SAN DIEGO -- Frankie the dog has been trained to sniff out thyroid cancer cells in urine with nearly 90% specificity, researchers said here.
The German Shepherd mix's "alert" correctly identified 30 out of 34 samples (88.2%) that later matched the final surgical pathology diagnosis, for a 86.6% sensitivity and 89.5% specificity, reported , of the University of Arkansas for Medical Sciences in Little Rock, and colleagues.
"Current diagnostic procedures for thyroid cancer often yield uncertain results, leading to recurrent medical procedures and a large number of thyroid surgeries performed unnecessarily," Bodenner said during a press conference at The Endocrine Society annual meeting.
Action Points
- Note that this study was published as an abstract and presented at a conference. These data and conclusions should be considered to be preliminary until published in a peer-reviewed journal.
"Based on preliminary data, scent detection by trained canines serves as a noninvasive, inexpensive and highly specific adjunct to current diagnostic practices," he said.
Other researchers have demonstrated similar findings over the past several years. Last year, an Italian group reported that dogs were able to sniff out prostate cancer cells (99.2% sensitivity, 97.1% specificity). ֱ has reported previously on studies where dogs sniffed breath samples to detect breast and lung cancer cells.
Although cancer-sniffing dogs turn in good results, sample sizes in the various studies have been small, and researchers have debated the usefulness of having dogs in a clinical setting.
"We argue among ourselves about this all the time," Bodenner acknowledged.
One of the best arguments in favor of using dogs is noninvasiveness. Thyroid nodules are often tested for malignancy with fine-needle aspiration, a procedure that is invasive and potentially uncomfortable for patients, pointed out , of Washington Hospital Center in Washington.
"There are some people who have multiple nodules that require repeated biopsies over time, and don't look forward to it," said Wexler, who moderated the press conference. "If you could design a technique where you have no invasive procedure, that could have tremendous widespread appeal."
Bodenner's group took 59 patients who presented at a thyroid clinic with more than one thyroid nodule at risk of being cancerous.
Frankie had been imprinted with urine, blood, and thyroid tissue from patients with metastatic thyroid carcinoma and was trained for more than 6 months to learn to discriminate cancer cells.
A handler, wearing gloves and blind to the sample, presented a 3-mL tube of urine to the dog, which would lie down if the cells were cancerous or turn away if benign. Only 34 of those samples, which were later confirmed to be benign or to have papillary thyroid cancer cells, were included.
Another advantage to scent-trained canines is a potentially lower price-tag. Next generation gene sequencing techniques for cancer diagnosis, such as and , can be "extremely expensive," Wexler said.
ThyGenX tests for the mutations that are likely to later turn into malignancy, while Afirma identifies benign nodules by analyzing the expression of 167 genes through messenger RNA. In 2013, ֱ reported that a diagnosis using Afirma costs $4,000.
Wexler noted that the use of cancer-sniffing dogs could potentially keep the costs of diagnosis down.
Bodenner said that in the future, cancer-sniffing dogs could be used as an adjunct in clinics. "It's not to say: 'It's benign -- go home'," he said, but it could work with other tests to identify cancerous cells.
In addition, the dogs could be helpful for patients who are at a high risk of recurrence. "The trained dogs could live with people at risk -- they could more or less adopt the dog," Bodenner told ֱ. He also said he was hopeful that the dogs could be used in developing countries, where access to more expensive tests and clinics are limited.
"If we can get sample numbers into the hundreds and still maintain a 90% accuracy, we're probably going to be ready for prime-time, but that's going to be several years from now," Bodenner said. "But it's very doable. We have to get the statistics to convince people that this isn't some stupid idea."
Disclosures
Bodenner and co-authors disclosed no relevant relationships with industry.
Primary Source
The Endocrine Society
Source Reference: Bodenner D, et al. "Scent-trained canine prospectively detects thyroid cancer in human urine samples" ENDO 2015; Abstract FRI-036.