Synopsis: Diseases apparently have distinctive odors that humans can’t detect. Researchers are using dogs, mice, rats and other animals to literally sniff out cancer and other diseases in the laboratory. In the 3rd World, rats are used to diagnose TB. Experts discuss the use of animals to diagnose disease and their efforts to build machines that can do the same thing.
Host: Reed Pence. Guests: Dr. Cindy Otto, Executive Director, Penn Vet Working Dog Center, University of Pennsylvania; Dr. Gary Beauchamp, Emeritus Director and President, Monell Chemical Senses Center; Dr. Christiaan Mulder, Director, TB program, Apopo
Links for more information:
16-17 Animal Disease Detection
Reed Pence: Early detection of disease almost always makes treatment easier and more effective, and doctors have come up with increasingly high tech tests and scans to find disease inside the body. But one new method of early detection is decidedly low tech — the use of dogs and other animals to sniff out disease.
Dr. Cindy Otto: One of the earliest reports was in 1989, when a physician listened to his patient, who said, “My dog keeps biting at this mole. Can you check it out?” And it turned out it was a melanoma. So that was the first time that it was published. And people were pretty skeptical.
Pence: That’s Dr. Cindy Otto, executive director of the Penn Vet Working Dog Center at the University of Pennsylvania.
Otto: Same sort of thing in the diabetic response, that there are a couple case reports where physicians heard this story of the dogs of these client acting differently when their blood sugar either went very very low or very high. So people started to hear this and think, huh, there might be something there. So there was an investigation and people started to look at it in a more scientific manner.
Pence: It should really be no surprise that animals can sense things that we can’t. Dr. Gary Beauchamp is emeritus director and president of the Monell Chemical Senses Center.
Dr. Gary Beauchamp: There’s a whole world of communication odors that’s really outside of our realm of the senses, or at least for most people’s realm of the senses. It’s as if they didn’t exist. Certain birds can detect wavelengths of light, insects as well, that we don’t detect. So there’s this whole area that we can’t really enter into unless we use approaches such as training the animals, having them signal the presence of the odor. So that’s one of the real challenges going forward in the future, and that is to try and understand how these animals are doing it, because eventually what we would like to do, obviously, is develop non-living techniques to do this — to detect these signals so that one would have a very very quick, easy, reproducible and hopefully non-invasive way of doing disease diagnosis that wouldn’t require bringing a dog or a ferret or a mouse into the hospital or into somebody’s house to do it.
Pence: But scientists have a long way to go before they can create a machine that can sniff odors as well as a dog.
Otto: Dogs can smell odors that are as tiny as parts per billion, which is a really hard concept. It’s like a teaspoon of sugar in a swimming pool. They can smell that small of an amount. Machines can’t even get that close and people are probably a thousand times to ten thousand times less sensitive than the dogs.
Pence: What’s more, Otto says dog’s brains are wired to make the most of that sensitivity.
Otto: I like to compare it to us finding “Where’s Waldo?” The dogs do the same sort of concept, because in vision we can block out all the things that are extraneous; we can hone in on that important thing that we’re looking for and dogs do the exact same thing. So all the colors and shapes and textures in our world become the odors that a dog uses for their landscape and can pick out just a very subtle difference, or subtle specific odor amongst thousands of other odors.
Beauchamp: Most animals I think use a sense of smell to communicate much like we use a sense of vision to communicate. They are able to get information about other members of their species of the same sort that we can get from vision — something about the individual identity of a person, their sex, their mood, their emotions, and whether they are sick or ill. An animal can do that with smell and they probably do many more things as well. So basically what the animal is doing is doing what it naturally does with smell. We’re trying to tap into that by using animals and then eventually probably with machines to identify what the messages are and to use those in some sort of diagnostic way to diagnose human disease.
Pence: Over the last couple of decades, dogs and other animals have been trained in research projects to detect prostate cancer in urine, lung cancer in breath samples, and much more. They’ve been able to predict seizures in epilepsy patients and find colorectal cancer, skin and breast cancer. Scientists are even investigating whether animals can detect undiagnosed Parkinson’s disease. Who knows what else?
Otto: We don’t know what the spectrum is. I would imagine that it’s probably pretty broad because anything that has a unique odor signature, it’s very possible to train a dog to recognize that, and then that becomes an opportunity to have them detect that. The question becomes what’s the practical application? For diabetics it’s a wonderful application because we pair a dog and a person and that dog can tell that person their blood sugar is dropping. And that often happens before the continuous glucose monitors will alarm. So, really powerful, really changes the lives of some of these people.
Pence: In research trials, dogs have been 98 to 99% accurate in detecting lung and prostate cancer. Otto says her dogs are about 90% accurate in detecting ovarian cancer.
Otto: We’re working with ovarian cancer, and our dogs are able to identify plasma samples from patients that have ovarian cancer. Our ultimate goal is to work with our collaborators who are building electronic sensors and our chemists who are trying to identify what this odor is so that we can have a very readily available inexpensive mass screening tool for hundreds of thousands of women that should be screened every year for this horrible disease.
Pence: But that’s all still in the laboratory. Out in the field, one of the few routine uses is in Tanzania and Mozambique, where giant pouched rats sniff out tuberculosis.
Dr. Christiaan Mulder: These studies showed that the rats they can find about 70% of the patients and they can rule out about 80% of the alphi individuals or known TB patients. So their sensitivity is about 70% and the specificity is about 80%.
Pence: Dr. Christiaan Mulder is head of the TB program for the Belgian-based organization, APOPO.
Mulder: Rats, they are really fast. They can screen 140 samples in less than 20 minutes. If you compare this with what a lab technician can do with a microscope it would take about 2-4 days. And also if you compare to gene experts, which is one of the latest molecular assays endorsed by the WHO organization, the rats are much faster, because for the gene expert it takes about two hours to process four samples. Your screening of individuals goes much faster and is much cheaper.
Pence: But if animals are so good at sniffing out diseases, why are scientists working so hard at creating a machine to do it that might never be as good? Otto says trusting animals to do the job can be a stretch.
Otto: I think that there are people who love dogs and get it that dogs have this capacity, but there are a lot of skeptics, and so it’s hard to do this research because all of the big agencies are like, “What are you doing? You’re doing this with dogs? You’re kidding.” So, in order to move it forward, whether it is moving it forward with dogs doing this and being able to be validated, or moving it forward and trying to get a piece of equipment that can come close to replicating the dogs, we need people to understand that this is viable. It does have huge potential and we need to find funding agencies that are willing to think that way.
Beauchamp: When you bring this to discussion with people who are in the business of doing these diagnostics they are very skeptical. There seems to be an inherent bias against using animals to make these diagnostic decisions. It’s interesting why that should be when obviously we use animals to make all sorts of other very very important decisions. For example, following individuals, looking at drugs, looking at explosives, things of that sort. But most of those are simpler in a sense. They’re signals that come from a single compound or a few compounds, whereas these odors of diseases, we don’t really know what they are. We don’t know what the animal’s actually sniffing. Because of that lack of knowledge I think there’s this concern that even though in some studies it shows it’s 98% accurate, that under normal conditions in testing in a hospital you might never reach that.
Pence: Otto and Beauchamp say that may be true. Reliability varies from one animal to another… and one day to the next.
Otto: We can’t say they do it all the time. And the reason that we can’t say that is because, they’re, I guess we can’t say they are human, but they’re like us in that sometimes we have bad days and sometimes we have good days, and sometimes we’re not as focused as other times. So we’re seeing that our dogs are accurate about 90% of the time.
Beauchamp: I can absolutely swear that I could train a mouse or a ferret or probably a dog, I don’t work with dogs, to make these discriminations at much better than chance level. Nevertheless, much greater than chance and 100% accuracy are not the same. If you have a dog or a pet you know that sometimes they behave very well and other times they don’t and it’s very difficult to control that sort of thing. There‘s a subjective problem with it. I think that people are nervous about whether the animal is sufficiently consistent over long periods of time to really depend upon that.
Pence: One other advantage to a machine is that it can run 24/7, and Otto says dogs can’t.
Otto: These dogs have to work really hard to find this tiny tiny odor and figure out which is which. So they can work for maybe 10 or 20 minutes, and then they need a break. So it’s not like we can screen as many samples. I think we can screen a large number of samples this way, but it’s still pretty expensive, plus we have all the training that goes into it, the maintenance of it, so it becomes economically fairly difficult. And I think it is certainly a possibility, but it’s not the most likely possibility.
Pence: Plus, in order to be accepted as a diagnostic test for cancer or other diseases, Otto says the use of animals would have to be FDA-approved. Among other hurdles, that would require trials that guarantee that the animal is doing it on his own, and not responding to the trainer.
Otto: We want to make sure that we’re not having some inadvertent influence on the dogs because they are so responsive to everything we do – body language. That is one of the biggest challenges that we face in trying to do the research — eliminating the human factor so that the dogs are really clearly telling us that this is the odor that they are detecting.
Pence: But what is the odor they’re detecting? It would help with credibility of the concept if we could say, yes, this is what they’re smelling. But do we know?
Otto: No, we don’t. [Laughter] We have some clues from our chemists of patterns, and that’s I think probably the most important thing to start thinking about. We used to think that odor must be an odor, a single molecule that they are smelling. I think what we’re seeing is it’s a pattern. Some odors change and go up and some odors change and go down. So the dogs can put together that pattern and recognize what that pattern is.
Pence: Otto says there’s value in skepticism because that makes scientists work harder to be perfect. But whether animals do the sniffing or machines eventually do it, Otto says what’s more important is the concept that we can diagnose cancer through its odor. Once we accept that, it has enormous potential.
Otto: I think it’s almost limitless because it opens up a whole new realm of how we think about disease and diagnosis. And, ovarian cancer – that it such a devastating disease, and it’s a silent killer and most women aren’t diagnosed until they are at stage three and four. And by then it has spread and their chances of survival are really low. But if we could pick it up at stage one, surgery can cure them. It’s a life changing possibility, and so that’s why we are so committed to this.
Pence: You can find out much more about all our guests through links on our website, radiohealthjournal.net, where you’ll also find archives of our shows. You can also find them on iTunes and Stitcher.
I’m Reed Pence.