Non-Invasive Estimation of Systemic Drug Exposure Using Exhaled Breath Metabolomics in Humans: A Translational Proof-of-Concept Study.

Accurate assessment of systemic drug exposure in humans remains challenging, as conventional approaches rely on intermittent blood sampling and laboratory-based assays that are invasive and provide limited temporal resolution. Exhaled breath contains volatile and semi-volatile compounds arising from drug metabolism and downstream biological processes, suggesting that it may serve as a non-invasive matrix for integrated pharmacokinetic sensing. We conducted a prospective observational study using exhaled breath metabolomics based on secondary electrospray ionization high-resolution mass spectrometry (SESI-HRMS) to estimate systemic drug exposure from a breath-sampling session. A total of 117 adults provided 127 paired breath-serum measurements under routine outpatient conditions. Model performance was evaluated using repeated k-fold cross-validation and quantified using concordance correlation coefficients. Breath-derived estimates of valproic acid, levetiracetam, lamotrigine, and lacosamide showed moderate concordance with measured serum concentrations, with concordance correlation coefficients ranging from 0.733 to 0.895 across medications. The lacosamide findings should be considered preliminary because of the limited subgroup size (n = 16). The observed concordance supports proof-of-concept exposure estimation but does not establish clinical interchangeability with serum-based therapeutic drug monitoring. The previously defined 11-feature VPA breath signature retained predictive utility in a separate adult cohort, supporting its cross-cohort transferability within the present analytical framework. These findings support the feasibility of exhaled breath analysis for non-invasive estimation of systemic drug exposure across pharmacologically distinct medications. Breath analysis should currently be considered complementary to serum-based monitoring, and external multicenter and longitudinal validation is required before clinical implementation.