Validating Asthma Biomarkers in Children with Real-Time Breath Analysis

<bold>Background:</bold> Discovery of asthma linked biomarkers in exhaled breath of children shows promise, however, biomarker validation remains limited. This study aimed to address this by validating previously identified <italic>m/z</italic> features as potential asthma biomarkers from an exploratory pilot study using secondary electro-spray ionization high-resolution mass spectrometry (SESI-HRMS). <bold>Methods:</bold> Two years after the pilot study, an independent cohort of 88 children (46 asthmatic and 42 healthy controls, aged 7-17) was recruited for validation. The breath measurements were performed on an AB SCIEX TripleTOF 5600+ HRMS coupled to SUPER-SESI, consistent with the pilot study. The previously identified 149 m/z features were analyzed for their presence, statistical significance, and diagnostic ability using supervised machine learning. <bold>Results:</bold> Over 70% (102) of the <italic>m/z</italic> features were successfully replicated in the validation cohort (mass tolerance ≤ 0.001). Of these, 68 were found to be significant again (FDR adjusted p < 0.05), confirming many of the previously detected metabolic pathways and chemical families including lysine degradation, tyrosine metabolism, 2-oxocarboxylic acid metabolism, fatty acid metabolites and monosaccharides. Furthermore, when applied to the new cohort, a disease prediction model trained on the pilot study data resulted in a ROC AUC of 0.79 (accuracy: 77%, sensitivity: 86%, specificity: 67%), demonstrating similar diagnostic performance as in the pilot study. <bold>Conclusion:</bold> This study represents the first successful validation of a subset of asthma associated biomarkers through real-time breath analysis using SESI-HRMS. The rich set of markers showed a promising diagnostic ability on previously unseen data.