Pulmonary Arterial Hypertension Induces a Distinct Signature of Circulating Metabolites

Pulmonary arterial hypertension (PAH) is an incurable, progressive disorder, and the early diagnosis and treatment of PAH are associated with increased survival [...].

a characteristic clustering of the PAH group, distinguishing them from DM and heart patients. The principal component analysis (PCA) of significantly changed metabolites ( Figure 1C,D) showed a clear separation of the PAH sub-population from DM and heart patients. Thus, the plasma metabolomics profile of patients with PAH is distinctly different compared with diabetic patients or patients with left heart diseases.
The principal component analysis (PCA) of significantly changed metabolites ( Figure 1C,D) showed a clear separation of the PAH sub-population from DM and heart patients. Thus, the plasma metabolomics profile of patients with PAH is distinctly different compared with diabetic patients or patients with left heart diseases. Figure 1. Metabolic profiling of plasma of pulmonary arterial hypertension (PAH) patients (n = 11), and comparison with diabetes mellitus (DM; n = 12) and left heart disease (heart; n = 11) patients. Our data indicate significantly (p < 0.05) altered metabolites in PAH vs. heart and PAH vs. DM analysis. (A,B) Heat maps and (C,D) principal component analysis (PCA) show the clustering of the PAH group that could differentiate PAH patients from DM and heart by circulating metabolites. Ellipses indicate the area of a 0.95 probability that the metabolic profiling of the patient from the same group will be inside an ellipse.
Moreover, we have identified 32 unique metabolites that are significantly altered in PAH compared with either control group and could be used as a fingerprint for PAH (Table 1). These unique metabolites could be classified into a few groups. The primary group consists of mitochondrial-derived metabolites, including tricarboxylic acid (TCA) metabolites and their derivatives, which have been reported as significantly altered in PAH subjects compared with healthy controls [13], and may indicate undergoing anaplerotic reactions. Thus, oxalic acid, the product of oxaloacetate decomposition, showed a ~20 fold increase in PAH compared with either control cohort. The next largest group consisted of carbohydrates that could become elevated as a result of the glycolytic shift associated with PAH pathogenesis. Vascular remodeling in PAH could be the main reason for a decreased level of circulating amino acids, the primary building blocks of proteins that are highly consumed by proliferating cells. Elevated levels of plasma myo-inositol and its derivatives strongly correlate with our recently published animal data, demonstrating that myoinositol could play an important role in proliferative signaling in PAH [14]. The last two groups contain metabolites associated with PAH-mediated damage, or with the altered gut microbiome, which may also be involved in PAH pathogenesis [15].
Although the analysis of PAH samples versus DM or heart cohorts was assessed in two different runs, we found a very robust reproducibility of the metabolic data ( Figure 2). Therefore, we were able to combine two experiments using the PAH group as a reference point and compare all three patients' show the clustering of the PAH group that could differentiate PAH patients from DM and heart by circulating metabolites. Ellipses indicate the area of a 0.95 probability that the metabolic profiling of the patient from the same group will be inside an ellipse.
Moreover, we have identified 32 unique metabolites that are significantly altered in PAH compared with either control group and could be used as a fingerprint for PAH (Table 1). These unique metabolites could be classified into a few groups. The primary group consists of mitochondrial-derived metabolites, including tricarboxylic acid (TCA) metabolites and their derivatives, which have been reported as significantly altered in PAH subjects compared with healthy controls [13], and may indicate undergoing anaplerotic reactions. Thus, oxalic acid, the product of oxaloacetate decomposition, showed a~20 fold increase in PAH compared with either control cohort. The next largest group consisted of carbohydrates that could become elevated as a result of the glycolytic shift associated with PAH pathogenesis. Vascular remodeling in PAH could be the main reason for a decreased level of circulating amino acids, the primary building blocks of proteins that are highly consumed by proliferating cells. Elevated levels of plasma myo-inositol and its derivatives strongly correlate with our recently published animal data, demonstrating that myo-inositol could play an important role in proliferative signaling in PAH [14]. The last two groups contain metabolites associated with PAH-mediated damage, or with the altered gut microbiome, which may also be involved in PAH pathogenesis [15].
Although the analysis of PAH samples versus DM or heart cohorts was assessed in two different runs, we found a very robust reproducibility of the metabolic data ( Figure 2). Therefore, we were able to combine two experiments using the PAH group as a reference point and compare all three patients' cohorts together ( Figure 3). Moreover, we undertook metabolite optimization and identified the minimal number of metabolites sufficient for a significant separation of the PAH group from DM and heart. In Figure 3, only eleven metabolites were used to distinguish the PAH samples from the other patients. This analysis indicates that PAH patients metabolically are well resolved from both diseases. Thus, this panel of eleven metabolites (oxalic acid, pseudouridine, gluconic acid, fumaric acid, uridine diphosphate (UDP)-glucuronic acid, aconitic acid, erythritol, 2-deoxytetronic acid, glutamic acid, inorganic phosphate, and 2-hydroxyglutaric acid) could be used for the pre-screening of patients to identify PAH at the early asymptomatic stage, or could help to minimize the time for PAH diagnosis after the onset of the initial symptoms, reported to be currently 47.1 ± 34.2 months [16]. cohorts together ( Figure 3). Moreover, we undertook metabolite optimization and identified the minimal number of metabolites sufficient for a significant separation of the PAH group from DM and heart. In Figure 3, only eleven metabolites were used to distinguish the PAH samples from the other patients. This analysis indicates that PAH patients metabolically are well resolved from both diseases. Thus, this panel of eleven metabolites (oxalic acid, pseudouridine, gluconic acid, fumaric acid, uridine diphosphate (UDP)-glucuronic acid, aconitic acid, erythritol, 2-deoxytetronic acid, glutamic acid, inorganic phosphate, and 2-hydroxyglutaric acid) could be used for the pre-screening of patients to identify PAH at the early asymptomatic stage, or could help to minimize the time for PAH diagnosis after the onset of the initial symptoms, reported to be currently 47.1 ± 34.2 months [16].  Optimization of the metabolic platform to distinct idiopathic PAH (IPAH) patients vs. DM or heart cohorts. The identified profile of eleven metabolites was sufficient to provide a significant separation of PAH patients (n = 11), from either patients with diabetes mellitus (DM; n = 12) or patients with left heart disease (heart; n = 11). Ellipses indicate the area of 0.95 probability that the metabolic profiling of the patient from the same group will be inside an ellipse.    (Figure 3). Moreover, we undertook metabolite optimization and identified the minimal number of metabolites sufficient for a significant separation of the PAH group from DM and heart. In Figure 3, only eleven metabolites were used to distinguish the PAH samples from the other patients. This analysis indicates that PAH patients metabolically are well resolved from both diseases. Thus, this panel of eleven metabolites (oxalic acid, pseudouridine, gluconic acid, fumaric acid, uridine diphosphate (UDP)-glucuronic acid, aconitic acid, erythritol, 2-deoxytetronic acid, glutamic acid, inorganic phosphate, and 2-hydroxyglutaric acid) could be used for the pre-screening of patients to identify PAH at the early asymptomatic stage, or could help to minimize the time for PAH diagnosis after the onset of the initial symptoms, reported to be currently 47.1 ± 34.2 months [16].  Optimization of the metabolic platform to distinct idiopathic PAH (IPAH) patients vs. DM or heart cohorts. The identified profile of eleven metabolites was sufficient to provide a significant separation of PAH patients (n = 11), from either patients with diabetes mellitus (DM; n = 12) or patients with left heart disease (heart; n = 11). Ellipses indicate the area of 0.95 probability that the metabolic profiling of the patient from the same group will be inside an ellipse.  Optimization of the metabolic platform to distinct idiopathic PAH (IPAH) patients vs. DM or heart cohorts. The identified profile of eleven metabolites was sufficient to provide a significant separation of PAH patients (n = 11), from either patients with diabetes mellitus (DM; n = 12) or patients with left heart disease (heart; n = 11). Ellipses indicate the area of 0.95 probability that the metabolic profiling of the patient from the same group will be inside an ellipse. There are limitations to this study. Thus, PAH-specific therapies or the size of the patient cohort could affect the outcome of metabolic profiling. Therefore, the larger cohort of patients and an analysis of the different PAH cohorts is required in order to confirm our conclusions and to estimate the prognostic value of metabolic profiling. Moreover, the future assessment of the other pulmonary hypertension (PH) World Health Organization (WHO) groups would estimate whether the discovered metabolic fingerprint could be applied toward the patients with different types of PH. Thus, it has been recently reported that not only PAH but other types of PH, including exercise-induced PH and chronic thromboembolic pulmonary hypertension (CTEPH), have a metabolomic pattern that is different compared with the control subjects [13,17]. However, while the PAH cohorts evaluated in these studies showed changes similar to our findings, such as increased levels of TCA cycle metabolites (fumarate, citrate, and malate), glycolysis intermediates (lactate and pyruvate), nucleosides (pseudouridine and urate), and ketone bodies (butyric acid derivatives), either exercise-induced PH or CTEPH produced less prominent changes and showed different types of metabolic alterations. These results suggest that an increase in the pulmonary pressure occurring not because of the over-proliferative processes in the pulmonary vasculature, but secondary to other triggers (WHO groups 2-5) or mild/undeveloped forms of PH (exercise-induced PH), could require individual metabolic profiling.
The heterogeneous nature of control cohorts (DM and heart) should also be considered, although both control groups were found to be metabolically more homogeneous than the IPAH cohort ( Figure 1C,D). We also understand that the conditions chosen are just a fraction of the diseases that could present in the general population. In the future, the same analysis should be extended to include other pathologies that metabolically can overlap with PAH patients, such as cancer, lung diseases, and systemic hypertension. Nevertheless, we believe that these findings will spark a discussion in the field on the potential value of metabolic profiling as a new diagnostic tool, and provide subsequent research with a specific set of preselected metabolites that could serve as a fingerprint of PAH. Furthermore, these results may highlight the potential value of the particular metabolites in dissecting the pathogenesis of PAH.