Plasma 12- and 15-hydroxyeicosanoids are predictors of survival in pulmonary arterial hypertension

Abstract Abstract

This study aimed to characterize alterations in select eicosanoids in experimental and human pulmonary arterial hypertension (PAH) and to assess their potential utility as predictors of outcome. Using liquid chromatography–mass spectrometry, we performed targeted lipidomic analyses of the lungs and right ventricles (RVs) of chronically hypoxic rats and plasma of consecutive PAH patients and healthy controls. In rat lungs, chronic hypoxia was associated with significantly decreased lung prostacyclin (PGI₂)/thromboxane B₂ (TXB₂) ratio and elevated lung 8-hydroxyeicosanoid (HETE) acid concentrations. RV eicosanoids did not exhibit any changes with chronic hypoxia. PAH treatment–naïve patients had significantly increased plasma concentrations of TXB₂ and 5-, 8-, 12-, and 15-HETE. The PGI₂/TXB₂ ratio was lower in PAH patients than in controls, especially in the treatment-naïve cohort (median: 2.1, 0.3, and 1.3 in controls, treatment-naïve, and treated patients, respectively, P = 0.001). Survival was significantly worse in PAH patients with 12-HETE_high (≥57 pg/mL) and 15-HETE_high (≥256 pg/mL) in unadjusted and adjusted analyses (hazard ratio [HR]: 2.8 [95% confidence interval (CI): 1.1–7.3], P = 0.04 and HR: 4.3 [95% CI: 1.6–11.8], P = 0.004, respectively; adjustment was performed with the REVEAL [Registry to Evaluate Early and Long-Term PAH Disease Management] risk score). We demonstrate significant alterations in eicosanoid pathways in experimental and human PAH. We found that 12- and 15-HETE were independent predictors of survival in human PAH, even after adjusting for the REVEAL score, suggesting their potential role as novel biomarkers.

Pulmonary arterial hypertension (PAH) is a life-threatening disease characterized by pulmonary vascular remodeling leading to impaired pulmonary hemodynamics, right ventricular (RV) failure, and death. While biomarkers can potentially be helpful in diagnosis, follow-up, and management of PAH patients, only brain natriuretic peptide (BNP) or its precursor is currently recommended at diagnosis and for longitudinal follow-up; neither, however, is specific to PAH.¹

Alterations in lipid metabolism of the arachidonic acid (AA) derivatives (Fig. 1), such as prostaglandins (PGs), leukotrienes, and hydroxyeicosanoids (HETEs), have been implicated in the pathogenesis of pulmonary hypertension (PH).^2,3 Studies suggest that metabolites of the AA derivatives mediate endothelium-dependent vasoconstriction.^4-7 The most studied of these in the pulmonary vasculature are prostacyclin (PGI₂), with its potent vasodilator, antiplatelet, antiproliferative, and anti-inflammatory properties,⁸ and thromboxane A₂ (TXA₂), which opposes the actions of PGI₂ by promoting vasoconstriction, platelet aggregation, and smooth muscle cell (SMC) proliferation.⁹ Lung expression of PGI₂ synthase is decreased¹⁰ and urinary levels of TXB₂ (the stable metabolite of TXA₂) are increased¹¹ in patients with PAH. The lung vasculature also synthesizes various HETEs, which are implicated pathogenetically in experimental models of PH^12-14 but not yet fully explored in PAH patients. Of these, 5-HETE is an enzymatic product of 5-lipoxygenase (5-LOX) and is thought to modulate the response of the lung vasculature in experimental PH;¹⁵12-HETE, the product of 12-LOX, is upregulated in the lungs of hypoxic rats and stimulates proliferation of pulmonary artery SMCs;¹⁴ 15-HETE, a precursor for lipoxins, promotes hypoxic pulmonary vasoconstriction, inhibits apoptosis of pulmonary artery SMCs, and promotes pulmonary vascular remodeling in some experimental PH models,^16-19 whereas in other models, it has vasodilatory effects on both the systemic and pulmonary vasculature.^20,21 It has also been found that 15-HETE possesses anti-inflammatory properties,²² which may be relevant, given the growing evidence of the role of inflammation in the pathogenesis of PAH.²³ Other products of AA metabolism, such as 8-HETE, have been investigated only superficially in PAH.²

Figure 1.

Arachidonic acid pathway.

The importance of lipid mediators derived from the AA metabolism in the pathophysiology of PAH is highlighted by the therapeutic use of PGI₂ or its analogs, with resultant amelioration of the disease.^24-27 However, the utility of these lipid mediators as biomarkers in PAH has not been investigated. In this study, we aimed to (1) characterize the alterations in targeted lipidomic profiles of AA derivatives of experimental and human PAH, specifically focusing on PGs and HETEs, (2) compare the lipidomic profiles of AA derivatives in experimental and human PAH, and (3) investigate the prognostic utility of these lipid mediators as potential biomarkers for PAH.

Methods

Hypoxic exposures and tissue sampling in rats

Thirty-six male rats (Sprague-Dawley, 250–300 g), 2–4 months old, were randomized to exposure: normoxia (21% oxygen saturation for 3 weeks, n = 22) or chronic hypoxia (11% oxygen saturation for 3 weeks, n = 14). Hemodynamic measurements were obtained at the end of the exposure period via a continuous pressure transducer catheter inserted through the right jugular vein and advanced to the right ventricle (RV).^14,28 After catheterization, the animals were euthanized, their thoraxes were opened, and their lungs and RVs were harvested immediately, weighed, and stored at −80°C. The protocol was approved by the Institutional Animal Care and Use Committee at Tufts Medical Center and was conducted in accordance with the National Institute of Health Guidelines for the Care and Use of Laboratory Animals.

Lung and RV lipidomic analyses

Lipid extraction and analyses were carried out on homogenates of harvested lungs and RVs and analyzed by liquid chromatography–mass spectrometry (LC/MS/MS; appendix).²⁸

Human subjects

The Tufts Medical Center Pulmonary Hypertension Database is a prospective registry of adult patients with a diagnosis of PH by right heart catheterization (RHC; appendix). The study was approved by the Tufts Medical Center Institutional Review Board, and informed consent was obtained from all subjects.

We reviewed data on consecutive patients evaluated between June 1, 2006, and March 31, 2012, and patients were classified into etiologic groups based on current guidelines.²⁹ PAH was defined as mean pulmonary artery pressure (mPAP) ≥ 25 mmHg, pulmonary capillary wedge pressure (PCWP) ≤ 15 mmHg, and pulmonary vascular resistance (PVR) > 3 Wood units.²⁹ Patients were included in the study if they had idiopathic PAH (IPAH) or connective-tissue disease (CTD)–associated PAH. Patients with other forms of group 1 PAH or group 2 (PH secondary to left heart disease), group 3 (PH secondary to chronic lung disease), group 4 (chronic thromboembolic PH), or group 5 (miscellaneous) PH were excluded. Patients were included in the study regardless of their treatment status (treatment naïve vs. treated).

Demographic and clinical characteristics were recorded, and blood samples were collected from fasting patients at the time of the RHC. Blood samples were also collected from healthy volunteers as controls. Death was determined from medical records and confirmed with the Social Security Death Index. All causes of death were considered in the survival analysis. Survival time was defined by the date of the RHC at which the blood samples were obtained. Survival data were censored on February 1, 2015.

Lipidomic analyses of human plasma

Plasma samples, collected at the time of the RHC in ethylene diamine tetra-acetic acid (EDTA) tubes, were centrifuged at 4°C, and plasma was isolated and stored in cryotubes at −80°C. Lipid extraction was carried out and analyzed by LC/MS/MS (appendix).²⁸

Statistical analysis

Continuous variables are presented as medians and interquartile range. Categorical variables are presented as frequencies and percentages. Wilcoxon rank-sum or Kruskal-Wallis tests were used for comparing tissue and plasma levels of eicosanoids, as appropriate. In order to evaluate the performance of different eicosanoids as predictors of outcome in PAH, the areas under the curve (AUCs) of the receiver operating characteristics (ROC) curves were calculated. The best AUCs were those of 12- and 15-HETE, and the optimal cutoff from each of their respective ROC curves was chosen. Kaplan-Meier survival curves were constructed, and the log-rank test was used to compared them. Cox proportional hazards were used to determine the hazard ratio of eicosanoids and survival in PAH; however, data are shown only for 12- and 15-HETE, as they were the only two biomarkers with significant associations. The results were then adjusted for the REVEAL (Registry to Evaluate Early and Long-Term PAH Disease Management) risk score.³⁰ Statistical analysis was performed with R, version 3.1.1, and the survival package (version survival 2.37–7). A P value of <0.05 was considered statistically significant.

Results

Experimental PH model

Chronic hypoxia was associated with significantly elevated RV pressures compared with normoxic controls (51 ± 6 vs. 24 ± 2 mmHg, P < 0.01) and RV hypertrophy compared with normoxia (RV mass/(mass of left ventricle and septum): 0.47 ± 0.05 vs. 0.25 ± 0.01, P < 0.01).

Hypoxia alters eicosanoid levels in rodent lungs

Levels of eicosanoids in the rat lungs were compared between the two exposures. In normoxia, several PG metabolites were below the limit of quantitation or had very low levels (Table S1).

Chronic hypoxia was associated with significant decreases in lung PGI₂/TXB₂ ratio (Fig. 2C). Other PG metabolites involved in bronchodilation and vasodilation, including prostaglandin E₂ (PGE₂), PGI₂, and TBX₂, were not significantly changed by hypoxia (Fig. 2A, 2B, 2D). Among different HETE molecules, chronic hypoxia was associated with high lung concentrations of 8-HETE. The 5-, 12-, and 15-HETE levels in the rat lungs were not significantly altered by hypoxia (Fig. 3A, 3C, 3D; Table S1). Levels of eicosanoids were very low in normal rodent RVs and did not exhibit any significant changes with chronic hypoxia (Table S2).

Figure 2.

Median (interquartile range) tissue levels of prostaglandin metabolites prostacyclin (PGI₂; A), thromboxane B₂ (TXB₂; B), and prostaglandin E₂ (PGE₂; D) and the PGI₂/TXB₂ ratio (C) in rat lungs during normoxia and chronic hypoxia.

Figure 3.

Median (interquartile range) tissue levels of 5-hydroxyeicosanoid (5-HETE; A), 8-HETE (B), 12-HETE (C), and 15-HETE (D) in rat lungs during normoxia and chronic hypoxia.

Human plasma lipidomic analyses

Subject characteristics

Twenty-two treatment-naïve PAH patients and 29 PAH patients treated with specific PAH therapies were included in the study. Fifteen healthy volunteers served as controls. The control group was younger and had more males than the PAH groups (Table 1). The majority of PAH patients had IPAH, and among those with CTD, systemic sclerosis was the most predominant etiology. Despite having similar right atrial pressures (RAPs) and mPAPs, treatment-naïve patients had higher PVRs and lower cardiac outputs and indices than treated patients (P = 0.01, 0.02, and 0.04, respectively). Patients and controls were followed for a median of 42 and 34 months, respectively. There were 20 deaths overall, 8 (40%) of whom were patients with CTD-associated PAH. Of the 20, 9 were treatment-naïve patients and 11 treated patients; there were no deaths among control subjects.

Table 1.

Demographic and clinical characteristics of pulmonary arterial hypertension (PAH) patients and control subjects

	Controls (n = 15)	Treatment-naïve PAH patients (n = 22)	Treated PAH patients (n = 29)
Age, years	46 (39–52)	65 (59–71)	63 (56–66)
Female, no. (%)	8 (53)	16 (73)	19 (66)
Etiology, no. (%)
Idiopathic	…	12 (55)	17 (59)
Systemic sclerosis	…	8 (36)	11 (38)
Systemic lupus erythematosis	…	2 (9)	…
Rheumatoid arthritis	…	…	1 (3)
Right atrial pressure, mmHg	…	7 (5–11)	9 (4–13)
Mean PA pressure, mmHg	…	46 (38–56)	47 (35–51)
Pulmonary vascular resistance, WU	…	10 (8–12)	6 (4–9)
Cardiac output, L/min	…	4.0 (3.2–4.6)	4.5 (4.1–6.0)
Cardiac index, L/min/m2	…	2.2 (1.8–2.5)	2.4 (2.2–3.1)
PAH-specific therapy, no. (%)
Prostacyclin	…	…	12 (41)
Endothelin receptor antagonist	…	…	15 (52)
Phosphodiesterase 5 inhibitor	…	…	15 (52)
Riociguat	…	…	2 (7)
Follow-up, months	34 (34–34)	42 (34–49)	41 (18–49)
Death, no. (%)	…	9 (41)	11 (38)

^NoteExcept as noted, results are reported as median (interquartile range). PA: pulmonary artery; WU: Wood units.

PAH is associated with significant alterations in plasma eicosanoid levels that change with PAH-specific treatment

Plasma levels of eicosanoids were not statistically significantly different between patients with IPAH and those with CTD-associated PAH; however, levels of all HETEs were higher in patients with IPAH than in patients with CTD-associated PAH (Tables S4, S5). Plasma levels of PGI₂ were not significantly different among groups; however, plasma levels of TXB₂ were significantly higher in the treatment-naïve PAH patients than in either controls or treated PAH patients (Fig. 4A, 4B). As a consequence, PAH patients had significantly lower PGI₂/TXB₂ ratios than controls. Moreover, treatment-naïve patients had the lowest ratios, significantly lower than those of the treated group (median of 2.1, 0.3, and 1.3 in controls, treatment-naïve, and treated patients, respectively, P = 0.001; Fig. 4C). In addition, treated patients had ratios significantly higher than those of treatment-naïve patients, though still significantly lower than those of controls. Plasma concentrations of PGE₂ were significantly lower in treatment-naïve PAH patients than in either controls or treated PAH patients (Fig. 4D). Levels of other measured PG metabolites were similar between PAH patients and controls (Table S3).

Figure 4.

Median (interquartile range) plasma levels of prostaglandin metabolites prostacyclin (PGI₂; A), thromboxane B₂ (TXB₂; B), and prostaglandin E₂ (PGE₂; D) and the PGI₂/TXB₂ ratio (C) in control subjects, treatment-naïve pulmonary arterial hypertension (PAH) patients, and treated PAH patients.

Treatment-naïve PAH patients had significant elevations in 5-, 8-, 12-, and 15-HETE levels (Fig. 5) compared to controls. The 5- and 8-HETE levels were considerably lower in treated PAH patients; however, 12- and 15-HETE levels, although lower than in treatment-naïve patients, still remained significantly higher in the treated patients than in controls. There was no significant difference in the levels of all HETEs measured between patients who were treated with PGI therapy and those who were not (data not shown). There were no trends observed to suggest decay over time in the levels of eicosanoids due to storage of plasma.

Figure 5.

Median (interquartile range) plasma levels of 5-hydroxyeicosanoid (5-HETE; A), 8-HETE (B), 12-HETE (C), and 15-HETE (D) in control subjects, treatment-naïve pulmonary arterial hypertension (PAH) patients, and treated PAH patients.

Plasma 12- and 15-HETE levels predict outcomes in PAH

Using plasma levels from all PAH patients, we generated ROC curves for PG metabolites (PGI₂, TXB₂, and PGE₂) and HETEs that were significantly different between PAH patients and controls. The best AUCs were those for 12- and 15-HETE, 0.752 and 0.663, respectively (Fig. 6). Combining 12- and 15-HETE did not improve the predictive value (data not shown). PGI₂, TXB₂, and PGE₂ curves had lower performance than 12- and 15-HETEs (AUC: 0.608, 0.605, and 0.531, respectively). Plasma levels of 12-HETE were positively correlated with mPAP and PVR (r = 0.35 and 0.33, P = 0.01 and 0.02, respectively) but not with RAP, PCWP, New York Heart Association functional class (NYHA FC), cardiac output, 6-minute walk distance (6MWD), or BNP levels. Similarly, plasma levels of 15-HETE were positively correlated with mPAP (r = 0.31, P = 0.03) but not with RAP, PVR, PCWP, cardiac output, NYHA FC, 6MWD, or BNP levels.

Figure 6.

Receiver operating characteristics curves constructed to determine the optimal thresholds for human plasma 12-hydroxyeicosanoid (12-HETE; A) and 15-HETE (B) for predicting survival in pulmonary arterial hypertension patients. AUC: area under the curve.

From the ROC analyses, we established threshold plasma levels of 57 pg/mL for 12-HETE and 256 pg/mL for 15-HETE to distinguish the survivors. On the basis of these thresholds, we divided the cohort of PAH patients into either 12-HETE_low (<57 pg/mL) and 12-HETE_high (≥57 pg/mL) or 15-HETE_low (<256 pg/mL) and 15-HETE_high (≥256 pg/mL) groups. Both the 12-HETE_high and 15-HETE_high groups had significantly decreased survival (Fig. 7). Elevated 12- and 15-HETE levels were both associated with a significantly increased risk of death (unadjusted hazard ratio [HR]: 3.4 [95% confidence interval (CI): 1.3–8.9] and 2.5 [1.0–6.2], respectively; Table 2). After adjustment for the REVEAL risk score, both the 12-HETE_high and 15-HETE_high cohorts remained associated with a significantly increased risk of death (adjusted HR: 2.8 [95% CI: 1.1–7.3] and 4.3 [1.6–11.8], respectively; Table 2).

Figure 7.

Pulmonary arterial hypertension (PAH) patient survival according to plasma hydroxyeicosanoid (HETE) levels. A, PAH patients with 12-HETE_low (<57 pg/mL) or 12-HETE_high (≥57 pg/mL). B, PAH patients with 15-HETE_low (<256 pg/mL) or 15-HETE_high (≥256 pg/mL).

Table 2.

Cox proportional-hazards models for 12-HETE_high and 15-HETE_high

	Unadjusted hazard ratio		Adjusted hazard ratioa
	Ratio	P value	Ratio	P value
12-HETEhigh	3.4 (1.3–8.9)	0.01	2.8 (1.1–7.3)	0.04
15-HETEhigh	2.5 (1.0–6.2)	0.04	4.3 (1.6–11.8)	0.004

^NoteResults reported as hazard ratio (95% confidence interval). HETE: hydroxyeicosanoid acid; 12-HETE_high: 12-HETE ≥ 57 pg/mL; 15-HETE_high: 15-HETE ≥ 256 pg/mL.

^aAdjusted for the REVEAL (Registry to Evaluate Early and Long-TermPAH Disease Management) risk score.

Discussion

This is the first study to characterize several biochemically active lipids in both experimental PH models and a well-phenotyped cohort of PAH patients. We demonstrate abnormalities in eicosanoid metabolites in the lungs of experimental rodent PH models and in the plasma of PAH patients, showing both similarities and differences. Moreover, we identified two novel potential biomarkers, 12- and 15-HETE, that were independent predictors of mortality in PAH patients, even after adjustment for known confounders.

PGs in rodent and human pulmonary vascular disease

In response to chronic hypoxia in rodents and despite lack of significant alterations in lung levels of PGI₂ and TXB₂, there was a decrease in PGI₂/TXB₂ ratio, supporting the notion of an imbalance between vasodilatory and vasoconstrictive mediators. These findings support that alterations in AA metabolism may play an active role in the pulmonary vascular response to chronic hypoxia. Our findings are in contrast to results from the Sugen 5416/hypoxia (SuHx) rat model of experimental PH, in which elevated levels of PGE₂ and TXB₂ were found at 3 weeks.³¹ This suggests that alterations in lung eicosanoids depend on the experimental model used and/or the type and severity of PH, as the SuHx rat model has been characterized as an irreversible, more severe form of PH, compared to the reversible model of chronic hypoxia.

In PAH patients, we observed elevated plasma levels of several AA metabolites in treatment-naïve PAH patients, abnormalities that were attenuated in the treated PAH patients. The alterations in PG pathways tended to be more substantial than those in the experimental PH model. The vasodilator PGE₂ changed in the direction opposite that in rodent lungs. PGE₂ was lower in treatment-naïve patients than in controls and not different between the treated patients and controls, while TXB₂ was elevated in treatment-naïve patients and restored in the treated PAH patients. More importantly, PGI₂/TXB₂ was significantly lower in our patients than in controls, and, although it increased in the treated patients, it remained abnormal. These findings are consistent with findings by Christman et al.¹¹ of higher urinary levels of TXB₂ and TXB₂/PGI₂ in 8 PAH patients compared to non-PAH controls, lending further support to the notion that impaired balance between PGI₂ and TXB₂ contributes to the pathogenesis of PAH in humans. In addition, our results suggest that PAH treatment may modify these PGs to partially restore the balance between vasoactive metabolites.

HETEs in rodent and human pulmonary vascular disease

Among the HETEs measured in rodent lungs with experimental PH, 8-HETE was significantly higher with exposure to chronic hypoxia. This is in accordance with increased 8-HETE levels in the SuHx rat model.³¹ Moreover, plasma 8-HETE levels were also higher in treatment-naïve than in treated PAH patients, suggesting that 8-HETE may play a role in pathogenesis of both experimental and human PH disease. It has been shown that 5-HETE plays a role in the monocrotaline experimental PH model;¹⁵ however, in our study, 5-HETE levels in the rat lungs were not affected significantly by chronic hypoxia. Previously, levels of 5-HETE in the lungs from patients with primary and secondary forms of PH were reported to be increased.² In our study, levels of 5-HETE were also elevated in the plasma of treatment-naïve PAH patients and were not different between control subjects and treated PAH patients, suggesting that they were ameliorated by PAH treatment.

By mediating barrier function disruption, neovascularization, and angiogenesis, 12-HETE has been implicated in the pathogenesis of vasculopathies in atherosclerosis, cancer, and diabetic retinopathy.^32-35 It also promotes platelet activation and aggregation,³⁶ and 12-HETE and the 12-LOX pathway have also been shown to play a role in the pathogenesis of the chronic hypoxia and SuHx rat models of PH.^14,31 We previously showed that 12-LOX is upregulated in the lungs of chronically hypoxic rats and that 12-HETE stimulates proliferation of pulmonary artery SMCs through an extracellular signal-regulated kinase–dependent mechanism.¹⁴In our study, although 12-HETE levels tended to increase in the lungs of rats exposed to chronic hypoxia, this increase did not achieve statistical significance. However, the levels of 12-HETE were significantly increased in the plasma of both treatment-naïve and treated PAH patients compared to those in controls.

Despite some conflicting reports, most studies have implicated 15-HETE in the pathogenesis of experimental models of PH through potentiating pulmonary artery SMC proliferation and protecting against apoptosis,^17-19,37 enhancing vasoconstriction during hypoxia via the protein kinase C pathway and downregulation of potassium channels³⁸ and stimulating the expression of angiogenic molecules such as vascular endothelial growth factor and fibroblast growth factor 2, promoting autocrine-mediated angiogenesis.^39,40 Both 15-HETE and 15-LOX have been implicated in platelet activation, pulmonary intravascular thrombosis, angiogenesis, and pulmonary vascular remodeling in experimental PH models.^41-43 Contrary to most studies in rodents, 15-HETE has vasodilatory effects on both the systemic and pulmonary vasculature in rabbits.^20,21 In our study, levels of 15-HETE were increased, although not significantly, in the lungs of rats exposed to chronic hypoxia but were significantly increased in the plasma of both treatment-naïve and treated PAH patients, similar to the changes in 12-HETE levels. These findings are similar to previously reported increased levels of 15-HETE in the SuHx rat model.³¹

In our study, 12- and 15-HETE correlated moderately with PVR and/or mPAP, suggesting a correlation with disease severity. In a prior report, 12- and 15-HETE levels in the lungs of patients with severe PAH were found to be elevated, while patients who were treated chronically with PGI tended to have lower 12- and 15-HETE lung levels.² In our cohort of patients, plasma levels of 12- and 15-HETE were similar between patients who were treated with PGIs and those who were not.

Most interestingly, we showed that plasma levels of 12- and 15-HETE were independent predictors of mortality in our cohort. We elected to use the REVEAL risk score for risk adjustment because it was derived from a large United States–based registry that included IPAH and CTD-associated PAH patients and has been validated in incident and prevalent PAH patients as well as cross validated in the French registry.^30,44,45 Also, the REVEAL risk score is based on the contribution of multiple variables that need not be measured concurrently.⁴⁶ Even after adjustment for the REVEAL risk score, 12- and 15-HETE remained independent predictors of mortality, implicating 12- and 15-HETE as two potential novel biomarkers for PAH. In our study, survival time was assessed from the time the blood samples were obtained in this cohort of newly diagnosed and prevalent patients with PAH. While the survival curves begin to separate after 2 years, the significantly increased risk of death associated with these markers remains an important independent prognostic factor because of its association with long-term outcomes in these patients. Taken together, the above findings suggest that 12- and 15-HETE and their pathways are involved in the pathogenesis of PAH, potentially through increased vasoconstriction, cell proliferation, and enhanced thrombosis of the pulmonary vasculature, and that these pathways may be modified by PAH treatment.

Our study has several limitations. First, we performed targeted lipidomic analyses focused on PGs and HETEs, and we did not evaluate other lipid molecules that could potentially be involved in the pathogenesis of PAH. Although we identified abnormalities in eicosanoid levels in the lungs of experimental PH rodent models and in plasma of human PAH patients, our design did not enable us to understand causal relationships between individual eicosanoids and the development of PH. Another limitation relates to our use of a convenience sample of controls who were not age and sex matched to our PAH patients. While this limits generalizability of our conclusion, our control data offer important insight into the pathobiology of eicosanoid imbalances in PAH, given the lack of data on normal values for eicosanoid levels in human plasma. Finally, our cohort included treatment-naïve and treated PAH patients; however, we did not have serial plasma samples from patients after initiation of PAH-specific therapy.

We conclude that alterations in eicosanoid metabolism are characteristic of both experimental models of PH and human PAH, suggesting a role for circulating eicosanoids in the pathogenesis of pulmonary vascular disease. While there were similarities between the rodent hypoxia model and human PAH, our study found several differences in eicosanoids, which emphasizes the lack of a representative experimental model of PAH. Furthermore, we identify two novel biomarkers that may be helpful in follow-up of PAH patients, assessment of treatment response, and prediction of outcome.

Appendix.

Supplementary Methods

RHC protocol

A Swan-Ganz catheter (Edwards Lifesciences, Irvine, CA) was inserted through an internal jugular vein under ultrasound and fluoroscopic guidance, as previously described.⁴⁷ The transducer was leveled at the midaxillary line. Parameters measured were mRAP, mPAP, PCWP, and cardiac output (thermodilution method). Cardiac index and PVR were calculated with standard formulas. Supplemental oxygen was administered as needed to maintain oxygen saturation of >90%.

Lung and RV lipidomic analyses

The harvested lung tissue was homogenized for 30 seconds in 2 mL of 66% methanol, after which 100 μL was taken for protein quantification. The aliquots were frozen at −80°C for 45 minutes for protein precipitation and then centrifuged at 4,500 rpm for 30 minutes at 4°C. The supernatant was diluted ≥10-fold with ice-cold high-performance liquid chromatography (HPLC)–grade water. The diluted supernatant was acidified on ice to a pH of 4.0 with 1N HCl. With an Agilent 25-μL glass syringe, a 20-μL aliquot of internal standard master mix (5-, 12-, 15-HETE-d₈, PGE₂-d₄, TXB₂-d₄) was added to each sample (total of 2 ng). The acidified supernatant containing the internal standard was loaded onto an Agilent C18 6-mL, 500-mg cartridge that was conditioned with 14 mL methanol, 14 mL water by low feeding speed in sequence. After the addition of the sample, the cartridge was washed with 14 mL of HPLC-grade water to remove any unabsorbed material and with 14 mL HPLC-grade ice-cold hexane to remove any interfering lipids. The analytes were collected in glass Pyrex tubes by elution with 7 mL HPLC-grade methanol. The solvent was evaporated under a gentle stream of nitrogen. The residue was reconstituted with 100 μL mobile phase, vortexed briefly, and transferred to an autosampler vial insert for LC/MS/MS analysis. All extraction procedures were performed under minimum-light conditions to avoid photodegradation of AA metabolites. Protein concentrations were determined with the Bradford assay.

Lipidomic analysis of human plasma

The plasma was acidified, and then a 20-μL aliquot of internal standard master mix (5-, 12-, 15-HETE-d₈, PGE₂-d₄, TXB₂-d₄) was added to each sample (total of 2 ng) and loaded onto an Agilent C18 6-mL cartridge. The analytes were collected in glass Pyrex tubes. The solvent was evaporated under a gentle stream of nitrogen. The residue was reconstituted with 100 μL mobile phase, vortexed briefly, and transferred to an autosampler vial insert for LC/MS/MS analysis. All extraction procedures were performed under minimum-light conditions to avoid photodegradation of AA metabolites. Extraction efficiencies were 70.3% ± 17.0%.

Analysis of lipid metabolites in tissue and plasma

HPLC simultaneous separation of AA metabolites

A Phenomenex Luna 5-micron C18 (Phenomenex, Torrance, CA; 150 mm × 2.00 mm × 5 µm) was used for HPLC.²⁸ The Agilent 1200 Series HPLC system consisted of a high-performance binary pump, a 108-well plate autosampler at 4°C, and a column compartment set at 25°C. The solvent system consisted of 100% methanol and 10 mM ammonium acetate, pH 8.5. Before each run, the analytical column was preequilibrated with methanol for 60 seconds, and the injection syringe was washed for 8 seconds in the flush port. Twenty-five microliters of the sample was drawn up at a speed of 200 μL/minute with a 100-μL syringe and ejected onto the column at a rate of 400 μL/min. The flow rate of solvent through the column was 400 μL/min. The analytical column was primed with 100% ammonium acetate for 0.11 minutes and immediately switched to 50% methanol and 50% ammonium acetate at 5 minutes. At 5 minutes, methanol was programmed to increase linearly from 50% to 100% over a period of 5 minutes, after which the flow remained at 100% methanol for 20 minutes to ensure elution of all analytes. At 30.10 minutes, the flow immediately switched to 100% ammonium acetate for the remainder of the 35-minute run, to flush the methanol entirely from the column. In order to get a complete separation of HETEs and PGs in the same run and under the same conditions, the column was flushed with 100% ammonium acetate for the last 4.90 minutes of each run.

MS/MS-targeted profiling of HETEs and PGs

Targeted profiling of HETEs and PGs was performed with a 5500 QTRAP (ABSciex) hybrid triple quadrupole linear ion trap mass spectrometer equipped with a turbo ion spray ionization source.^28,48 Multiple-reaction monitoring (MRM) experiments were performed, using negative ion spray mode. Detection of eicosanoids was separated into two periods, and the dwell time used for MRM experiments was 400 milliseconds for HETEs and 120 milliseconds for PGs and TXB₂. The collision exit potential was −11 V for HETEs and −13 V for PGs and TXB₂. The declustering potential was −250 V for HETEs and −150 V for PGs and TXB₂. The source-dependent MS parameters temperature and ion-spray voltage were set at 350°C and −4,500 V, respectively. Collision energy was determined for each specific compound. With a Valco valve diverter, analytes were sent to waste for the first 8 minutes of the run, then sent to the mass spectrometer, where they were collected for 22 minutes of the run, and then diverted back to waste for the remainder of the run.

Materials for LC/MS/MS

The following compounds were purchased from Cayman Chemical (Ann Arbor, MI): 5-hydroxy-6E,8Z,11Z,14Z-eicosatetraenoic acid (5-HETE); 8-hydroxy-5Z,9E,11Z,14Z-eicosatetraenoic acid (8-HETE); 12-hydroxy-5Z,8Z,10E,14Z-eicosatetraenoic acid (12-HETE); 15-hydroxy-5Z,8Z,11Z,13E-eicosatetraenoic acid (15-HETE); 9-oxo-11R,15S-dihydroxy-5Z,13E-prostadienoic acid (PGE₂); 9-oxo-11S,15S-dihydroxy-5Z,13E-prostadienoic acid (11β-PGE₂); 9,15-dioxo-11R-hydroxy-5Z,13E-prostadienoic acid (15-keto-PGE₂); 9,15-dioxo-11R-hydroxy-5Z-prostenoic acid (13,14-dihydro-15-keto-PGE₂); 9S,15S-dihydroxy-11-oxo-5Z,13E-prostadienoic acid (PGD₂); 11,15-dioxo-9S-hydroxy-5Z-prostenoic acid (13,14-dihydro-15-keto-PGD₂); 9S,11R,15S-trihydroxy-5Z,13E-prostadienoic acid (PGF_2α); 9S,11S,15S-trihydroxy-5Z,13E-prostadienoic acid (11β-PGF_2α); 6-oxo-9S,11R,15S-trihydroxy-13E-prostenoic acid (6-keto-PGF_1α); 9S,11,15S-trihydroxy-thromboxa-5Z,13E-dien-1-oic acid (TXB₂); 9-oxo-11R,15S-dihydroxy-5Z,13E-prostadienoic acid-cyclo[8S,12R] (8-iso-PGE₂), 9S, 11S, 15S-trihydroxy-(8β)-prosta-5Z, 13E-dien-1-oic acid (8-iso-PGF_2α); 9S,11R-dihydroxy-15-oxo-5Z-prostaenoic acid-cyclo[8S,12R] (8-iso-13,14-dihydro-15-keto-PGF_2α).

The internal standards mixture (Cayman Chemical) consisted of 5S-hydroxy-6E,8Z,11Z,14Z-eicosatetraenoic acid (5,6,8,9,11,12,14,15-d8) [5(S)-HETE-d₈]; 12S-hydroxy-5Z,8Z,10E,14Z-eicosatetraenoic acid-d8 [12(S)-HETE-d₈]; 15S-hydroxy-5Z,8Z,11Z,13E-eicosatetraenoic acid-d8 [15(S)-HETE-d₈]; 11R,15S-dihydroxy-9-oxo-5Z,13E-prostadienoic acid (3,3,4,4-d4) (PGE₂-d₄); and 9S,11,15S-trihydroxy-thromboxa-5Z,13E-dien-1-oic acid-d4 (TXB₂-d₄). Gradient-grade methanol, water, hexane, and ammonium acetate for HPLC ≥ 99.9% were purchased from Sigma-Aldrich (St. Louis, MO).

REVEAL risk score parameters

The REVEAL risk score includes 19 variables, including age, sex, World Health Organization (WHO) group 1 subgroup, NYHA FC, heart rate, systolic blood pressure, 6MWD, BNP, echocardiogram findings, lung diffusion capacity, mRAP, and PVR.^30,44

Supplementary tables

Table S1.

Levels of eicosanoids in rat lungs

	Normoxia (n = 22)	Chronic hypoxia (n = 14)	P valuea
PGE2	0.03 (0.02–0.12)	0.11 (0.003–0.15)	0.94
11β-PGE2	0.06 (0–0.10)	0 (0–0.22)	0.80
15-keto-PGE2	ND	ND	…
13,14-dihydro-15-keto-PGE2	ND	ND	…
PGD2	0 (0–0.02)	0 (0–0)	…
13,14-dihydro-15-keto-PGD2	ND	ND	…
PGF2α	ND	ND	…
11β-PGF2α	ND	ND	…
6-keto-PGF1α	0.04 (0–0.23)	0 (0–0)	0.07
2,3-dinor-6-keto-PGF1α	0.06 (0–0.11)	0.02 (0–0.37)	0.76
PGI2	0.20 (0.04–0.41)	0.05 (0–0.49)	0.59
TXB2	0.16 (0.09–0.41)	0.48 (0.11–0.96)	0.18
PGI2/TXB2	0.95 (0.25–1.60)	0.17 (0–0.74)	0.04
8-iso-PGE2	0.0003 (0–0.0009)	0.002 (0–0.003)	0.23
8-iso-PGF2α	ND	ND	…
8-iso-13,14-dihydro-15-keto-PGF2α	ND	ND	…
5-HETE	0.02 (0.01–0.04)	0.03 (0.01–0.04)	0.47
8-HETE	0.043 (0.02–0.06)	0.08 (0.06–0.14)	0.02
12-HETE	0.45 (0.29–1.42)	0.72 (0.47–1.90)	0.11
15-HETE	0.06 (0.04–0.15)	0.14 (0.05–0.19)	0.24

^NoteResults presented as median (interquartile range); except for the ratio PGI₂/TXB₂, the unit of measure is pg/μg of protein. PGE: prostaglandin E; PGD: prostaglandin D; PGF: prostaglandin F; PGI: prostacyclin; TX: thromboxane; HETE: hydroxyeicosanoid; ND: not detected.

^aFrom Wilcoxon rank-sum test.

Table S2.

Levels of eicosanoids and HETEs in rat right ventricles

	Normoxia (n = 13)	Chronic hypoxia (n = 12)	P valuea
PGE2	0.03 (0.03–0.03)	0.03 (0.02–0.03)	0.02
11β-PGE2	ND	ND	…
15-keto-PGE2	ND	ND	…
13,14-dihydro-15-keto-PGE2	ND	ND	…
PGD2	ND	ND	…
13,14-dihydro-15-keto-PGD2	ND	ND	…
PGF2α	ND	ND	…
11β-PGF2α	ND	ND	…
6-keto-PGF1α	ND	ND	…
2,3-dinor-6-keto-PGF1α	0.0003 (0–0.0004)	0 (0–0)	0.004
PGI2	0.0003 (0–0.001)	0 (0–0)	0.05
TXB2	0.009 (0.005–0.03)	0.02 (0.01–0.02)	0.27
PGI2/TXB2	0.04 (0–0.06)	ND	0.03
8-iso-PGE2	ND	ND	…
8-iso-PGF2α	ND	ND	…
8-iso-13,14-dihydro-15-keto-PGF2α	ND	ND	…
5-HETE	0.02 (0.01–0.07)	0.02 (0.02–0.05)	0.77
8-HETE	0.03 (0.02–0.04)	0.03 (0.03–0.07)	0.38
12-HETE	0.31 (0.21–0.40)	0.22 (0.12–0.53)	0.94
15-HETE	0.03 (0.02–0.05)	0.05 (0.03–0.09)	0.03

^NoteResults presented as median (interquartile range); except for the ratio PGI₂/TXB₂, the unit of measure is pg/μg of protein. PGE: prostaglandin E; PGD: prostaglandin D; PGF: prostaglandin F; PGI: prostacyclin; TX: thromboxane; HETE: hydroxyeicosanoid; ND: not detected.

^aFrom Wilcoxon rank-sum test.

Table S3.

Human plasma eicosanoid levels

	Controls (n = 15)	Treatment-naïve PAH patients (n = 22)	Treated PAH patients (n = 29)	P valuea
PGE2	19.7 (14.7–24.6)	6.6 (3.8–14.7)	18.8 (7.0–44.7)	0.02
11β-PGE2	ND	0 (0–49.0)	0 (0–86.1)	0.39
15-keto-PGE2	0 (0–0)	0 (0–3.6)	0 (0–3.4)	0.49
PGF2α	0 (0–5.5)	ND	0 (0–8.0)	0.23
6-keto-PGF1α	150.2 (76.7–296.9)	36.0 (2.7–108.4)	49.7 (21.4–111.9)	0.005
2,3-dinor-6-keto-PGF1α	69.1 (35.6–127.0)	37.2 (2.3–104.4)	68.2 (10.8–126.6)	0.57
PGI2	221.5 (156.4–388.8)	73.6 (24.9–305.6)	145.0 (27.0–224.3)	0.08
TXB2	127.4 (106.7–167.2)	289.6 (195.2–577.8)	140.5 (56.3–207.4)	0.002
PGI2/TXB2	2.1 (1.5–2.2)	0.3 (0.1–1.3)	1.3 (0.7–2.1)	0.001
8-iso-PGE2	ND	0.5 (0.1–1.7)	0.4 (0–0.7)	0.009
8-iso-PGF2α	0 (0–4.8)	ND	3.4 (0–12.1)	0.048
PGD3	ND	ND	ND	…
5-HETE	17.0 (3.2–37.5)	192.9 (10.1–4092.0)	23.1 (8.2–178.5)	0.05
8-HETE	6.8 (2.7–31.9)	65.0 (11.3–1014.0)	14.2 (7.5–89.0)	0.05
12-HETE	20.1 (11.0–30.8)	104.1 (25.4–1508.0)	25.8 (18.0–277.5)	0.02
15-HETE	5.2 (4.0–11.4)	127.0 (5.5–414.0)	20.9 (14.9–38.9)	0.01

^NoteResults presented as median (interquartile range); except for the ratio PGI₂/TXB₂, the unit of measure is pg/mL. PAH: pulmonary arterial hypertension; PGE: prostaglandin E; PGD: prostaglandin D; PGF: prostaglandin F; PGI: prostacyclin; TX: thromboxane; HETE: hydroxyeicosanoid; ND: not detected.

^aFrom Kruskal-Wallis test.

Table S4.

Human plasma eicosanoid levels of treatment-naïve PAH patients

	IPAH (n = 12)	CTD-PAH (n = 10)	P valuea
PGE2	6.6 (3.8–16.2)	6.0 (3.8–9.6)	0.82
11β-PGE2	0 (0–41.8)	0.9 (0–45.1)	0.83
15-keto-PGE2	0 (0–1.9)	0 (0–5.7)	0.85
PGF2α	0 (0–0.9)	0 (0–0)	0.43
6-keto-PGF1α	26.5 (0.6–50.5)	64.5 (15.6–183.9)	0.41
2,3-dinor-6-keto-PGF1α	15.5 (0–54.9)	83.5 (18.0–201.7)	0.06
PGI2	56.5 (18.5–122.4)	184.8 (55.2–380.7)	0.29
TXB2	257.0 (186.1–499.5)	323.5 (213.3–596.7)	0.58
PGI2/TXB2	0.1 (0.06–1.1)	0.7 (0.2–1.3)	0.39
8-iso-PGE2	0.2 (0–0.7)	1.3 (0.3–2.3)	0.09
8-iso-PGF2α	0 (0–0.8)	0 (0–0)	0.49
PGD3	0 (0–68.7)	0 (0–0)	0.89
5-HETE	254.7 (16.7–3,882.0)	93.2 (7.4–3,350.0)	0.63
8-HETE	281.6 (19.1–1,062.0)	29.3 (4.5–875.6)	0.58
12-HETE	471.1 (46.1–1,552.0)	42.2 (11.1–1,282.0)	0.46
15-HETE	183.0 (12.2–463.8)	23.2 (2.1–284.0)	0.35

^NoteResults presented as median (interquartile range); except for the ratio PGI₂/TXB₂, the unit of measure is pg/mL. PAH: pulmonary arterial hypertension; IPAH: idiopathic PAH; CTD-PAH: connective-tissue disease–associated PAH; PGE: prostaglandin E; PGD: prostaglandin D; PGF: prostaglandin F; PGI: prostacyclin; TX: thromboxane; HETE: hydroxyeicosanoid.

^aFrom Wilcoxon rank-sum test.

Table S5.

Human plasma eicosanoid levels of treated PAH patients

	IPAH (n = 17)	CTD-PAH (n = 12)	P valuea
PGE2	18.8 (4.0–44.7)	21.3 (10.1–39.4)	0.81
11β-PGE2	0 (0–71.2)	23.4 (0–86.9)	0.72
15-keto-PGE2	0 (0–3.6)	0 (0–0)	0.22
PGF2α	0 (0–8.0)	0 (0–6.9)	0.70
6-keto-PGF1α	46.9 (7.4–93.3)	55.3 (35.4–112.1)	0.53
2,3-dinor-6-keto-PGF1α	51.8 (10.8–89.3)	94.9 (31.4–171.4)	0.54
PGI2	104.1 (22.9–165.6)	202.4 (69.4–239.9)	0.35
TXB2	140.5 (56.3–243.1)	127.7 (59.8–192.7)	0.91
PGI2/TXB2	1.0 (0.7–2.1)	1.3 (1.2–1.7)	0.42
8-iso-PGE2	0.6 (0–0.7)	0.1 (0–0.5)	0.19
8-iso-PGF2α	3.4 (0–14.9)	3.8 (0–10.1)	0.83
PGD3	0 (0–0)	97.4 (0–620.8)	0.15
5-HETE	38.8 (13.9–662.7)	16.1 (6.6–66.6)	0.35
8-HETE	22.9 (13.2–451.6)	8.7 (6.9–16.2)	0.14
12-HETE	47.5 (24.9–563.2)	18.7 (14.8–28.8)	0.02
15-HETE	20.1 (14.9–182.2)	21.0 (17.0–38.0)	0.99

^NoteResults presented as median (interquartile range); except for the ratio PGI₂/TXB₂, the unit of measure is pg/mL. PAH: pulmonary arterial hypertension; IPAH: idiopathic PAH; CTD-PAH: connective-tissue disease–associated PAH; PGE: prostaglandin E; PGD: prostaglandin D; PGF: prostaglandin F; PGI: prostacyclin; TX: thromboxane; HETE: hydroxyeicosanoid.

^aFrom Wilcoxon rank-sum test.

Source of Support: NA is supported by the National Center for Advancing Translational Sciences, National Institutes of Health grants UL1 TR001064 and TL1 TR001062. IRP was supported by National Heart, Lung, and Blood Institute grant K08 HL7734.

Conflict of Interest: NSH and IRP report grants and personal fees from Actelion, Bayer, Gilead, and United Therapeutics unrelated to the submitted work. All other authors report no conflict of interest.

Supplements

• Appendix ^{(455.9KB, pdf)}