What Does Cancer Smell Like? Why scientists are putting stock in an electronic nose

November 19, 2013

What Does Cancer Smell Like?

By VERONIQUE GREENWOOD

On a lab bench in Philadelphia sits a tiny box lined with nearly invisible nanotubes and gold. A clear plastic pipe runs through it, and a thicket of pins, each sprouting a red or blue wire, protrudes from its end. As air from the pipe wafts over the nanotubes, electrical signals surge out of the box along the wire threads. The whole apparatus is situated near a vial of blood, “sniffing” the air above it through the pipe.The box, an electronic nose, is a key part of a theory being explored by George Preti, an organic chemist at the Monell Chemical Senses Center, and an interdisciplinary team that includes physicists and veterinarians at the University of Pennsylvania. Preti is an expert on human odors, having studied them for more than 40 years. He has sniffed — both with machines and with his nose — breath, sweat and other secretions in search of answers about why we smell the way we do. This latest project seeks to answer a question others might have never thought to ask: Does ovarian cancer have a smell?

In modern cancer medicine, doctors tend to rely on advanced imaging techniques and the detection of lumps. The widely acknowledged problem with these methods, though, is that by the time doctors have reason to order a scan or feel something, it’s often too late. Ovarian cancer has usually spread to other organs by the time it’s detected. If it is caught early — which happens only 15 percent of the time, often by accident when doctors are looking for something else — 92 percent of patients live for at least five years. But when it’s caught late, that rate drops to 27 percent. Scent might be a way to get there sooner.

Discovering earlier and better markers for all kinds of cancer, especially in blood, is a priority, said Dr. J. Leonard Lichtenfeld, deputy chief medical officer of the American Cancer Society. Ovarian cancer already has a blood test that has turned out to be not as useful as hoped — giving out both false positives and negatives. A smell-based test would need to perform better.

Diseases can subtly alter people’s fragrance. In the normal course of metabolism, thousands of waste products are swept out in our breath, blood and urine, or simply released into the air above the skin. Metabolic disorders, like diabetes, interfere with the way the body breaks down nutrients and thus make that exhaust especially stinky. People with phenylketonuria (or PKU) tend to smell musty. A faulty or missing digestive enzyme makes people with trimethylaminuria (or TMAU) smell fishy. Untreated diabetics can smell like nail-polish remover: Unable to get energy from sugar, their bodies burn fat for fuel and release acetone as a by-product. (These scents don’t always smell bad; there exists a disorder known as “maple syrup urine disease.”) For Preti, originally from Brooklyn, this makes a subway ride unusually informative. “I often tell people I work with, ‘I bumped into the guy with isovaleric acidemia today.’ ”

Cancer cells, though they don’t alter human metabolism overall, can have altered metabolisms themselves. That means the substances they release could differ from those generated by healthy cells. This idea has been around for decades, but only very recently have biochemical and sensor technology advanced to the point where we can develop portable, hand-held sniffing machines.

Electronic noses have the potential to detect even very small amounts of molecules — but they need to be programmed to look for specific signs wafting up from patient samples. To do that, A.T. Charlie Johnson, a physicist and collaborator of Preti’s at Penn, has the electronic nose sniff blood samples from both sick and healthy patients. As the air passes through the tube, molecules from the samples alight on strands of sticky DNA attached to the carbon nanotubes, changing the electrical signals running out of the box. The team can look for patterns in the signals and use the difference — if there is one — between cancer samples and healthy samples to create an odor-based ovarian cancer test. (Preti is also attempting to identify the specific molecules present in ovarian cancer sufferers’ blood using a much larger machine called a gas chromatograph-mass spectrometer.)

A work in progress, the electronic nose is, for now, an example of how modern medicine can look for answers in unusual places. The impetus that finally pushed Preti and his team to seriously investigate the possibility of cancer detection by smell traces its roots to a dog. In 1989, a letter published in The Lancet reported that a woman had come into the doctor’s office to have a mole looked at. She hadn’t noticed it until her collie-Doberman mix began to sniff the spot intently — even through her pants — and tried to bite it off when she wore shorts. The mole turned out to be an early-stage malignant melanoma, inspiring researchers to test whether dogs, whose smell machinery is at least 10,000 times as sensitive as ours, can tell healthy samples from cancerous ones.

The results from the dog tests have been inconclusive, but to Preti, who has mulled the idea that hidden cancers could be detected from smell molecules since the 1970s, they suggested that there was a real possibility for a new diagnostic. “We think that they’re present very early in the carcinoma process,” Preti said of the scents. “The main question is: Can we be as sensitive as the dogs in picking these things up?”