Best Doctors 2013: On the Scent

Cleveland Clinic's Dr. Peter Mazzone has discovered a new way to detect lung cancer — with an electronic nose.

Like most doctors, Dr. Peter Mazzone had long known that certain diseases produce substances in the body that yield their own signature scents. Patients suffering from cirrhosis of the liver, for example, have a slight odor of ammonia about them, while those with uncontrolled diabetes actually smell sweet. The Cleveland Clinic Respiratory Institute pulmonologist and researcher was also aware of the instruments developed by various industries to detect unsafe chemical levels in the air. But it wasn't until two of his colleagues returned from a 2003 conference with news of a sensor designed to detect breath chemicals that he came up with the idea of developing an electronic nose for lung cancer.

"Cancer cells work a little differently than normal cells," he explains. "We know that because they behave differently: They grow faster, they have different properties than normal cells. Their exhaust systems give off different chemicals than the exhaust system of a normal cell would. That exhaust is carried in our blood to our lungs, where we breathe it out into the air."

Ten years and three studies later, the sniff test performed by Mazzone's electronic nose is accurately detecting lung cancer 80 to 85 percent of the time. The device's key component is a thick, credit-card-sized sensor covered with 130 dots, each made of pigments that react to chemicals in the breath by changing color. The sensor is housed in a machine resembling a desktop computer tower. During previous studies, subjects diagnosed with lung cancer or determined to be cancer-free exhaled into a collection instrument that removes the first expulsion of air — a process that exposes the sensor to air from deep in the lungs, where the highest chemical concentrations are found. The sensor is then photographed by a camera built into the device and researchers analyze the images.

Mazzone stresses that the sniff test doesn't identify the exact chemicals responsible for pigment changes. Studies conducted by other researchers to determine specific chemical differences in healthy and cancerous breaths have yet to produce consistent results. He compares his test to the human ability to identify, say, the smell of coffee without discerning the combination of chemicals that produce it.

"It's just the pattern of changes in the sensor that we're looking at, to see if that's a pattern in keeping with cancer or not," he says.

Such a screening is needed. Mazzone notes that although lung cancer isn't the most common cancer in the U.S., it is the most lethal, claiming 160,000 American lives each year. Only 16 percent of patients survive, often because they experience no symptoms and aren't diagnosed until the disease reaches an advanced stage. In the last year, the CT scan has become a generally accepted screening tool for high-risk individuals, defined as smokers ages 55 to 74 who have smoked at least 30 "pack years," meaning they smoked a pack of cigarettes a day for 30 years or two packs a day for 15 years. But the CT scan is an expensive test that emits a low dose of radiation and produces images identifying scars caused by infections such as pneumonia, which can be mistaken for tumors. The sniff test is safe, quick and inexpensive. Mazzone estimates the disposable single-use sensor will cost no more than $5 to make.

"It would also open up testing to people who might not have quite as high a risk of developing lung cancer," he adds.

Mazzone is striving to attain an accuracy rate of 95 percent in a yearlong study concluding at the Cleveland Clinic and National Jewish Hospital in Denver. The results will hopefully lead to a three-year study with thousands of lung-cancer and control subjects instead of the hundreds tested in previous pilots. Mazzone estimates another five years could pass before the electronic nose is in doctors' offices, regardless of how compelling results from a large-scale study may be.

"That sort of a study requires an awful lot of support," he says. "And support in research is very difficult to find."

Share this story: