One study, by researchers from Kyushu University in Fukuoka, Japan, involved 48 patients with confirmed bowel cancer and 258 controls. It showed that a Labrador specially trained in scent detection was able to distinguish between cancerous and non-cancerous samples in 33 out of 36 breath tests and in 37 out of 38 stool tests-a success rate of more than 90 per cent.
Compared with conventional diagno-sis by colonoscopy-an invasive proce-dure involving a camera inserted via the rectum-the dog's sensitivity (number of people with cancer correctly identified as such) was 91-per-cent accurate for breath samples and 97-per-cent for stools. The dog's specificity (number of people without cancer correctly identi-fied as such) was 99 per cent for both the stool and breath samples. For early-stage cancer, the accuracy of the dog's disease-detecting ability was even higher (Gut, 2011 Jan 31; Epub ahead of print).
Similar results were found in another study, looking at a Belgian (Malinois) Shepherd's ability to detect prostate cancer in urine samples from 33 patients with cancer and 33 without. After about a year of training, the dog was able to correctly identify the cancer samples in 30 out of 33 cases, equating to a sensitivity and specificity of 91 per cent.
However, of the three cases the dog wrongly classified as cancer, one patient was re-biopsied and subsequently diagnosed with the cancer, suggesting that the dog's powers of detection were even better than those of the standard biopsy test (Eur Urol, 2011; 59: 197-201).
The two studies indicate that cancer has a specific scent that can be detected by trained dogs in a person's breath, urine or faeces. Indeed, there are a number of earlier studies that support this theory. A 2006 study reported that, with only a few weeks of training, ordinary household dogs were able to accurately distinguish the breath samples of lung and breast cancer patients from those of controls (Integr Cancer Ther, 2006; 5: 30-9). Another study concluded that dogs can be trained to distinguish patients with bladder cancer on the basis of urine odour (BMJ, 2004; 329: 712).
The smell of a person's blood or skin can also signal the presence of cancer, it appears. A Swedish study found that two Giant Schnauzers could accurately identify people with ovarian cancer by sniffing blood samples with a sensitivity of 100 per cent and specificity of 98 per cent (BMC Cancer, 2010; 10: 643), while a US study reported that dogs can sniff out human skin cancer (Appl Anim Behav Sci, 2004; 89: 107-16).
These intriguing findings could lead to improved methods of cancer detec-tion, including tools for very early cancer detection. Early diagnosis is critical for the successful treatment of cancer, but many of the currently available tests are ineffective at picking up the disease in its earliest stages. The faecal occult blood test, for example-a non-invasive method of screening for bowel cancer-is only able to identify early-stage disease in one in 10 cases. Moreover, some tests, such as the PSA (prostate-specific antigen) test for prostate cancer, have high rates of false-positives (indicating cancer when there is none), which can lead to unnecessary treatment.
While it's unlikely that canine cancer-detection will become the norm in clinical practice, scientists are currently working on the idea of 'electronic noses' that can sniff out cancer in a similar way to dogs. Whether these machines will be as good as the real thing, however, remains to be seen.
Dogs can also sense other types of illness, evidence suggests. There are numerous anecdotal reports of dogs predicting seizures in their epileptic owners, and even a few scientific studies exploring the phenomenon.
In a study carried out by the charity Support Dogs, Val Strong and colleagues successfully trained several 'seizure-alert dogs'. "As training progressed, the dogs were able to provide overt signals to their owners within time periods varying from 15 to 45 minutes prior to a seizure occurring," the authors said. Each dog had an accurate prediction time and, in every case, the owner's seizure frequency was reduced (Seizure, 1999; 8: 62-5).
This apparent ability of the dogs to actually reduce seizures was investigated in another study by Support Dogs of 10 people with tonic-clonic seizures. The patients were monitored for a total of 48 weeks, both before and after being provided with a seizure-alert dog. On comparing their baseline seizure fre-quencies with the last 12 weeks of follow-up, the researchers reported a mean reduction of 43 per cent. Nine out of
10 participants showed a 34-per-cent or greater reduction and four out of 10 showed a 50-per-cent or greater reduction (Seizure, 2002; 11: 402-5).
Another study, by University of Florida researchers, reviewed all the evidence on seizure-alert dogs and interviewed 63 people with epilepsy, 29 of whom had pet dogs. They concluded that "some dogs have innate ability to alert and/or respond to seizures". They noted that the success of the dogs depended largely on the owner's/handler's awareness and response to the dog's alerting behaviours (Seizure, 2003; 12: 115-20).
Other evidence suggests that dogs have an inbuilt ability to detect hypo-glycaemia (low blood sugar) in diabetic owners. One study from Belfast, North-ern Ireland, and published in the Journal of Alternative and Complementary Medicine revealed that "behavioural reactions to hypoglycaemic episodes in pet owners with type 1 diabetes commonly occur in untrained dogs".
The researchers analyzed data from 212 dog owners with type 1 diabetes and reported that 65 per cent of them said their dog had shown a behavioural reac-tion to at least one of their hypoglycaemic episodes, with 32 per cent of the dogs reacting to 11 or more events. In all, 36 per cent of owners believed their dogs reacted mostly when they went "low", and 34 per cent said that their pets reacted before they themselves were aware that they were hypoglycaemic.
Although dog reactions varied, most behaved in a way that was suggestive of getting their owners' attention-for example, barking, licking or jumping on them. Interestingly, the gender, age and breed of the dogs appeared to have no bearing on their hypoglycaemia-response ability, and neither did the duration of pet ownership (J Altern Complement Med, 2008; 14: 1235-41).
Precisely how dogs are able to predict seizures or low blood sugar in their owners is a mystery but, as the cancer-detection studies suggest, it could be that they are reacting to a certain smell. Indeed, according to Dr John Kirkwood of the University of Pittsburgh Cancer Institute, "it might well be that [dogs] have such high acuity of smell that they can smell things like the immune system can sense antigens. Their sense of olfaction may be in a sense akin to the immunological ability to recognize different things on the basis of molecular constituents."
Alternatively, it could be that the dogs are responding to minute changes in their owner's behaviour or physicality. However, with reports of dogs being out of sight of their handlers, and then suddenly approaching them and alerting them (Seizure, 2003; 12: 115-20), this explanation seems to be far too simple.
A furry future?
Ultimately, more studies are needed on dogs' disease-detecting abilities, as this novel area of research could lead to future advancements in medicine and technology. Discovering how dogs are able to 'smell' a seizure or cancer cells, for instance, could set the stage for the development of new technologies to detect odorous molecules in a similar way.
In the meantime, specially trained medical response dogs are already helping hundreds of people with life-threatening chronic conditions, providing them with a crucial early-warning system (see the websites at www.medicaldetectiondogs.org.uk and www.support-dogs.org.uk).
As for all the other dog owners out there, if your companion starts acting strangely around you-sniffing or licking you excessively, for example-it might be a good idea to have yourself checked out by a physician.
WDDTY VOL. 22 NO. 2