A-270 BreathBiomics: an exhaled breath-based noninvasive diagnostic device for respiratory tract infections

Background Accurate diagnosis of lower respiratory tract infections (LRTIs) in critical care settings is crucial for effectively administering antibiotics and mitigating mortality rates. Current diagnostic methodologies using clinical manifestations and microbiological cultures show poor sensitivity and specificity. Molecular diagnostic tools, such as nucleic acid amplification tests, fail to differentiate bacterial colonization from an actual infection, wherein identical test outcomes may suggest disparate clinical trajectories. To address this limitation, host proteases, particularly those derived from neutrophils, have been investigated due to the substantial evidence showing their dysregulation during LRTIs. Previously, we identified neutrophil proteases in exhaled breath samples from intubated patients with mechanical ventilation. This study investigated whether detecting protease activity in breath could be developed into an in vitro diagnostic approach for LRTIs. Methods We designed a substrate sensor targeting human neutrophil elastase (HNE). Breath samples were collected from intubated patients, and a multiplexed in vitro assay was developed to detect the HNE activity in breath samples using a rapid mass spectrometry technique. The quantitative assessment of HNE protease activity is achieved by the calculated ratio of the intact substrate to its corresponding cleavage product, as extracted from the mass peak intensities measured via mass spectrometry. Results We engineered an HNE substrate sensor, PEG36-Nle(OBzl)-Met(O)2-Oic-Abu-ACC, for the in vitro assay. A subsequent covalent modification at the C-terminus with 7-amino-4-carbamoylmethylcoumarin (ACC) facilitates its removal by HNE, and the N-terminus is covalently bound to amino-PEG36-acid (polyethylene glycol). We applied this in vitro assay to 13 LRTI cases, 15 non-LRTI cases, and 19 healthy volunteers. The results showed that the LRTI group significantly differed from non-infection cases, with a 9.2 mean difference. The Receiver Operating Characteristic (ROC) curve delivered an Area Under the Curve (AUC) value of 0.987 between LRTI and nonLRTI groups, showing the diagnostic potential of our approach. Using Youden*s index on the ROC curve, an optimal HNE threshold of 0.2 pM was used, which offered a sensitivity of 1 and specificity of 0.867. Conclusion Using human breath and specific substrate sensors, we developed an in vitro assay and a noninvasive approach for LRTI diagnosis in critical care. We demonstrated that the substrate sensors are sensitive, which is critical to detect low concentrations of proteins in breath. We applied this noninvasive in vitro approach to clinical breath samples and demonstrated the diagnostic potential of our approach. The approach we developed in this study could make a significant impact on LRTI management.