Exogenous Volatile Organic Compound (EVOC^®) Breath Testing Maximizes Classification Performance for Subjects with Cirrhosis and Reveals Signs of Portal Hypertension

Abstract

Background: Cirrhosis detection in primary care relies on low-performing biomarkers. Consequently, up to 75% of subjects with cirrhosis receive their first diagnosis with decompensation when causal treatments are less effective at preserving liver function. We investigated an unprecedented approach to cirrhosis detection based on dynamic breath testing. Methods: We enrolled 29 subjects with cirrhosis (Child–Pugh A and B), and 29 controls. All subjects fasted overnight. Breath samples were taken using Breath Biopsy® before and at different time points after the administration of 100 mg limonene. Absolute limonene breath levels were measured using gas chromatography–mass spectrometry. Results: All subjects showed a >100-fold limonene spike in breath after administration compared to baseline. Limonene breath kinetics showed first-order decay in >90% of the participants, with higher bioavailability in the cirrhosis group. At the Youden index, baseline limonene levels showed classification performance with an area under the roc curve (AUROC) of 0.83 ± 0.012, sensitivity of 0.66 ± 0.09, and specificity of 0.83 ± 0.07. The best performing timepoint post-administration was 60 min, with an AUROC of 0.91, sensitivity of 0.83 ± 0.07, and specificity of 0.9 ± 0.06. In the cirrhosis group, limonene bioavailability showed a correlation with MELD and fibrosis indicators, and was associated with signs of portal hypertension. Conclusions: Dynamic limonene breath testing enhances diagnostic performance for cirrhosis compared to static testing. The correlation with disease severity suggests potential for monitoring therapeutic interventions. Given the non-invasive nature of breath collection, a dynamic limonene breath test could be implemented in primary care.

Keywords

• breath biopsy
• volatile organic compounds
• functional diagnostics
• non-invasive
• MELD