Paging Dr. Canine

During the last 15 years, cancer researchers have been experimenting to see if dogs can sniff out the unique chemicals emitted by various cancers. The general consensus is that dogs can be diagnosticians, but whether they will be a regular part of health screenings remains to be seen.

Dog identified cancer in owner's leg

The dog that started it all was Trudi, a Dalmatian with a keen nose who, nearly 30 years ago, found a mole on the leg of her 19-year-old owner Gill Lacey. The dog made such a fuss that Lacey saw a doctor and discovered that the mole was a malignant melanoma, a deadly form of cancer.

Trudi's story, and others like it, came to the attention of scientists who saw the sensitive noses of man's best friend as a way for researchers to identify cancer.

Mixed results

One of the earliest studies was conducted at Oxford University in England, where the dogs attained a 41 percent success rate in identifying cancers in urine (pure chance would be 14 percent) proving the study's hypothesis, that cancer has a unique odor. But there was a bonus. The six cancer-sniffing dogs all detected cancer in the urine of a control - a man thought to be cancer-free. He was tested and found to have a malignant kidney tumor. The dogs saved his life.

"I am confident that it can work," said Robert Kratzke, a lung cancer researcher at the University of Minnesota. "We've been able to detect compounds in the breath of lung cancer patients. When this was first discussed, we thought we should be able to use gas chromographs, the advantage being that you don't have to feed, pet or take a gas chromograph for a walk. But the advantage of a dog over a machine is that it's not just one chemical that's indicative of cancer, it's 12 or more. A dog's brain could sort out a mix of odors and integrate them into one identifiable smell."

In the United States, funding for this research is hard to secure. Dr. Lawrence Meyers, a researcher at the Department of Anatomy, Physiology and Pharmacy at Auburn University in Alabama, conducted one study where dogs were trained to detect mammary and prostate cancer in urine samples. Because his study was unfunded, Meyers reached out to the community to find volunteer dogs and handlers.

"We used a large number of subjects, gave them a protocol to follow and used a lot of controls," he said. "After training the dogs, we tested them. They were 80 percent accurate. But in a double-blind test, where no one near the dogs knew which samples were from the cancer patients, the result was statistically no better than chance."

The problem, he says, is that dogs have an inborn ability to follow subtle human social cues, so it's quite possible that while dogs accurately located cancers, they may have also been sensing their handlers' unconscious physical responses to correct answers.

It's all in the training

Duane Pickles, the dog trainer for a Florida State University study, agrees that the problem is not the science but in the training. Pickles has 31 years experience training police dogs. In 1993, Florida State researchers asked him to train dogs to sniff out the chemical markers for three types of skin cancer and five types of lung cancers.

One dog that Pickles trained was George, a former bomb-sniffing dog. George and the other dogs in the study were given the job of identifying samples of cancerous tissue taped under bandages on their handler's body, as well as samples hidden in one of the ten holes drilled into a plastic tube. George's success rate on both tests was greater than 99 percent.

"You can't just use someone's pet," Pickles said. "You have to have dogs that are certified utility dogs (the highest AKC obedience rating). You need a professional handler with several [obedience trial] wins. The dogs need to know 150 commands just to start and have to know 400 commands by the end of training."

"A dog can be a real-time chemical detection system," Meyers said. "We're talking about health screenings or the use of dogs in certain third-world situations. Cancer screening by dogs is non-invasive and can find cancers very early, before symptoms appear. But, there are still questions on whether you can do this reliably.

"Right now," said Meyers, "we're simply trying to find ways to improve the early detection of cancers. There's still an enormous amount of work to be done."

Stephanie Fox is a freelance writer in Minneapolis.