Dogs are often used to sniff out everything from illegal drugs to explosives. But new research shows that they can also smell cancer.
Canine cancer detectors. I'm Bob Hirshon and this is Science Update.
You’ve heard of bomb-sniffing dogs, but it turns out dogs can sniff cancer too. Researchers at the Pine Street Foundation in California recently trained five different dogs to smell breast and lung cancer on a patient’s breath. The dogs’ diagnoses were 88 to 99 percent accurate, even when odiferous factors like smoking were taken into account.
So what was tipping the dogs off? Research director Michael McCullogh says figuring that out is the next step. That could pave the way for new technologies that could screen for cancer with a simple breath test.
The fact that it was dogs is almost beside the point. Although I should add that the dogs performed so well that now technology really has to rise to the challenge that they laid down.
I'm Bob Hirshon, for AAAS, the science society.
Making Sense of the Research
When you first hear about this study, the idea of dogs smelling cancer probably seems like the main point—and the most interesting one. But, as the story says, the goal of this study isn’t to put a German shepherd in every doctor’s office. The dogs were simply used to prove a concept: that cancer leaves some kind of distinct chemical signature on the breath.
In fact, other scientists are already developing machines that can diagnose other diseases by sampling a patient’s breath: diseases like pneumonia, stomach ulcers, and even the same kinds of cancer studied here. That’s because certain diseases cause the body to produce chemicals that we normally don’t produce, or in different amounts than a healthy person would produce. Some of these chemicals find their way into the air we exhale, which means they can be smelled—if not by a human, then by an animal or a machine.
So, what’s the point of the dogs? Well, for one thing, scientists don’t yet know exactly how having cancer might change your breath, which means they don’t know what to look for. So it’s tough to build a machine—an “electronic nose”—that’s specifically tailored to look for cancer. There are electronic noses out there that just look for overall differences in odors, rather than specific chemicals, but so far, they haven’t matched the accuracy of the dogs. (One device, tested recently at the Cleveland Clinic, was 66 percent accurate in diagnosing cancer.)
Since dogs have excellent noses and are easily trained to recognize just about any odor, the dogs are like odor-detecting machines that the scientists don’t have to build. In this experiment, the researchers exposed the dogs to breath samples from known cancer patients and breath samples from healthy patients, and simply rewarded them for picking out the cancer breath. Once the dogs got the hang of it, the researchers conducted a “double-blind” trial, where the origin of the breath samples was concealed not just from the dogs, but also from the researchers themselves. With the dogs achieving up to 99 percent accuracy in the double-blind trials, the experiment shows that it should be possible to build a breath-diagnosing machine that gets it right just about every time.
The challenge is to figure out what chemicals to look for. That’s tougher than it sounds, because the smell of cancer is probably made up of many different odor-causing chemicals. (This is true of most smells in the real world as well.) What’s more, each type of cancer may have a unique chemical signature. It will take many more experiments—involving humans, dogs, and machines—to find just the right formula. But if it works, it could make screening for deadly diseases a lot cheaper and easier.
Now try and answer these questions:
- Why was it possible for the dogs to smell cancer on a patient’s breath?
- What are the differences between experimenting with dogs and experimenting with machines? Give advantages and disadvantages for each.
- What is a double-blind trial? Why was it important in this experiment? In science in general?
- Suppose the dogs appeared to distinguish cancer breath from normal breath when they were being trained, but then they failed the double-blind trial. What might that suggest about the way the dogs were trained?
You may want to check out the February 10, 2006 Science Update Podcast to hear this Science Update and the other programs for that week. This podcast's topics include: Cancer-sniffing dogs; the origin of the moon; how saturated fats kill; a new neighbor galaxy; comedy is healthy.
The National Geographic News Article Dogs Smell Cancer discusses this research. Their article Robo-Nose features a high-tech electronic nose that sniffs out bombs.
The American Association for Artificial Intelligence’s Artificial Noses page has links to many other articles about electronic noses.