In 1989, Williams and Pembroke described in The Lancet the case of a woman who had become suspicious of a lesion she had on her thigh because her dog spent several minutes a day sniffing intently at it and even attempted to bite it off. An histological examination revealed the lesion was indeed a malignant melanoma. Twelve years later, a similar story was reported by Church and Williams (2001).
These two studies raised the interest of the scientific community and paved the way to more systematic investigations examining dogs’ olfactory ability to detect various forms of cancer.
In early 2011, the media largely covered a new study published in the journal GUT, in which Marine, an eight-year-old Labrador retriever, had successfully detected bowel cancer from breath and stool samples (Sonoda et al., 2011). Five samples were randomly placed into five boxes: one obtained from patients with colorectal cancer (CRC) and four obtained from people without any cancer used as controls. Marine first smelled a standard sample from a patient with CRC. Then, she smelled each sample station and sat down in front of the station in which she detected cancer scent. She repeated this test with 33 groups of breath samples and 37 groups of stool samples. As in previous studies testing dogs’ olfactory ability to detect cancer, the sensitivity (the proportion of cancer which were correctly identified) and specificity (the proportion of cases without cancer which were correctly identified) of Marine’s scent detection was impressive (see box).
The specificity and sensitivity of canine scent detection according to Sonoda et al. findings (2011)
However, despite these encouraging results, canine olfactory detection of cancer seems unlikely to be introduced into mainstream clinical practice. Why not?
A heterogeneous performance
Reviewing the literature, Lippi and Cervellin (2011) observe that the performance of the dogs used in studies conducted since 1989, is heterogeneous. According to them, this can be due to the different breeds of the dogs, suggesting that their genotypes may influence their perceptions of certain odours. Sonoda et al. (2011) suggest that dogs’ scent ability and concentration not only vary between different types of dogs, but also for the same dog on different days. This observation challenges the reliability of dogs as “screening tools”.
Can one dog detect all types of cancer?
In each reviewed study, dogs have been trained to detect a specific type of cancer (colorectal, lung, breast, prostate, ovarian, bladder, etc.) from a specific type of biological matrix (blood, tissues, breath, urine, stools, etc.). If researchers seem to agree that cancer produce volatile organic compounds (VOC), it is not known for sure yet whether all types of cancer have the same odour signature. This raises the question of whether a single dog could eventually be trained to detect all types of cancer.
Though the study by Sonoda et al. (2011) states that canine scent detection was not confounded by current smoking or benign colorectal disease, Lippi and Cervellin (2011) argue that further investigation is needed. In the reviewed studies, dogs have been tested to distinguish between normal and cancer samples, “but they have not been tested so far to differentiate cancer patients from those suffering from other comorbidities” (2011, p.3). Further investigation is needed to ascertain whether dogs could be misled by non-cancerous diseases which may produce confounding or over-lapping biochemical signals.
What about early forms of cancer?
While the study of Sonoda et al.(2011) suggests the accuracy of canine scent detection was high even for early cancer, Lippi and Cervellin (2011) raise a flag. They argue that dogs used in the reviewed studies (which do not include the one by Sonoda et al.) were trained to identify established and evident malignancies. According to them, dogs’ performance for screening early forms of cancer – and thus expediting provision of timely treatments – remains to be defined.
And the perennial question of cost-effectiveness…
Training dogs is time consuming and expensive and, of course, dogs require care throughout their life. Marine’s training took place over several months and necessitated the expertise of a dog handler. Moreover, hundreds of breath samples from patients with cancer and 500 samples from volunteers were needed to complete her training. The cost-effectiveness of canine scent detection thus requires further analyses.
Hence, what dogs have taught us so far about the detection of cancer is that various types of cancer have detectable odours and that odour material may become effective cancer screening tools. For the scientific community, the next step then is to identify the cancer-specific volatile organic compounds that dogs pick up and to develop technological sensors that could detect them. Yet, this is not easy since researchers have already identified a large number of biomarkers related to specific types of cancer and, as each biomarker may potentially have its own odour signature, determining which compounds dogs actually detect in cancer samples will prove challenging.
|Author :||Myriam Hivon, Ph.D.
Church J., Williams H. (2001). Another sniffer dog for the clinic? Lancet, 358: 930.
Hideto Sonoda, Shunji Kohnoe, Tetsuro Yamazato, Yuji Satoh, Gouki Morizono, Kentaro Shikata, Makoto Morita, Akihiro Watanabe, Masaru Morita, Yoshihiro Kakeji, Fumio Inoue, Yoshihiko Maehara. (2011). Colorectal cancer screening with odour material by canine scent detection. Gut, 60: 814-819 originally published online January 31, 2011. doi: 10.1136/gut.2010.218305
Lippi G., Cervellin G. (2011). Canine olfactory detection of cancer versus laboratory testing: myth or opportunity? Clin Chem Lab Med, 49 (10). DOI 10.1515/CCLM.2011.672
Williams H., Pembroke A. (1989) Sniffer dogs in the melanoma clinic? Lancet, i: 734.