Supplementary Material for: Breathprint of Pulmonary Sarcoidosis: Development of a Non-Invasive Diagnostic Tool

Patients with sarcoidosis were further stratified into active and inactive disease states. Molecular breath profiles were compared using t-test and prediction models were developed based on ReliefF feature selection and Naïve Bayes algorithms.</p><p dir="ltr"><b>RESULTS</b></p><p dir="ltr">39 patients with pulmonary sarcoidosis, 39 matched controls, and 25 patients with pulmonary fibrosis were included. Sarcoidosis could be discriminated from matched controls with a repeated cross-validated ROC-AUC of 0.76, with the main metabolic changes in fatty acid and cysteine metabolism.

Active sarcoidosis was discriminated from inactive disease with a repeated cross-validated ROC-AUC of 0.77, with linoleate metabolism playing a central role.</p><p dir="ltr"><b>CONCLUSION</b></p><p dir="ltr">Exhaled breath analysis shows potential to distinguish patients with pulmonary sarcoidosis from matched controls and patients with pulmonary fibrosis, and to differentiate between active and inactive sarcoidosis. Metabolic signatures, particularly involving fatty acid and linoleate metabolism, support its potential as a non-invasive, real-time tool for diagnosis and monitoring of pulmonary sarcoidosis, pending validation in larger cohorts.INTRODUCTION</p><p dir="ltr">Sarcoidosis is a systemic inflammatory disease of unknown origin, primarily affecting the lungs and lymphatic system.

Diagnosis relies on clinical assessment, histological confirmation of non-necrotizing granulomas, and exclusion of other conditions. However, current diagnostic methods are invasive and costly, making them unsuitable for disease monitoring. Existing biomarkers lack accuracy and clinical utility.

The objective was to evaluate exhaled breath analysis as a non-invasive, real-time diagnostic and monitoring tool for pulmonary sarcoidosis, by identifying disease-specific metabolic signatures.</p><p dir="ltr">METHODS</p><p dir="ltr">Exhaled breath samples were collected from patients with pulmonary sarcoidosis, matched controls without lung disease, and patients with pulmonary fibrosis. Patients with sarcoidosis were further stratified into active and inactive disease states.

Molecular breath profiles were compared using t-test and prediction models were developed based on ReliefF feature selection and Naïve Bayes algorithms.</p><p dir="ltr">RESULTS</p><p dir="ltr">39 patients with pulmonary sarcoidosis, 39 matched controls, and 25 patients with pulmonary fibrosis were included. Sarcoidosis could be discriminated from matched controls with a repeated cross-validated ROC-AUC of 0.76, with the main metabolic changes in fatty acid and cysteine metabolism.

Active sarcoidosis was discriminated from inactive disease with a repeated cross-validated ROC-AUC of 0.77, with linoleate metabolism playing a central role.</p><p dir="ltr">CONCLUSION</p><p dir="ltr">Exhaled breath analysis shows potential to distinguish patients with pulmonary sarcoidosis from matched controls and patients with pulmonary fibrosis, and to differentiate between active and inactive sarcoidosis. Metabolic signatures, particularly involving fatty acid and linoleate metabolism, support its potential as a non-invasive, real-time tool for diagnosis and monitoring of pulmonary sarcoidosis, pending validation in larger cohorts.</p><p dir="ltr"><br></p>