Medical Examiner

Roll Over! Shake! Smell This Mole!

Are dogs really a good way to screen for cancer?

The doctor will sniff you now.

Photo illustration by Slate. Photo by Shutterstock.

Dogs can reliably detect prostate cancer by sniffing a man’s urine, according to a widely publicized study released this month. I’ll take a dog over a digital rectal exam any day. Plus, dogs are our most faithful friends—who better to save you from cancer than your best pal in the whole world? Is this the beginning of a productive partnership between oncologists and dogs?

Don’t hold your breath. Anecdotes about cancer-sniffing dogs have been floating around the medical world for a quarter of a century, and the first clinical trials came out about 10 years ago. But there are good reasons the idea hasn’t caught on, and the most recent study doesn’t change that. Most of the cancer dog studies are methodologically weak. The logistics of bringing dogs into medical practice are a nightmare. Most importantly, machines are easier to train and more predictable over the long term than dogs. 

The following passage appeared in a 1989 letter to the medical journal The Lancet:

The patient first became aware of the lesion because her dog (a cross between a border collie and a doberman) would constantly sniff at it. The dog (a bitch) showed no interest in the other moles on the patient’s body but frequently spent several minutes a day sniffing intently at the lesion, even through the patient’s trousers. As a consequence the patient became increasingly suspicious. This ritual continued for several months and culminated in the dog trying to bite off the lesion when the patient wore shorts.

It seemed like the story of one miraculously astute canine that saved its owner from melanoma. When a second cancer-detecting dog was reported in the same journal in 2001, however, scientists took notice. In the past 10 years, researchers have produced nine trials on cancer-detecting dogs. Canines have been tested for their ability to sniff out bladder and prostate cancer in urine, melanoma and breast cancer on skin, ovarian cancer from tissue samples, and lung cancer on a patient’s breath. Some studies claim dogs can detect cancer with more than 95 percent accuracy.

Most of the studies, however, are overblown. “I’ve been working in this community a long, long time, and people often make outrageous claims,” says Lawrence Myers of Auburn University, who has been studying the sensory function and behavior of detector dogs for 32 years. Myers says most of the cancer-sniffing dog studies—especially those showing more than 95 percent accuracy—are inadequately blinded. If the handler knows which samples come from cancer patients, the dog will react to the human’s cues, stopping in front of the sample from the cancer patient when the handler hesitates.

“I’ve watched video [from some studies] frame by frame,” Myers says, “and the handlers stopped before the dogs. It’s a fraction of a second, but the dogs pick up on that.”

Many studies also don’t have enough samples. A small number of training runs followed by a small number of tests is not enough to prove that a dog can reliably detect cancer over the course of its lifetime. Myers suggests the researchers need thousands of samples, which can be difficult to obtain. Another problem with translating these studies into a clinically relevant procedure is the stage of cancer. In the recent prostate cancer study, for example, the researchers included urine samples from patients with advanced disease. Detecting advanced prostate cancer is easy enough already.

A lack of compelling evidence isn’t the only problem. Doctors aren’t particularly interested in working with dogs. “Cool” was the word that kept coming up in my conversations with oncologists. Cancer-sniffing dogs are cool like stomp rockets or didgeridoos—interesting, but not to be taken seriously. They face a long list of obstacles.

The first and most obvious issue is what exactly the dogs are smelling. “For all other purposes sniffer dogs have been used for so far, you have the pure substance for training them: dynamite, drugs, or even the smell of a person,” says Enole Boedeker, a doctor in Germany who has published her own cancer-dog studies. “We don’t know the origin of [the cancer] smell.”

If we don’t know what chemical substances dogs are detecting in urine, what should we do when one or more dogs inevitably begins to underperform? Do we assume the problem is isolated to a single dog? Do our training methods need to improve? Should we retrain and retest the dogs more frequently?

Oncologists also point out that, for some cancers, we don’t need more screening tests. Reliable cancer tests are available for some of the conditions dogs are supposedly able to detect. If anything, we’re already hurting some patients by detecting and treating harmless cancers.

“We don’t have a problem detecting prostate cancer—we have a problem matching the level of treatment to the level of risk,” says Charles Ryan, a professor of medicine and urology at the University of California–San Francisco. “In that sense, the cancer-sniffing dog doesn’t represent a significant new technology.”

The legal obstacles may also be insurmountable. The FDA regulates medical devices. Is the FDA really going to oversee the training, housing, care, and disposal of thousands of dogs? If a radiologist misses an obviously cancerous breast mass, the patient can sue him or her for medical malpractice. What will happen when a cancer dog misses a diagnosis, or misdiagnoses a harmless sample as cancerous? Can you sue the dog’s trainer? Who will provide medical malpractice insurance to dog trainers?

Cancer kills more than 550,000 Americans annually, and early detection could potentially prevent as many as one-third of those deaths. Cancer-sniffing dog studies may be an important step in improving our save rate, but not because we will one day rely on bloodhounds to diagnose disease. Instead, these dog studies suggest that tumors leach volatile organic compounds in detectable concentrations, and that these emissions differ from those of noncancerous cells. The next step isn’t training up an army of cancer dogs. But perhaps we need a small core of well-trained animals to help us figure out what, exactly, the chemical signals are. Machines—man’s other best friend—can take it from there.