary physician at Krems University Hospital, in Austria. (In the double-blinded studies, neither the dogs nor their handlers knew which samples were cancerous.)

Soon, there were countless studies showing that trained dogs could detect specific cancers by sniffing biological samples, such as a person's breath or urine. That's because cells, even cancerous ones, give off volatile organic compounds (VOCs). Each type of cancer likely has a distinct VOC, meaning it has a different odor compared with other cells, Hackner said.

Given that dogs have more than 220 million smell receptors in their noses, they're excellent animals for sniffing out disease, Hackner said. In comparison, humans have a "mere" 5 million smell receptors in their noses, he said. [Why Do Dogs Have Whiskers?]

Most dogs can be trained to recognize the odor of a specific cancer in about 6 months, Hackner said. However, many studies had setups that work in laboratories, but not the real world: often, the dog would be given five samples that always had one cancerous specimen. In reality, depending on the type of cancer, a sniffer dog might find just four cancerous specimens out of a batch of 1,000, he said.

If neither the dog nor the handler knows which four out of those 1,000 samples are cancerous, the handler can't give the dog positive reinforcement when the dog picks the right specimen, Hackner said.

"I think this was one main point for why our study failed," said Hackner, whose 2016 work, which had a real-world-like setup, was published in the Journal of Breath Research. "We were not able to provide positive feedback because neither one knew in the screening situation if the dog was right or not. This was stressful for both the dogs and the handlers."

This situation could be remedied if there was always a planted cancerous sample in each set, so the dog could get a reward and wouldn't be bored after sniffing thousands of noncancerous samples from patients, he said.

But even if the setup could be changed to accommodate the dogs, it wouldn't be a realistic way to screen patients, Brodie said. It would take an immense amount of resources to train dogs to recognize the many types of cancer that can affect humans. In addition, while no test is perfect, at least doctors know how accurate different tests, such as mammograms, are, and at what rate they produce false positives and false negatives. But these rates would vary for each dog, Brodie said.

Moreover, dogs can get bored, hungry and "have bad days, just like you and I," Brodie said. "You'd have to be carefully monitoring their effectiveness throughout their cycles."

Rather, Brodie and Hackner envision dogs helping researchers create and refine biochemical "nose" machines, known as e-noses, that could "sniff" patients and deliver diagnoses, they said. These machines already exist for certain medical conditions, but could be made more sensitive and applicable to more diseases with the help of dogs, Brodie said. But the research isn't there yet, he noted.

In one project, Brodie and his colleagues were studying whether dogs could detect volatile organic compounds from head and neck cancer patients by smelling the breath patients had exhaled into a container. But the researchers put the project on hold after the dog trainer began broadcasting that her dogs could sniff out cancer.

"We didn't want to be affiliated with that," Brodie said. "We wanted to prove that they're detecting it, not state that they're detecting it and then prove it. You've got to do the science first. This is not even close to or near prime time."

Original article on Live Science.