Molecular breath profile of acute COPD exacerbations.

Acute exacerbations of chronic obstructive pulmonary disease (AECOPD) show high variability in individual susceptibility and promote disease progression; thus, accurate diagnosis and treatment is essential. Unravelling the molecular metabolic changes during AECOPD in breath could promote understanding of AECOPD and its treatment. Our objective was to investigate the metabolic breath profiles during AECOPD for biomarker detection. We conducted real-time breath analysis in patients with COPD during AECOPD and during subsequent stable phase. Molecular breath patterns were compared between AECOPD and stable phase by dimension reduction techniques and paired t-tests. Pathway enrichment analyses were performed to investigate underlying metabolic pathways. Partial least-squares discriminant analysis and XGboost were utilised to build a prediction model to differentiate AECOPD from stable state. 35 patients (60% male) with a mean age of 65 (10.2) yr with AECOPD were included. AECOPD could be predicted with a high sensitivity of 82.5% (95% confidence interval of 68.8%-93.8%) and an excellent discriminative power (AUC = 0.86). Metabolic changes in the linoleate, tyrosine, and tryptophan pathways during AECOPD were predominant. Significant metabolic changes occur during COPD exacerbations, predominantly in the linoleate, tyrosine, and tryptophan pathways, which are all linked to inflammation. Real-time exhaled breath analysis enables a good prediction of AECOPD compared to stable state and thus could enhance precision of AECOPD diagnosis and efficacy in clinical practice.