Validation of volatile metabolites of pulmonary oxidative injury: a bench to bedside study

**Background:** Oxidative stress is implicated in many pulmonary disorders, and exhaled volatile organic compounds (VOCs) have been proposed as noninvasive biomarkers. However, identified VOCs are inconsistent across studies, and translational validation from in vitro to clinical settings is lacking. This study aimed to validate known and novel VOCs of oxidative stress using a combined bench-to-bedside approach.

**Methods:** In the in vitro component, A549 alveolar epithelial cells were exposed to 1 mM hydrogen peroxide (H2O2) to induce oxidative stress. VOCs were captured from the headspace of sealed glass vials using high-capacity polydimethylsiloxane sorbent fibres (HiSorbs) and analyzed by gas chromatography-mass spectrometry (GC-MS). Both targeted analysis (literature-reported VOCs) and untargeted discovery analysis were performed. Cell stress was confirmed by lactate dehydrogenase (LDH) and interleukin-8 (IL-8) measurements. For the clinical component, breath samples were collected from 489 invasively ventilated ICU patients from the DARTS multicenter cohort study. Samples were collected on Tenax GR sorbent tubes and analyzed by the same GC-MS method. Pulmonary oxidative stress risk was defined by FiO2 ≥0.6. Spearman rank correlation and area under the receiver operating characteristic curve (AUROCC) were used to assess associations between VOCs and oxygen exposure.

**Key Results:** In vitro, H2O2 exposure caused significant cellular stress (increased LDH and IL-8). Of the targeted VOCs, only octane showed a significant difference—but in the opposite direction, with lower concentrations in stressed cells (p=0.0013). Cyclohexane, decane, heptane, hexane, and undecane showed no significant changes. Untargeted analysis of 357 VOCs identified only one differentially expressed compound: 2-ethyl-1-hexanol, which was increased in H2O2-exposed cells (p=0.00014). In the clinical cohort (n=489; 90 high-risk, 399 low-risk), only three VOCs (decane, hexane, octane) were detectable in patient breath. None showed significant correlation with FiO2 (Spearman Rs range: −0.015 to −0.065). AUROCC values for discriminating FiO2 ≥0.6 ranged from 0.48 to 0.55, indicating no discriminatory ability. 2-ethyl-1-hexanol was not detected in patient breath. Sensitivity analyses using different FiO2 thresholds or prolonged oxygen exposure showed no improvement.

**Clinical Implications:** This study failed to validate any known or novel volatile biomarkers of oxidative stress despite a rigorous translational design, a contamination-minimized in vitro model, and one of the largest clinical breath cohorts to date. The findings challenge the widely held assumption that exhaled hydrocarbons and related VOCs reliably reflect pulmonary oxidative injury. The inconsistencies underscore the risk of false discovery in breath research, the importance of exogenous contaminants, and the need for better understanding of VOC biosynthetic pathways. The authors caution against attributing clinical breath VOC findings to oxidative stress without direct mechanistic evidence.