Right now, as we speak, there exists a relatively inexpensive method for detecting certain early
cancers with an accuracy rate of up to ninety-seven percent. You might even have one in your room
with you right now home. And it probably goes wild when you reach for a tennis ball or the doorbell
Yes — we’re talking about the domestic dog. So why aren’t there domestic dogs trained to detect
cancer in every hospital on the planet? It’s complicated. Anecdotally, there are a lot of stories out
there about dogs who seemed to smell lung cancer on their owner’s breath or became unusually
interested in a particular mole later found to be cancerous. But there’s also more rigorous evidence
too. A 2019 study found that some dogs can detect lung cancer in blood samples with almost 97%
accuracy. And that’s not too surprising when you consider that dogs can learn a huge amount about
other dog's health, age, and reproductive status by smelling their butts. But how they’re able to
separate the smell of healthy cells from cancerous cells is pretty remarkable.
Cancer cells produce odiferous chemicals called volatile organic compounds, or VOCs. And dogs
sometimes actually detect particular cancer’s signature blend. The particular group of VOCs
produced by a given type of cancer can act as a biomarker for that cancer. A biomarker is any
biological characteristic that can be measured and then used to identify some sort of process in the
body. If that sounds vague, it’s because biomarkers can be a huge range of things. But in this case,
they make it possible to tell healthy tissue from cancerous cells. VOCs are produced during normal
biological processes. They might be present in skin, urine, blood, sweat, stool, or even a person’s
breath, where they can serve as biomarkers for both healthy and diseased biological processes.
That’s because the VOC profile you see in these places is different in people with, say, lung cancer
then it is for healthy people. And it’s different in people with lung cancer than it is in people with
ovarian cancer, or bladder cancer, or breast cancer. Detecting cancer-specific VOCs is complicated for
us humans because there are so many to sort through — there can be as many as 3000 different
VOCs in a person’s healthy exhaled breath. In 2003, researchers were able to identify two classes of
VOCs — alkanes and monomethylated alkanes — as possible tumor markers in lung cancer patients.
The specific biomarkers of many other cancers haven’t been identified yet. That’s mostly because it’s
challenging to pick out the few VOCs that are specific to a given cancer from all the other given
compounds that might be present in a sample. But if that two 2019 study is any indication, some of
our fine furry friends can be trained to sniff them out.
So you’d think every hospital and doctor’s office and diagnostic lab that doesn’t already have a
cancer-sniffing dog would be trying to get one. But there are some technical barriers, despite these
pooches’ impressive accuracy. Dogs thrive in jobs that require a close relationship with a handler.
Dogs that sniff out drugs or bombs, or who look for survivors of disasters, work in a very engaging
environment. These jobs are exciting for dogs.
The animals are working in an environment with a lot of external stimuli, and that moment where a
survivor is found under a pile of rubble can be, like, emotional for everyone involved. And a search
and rescue dog’s handlers can confirm that the dog has found its target, which means the dog will
always get the right reinforcement at the right time. Compare that to a bunch of dogs in a lab sniffing
out sample after sample, most of which probably won’t be positive. That could be frustrating for a
dog. But the biggest problem in that real-life scenario is that the handlers can’t give the dogs any
feedback if they don’t know which samples are cancerous. If they reward the dog for every alert,
they could unknowingly be rewarding false positives, which will ultimately impact the dog’s accuracy.
No rewarding means no joy, and those cancer-sniffing canines could easily start regretting their
career choice. And even though the dogs can be pretty accurate in theory, things could be different
in practice… Like people, dogs are fallible. I don’t want to… don’t send me to hate mail! And unlike a
hypothetical cancer-sniffing machine, a dog might work better in the morning and not so great in the
afternoon, which would result in different rates of accuracy. You don't have that kind of problem
with a machine, which operates the same way every time you use it.
Finally, it’s a huge investment of both time and money to train a cancer-sniffing dog. A single dog
needs to spend six to eight months sniffing a minimum of 300 biological samples before they can be
a certified cancer-sniffer. But that doesn’t mean we can’t learn a lot from cancer-sniffing canines.
Dogs can still help us identify the odor signatures of different cancers. We could use that information
to build electronic noses — a robot cancer-sniffing dog. Like, not literally, but it was fun to say… These
devices are already being tested, but more work needs to be done to determine which VOCs the
noses should be looking for, as well as to improve their overall accuracy. So far, some studies are
using electronic noses to look at patterns of VOCs, rather than picking out specific ones, because the
specific VOCs aren’t always known. And electronic noses still aren’t as accurate as some dogs are.
Other medical conditions, for example, can sometimes interfere with the machine’s ability to spot
cancer. One study on using VOCs to detect bladder cancer found that the presence of infection could
reduce the accuracy of the electronic nose. Which means there’s more progress to be made. Still,
while it’s fun to imagine legions of lab coat-wearing canines quietly saving lives behind the scenes,
it’s probably not going to happen in quite that way. But the things we’ve learned from dogs and their
amazing olfactory abilities are already helping us build technologies that will save lives through early