Together we will beat cancer

A dog's nose

Can our furry friends detect cancer?

[Updated below with information about a new UK clinical trial to see if dogs can aid in the detection of prostate, kidney and bladder tumours, 10/08/15 KA]

Our four-legged friends have a range of impressive skills. They can lead the blind, rescue people up mountains, and find the tiniest traces of explosives.

But can they sniff out cancer?

The idea that dogs can detect cancer has been around for a while, perpetuated on the internet (for example here) and in a BBC4 documentary first broadcast in 2006.

There are many anecdotes of family pets persistently sniffing or worrying at a certain area on their owner, which subsequently turns out to harbour a tumour.

As a result, we receive numerous enquiries asking whether Cancer Research UK is investigating this phenomenon.

But is there any scientific basis behind the idea?  And if there is, are dogs really a practical solution for cancer screening?

Picking up the scent
All smells – from the pleasant aroma of freshly cut grass to the whiff of Stinking Bishop – are due to molecules diffusing in the air, given off by whatever is causing the pong.

These so-called volatile molecules are detected by scent (olfactory) receptors in the noses of humans and animals, sending signals to the brain which then interprets the smell. Dogs have an astounding sense of smell, because their noses are packed with many times more scent receptors than humans have.

We know that some tumours produce unusual volatile molecules (such as lung cancer, breast cancer, and melanoma) which are presumably being picked up by the dogs in the stories mentioned above.  But does this mean that we should have “cancer sniffer dogs” in every GP surgery?

Putting it to the test
A small number of studies have set out to directly investigate whether dogs can detect cancer under controlled conditions, and one of the key pieces of research is a paper by Carolyn Willis and her colleagues in the UK.

Her team trained six dogs to tell the difference between urine samples from bladder cancer patients and samples from healthy people.  When the dogs were presented with new samples – six from healthy people and one from a cancer patient – the scientists found they could pick out urine samples from people with cancer 41 per cent of the time (22 out of 54 tests).  If they were correct only by chance, we would expect the dogs to pick the right sample only 14 per cent of the time (i.e. one in seven).

Michael McCulloch in the US has also looked into the phenomenon.  His team trained 5 ordinary dogs using breath samples from lung and breast cancer patients, as well as 83 samples from healthy people.  Then they tested the dogs with new samples from patients or healthy people that they hadn’t previously sniffed.

The test was double-blinded, meaning that neither the dog handlers nor the researchers watching them knew which samples were which. The results were promising – with only a few weeks training, the team found that the dogs could correctly spot the samples from cancer patients with high levels of accuracy.

But another study by Robert Gordon and his colleagues in California showed less impressive results.   In a study involving six dogs, only two performed better at detecting urine samples from patients with breast cancer than you might expect by chance.  And in a test of four dogs, only two could pick out urine samples from prostate cancer patients at a rate better than predicted by chance.

In August 2015, the charity Medical Detection Dogs announced that they are launching a large-scale clinical study in the UK, using nine dogs to test samples from 3,000 patients to check for prostate, bladder and kidney cancer. We provided this comment to the media about the study, which was used in part in some of the coverage:

Dr Kat Arney, Cancer Research UK’s science information manager, said: “Any test for cancer must be shown to be reliable, specific and practical, and large-scale clinical trials are an essential part of this process. It will be interesting to see whether this new trial shows that dogs can identify prostate cancer better than current tests, but it’s unlikely that canine cancer screening would be practical in the clinic on a wider scale. In the long term it would be useful to discover the identities of the molecules the dogs are sniffing, which could lead to more accurate lab tests to diagnose cancer.” [Updated 10/08/15 KA]

Practical problems
It’s certainly a nice idea – man’s best friend sniffing out tumours.  But although these small research studies show that dogs could have the potential to pick up cancer, their accuracy is questionable.

This is problematic because in the case of tests for cancer, it’s important to have a test that is as reliable and accurate as possible – although, of course, no test is ever 100 per cent infallible. This cuts down on the chance that cancers will be missed (false negatives), or that a diagnosis of cancer will be wrongly made (false positives).

Also, it’s simply not realistic to transfer this to a clinical setting.  For a start, there are the practical considerations.  Dogs would need to be trained, housed and fed, and would not be able to work for more than a few hours at a time for the sake of their welfare.

More importantly, it’s still not clear from the research that the dogs are picking up specific molecules originating from a tumour, or are just picking up molecules associated with more general illness. And there’s also the chance that  they may be confused by other scent molecules on the breath or in urine, such as garlic, asparagus, tobacco, alcohol or other pungent foods.

Dogs are also – obviously – unable to tell doctors exactly which molecules are faulty in an individual’s cancer – information that can be obtained using molecular diagnostic techniques.  This is important because such information is increasingly being used to tailor treatment to an individual’s cancer.

But although Fido won’t be wearing a canine labcoat any time soon, there’s a lot of research going on into detecting the scent of cancer.

Introducing the electronic nose
Although they don’t have furry coats, there are “electronic noses” that can pick up the volatile molecules produced by cancer cells with greater accuracy than a real life cold, wet one.  This kind of technology is being tested in a range of cancers, including breast and lung.

It’s still early days for electronic noses, and we need to see a lot more research to make sure the machines are accurate and reliable at detecting cancer.  But once the technology has advanced further, it’s likely that they will have wide applications for detecting many different types of cancer.  And they won’t need feeding or walking either.



C DENG (2004). Investigation of volatile biomarkers in lung cancer blood using solid-phase microextraction and capillary gas chromatography?mass spectrometry Journal of Chromatography B, 808 (2), 269-277 DOI: 10.1016/j.jchromb.2004.05.015

Michael Phillips et al (2006). Prediction of breast cancer using volatile biomarkers in the breath Breast Cancer Research and Treatment, 99 (1), 19-21 DOI: 10.1007/s10549-006-9176-1

A. D’Amico et al (2008). Identification of melanoma with a gas sensor array Skin Research and Technology, 14 (2), 226-236 DOI: 10.1111/j.1600-0846.2007.00284.x

C. M Willis (2004). Olfactory detection of human bladder cancer by dogs: proof of principle study BMJ, 329 (7468) DOI: 10.1136/bmj.329.7468.712

M. McCulloch (2006). Diagnostic Accuracy of Canine Scent Detection in Early- and Late-Stage Lung and Breast Cancers Integrative Cancer Therapies, 5 (1), 30-39 DOI: 10.1177/1534735405285096

Robert T. Gordon et al (2008). The Use of Canines in the Detection of Human Cancers The Journal of Alternative and Complementary Medicine, 14 (1), 61-67 DOI: 10.1089/acm.2006.6408

K GENDRON et al (2007). In vitro discrimination of tumor cell lines with an electronic nose Otolaryngology – Head and Neck Surgery, 137 (2), 269-273 DOI: 10.1016/j.otohns.2007.02.005

H CHAN, C LEWIS, P THOMAS (2009). Exhaled breath analysis: Novel approach for early detection of lung cancer Lung Cancer, 63 (2), 164-168 DOI: 10.1016/j.lungcan.2008.05.020


George Lavery February 15, 2012

Any thoughts from bladder cancer patients who have to have regular cystoscopies often with irritation and sometimes infection. Give me the “dog test” anytime,
Comfort,economy and less chance of infection and scarring.Having nursed for 50 and more years and now,personally,with B.C. I think I can claim some inside knowledge.
Yours etc.,

emma June 10, 2010

i think that the trials are great but in 90% of cases that were studied, the dog did not know the person! i believe dogs have have this ability to know when their owners DEVELOPE cancer rather than have it. their sense of smell is so great that they sense a change if we develope cancer. response please!

Dawn June 28, 2009

I find the subject interesting but understand there are probably practical difficulties that would need to be overcome before using dogs to detect cancer could become more widespread. It may be that they can detect some cancers more easily than others, but any dog can have an off day and cancer detection is too important to be left with something that isn’t as reliable as possible. I say ‘as possible’ as I don’t think any detection means can ever be 100%.

But it’s an interesting field of study and shows us how valuable dogs can be to humans. I’ve also read of cats that can detect cancer, though the accounts I read about were anecdotal, ie from the cats’ owners. I imagine that cats wouldn’t be trainable in the same way as dogs; if a cat doesn’t want to bother doing something all the rewards in the world won’t make it! We hear of ‘working dogs’, but who ever heard of a ‘working cat’?

Anyhow, I’m glad the dogs and their abilities are getting publicity and I hope in future they can be trained to help even further in the fight against this horrible disease.

sagar gupta March 13, 2009

i am happy to see this as every animal in this world had got some unique in them ,so whats bad in using this to make our work easier.scientist should work more on this

Kat Arney March 3, 2009

Hi Elizabeth,

I’m sorry you feel this way. But as a charity we have limited resources, and have to make tough choices about where to focus our research.

As I’ve explained in the original post and the answers to other comments, there are significant barriers to the practical use of dogs to detect cancer in the clinic.

We can only afford to invest in the most promising areas of research that will have wide practical applicability to improve survival for cancer patients. As I’ve shown, there simply isn’t enough scientific evidence to support research into the use of dogs in cancer detection.

The scientific evidence does point us in a direction though – investigating automated, electronic ways of detecting highly specific molecules associated with cancer that will produce efficient, cost-effective tools, which – once they are developed – can be used widely across cancer services. This will bring significant benefit to patients by reliably detecting cancers earlier, as well as providing vital information about the specific cancer. These are things that dogs would not be able to do.

Once again, I’m sorry you feel that you are unable to support our life-saving work, and hope that you will reconsider your decision.

There’s more about the research we fund on our website here: and examples of the progress we’ve made here:


Elizabeth Wilkinson March 3, 2009

What a negative article, I know that my mothers dog was right about her cancer and of equal importance to me (at least) my fathers prostate cancer. What would be wrong in funding research into this area as you my find there is some value.
By the way I have a hypo alert dog which I know has saved my life. Cancer research has now lost my support and I will put my hard earned pennies to a more open cause, maybe mor people should withdraw support from what appears to be am organisation not open to all possible answers.

oklahomacitymesothelioma February 6, 2009

I beleive that dogs may be able to smell it. But at what point when a dog sniffs you do you think oh no this may be cancer he is picking up on?

Kat February 4, 2009


With regard to prostate cancer, Cancer Research UK scientists have been working on a way to detect specific proteins (called MCM proteins) in urine and other fluid samples, which we hope will lead to an automated, reliable way to diagnose prostate and bladder cancer. See

There is a significant body of evidence to show that automated tests for specific proteins will lead to effective diagnostics for cancer in the future (for example, follow some of the links to PubMed, the medical publications database in the original post).

Weighed against that, there are two papers suggesting that a handful of dogs can detect cancer under certain conditions, and there’s one paper to show they don’t. And this still doesn’t take into account the significant practical and welfare issues involved in using dogs in a medical setting.

Cancer Research UK will only fund world-class science, where there is evidence to suggest that the research will lead to effective new ways to diagnose, treat and prevent cancer.

We give out funding based on grant applications by researchers – we do not “commission” research. If a UK scientist was to make an application to our science funding committee to undertake a project testing dogs for diagnosing cancer, it would be scrutinised by a panel of expert scientists, as all our grant applications are.

If the science was felt to be of high enough calibre, and fitted with our strategic objectives, then the grant would be funded. But, to the best of our knowledge, we have never received any such applications for funding.


Anne Good February 4, 2009

I would not ever doubt this but hopefully more research in this area is put forth.

Jemima Harrison February 3, 2009

The McCulloch study showed dogs were 90 per cent accurate in two different types of cancer. It might turn out that there are other explanations for the success, but I think it’s interesting enough to pursue.

Let’s take prostrate cancer. Current method of diagnosis? Not that nice. If dogs were able to detect via urine or prostatic fluid or even breath exhalate (as was used in the McCulloch study), i.e. non-invasively, that would be incredibly useful. And you would not have to rely on one dog on one day for diagnosis – it would be perfectly possible to run several dogs by an individual sample.

All I’m arguing is that it should be investigated further for the lessons that could be learned from dogs. I’m not saying this as a “dog person” blinded by their wonderfulness – I’m saying it because I think the preliminary evidence warrants it.

By fuzzy logic I mean that dogs may be able to reason/assess in a way that an e-nose may not – to detect a pattern that might be variable. There is some evidence that they can do this from the drugs/explosive work.

I think it is particularly interesting, for instance, that dogs have been taught to alert over a specific threshold (as with the currency dogs used by Customs + Excise).

Of course I can see the problems. But, equally, I still think this should have some proper money thrown at it and be explored properly. Cancer Research is in a unique position to do so, it would not take very much money and I’d wager that it would be a particularly effective PR tool for CR.

It would make a terrific follow-up film, too – a real opportunity to educate about the scientific method.


Jemima February 3, 2009

Yes, absolutely dogs and smell cancer (in particular skin cancer) this has been known about for years.

Here’s some helpful YouTube video links about this:


Kat February 3, 2009

Thanks Jemima. You’re absolutely right – dogs have been invaluable to humankind over the ages. And nobody would doubt their incredible abilities in finding bombs or drugs, guiding the blind or just generally being a good buddy. But we should be mindful of stretching that admiration too far (and I say that as a ‘dog’ person by the way).

As we pointed out in the post, the studies on dogs have been small, and although more accurate than pure chance, are still not accurate enough to be a good basis for cancer screening – not to mention all the practical problems of using dogs in a hospital setting for cancer detection. Also, you are confusing the mathematical and computing term “fuzzy logic” [see] with inaccuracy.

Just because dogs are not 100% accurate at detecting cancer samples does not mean that they are using fuzzy logic. It means that they are not accurate – and that is the most serious problem. I love dogs but I wouldn’t want to risk my own cancer diagnosis because Rover was having an ‘off’ day, fancied a bone, spotted a cat or really needed a run in the park.

As raised in the article, there is plenty of research going on around the world into the kinds of volatile chemicals that are given off by tumours, and development of machines that can use them to reliably and accurately detect cancer – the basis of any successful screening programme or diagnostic.

Cancer Research UK is investing in research into biomarkers – molecular “clues” to cancer that will enable us to do just that.


Simon K February 3, 2009

Jemima Harrison says that “There is no doubting the impressiveness of the Willis and McCulloch studies”. While there is no doubting that some dogs are able to detect something, some of the time, I don’t think that a test which only has a 41% accuracy rate (in other words, is wrong more often that it’s right) can really be described as “impressive”.

She also says that “there is no electronic nose that can replicate the dog’s fuzzy logic”. And yet the links provided in the article show that electonic noses have demonstrated accuracy levels of over 90% – way above what any dog has been shown to be capable of.

I’d love the idea of replacing mammograms with spanielgrams and laparoscopies with labradoroscopies, but sadly, I don’t think it’s going to happen.

Jemima Harrison February 3, 2009

I am disappointed to see Cancer Research still taking such a sceptical view of dogs’ potential ability to detect cancer. There is no doubting the impressiveness of the Willis and McCulloch studies. And there is as yet no electronic nose that can replicate the dog’s fuzzy logic. No one is suggesting that it is very practical to use dogs as diagnositc tools at the end of the day, although I dont really see why not – after all we train and trust dogs to make life and death decisions for us in other areas such as bomb detection. At the very LEAST Cancer Research should be funding research into working out what it is the dogs in the successful studies are detecting – and/or perhaps finding the small amount of money that would be necessary for a larger, more definitive study.

Jemima Harrison
“Can Dogs Smell Cancer?”

Abhishek Tiwari February 3, 2009

Let assume that dogs can smell it, but they can not tell it, at the end we need to rely on difference diagnostic methods. This may be interesting to find how dogs can smell cancer or what is the mechanism behind it.