Metabolomic Insights in Pulmonary Arterial Hypertension

One of the first applications of metabolomics to samples from patients with PAH was performed on lung tissue.7 A comparison of 8 PAH lung samples collected at transplantation with 8 lobectomy samples identified increased glycolysis metabolites and altered fatty acid profiles, alongside changes in gene expression; the 4 genes that encode the enzymes fatty acetyl-CoA L1 (ACSL1), Acyl CoA dehydrogenases (ACADM), Acetyl–CoA acetyltransferase1 (ACAT1), and Acetyl-CoA carboxylase (ACACA), respectively, were all increased in PAH tissue. Tricarboxylic acid (TCA) cycle metabolites including citrate, succinate, cisaconitate, and genes including SUCLA2, encoding succinate-CoA ligase, the gene encoding fumarate hydratase (FH), and the genes encoding aconitase (IREB1) and the second isoform of aconitase, iron responsive element binding protein 2 (IREB2), were all upregulated in PAH tissues. Whether this reflects increased TCA cycle activity or a buildup of intermediates due to a block in the cycle cannot be concluded from these data, but clearly the energetics of pulmonary cells are disturbed in PAH patients. Previous data on alterations in metabolism driven by BMPR2 mutations (the principle cause of heritable PAH8) in human pulmonary artery endothelial cells also demonstrated the relevance of energetics, but with some differences; several TCA intermediates were reduced, except citrate, suggesting a possible block downstream of citrate, and acylcarnitines and glutamate were also lower in cells expressing the mutant BMPR2.9 Both studies showed elevated gene expression of isocitrate dehydrogenase1 (IDH1) and Fessel et al suggested serum activity may be a biomarker of disease severity.9 

A later analysis of the lung metabolite data demonstrated further abnormalities including increased sphingolipids, heme breakdown products (biliverdin, bilirubin), and arginine metabolism (leading to polyamine, putrescine, spermidine, and spermine production) in PAH.10 This was again paired with altered gene expression, including elevated arginase1 (ARG1), sphingomyelin synthase 2 (SGMS2), and N-acylsphingosine amidohydrolase (non-lysosomal ceramidase) 2 (ASAH2) in PAH tissue.

A targeted study of dyspneic individuals where plasma concentrations of 105 metabolites were analyzed and related to hemodynamic measures, including resting right atrial pressure, mean pulmonary artery pressure (PAP), pulmonary vascular resistance (PVR), and PVR and PAPflow response (DPQ) during exercise, identified indoleamine 2,3-dioxygenase (IDO)–dependent tryptophan metabolites (TMs; anthranilate, quinolinate, and kynurenine) (Figure 1) release higher levels in radial artery than proximal pulmonary artery in patients with PVR >2 Woods units and higher circulating levels in PAH patients than controls in an independent validation cohort.11 Arginine and TCA cycle intermediates (aconitate, isocitrate, malate, and succinate methylmalonate) were associated with the severity of PH in the discovery analysis (n=71) and derived metabolite scores were associated with mean PAP in a validation cohort (n=71).

The specificity of the IDO-TM changes was perhaps surprising, as the other pathway of tryptophan metabolism produces serotonin (via tryptophan hydroxylase). Perturbation of serotonin activity is linked with PAH12,13 but metabolites of this pathway were not associated with the measures studied.

The adult heart normally obtains 50% to 70% of its energy from fatty acid β-oxidation.14 In heart failure, circulating levels of fatty acids are elevated, but uptake to cardiac tissue can be inhibited by concomitantly elevated ketone bodies (β-hydroxybutyrate and acetoacetate). A study of nonesterified free fatty acids (FFAs) and acylcarnitines, which shuttle fatty acids across membranes, demonstrated FFAs were around twice as high in 19 patients with PAH than in 22 controls and long chain (defined as ≥14 carbons) acylcarnitines were also elevated. 15 Higher levels of the long chain acylcarnitines, namely, palmitoylcarnitine, stearoylcarnitine, oleoylcarnitine, and linoleoylcarnitine, were associated with worse functional class. Proton magnetic resonance spectroscopy was used to demonstrate elevated lipid levels in right ventricular (RV) tissue from living patients with PAH (Figure 2). However, measurements of PAH RV tissue showed significant reductions in the levels of acylcarnitines compared to unmatched donor hearts and explanted hearts from patients with dilated cardiomyopathy (DCM), as diseased comparators. RV tissue from BMPR2 mutant mice was significantly less able to metabolize long chain fatty acids than control tissue, suggesting that PAH RV tissue may have elevated fatty acids due to reduced processing to acylcarnitines. Indeed, elevated fatty acid levels were measured in PAH tissue, along with elevated ceramide, a potentially toxic alternative breakdown product.

Rhodes et al performed the first unbiased (in terms of target metabolite selection) screen of plasma metabolomics in patients with PAH using UPLC-MS methodologies provided by Metabolon Inc. (Durham, NC, USA).16 

Nonfasting plasma samples were provided by healthy controls (n=121 in 2 groups), disease controls (symptomatic patients referred to the PH clinical service at Hammersmith Hospital but found not to have PH, n=139 in 2 groups) and patients with idiopathic or heritable PAH (n=365 in a discovery and validation group from Hammersmith Hospital plus a further validation group from expert centers across the UK [The National Cohort Study of PAH, see ipahcohort.com]). Detected and measured in over 95% of samples were 686 biological metabolites. Plasma levels of 53 metabolites were significantly different in patients with PAH compared with healthy controls in all 3 cohorts after Bonferroni correction for multiple comparisons, and after further analyses correcting for potential confounders such as liver and renal function, age, sex, ethnicity, body mass index, and drug therapies. Twenty of these metabolites distinguished the PAH cohorts from the disease controls, and a network analysis identified the modified nucleoside, N2,N2-dimethylguanosine, and the TCA cycle intermediate, malate, as key hub metabolites. Discriminant scores based on just 7 and 4 of these metabolites were able to distinguish patients with PAH from healthy controls with 93% to 95% accuracy and from disease controls with 72% to 75% in validation cohorts. Interestingly, patients with PAH deemed to be vasoresponders, a subgroup who have relatively good clinical outcomes on calcium channel blocker therapy, had metabolite profiles more similar to healthy controls than patients with PAH. The patients in this study had established disease (ie, were not treatment naive), and it would be of great interest to study these metabolites before and after initiation of vasodilator therapy to observe whether vasoresponders have more “healthy” metabolomes at diagnosis or whether it is corrected by successful treatment.

Clustering analysis of these metabolites suggested that the most significant shifts in metabolite levels were associated with patients who died during follow-up. Consistent with this, 36 metabolites were prognostic in 2 cohorts of PAH patients, independent of NT-proBNP, 6-minute walk, and red cell distribution width data (Figure 3), a known highly prognostic combination. 17 Serial samples from patients who died during follow-up were more likely to show a further increase in prognostic metabolites, including circulating modified nucleoside levels, than patients who survived, suggesting that monitoring metabolite levels over time could be useful in judging response to therapy.

The elevated circulating modified nucleosides (N1-methylinosine and N2,N2-dimethylguanosine) originate from transfer RNA molecules and their release into the circulation during stress follows cleavage by the ribonuclease angiogenin (Figure 4).18 Elevated plasma angiogenin levels correlated with higher levels of modified nucleosides. Increased angiogenin levels have been reported in breath condensates19 of patients with PAH, which implicates the lung as a potential source. Furthermore, elevated tRNA nucleosides are observed in hyperproliferative cancers, most likely reflecting the general upregulation of the translational apparatus.20,21 This suggested that levels of these metabolites may be reporting on stressed, proliferative pulmonary vascular cells in patients with PAH.

Consistent with previous studies, we found alterations in several energetic metabolism pathways, including elevated acylcarnitines, glutamate, and altered TCA cycle intermediates. The buildup of precursors is consistent with dysfunction in the TCA cycle and/or the electron transport chain downstream, preventing cells from matching the energy demands forced upon them by the disease pathology (Figure 5). Not all of the metabolites of the TCA cycle are easily quantified, so identification of the specific steps that may be affected was not possible in this study. As demonstrated by Brittain et al,15 circulating metabolite levels do not necessarily represent levels in relevant tissues and further study of specific pathways in pulmonary vascular and myocardial cells could illuminate these alterations.

There were reductions in multiple lipid species, including sphingomyelins and phosphatidylcholines and steroids including dehydroisoandrosterone-sulfate (DHEA-S) and metabolites, in patients with PAH, especially in those with poor clinical outcomes. Some of these may relate to developing insulin resistance but may also act more directly as sources of cellular signaling molecules, including eicosanoids, prostacyclin being the most relevant to PAH. In contrast to previous reports, circulating tryptophan metabolites were increased, but this tracked with a marker of liver dysfunction (bilirubin levels). This emphasizes the importance of interpreting circulating metabolomic profiles in the context of systems biology, and investigating the relevance to diseased tissue measurements can provide more direct evidence of action at the site of pathology.

The observation that metabolite profiles are most disturbed in PAH patients with a poor clinical outcome has 2 implications. First, measurements of circulating metabolites in the clinic could provide useful information in terms of stratifying patients by prognosis. Second, medications capable of reversing the metabolic changes observed may be more likely to succeed in treating the disease. But these claims need to be examined prospectively and compared with prognostic risk equations such as the REVEAL score,22 and alternative candidates such as proteomic measurements from patients with PAH.23 Establishing the role of these metabolites in the pathology of PAH is also key to adoption of measurements in clinical practice.

There is considerable interest in whether metabolomic profiles differentiate subclasses of PH. In the first instance, this will require broadening cohort studies to collect samples from patients presenting with a tentative diagnosis of PH. Studies showing altered metabolite profiles including blood tyrosine, ornithine, thyroid stimulating hormone, and phenylalanine levels in persistent PH of newborns already highlight that metabolic disturbances may occur early in multiple presentations of PH.24 The question of tissue origin of the circulating metabolites of interest in PAH is also a pressing concern. Sampling of patients at multiple anatomic locations as well as studies in specific tissues and cells will help answer this. Efforts toward these goals are underway in the United Kingdom and in the United States as part of the PVDomics study.25 

High-throughput technologies have broadened our understanding of the metabolic disturbances in PAH far beyond the earlier studies focusing on the Warburg effect of switching to glycolytic metabolism as observed in cancers. Multiple studies have made complementary findings that encourage belief in the robustness of the associations between PAH disease burden and shifts in the way metabolites are transported and used, for example, in the failing right ventricle. The clinical utility of these measurements will depend on the development of easy-to-use assay formats that can be deployed in the clinic or in the longer term, assimilation of full metabolic profiling into clinical laboratories. Preclinical studies will define the importance and function of these metabolites, and therapeutic strategies to correct those implicated directly in the pathogenesis of PAH need to be explored.