It will take years, but the researchers intend to bioengineer a nanotechnology sensor - an "electronic nose" - to screen healthy women's blood for the odor of nascent ovarian cancer. The malignancy now defies detection before it has spread, which is why it kills 14,000 women a year.
The project, launched by Penn and the Monell Chemical Senses Center, with an $80,000 grant from the Kaleidoscope of Hope Ovarian Cancer Foundation, involves experts in such disparate fields as veterinary science, oncology, physics, organic chemistry, and nanotechnology.
But they need the four-legged olfactory experts for a breakthrough.
George Preti, a Monell analytical chemist and coleader of the research, noted that sonar and ultrasound technology grew out of studying echolocation in bats and dolphins. In the same way, he said, "we hypothesize that the study of disease-detecting olfaction in dogs can facilitate the development of nanotechnology-based olfactory sensors."
The evidence supporting that hypothesis is tantalizing.
Canines' sense of smell is estimated to be one million times keener than that of humans. Dogs have been taught to sniff out bombs, drugs, termites, bedbugs, disaster victims, even low glucose levels in diabetics.
Various types of cancer are known to produce volatile chemical compounds that are released into the blood, urine, sweat, and breath. Many of these chemicals have already been isolated and identified using gas chromatography and other arcane technology.
Anecdotal cases and studies suggest that humans' best friend can detect the "odor signature" of certain cancers.
The first published report, in 1989, was of a border collie-Doberman mix that persistently sniffed at a mole on his owner's leg. She consulted a doctor after the mutt tried to bite off the mole. It was melanoma, a potentially deadly skin cancer.
In 1995, a Florida dermatologist and a canine-unit police officer trained a schnauzer and a golden retriever to sniff out melanoma on patients whose lab tests were inclusive. The dogs' accuracy was high, although not perfect.
In five larger, more carefully controlled studies, dogs were trained to distinguish normal from malignant samples of breast, bladder, ovary, and skin by smelling tissue, urine, breath, or blood. The dogs' accuracy ranged from stunning to no better than chance.
The reasons for this lack of reliability are unclear. It may have to do with the nature of some dogs - or of some cancers. Or maybe some trainers introduced contamination or bias.
"Is it the dogs who are unpredictable or the training? We don't know," said veterinarian Cindy Otto, head of Penn's Working Dog Center, part of the veterinary school.
Otto cofounded the facility in 2012, drawing on decades of work with search-and-rescue dogs. The center, in a former DuPont lab at 3401 Gray's Ferry Road, is now training a dozen carefully selected pooches for rarefied jobs in the military, law enforcement, and disaster relief.
McBaine and two other furry phenoms - Ohlin and Tsunami - were chosen as the inaugural cancer-sniffers.
A champion nose
Anyone trapped in earthquake rubble would be lucky to be sought by Tsunami, an 8-month-old German shepherd with a champion nose and commanding bark.
But on the first day of cancer training, "Tsu" seemed clueless, unlike slightly older McBaine and Ohlin.
Tsu knew they were playing the search-and-find game. But what was he supposed to find? He saw Otto holding out the vial. He sniffed Otto, ignoring the vial.
"No connection yet," said DeAngelo, the trainer.
The masters teach the dogs to recognize, or "imprint," a scent by linking it to a reward such as food or play.
But within that framework, the humans are constantly adapting and honing the process.
That's why, a few days after training began, they pulled a "daisy wheel" out of storage. The rotating contraption, made by Penn engineering students, had six paint cans attached to the ends of planks. Five were empty; the sixth can held the cancer sample and a scrap of a terry-cloth tug toy. The scrap, replaced by smaller and smaller swatches, would eventually disappear.
On the seventh day of training, handler Donna Magness brought Tsu in on a leash. He walked from one paint can to the next, taking a whiff.
At the fourth can, which held the tissue sample and cloth, he stopped, put his snout deep inside for the scent, then sat down to signal his discovery.
"Good boy!" DeAngelo yelled as she tossed him his terry-cloth toy for a tug with Magness.
Each dog searched seven times that morning, with DeAngelo spinning the wheel between finds. There wasn't a single miss.
Despite the dogs' rapid progress - on Wednesday, McBaine correctly sniffed out the cancer sample even though the cloth was gone - they have a long way to go. They will be steadily challenged with more types of ovarian cancer, benign ovarian disorders, blood rather than tissue from 30 cancer patients, and normal as well as cancerous samples.
If successful, the dogs will distill the odorous essence of the disease.
"If we can separate half a dozen compounds that change in ovarian cancer, isolate them, and present them back to the dog," Preti said, "then we can see if they react the same way" as they react to the cancer samples.
"It's going to take years of work," he added. "But this is a start."
Contact Marie McCullough at 215-854-2720 or firstname.lastname@example.org.