Sophie Lutter takes a look at the research behind the recent headlines on dogs detecting prostate cancer with their noses, and asks if we’re on the scent of something promising, or barking up the wrong tree.
Humans have been teaming up with dogs to make use of their heightened sense of smell for centuries, and now they’re invaluable members of our police and military forces.
More recently there have been stories trickling through about the use of dogs in detecting various medical conditions. The most recent headlines concentrate on prostate cancer, and the abilities of two trained explosive-sniffing German shepherds to detect prostate cancer from urine samples.
A team of researchers based in the Humanitas Clinical and Research Center in Italy carried out the study, and published it in the peer-reviewed Journal of Urology. Four people worked full time to train the dogs, and then recorded their ability to tell the difference between 362 urine samples from men with prostate cancer, and 540 urine samples from a control group of both men and women without prostate cancer.
The dogs sniffed out the prostate cancer samples, and got it right, 100 per cent and 98.7 per cent of the time, and ignored the non-prostate cancer samples properly 98.6 per cent and 97.6 per cent of the time. Both of which are obviously very good scores.
So we know that some dogs at least can smell something in a urine sample that means prostate cancer is present. This is interesting.
The trouble with stories like this is sometimes a problem with the way the research was done, but not in this case – it was a good study, run well and with interesting results. Having said that, we do always like to read the small print and, as the NHS choices report on this story pointed out, the dogs were tested with some of the same urine samples that they were trained with. It would have been better if they were trained and tested with different samples, because otherwise there’s a risk that the dogs are just responding to their training, not to a whiff of prostate cancer.
So, if there isn’t a problem with the way the research was done, why aren’t we jumping up and down with excitement about this story? Well, that’s because there are still some big questions to answer – some of which the researchers point out themselves. For us, the biggest question we need an answer to is how this will eventually benefit men with prostate cancer.
Let’s face it. It’s highly unlikely that it will ever be feasible to include sniffer dog tests as a routine part of prostate cancer diagnosis for all men in the UK. Apart from anything else, the logistics of it are mind-boggling. How many dogs would you need? Where would they live? Who would look after them? What happens if they get a doggy cold, or other illness that affects their nose? How many staff would you need to employ to look after and train them? If four people needed to work full time just to train the two dogs, this isn’t going to be cheap.
But that doesn’t mean that there’s nothing to follow up on.
We’d still like to know whether the dogs can diagnose prostate cancer. We know that they can be trained to recognise prostate cancer in samples where the researchers knew it was present. But can they accurately find prostate cancer in the urine of men who haven’t been diagnosed? The researchers plan to test this next, which will give a better sense of whether whatever the dogs can smell could be useful for diagnosis.
The problem is that to really prove the quality of a diagnostic test, you need to be able to show not only that it does detect prostate cancer when it’s there, but also that that it doesn’t detect prostate cancer when it isn’t. And this needs to be not just in young men, women and older men with a low PSA as in this study, but also in potentially ‘suspicious’ samples from men of the same age as the prostate cancer group. This would include men with a raised PSA and possibly other prostate problems, but who don’t have prostate cancer. The difficulty is that, as the researchers point out, you just can’t go around doing biopsies on men at low risk of prostate cancer (the healthy controls). It’s not ethical, and who would volunteer for that?
Another big question is, even if the dogs can diagnose prostate cancer, will it reduce the number of biopsies? This research suggests that the dogs responded in the same way to advanced prostate cancer that had spread outside the prostate as they did to low-risk prostate cancer in men on active surveillance. Is this useful? If men still need a biopsy to confirm the results of the sniff-test and see how aggressive the cancer is, there’s still exactly the same risk of men with very low-risk prostate cancer undergoing unnecessary and potentially harmful biopsies, and of the biopsy missing an important cancer as there is now.
And finally, probably the most important question is, what is it that the dogs can smell? Perhaps if we could answer this question, we could design a laboratory test to reproduce these results. We could tell whether there’s any possibility of it being useful to distinguish between aggressive and non-aggressive prostate cancer – which, let’s not forget, is what we’re really in need of – and whether it could be done on a large enough scale to be of use to all men with or at risk of prostate cancer.
So overall our feelings on these results are a little bit ‘so what?’ Some dogs can smell prostate cancer in urine samples. But that just isn’t the same thing as a reliable, accurate diagnostic test that can be used as part of routine care to diagnose men with aggressive prostate cancer, and avoid the need for a biopsy in men with non-aggressive or no disease. That’s what we’d like to see.
Director of Research, Dr Iain Frame sums it up in his comment: “The results of this research are only interesting if we investigate further through rigorous research exactly what the dogs are detecting. Whilst quirky and appealing to some, the idea of dogs being used routinely to detect prostate cancers is unrealistic and could do more harm than good based on the evidence we have now. Prostate cancer diagnosis continues to be one of the major unsolved problems facing the disease with over diagnosis and over treatment of non-aggressive cancers remaining a big issue. These results do little to alleviate these concerns at the moment.
“We all want to see men get a better deal at the point of diagnosis. That’s why we’re committed to bringing together the top research minds in this field to find a feasible solution as soon as possible. Using dogs to detect a scent in urine is unlikely to be the answer.”
Dr Frame will soon be meeting with Dr Claire Guest, CEO of the charity Medical Detection Dogs, to talk about how we can help them get robust data from properly controlled trials and examine what might be being detected by the dogs.