Canine olfactory detection of malignant melanoma

Abstract

Our patient is a 75-year-old man who presented after his pet dog licked persistently at an asymptomatic lesion behind his right ear. Examination revealed a nodular lesion in the postauricular sulcus. Histology confirmed malignant melanoma, which was subsequently excised. Canine olfactory detection of human malignancy is a well-documented phenomenon. Advanced olfaction is hypothesised to explain canine detection of bladder, breast, colorectal, lung, ovarian, prostate and skin cancers. Further research in this area may facilitate the development of a highly accurate aid to diagnosis for many malignancies, including melanoma.

Background

Malignant melanoma (MM) is the most serious form of the common skin cancers. MM is the sixth most common cancer in the UK¹ with a crude incidence of 15.9/100 000, equating to 9500 new cases per annum.² Breslow thickness, the thickness of a melanoma in millimetres measured from the top of the granular layer of the epidermis,³ is the most important prognostic indicator.⁴ Excision of small, thin melanomas is potentially curative. Therefore, early detection is essential to improve 5-year survival rates.

MM diagnosis is made on the basis of clinical examination, which is inherently subjective relying on the physician's training and experience, and therefore results in a high-false-positive rate. An analysis of UK data revealed that the ratio of benign moles for each MM excised was 6:3.⁵ Dermoscopy enhances diagnostic accuracy,⁶ however, development of an independent objective technique may be a valuable adjunct in the early detection of MM.

Williams and Pembroke reported the first case of canine olfactory detection of MM in 1989.⁷ The patient's dog, a Border Collie-Doberman cross, would sniff repeatedly at the patient's lesion even through clothing, revealing an olfactory origin to the behaviour. The authors surmised the dog had probably saved her owner's life by prompting her to seek treatment while the lesion, in this case a superficial spreading MM, was still at a ‘thin and curable’ stage.

This case report adds further anecdotal evidence of canine olfactory detection of MM and prompts further debate about the potential application of canine olfaction to early MM diagnosis in clinical practice.

Case presentation

A 75-year-old white Caucasian man presented to the dermatology clinic after his pet dog licked persistently at an asymptomatic lesion behind his right ear. The patient was previously unaware of the lesion. The dog, a female Alsatian (figure 1), was acquired from a rescue centre 4 years previously. Therefore, the dog's previous history and exact age was unknown. The patient had a history of significant sun exposure. He also suffered with eczema as an adolescent. There was no family history of skin disease or skin cancer. Medical history included multiple myeloma diagnosed 4 years previously, for which the patient had received chemotherapy, bortezomib and radiotherapy. At the time of presentation he was in slow relapse.

Figure 1.

The patient's dog.

Clinical examination revealed a 2 cm, dry, scaly, nodular lesion in the right postauricular sulcus. It was asymmetrical with an irregular border and a peripheral ring of pigmentation (figure 2). It was initially suspected to be a basal cell carcinoma. The differential diagnosis included squamous cell carcinoma and MM and therefore a biopsy was taken to plan treatment.

Figure 2.

The patient's lesion.

Investigations

A 3 mm diameter punch biopsy was sent for histological evaluation. A microscopic analysis revealed an invasive, ulcerated MM, with an estimated Breslow thickness of 2.5 mm. Diagnosis was confirmed with Panmelanoma staining—an antibody cocktail (HMB45, anti-MART1, Melan A and antityrosinase) designed to detect melanocytic differentiation.^{8 9}

Differential diagnosis

The differential diagnosis of pigmented skin lesions includes: benign naevus, dysplastic naevus, seborrhoeic keratosis, pigmented-basal cell carcinoma, pigmented-squamous cell carcinoma and dermatofibroma.

Treatment

The MM was removed surgically with a 1 cm wide local excision, including the underlying cartilage. A concurrent full-thickness skin graft was performed.

Outcome and follow-up

The histological appearance of the specimen was of a completely regressed MM, with a small focus of residual lentigo maligna. The lesion was completely excised by a margin of 2.5 mm. Subsequent follow-up revealed a well-healed surgical site, with no evidence of local recurrence. A CT scan revealed no MM-related pathology, although lytic lesions in the axial skeleton associated with the patient's myeloma were observed. The patient had regular dermatology follow-up, with no evidence of recurrence of his MM, until his death 11 months later from progression of his multiple myeloma.

Discussion

Since the first case report of canine olfactory detection of MM in 1989 there has been limited exploration of this phenomenon. The sole prospective trial to date involving MM¹⁰ demonstrated sensitivity of 82% in seven patients. While these results are certainly encouraging, the authors noted several limitations to the study including the small patient cohort and the selection bias of the patients involved, who had been preselected by a dermatologist as likely having MM.

The original paper by Williams and Pembroke did not comment on the presence or absence of ulceration. In the cases described by Pickel et al, the melanomas were not ulcerated. In our case, the MM was ulcerated and it is possible that the dog merely detected the lesion because it smelled different to the rest of the skin, rather than detecting any melanoma-specific chemicals. However, evidence exists that dogs can be trained to detect MM and other cancers. It is the reproducibility of these results between patients that suggests the existence of malignancy-specific volatile organic chemicals (VOCs).

Considering other malignancies, there have been numerous case-reports and some small-scale trials demonstrating canine ability to detect bladder, breast, colorectal, lung, ovarian and prostate cancers (table 1).^10–15

Table 1.

Summary of the published evidence of canine olfactory detection of human malignancy

Malignancy	Study type	Canine characteristics	Substrate	Number of cases (excluding controls)	Positive response	Sensitivity (%)	Specificity (%)
Melanoma10	Prospective	1 Standard Schnauzer (m), 1 Golden Retriever (f)	In vivo MM	7	Sitting and pawing or mouthing at sample	82	–
Bladder11	Prospective	6 dogs, no further details published	Urine	27	Lying next to positive sample	41	–
Breast12	Prospective	3 Labrador Retrievers (2m, 1f) 2 Portuguese Water Dogs (1m, 1f)	Exhaled breath	31	Sitting or lying in front of sample	88	98
Colorectal13	Prospective	1 Labrador Retriever (f)	Exhaled breath (B) and watery stools (S)	B 33 S 37	Sitting in front of sample	91	99
Lung12	Prospective	3 Labrador Retrievers (2m, 1f) 2 Portuguese Water Dogs (1m, 1f)	Exhaled breath	55	Sitting or lying in front of sample	99	99
Ovarian14	Prospective	2 Giant Schnauzers (2f)	Tissue (T) and blood (B)	T 10 B 10	Scratching at sample with forelegs	T 100 B 100	T 95 B 98
Prostate15	Prospective	Belgian Malinois (gender not published)	Urine	33	Sitting in front of sample	91	91

MM, malignant melanoma.

The literature suggests that canine olfactory detection of malignancy may have potential, both in terms of application in contemporary clinical practice and directing future research.

Further validation of canine olfactory detection of MM may provide evidence to support the introduction of ‘sniffer dogs’ to dermatological practice. Dogs are already widely used in healthcare, for example, as guide dogs, hearing dogs and medical alert dogs. Thus large-scale patient acceptance is already demonstrated. In terms of financial viability, the training of scent detection dogs could be adopted from current training regimes used in other industries. This might offer a cost effective adjunct to diagnosis, particularly in resource poor settings where investigations are limited.

Alternatively, identification of the specific VOCs being detected by canine olfaction could facilitate the development of a technologically based scent detection system, a so-called ‘electronic nose’.¹⁶ A recent study has identified three VOCs preferentially expressed in MM: 4-methyl decane, dodecane and undecane.¹⁷ This study involved various melanoma types, including: MM in situ, recurrent, superficial spreading, nodular and metastatic, revealing common VOC expression. While research into VOC detection has been pursued for decades, usually based on gas chromatography, currently canines significantly out-perform the technological equivalents. Further study of canine olfaction may help to refine the technology potentially improving the detection of melanoma and a variety of other malignancies, including those which are currently particularly difficult to detect at an early stage.

Learning points.

• There is extensive peer-reviewed evidence of canine olfactory detection of volatile organic chemicals associated with human cancers, including melanoma, bladder, breast, colorectal, lung, ovarian and prostate.

• Existing canine training regimes for example, explosive/narcotic detection, could be easily modified for cancer detection, making this a financially attractive diagnostic tool.

• Identification of the volatile organic chemicals being detected by canine olfaction may facilitate the development of a technologically based method for diagnosing a variety of malignancies.