A labrador retriever was trained to detect cancer in patients just by smelling their breath.
Maybe you’ve heard that line about dogs being able to tell whether or not someone has cancer just by smelling their breath. And maybe you think it’s baloney, kind of like eating chocolate will kill a dog, or cats can tell when a person is going to die?
Wrong. Wrong. Um, I have no idea.
It may surprise you to know that the ability of dogs to smell cancer is well documented. A 1989 report marked the first medical record of what had been known anecdotally in communities around the world for hundreds of years. Subsequent studies have quantified just how skilled dogs are at detecting different cancers including lung, breast, ovarian, and bladder cancers, and skin melanoma. Incredibly, they can do this just by smelling the patient’s breath.
The top three killer cancers worldwide are lung, stomach, and liver cancers. Number four is colorectal cancer (1.2 million new cases in 2008). A recent study published in the journal Gut, adds colorectal cancer (CRC) to the list of cancers that our canine friends can sniff out. Kyushu University’s Yoshihiko Maehara and colleagues began by training a Labrador retriever (named Marine) in scent detection of cancer by exposing it to the breaths of healthy people without cancer and patients with confirmed colorectal cancer. They also trained Marine with watery stool samples to compare her performance to the fecal occult test, a common CRC screen. Thirty-three patients and 132 healthy controls were tested. For each trial Marine sniffed five samples, four from healthy controls and one from a patient. She was taught to lie down on the floor upon positive identification of a patient. Marine’s accuracy was astounding. Of the patients conventionally diagnosed by colonoscopy she correctly identified 91 percent based on the smell of their breaths—making the distinction between cancerous and benign polyps—and she was correct in excluding 99 percent of the healthy samples. When she was presented with the water stool samples her accuracy increased to 97 percent. Compare that to a 70 percent accuracy of fecal occult tests. What’s more, Marine was better at detecting early-stage cancers.
I told you it was astounding.
The potential here is obvious. Every dog park in the world is populated by highly-accurate, non-invasive cancer screens. Patients with familial adenomatous polyposos or Lynch syndrome—different forms of CRC—are subject to frequent colonoscopy to minimize the morbidity and mortality associated with CRC. Imagine if they had only to blow into a bag.
So why not bring Marine and her friends into the clinic?
“Come right in Mr. Jones. Put on this robe and have a seat on the table next to Fido.”
In fact, some people are pushing to get dogs into the clinic. But there’s resistance to that, and it’s understandable. The oft cited argument against this was in fact mentioned in the current study: “It may be difficult to introduce canine scent judgement into clinical practice owing to the expense and time required for the dog trainer and for dog education.”
I don’t buy that. Recent research shows that among the causes of death in the world cancer has the most devastating impact on economy. I admit I haven’t crunched the numbers to determine just how much a dog would cost a clinic—but I’m still not buying it.
I suspect it’s something else. I suspect it comes down to confidence in Fido, despite what the data says they can do. Usually when a person is diagnosed with cancer he’s looking at a mass in an MRI scan or results from a biopsy. This is hard, tangible evidence with a good track record. How many of us would immediately commit to the frightful regimen of chemotherapy and surgery that follows a cancer diagnosis because a dog lay down on the floor? What if the dog was too good at early detection and follow-up verification showed negative? What do you do then? What would you do?
The much more accepted approach is to bring dogs not to the clinic, but to the lab. The goal here is to identify the compounds in a person’s breath that signals to the dog cancer. But that’s a really, really tough task.
We all know that a dog’s sense of smell is far better than that of a human. A dog can detect odor molecules at concentrations as low as one part per trillion while the limits of human detection top out at about one part in several billion. Obviously there’s something unique about the odor of cancers otherwise the dogs wouldn’t be able to distinguish diseased breath from normal breath. Numbered among the normal odorants in the patient’s breath is (are) molecule(s) produced solely by the tumor, and the fact that they’re detectable in the breath means they’re systemic.
A cell becomes cancerous when the replication machinery malfunctions and the cell proliferates out of control, resulting in a tumor. This process involves gene and/or protein changes that may result in peroxidation of lipids that make up the cellular membranes. These peroxidized lipids then generate what are called volatile organic compounds (VOCs). Dogs, as well as other animals, are currently used in research to screen for tumor-specific VOCs in the hopes of identifying them. If we can identify them, then we could build chemical sensors to detect them. But it’s estimated that a single breath contains hundreds of thousands of compounds. To identify them all and rule out those not produced by the tumor is a daunting task to say the least.
But maybe we don’t have to.
Dr. Hossam Haick of Israel Institute of Technology invented an "electronic nose" able to detect cancer in the breaths of patients. He was named one of the "Ten Most Promising Young Israeli Scientists" in 2010.
Gene arrays that compare gene expression patterns between healthy and diseased cells often show differences in tens or hundreds of different genes. From a diagnostic standpoint, that’s enough. If, instead of trying to identify the VOCs generated by tumors, what if we simply compared VOC concentrations in healthy individuals with those of cancer patients. If the VOC profiles between the two groups are sufficiently different it would make for a very effective screen.
Hossam Haick at the Russell Berrie Nanotechnology Institute, Israel Institute of Technology, has built such a screen. Described in a study last August, Dr. Haick’s group developed a kind of “electronic nose” that can “smell” VOCs. The nose is actually a nanosensor array of organically functionalized gold particles that can detect trace amounts of VOCs. Single, exhaled breaths of patients with lung, colon, breast, and prostate cancer were analyzed with the array and compared with the breaths of people without cancer. The results showed a distinct difference in the VOC profile of cancer patients compared to the cancer-free group. The differences were then confirmed with gas chromatography linked to mass spectrometry, technology commonly used to quantify chemical concentrations.
This is an incredible example of technology imitating biology and I’m extremely eager to see how the electronic nose is put to use. If we do find ourselves one day soon, blowing into a bag as a normal part of our checkup, we should remember to thank our canine friends when we get home. Were it not for dogs like Marine pointing the way, researchers like Haick might not have known to look in the first place. And while the electronic nose looks promising, it is as yet still unproven in the clinic. That leaves us with only our dogs to sniff out cancer. And thus it is unfortunate that their use in cancer detection is not more widespread as a tool that could at least assist, if not replace current diagnostic tools.
[image credit: Peter Wadsworth via WikiCommons]
[image credit: Russell Berrie Nanotechnology Institute at the Technion -- Israel Institute of Technology]
Image 1: http://commons.wikimedia.org/wiki/File:Labrador_Retriever_black_portrait_Flickr.jpg
Image 2: http://rbni.technion.ac.il/.upload/hossam35.JPG
Comments
On the Japanese dog study: I reviewed the actual paper (my day job is developing products and getting FDA approval, but I love dogs and they are excellent in lots of tasks).
1. The study was poorly designed and does not show the dogs “smelled” cancer. There were very large differences between the two test groups (specimens from patients with cancer and from patients either normal or “cured”). These included prevelance of smoking, use of certain drugs and other conditions, all of which are even more likely to provide detectable chemical (by assay and, I expect canine) traces in stool/saliva. The differences were more than significat, some were at the level equivlent to the dog detection (they do not give the individual patient data so it is hard to tell the true magnitude of the effect.
That said, while it is physiologically plausible that dogs can detect when perosn are affected by serious disease, like advanced cancers that affect metabolism in ways that create odor signatures, or are are highly exposed (eg skin and maybe upper oral and respitory tract). It is equally unlikely that they can detect early cancers. The problem with detecting cancers is that it is your own cells with surprizingly few genetic and exogenetic changes. The tiny signals is lost among the noise of the body untill there is so much cancer it can be picked up.
Several points…
Has anybody performed tests on how quickly a dogs nose fatigues after smelling clinical samples over time? Is there cancer detection sensitivty and specificty going to change over the weeks and years of working in a lab?
Another study published in the journal Animal Cognition demonstrated that drug sniffing dogs false positive rate was heavily influenced by their handlers expectations. Would medical dogs require special double blinds to prevent such confounding factors?
Creating artificial noses has largely failed over the past decade. Congress recently took DARPA to task for spending vast sums of money over the past decade in the pursuit of an e-nose. DARPA’s results have so far been significantly less impressive than good ol mother nature.
One possible reason for this could very well be that the proteins used for sensing chemicals in olfaction don’t work as we expected. Recent research (Molecular vibration-sensing component in Drosophila melanogaster olfaction) indicates that instead of using the lock-and-key model olfactory proteins may actually gain the specificity by registering the specific vibration of a molecules electron bonds (a separate feature from the molecules shape).
Thank you for your comment, Freud (any relation?).
You’re absolutely right that there’s a confound with the animal handlers’ expectations, and they would need to be dealt with. Dr. Gary Beauchamp mentioned this on Science Friday a couple weeks ago. He’s doing similar studies with mice, and he’s noticed that even they, and rats, and other animals will react according to the experimenter’s expectations.
As I’m sure you know, the vibration-sensing theory of olfaction has quite an interesting and controversial history. The vast body of literature supports the structural, “lock-and-key” model, and I think it’d be a hard sell to encourage researchers to shift emphasis to a vibrational model.
No relation but I have a beard
You are right that most researchers are going to have a hard time accepting a vibration only model of smell. I misstated what the vibration hypothesis entails, rather than replacing the lock-and-key model Franco et al. believe that both mechanisms are necessary to explain the data.
How can a dog smell colon cancer in stool, much less in a person’s breath? Is he smelling the actual DNA mutation? I don’t think that’s possible. How can a DNA or RNA mutation from a colon polyp make its way all the way up to the respiratory tract and be exhaled?
Does anyone have a theory for this?
honestly, i am not surprised that dogs can do that. another question is why do we need this ability? does it help? how? why? can they detect it on early staged when person still can have a treatment and a kind of a hope?
and this question in the topic – why aren’t dogs in the clinic… stupid question.. clinic must be as clean as it is possible. and any dog ill never help in this no matter how often you will bathe it and how good you would look after it
Theobromine is toxic to dogs as they metabolise it so slowly.