Can Dogs Detect Cancer?
+Nick Kai Nielsen tagged me about a recent article in The Guardian about a story where a woman’s assistance dog detected her breast cancer. The story is remarkable.
“One evening in November 2011, I was at my computer when Mia leapt on to my lap and nuzzled into the flesh at the top of my left breast. She closed her eyes and licked furiously. That frightened me because it’s what she does when I have a bruise or cut. I had an ultrasound within a week and, sure enough, there was a lump that a biopsy later confirmed was grade 2a breast cancer”
Cancer detection through dogs is not unheard of. What is the theory behind this approach? We know that dogs have extremely sensitive noses, and it has been reported that bloodhounds have upto 4 billion olfactory receptor cells, compared to just 5 million in a human. What this means is that our canine friends can discern the presence of a molecule, even when it is greatly diluted. We also know that cancer cells have abnormal metabolic pathways to generate energy, and it is possible that tumours can give off specific volatile chemicals that could be detected by dogs with extremely sensitive noses. But, we do not yet know what these molecules are, much less if and how they are produced by cancers. Clearly, this is a topic that needs to be researched.
In the UK there is a charity called Medical Detection Dogs, and a quick search through Pubmed indicates that there are several studies to evaluate the efficacy of this novel detection method. However, they are a bit hit-and-miss, and some of the studies seem poorly designed. I’ll summarise a couple below:
In 2013, a retrospective study was carried out where the dogs were trained to detect ovarian cancer in blood samples taken from patients. They compared patients that had clinical remission after undergoing chemotherapy with healthy controls. Generally, we do not know how many patients will have residual cancer cells after complete clinical remission is declared. We also do not know if the dogs were detecting residual chemotherapy drugs in the patient blood samples, or actual cancer cells. I am struck by how the study design did not include another control where the patients were diagnosed with cancer but had not undergone chemotherapy yet, to ensure that the dogs actually detected cancer as opposed to chemo drugs.
Another study looking at prostate cancer used dogs that learned to positively identify prostate cancer samples from controls. But when the dogs were tested under double-blind conditions (where neither the dog, nor the handlers, nor the scientists know which samples belong to the control group vs the test group), they did not identify prostate cancer samples more frequently than expected by chance. The study warns that “it is possible that the dogs may memorise the individual odours of large numbers of training samples rather than generalise on a common odour”.
As tempting as it is to look at stories such as this Guardian article with optimism, it is critical to approach them with a degree of realism. These are anecdotal stories. Compelling yes, but still anecdotal.
There are so many questions we still need to answer:
- Are there molecules produced by cancer cells that can be detected by dogs?
- How are they produced by the cancer cells?
- Can dogs actually smell them?
- Can the dogs be trained to indicate that they smell them?
- How reliable are the dogs at smelling them?
Until this approach has been rigorously tested, we cannot abandon traditional cancer screening and detection programs in favour of cancer sniffing dogs.