Identification of Pseudomonas aeruginosa and airway bacterial colonization by an electronic nose in bronchiectasis

• Published in: Respiratory medicine 2018-03, Vol.136, p.111-117
• Main Authors: Suarez-Cuartin, Guillermo, Giner, Jordi, Merino, José Luis, Rodrigo-Troyano, Ana, Feliu, Anna, Perea, Lidia, Sanchez-Reus, Ferran, Castillo, Diego, Plaza, Vicente, Chalmers, James D., Sibila, Oriol
• Format: Article
• Language: English
• Subjects: Aged; Analysis of Variance; Antibiotics; Asthma; Bacteria; Bacterial infections; Bacteriology; Bronchi - microbiology; Bronchiectasis; Bronchiectasis - microbiology; Bronchiectasis - physiopathology; Chronic obstructive pulmonary disease; Colonization; Cross-Sectional Studies; Cystic fibrosis; Discriminant analysis; Electronic Nose; Electronic noses; Female; Forced Expiratory Volume - physiology; Humans; Male; Microorganisms; Middle Aged; Mortality; Organic compounds; Patients; Principal components analysis; Pseudomonas aeruginosa; Pseudomonas aeruginosa - isolation & purification; Pseudomonas Infections - diagnosis; Pseudomonas Infections - physiopathology; Respiratory diseases; Respiratory tract; Software; Sputum; Streptococcus infections; Variance analysis; Vital Capacity - physiology; VOCs; Volatile organic compounds
• ISSN: 0954-6111; 1532-3064
• DOI: 10.1016/j.rmed.2018.02.008

Airway colonization by Potentially Pathogenic Microorganisms (PPM) in bronchiectasis is associated with worse clinical outcomes. The electronic nose is a non-invasive technology capable of distinguishing volatile organic compounds (VOC) in exhaled breath. We aim to explore if an electronic nose can reliably discriminate airway bacterial colonization in patients with bronchiectasis. Seventy-three clinically stable bronchiectasis patients were included. PPM presence was determined using sputum culture. Exhaled breath was collected in Tedlar bags and VOC breath-prints were detected by the electronic nose Cyranose 320®. Raw data was reduced to three factors with principal component analysis. Univariate ANOVA followed by post-hoc least significant difference test was performed with these factors. Patients were then classified using linear canonical discriminant analysis. Cross-validation accuracy values were defined by the percentage of correctly classified patients. Forty-one (56%) patients were colonized with PPM. Pseudomonas aeruginosa (n = 27, 66%) and Haemophilus influenzae (n = 7, 17%) were the most common PPM. VOC breath-prints from colonized and non-colonized patients were significantly different (accuracy of 72%, AUROC 0.75, p