It’s so quiet you could hear a pin drop. The only sound is that of a Springer Spaniel methodically sniffing twelve ports on a stainless steel wheel. Suddenly he stops and sits.
“Yes! Good boy, McBaine! Way to go, bud!” exclaims one of the trainers, as she rewards him with a treat. McBaine has just successfully identified a malignant ovarian cancer blood sample.
McBaine is one of three cancer detection dogs in training at the Penn Vet Working Dog Center. Along with Labrador Retriever Ffoster and German Shepherd Tsunami, he is using his incredible sense of smell to help create an early cancer detection system that could one day save lives.
The Center is currently focusing on ovarian cancer, known as “the silent killer” because there are often no symptoms of the disease until it has progressed to an advanced stage. Thanks to highly trained dogs like McBaine, there is hope for the future.
“Smelling in Color”
Anecdotes of dogs alerting their owners to undiagnosed cancer have permeated the news for years. In 2013, the Working Dog Center began to study the science behind these stories by launching a cancer detection program with a grant from the Kaleidoscope of Hope Ovarian Cancer Foundation.
“We’ve always known that dogs have an incredible sense of smell. I like to say that they ‘smell in color,’” said Dr. Cindy Otto, Executive Director of the Center. “What we’re trying to do with this study is harness that remarkable ability to inform and create an electronic system for detecting early-stage ovarian cancer.”
A dogs’ nose is indeed remarkable. Canines can smell 10,000 to 100,000 times more acutely than humans. “They really see the world through their noses,” said Otto. With up to 300 million olfactory receptors – compared to six million in humans – and a surface area that is 30 times larger than ours, dogs are uniquely positioned to aid in disease detection.
So it was no surprise when the dogs quickly took to the training.
The Sniff That Saves Lives
The Center initially imprinted the dogs on tissue samples donated from ovarian cancer patients. They have since moved on to imprinting on plasma (blood samples). During the imprinting phase, the dogs are introduced to the scent and rewarded for a sniff. The dogs then progress to using the scent wheel to identify samples.
A key component of the Center’s training is the use of positive reinforcement. When the dogs correctly identify a sample at the wheel, they are rewarded with a treat or toy.
“The goal is to keep it fun,” said Annemarie DeAngelo, Training Director. “Our dogs love to work because it’s a game for them.”
Now three years in, the training has advanced. The cancer wheel is outfitted with twelve ports containing a malignant sample, benign samples, and normal samples, as well as a mix of other “distractors” such as pieces of gloves, paper towels, and other items from the lab environment. The dogs are successfully identifying the malignant sample up to 90% of the time.
“I am absolutely amazed by how quickly the dogs have progressed,” said DeAngelo. “We continue to learn something new every day.”
Among the important discoveries: the need for eliminating human influence.
“We have tweaked the environment in which the dogs work to optimize and validate the results,” said Dr. Lorenzo Ramirez, post-doctoral fellow at the Working Dog Center. For example, dog handlers are no longer permitted in the room during training sessions.
“This enables us to eliminate the ‘Clever Hans’ effect of unconscious cuing,” said Ramirez. “It was really validating to see that, even after removing people from the equation, the dogs were definitely using their noses and brains to figure it out.”
The Center is now conducting double-blind studies, meaning both the dog and the person administering the training session do not know which sample is malignant.
“This is an extremely rewarding project to be involved in,” said Ramirez. “One of my advisors once told me that you know you love your science when it keeps you up at night. And that’s definitely true about this work. We’re always learning news things and thinking about ways to adjust the process.”
Throughout it all, the team has documented their discoveries. “We are very open to sharing ideas and collaborating with others,” added Ramirez. “We want people to learn from the work we are doing.”
However, Otto cautions that these studies are not easy to reproduce. “This work requires a very controlled environment. We’ve learned a lot about what distractions need to be eliminated.”
Partners in Science
A key component to the Center’s success is collaboration. In order to determine exactly what the dogs are smelling – and to reproduce the dogs’ abilities into an early detection system – the Center is partnering with physicians, chemists, and physicists on an interdisciplinary approach.
Dr. Janos Tanyi from Penn Medicine’s division of gynecologic oncology acquires blood samples from ovarian cancer patients that he is treating. These samples are given to the Working Dog Center, as well as to Dr. George Preti, an analytical chemist at the Monell Chemical Senses Center, and Dr. A.T. Charlie Johnson, Director of Penn’s Nano/Bio Interface Center.
“We’re all working in parallel with the same samples,” explained Otto. “We know that there is an odor signature – a unique chemical ‘fingerprint,’ if you will – of malignant ovarian cancer, so we are all working backwards to determine what that is.”
By isolating the chemicals of cancer that the dogs can smell, the team hopes to one day create an electronic sensor that can detect cancerous tissue early.
“The end goal is not to have dogs in doctor’s offices,” explained Otto. “We actually want the dogs to work themselves out of a job. If our team of scientists can identify what it is exactly that the dogs are detecting, we can take [the dogs] out of the equation.”
At Monell, Preti is working to isolate and identify the volatile organic compounds (VOCs) of ovarian cancer’s unique odor signature. VOCs are odorants, and Preti is looking for the odors that are altered in ovarian cancer, particularly those that signal the disease’s earliest stages. As Preti isolates candidate compounds, he plans to send them back to the dogs to test whether they respond and validate the results.
Meanwhile, Johnson, a physicist, is creating an electronic nose comprised of nanosensors that mimic the abilities of the dog’s nose. In his laboratory, Johnson releases concentrated blood samples over sensors he created that include single strands of DNA that bond with specific chemical compounds. Eventually, the sensors will be designed to specifically bind to the compounds that the dogs identify.
“Right now, the dogs are informing and refining the electronic system,” said Otto. “Eventually, the machine will be able to process hundreds of thousands of samples based on what the dogs have taught us. And that’s really exciting.”