Introduction
Since the time of the ancient Greeks, physicians have been aware that it is possible to smell disease on the breath of the afflicted. An obvious example of this is the pear-drop-like smell of acetone on the breath of a diabetic. The human nose can be highly trained as an analytical tool for the detection and discrimination of food odours (e.g. fish, meat, cheese and wine) and perfumes (e.g. in cosmetics and toiletries). However, there may be issues regarding health and safety, and also regarding inter-observer variability and sensitivity over time.Reference Nagle, Gutierrez-Osuna and Schiffman1 To overcome these difficulties, sensor arrays have been used to develop electronic nose (e-nose) technology which can mimic human olfactory discrimination, based on volatile substance production patterns.
Since the 1960s, attempts have been made to build an artificial olfactory model which can mimic the mammalian olfactory sense in detecting and discriminating between volatile compounds. Early instruments detected odours using mechanical noses based on redox reactions. The 1980s saw the early development of e-nose technology. The currently accepted definition of an electronic nose was proposed by Gardner and Bartlett in 1994, as follows.
An electronic nose is an instrument which comprises an array of electronic chemical sensors with partial sensitivity and an appropriate pattern recognition system, capable of recognising simple or complex odours.Reference Gardner and Bartlett2
The odorant molecules which interact with the sensor array surfaces in the e-nose (and in the olfactory system) are typically hydrophobic and volatile due to their molecular weight (30–300 Da) and to the strength of the interactions between the molecule's components and its shape.
Volatile organic compounds are organic compounds that can easily become gases or vapours; as such, they can be detected in breath and bodily fluids. Many studies have shown that so-called ‘volatile fingerprints’ (i.e. specific patterns of volatile organic compounds) can be detected for certain pathological conditions, such as infection, malignancy, and liver and cardiopulmonary disease. There is thus the potential to utilise this approach to improve the accuracy of disease diagnosis.
An example of an e-nose device is shown in Figure 1. Such devices mimic the olfactory sense of mammals. In humans, smells are detected by olfactory sensory neurons in the olfactory epithelium. Volatile molecules bind to olfactory neuron cell receptors, producing a change in conformation which induces signal transduction to the brain. In an e-nose device, the active materials of the sensor array (equivalent to the olfactory receptors in a mammalian nose) interact with volatile organic compounds, resulting in a change in the binding surface of each sensor (i.e. a change in resistance or conductance) which is transmitted to the pre-processor (equivalent to the olfactory bulb), where the output is amplified and signal noise reduced. In the final stage, the simplified signals (equivalent to nerve impulses) are processed as patterns of response by the data analysis system (equivalent to the hypothalamus and olfactory cortex in the brain).
An NST 320 electronic nose device (Applied Sensors, Linköping, Sweden), used for clinical studies.
This type of technology can be used to develop novel rapid diagnostic tools and may also improve therapeutic drug selection.Reference Persaud and Dodd3 The ultimate goal of e-nose technology evolution is the development of clinical devices which are portable and inexpensive and which enable rapid bedside diagnosis. Such devices could be widely deployed.
The cost of such devices can vary from approximately £7500–10 000 for a hand-held device to £15 000–30 000 for more complex, centralised devices which include an autosampler. However, it may be possible to improve the cost–benefit ratio by using simple, portable devices to make measurements at the bedside, and then sending the collected information remotely to a central processing facility, which could undertake both patient analysis and also data processing (using appropriate bioinformatics software programs specific to the relevant disease application).
Otolaryngology has probably the highest number of patients with the commonest range of symptoms in the field of medicine (otitis externa, rhinorrhoea, discharging ears etc). Electronic nose technology offers many potential benefits to the speciality, in the form of non-invasive, rapid and relatively portable diagnostic tools. This review will demonstrate that preliminary applications of e-nose technology have already demonstrated successful results. We believe that such applications could have significant benefit for the modern ENT department.
One area of particular interest is the diagnosis of biofilms. Two very extensive and interesting reviews have recently highlighted the importance of biofilms in otolaryngology.Reference Vlastarakos, Nikolopoulos, Maragoudakis, Tzagaroulakis and Ferekidis4, Reference Macassey and Dawes5
Search strategy
In order to establish the current evidence for the role of e-nose technology in otolaryngology, a literature review was undertaken. The Medline (1952–2009), Embase (1974–2009), Cinahl (1937–2009) and Cochrane Library (1996–2009) electronic databases were searched for relevant studies, references and reviews. Initially, a review of all applications of e-nose technology was undertaken; we then focused on clinical applications. Finally, combinations of the following keywords were used to identify applications relevant to otolaryngology: ‘e-nose’, ‘electronic nose’, ‘infection’, ‘otolaryngology’, ‘ear nose and throat’, ‘ENT’ and ‘biofilms’.
Four studies matched our criteria for the use of e-nose technology in ENT infections; these are detailed in Table I. One of these studies demonstrated the benefit of e-nose technology in detecting biofilms.
published studies of E-nose technology applications in otolaryngology
*Compared with microbiological culture ‘gold standard’. E-nose = electronic nose; pts = patients; CRS = chronic rhinosinusitis; ?? = bacterial classification
Applications of electronic nose technology
Non-medical applications
Electronic nose technology has been widely employed outside the medical field. Much work has been done in the food industry on detecting early microbial spoilage and non-microbial tainting of food. Applications have ranged widely, from monitoring the freshness of stored fishReference Du, Kim, Cornell, Huang, Marshall and Wei10 and meat, through to the classification of coffee, beer and wine.Reference Di Natale, Macagnano, Davide, D'Amico, Paolesse and Boschi11 There is significant interest in detecting the level of fungal mycotoxins in many food products. Thus, diverse studies have been conducted to differentiate between mycotoxigenic and non-mycotoxigenic strains of fusarium, aspergillus and penicillium in raw food materials.Reference Keshri and Magan12, Reference Sahgal, Needham, Cabañes and Magan13
Electronic nose technology has also been employed to detect mould growth on library books.Reference Canhoto, Pinzari, Fanelli and Magan14
Clinical applications
Pilot studies have assessed the use of e-nose technology for the diagnosis of tuberculosis,Reference Pavlou, Magan, Meecham Jones, Brown, Klaster and Turner15 Helicobacter pylori,Reference Pavlou, Magan, Turner, Sharp, Brown and Barr16 renal failure,Reference Montag, Frank, Ulmer, Wernet, Gopel and Rammensee17 urinary tract infection,Reference Pavlou, Magan, McNulty, Meecham Jones, Sharp and Brown18 fungal skin infection,Reference Sahgal, Monk, Wasil and Magan19 ventilator-acquired pneumonia,Reference Hockstein and Thaler20, Reference Adrie and Monchi21 cancer (lungReference Phillips and Gleeson22, Reference Di Natale and Macagnano23 and breastReference Phillips and Cataneo24) and acute asthma.Reference Dragonieri, Schot, Mertens, Le Cessie, Gauw and Spanevello25 Each of these studies demonstrated promising results for the use of e-nose technology as an adjuvant method of disease detection, offering the potential for a rapid and non-invasive diagnostic method currently unavailable to the clinician. Electronic nose technology has also been explored as a potential rapid method for screening the efficacy of antifungal drugs in cases of skin infection.Reference Naraghi, Sahgal, Adriaans, Barr and Magan26
A number of reviews of the clinical applications of e-nose technology have also been published.Reference Turner and Magan27–Reference Thaler, Kennedy and Hanson29
Despite the wide-ranging clinical applications of e-nose technology published in the literature, we identified only four studies matching our criteria for the use of e-nose technology in ENT infections, one of which demonstrated benefits in biofilm detection.
Electronic nose technology and otolaryngology
Current applications
Many recent clinical applications of e-nose technology have focused on the field of otolaryngology. Various studies have demonstrated evidence for the effectiveness of e-nose diagnosis in common otolaryngological conditions such as ENT infectionReference Shykhon, Morgan, Dutta, Hines and Gardner6 and acute and chronic sinusitis.Reference Thaler and Hanson7, Reference Bruno, Alessandrini, Ottaviani, Delfini, Pierro and Camillo8
Shykhon et al. Reference Shykhon, Morgan, Dutta, Hines and Gardner6 used a commercial, portable e-nose device (Cyranose C320), and recruited 90 patients with various types of ENT pathology including otitis externa, chronic suppurative otitis media and nasal vestibulitis. Swabs were taken and analysed using the e-nose device, before microbiological culture. The e-nose had a sensitivity of 88.2 per cent for the discrimination of controls, Pseudomonas aeruginosa, staphylococcus aureus and mixed skin organism compared with the gold standard of microbiological culture.
In 2006, Thaler and HansonReference Thaler and Hanson7 showed the potential of an e-nose approach to the diagnosis of bacterial sinusitis. They used a nasal continuous positive airway pressure mask to sample gas exhaled through the nose of patients with sinusitis, and compared this with controls. Patients were considered infected if there was positive growth from a swab sent for microbiological culture. These authors were able to correctly predict the diagnosis of sinusitis in at least 72 per cent of samples, and to correctly identify infected versus non-infected patients in 82 per cent of cases.
Bruno et al. Reference Bruno, Alessandrini, Ottaviani, Delfini, Pierro and Camillo8 demonstrated diagnostic use of an e-nose device in patients with chronic rhinosinusitis. They performed a controlled study comparing chronic rhinosinusitis patients with healthy subjects. Electronic nose technology was able to correctly identify pathological bacteria in up to 85 per cent of patients with chronic rhinosinusitis.
One of the most exciting recent developments has been Thaler and colleagues'Reference Thaler, Huang, Giebeig, Palmer, Lee and Hanson9 work demonstrating successful e-nose identification of biofilm-producing versus non-biofilm-producing bacterial strains. The e-nose correctly identified biofilm-producing and non-biofilm-producing mutant strains of pseudomonas and staphylococcus species, with an accuracy of 72.2 to 100 per cent.
This latter work has opened a new chapter in the clinical application of e-nose technology. Recent data has linked biofilms to common types of antibiotic-resistant ENT infection. Biofilms are complex aggregations of micro-organisms which have been implicated in many cases of chronic antibiotic-resistant otolaryngological infection.Reference Palmer30 Biofilms have been found in cases of cholesteatoma,Reference Chloe and Faddis31 tympanostomy tube use,Reference Mehta, Lee, Stevens and Antonelli32 chronic tonsillitis,Reference Chloe and Faddis33 otitis media with effusion,Reference Ehrlich, Veeh, Wang, Costerton, Hayes and Hu34 speech valve use,Reference Everaert, Mahieu, Chung, Verkerke, van der Mei and Busscher35, Reference Leunisse, van Weissenbruch, Busscher, van der Mei, Dijk and Albers36 cochlear implantationReference Antonelli, Lee and Burne37, Reference Pawlowski, Wawro and Roland38 and sinusitis.Reference Palmer30, Reference Sanderson, Leid and Hunsaker39–Reference Perloff and Palmer42 Biofilms are very difficult to eradicate because they are enclosed within a matrix that can restrict the diffusion of substances and bind antimicrobials. This confers effective resistance to biofilm cells against large molecules, especially anti-microbial proteins such as lysozyme and complement.Reference Ishida, Ishida, Kurosaka, Otani, Sato and Kobayashi43 Therefore, a non-invasive, rapid method of diagnosing biofilms in the clinic would be of great benefit. Further investigation of the application of e-nose technology in this field is warranted.
Proposed research avenues
One potential future application of e-nose technology is in the field of rhinosinusitis. This condition is common, affecting around 15 per cent of the population, and causes a significant reduction in patients' quality of life.Reference Scadding, Durham, Mirakian, Jones, Drake-Lee and Ryan44 Despite surgical and medical developments, it remains a significant clinical problem; sinusitis is a major reason for antibiotic prescriptions and lost workplace productivity.Reference Gliklich and Metson45 (In North America, a conservative estimate of the cost of decreased productivity due to allergic rhinitis or its treatment is four billion dollars per year.)Reference Fireman46 The latest ENT-UK research suggests that 31 per cent of patients attending the out-patient department of a general ENT clinic have nasal symptoms.Reference Kinshuck and Lesser47 As mentioned above, biofilms have recently been implicated in the pathogenesis of recalcitrant chronic rhinosinusitis;Reference Prince, Steiger, Khalid, Dogrhamji, Reger and Eau Claire48, Reference Palmer49 furthermore, e-nose technology has been successfully used to differentiate between biofilm-producing and non-biofilm-producing bacterial strains.Reference Bruno, Alessandrini, Ottaviani, Delfini, Pierro and Camillo8 Therefore, studies are warranted to evaluate the performance of e-nose technology as an adjuvant tool in the investigation of chronic rhinosinusitis. One possibility is the use of a thin probe to collect deep nasal cavity samples during flexible nasendoscopy, which could be analysed by e-nose technology to facilitate early, targeted antibiotic therapy, with the aim of improving symptoms and reducing the need for surgical intervention (with its associated possible complications).
Another potential area to be explored is the potential benefit of e-nose diagnosis of chronic suppurative otitis media (CSOM), one of the most common childhood infections and associated with significant morbidity.Reference Verhoeff, van der Veen, Rovers, Sanders and Schilder50 A recent study found that biofilms were present in 60 per cent of paediatric cases of CSOM.Reference Lee, Pawlowski, Luong, Furze and Roland51 Electronic nose technology could potentially be used to detect pathogens in ear discharge samples, enabling early identification of biofilms and, hopefully, improved clinical outcomes and fewer surgical interventions.
Discussion and conclusions
Electronic nose technology has great potential for application in the field of disease diagnosis, due to its rapidity, simplicity, non-invasive nature and great sensitivity (as demonstrated in pilot studies). This technology has particular potential for the identification of patients with more virulent and/or medically recalcitrant forms of chronic otolaryngological infection. Especially exciting is the recent report that e-nose technology can differentiate between biofilm-producing and non-biofilm-producing bacterial species. In the field of otolaryngology, e-nose technology has many potential applications. Many patients seen by otolaryngologists have very common conditions which are difficult to differentiate in the out-patient setting, but which may possibly be identified by e-nose technology, enabling earlier, better targeted treatment.
Electronic nose technology offers the potential for development of rapid, bedside testing methods enabling the identification of causative microbes and thus facilitating immediate prescription of targeted antibiotics. Multi-centre clinical studies are required to validate the results of pilot studies already undertaken, and to further evaluate the potential of this exciting technology.
Acknowledgements
Dr Catherine Kendall is a Royal Society Dorothy Hodgkin Research Fellow and gratefully acknowledges their support. Professor Hugh Barr has been an inspiration to all involved in this work.
