Olfactory diagnostics

Olfactory diagnostics

Smelling bad

Aug 28th 2008
From The Economist print edition

Doctors may soon have a new diagnostic tool in their kit bags

Illustration by Stephen Jeffrey

SINCE time immemorial—or at least as far back as Hippocrates—novice physicians have been taught to smell patients’ breath for signs of illness. Though unpleasant for the doctor, it is a useful trick. The sweet smell of rotten apples, for instance, indicates diabetes. Liver disease, by contrast, often causes the breath to smell fishy. But the human nose cannot detect all the chemical changes brought about by disease. Science, therefore, seeks to smell what human doctors cannot. The aim is to create a diagnostic nose as discriminating as those of perfume mixers or wine buyers. Such a nose would, however, be sensitive not to life’s pleasures, but to its pains.

The idea of creating a diagnostic nose goes back to the 1970s. In that decade Linus Pauling, a Nobel-prize-winning chemist, performed the first serious scientific analysis of human breath. He used a technique called gas chromatography, which enables complex mixtures to be separated into their components, to detect some 250 volatile organic compounds in the air exhaled from lungs. Gas chromatography by itself, however, does not allow you to identify each component—it is merely a way of separating them. To make the identifications, you need to add a second step, called mass spectrometry. This, as its name suggests, works out the weight of the molecules in each component. Often, weight is enough by itself to identify a molecule. But if two molecules happen to have the same weight, they can be analysed by breaking them up into smaller, daughter molecules. These are almost certain to differ in weight.

Using gas chromatography and mass spectrometry, researchers have, over the years, identified more than 3,000 compounds that are regularly exhaled, excreted or exuded from the body. The search, now, is to understand how changes in the mixture of these compounds may indicate disease, and to find ways of recognising such changes routinely and robustly.

Exhaustive analysis

One of the first practitioners of the field of olfactory diagnosis, Carolyn Willis of Amersham Hospital in Britain, decided to contract the job out to dogs. They, she reckoned, have the necessary nasal apparatus to sniff out illness, and there was already some anecdotal evidence that they could, indeed, smell people with cancer. It worked. For the past four years her sniffer dogs have been diagnosing bladder cancer. She is now training them to detect prostate cancer and skin cancer as well.

But training dogs is probably not the best solution. It takes time and needs special skills, so mass-producing sniffer dogs would be hard. Moreover, a dog can give you only a yes-or-no answer. It cannot describe nuances, even if it detects them. Boguslaw Buszewski of Nicolaus Copernicus University in Torun, Poland, compares this approach to checking for fever by touching a patient’s forehead. That tells you he is ill. However, it is only by measuring his temperature with a thermometer that you can discover how serious his condition is. In Dr Buszewski’s view the breath-analysis equivalent of the thermometer is the mass spectrometer, and that is where effort should be concentrated.

Other researchers agree. Earlier this month Michelle Gallagher, of the Monell Chemical Senses Centre in Philadelphia, announced the results of a study that uses this approach. She confirmed that the early stages of basal-cell carcinoma, a type of skin cancer, can be detected by analysing the odour of a person’s skin using gas chromatography and mass spectrometry. To do so, she sampled the air immediately above the tumours and compared its composition with that of air from the same sites in healthy individuals. She also checked the composition of the air in the room when nobody was present, as an extra control. She found that although air collected from both groups contained the same chemical substances, there was a difference in the amounts of some of them. This finding allowed her to produce what is known as a biomarker profile for the illness. That means it can be diagnosed reliably and—crucially—early on.

The combination of gas chromatography and mass spectrometry thus works. It can, nevertheless, take up to two days to run the tests. Dr Buszewski hopes to refine and speed up the process so that it can be carried out within an hour.

To do this, he has developed a device that can be tuned to pick up and concentrate the most relevant molecules. With patents still pending, he is cagey about the details, but the principle is to trap relevant molecules using columns made of metal or silica that are the width of a human hair. Each column is coated with special polymers tweaked so that they bind preferentially to particular compounds found in the breath. Pass a sample through a forest of these columns and the molecules of interest will be sucked out. They can then be flushed into the analytical machinery and a result quickly emerges.

Dr Buszewski is now tweaking his device so that it works with the biomarker profiles of a range of diseases. If he can do this successfully, olfactory diagnosis could become mainstream without a wagging tail in sight.