Microbial and non-microbial volatile fingerprints: Potential clinical applications of electronic nose for early diagnoses and detection of diseases

Date published

2008-01

Free to read from

Supervisor/s

Journal Title

Journal ISSN

Volume Title

Publisher

Cranfield University

Department

Type

Thesis or dissertation

ISSN

Format

Citation

Abstract

This is the first study to explore the potential applications of using qualitative volatile fingerprints (electronic nose) for early detection and diagnosis of diseases such as dermatophytosis, ventilator associated pneumonia and upper gastrointestinal cancer. The investigations included in vitro analysis of various dermatophyte species and strains, antifungal screening, bacterial cultures and associated clinical specimens and oesophageal cell lines. Mass spectrometric analyses were attempted to identify possible markers. The studies that involved e-nose comparisons indicated that the conducting polymer system was unable to differentiate between any of the treatments over the experimental period (120 hours). Metal oxide-based sensor arrays were better suited and differentiated between four dermatophyte species within 96 hours of growth using principal component analysis and cluster analysis (Euclidean distance and Ward’s linkage) based on their volatile profile patterns. Studies on the sensitivity of detection showed that for Trichophyton mentagrophytes and T. rubrum it was possible to differentiate between log3, log5 and log7 inoculum levels within 96 hours. The probabilistic neural network model had a high prediction accuracy of 88 to 96% depending on the number of sensors used. Temporal volatile production patterns studied at a species level for a Microsporum species, two Trichophyton species and at a strain level for the two Trichophyton species; showed possible discrimination between the species from controls after 120 hours. The predictive neural network model misclassified only one sample. Data analysis also indicated probable differentiation between the strains of T. rubrum while strains of T. mentagrophytes clustered together showing good similarity between them. Antifungal treatments with itraconazole on T. mentagrophytes and T. rubrum showed that the e-nose could differentiate between untreated fungal species from the treated fungal species at both temperatures (25 and 30°C). However, the different antifungal concentrations of 50% fungal inhibition and 2 ppm could not be separated from each other or the controls based on their volatiles. Headspace analysis of bacterial cultures in vitro indicated that the e-nose could differentiate between the microbial species and controls in 83% of samples (n=98) based on a four group model (gram-positive, gram-negative, fungi and no growth). Volatile fingerprint analysis of the bronchoalveolar lavage fluid accurately separated growth and no growth in 81% of samples (n=52); however only 63% classification accuracy was achieved with a four group model. 12/31 samples were classified as infected by the e-nose but had no microbiological growth, further analysis suggested that the traditional clinical pulmonary infection score (CPIS) system correlated with the e-nose prediction of infection in 68% of samples (n=31). No clear distinction was observed between various human cell lines (oesophageal and colorectal) based on volatile fingerprints within one to four hours of incubation, although they were clearly separate from the blank media. However, after 24 hours one of the cell lines could be clearly differentiated from the others and the controls. The different gastrointestinal pathologies (forming the clinical samples) did not show any specific pattern and thus could not be distinguished. Mass spectrometric analysis did not detect distinct markers within the fungal and cell line samples, but potential identifiers in the fungal species such as 3-Octanone, 1-Octen-3-ol and methoxybenzene including high concentration of ammonia, the latter mostly in T. mentagrophytes, followed by T. rubrum and Microsporum canis, were found. These detailed studies suggest that the approach of qualitative volatile fingerprinting shows promise for use in clinical settings, enabling rapid detection/diagnoses of diseases thus eventually reducing the time to treatment significantly.

Description

Software Description

Software Language

Github

Keywords

DOI

Rights

© Cranfield University 2008. All rights reserved. No part of this publication may be reproduced without the written permission of the copyright owner.

Relationships

Relationships

Supplements

Funder/s