Diagnosing illness by smell
A prototype device to detect the scent of disease
ONE of a doctor’s most valuable tools is his nose. Since ancient times, medics have relied on their sense of smell to help them work out what is wrong with their patients. Fruity odours on the breath, for example, let them monitor the condition of diabetics. Foul ones assist the diagnosis of respiratory-tract infections.
But doctors can, as it were, smell only what they can smell—and many compounds characteristic of disease are odourless. To deal with this limitation Hossam Haick, a chemical engineer at the Technion Israel Institute of Technology, in Haifa, has developed a device which, he claims, can do work that the human nose cannot.
The idea behind Dr Haick’s invention is not new. Many diagnostic “breathalysers” already exist, and sniffer dogs, too, can be trained to detect illnesses such as cancer. Most of these approaches, though, are disease-specific. Dr Haick wanted to generalise the process.
As he describes in ACS Nano, he and his colleagues created an array of electrodes made of carbon nanotubes (hollow, cylindrical sheets of carbon atoms) and tiny particles of gold. Each of these had one of 20 organic films laid over it. Each film was sensitive to one of a score of compounds known to be found on the breath of patients suffering from a range of 17 illnesses, including Parkinson’s disease, multiple sclerosis, bladder cancer, pulmonary hypertension and Crohn’s disease. When a film reacted, its electrical resistance changed in a predictable manner. The combined changes generated an electrical fingerprint that, the researchers hoped, would be diagnostic of the disease a patient was suffering from.
To test their invention, Dr Haick and his colleagues collected 2,808 breath samples from 1,404 patients who were suffering from at least one of the diseases they were looking at. Its success varied. It could distinguish between samples from patients suffering from gastric cancer and bladder cancer only 64% of the time. At distinguishing lung cancer from head and neck cancer it was, though, 100% successful. Overall, it got things right 86% of the time. Not perfect, then, but a useful aid to a doctor planning to conduct further investigations. And this is only a prototype. Tweaked, its success rate would be expected to improve.