Although progress has been made in the therapy of pulmonary arterial hypertension (PAH), diagnosis is often late and therapy aims at vasoconstriction instead of the underlying mechanisms of pulmonary arterial (PA) remodeling. There is a need for biomarkers that are functional, diagnostic and prognostic and provide new targets for treatment.

We used a broad metabolomics approach with machine learning analysis to develop diagnostic and prognostic pulmonary hypertension (PH) biomarkers using a training cohort and an independent multicentre validation cohort of 233 patients. In addition, we investigated lipid homeostasis in the pulmonary arteries of idiopathic PAH (IPAH) lungs.

We identified a set of metabolites as diagnostic PH markers and showed that the same set had an excellent prognostic power. Generally, lipophilic metabolites were more robust predictors than hydrophilic metabolites. The ratio of three distinct fatty acids (FA) to three distinct lipids performed very well for PH prediction (AUC 0.89 and 0.90 in the training and validation cohort, respectively) and survival prediction, which was age-independent and improved the risk prediction derived from the established clinical scores FPHR4p and COMPERA 2.0 (from 1.8 to 3.0 and from 1.9 to 3.3, respectively). Explanted PAs from IPAH patients showed lipid accumulation and altered expression of genes related to lipid homeostasis. Our functional studies on PA endothelial and smooth muscle cells showed that elevated free FA levels leads to hallmarks of pulmonary arterial hypertension.

In summary, lipidomic changes in PH can be used as novel diagnostic and prognostic biomarkers that could lead to the discovery of new therapeutic targets.