Can Dogs Smell Cancer?

IRA FLATOW, host:

This is SCIENCE FRIDAY. I'm Ira Flatow. Researchers in Japan say they have trained a Labrador retriever to sniff out cancer in people. Writing in the journal Gut, the scientists say the dog was better than conventional tests in identifying people with colorectal cancer.

The dog, named Marine, sniffed the breath and the stool samples of more than 300 people and had a whopping 98 percent accuracy rate, picking out the 40 people in that group who actually had cancer.

What is going on here? Do diseases have certain smells? And how can dogs smell when something is wrong with our bodies? And could they one day be used - dogs -as diagnostic tools?

That's what we'll be talking about. Our number is 1-800-989-8255, 1-800-989-TALK. You can tweet us, @scifri - @-S-C-I-F-R-I. Gary Beauchamp is the director of the Monell Chemical Senses Center in Philadelphia, Pennsylvania. Thank you for being with us today.

Dr. GARY BEAUCHAMP (Director, Monell Chemical Senses Center): It's good to be here.

FLATOW: Now, this is not - I understand - this is not strange to you. You hear about all types of studies, you know, where dogs can sniff out all kinds of things.

Dr. BEAUCHAMP: Yes, and for hundreds of years, it's been thought that dogs and many other species, including people, can actually detect diseases based on smells coming from patients. So this idea has a very long history in medicine.

FLATOW: So it is plausible and believable, this study - there's no reason to believe this study is inaccurate?

Dr. BEAUCHAMP: No, in fact there were many anecdotes for many, many years, but it's only been recently that true, experimental studies have been conducted - like this one. And many of them are certainly showing that dogs are able to make diagnoses that are quite good.

FLATOW: Do we know what is in the odor that they're smelling?

Dr. BEAUCHAMP: Well, that's the profound question. Although many people believe that dogs can do this - or other species as well - I doubt that it will actually become a useful technique for animals to make the diagnosis. So we really want to know what the compounds are, what are the odors, so that we can develop devices or instruments that could actually make these distinctions.

FLATOW: Because there are a lot of different compounds there.

Dr. BEAUCHAMP: Probably hundreds of thousands, perhaps. For any one, particular cancer, it could be one compound changes in concentration; it could be many compounds. We really don't know what the dog is sniffing. And for most people, probably, we can't detect it ourselves.

FLATOW: But the dog certainly knows it's there.

Dr. BEAUCHAMP: Definitely, the dog knows it's there. The dog is proof of principle that there is a signal there. And it's our job, as researchers, to try to figure out what that signal is.

FLATOW: Now, the - now your center has done studies using mice instead of dogs.

Dr. BEAUCHAMP: Right. Dogs are rather large, and so we've gone to mice. And in fact, the nose of the mouse is almost as good as the nose of a dog. And it's much, much better, I think, than humans for many things.

So what we've done is, we've trained animals to discriminate other animals, and now we're working on humans with different diseases.

FLATOW: Is it possible to winnow out exactly what is - what they're smelling?

Dr. BEAUCHAMP: Well, the trick is that the sense of smell is an almost miraculous sense. It can detect very, very low concentrations of compounds. It can detect patterns of those compounds. The system is made up of a set of receptors - maybe in dogs, of over 1,000 different receptors - that detect different compounds.

So you've got an instrument here that is really beyond anything that we have in the chemical arena. And so we're not yet there, in terms of being able to pull apart the odors that the dogs are detecting.

FLATOW: Let's talk about your mice, for example. And I'm wondering: Why can't you just compare the mice with cancer to the mice without cancer, and see what's the difference, what the odor is there?

Dr. BEAUCHAMP: Well, that's a well-designed experiment, and that's exactly what we do. The trouble is that there are many, many different things. And some of them are very, very hard to detect. And it's not clear that those are the ones that the animals are actually using to detect the odors.

So it's a complex problem. It's a problem of almost too much data. And so one has to develop ways of reducing those data, and that's what the brain of the dog or the brain of the mouse does. And we're trying to figure out how we can mimic that.

FLATOW: I saw a report earlier this week - it's coming out this week - from U.C. Davis, that said: Drug-sniffing dogs could pick up on cues from their handlers and would be swayed by what the handlers think.

You know, if you take a dog and you run it through, let's say, an airport looking for drugs, and the dog comes up and sniffs some of your luggage, the report was saying there's actually an influence of the handler wanting the dog -maybe subconsciously or unconsciously - to find something, and that influences what the dog is doing.

Dr. BEAUCHAMP: That's absolutely true, and we actually did a little work with dogs and found the same thing. In our study with mice, we do a whole bunch of controls to make absolutely sure that won't happen - the most important control being that nobody knows what the answer is, that's actually working with the animals.

I know when I first began working with animals, and I was testing them and knew what I wanted them to do, and they were doing something wrong, I developed the most tremendous urge to sneeze. And it's a very powerful phenomenon, and so one has to really protect against it.

FLATOW: So we're not envisioning a time here when you might walk into your doctor's office, and the doctor's going to introduce you to a dog?

Dr. BEAUCHAMP: Well, I don't think so. In principle, it would work, and of course we use dogs all the time for drug sniffing and things of that sort. On the other hand, my guess is that for doing diagnosis of disease, which would lead to some sorts of maybe serious kinds of surgery, they would be reluctant to do it.

And I should also say that I think the disease diagnosis by smell is never going to be the only diagnostic tool. It's going to be one that potentially is very, very cost-effective, very non-invasive. But I think it would be in conjunction with other diagnostic tools, to get an overall view of the likelihood of a particular disease.

FLATOW: Let's go to Joey(ph) in Sacramento. Hi, Joey.

JOEY (Caller): Hi.

FLATOW: Hi there.

JOEY: I was just calling in to say that I have a MedicAlert dog. I'm diabetic. And he sniffs my breath, and he actually caught a low-blood-sugar event that had no warning signs, and got me to sit down and be safe before it got bad or there were any symptoms.

And had he not caught it, it would have hit me, and I would have been helpless. But he sniffs my breath all the time.

FLATOW: I was going to say, do you just purposely breathe on the dog to - alert you, or is this something that, you know, the dog just likes to lick your face or something and discovers it?

JOEY: Well, I know one other person who has a MedicAlert dog who's diabetic, and her dog licks her skin. My dog will sniff my breath.

FLATOW: So the dog is trained to do this, you're saying?

Dr. BEAUCHAMP: No - well, a lot of MedicAlert dogs figure it out on their own. Mine learned it from another dog who sniffed our breath, and would let us know if we were getting cold. He started sniffing my breath as a puppy, and I became diabetic. And it's like the lights went on, and he went, oh.

And he'll sniff my breath regularly to check me, but he doesn't show any interest in it unless my blood sugar gets low enough.

FLATOW: Gary Beauchamp?

JOEY: Pardon?

FLATOW: I'm getting a reaction from Gary.

Dr. BEAUCHAMP: This is one of many such stories I've heard, and I believe them to be absolutely true. One of the most famous ones, I think, in our field is the description of a woman who had a dog who all of a sudden started sniffing avidly at one of her breasts, and she couldn't stop him. And so she finally went into the doctor, and they discovered a breast cancer on that breast.

So I think these things, the sensitivity of the nose - in particular, the dog's nose or the mouse's nose - to changes are - it's just phenomenal. The trick is to try to figure out what it is that those animals are cuing in on.

FLATOW: Do you think that there might be an aspect of this story where people want to believe that their dog is helping them or looking out for them?

Dr. BEAUCHAMP: Yeah, and that's part of the issue of the other citation you talked about, that there is a strong desire for this to be the case, and particularly for dogs. I mean, dogs have this special relationship with people, and the idea that the dog would do this sort of signals that the dog has particular love for that individual.

So you can see the strong desire for that. Nevertheless, I believe that there's something to these stories.

FLATOW: And how does the dog communicate that it's found something?

Dr. BEAUCHAMP: Well, I think it varies, and what she was saying is that it pays no attention to the breath until there's some change. In the example I was saying, it sniffed much more frequently, and pushed and poked. I mean, I think there's many ways the dog could signal that.

FLATOW: Let's go to Todd(ph) in Boulder. Hi, Todd.

TODD (Caller): Hi there. I just have a comment that, you know, this is not at all surprising to me. I'm a nurse, and I've worked on many oncology floors and actually, in GI clinics where, you know, finding cancer is pretty common.

And it just seems completely logical to me because in - at times where we come across people like, where you're taking care of somebody that has advanced cancer, the smell is very strong. You know, you can smell it in the room, and you can smell it on the patient. So it seems just simply logical to me that a dog who has, you know, just a higher sense of smell would pick this out at a much earlier stage.

FLATOW: All right, thanks, Todd.

TODD: Sure.

FLATOW: He - obviously people, as you say, have been talking about this for a while. But how do we take it from the dog-smelling stage into more of a scientific, reproducible stage?

Dr. BEAUCHAMP: Well, I think the first step is what the paper that's just come out, and several other papers, similar, are doing, and that is to put it into an experimental situation, where you're really testing the hypothesis.

Then the next step is doing also exactly what you said, which is to try to take samples - we use urine; you could use breath, what have you - from patients and controls, and making sure you have the appropriate control people, and try to identify the compounds and develop ways of discovering what the odorants are, and then using and developing devices that can be routinely used to detect them.

FLATOW: But cancer is just so many different kinds of cancer. There's not just one kind of cancer.

Dr. BEAUCHAMP: Right.

FLATOW: Wouldn't the odor vary by cancer to cancer, and person to person?

Dr. BEAUCHAMP: Well, so that's a very, very important point. And it's a point that the previous caller, I think, alluded to as well. And that is when people are very sick with cancer, there may be a general odor that signals illness, that is not specific.

What one wants for something like this is a very specific odor, and you'd really like to have one that differentiated different kinds of cancer. And I don't think we know the answer to that yet. I think that's also a major area for investigation.

Does, say, lung cancer produce one kind of odor, and colorectal cancer a different - or do the different classes of cancers produce different odorants? That's what you'd want. You'd want something that was very specific. And we don't know, really, how specific these odors are.

FLATOW: Thank you very much, Dr. Beauchamp, for being a guest.

Dr. BEAUCHAMP: My pleasure.

FLATOW: Gary Beauchamp is director of the Monell Chemical Senses Center in Philadelphia. We're going to take a break and when we come back, we're going to talk about when politics first met science. It's all about science and politics. So stay with us. We'll be right back after this break.

Copyright © 2011 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information.

NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.