Exhaled breath profiling for diagnosing acute respiratory distress syndrome

Abstract

Background: The acute respiratory distress syndrome (ARDS) is a common, devastating complication of critical illness that is characterized by pulmonary injury and inflammation. The clinical diagnosis may be improved by means of objective biological markers. Electronic nose (eNose) technology can rapidly and non-invasively provide breath prints, which are profiles of volatile metabolites in the exhaled breath. We hypothesized that breath prints could facilitate accurate diagnosis of ARDS in intubated and ventilated intensive care unit (ICU) patients.

Methods: Prospective single-center cohort study with training and temporal external validation cohort. Breath of newly intubated and mechanically ventilated ICU-patients was analyzed using an electronic nose within 24 hours after admission. ARDS was diagnosed and classified by the Berlin clinical consensus definition. The eNose was trained to recognize ARDS in a training cohort and the diagnostic performance was evaluated in a temporal external validation cohort.

Results: In the training cohort (40 patients with ARDS versus 66 controls) the diagnostic model for ARDS showed a moderate discrimination, with an area under the receiver-operator characteristic curve (AUC-ROC) of 0.72 (95%-confidence interval (CI): 0.63-0.82). In the external validation cohort (18 patients with ARDS versus 26 controls) the AUC-ROC was 0.71 [95%-CI: 0.54 - 0.87]. Restricting discrimination to patients with moderate or severe ARDS versus controls resulted in an AUC-ROC of 0.80 [95%-CI: 0.70 - 0.90]. The exhaled breath profile from patients with cardiopulmonary edema and pneumonia was different from that of patients with moderate/severe ARDS.

Conclusions: An electronic nose can rapidly and non-invasively discriminate between patients with and without ARDS with modest accuracy. Diagnostic accuracy increased when only moderate and severe ARDS patients were considered. This implicates that breath analysis may allow for rapid, bedside detection of ARDS, especially if our findings are reproduced using continuous exhaled breath profiling.

Trial registration: NTR2750, registered 11 February 2011.

Figure 1

Sample collection and data analysis. Exhaled breath was sampled and analyzed using an electronic nose with a side–stream connection distal from the endotracheal tube. This resulted in a response for the 32 polymer sensors in the nano–composite sensor array. The eNose was trained using sparse–partial least square (SPLS) logistic regression with 10.000–fold cross–validation. Data from the training cohort was split into a fraction for model building and model evaluation (10 cases and 10 controls). The algorithm that provided the best internally validated diagnostic accuracy, evaluated by the area under the receiver operating characteristics curve (ROC–AUC), was selected for blind testing in the validation cohort and the ROC–AUC with optimal sensitivity and specificity was reported. Differences in the predictive algorithm between different subgroups (severity of disease, pulmonary and non–pulmonary ARDS) were analyzed using non–parametric tests and the ROC–AUC was reported. Furthermore, the ROC–AUC for distinguishing CPE and pneumonia from ARDS and moderate/severe ARDS only was calculated. A sensitivity analysis was performed using logistic regression on comorbidities that are known to influence breath prints, the PaO2/FiO2 ratio, minute volume ventilation and APACHE II and SAPS II scores.

Figure 2

Flow of patient inclusion. ARDS: acute respiratory distress syndrome; CPE: cardiogenic pulmonary edema; MV: mechanical ventilation.