Kynurenine pathway metabolism evolves with development of preclinical and scleroderma-associated pulmonary arterial hypertension

Abstract

Understanding metabolic evolution underlying pulmonary arterial hypertension (PAH) development may clarify pathobiology and reveal disease-specific biomarkers. Patients with systemic sclerosis (SSc) are regularly surveilled for PAH, presenting an opportunity to examine metabolic change as disease develops in an at-risk cohort. We performed mass spectrometry-based metabolomics on longitudinal serum samples collected before and near SSc-PAH diagnosis, compared with time-matched SSc subjects without PAH, in a SSc surveillance cohort. We validated metabolic differences in a second cohort and determined metabolite-phenotype relationships. In parallel, we performed serial metabolomic and hemodynamic assessments as the disease developed in a preclinical model. For differentially expressed metabolites, we investigated corresponding gene expression in human and rodent PAH lungs. Kynurenine and its ratio to tryptophan (kyn/trp) increased over the surveillance period in patients with SSc who developed PAH. Higher kyn/trp measured two years before diagnostic right heart catheterization increased the odds of SSc-PAH diagnosis (OR 1.57, 95% CI 1.05–2.36, P = 0.028). The slope of kyn/trp rise during SSc surveillance predicted PAH development and mortality. In both clinical and experimental PAH, higher kynurenine pathway metabolites correlated with adverse pulmonary vascular and RV measurements. In human and rodent PAH lungs, expression of TDO2, which encodes tryptophan 2,3 dioxygenase (TDO), a protein that catalyzes tryptophan conversion to kynurenine, was significantly upregulated and tightly correlated with pulmonary hypertensive features. Upregulated kynurenine pathway metabolism occurs early in PAH, localizes to the lung, and may be modulated by TDO2. Kynurenine pathway metabolites may be candidate PAH biomarkers and TDO warrants exploration as a potential novel therapeutic target.

NEW & NOTEWORTHY Our study shows an early increase in kynurenine pathway metabolism in at-risk subjects with systemic sclerosis who develop pulmonary arterial hypertension (PAH). We show that kynurenine pathway upregulation precedes clinical diagnosis and that this metabolic shift is associated with increased disease severity and shorter survival times. We also show that gene expression of TDO2, an enzyme that generates kynurenine from tryptophan, rises with PAH development.

INTRODUCTION

The molecular determinants of pulmonary arterial hypertension (PAH) development and progression remain uncertain, though altered metabolism is a defining feature of PAH pathobiology, and metabolic disturbances may contribute to disease development and progression (1–3). PAH affects 10%–12% (4, 5) of patients with systemic sclerosis (SSc) and is a leading cause of death in SSc (6–8). Patients with SSc therefore comprise an at-risk population in which PAH screening is regularly conducted (9, 10). PAH surveillance in SSc represents a unique opportunity to investigate evolving metabolic variation related to PAH development, enabling the identification of fundamental pathobiological processes and molecular targets. We hypothesized that metabolic disturbances would evolve before PAH diagnosis in SSc, and that early metabolic evolution would relate to risk of disease development, disease severity, and outcomes.

In this study, we examined metabolic change over time in an SSc cohort longitudinally monitored for the development of PAH, with samples analyzed before and near to the time of PAH diagnosis. Metabolic differences in PAH were validated in a second multicenter cohort, the pulmonary vascular disease phenomics (PVDOMICS) cohort, with cross-sectional and longitudinal clinical data. To recapitulate metabolic and hemodynamic evolution in a robust animal model that would enable access to heart and lung tissues, we characterized PAH progression in the Sugen hypoxia (SuHx) rodent model at predefined time intervals and performed metabolomic analysis in plasma and tissue. For metabolites implicated in our analyses, we analyzed clinical and preclinical transcriptomic data to query differential gene expression corresponding to the observed metabolic differences.

METHODS

Full methods are available in the online Supplemental Methods (all Supplemental material is available at https://doi.org/10.6084/m9.figshare.23289872).

Clinical Samples

All subjects provided written informed consent before enrollment into research protocols approved by the Johns Hopkins University Institutional Review Board. We searched the Johns Hopkins Scleroderma Center (JHSC) Research Registry to identify subjects with SSc who developed PAH while under surveillance and contributing serial serum samples to the JHSC biorepository. Eighty-one subjects with one “proximate sample” within 12 mo of PAH diagnosis by right heart catheterization (RHC), and a prior “distant sample” within 5 years before PAH diagnosis were included as cases. Eighty-one SSc comparators were identified from among subjects who contributed at least two serial samples to the biorepository but did not develop PAH while under routine surveillance. Comparators were matched one to one to cases on the basis of age, sex, and duration of SSc at time of serum contribution.

Sixty-two subjects with SSc-PAH, 19 SSc comparators without PAH, and 85 healthy controls without SSc in the pulmonary vascular disease phenomics (PVDOMICS) program cohort (NCT02980887) were examined for validation of metabolic differences and assessment of metabolite-phenotype associations. PVDOMICS subjects underwent detailed physiologic phenotyping, including broad-based metabolomics (Metabolon, Morrisville, NC; 11, 12), invasive hemodynamic measurements, echocardiography, and cardiac MRI (13–15). A subset has paired mixed venous and wedged samples for analysis of transpulmonary metabolic gradients.

Animal Samples

Male Wistar rats received Sugen or vehicle followed by chronic hypoxia or normoxia for 3 wk. Six SuHx animals and six sham animals were euthanized at predefined timepoints (7, 14, and 21 days post-Sugen) to create a time course. Before animals were euthanized, in vivo invasive hemodynamics were obtained, plasma was collected, and heart and lung tissues were harvested. Animal protocols were approved by the Johns Hopkins Animal Care and Use Committee (A3272-01).

Metabolomic Analysis

Metabolomics experiments were performed by the Johns Hopkins Molecular Determinants Core at All Children’s Hospital for JHSC and rodent samples and by Metabolon for PVDOMICS samples. Metabolite data were normalized and scaled as appropriate. Univariable comparisons were made using fold-change (FC) analyses, t tests, and rank-sum tests, as appropriate. Logistic regressions measured the odds of PAH associated with metabolite abundance and metabolite change over time. Metabolites that surpassed a Benjamini–Hochberg false discovery rate (FDR) correction for multiple comparisons (q ≤ 0.05) in the JHSC surveillance cohort were investigated further. Correlations and linear regressions were performed to analyze relationships with clinical variables. Cox models measured relationships with transplant-free survival.

Transcriptomic Analysis

Real-time reverse transcription polymerase chain reactions were performed to investigate gene expression in SuHx heart and lung tissues. Gene expression in PAH versus non-PAH lung tissue was examined in two independent datasets in the NCBI Gene Expression Omnibus (GEO) database (16) [GSE53408 (17) and GSE113439 (18)]. Differential gene expression was investigated using FC analyses and t tests of differences. The Benjamini–Hochberg FDR correction was applied to determine the significance of differences in gene expression in clinical datasets. Relationships between gene expression levels and other variables were assessed using Pearson or Spearman correlation, as appropriate.

Statistical analyses were performed using R Statistical Software (v4.1.2; R Core Team 2021) via the MetaboAnalyst package, the GEO2R package, and Stata Statistical Software (v17, College Station, TX).

RESULTS

Demographics and clinical characteristics are provided for each cohort in Table 1. There were no significant differences in demographics between SSc-PAH cases versus SSc comparators in either the JHSC or the PVDOMICS cohort. In both cohorts, subjects with SSc-PAH tended to be white, female, and 60–65 yr old, with moderate elevations in mPAP and PVR, and preserved cardiac output and index. The JHSC cohort captured incident, treatment-naïve patients with SSc-PAH, whereas the PVDOMICS cohort enrolled mostly prevalent patients who were then followed prospectively over time. The median duration of PAH at PVDOMICS enrollment was 2.35 years [interquartile range (IQR) 0.05–4.75 years], with only 10% of subjects with PAH enrolled within 6 mo of diagnosis. Among prevalent patients with SSc-PAH, at the time of enrollment, 76% were on a phosphodiesterase-5 inhibitor, 51% were on an endothelin receptor antagonist, and 19% were on prostacyclin therapy, with the majority of subjects receiving two or more pulmonary hypertension (PH)-specific medications in combination.

Table 1.

Demographic and Clinical Characteristics of SSc-PAH, SSc Comparators Without PH, and Healthy Controls

	JHSC Cohort		PVDOMICS Subcohort
Characteristic	Case	Comparator	Control	Case	Comparator
Demographics
Subjects, n	70	80	85	62	19
Age, yr	61 (13)	57 (14)	49 (14)	65 (10)	67 (8)
Sex, n (% female)	62 (89)	70 (88)	60 (71)	45 (73)	17 (89)
Race, n (% white)	51 (73)	66 (83)	75 (88)	50 (81)	16 (84)
BMI, kg/m2	25 (4)	27 (5)	28 (6)	27 (6)	26 (6)
NYHA FC, n (%III/IV)	—	—	—	35/61	27/61
6MWD, m	—	—	534 (99)	338 (102)	375 (113)
Laboratory values
Cr, mg/dL	1.06 (0.70)	0.91 (0.36)	0.83 (0.15)	1.01 (0.49)	0.95 (0.39)
proBNP, pg/mL (median, IQR)	746 (414, 3,924)*	183 (91, 347)	69 (69)	2,105 (4,722)*	412 (1,097)
Hemodynamics
RAP, mmHg	8 (5)	—	—	6 (5)	4 (2)
mPAP, mmHg	39 (12)	—	—	38 (11)*	17 (3)
PAWP, mmHg	11 (5)	—	—	10 (5)	7 (3)
PVR, Wood units	8 (5)	—	—	6.2 (4.4)*	2.1 (0.9)
Cardiac output, L/min	3.9 (1.0)	—	—	5.1 (1.9)	5.1 (1.9)
Cardiac index, L/min/m2	2.3 (0.6)	—	—	2.8 (1.0)	2.8 (0.9)
PA Sat %	—	—	—	66.5 (8.8)	70.3 (6.1)
Pulmonary Function Tests
FEV1, % predicted	—	—	101 (15)	76 (16)*	97 (22)
FVC, % predicted	69 (23)	80 (21)	104 (13)	77 (18)*	98 (20)
FEV1/FVC, %	—	—	79 (7)	79 (9)*	78 (6)
DLCO, % predicted	49 (17)	61 (25)	90 (17)	34 (13)*	64 (21)

Values are mean (SD) unless otherwise specified. For the JHSC cohort, clinical characteristics reflect the timepoint nearest PAH diagnosis for matched subjects with sufficient serum for metabolomics. JHSC cases have SSc-PAH, and comparators are SSc subjects without PAH. Eleven JHSC cases and 1 comparator did not have sufficient serum for metabolomics. For the PVDOMICS subcohort, clinical characteristics reflect cohort enrollment. Controls are healthy controls, cases have SSc-PAH, and comparators are SSc subjects without PAH. BMI, body mass index; Cr, creatinine; DLCO, diffusing capacity of the lung for carbon monoxide; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity; JHSC, Johns Hopkins Scleroderma Center; mPAP, mean pulmonary arterial pressure; NYHA FC, New York Heart Association Functional Class; PAH, pulmonary arterial hypertension; PA Sat, pulmonary arterial saturation; PAWP, pulmonary artery wedge pressure; proBNP, brain natriuretic peptide pro-hormone; PVDOMICS, pulmonary vascular disease phenomics; PVR, pulmonary vascular resistance; RAP, right atrial pressure; 6MWD, 6-min walk distance; SSc, systemic sclerosis. *Statistically significant differences. ^— Data are not available.

Clinical Metabolomics

In the JHSC surveillance cohort, by the time of PAH diagnosis, nine metabolite features were significantly higher in patients with SSc-PAH than in patients with SSc without PH, with four surpassing the FDR correction for multiple testing. Kynurenine (FC 1.49, P = 1.49 × 10⁻⁴), a product of tryptophan catabolism via enzymatic action of indoleamine 2,3 dioxygenase (IDO) or tryptophan 2,3 dioxygenase (TDO), and the kynurenine to tryptophan ratio (kyn/trp) (FC 1.49, P = 8.26 × 10⁻⁵), a surrogate for the rate-limiting step in kynurenine pathway metabolism, exhibited the largest significant FC differences between SSc-PAH compared with SSc without PH. The four features that surpassed the FDR-corrected significance threshold (kyn/trp, kynurenine, N-acetylputrescine, and 1-methyladenosine) were all significantly higher in SSc-PAH compared with both SSc comparators and healthy controls in the PVDOMICS cohort (Fig. 1 and Supplemental Tables S1 and S2).

Figure 1.

Volcano plots depict the fold-change (FC; x-axis) and significance (y-axis) of metabolite differences in SSc-PAH vs. SSc without PH (A) and healthy controls (B) in PVDOMICS. Dots represent individual metabolite features. Features up in SSc-PAH are in red, whereas features down in SSc-PAH are blue. Features in common with the surveillance cohort that surpassed a FDR correction for multiple testing are denoted in purple. Features in common with the surveillance cohort that did not surpass the multiple testing threshold are denoted in green. FDR, false discovery rate; PAH, pulmonary arterial hypertension; SSc, systemic sclerosis.

Fold-change evolution over time for metabolites higher at SSc-PAH diagnosis is depicted in Fig. 2. The largest absolute fold-change difference over time occurred in kynurenine, which was 16% higher in cases 2.18 years before diagnosis (IQR 1.11–2.77 years), and 49% higher near PAH diagnosis (within 90 days, IQR 35–208 days). With normalization for time interval variation, kyn/trp (P = 0.004), N-acetylputrescine (P = 0.028), and 1-methyladenosine (P = 0.019) rose faster in subjects with SSc who went on to develop PAH, compared with subjects with SSc who would not (Supplemental Fig. S1). Kynurenine and 1- and 3-methylhistidine also rose more steeply in cases, however, these differences were borderline significant. For kyn/trp, kynurenine, and N-acetylputrescine, the rate of rise over time predicted risk of PAH development and mortality (Table 2). When sampled 2 years before RHC, each standard deviation higher kyn/trp increased the odds of an eventual SSc-PAH diagnosis by 57% (OR 1.57, 95% CI 1.05–2.36, P = 0.028) (Supplemental Table S3). In both cohorts, higher kynurenine, kyn/trp, and N-acetylputrescine at diagnosis were associated with greater mortality risk (Table 2 and Supplemental Table S4).

Figure 2.

Fold-change differences in SSc-PAH at Time 1 (the distant timepoint, 2.18 years before PAH diagnosis) and Time 2 (the proximate timepoint, within 90 days of PAH diagnosis) for the nine metabolites significantly higher by the time of diagnosis in the local surveillance cohort. PAH, pulmonary arterial hypertension; SSc, systemic sclerosis.

Table 2.

Associations between absolute metabolite values and rate of metabolite rise and diagnostic and prognostic risk in the JHSC cohort

	Metabolite Abundance at Diagnosis		Metabolite Slope Change Pre-Diagnosis
Metabolite Feature	OR for PAH	HR for death	OR for PAH	HR for death
Kynurenine	2.98 (1.65–5.41, <0.001)	1.24 (0.98–1.58, 0.074)	2.43 (1.07–5.52, 0.034)	1.46 (1.07–2.00, 0.018)
Kyn/trp	2.66 (1.48–4.77, 0.001)	1.51 (1.27–1.81, <0.001)	3.00 (1.21–7.45, 0.018)	1.81 (1.35–2.44, <0.001)
N-acetylputrescine	2.32 (1.41–3.83, 0.001)	1.36 (1.10–1.69 (0.004)	3.08 (1.35–7.04, 0.008)	1.23 (1.03–1.46, 0.020)
1-methyladenosine	2.03 (1.21–3.42, 0.007)	1.18 (0.91–1.53, 0.205)	2.29 (1.08–4.86 (0.030)	1.36 (0.90–2.03, 0.142)
1-methylhistidine	2.52 (1.38–4.59, 0.003)	1.31 (10.2–1.69, 0.037)	1.71 (0.91–3.22 (0.093)	1.10 (0.80–1.53, 0.556)
3-methylhistidine	2.59 (1.40–4.67, 0.002)	1.27 (0.99–1.65, 0.062)	1.70 (0.90–3.18 (0.099)	1.11 (0.80–1.54, 0.523)

Data are presented as OR or HR (95% CI, P value) and reflect logistic or Cox regressions performed in the local surveillance cohort. Regressions are adjusted for age and sex. Metabolite data normalized by mean centering and dividing by standard deviation for each feature. HR, hazard ratio; Kyn/trp, kynurenine and its ratio to tryptophan; PAH, pulmonary arterial hypertension.

Metabolite correlations with clinical parameters in PVDOMICs are listed in Table 3. Kyn/trp, kynurenine, N-acetylputrescine, and 1-methyladenosine all correlated positively with right atrial pressure (RAP), mean pulmonary arterial pressure (mPAP), and pulmonary vascular resistance (PVR), and negatively with pulmonary arterial compliance, echocardiographic tricuspid annular plane systolic excursion (TAPSE), and MRI RV ejection fraction (RVEF). Our metabolites of interest were also significantly correlated with several commonly used multivariable PAH risk models (19, 20). There were no associations with pulmonary capillary wedge pressure or cardiac output or index. Kynurenine and kyn/trp relationships with RAP, mPAP, and PVR in patients with versus without PAH are plotted in Supplemental Fig. S2.

Table 3.

Correlations between significant metabolite features and clinical variables in PVDOMICS

	Kynurenine		Kyn/Trp		N-Acetylputrescine		1-Methyladenosine
	Coefficient	P Value	Coefficient	P Value	Coefficient	P Value	Coefficient	P Value
Hemodynamic Variables
RAP	0.29	<0.01	0.22	0.04	0.29	<0.01	0.24	0.03
mPAP	0.44	<0.01	0.42	<0.01	0.41	<0.01	0.42	<0.01
PCWP	0.14	0.21	0.14	0.22	0.11	0.32	0.13	0.25
PVR	0.36	<0.01	0.42	<0.01	0.40	<0.01	0.35	<0.01
CO	−0.02	0.83	−0.12	0.27	−0.19	0.09	−0.09	0.44
CI	−0.02	0.83	−0.03	0.78	−0.17	0.12	−0.15	0.17
Stroke volume	0.04	0.71	−0.13	0.24	−0.18	0.10	−0.13	0.25
PA compliance	−0.27	0.01	−0.36	<0.01	−0.36	<0.01	−0.32	<0.01
RV functional variables
TAPSE	−0.34	<0.01	−0.41	<0.01	−0.36	<0.01	−0.29	<0.01
RVEF	−0.28	<0.01	−0.35	<0.01	−0.24	0.01	−0.32	0.01
Multivariable risk models
COMPERA 2.0 4-strata model	0.35	<0.01	0.43	<0.01	0.39	<0.01	0.39	<0.01
REVEAL Lite 2.0	0.42	<0.01	0.51	<0.01	0.47	<0.01	0.44	<0.01
French noninvasive risk model	0.35	<0.01	0.42	<0.01	0.39	<0.01	0.38	<0.01

CI, cardiac index; CO, cardiac output; mPAP, mean pulmonary arterial pressure; PA, pulmonary artery; PAWP, pulmonary artery wedge pressure; PVR, pulmonary vascular resistance; RAP, right atrial pressure; RVEF, right ventricular ejection fraction by MRI; TAPSE, tricuspid annular plane systolic excursion by echo.

A subset of 44 subjects (32 with SSc-PAH and 12 SSc comparators) in PVDOMICS had paired mixed venous and wedged samples available for assessment of transpulmonary metabolite gradients. Kynurenine and kyn/trp were significantly higher in wedged samples than in mixed venous samples, suggesting kynurenine pathway metabolism and metabolite release across the pulmonary circulation. In subjects with PAH, kynurenine was 7.68% higher in wedged samples compared with mixed venous samples (P < 0.001); and kyn/trp was 5.52% higher in wedged samples (P < 0.002) (Supplemental Fig. S3). There were no differences in N-acetylputrescine or 1-methyladenosine concentrations across the pulmonary circulation.

SuHx Plasma Metabolomics

In the SuHx model, we first examined the plasma metabolome as disease evolved. By day 7 post-Sugen, SuHx animals could be discriminated from sham animals on the basis of plasma metabolomics alone (Supplemental Fig. S4A), despite only mild elevation in right ventricular systolic pressure (RVSP). Both RVSP and RV hypertrophy later increased through day 21 to manifest disease (Supplemental Fig. S5). Higher kynurenate was the fourth most important plasma metabolite for discriminating SuHx from sham animals at day 7 (FC 2.33, P = 0.018) (Supplemental Fig. S4B), and plasma kynurenine pathway metabolites were significantly correlated with RVSP and RV mass (Supplemental Fig. S6).

SuHx Tissue Metabolites and Gene Expression

We next examined kynurenine pathway metabolites and genes in SuHx tissues. Lung kyn/trp tended to rise early (days 7 and 14) as disease evolved, then fall at day 21 (Fig. 3A), which was similar to our findings in plasma. Because tryptophan conversion to kynurenine can be catalyzed by either TDO, IDO1, or IDO2 in the first, rate-limiting step of kynurenine pathway metabolism (Supplemental Fig. S7), to investigate molecular modulation of the pathway, we examined gene expression over time for TDO2, IDO1, and IDO2 in SuHx lung and RV tissue. Lung TDO2 expression exhibited a stepwise increase over time as disease developed (Fig. 3B). IDO2 increased by day 14 and IDO1 increased by day 21. TDO2 expression correlated significantly with lung kynurenine and RV hypertrophy (Fig. 3C). No significant differences in gene expression for IDO1, IDO2, or TDO2 were observed in the RVs of SuHx versus sham animals.

Figure 3.

Lung kynurenine pathway gene expression in the SuHx animal model in sham animals (n = 6) compared with Sugen hypoxia animals (n = 6) at weeks 1, 2, and 3. A: abundance of kyn/trp in sham and SuHx lungs. B: expression of TDO2, IDO1, and IDO2 in sham and SuHx lungs. C: a heatplot of correlations between metabolites, gene expression, and phenotypic features. Red-blue coloring corresponds to the absolute value of the correlation coefficient from the most positive value (red) to the most negative value (blue) depicted. *Significant correlation coefficients. Male Wistar rats were used for all animal experiments. Kyn/trp, kynurenine and its ratio to tryptophan; SuHx, Sugen hypoxia.

Human Lung Tissue Gene Expression

We next examined kynurenine pathway differential gene expression in human PAH versus non-PAH lungs using publicly available GEO data. TDO2 was among the top-most differentially expressed genes in two independent datasets, significantly higher in PAH lungs. Data from GSE113439 is shown in Fig. 4. In addition to TDO2, KMO (kynurenine mono-oxygenase), KYAT3 (kynurenine aminotransferase 3), and KYNU (kynureninase), which encode enzymes that convert kynurenine to downstream metabolites in the pathway (Supplemental Fig. S7), were also differentially expressed in PAH lungs (Fig. 4B and Supplemental Fig. S8). By contrast, there were no significant differences in IDO1 or IDO2 expression. Expression of TDO2, but not IDO1 or IDO2, was correlated with genes expressing downstream kynurenine pathway enzymes (Fig. 4C).

Figure 4.

Kynurenine pathway gene expression in the human PAH lung in GEO dataset 113439. A: a volcano plot denoting the magnitude and significance of differentially expressed genes in lung tissues obtained from subjects with vs. without PAH. B: violin plots of differentially expressed kynurenine pathway genes. C: a correlogram displaying relationships between relevant differentially expressed genes. GEO, Gene Expression Omnibus; PAH, pulmonary arterial hypertension. Asterisks denote statistical significance of correlations.

To gain insight into whether differential kynurenine pathway gene expression demonstrated in bulk lung tissue might be cell-type specific, we examined publicly available single-cell RNA sequencing (scRNAseq) data from a previously published study of three idiopathic pulmonary arterial hypertension (IPAH) versus three control lungs (Supplemental Fig. S9, A–E; 21). In this data set, differential TDO2 expression was cell-type specific: TDO2 was four times higher in IPAH fibroblasts compared with control fibroblasts (FC 3.942). IPAH dendritic cells, immune cells that serve as antigen-presenting cells in the lung, demonstrated higher expression of both IDO1 (FC 2.148) and IDO2 (FC 3.484) compared with control dendritic cells. Two other kynurenine pathway genes differentially expressed in PAH bulk lung and available for analysis in the scRNAseq data set, KMO and KYNU, did not demonstrate cell-type specific differential gene expression in IPAH versus control specimens. KMO and KYNU transcripts were expressed primarily in various immune cells (e.g., monocytes, macrophages, and dendritic cells). Our analysis of publicly available single-nucleus RNA sequencing data from healthy human lungs confirms KMO and KYNU expression in immune cells (22). In the healthy human lung, TDO2, as well as IDO1 and IDO2, are expressed at low levels across cell types (Supplemental Fig. S10).

Metabolites As Biomarkers

After completion of our -omics analyses, we returned to our human cohorts for assessment of kynurenine pathway features as potential PAH biomarkers. In PVDOMICS, kynurenine and kyn/trp were higher in all PAH subgroups compared with subjects without PH, with the highest concentrations seen in CTD-PAH (Supplemental Fig. S11). Discrimination of PAH from among non-PH disease comparators and healthy controls was best in CTD-PAH (and in SSc-PAH in particular, with kyn/trp C-statistic 0.86) and in incident PAH, with the kyn/trp ratio consistently exhibiting better discrimination than kynurenine (Supplemental Table S5). Kyn/trp discrimination was good, but not excellent, in the JHSC cohort (C-statistic 0.72). For additional validation, we performed targeted metabolomics in a small, third independent cohort of SSc subjects with and without PAH (n = 24) identified at our center. In this third cohort, the C-statistic for kyn/trp was 0.87 (Fig. 5). We next compared kyn/trp to the most common PAH biomarker, the brain natriuretic prohormone (proBNP), in each of these cohorts. In all three cohorts, the area under the curve for proBNP was numerically, though not significantly, higher than the area under the curve for kyn/trp.

Figure 5.

Receiver operating characteristics (ROC) curves demonstrating ability of the kyn/trp ratio to discriminate subjects with PAH in the PVDOMICS cohort (A), the JHSC cohort (B) and in a validation cohort of subjects from the Johns Hopkins Pulmonary Hypertension Program (C). ROC curves representing discrimination of the brain natriuretic prohormone (proBNP) are superimposed in light blue-gray. P values represent ROC comparisons between kyn/trp and proBNP. JHSC, Johns Hopkins Scleroderma Center; Kyn/trp, kynurenine and its ratio to tryptophan; PAH, pulmonary arterial hypertension; PVDOMICS, pulmonary vascular disease phenomics.

Given the availability of granular SSc-specific data captured by the JHSC, we also examined associations between kynurenine pathway features and markers of SSc activity in the JHSC cohort. Kynurenine and kyn/trp displayed no association with the presence or absence of Raynaud’s phenomenon or telangiectasias. Kyn/trp was numerically, but not significantly, higher in subjects with forced vital capacity ever <70% predicted, which we interrogated as a surrogate for the possible presence of interstitial lung disease (Supplemental Fig. S12).

DISCUSSION

To the best of our knowledge, this is the first study to examine early and evolving metabolic divergence between subjects who do versus do not develop PAH in an at-risk cohort. Taken all together, our results show increased kynurenine pathway metabolism is an early event in PAH pathobiology. Our human data show increased kynurenine pathway metabolism precedes a clinical diagnosis of SSc-PAH, the rate of kynurenine conversion from tryptophan rises over time as PAH develops, and that rate of rise is associated with greater disease risk and shorter survival times. Metabolic alterations observed in the human data were recapitulated in our animal model, including upregulated kynurenine pathway metabolism by day 7 in the SuHx model, before disease is fully manifest. In both human and experimental PAH, kynurenine pathway features were tightly correlated with adverse hemodynamics and RV remodeling.

Our results also suggest abnormal kynurenine pathway metabolism may localize to the lung in PAH. In PVDOMICS subjects, multisite sampling demonstrated kynurenine release across the pulmonary circulation, with higher abundance in wedged samples compared with mixed venous samples. This localization is further supported by transcriptomic analyses of human and SuHx lung tissue, which demonstrate that genes encoding kynurenine pathway enzymes are upregulated in the PAH lung. Our secondary analysis of scRNAseq data suggests TDO2 upregulation may be specific to diseased lung fibroblasts. Notably, in the microarray data deposited by Mura et al. (18) (GSE113439), TDO2 was the fourth most upregulated gene in PAH, while IDO1 and IDO2 were not differentially expressed. TDO2 transcriptional regulation of kynurenine pathway metabolism was further supported by our animal model.

Kynurenine pathway metabolites were initially associated with PH in 2016 reports by Lewis et al. (23) and Jasiewicz et al. (24). Lewis et al. (23) identified strong associations between kynurenine pathway metabolites and adverse hemodynamics in subjects with RV-pulmonary vascular dysfunction. The group further demonstrated upregulation of kynurenine and kyn/trp in the lungs of mice with hypoxia-induced PH, relative to normoxic control mice. Jasiewicz et al. (24) showed that kyn/trp and kynurenine were correlated directly with regulatory T cells and indirectly with Th17 lymphocyte subsets in PAH, demonstrating an association between kynurenine pathway metabolism and dysregulated immunity. Most recently, Cai et al. (25) redemonstrated associations with hemodynamics, and showed that kynurenine pathway metabolite trajectory after PAH treatment initiation was predictive of outcomes in a small cohort .

Tryptophan can be metabolized via the kynurenine pathway by IDO1, IDO2, or TDO, and the ratio of kynurenine to tryptophan is considered a surrogate of rate-limiting IDO/TDO activity (Supplemental Fig. S7; 26). The kynurenine pathway is a major route for tryptophan metabolism in mammals and results in the generation of nicotinamide adenine dinucleotide (NAD). Kynurenine, the first stable metabolite produced, functions as an immunomodulator that promotes differentiation of regulatory T cells, suppresses allogenic T-cell proliferation, and inhibits T-cell responses (27). Upregulated kynurenine pathway activity has been associated with tumor growth via local immune suppression in the tumor microenvironment. Kynurenine and kynurenate are among several endogenous ligands for the nuclear transcription factor AhR (aryl hydrocarbon receptor), and AhR activation is also thought to promote tumor immune escape mechanisms (28). Inhibition of the Trp-Kyn-AhR axis is thus under investigation as a potential therapeutic strategy for cancer immunotherapy (29).

There is evidence to suggest AhR is also crucial to PAH pathobiology: Masaki et al. (30) showed that an AhR agonist, together with hypoxia, induces severe experimental PH with plexiform-like lesions, and AhR knockout rats do not develop experimental PAH in response to Sugen and hypoxia. AhR pharmacologic inhibition is sufficient to reverse RVSP elevation and vascular remodeling (31). Interestingly, in our human transcriptomic analysis, AHR gene expression was correlated with expression of TDO2 and other kynurenine pathway regulators, suggesting a link between the kynurenine pathway and AHR in the PAH lung.

Our rodent time course suggests the early rise in kynurenine pathway metabolites is followed by a decline in both the circulation and the lung after disease is established. This observation aligns with data reported by Cai et al. (25) in which kynurenine abundance fell over time after PAH diagnosis, and with data reported by Nagy et al. (32) in which lung explants from patients with PAH with end-stage disease had lower kynurenine abundance compared with controls. The pattern of kynurenine rise in early disease then fall with established disease, taken together with the stepwise increase over time in TDO2 expression, and higher TDO2 expression even in end-stage PAH, suggests the trp-kyn enzymatic reaction may be substrate-limited by tryptophan depletion as disease progresses.

In addition to providing insight into kynurenine pathway temporal dynamics, our work highlights the importance of TDO to the pathway. Much of the investigation of kynurenine pathway metabolism in cardiovascular diseases has centered on IDO, known to link kynurenine pathway metabolism with immunoinflammatory responses (33), and less is known about TDO in this context. However, in bone marrow transplanted mice, kynurenine generated by TDO, not IDO, was necessary for Th17-dependent pulmonary fibrosis in response to gamma herpesvirus infection (34). Pharmacologic blockade of TDO resulted in reduced Th17 lymphocyte counts, partially normalized the Tr1/Th17 ratio, and ameliorated the fibrotic response to infection (34). In human brain gliomas, TDO, rather than IDO, mediates kynurenine release. In addition, TDO itself exerts autocrine and paracrine effects on tumor cells, suppressing antitumor immune responses and enhancing tumor growth (27). Our results therefore add to a body of evidence suggesting TDO as a potential target for inhibition and subsequent examination of downstream effects in PAH.

The current study has several limitations. Because our local samples for metabolomics were collected and later analyzed retrospectively, samples are subject to variations in time of day, fasting versus feeding states, and other environmental variables. Our use of targeted metabolomics in the surveillance cohort limits the metabolites that can be validated in PVDOMICS. Because metabolite abundance is relatively rather than absolutely quantitated, optimal thresholds for candidate biomarkers cannot be examined, and model calibration cannot be assessed. Use of kyn/trp as a PAH biomarker may be limited by its lack of specificity for PAH pathobiology. Prior studies demonstrate associations between kynurenine pathway metabolites and fibrosing disease processes, including skin graft versus host disease, liver fibrosis, and SSc itself (35–37). Kyn/trp may ultimately prove to be a marker of abnormal fibroblast activation, which may explain why discrimination of kyn/trp is better for SSc-PAH compared with other forms of PAH in the current study. Because kyn/trp does not improve upon the discrimination afforded by proBNP, it is unlikely to be clinically useful as a single biomarker. However, given its discriminatory ability, it warrants inclusion in investigations of predictive multimarker sets. Despite the temporal nature of our results, whether kynurenine pathway metabolism contributes to PAH development directly or is merely reflective of evolving PAH pathophysiology remains uncertain. Our focus on SSc may limit the generalizability of our results, although our cross-sectional findings are generally consistent with previous metabolomics reports in PAH (23, 38–40). Finally, the analyses presented here are based on metabolomic associations with gene expression data only. Given the potential for discordance between gene expression and protein abundance, the lack of protein data, particularly from human lung tissues, is a limitation of the study.

Our study also has several strengths. Ours is the first study to examine metabolic evolution as PAH develops in an at-risk cohort. This is the largest study to examine kynurenine pathway associations in PAH, and it is the only study to confirm statistical associations in external validation cohorts. Our clinical cohorts have complementary strengths: the JHSC surveillance cohort captures subjects as they develop incident, treatment-naïve PAH. We are able to leverage this surveillance paradigm to show that kynurenine pathway metabolism is altered 2 years before a clinical diagnosis of PAH. The PVDOMICS cohort is multicentered and meticulously phenotyped, which allows the assessment of metabolite relationships with measures of disease severity specific to the pulmonary circulation and the RV. PVDOMICS captures patients with prevalent disease. Demonstrating consistent results across human and animal data paves a path forward for future mechanistic studies in a relevant animal model. Integration of lung gene expression gives insight into molecular modulation of the pathway and bolsters the pathobiologic relevance of our findings within the organ that primarily manifests disease.

In conclusion, this report demonstrates that kynurenine pathway metabolites, among others, increase in abundance as subjects with SSc develop PAH. Upregulation of kynurenine pathway metabolism may be an early event in PAH pathobiology. Future in vitro and in vivo studies are needed to clarify mechanisms and investigate the potential for therapeutic kynurenine pathway targeting in PAH. Future clinical studies should absolutely quantify kynurenine pathway metabolites and evaluate them rigorously as diagnostic biomarkers.

DATA AVAILABILITY

Data will be made available upon reasonable request.

SUPPLEMENTAL DATA

Supplemental Figs. S1–S12, Supplemental Tables S1–S5, and Supplemental Methods: https://doi.org/10.6084/m9.figshare.23289872.

GRANTS

This study was supported by National Institutes of Health/National Heart, Lung, and Blood Institute awards K23HL153781 (to C.E.S.), R01HL132153 (to R.L.D., P.M.H.), U01HL125175 (to P.M.H., S.C.M.), U01HL125177 (to S.C.E., G.J.B.), U01HL125218 (to E.S.B.-R., E.M.H.), U01HL125205 (to R.P.F.), U01HL125212 (to A.R.H.), U01HL125208 (to F.R.), U01 HL125215 (to J.A.L.), a National Scleroderma Foundation New Investigator Award (to C.E.S.), and the Pulmonary Hypertension Association. The Johns Hopkins Scleroderma Center Research Registry and Biorepository are supported by the Johns Hopkins inHealth Initiative, the Donald B and Dorothy L Stabler Foundation, the Nancy and Joachim Bechtle Precision Medicine Fund for Scleroderma and the Manugian Family Scholar.

DISCLOSURES

C. E. Simpson, A. A. Shah, L. K. Hummers, A. R. Hemnes, J. A. Leopold, W. H. W. Tang, M. A. Aldred, E. M. Horn, E. S. Berman-Rosenzweig, S. C. Mathai, P. M. Hassoun, and R. L. Damico report research grants from the National Institutes of Health. C. E. Simpson reports research grants from the National Scleroderma Foundation and the Pulmonary Hypertension Association. S. C. Mathai reports research grants from the U.S. Department of Defense. L. K. Hummers reports clinical trial grants from Prometheus, Cumberland Pharmaceuticals, Mitsubishi Tanabe, Horizon Pharmaceuticals, Arena Pharmaceuticals, Medpace LLC, and Kadmon Corporation that are unrelated to the present work. A. A. Shah reports clinical trial grants from Eicos Sciences, Arena Pharmaceuticals, Kadmon Corporation, and Medpace LLC that are unrelated to the present work. W. H. W. Tang is a consultant for Sequana Medical, Cardiol Therapeutics, Genomics plc, Zehna Therapeutics, Renovacor, WhiteSwell, Kiniksa Pharmaceuticals, Boston Scientific, CardiaTec Biosciences, and has received honoraria from Springer Nature and American Board of Internal Medicine. A. R. Hemnes reports consulting fees from Bayer, MSD, United Therapeutics, GossamerBio, Tenax Therapeutics, and Janssen and holds stock in Tenax Therapeutics. S. C. Mathai reports fees from Actelion, United Therapeutics, Janssen, MSD, and Clinical Viewpoints. R. L. Damico has received payments for expert testimony concerning unrelated matters. A. R. Hemnes has served on an Advisory Board for Janssen and serves on the Board of Directors for the Pulmonary Vascular Research Institute. S. C. Mathai has served on an Advisory Board for Bayer and reports a leadership/fiduciary role with the Patient-Centered Outcomes Research Institute. P. M. Hassoun serves on a scientific advisory steering committee for MSD, an activity unrelated to the current work. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

C.E.S., P.M.H., and R.L.D. conceived and designed research; A.S.A., R.H., A.R., D.G., and S.A.A.C. performed experiments; C.E.S., A.S.A., R.H., A.R., S.A., and D.G. analyzed data; C.E.S., A.S.A., A.A.S., L.K.H., A.R.H., J.A.L., E.M.H., E.S.B.-R., G.G., M.A.A., J.B., W.H.W.T., M.G., F.R., R.P.F., S.C.E., G.J.B., N.S.H., S.C.M., P.M.H., and R.L.D. interpreted results of experiments; C.E.S. and A.S.A. prepared figures; C.E.S. drafted manuscript; C.E.S., A.S.A., R.H., A.R., S.A., D.G., A.A.S., L.K.H., A.R.H., J.A.L., E.M.H., E.S.B.-R., G.G., M.A.A., J.B., S.A.A.C., W.H.W.T., M.G., F.R., R.P.F., S.C.E., G.J.B., N.S.H., S.C.M., P.M.H., and R.L.D. edited and revised manuscript; C.E.S., A.S.A., R.H., A.R., S.A., D.G., A.A.S., L.K.H., A.R.H., J.A.L., E.M.H., E.S.B.-R., G.G., M.A.A., J.B., S.A.A.C., W.H.W.T., M.G., F.R., R.P.F., S.C.E., G.J.B., N.S.H., S.C.M., P.M.H., and R.L.D., approved final version of manuscript.