S28 Paws for thought: sniffer dogs for infection surveillance in non-sputum producing people with CF

<h3>Background</h3> Prompt detection of <i>Pseudomonas aeruginosa</i> (Pa) may prevent chronic infection in CF but challenges exist. Cough swabs are insensitive compared with sputum; young patients and many adults on CFTR modulators cannot expectorate sputum; standard culture intervals may miss infection and costs preclude frequent surveillance. Medical Detection Dogs demonstrated excellent sensitivity and specificity in detecting Pa in culture broths (Davies JC; Eur Respir J, 2019. 54(5)). We are developing a method suitable for frequent home screening. Gauze loaded into a mouthpiece provides a large surface area onto which the subject ‘huffs’; this is then rapidly screened for Pa by trained dogs. <h3>Methods</h3> 10 clinical Pa &amp; 30 non-Pa (negative control) isolates were cultured in triplicate, adjusted by optical density and diluted to generate high, medium and low concentrations. 20 µl drops of these (~1,000,000, 1000 &amp; 10–100 CFU respectively) were placed on the gauze (n=30 Pa, n=90 neg) mimicking a ‘huff’. Sterile gauze (n=45) served as blanks. Gauze was incubated overnight in Pa-selective broth (cetrimide). Samples were presented to 3 trained dogs in randomised, double-blind study. Dogs indicate positive response by standing still; immediate trainer unblinding allowed rewards for correct responses. Data are means (95%CI). Clinical pilot recruited 20 adults with CF, 10 chronic Pa and 10 with other/no chronic infection, who can reliably produce sputum. Three huff samples (n=60) and one comparative sputum (n=20) are collected over one week. <h3>Results</h3> In the laboratory-model, dogs ignored 99.3 (95.9,100)% blanks and 90.0 (85.8,93.0)% negative controls. They correctly identified Pa in 100 (95.9,100)% high, 97.8(92.3,99.4)% medium and 63.3 (53.0,72.6)% low starting concentration samples. In initial clinical pilot, cetrimide proved insufficiently supportive of the very low bacterial numbers produced. However, subsequent laboratory-model trial of non-selective broth (Luria broth) showed persistence of high specificity/sensitivity. A repeat clinical pilot with Luria broth is underway. <h3>Discussion</h3> Trained dogs rapidly identify Pa from gauzes after overnight enrichment even from very low bacterial numbers, showing potential for large scale, home screening for Pa in non-sputum producers. Samples from current clinical pilot will be presented to the dogs in a randomized, double blind test and results presented at conference. Please refer to page A208 for declarations of interest related to this abstract.