Science: Smelling Faults

Soon an electronic nose could be used to sniff out a range of illnesses from schizophrenia to lung cancer to cirrhosis of the liver. Hugh Aldersey-Williams reports on the progress tronic nosetandfirst to go here tandfirst to go here tandfirst to

SCHIZOPHRENIA is the greatest single drain on health resources. It may affect just one person in 100, but it can be a lifelong illness, and costs pounds 1.5bn annually. Yet there is no drug-based cure and no clinical diagnosis for this illness.

Schizophrenia and a range of other mental disorders are thought to be triggered by biochemical imbalances. These imbalances also alter the composition of exhaled breath and body secretions and physicians as long ago as the 18th century noted the anomalous breath odour of patients who would today be described as schizophrenic. Reports to this effect appear in the medical literature up to the first half of this century. But, following the Second World War, increased standards of hygiene discouraged such observations. Now, scientists believe that this lost skill may be regained and improved in order to provide a diagnostic tool for schizophrenia.

"We went recently to Romania, and there it's part of the routine teaching that people should recognise the smell of schizophrenia," says Dr Iain Glen of the Highland Psychiatric Research Group (HPRG) at Inverness's Craig Dunain Hospital. "Here, medical students are seldom taught about it. There have been attempts to isolate the smell of schizophrenia from sweat, but, until now, not using recent technology."

Now, new analytical techniques are making it possible to accurately identify chemicals present in minute amounts. The HPRG, which has already developed a promising skin test, is now involved in a multi-disciplinary collaboration with Scotia Pharmaceuticals and odour- sensing technology pioneer, Dr George Dodd, to develop a powerful and versatile breath test.

The HPRG's research so far has shown that the nerve-cell membranes in schizophrenics are deficient in two acids - arachidonic acid and docosahexaenoic acid. "If there is a loss of arachidonic acid, where is it going?" asks Dr Glen. "We think it is being oxidised and going into the breath." Ethane and pentane have been detected on the breath. Other researchers have shown that levels of the enzymes that convert these acids into prostaglandins are high during acute periods of illness. In 1996, it was found that a gene associated with one of the enzymes had abnormal forms in schizophrenia patients. Both environment, through the intake of unsaturated fatty acids, and genetics play their part. The picture was complete.

It has taken the HPRG team 15 years to get to this stage, and to overcome scepticism from the medical community, to get smell accepted as a diagnostic tool for schizophrenia. The Scottish researchers recently obtained their first government funding to continue their work towards a clinical diagnosis and possible drug treatment of the root causes of the disease. "The skin test has produced major interest," says Dr David Horrobin, director of Scotia Pharmaceuticals and president of the Schizophren-ia Association of Great Britain. "Patients, too, are fascinated. They say, 'I always knew it wasn't all in my head.' For many, it's an immense revelation to see a skin test on their forearm. But the breath test is potentially enormously more sophisticated. It can give you answers potentially to many more diseases."

Construction started last month in Inverness on the breath-test apparatus. This is a modification of technology originally developed by the US Environmental Protection Agency to test for contamination by organic chemicals in industry. "We're talking about picograms [million-millionth parts of a gram] in the breath, so we needed a technology to detect molecules at very low levels," says Dr Glen.

For the patient the experience is much like using a breathalyser. In the laboratory, the stored breath sample is heated to liberate the breath constituents which are then separated using gas chromatography. These individual substances are analysed in a mass spectrometer, and the data obtained used to determine their composition and concentration.

But the fundamental science of smell is still in its infancy. Some of the leading players in the field are showmen as much as scientists. Dr Dodd, for example, who set up the Institute of Olfactory Research at Warwick University in the 1970s, has developed a perfume as well as a fish attractant for anglers based on synthesised human female pheromones. He believes that the technology now available to the HPRG might also be used to reveal the secrets of human sexual attraction. "With this instrument I'd be disappointed if we weren't able within six months for the first time to obtain individual pheromone profiles and then carry out pheromone profiling on a large scale, and for the first time ever begin to map out the individual differences in human body odour."

Because so little is known about the human sense of smell, the focus in perfumery and other fields where odour is important has been on the empirical relation between a smell as smelled and the mood the smeller reports it induces. Knowledge of how odour is sensed and decoded by the nose and brain is not absolutely necessary in order to achieve desired results in these applications which is one reason why little research has been done in this area.

Without a fundamental understanding of how smell works, the first generation of "electronic noses" is aimed at mimicking animal noses. "Nature has solved the problem already," says Dr Dodd. "A sniffer dog has sensitivity far exceeding any instrument we have for detecting smells." Although typically trained to sniff out drugs and explosives, their ability to detect minute amounts of key substances is quite general. "We could train them on wine if we chose."

"A biological system has an array of perhaps up to a thousand sensors," Dr Dodd explains. "These are connected in complex ways to central processors which interrogate the sensors in ways that we do not understand." In place of biological receptors, electronic noses use electrically conducting polymer substrates sensitive to the presence of certain groups of molecules, although exactly how these work is also unclear. Variations in the fabrication process and moisture in the sample greatly affect their performance. An electronic nose that can distinguish dry samples of different coffee varieties might still have trouble telling a steaming cup of Java from hot water.

When it detects certain molecules, the polymer produces an electrical signal. The whole array of sensors will produce a mass of such signals of different intensities according to the composition of the gas mixture introduced. Statistical procedures may be used to analyse the mixture, or a neural network processor behind the sensors may be "trained" to distinguish particular odours from the background much like sniffer dogs or wine experts.

One challenge is to balance the number of sensors and the computer power behind them. Following nature, it would seem that the more sensors the better - dogs have millions of them. But adding too many sensors can add to the "background noise" in the detected signal and slow down data processing.

For many applications, the ability to pick out target constituents from a mixture of odorants is more important than sensitivity to very low concentrations. While developing his electronic nose, Dr Dodd carried out development work with the brewer, Bass. The brewing process can go wrong in many ways, each producing its own cocktail of chemical contaminants. The requirement to distinguish between them directed the project away from increasing the sensitivity of the array towards developing the "parallel" computing processing that would enhance discrimination. Since those early days, both substrate sensitivity and the ability to process the signal arrays that they generate have improved. But electronic noses are still only good at detecting substances they have been "taught" to identify.

Nevertheless, progress in the field is keenly watched in industries such as food and tobacco where the aroma is a reliable indicator of its quality. A wide variety of applications are imagined. The Royal Veterinary College is developing technology to detect oestrus in dairy cattle. General Motors wants to identify the ingredients of "new car smell". Unilever wants machines to replace the people who assess the effectiveness of under-arm deodorants.

Unlike either canine or electronic noses, the gas chromatographic mass spectrometry to be employed by the HPRG can, in principle, detect and identify any airborne substance. Although more sensitive than other devices, it is unfortunately an expensive and bulky item of equipment. "To be useful in the diagnosis of disease it's got to be useful in 50,000 general practices," says Dr Horrobin.

The sensitivity of mass spectrometry also offers hope for the perfume industry. Instead of brutally extracting the delicate raw materials from flowers to create scent, manufacturers might simply sample the air above a rose and synthetically recreate what we actually smell. But there are practical barriers. "The hype is that you can replicate whatever's in the air," says Dr Luca Turin of University College London, a perfumer and researcher into the molecular basis of smell. "But chemicals fall into two classes: ones you already have on your shelf, and ones that it takes years of work to make. So you can guess which ones get chosen. The reality is that it still takes a very good chemist and perfumer to figure out which are the important components."

One advantage of an electronic analogue of a well-trained human "nose" is that it might be linked with other electronic systems. Although a recent application for European funds was turned down, the HPRG team hope ultimately that it might use the digital telephone network to provide home diagnosis of a range of illnesses. "We reckon we can have a sensitive test before people feel the need to go to a doctor," says Dr Dodd.

There is already some evidence to indicate that conditions such as depression, which have similarities with schizophrenia, may also be diagnosed by testing skin or breath samples. Acetone is exhaled in the breath of some diabetics. It is known that bacteria in the colon produce other substances, so there may be a useful link to colon diseases. It is intuitively obvious that diseases from lung cancer to cirrhosis of the liver may affect the breath. There is also the possible biochemical connection between smelling abnormal and losing one's sense of smell; anosmia. The loss of sense of smell is an early sign of schizophrenia as well as of Parkinson's and Alzheimer's diseases.

"Thirty years from now one could take a breath test rather than use intravenous blood sampling and pick up quite a range of diseases," believes Dr Horrobin. "The aim really is to provide a much more sensitive and specific diagnosis for smell symptoms known in the 18th century." !