Background: Asthma is a heterogeneous disease with poorly defined phenotypes.

Aim: To identify metabolic dysregulations associated with asthma severity and evaluate the effects of asthma medication upon observed metabotypes.

Methods: Baseline urine was collected from healthy controls (HC, n=108), mild-to-moderate asthmatics (MMA, n=87) and severe asthmatics (SA, n=418) from the U-BIOPRED cohort. 12-18 month longitudinal samples were collected from the SA cohort (n=305). Metabolomic data were acquired using mass spectrometry and analyzed using multivariate statistics. Gene set variation analysis (GSVA) was performed on bronchial brushing transcriptomic data.

Results: 90 metabolites were identified with 40 altered in asthma (FDR<0.1). Multivariate modeling showed that HC and MMA differed significantly from all SA (p=1.4 ×10-14) and that oral corticosteroid (OCS)-treated asthmatics differed significantly from non-treated (p=9.52 ×10-4). Longitudinal samples were metabolically stable relative to baseline. OCS affected the levels of 25% of the metabolites, while theophylline affected 12%, and omalizumab had a minimal effect. Carnitine levels decreased in SA in an OCS-independent fashion. Carnitine is involved in long-chain fatty acid metabolism in mitochondria, which decreased along with levels of the carnitine transporter SLC22A5 in association with asthma severity in bronchial brushings, with differences strengthened by Th2 high/low stratification.

Conclusions: SA have a dysregulated urinary metabolic profile that is strongly confounded by OCS treatment. Altered carnitine metabolism is independent of OCS and associated with mitochondrial dysfunction, presenting a potential target for intervention.