"DETECTION OF BACTERIAL INFECTION"
FIELD OF THE INVENTION
This invention relates to method and apparatus for detection of a bacterial infection in mammals, particularly but not exclusively humans. BACKGROUND OF THE INVENTION
Detection of mammalian bacterial infection, particularly at a stage when symptoms are not readily apparent by simple means, is highly desirable but not always easy. In particular detection techniques which do not require skilled operators and/or complex equipment, and can provide results in minimal time, are sought.
For example, Helicobacter pylori is a major cause of peptic ulceration and is associated with the development of gastric cancer; the early detection and eradication of this bacteria is of prime importance. Currently detection of H pylori is via the 13C or 14C labelled urea breath test (13C UBT)1. The 13C UBT involves collection of breath samples before and after administering 13C labelled urea and determination of the ratio of 13C present in the two samples. H pylori , if present, is resident in the mucosal layer of the stomach lining and produces urease as a self defence mechanism against the low pH of the stomach.
Urease converts urea to ammonium and bicarbonate ions. The bicarbonate results in exhalation of C02. Hence a ratio 13C02Post-urea/13C02Pre-urea >3.5 is taken as an indication of the presence of H pylori . Determination of the isotope mass ratio requires expensive equipment and extensive sample pre-treatment to remove oxides of nitrogen by gas chromatography prior to isotope ratio mass spectrometry. Also the technique, due to the nature of the analysis and equipment, has typically a delay of up to two weeks before a result is available.
It has long been known that the condition of human breath is indicative of the presence of certain disorders, including some diseases caused by bacterial infection. However, it does not appear to have been noted that a non-pulmonary bacterial infection as such, for example in the gastrointestinal tract, generates a specific pattern of volatile organic compounds (VOCs) in the breath.
It has been reported2 that analysis of eleven VOCs in breath enabled identification of patients with schizophrenia with high confidence. The VOCs were trapped by adsorption and then assayed by gas chromatography/mass spectrometry. The same technique has been suggested3 as a route for the possible detection of several common disorders, including pneumonia, although no specific correlation of a VOC with a disease is
mentioned. Detection of pulmonary infection by an "electronic nose" using semiconducting polymer elements has also been suggested7.
Techniques are under development for the detection and recognition of micro-organisms outside the human body, from the VOCs present in their headspace . VOCs arising from microbial cultures on building materials have been studied, using HPLC and GC-MS4. Spoilage fungi have been distinguished from each other by detection of the VOCs which they produced, using a commercial
"electronic nose unit" consisting of 14 polymer sensors and mathematical analysis of the data by principal component analysis and discriminant function analysis5.
In connection with the disclosure of the invention below, it should also be mentioned here that quartz crystals were first described as single element mass sensitive sorption detectors for single analytes and used in applications such as relative humidity sensing. More recently array based gas sensing has utilised data acquisitions systems and mathematical algorithms such as principal component analysis and discriminant function analysis which have been applied to complex multivariate systems .
Arrays of quartz crystal microbalances (QCMBs) have been used to determine chemical composition and concentration of gases and vapours of complex mixtures of
volatile organic compounds and have been used to distinguish between enantiomeric forms of various monoterpenes . SUMMARY OF THE INVENTION The object of the invention, in the light of the above discussion, is to provide improved method and apparatus for detection of non-pulmonary bacterial infection, which can be used simply and quickly.
According to this invention in one aspect, there is provided a method of diagnosis of a non-pulmonary bacterial infection of a mammal, including the steps of i) performing an analysis of a plurality of volatile organic compounds in exhaled breath of the mammal , and ii) determining from the analysis result whether or not the bacterial infection is present. In particular, step (ii) is performed by comparing the analysis result with a predetermined pattern of occurrence of volatile organic compounds in breath indicative of the presence of the bacterial infection.
It has surprisingly been found that a non-pulmonary bacterial infection, i.e. an infection which is not in the respiratory system, can be detected from the pattern of VOCs in the exhaled breath, particularly in a case of infection of the gastro-intestinal system. It seems likely that this is direct detection of the bacterial
infection, though it cannot be ruled out that the VOC pattern results from a disorder caused by the infection. On this basis it is reasonably predictable that many non- pulmonary bacterial infections give rise to characteristic and detectable VOC patterns in exhaled breath. Different bacteria will in general give different patterns, enabling not only detection of a bacterial infection, but also possibly distinguishing between them. The exhaled breath is typically breath from the lungs, exhaled during normal breathing. The invention is particularly applicable to detection of gastro-intestinal bacterial infections, i.e. gastric infection or intestinal infection, but is not limited thereto. Since in a healthy mammal there are normally bacteria present in the gastro-intestinal system, the method of the present invention is detecting abnormal levels of bacteria, i.e. an infection which is deleterious to health.
Preferably the analysis of step (i) is performed using an array of sensors which are sensitive to volatile organic compounds and have mutually different sensing characteristics, and obtaining an analysis result characteristic of the analysed breath by means of a multivariate system mathematical algorithm, such as principal component analysis and discriminant functional analysis, which are known and readily available
techniques. Suitable sensors are oscillating quartz crystal microbalances (QCMBs) . Such sensor arrays are preferably of the type in which each sensor has a surface which is contacted with the breath sample, the surfaces of the respective sensors having different chemical interactions with volatile organic compounds. These different molecular interactions may be selected from dispersion interactions, interactions due to polarizability, dipolar interactions,
H-bond interactions (e.g. acid and base) . Any suitable materials providing these interactions may be employed.
In a second aspect, the invention consists in sensor apparatus for diagnosis of a non-pulmonary bacterial infection of a mammal, comprising:
(a) sensor means arranged and adapted to perform an analysis of a plurality of volatile organic compounds in a sample of exhaled breath, and (b) a comparator arranged and adapted to compare an analysis result from said sensor means with a predetermined pattern of occurrence of volatile organic compounds in breath indicative of the presence of the non-pulmonary bacterial infection. The present inventors have found that an array of quartz crystal microbalance sensors is especially
suitable for the analysis of breath samples as here described, because of the ability to load them with a range of compounds providing a suitable range of interactions and because their high suitability (avoidance of drift) allows repeatable analyses and calibration.
The breath sample may be captured as described below, or using a standard manufactured breath sampler, e.g. such as that supplied by the UK Government Health and Safety Executive.
It is to be noted that the present inventors have used breath samples which emanate from the lungs, and avoided, as far as possible, samples of exhaled breath which included gas ejected directly from the stomach (i.e. "burp" gas) .
BRIEF INTRODUCTION OF THE DRAWINGS
Fig. 1 is a graph plotting results of principal component analysis in a test of the apparatus described below. DESCRIPTION OF THE EMBODIMENT
An embodiment of the invention will now be described, in which the method of the invention was tested alongside 13C UBT in order to assess its accuracy.
Quartz Crystals
Eight AT-cut crystals oscillating at approximately
30 MHz were each spray-coated with a different chemical The coating materials (Table 1) were selected based on their diverse chemical nature and range of reversible chemical interactions.
Table 1 : Coating compound, coating frequency (i . e . the measured difference in frequency before and after coating) and coating/analyte molecular interaction .
These quartz crystals were arranged as a QCMB array, for testing of breath samples . Array sensors of this general type are now available from Quartz Technology Ltd, London and are described in reference 6 .
Experimental Protocol
In accordance with standard breath test protocols, patients fasted for a minimum of 4 hours, avoided proton pump inhibitors for 28 days and H2 receptor antagonists for 7 days before test. In addition to the samples collected for isotope mass ratio determination, two 7ml breath samples were collected in vacutainer tubes from 23 patients attending a 13C UBT clinic, by exhaling into a seven ml glass tube (a vacutainer) via a straw, followed by sealing with a septum. The sample was later removed by piercing the septum with a syringe needle. The first sample was taken prior to administering the test meal and urea dose, while the second breath sample was taken 40 minutes later. Both breath samples for QCMB analysis were taken concurrently with the 13C breath sampling, and QCMB analysis was carried immediately. Sample and purge times for the pumped QCMB array were both 70 seconds.
Results In order to validate the principle of operation water, methanol, propan-1-ol and propan-2-ol were passed over the same QCMB array twice and the frequency shifts obtained after 10 seconds were examined using principal component analysis (PCA) . PCA is an unsupervised mathematical algorithm that maximises the variance within the data with no prior information. The first and second
principal components are shown in attached Fig. 1, which shows the principal component analysis of frequency responses from the 8 coated QCMBs after ten seconds exposure to water (unfilled circles) , propan-1-ol (triangles) , propan-2-ol (filled circles) and methanol (diamonds) . Repeat values are also shown for each compound .
It is apparent from Fig. 1 that discrimination between the four chemicals is possible. Even the structural isomers, propan-1-ol and propan-2-ol are readily distinguished.
The reduction in frequency of each of the 8 coated quartz crystals on exposure to the breath sample was taken after 30 seconds and discriminant function analysis (DFA) used to maximise the variance within the data according to the patients of known Helicobacter pylori status, as shown by the 13C UBT. Hence DFA is described as a supervised technique.
DFA allows multidimensional data to be represented in a reduced, two-dimensional format and for the axes to be set such that variance is maximised according to a predetermined property, in this case Helicobacter pylori colonisation. Significance testing of the null hypothesis; Ho: There is no significant difference (P=0. 05) between the two groups of Helicobacter pylori
posi tive and negative patients (n=46) based on the results obtained show that there is a significant difference between the two groups, the mean values and standard deviations being 0.26 and 0.134 respectively for the positive group and 0.13 and 0.137 respectively for the negative group.
The technique of array-based gas sensing has thus been applied to discrimination of patients with and without Helicobacter pylori colonisation and the results show that there is a significant difference (P=0.05) between the two groups. The relative distance between the mean value of the two populations when considered with respect to the magnitude of the standard deviations is such that at present the technique can be considered in statistical terms as accurate but imprecise. To enhance precision, that is to reduce the standard deviation, there are proposed increased coating film thickness, calibration of the array prior to sample introduction and improved temperature control of both sample and crystal array.
Calibration enables chemical and viscoelastic changes within the polymer coating materials to be corrected for.
Temperature control of the sample and QCMB array reduces three potential sources of error; (i) the vapour pressure and subsequent vapour phase concentration
effects, that are associated with humid alveolar breath, (ii) the viscoelastic changes associated with the coating materials and resultant changes in diffusion rates, and (iii) and the temperature frequency profile of the AT-cut bulk acoustic wave device.
Particularly, calibration and thus comparison, e.g. by data handling means, of a test result on a section sample with one or more predetermined and stored VOC patterns indicative of specific non-pulmonary bacterial infections will permit rapid and simple diagnosis of the presence of such infections. The task of analyzing the data from the sensor and comparing it with a predetermined VOC pattern is readily performed by a computer . The programming and running of such a procedure does not need description here, requiring only ordinary skill and knowledge.
Real time on-line diagnosis of the presence of gastrointestinal H pylori without the need for expensive equipment or trained personnel is thus possible.
References :
1. J. E. Dominguez-Munoz; A. Leodolter; T. Sauerbruch; P. Malfertheiner .
(1997) A citric acid solution is an optimal test drink in the 13C-urea breath test for the diagnosis of Helicobacter pylori infection. Gut. 40. 459-62
2. M. Phillips; G.A. Erickson; M. Sabas; J.P. Smith; J. Greenberg.
(1994) Volatile organic compounds in the breath of patients with schizophrenia. J. Clin. Pathol.48/ 466-9
3. M. Phillips.
(1997) Method for the collection and assay of volatile organic compounds in breath. Analytical Biochemistry 247, 272-278
4. A. Korgi; A-L. Pasanen; P. Pasanen.
(1988) Volatile compounds originating from mixed microbial cultures on building materials under various humidity conditions. Applied and Environmental Microbiology 64, 2914- 2919.
5. G. Keshri; N. Magan; and P. Voysey. Letters in Applied Microbiology, 1998, 27, 261-264. Use of an electronic nose for the early detection and differentiation between spoilage fungi
6. K-L Lau; J. Micklefield; J.M. Slater. The
optimisation of sorption sensor arrays for use in ambient conditions.
Sensors and Actuators, B, Chemical, 50., (1998) , 68 79. C.W. Hanson III; H.A. Steinberger. The use of a novel electronic nose to diagnose the presence of intrapulmonary infection. ASA abstract A269, Anesthesiology, Sept. 1997.