Any disease that produces a scent is potentially suitable for detection using electronic-nose (e-nose) technology. Human metabolism produces myriad volatile organic compounds (VOCs), such as alcohols and ketones, some of which can be detected on the breath or skin, others are removed in waste products such as urine and faeces. When people have certain diseases, their metabolism changes, leading to the production of different VOCs. ‘It is this difference, either in the relative amounts expressed or the creation of disease-specific, novel compounds, that our systems detect,’ explains George Preti, organic chemist and member of the Monell Chemical Sensors Center in Philadelphia, Pennsylvania, US.
In general, e-noses have an array of up to 30 sensors that react with VOCs on contact and produce a physical change, which is recorded and converted to a digital signal. The computer combines the response from all the sensors and performs a ‘global fingerprint analysis’, which is then compared with reference patterns in a database. The process is similar to how the human nose works, which recognises an odour signature or pattern of VOCs, rather than the specific compounds.
‘The idea is like a fingerprint analysis,’ explains Graham Timmins, associate professor of medicinal chemistry at the University of New Mexico, US. ‘We don’t know the compounds but we can make a fingerprint pattern and match that against what’s going on in the patient.’ In this way, e-noses differ from mass spectrometers, which provide structural information on individual VOCs. At least this is the case currently, he notes, but technologies may become more specific with future advances.
The e-noses are small, fast and can be used at point-of-care, making them a potential game changer in diagnostic medicine. ‘E-noses could do to healthcare what blood analytics did a few decades ago,’ says Antti Roine, a doctor in the school of medicine at the University of Tampere, Finland, who has used e-noses to detect prostate cancer from urine samples. Moreover, the technology could be used for early diagnosis of any disease, as well as understanding disease progression and the effect of therapeutic interventions.
In cancer, such diagnostics would be hugely beneficial, particularly for those cancers with a poor prognosis that tend to be discovered late when early symptoms are lacking; ovarian cancer is a case in point. ‘Today’s tools for detecting ovarian cancer – pelvic exam, radioimmunoassay or ultrasound – are so invasive or expensive that they are limited to high-risk women,’ says Hossam Haick, head of the Laboratory for Nanomaterial-Based Devices at Technion – Israel Institute of Technology. ‘A non-invasive test, cheap enough to use for screening, would increase the chances to detect ovarian cancer at its earliest stages and therefore increase survival rates,’ he says.
Haick’s team has developed a sensor that comprises a flexible, thin film of gold nanoparticles that are linked to a variety of electron withdrawing and donating ligands. The film is placed on a flexible polyimide substrate, which sits between two electrodes.1
‘Ovarian cancer, it turns out,’ says Haick, ‘gives a mixture of styrene, nonanol, 2-ethylhexanol, 3-heptanone, decanal, and hexadecane in the patient’s breath.’ These VOCs interact with the ligands causing a measurable change in resistance, which is used to diagnose the disease. By bending the device, both large and small molecules can bind. ‘The measured changes in strain and resistance provide more data per sensor,’ Haick says. In tests from breath samples of 43 volunteers, 17 of whom had ovarian cancer, the e-nose diagnosed ovarian cancer correctly in 82% of the patients.
Meanwhile, Roine and his team have developed an e-nose to detect prostate cancer in urine sample.2 Using a sensor system based on ion mobility spectrometry, which ionises the sample gas, the team found the e-nose could discriminate between patients diagnosed with prostate cancer and those with benign prostatic hyperplasia, ie an enlarged prostate, with an accuracy of 78% and a specificity of 67% (the probability of the test being negative when cancer is absent). ‘The next step is to investigate whether the e-nose can discriminate between aggressive and non-aggressive tumours and thus decrease the need for biopsies,’ says Roine who hopes the device will be available for use in doctors’ offices within the next five to 10 years.
The e-nose technology has also been used successfully to detect lung and breast cancers but so far, like the other studies, all the results fall short of the 100% accuracy mark. Peter Sterk, professor of pathophysiology and phenotyping of asthma and COPD at the University of Amsterdam, The Netherlands, says this is something that needs to be improved if e-noses are to be used widely in clinical applications. ‘Sometimes 90% probability for establishing or excluding a diagnosis is ok, for example, in asthma. But for other diseases, such as lung cancer or TB, this has to be close to 100% in order to establish or rule out the disease,’ he explains. If e-noses are to be the first-line screening, thus avoiding complex, expensive and often invasive diagnostic procedures, this level of accuracy is necessary, he adds.
For the detection of colon cancer, the e-nose technology is a step closer to the clinic. In the UK, Cambridge University spin-out Owlstone has developed an e-nose for detecting colon cancer that the company says is virtually 100% accurate.3 The Owlstone e-nose was used in a pilot study to sniff out the differences in urine odours between 20 colon cancer patients, 20 with ulcerative colitis (UC) and seven healthy people. Results confirmed the sensor identified colon cancer patients as a distinct and tightly grouped set, compared with the UC patients and healthy subjects.
The researchers concluded the e-nose had a high potential to detect colon cancer. A larger trial with 133 subjects also showed the e-nose distinguished cancer patients from healthy people.4 Co-founder Billy Boyle aims to develop a small and cost-effective device within two years that can be placed in the hands of GPs.
The Owlstone sensor is also being trialled in the UK to detect lung cancer as part of the LuCID (Lung Cancer Indicator Detection) project, with funding from the NHS Small Business Research Initiative. Boyle believes the detection of early-stage lung cancer could be increased from the current 14.5% to 25% by 2020 and in the process save 10,000 lives and £245m for the NHS. If results, expected in 2016, are positive, the technology will be evaluated in hospitals and GP clinics. ‘For the first time there is the real potential for routine screening and improved rates of early diagnosis,’ Boyle says.
Meanwhile, Timmin’s research team is taking a different approach to the technology. ‘We’ve decided to look at stable isotopes of microbial pathways rather than volatiles,’ he explains.5 His team is looking to develop a speedy breath test for tuberculosis, which could be useful for developing countries and for determining drug resistance. In principle, people would inhale a tracer of 13C-urea, which is metabolised by bacteria into 13CO2 and excreted in the breath. The isotope ratios would then be measured, not by an e-nose but by infrared spectroscopy, with results in as little as 10–15 minutes. Moreover, this technology samples the whole lung, unlike current sputum tests, which only detect a sample of the possible infected area.
Timmins says there are many advantages to this method. ‘We can look at the metabolism of drugs and measure the treatment response; we can measure the amount of bacteria based on the signal we receive; we can probe the chemistry specifically and we can even identify the species of microbes in some cases. It’s like a thermometer to measure bad bacteria.’ The only downside, Timmins notes, is the regulatory aspect because patients would have to inhale a chemical but he hopes his breath test will be at least complementary to e-noses.
Improving sensor arrays
A German team is focusing its efforts on improving the sensor arrays used in e-noses. Martin Strauch, a postdoctoral researcher at the RWTH (Rheinisch-Westfälische Technische Hochschule) Aachen University in Germany, is studying the antenna of a fruit fly as a potential chemosensor that he says would distinguish cancer cells from healthy cells just by their smell.6
Strauch explains that animals smell odours with the help of hundreds of olfactory receptor neurons (ORNs), which detect the presence of chemical features in an odour molecule. ‘In an abstract sense, the ORNs are equivalent to the photoreceptor cells on the retina of the eye that detect light,’ he says. He believes a fruit fly’s ORNs, situated in the antenna, have a good coverage of the general odour space and could be the basis of an artificial chemosensor or e-nose. ‘Evolution,’ he says, ‘has led to very good natural sensor arrays that are better than the electronic sensors used in standard e-noses.’
Strauch and his collaborators used in vivo calcium imaging to measure the activity of the neurons. With computer analysis they found that cell volatiles from different cell types lead to different responses in the ORNs, thereby discriminating healthy from cancerous cells.
Further research showed the ORNs could differentiate between five types of breast cancer. ‘Future work is targeted at bringing the system closer to clinical applications,’ Strauch says. ‘Eventually – but this is a long way off – we would have an electronic device that contains biological sensors borrowed from an animal. Thus we would have individual neurons or receptor proteins integrated into the device instead of using a living fruit fly.’
Finally, in October 2015, Parkinson’s UK announced a trial to identify the molecules on skin that appear to give Parkinson’s disease patients a subtle but distinctive musty smell. The charity is funding researchers in the universities of Manchester, Edinburgh and London in the UK, who will use mass spectrometry to look for the molecular signature believed to be present in the oily skin secretion of Parkinson’s patients. ‘Once we have identified the biomarkers we may develop sensors or more portable mass spectrometry devices for targeted analysis,’ says Perdita Barran, professor of mass spectrometry at the University of Manchester, UK, noting that a test for Parkinson’s would be revolutionary. ‘We have certainly thought about an e-nose but we are not yet at that stage of the research.’
With developments continuing apace, the sector is certainly one to watch. Limitations still exist, such as standardising the technologies and ensuring cost effectiveness, as well as accuracy issues, but scientists are hopeful e-noses will become commonplace diagnostics for a range of diseases.
‘I am certain that by the year 2030, e-nose technology will be widespread with daily clinical use,’ says Roine.
And with e-nose smartphone accessories already in development, he says that by 2030 we might all carry such a device for personal use.
1 N. Kahn et al, Nano Lett., 2015, 15(10) 7023
2 A. Roine et al, J. Urology, 2014, 192, 230
3 J. Covington et al, Gut, 2013, 62, A218
4 R. Arasarad et al, PLoS One, 2014, 9(9), e108750
5 M. Maiga et al, mBio., 2014, 5(2), e00990-1.4
6 M. Strauch et al, Nature Scientific Reports, 2014, 4, 3576
Katrina Megget is a science and healthcare writer based in London, UK