Waiting to exhale

Scientists sift the chemical potpourri that escapes our lungs for new ways to diagnose disease

CATCHING YOUR BREATH  The chemicals that escape our lungs could give doctors new ways to diagnose disease.

© Blend Images/Alamy

The 800 or so breaths you release each hour contain more than just spent air. Along with familiar gases like carbon dioxide, nitrogen and oxygen, each breath holds a vaporized record of the foods you’ve eaten, the places you’ve been, the drugs you’ve taken, the pollutants you’ve encountered and the general operation of your internal organs. It’s a chronicle of daily living that doctors have been largely unable to read.

But a handful of researchers are getting better at deciphering these gaseous clues, bringing us closer to the day when a kind of disease breathalyzer could be part of a routine checkup or maybe even a cell phone app.

The idea of medical diagnosis through exhaled odor is as old as the practice of medicine itself. Hippocrates wrote a treatise on “fetor hepaticus,” or the fishy aroma of liver failure, and noted the sour-scented breath of those with failing kidneys. Only recently, however, has the mainstream medical establishment come to view the idea of breath analysis as anything more than, as one researcher described it, a bizarre curiosity.

JUST BLOW More than 99 percent of a person’s exhaled breath consists of gases already present in the atmosphere. But the remaining 1 percent consists of volatile organic compounds, some of which rise and fall depending on the person’s health, metabolic processes and environmental exposures. Source: R. Dweik M. Atarod

“When I go and talk to a bunch of doctors, they don’t laugh at me any more,” says internist Michael Phillips, who founded Menssana Research in Fort Lee, N.J., to develop breath tests for disease.

While the experimental technology still tends to be cumbersome — some breath-reading devices are the size of small refrigerators — it has progressed far enough in recent years that scientists have moved beyond theoretical scenarios. Doctors already have a few breath tests at their disposal, and researchers are making headway on the breath detection of tuberculosis, heart failure, environmental exposures, liver disease and even some cancers.

“The really good thing about breath is, you can measure it anywhere, in any time frame you want. You don’t run out of it,” says Joachim Pleil, an analytical chemist at the U.S. Environmental Protection Agency. “You can collect a breath sample from a guy standing on a street corner, and you can collect five samples or more in 10 minutes.” Breath testing is as convenient and painless as medical tests come, even for the most vulnerable newborn or the frailest centenarian. The patient doesn’t even have to be conscious.

For the idea to work, researchers have to mine breath samples to find airborne nuggets of disease, then build machines that detect those molecules with the kind of sensitivity (the ability to correctly identify people who are sick) and specificity (the ability to rule out people who are not) necessary for a medical diagnosis.

Scientists know of more than 3,000 chemical compounds that can come out of the lungs, but often in infinitesimally small concentrations. Some signs of disease waft out at concentrations of parts per billion. The challenge is not just finding a needle in a haystack, but a needle in a haystack the size of a townhouse. And doing it cheaply enough to be practical. “We have to come up with a test that is useful, not just show-and-tell,” says Raed Dweik, a pulmonologist at Cleveland Clinic.

Nonetheless, recent experiments provide reason for optimism. One of Dweik’s research interests is heart failure, which affects more than 5 million Americans and is one of the most common reasons that older adults are admitted to the hospital. The disease develops when the heart muscle becomes damaged and unable to properly pump blood. The body’s tissues are gradually starved of fuel. Among the chemical calling cards of heart failure is excess production of acetone and pentane by cells that are running on too little oxygen.

Writing in the Journal of the American College of Cardiology in April, Dweik and his colleagues described a study of 41 patients, two-thirds of whom had been hospitalized with heart failure. Based on a single exhalation, Dweik’s breath test could separate the people with heart failure from the rest of the patients based on their acetone and pentane levels. He says his heart failure breath test has so far been accurate every time. Proof-of-principle experiments like these are important, he says, because researchers have to demonstrate that the technology can identify patients who are known to be sick before moving on to testing in those whose health is unknown.

Dweik is also using acetone, pentane and another compound, trimethylamine, as the foundation of an experimental breath test for liver disease. Trimethylamine levels rise in people with diseased livers because the enzyme that usually removes the compound from the body has dwindled. Dweik and colleagues reported in the Sept. 10 Clinical Gastroenterology and Hepatology that among patients with alcoholic hepatitis, a precursor to liver failure, the three compounds exceeded a certain threshold level about 90 percent of the time. (For comparison, mammography for breast cancer screening has an overall sensitivity of about 79 percent.)

DIAGNOSTIC LADDER A breath test for a disease such as breast cancer would not be the final say in diagnosis. Researchers generally think of it as a first step to determine the need for more expensive and invasive tests.

Rapid TB test

A breath test for tuberculosis has amassed a similarly encouraging track record. TB is second only to HIV as a causeof death from infection, and most of those affected live in countries with poor health care and scant resources. Rapid identification of those who are ill could help control outbreaks and lower mortality by getting treatment started sooner.

Definitive diagnosis of infection requires a chest X-ray or microscopic examination of lung sputum, both of which are time consuming and require access to a hospital and laboratories.

“One of the biggest problems, especially in developing countries, is they don’t have much money to go out and find people with tuberculosis,” says Menssana’s Phillips, who is also on the clinical faculty of the New York Medical College.

He set out to develop a TB breath test that provides almost immediate results. His device measures compounds like cyclohexane, an emission of TB-causing bacteria. Last year in the journal Tuberculosis, Phillips and his colleagues reported the results of a trial involving more than 250 patients in three countries. Six minutes after collecting a breath sample, the device detected tuberculosis with 80 percent accuracy.

Phillips is also one among a number of researchers in the United States and internationally who are trying to develop a breath test for cancer, notably lung cancer. Heavy smokers are now offered CT scans, but the method is costly and produces many false alarms. A cheap, widely available breath test could help decide who needs a CT screening, says Anton Amann of Innsbruck Medical University in Austria. He is developing a breath test based on the levels of compounds that are found in the lungs of cancer patients but not healthy volunteers, regardless of whether they are smokers or nonsmokers. That’s an important distinction, because a breath test would need to tell the difference between cancer and general lung damage from tobacco. Around half of lung cancer patients are smokers at the time of diagnosis.

In 2009 in the journal BMC Cancer, Amann reported that testing for 15 different compounds allowed his researchteam to identify the breath of lung cancer patients 71 percent of the time. When they expanded to test for 21 compounds, sensitivity rose to 80 percent.

BREATHPRINT Trace levels of organic chemicals in the breath can reliably distinguish heart failure from other cardiac conditions. Based on statistical profiles of breath constituents, researchers developed a “breathprint” that can be illustrated on a two-dimensional plot. After plotting 25 known cases of heart failure (blue dots) against controls (orange dots), they validated the method by correctly flagging 36 cases (green dots) from unknown samples. Source: M. Samara et al/J. Am. Coll. Cardiol. 2013 M. Atarod

While promising, the results are still far from an actual test. For one thing, researchers must establish that a breath test could identify the malignancy at an early stage. Any test would also have to take into account the metabolic signatures of other tobacco-driven diseases, like emphysema, which could confuse the results of a breath screen.

Lung cancer is not the only malignancy under investigation. Scientists from Technion–Israel Institute of Technology in Haifa are developing a stomach cancer test with colleagues from China and Latvia. Writing this spring in the British Journal of Cancer, the researchers found that a prototype breath test could single out stomach cancer 89 percent of the time by searching for five particular compounds.

Aside from the obvious targets, breath tests are opening the door to other kinds of early detection. The EPA’s Pleil is hoping to see screening for exposures to environmental pollutants that would produce instant results by measuring inflammatory reactions in the lungs — and with greater ease than urine or blood tests. He is also working with researchers from the University of North Carolina at Chapel Hill to develop a way to detect early lung infections among patients with cystic fibrosis. The disease causes mucus to build up in a patient’s lungs, and this warm, moist environment offers an inviting home for microbes. “It behooves the medical community to be able to diagnose a ramping infection before the person actually gets sick,” Pleil says, to give the right antibiotic as soon as possible. He expects to publish data soon.

Easier said than done

Although encouraged by the progress, researchers concede that breath testing is more challenging than it may seem. Clinicians have come to appreciate the complex interplay between body and breath. Breath testing is possible because the lungs catch molecules that escape from the bloodstream through small capillaries nestled against air sacs deep in the lungs. Some of these molecules come from substances that have been taken in and made their way into the blood, which is why an alcohol breath test works. Other chemicals are the natural remnants of metabolism.  Then there are the bystander molecules that have been inhaled and re-released. “If you put gas in your car, you’ll be breathing out gas the rest of the day,” says Terence Risby, an environmental health scientist at Johns Hopkins School of Public Health. And not all of it came from fumes; substances absorbed through the hands and into the blood can also exit by air.

BREATH TEST PIPELINE Some tests that use breath to detect disease are already on the market, with more under development. Source: K. Paschke et al/F1000 Medicine Reports 2010

Bacteria, which outnumber the body’s human cells by 10 to 1, add to the mix as well as they constantly belch waste products that make their way into the airways of their hosts. The bacteria within might also have surprising influences on health. In April, researchers from Cedars-Sinai Medical Center in Los Angeles reported in theJournal of Clinical Endocrinology & Metabolism that a breath heavy with hydrogen and methane was associated with a higher body mass index and more body fat, possibly because certain bacteria living in the intestine are affecting metabolism.

With this vast chemical potpourri traveling through the lungs, breath tests won’t work unless scientists can tease out molecules that are biologically significant from those that aren’t. One more problem: Collecting a pure breath sample is not as easy as puffing into a police officer’s roadside breathalyzer. The signals of illness are far too diffuse, and too easily drowned out by background noise. Even the act of breathing into a machine — a person tends to hyperventilate from trying too hard — or being in a hospital alters the composition of the breath. “If you don’t control how you’re sampling, you’re analyzing garbage,” says Risby, of Johns Hopkins.

His point: Breath analysis is not just about detecting the presence or absence of molecules. The breath changes when the body becomes ill, but usually it doesn’t produce anything new or as obviously abundant as the steamy cloud of ethanol exhaled by a drunk driver. Instead, chemicals that occur naturally become more or less common, just as the same ingredients can produce a cupcake or a muffin depending on their proportions. The balance of chemicals, not their presence or absence, is what usually gives an illness a distinctive breath.

“I can smell an orange, but can’t tell you what about it makes it an orange,” Dweik says. So it is for the molecular composition of your breath. Changes occur, but the precise recipes for “cancer breath” or “kidney disease breath” are still being worked out. (Which is one of the limitations of a similar line of work, Dweik says: training dogs to detect disease in a whiff of a person’s breath. A canine nose may be able to sniff cancer, but scientists still can’t say what exactly the dogs are smelling, which may limit the future of dog detection.)

SMELL PHONE An inexpensive cell phone–mounted device already exists to measure blood alcohol level by analyzing breath. In the future, similar devices may turn smartphones into medical diagnostic devices. Courtesy Alcohoot, istockphoto

To better define the aerosol signature of disease, researchers are stepping back and examining dynamic shifts in breath not only from person to person, but even in the same individual over time. In April, scientists from Zurich reported in PLOS ONE the analysis of breaths sampled from 11 volunteers over nine days. Using technology that separates chemicals based on mass, the researchers found that although a person’s breath changed slightly depending on the time and day the sample was taken, each volunteer had a unique “breathprint,” or chemical composition, that remained distinct and relatively stable.

This seems like basic information, yet establishing what a healthy breath is supposed to be is necessary before doctors can single out an unhealthy one. No one in the field doubts they will eventually succeed. In some cases, they already have. A breath test is now used to detect infection with the bacterium H. pylori, the cause of stomach ulcers, and to help measure the odds that a heart transplant recipient will reject a donor organ within the first year after transplantation. The nitric oxide concentration in a person’s breath is measured for managing asthma.

Once researchers know what a device needs to measure, and how, the next step will be to make the technology as compact and inexpensive as possible. There are precedents: Nitric oxide monitors are small enough now for children to hold. The first alcohol breath test, the Drunkometer, was introduced in 1938 and was bigger than a shoebox. Today, alcohol detectors that plug into a smartphone sell for as little as $30.

If medical breath testing follows a similar pattern of shrinking size and price, the next generation of texters may be able to blow into a thumb-sized device attached to a smartphone to learn about their health. “I don’t think you’re going to have one in your cell phone next week. But it’s possible you could have one in a few years,” Phillips says. “It’s just a matter of ramping up the sensitivity.” 

More Stories from Science News on Health & Medicine

From the Nature Index

Paid Content