Bone morphogenetic protein 9 is a mechanistic biomarker of portopulmonary hypertension Ivana Nikolic, M.D.1#; Lai-Ming Yung, Ph.D.1#; Peiran Yang, Ph.D.1; Rajeev Malhotra, M.D.2; Samuel D. Paskin-Flerlage, B.S.1; Teresa Dinter, B.S.1; Geoffrey A. Bocobo, B.S.1; Kathleen E. Tumelty, Ph.D.3; Anthony J. Faugno, M.D.4; Luca Troncone, Ph.D.1; Megan E. McNeil, B.S. 1; Xiuli Huang, M.Sc.5; Kathryn R. Coser, M.Sc.3; Carol S.C. Lai, M.D.1; Paul D. Upton, Ph.D.6; Marie Jose Goumans, Ph.D.7; Roham T. Zamanian, M.D.8; C. Gregory Elliott, M.D.9; Arthur Lee, M.Sc.5; Wei Zheng, Ph.D.5; Stephen P. Berasi, Ph.D.3; Christine Huard, M.Sc.3;Nicholas W. Morrell, M.D.6; Raymond T. Chung, M.D.11; Richard W. Channick, M.D.10; Kari E. Roberts, M.D.4; Paul B. Yu, M.D., Ph.D.1* 1Brigham and Women’s Hospital and Harvard Medical School, Division of Cardiovascular Medicine, Department of Medicine, Boston, MA 2Massachusetts General Hospital and Harvard Medical School, Division of Cardiology, Department of Medicine, Boston, MA 3Pfizer Centers for Therapeutic Innovation, Cambridge, MA 4Tufts Medical Center, Division of Pulmonary, Critical Care and Sleep Medicine, Department of Medicine, Boston, MA 5National Center for Advancing Translational Sciences, Therapy for Rare and Neglected Diseases Program, Rockville, MD 6University of Cambridge School of Clinical Medicine, Addenbrooke’s and Papworth Hospitals, Division of Respiratory Medicine, Department of Medicine, Cambridge, United Kingdom 7Leiden University Medical Centre, Department of Molecular Cell Biology and Cancer Genomics Centre Netherlands, Leiden, Netherlands 8Stanford University Medical Center, Division of Pulmonary and Critical Care Medicine, Department of Medicine, Stanford, CA 9Intermountain Medical Center and University of Utah, Department of Medicine, Salt Lake City, UT 10Massachusetts General Hospital and Harvard Medical School, Division of Pulmonary and Critical Care Medicine, Department of Medicine, Boston, MA 11Massachusetts General Hospital and Harvard Medical School, Gastrointestinal Unit and Liver Center, Department of Medicine, Boston, MA #These authors contributed equally *Address correspondence to: Paul B. Yu, M.D., Ph.D., Division of Cardiovascular Medicine, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115 Email: pbyu@bwh.harvard.edu; Voice: 857-307-0390; Fax: 617-278-5509 Author Contributions: IN, LMY, PY, PDU, NWM, RTC, RWC, KER and PBY conceived of and designed the study. IN, LMY, PY, SDF, TD, GAB, KET, LT, MEM, XH, KRC, AL and PBY performed experiments. IN, SDP, AJF, CSL, RTZ, CGE, RTC, RWC, KER and PBY enrolled patients and reviewed clinical data. IN, LMY, PY, RM, GAB, PDU, NWM, RTC, KER and PBY analyzed data. IN, LMY, PY, RM, GAB, PDU, MJG, RTZ, CGE, AL, WZ, SPB, CH, NWM, KER and PBY drafted and revised the manuscript. Funding: This work was supported by funding from the U.S. National Institutes of Health (PBY: HL079943, HL131910, HL132742, AR057374; IN: T32HL007604; RTC: DK108370, DK098079), the John S. LaDue Fellowship at Harvard Medical School (IN), a Gilead Sciences Research Scholars Program award in pulmonary arterial hypertension (LMY), a Pulmonary Hypertension Association Clinician-Scientist Award (PBY), a Leducq Foundation Transatlantic Network of Excellence Award (PBY, NWM), the British Heart Foundation (NWM, PDU), a Howard Hughes Medical Institute Early Career Physician-Scientist Award (PBY), a research grant from the Pfizer Centers for Therapeutic Innovation (PBY), and with support from Pfizer Worldwide Research and Development (KET,KRC, SPB, and CH). Running title: Diminished BMP9 is a risk factor for PoPH Category: 17.6 Pulmonary Hypertension: Experimental Word count: 3,744 words At a Glance Commentary: Loss of function mutations in BMP9, its receptors, and downstream signaling molecules have been implicated in heritable pulmonary arterial hypertension (PAH). In the current study diminished levels of circulating BMP9 were found in portopulmonary hypertension (PoPH), but not other etiologies of PAH, or in end-stage liver disease without PAH, suggesting its utility as a unique biomarker of PoPH. Diminished BMP9 levels were found in animals which have pulmonary hypertension associated with portal hypertension and cirrhosis, but not other animal models of pulmonary hypertension. Pharmacologic sequestration of BMP9 exacerbated pulmonary hypertension and vascular remodeling in mice exposed to hypoxia, suggesting a synergy between loss of BMP9 signaling and vascular injury. Acquired loss of BMP9 function appears to be a contributor or risk factor for the development of PAH, and provides a mechanistic link between portopulmonary hypertension and heritable PAH. This article has an online data supplement, which is accessible from this issue's table of content online at www.atsjournals.org 1 3 ABSTRACT Rationale: Bone Morphogenetic Protein 9 (BMP9) is a circulating endothelial quiescence factor with protective effects in pulmonary arterial hypertension (PAH). Loss-of-function mutations in BMP9, its receptors and downstream effectors have been reported in heritable PAH. Objectives: We sought to determine how an acquired deficiency of BMP9 signaling might contribute to PAH. Methods and Results: Plasma levels of BMP9 and antagonist soluble Endoglin (sEng) were measured in Group 1 PAH, Group 2 and 3 pulmonary hypertension (PH), and in patients with severe liver disease without PAH. BMP9 levels were markedly lower in portopulmonary hypertension (PoPH) vs. healthy controls, or other etiologies of PAH or PH, distinguished PoPH from patients with liver disease without PAH, and was an independent predictor of transplant-free survival. BMP9 levels were decreased in mice with PH associated with CCl4-induced portal hypertension and liver cirrhosis, but were normal in other rodent models of PH. Administration of BMP9 ligand trap ALK1-Fc exacerbated PH and pulmonary vascular remodeling in mice treated with hypoxia vs. hypoxia alone. Conclusions: BMP9 is a sensitive and specific biomarker of PoPH, predicting transplant-free survival and the presence of PAH in liver disease. In rodent models, acquired deficiency of BMP9 signaling can predispose to or exacerbate PH, providing a possible mechanistic link between PoPH and heritable PAH. These findings describe a novel experimental model of severe PH that provides insight into the synergy between pulmonary vascular injury and diminished BMP9 signaling in the pathogenesis of PAH. Word count: 240 Key words: bone morphogenetic protein, signaling, portopulmonary hypertension, portal hypertension, cirrhosis, pulmonary arterial hypertension INTRODUCTION The World Health Organization (WHO) Classification defines Group 1 pulmonary arterial hypertension (PAH) by a mean pulmonary artery pressure (mPAP) 25mmHg, pulmonary vascular resistance (PVR) > 3 Wood units, and pulmonary capillary wedge pressure (PCWP) ≤ 15 mmHg, in the absence of left-sided heart disease, severe lung disease, or chronic thromboembolic disease.(1) Group 1 PAH includes patients with heritable PAH (HPAH), idiopathic PAH (IPAH), PAH associated with systemic conditions including toxin or stimulant drug exposure (APAH-STIM), connective tissue disease (APAH-CTD), and portopulmonary hypertension (PoPH).(2) The majority of HPAH disease is attributed to heterozygous loss-of-function mutations in the bone morphogenetic protein (BMP) signaling pathway,(3, 4) including mutations affecting BMPR2 encoding the BMP type II receptor,(5) and ACVRL1 and ENG encoding the BMP9 receptor and co-receptor ALK1 and Endoglin, respectively.(6, 7) Homozygous nonsense mutations in BMP9, encoding the cognate ligand itself, have been reported in a child with severe PAH.(8) While the mutations implicated in HPAH are incompletely penetrant, they subtend a set of genes critical for the transduction of BMP9 by endothelial cells, supporting a pivotal contribution of this signaling axis to PAH. The etiology of portopulmonary hypertension (PoPH) is poorly understood, and its mechanistic relationship to other etiologies of WHO Group I pulmonary arterial hypertension (PAH) is unknown. We previously reported that circulating levels of the soluble form of Endoglin (sEng), an antagonist of BMP9, are a sensitive biomarker of PAH.(9, 10) We recently found that exogenous BMP9 attenuates pulmonary hypertension (PH), vascular remodeling, and RV hypertrophy in several animal models of PH,(11) further implicating the BMP9/BMPR2/Endoglin/ALK1 signaling axis in PH. Since BMP9 is present in the circulation at biologically active levels,(12-15) and is synthesized by the liver, we hypothesized that altered expression of BMP9 might serve as a sensitive mechanistic biomarker of PAH associated with liver disease. In two independent cohorts of patients with diverse etiologies of PAH, we found profoundly diminished circulating BMP9 among patients with PoPH, suggesting a role of impaired signaling. Similarly diminished BMP9 levels were found in rodents with PH associated with portal hypertension and cirrhosis, but not in other animal models of PH. Importantly, administration of the BMP9 ligand trap ALK1-Fc exacerbated PH and pulmonary vascular remodeling in mice exposed to hypoxia, directly demonstrating a protective effect of endogenous BMP9. Diminished BMP9 thus appears to be both a risk factor and sensitive biomarker of PoPH. The acquired deficiency of BMP9 signaling in PoPH may provide a mechanistic link to heritable forms of PAH, and could represent a clinical screening opportunity for the diagnosis and management of PoPH in high-risk populations. MATERIALS AND METHODS Detailed materials and methods are available in an on-line supplement. RESULTS Serum BMP9 induces SMAD1/5/8 phosphorylation in endothelial cells. As previously described,(11, 16) recombinant mature BMP9 homodimeric protein activated BMP-responsive SMAD1/5/8 potently in endothelial cells, assayed by immunoblot, in-cell Western, and SMAD1/5/8 nuclear translocation assays (Fig. 1). BMP9-mediated activation of SMAD1 was dose-dependently inhibited by a potent ALK1/ALK2 kinase inhibitor, TRND-348345, or by the BMP9 ligand trap ALK1-Fc (Fig. 1A). Exposure of cultured human pulmonary microvascular endothelial cells (PMVECs) to pooled human serum (20%) also resulted in activation of SMAD1/5/8, which was inhibited to baseline levels by ALK1-Fc, but not BMP2/4 ligand trap ALK3-Fc (Fig. 1B), suggesting BMP9 is the predominant endothelial SMAD1/5/8-activating factor in the circulation. A phospho-SMAD1/5/8 nuclear translocation assay using cultured PMVECs (Figs. 1C-D and S1A-B) demonstrated potent activation of endothelial signaling by BMP9 (EC50~16 pg/mL) and pooled human serum (EC50~1.6%). BMP9-mediated activation of endothelial SMAD1/5/8 was abrogated by co-treatment with ALK1-Fc or a monoclonal anti-BMP9 (MAB3209, Fig. 1E). Serum-induced activation of endothelial SMAD1/5/8 was similarly inhibited by ALK1-Fc or anti-BMP9 (Fig. 1F), accounting for 80%-100% of activity, confirming that the principal endothelial SMAD1/5/8-activating factor in circulation is BMP9. To detect BMP9 in plasma and serum, a sandwich ELISA with a sensitivity of ~1.6 pg/mL for recombinant mature BMP9 and no cross-reactivity for highly homologous BMP10 was used (Fig. S1C). Specificity was confirmed by the observation that 100% of BMP9 could be neutralized by exogenous ALK1-Fc (Fig. S1D) in a dose-dependent fashion (IC50~14 ng/mL), and similarly for BMP9 activity in normal human and mouse plasma (Fig. S1E). To ascertain the protein species measured by this antibody pair, an immunoprecipitation reaction using the capture antibody (MAB3209) with pooled human serum was resolved by immunoblot using the detection antibody (BAF3209). Under reducing conditions, the predicted 12.5 Kd BMP9 monomer was visualized with the immunoprecipitation of human serum (HS) with MAB3209 but not control IgG, consistent with the monomer from reduced recombinant mature human BMP9 (Fig. S1F). Circulating BMP9 is diminished in portopulmonary hypertension but not other etiologies of Group 1 PAH Levels of circulating BMP9 were measured in distinct derivation and validation cohorts of Group 1 PAH, as well as Group 2 and Group 3 PH patients, with baseline demographic and clinical data presented in Tables 1 and 2. BMP9 levels in healthy individuals were comparable to previously reported average concentrations.(17) Within the derivation cohort (Fig. 2A), there was no significant difference in circulating BMP9 between healthy volunteers and patients with Group 1 PAH (210 pg/mL [IQR 190-246 pg/mL] vs. 217 pg/mL [155-297 pg/mL], p>0.05), Group 2 PH (232 pg/mL [174-264 pg/mL], p>0.05 vs. controls), or Group 3 PH (220 pg/mL [151-272 pg/mL], p>0.05 vs. controls). However, among the etiologies of Group 1 PAH, plasma levels of circulating BMP9 were markedly diminished in patients with PoPH compared to controls (46 pg/mL [21-71 pg/mL]; p<0.0001 vs. controls; Fig. 2B), while no difference was observed with other etiologies of PAH, including IPAH, APAH-CTD, or APAH-STIM. In the derivation cohort, BMP9 levels distinguished PoPH vs. non-PoPH (controls, Group 1 PAH, and Group 2 and 3 PH, with a ROC-AUC of 0.999±0.002 (p<0.001). At a cutoff of <132 pg/mL, BMP9 levels identified PoPH with 100% sensitivity , 96% specificity, 65% positive predictive value (PPV), and 100% negative predictive value (NPV). Similarly in the validation cohort (Fig. 2C), BMP9 levels did not differ between controls and Group 1 PAH, but were markedly decreased in patients with PoPH (81 [48-131 pg/mL] vs. 184 pg/mL [126-227 pg/mL], p=0.0028) and not IPAH or APAH-CTD (Fig. 2D). In the validation cohort, BMP9 levels distinguished PoPH vs. non-PoPH (controls and Group 1 PAH) with a ROC-AUC of 0.827±0.054 (p<0.001). At a cutoff of <132 pg/mL, BMP9 levels identified PoPH with 76% sensitivity, 67% specificity, 43% PPV, and 90% NPV. To address relatively small numbers of PoPH patients in each cohort, the derivation and validation cohorts were pooled and analyzed, revealing an aggregate ROC-AUC of 0.956±0.015 (p<0.0001, Fig. 2E), with 86% sensitivity, 89% specificity, 51% PPV, and 98% NPV. PoPH patients exhibit dysregulated BMP9-Endoglin signaling Endoglin functions as a BMP9 co-receptor on the cell membrane, whereas soluble Endoglin (sEng) functions as a BMP9 ligand trap in the circulation.(12) sEng dose-dependently inhibited BMP9 signaling activity in PMVEC with an IC50 of ~250 ng/mL (Fig. S2A). We confirmed as previously reported that serum levels of sEng are increased in the Group 1 PAH patients(9) as compared to controls (5.6 ng/mL [IQR 4.5-7.0 ng/mL] vs. 4.0 ng/mL [3.4-4.9 ng/mL, p<0.0001, Fig. S2B). In contrast to BMP9, sEng was increased in IPAH (5.4 ng/mL [4.2-6.4], n=35, p=0.0142 vs. controls), and yet more elevated in PoPH versus other etiologies (6.5 ng/mL [5.9-9.1 ng/mL], p<0.0001 vs. control, p=0.01 vs. IPAH, p=0.036 vs. APAH-CTD, and p=0.012 vs. APAH-STIM, Fig. S2C). Levels of sEng correlated minimally with levels of BMP9 (r=-0.11, 95% CI -0.24 to 0.026, p=0.11). Despite the greater levels of sEng seen in PoPH than other Group I PAH, the ability of the BMP9/sEng ratio to distinguish PoPH from other Group 1 PAH was not significantly different than BMP9 alone (AUC=0.90±0.03, p<0.0001, Fig. S2D-E and Fig. 2F). BMP9 levels predict transplant-free survival in Group 1 PAH patients The 5-year transplant-free survival for Group 1 PAH patients, defined as survival without lung transplantation, was 73.5%. Among the hemodynamic and clinical parameters (Table 2), univariate predictors of transplant-free survival included age, NYHA functional class, six-minute walk distance (6MWD), brain natriuretic peptide (BNP), and right atrial pressure (RAP) measured by right heart catheterization (Table 3). Importantly, BMP9 predicted transplant-free survival among Group 1 PAH patients with a Cox HR of 0.66 for every 50 pg/mL increase (95%CI 0.54-0.81, p<0.001, Table 3), and was an independent predictor of transplant-free survival among PAH patients even after adjusting for age, BNP levels, NYHA class, 6MWD, and RAP (Cox HR 0.74 for every 50 pg/mL increase, 95%CI 0.56-0.97, p=0.03). Kaplan-Meier survival analysis within this cohort of PAH patients was performed with BMP9 levels dichotomized around the median value of 176 pg/mL (Fig. 3A), revealing a 76% reduced hazard for death or transplant for patients with BMP9 ≥176 pg/mL (Cox HR 0.24, 95% CI 0.12-0.47, p<0.0001) compared to those with levels <176 pg/mL. The association of BMP9 with survival was not due strictly to PoPH patients, as this association was still significant among PAH excluding PoPH (Cox HR 0.77 for every 50 pg/mL increase, 95% CI 0.63-0.95, p=0.013), and among PAH adjusting for PoPH (Cox HR 0.76, 95% CI 0.62-0.93, p=0.007). In contrast to BMP9, sEng did not predict transplant-free survival (Cox HR 1.08 for every 1 ng/mL increase, 95% CI 0.95-1.24, p=0.242; Table 3). Among Group 1 PAH patients, BNP was elevated primarily in highly symptomatic disease (NYHA III-IV) and correlated with increasing NYHA functional class (Fig. S3A). As previously reported,(9) sEng was elevated even in mildly symptomatic disease (NYHA I-II) compared to healthy controls, but in contrast to BNP neither sEng nor BMP9 correlated significantly with NYHA functional class (Fig S3B-C). Diminished BMP9 levels sensitively predict PoPH in patients with liver disease Circulating BMP9 was measured in 42 patients with liver disease of varying etiology and severity, without evidence of PAH based on negative screening echocardiograms and lack of clinical symptoms. Baseline demographic and clinical data for this patient population and 28 PoPH patients are presented in Table 4. BMP9 was significantly lower in patients with PoPH compared to patients with liver disease without evidence of PAH (63 pg/mL [IQR 37-104 pg/mL] vs. 223 pg/mL [153-282 pg/mL], p<0.0001, Fig. 3B). In contrast to BMP9 there was no difference in sEng levels between PoPH and liver disease without PAH (6.5 ng/mL [5.9-9.1 ng/mL] vs. 6.2 ng/mL [5.4-8.8 ng/mL], p>0.05, Fig. 3C). In this population, BMP9 correlated positively with albumin (r=0.69, p<0.0001) and platelet count (r=0.53, p<0.0001), and negatively with total bilirubin (r=-0.71, p<0.0001) and international normalized ratio (INR) (r=-0.72, p<0.0001), but demonstrated no significant correlation with creatinine (p=0.07, Fig. S4A-F). In agreement, there was a strong negative correlation between BMP9 and the Model for End-Stage Liver Disease (MELD) score (r=-0.84, p<0.0001, Fig. S4E), a composite measure of liver disease severity and prognosis. We tested the correlation of BMP9 levels with the degree of histologic liver fibrosis according to Ishak fibrosis stage, scored in liver biopsy samples available from 34 liver disease patients without known PAH. There was a marginal negative correlation between BMP9 levels and the Ishak fibrosis score (r=-0.31, p=0.07, Fig. S5A). Since MELD scores were higher among PoPH than with liver disease without PAH (Table 4), two approaches were used to test if BMP9 could discriminate PoPH versus liver disease independently of disease severity. First, stratification of individuals into groups with mild (MELD≤10) or moderate-severe disease (MELD>10) revealed significantly lower BMP9 levels in PoPH, but only among the mild disease group (84 pg/mL [54-148 pg/mL] vs. 239 pg/mL [205-293 pg/mL], p<0.0001, Fig 3D). Second, a multivariable logistic regression was performed comparing BMP9 to liver disease indices albumin, INR, creatinine, platelet count, and MELD score as predictors of PoPH. When compared to each of these liver function indices individually or as a group, BMP9 was the only significant predictor of PoPH, with an odds ratio of 0.325 (per 50 pg/mL increase in BMP9, 95%CI 0.157-0.673, p=0.002). Consistent with these results, ROC analysis of BMP9 as a discriminator of PoPH among all liver disease patients yielded an AUC of 0.844±0.050, p<0.001, with BMP9 levels of <154 pg/mL exhibiting sensitivity of 96% and a specificity of 76% for PoPH (Fig. 3E). Similarly, among liver disease with MELD ≤10, BMP9 levels <154 pg/mL were 88% sensitive and 94% specific for PoPH (AUC=0.974±0.022, p<0.0001, Fig. S5B). CCl4-induced cirrhosis and portal hypertension in mice is accompanied by mild pulmonary hypertension and diminished circulating BMP9 To test if the clinical association of PoPH and BMP9 deficiency could be replicated in an animal model, mice were induced to develop cirrhosis and portal hypertension after receiving CCl4 for 16 or 20 weeks. As previously described,(18) splenomegaly, as evidence of portal hypertension, was documented by elevated spleen to body weight ratios in CCl4-treated mice compared to controls (0.63±0.04 vs. 0.30±0.03, mean±SEM, p<0.0001, Fig. 4A). The impact of cirrhosis upon the pulmonary circulation was assessed by echocardiography and confirmed by right heart catheterization. At 16 weeks, CCl4-treated mice exhibited reduced pulmonary artery acceleration time (PAT) (11.9±0.3 ms vs. 19.3±0.8 ms in controls, p<0.0001) and PAT to pulmonary ejection time (PET) ratios (PAT/PET) (21.5±0.6 vs. 34.2±1.2 in controls, p<0.0001, Fig. 4B), similar to previous.(18) CCl4-treated mice demonstrated mildly elevated RV pressures compared to controls (26.2±1.0 mmHg vs. 19.6±0.9 mmHg, p=0.0002, Fig. 4C). Consistent with mild PH, there was no difference in Fulton’s index between CCl4-treated and control mice (Fig. 4D), however, CCl4-treated mice exhibited significantly increased muscularization of small pulmonary arterioles (<50 µm, Fig. 4E-F). After 16 weeks of CCl4, circulating levels of BMP9 were decreased in CCl4 treated mice vs. controls (88.2±9.6 pg/mL vs. 162.9±12.3 pg/mL, p=0.0003), and yet lower after 20 weeks of CCl4 (49.5±2.3 pg/mL vs. 116.6±5.88, p<0.0001 vs. controls, Fig. 4G), comparable to the relative decreases in BMP9 levels in PoPH patients vs. healthy controls. Supporting the specificity of decreased BMP9 in PH associated with cirrhosis and portal hypertension, BMP9 levels were preserved in all other rodent models of pulmonary hypertension examined, including mice and rats with severe PH induced by SU5416 and hypoxia, and rats with severe PH induced by monocrotaline (Fig. 4G). Furthermore, BMP9 levels correlated inversely with RVSP measured by catheterization (r=-0.83, p=0.0005), and correlated with PAT (r=0.60, p=0.02, data not shown) and PAT/PET ratio (r=0.66, p=0.01, Fig. S6A-B), suggesting the severity of PH may be related to the degree of BMP9 deficiency. BMP9 ligand trap ALK1-Fc potentiates hypoxia-induced pulmonary hypertension To determine whether neutralizing circulating BMP9 could trigger or potentiate PH, we administered the BMP9 ligand trap, ALK1-Fc, to mice under normoxic or hypoxic conditions. Exogenous ALK1-Fc was previously shown to antagonize angiogenic activity and reduce tumor burden in a tumor xenograft model.(19) A single injection of ALK1-Fc (5 mg/kg i.p.) reduced plasma BMP9 levels at 3, 24 and 48 h following injection (n=4, p<0.01, Fig. 5A), and reduced expression of BMP target genes Id1 and Id3 in whole lung tissues of ALK1-Fc vs. vehicle-treated mice at 48 hours (n=4, p<0.05, Fig. S7). Administration of ALK1-Fc (5mg/kg i.p. weekly for 3 weeks) to adult mice under normoxic conditions did not lead to measurable PH or RV hypertrophy, but caused a non-significant decrease in PAT/PET ratio compared to normoxic controls (p=0.06, Fig. 5B-D) without inducing a change in cardiac function measured by left ventricular ejection fraction (Fig. 5E). Exposure to hypoxia (FIO2=0.10) alone caused a typical moderate degree of PH, which was markedly worsened by ALK1-Fc (31.3±1.3mmHg vs. 42.8±4.6mmHg, p=0.04). ALK1-Fc did not exacerbate the RV hypertrophy associated with hypoxia alone (Fig. 5C), but markedly worsened the degree of pulmonary vascular remodeling based on increased muscularization of small arterioles (<50 µm diameter, Fig. 5F-G, Fig. S8). In contrast, administration of soluble Endoglin-Fc or ALK3-Fc neither induced or exacerbated pulmonary hypertension in normoxic or hypoxia-exposed mice (Fig. S9). Taken together, these data suggest that a high-affinity BMP9 ligand trap such as ALK1-Fc may exacerbate PH induced by exposure to hypoxia, whereas a relatively lower-affinity trap such as sEng-Fc, or a BMP2/4 ligand trap such as ALK3-Fc do not exhibit this effect. DISCUSSION The ligand BMP9, its cognate receptors BMPR2 and ALK1, and co-receptor Endoglin represent a discrete signaling axis that directly targets the vascular endothelium,(13) with potent anti-apoptotic and homeostatic functions in the pulmonary vasculature.(11, 15, 20) Loss-of-function mutations in each of these signaling molecules have been previously associated with HPAH and PH overlap syndromes including hereditary hemorrhagic telangiectasia and juvenile polyposis syndromes.(21, 22) We investigated circulating levels of BMP9 and its secreted antagonist sEng as candidate biomarkers in PAH. Levels of circulating BMP9 were sufficiently high in healthy individuals (~200 pg/mL) in comparison to its measured EC50 of 16 pg/mL in endothelial cells to elicit signaling, and accounted for 80%-100% of the endothelial BMP signaling activity in human serum (Fig. 1 and S1). We found that BMP9 is a sensitive biomarker that segregates PoPH from other types of PAH, as well as Group 2 and 3 PH, and may represent a risk for PoPH that is not represented by standard indices of liver disease. Consistent with human disease, circulating BMP9 was depleted in rodents with PH due to cirrhosis and portal hypertension, but remained intact in other rodent models of PH. Consistent with the idea that diminished BMP9 might be a risk factor or driver for PH, the potent BMP9 ligand trap ALK1-Fc did not trigger PH, but severely exacerbated PH and pulmonary vascular remodeling due to hypoxia. While sEng was higher in PoPH than IPAH or APAH-CTD, sEng did not discriminate between patients with liver disease without PAH vs. PoPH (Fig. 3B-C). The modest ligand-trapping effect of sEng (Fig. S2A) could potentially compound the effects of diminished circulating BMP9 and contribute to the pathogenesis of PoPH, yet an effect of sEng-Fc was not observed in hypoxic mice (Fig. S9). The role of BMP9 highlighted by the human genetics of PAH, its impact in experimental PH, and its characteristics as a potential biomarker of PoPH support the concept that acquired deficiency of BMP9 may link the pathogenesis of PoPH to heritable, syndromic forms of PAH. The localization of BMP9 expression to hepatocytes, hepatic stellate cells and/or biliary epithelial cells is consistent with its diminished expression in PoPH.(20, 23) Recently it was reported that in addition to BMP9 homodimers, BMP9/BMP10 heterodimers appear to exist in the circulation based on immunodetection, immune depletion, and genetic depletion methods, and may arise from co-expression of these ligands in hepatic stellate cells.(24) Both BMP9 homodimer and BMP9/BMP10 heterodimer would be detected by the assays of this study, and would be inhibited by ALK1-Fc or neutralizing antibody, making observations of this study attributed to BMP9 homodimer applicable to BMP9/BMP10 heterodimer. Direct biochemical assessments of the abundance and function of these species would be needed to discern their relative contributions to the pathophysiology of PoPH or PAH. We acknowledge limitations of the current study due to the size of the cohorts, the inclusion of incident and prevalent cases of PAH, and differences in the source populations and care settings for obtaining PAH, PoPH, liver disease and control samples. We acknowledge the limitations of enriching this study with enrollment from two liver transplant screening programs with a higher prevalence of PoPH than the estimated 2-5% among patients with portal hypertension or 8.5% among those undergoing liver transplant (25). We note also that a sizable fraction of liver disease patients without known PAH had BMP9 levels less than the proposed cutoff of 154 pg/mL (Fig. 3B), but whether or not these levels may predict risk for PoPH is not known. Prospective longitudinal studies evaluating the performance of BMP9 as a predictor of PoPH among at-risk patients with liver disease with a typical prevalence of disease would be required to validate these concepts. There remains an unmet need for biomarkers that can detect pulmonary vascular disease before the onset of severe symptoms and end organ damage. Since PoPH carries a worse prognosis than other etiologies of PAH,(26) with a 5-year mortality of 55% without liver transplant,(25) earlier identification could facilitate medical or surgical treatment when it may still be effective or feasible. In a stratified analysis of patients with mild liver disease, BMP9 levels segregated PoPH patients from those without PAH. in a multivariable logistic regression, BMP9 outperformed other liver disease indices in predicting PoPH among liver disease patients, suggesting BMP9 is more than a marker of liver disease itself. Accordingly, BMP9 did not correlate with Ishak liver fibrosis scores. BMP9 was predictive of transplant-free survival in Group 1 PAH patients, including or excluding PoPH patients, yet was not correlated with NYHA functional class, RAP, or BNP (Fig. S3C, and data not shown) among PAH patients, suggesting BMP9 predicts survival independent of right ventricular function. Among the animal models of PH, BMP9 was diminished only in rodents with severe cirrhosis, but preserved in SU5416/hypoxia-treated mice and rats, and monocrotaline-treated rats despite the known hepatotoxicity of pyrrolizidine alkaloids.(27) Taken together, these results suggest that BMP9 levels are linked to PoPH via portal hypertension rather than liver injury or synthetic dysfunction alone. The concept that BMP9 deficiency may contribute to PoPH is further supported by our previous work in which administration of exogenous BMP9 ameliorated PH due to toxin exposure or loss of BMPR2 function.(11) While potential hurdles may exist in the therapeutic use of BMP9, including the theoretical risk of heterotopic ossification,(28) other strategies to modulate BMP9 signaling could potentially be applied to PoPH or PAH. Conversely, neutralizing BMP9 with ALK1-Fc does not cause PH by itself, but potentiates hypoxia-induced PH in mice, suggesting deficient BMP9 signaling may synergize with other pulmonary vascular insults present in end-stage liver disease, akin to unknown “second hits” required for the penetrance of HPAH-causing mutations. We speculate that the chronic high-flow, high-output state associated with portal hypertension and cirrhosis may interact with diminished BMP9 homeostatic activity to precipitate PoPH. We note beyond the potential endocrine impact of hepatic BMP9 on the pulmonary circulation that BMP9 has important autocrine and paracrine effects on hepatocyte growth and function, liver fibrosis and remodeling,(23, 29, 30) functions that may impact the evolution of underlying liver disease as well as therapeutic efforts to modulate BMP9. In summary, we have identified acquired loss of BMP9 as a potential risk factor for PoPH, possibly due to a loss of tonic protective signaling, and have developed a novel rodent model of severe PH and remodeling due to acquired loss of BMP9 function. These findings highlight a previously unappreciated mechanistic link between non-genetic and genetic forms of PAH, and may provide clues as to the identity of other penetrance factors or environmental “hits” which synergize with diminished BMP signaling to cause PAH. Acknowledgements: The authors wish to thank Drs. Kenneth Bloch, Donald Bloch, and Peter ten Dijke for their critical feedback on this work. DISCLOSURES: P.B.Y. has served as a consultant to Acceleron Pharma, Inc., for the development of sotatercept (ACTRIIA-Fc), an experimental treatment for pulmonary arterial hypertension that may act by modulating the balance of activin versus BMP signaling. K.E.T., K.R.C., S.P.B., and C.H. are employees of Pfizer. REFERENCES 1. Hoeper MM, Bogaard HJ, Condliffe R, Frantz R, Khanna D, Kurzyna M, Langleben D, Manes A, Satoh T, Torres F, Wilkins MR, Badesch DB. Definitions and diagnosis of pulmonary hypertension. Journal of the American College of Cardiology 2013; 62: D42-50. 2. Simonneau G, Gatzoulis GM, Adatia I, Celermajer D, Denton C, Ghofrani A, Gomez Sanchez MA, Kumar RK, Landzberg M, Machado RF, Olschewski H, Robbins IM, Souza R. Updated clinical classification of pulmonary hypertension. Journal of the American College of Cardiology 2013; 62: D34-41. 3. International PPHC, Lane KB, Machado RD, Pauciulo MW, Thomson JR, Phillips JA, 3rd, Loyd JE, Nichols WC, Trembath RC. Heterozygous germline mutations in BMPR2, encoding a TGF-beta receptor, cause familial primary pulmonary hypertension. Nature Genetics 2000; 26: 81-84. 4. Deng Z, Morse JH, Slager SL, Cuervo N, Moore KJ, Venetos G, Kalachikov S, Cayanis E, Fischer SG, Barst RJ, Hodge SE, Knowles JA. Familial primary pulmonary hypertension (gene PPH1) is caused by mutations in the bone morphogenetic protein receptor-II gene. American Journal of Human Genetics 2000; 67: 737-744. 5. Soubrier F CW, Machado R, Grünig E, Aldred M, Geraci M, MD, Loyd JE, Elliott CG, Trembath RC, Newman JH, Humbert M. Genetics and genomics of pulmonary arterial hypertension. Journal of the American College of Cardiology 2013; 62: D13-21. 6. McAllister KA, Grogg KM, Johnson DW, Gallione CJ, Baldwin MA, Jackson CE, Helmbold EA, Markel DS, McKinnon WC, Murrell J. Endoglin, a TGF-beta binding protein of endothelial cells, is the gene for hereditary haemorrhagic telangiectasia type 1. Nature Genetics 1994; 8: 345-351. 7. Berg JN, Gallione CJ, Stenzel TT, Johnson DW, Allen WP, Schwartz CE, Jackson CE, Porteous ME, Marchuk DA. The activin receptor-like kinase 1 gene: genomic structure and mutations in hereditary hemorrhagic telangiectasia type 2. American Journal of Human Genetics 1997; 61: 60-67. 8. Wang G, Fan R, Ji R, Zou W, Penny DJ, Varghese NP, Fan Y. Novel homozygous BMP9 nonsense mutation causes pulmonary arterial hypertension: a case report. BMC Pulmonary Medicine 2016; 16: 1-4. 9. Malhotra R, Paskin-Flerlage S, Zamanian RT, Zimmerman P, Schmidt JW, Deng DY, Southwood M, Spencer R, Lai CS, Parker W, Channick RN, Morrell NW, Elliott CG, Yu PB. Circulating angiogenic modulatory factors predict survival and functional class in pulmonary arterial hypertension. Pulm Circ 2013; 3: 369-380. 10. Venkatesha S, Toporsian M, Lam C, Hanai J, Mammoto T, Kim YM, Lim KH, Yuan HT, Libermann TA, Stillman IE, Roberts D, D'Amore PA, Epstein FH, Sellke FW, Romero R, Sukhatme VP, Letarte M, Karumanchi SA. Soluble endoglin contributes to the pathogenesis of preeclampsia. Nature medicine 2006; 12: 642-649. 11. Long L, Ormiston ML, Yang X, Southwood M, Graf S, Machado RD, Mueller M, Kinzel B, Yung LM, Wilkinson JM, Moore SD, Drake KM, Aldred MA, Yu PB, Upton PD, Morrell NW. Selective enhancement of endothelial BMPR-II with BMP9 reverses pulmonary arterial hypertension. Nature medicine 2015; 21: 777-785. 12. Castonguay R, Werner ED, Matthews RG, Presman E, Mulivor AW, Solban N, Sako D, Pearsall RS, Underwood KW, Seehra J, Kumar R, Grinberg AV. Soluble endoglin specifically binds bone morphogenetic proteins 9 and 10 via its orphan domain, inhibits blood vessel formation, and suppresses tumor growth. Journal of Biological Chemistry 2011; 286: 30034-30046. 13. David L, Mallet C, Mazerbourg S, Feige JJ, Bailly S. Identification of BMP9 and BMP10 as functional activators of the orphan activin receptor-like kinase 1 (ALK1) in endothelial cells. Blood 2007; 109: 1953-1961. 14. Nolan-Stevaux O, Zhong W, Culp S, Shaffer K, Hoover J, Wickramasinghe D, Ruefli-Brasse A. Endoglin requirement for BMP9 signaling in endothelial cells reveals new mechanism of action for selective anti-endoglin antibodies. PLoS ONE [Electronic Resource] 2012; 7: e50920. 15. Upton PD, Davies RJ, Trembath RC, Morrell NW. Bone morphogenetic protein (BMP) and activin type II receptors balance BMP9 signals mediated by activin receptor-like kinase-1 in human pulmonary artery endothelial cells. Journal of Biological Chemistry 2009; 284: 15794-15804. 16. David L, Mallet C, Keramidas M, Lamande N, Gasc JM, Dupuis-Girod S, Plauchu H, Feige JJ, Bailly S. Bone morphogenetic protein-9 is a circulating vascular quiescence factor. Circ Res 2008; 102: 914-922. 17. Kienast Y, Jucknischke U, Scheiblich S, Thier M, de Wouters M, Haas A, Lehmann C, Brand V, Bernicke D, Honold K, Lorenz S. Rapid Activation of Bone Morphogenic Protein 9 by Receptor-mediated Displacement of Pro-domains. J Biol Chem 2016; 291: 3395-3410. 18. Das M, Boerma M, Goree JR, Lavoie EG, Fausther M, Gubrij IB, Pangle AK, Johnson LG, Dranoff JA. Pathological changes in pulmonary circulation in carbon tetrachloride (CCl4)-induced cirrhotic mice. PLoS One 2014; 9: e96043. 19. Mitchell D, Pobre EG, Mulivor AW, Grinberg AV, Castonguay R, Monnell TE, Solban N, Ucran JA, Pearsall RS, Underwood KW, Seehra J, Kumar R. ALK1-Fc inhibits multiple mediators of angiogenesis and suppresses tumor growth. Mol Cancer Ther 2010; 9: 379-388. 20. Bidart M, Ricard N, Levet S, Samson M, Mallet C, David L, Subileau M, Tillet E, Feige J-J, Bailly S. BMP9 is produced by hepatocytes and circulates mainly in an active mature form complexed to its prodomain. Cellular and Molecular Life Sciences 2012; 69: 313-324. 21. Wooderchak-Donahue WL, McDonald J, O'Fallon B, Upton PD, Li W, Roman BL, Young S, Plant P, Fulop GT, Langa C, Morrell NW, Botella LM, Bernabeu C, Stevenson DA, Runo JR, Bayrak-Toydemir P. BMP9 mutations cause a vascular-anomaly syndrome with phenotypic overlap with hereditary hemorrhagic telangiectasia. Am J Hum Genet 2013; 93: 530-537. 22. McDonald JE, Miller FJ, Hallam SE, Nelson L, Marchuk DA, Ward KJ. Clinical manifestations in a large hereditary hemorrhagic telangiectasia (HHT) type 2 kindred. Am J Med Genet 2000; 93: 320-327. 23. Breitkopf-Heinlein K, Meyer C, Konig C, Gaitantzi H, Addante A, Thomas M, Wiercinska E, Cai C, Li Q, Wan F, Hellerbrand C, Valous NA, Hahnel M, Ehlting C, Bode JG, Muller-Bohl S, Klingmuller U, Altenoder J, Ilkavets I, Goumans MJ, Hawinkels LJ, Lee SJ, Wieland M, Mogler C, Ebert MP, Herrera B, Augustin H, Sanchez A, Dooley S, Ten Dijke P. BMP-9 interferes with liver regeneration and promotes liver fibrosis. Gut 2017; 66: 939-954. 24. Tillet E, Ouarne M, Desroches-Castan A, Mallet C, Subileau M, Didier R, Lioutsko A, Belthier G, Feige JJ, Bailly S. A heterodimer formed by bone morphogenetic protein 9 (BMP9) and BMP10 provides most BMP biological activity in plasma. J Biol Chem 2018; 293: 10963-10974. 25. Saleemi S. Portopulmonary hypertension. Ann Thorac Med 2010; 5: 5-9. 26. Krowka MJ, Miller DP, Barst RJ, Taichman D, Dweik RA, Badesch DB, McGoon MD. Portopulmonary hypertension: a report from the US-based REVEAL Registry. Chest 2012; 141: 906-915. 27. Yamashita Y, Fujise N, Imai E, Masunaga H. Reduction of monocrotaline-induced hepatic injury by deleted variant of hepatocyte growth factor (dHGF) in rats. Liver 2002; 22: 302-307. 28. Leblanc E, Trensz F, Haroun S, Drouin G, Bergeron E, Penton CM, Montanaro F, Roux S, Faucheux N, Grenier G. BMP9-Induced muscle heterotopic ossification requires changes to the skeletal muscle microenvironment. J Bone Miner Res 2010. 29. Song JJ, Celeste AJ, Kong FM, Jirtle RL, Rosen V, Thies RS. Bone morphogenetic protein-9 binds to liver cells and stimulates proliferation. Endocrinology 1995; 136: 4293-4297. 30. Bi J, Ge S. Potential roles of BMP9 in liver fibrosis. Int J Mol Sci 2014; 15: 20656-20667. FIGURE LEGENDS Figure 1. BMP9 accounts for the majority of serum-induced SMAD1/5/8 phosphorylation in endothelial cells. (A) Western blot detecting phosphorylated forms of SMAD1 and SMAD3 in cultured BAEC revealed BMP9-mediated activation of SMAD1 (1 ng/mL) is inhibited by ALK1/ALK2 kinase inhibitor TRND-348345 or BMP9 ligand trap ALK1-Fc. (B) In-cell Western revealed activation of SMAD1/5/8 following stimulation of HPMVECs with human serum (20%), with inhibition of phospho-SMAD1/5/8 to baseline levels by ALK1-Fc, but not BMP2/4 ligand trap ALK3-Fc. (C-D) High content phospho-SMAD1/5/8 nuclear translocation assay revealed activation of endothelial signaling by BMP9 (EC50~16 pg/mL) and pooled human AB serum (EC50~1.6%) in cultured PMVECs. (E-F) BMP9-mediated or serum-induced activation of endothelial SMAD1/5/8 were abgrogated by co-treatment with ALK1-Fc or a monoclonal anti-BMP9 Ab. Figure 2. Circulating BMP9 is profoundly decreased in portopulmonary hypertension (PoPH) and is a sensitive and specific predictor of PoPH. (A) Circulating BMP9 was measured in a derivation cohort comprised of 49 healthy controls, 90 WHO Group 1 pulmonary arterial hypertension (PAH) patients, 20 Group 2 PH patients, and 21 Group 3 PH patients. There are no significant differences in the abundance of BMP9 between Group 1, 2 or 3 patients versus healthy controls (Kruskal-Wallis test with Dunn’s multiple comparison). (B) Among Group 1 PAH patients in the derivation cohort, circulating BMP9 was markedly diminished in patients with PoPH (n=11), a difference is not observed in other etiologies of Group 1 PAH, including idiopathic PAH (IPAH, n=41), PAH associated with connective tissue disease (APAH-CTD, n=20), or PAH associated with stimulant use (APAH-STIM, n=9) (Kruskal-Wallis test with Dunn’s multiple comparison). (C) There was a non-significant trend towards decreased BMP9 levels in Group 1 PAH patients from a distinct derivation cohort (58 PAH patients), with a similarly marked decrease in circulating BMP9 among 17 PoPH patients (Mann-Whitney test) (D) but not among 27 IPAH patients and 14 APAH-CTD patients (Kruskal-Wallis test with Dunn’s multiple comparison). (E) ROC analysis for BMP9 as a predictor of PoPH versus healthy individuals and (F) versus patients with Group 1 PAH; black line = no discrimination. Data are expressed as median±interquartile range (IQR), p values as indicated, NS signifies p>0.05 vs. healthy controls. Figure 3. BMP9 predicts transplant-free survival in PAH patients and predicts the presence of PoPH in liver disease. (A) Cumulative survival without lung transplant in Group 1 PAH patients is depicted by Kaplan-Meier survival curves dichotomized about the median value for circulating BMP9 of 176 pg/mL. PAH patients with a BMP9 level ≥176 pg/mL (red line) exhibit a 76% reduced hazard for death or lung transplant (Cox HR 0.240, 95% CI 0.123-0.468, p<0.0001) compared to those with BMP9 <176 pg/mL (black line). Transplant-free survival curves were compared using the Log Rank test. (B) Circulating BMP9 was measured in 60 healthy, control individuals, 42 patients with liver disease without known pulmonary vascular disease, and 28 patients with PoPH. BMP9 levels were substantially lower in PoPH compared to patients with liver disease without known PAH (Kruskal-Wallis test with Dunn’s multiple comparison). (C) Circulating sEng was measured in plasma collected from 26 PoPH patients and 39 patients with liver disease without a diagnosis of PAH. In contrast to BMP9, no significant difference was found in circulating sEng between PoPH and patients with liver disease without PAH. Data are expressed as median±interquartile range (IQR), p values as indicated, NS signifies p>0.05 vs. control (Kruskal-Wallis test with Dunn’s multiple comparison). (D) Among patients with less severe liver dysfunction defined by Model For End-Stage Liver Disease (MELD) scores ≤10, BMP9 was significantly lower in PoPH vs. liver disease without PAH. There was no significant difference in BMP9 levels among patients with severe liver disease defined by MELD score >10. Data are expressed as median±IQR, p value as indicated, NS signifies p>0.05 (Separate Mann-Whitney tests for MELD≤10 and >10). (E) ROC analysis reveals that diminished circulating BMP9 is a highly sensitive and specific predictor of PoPH among all patients with known liver disease, with overall comparable performance to (F) echocardiographic estimated RVSP based on ROC analysis, but with distinct sensitivity at cutoffs of comparable specificity (dotted blue lines); black line = no discrimination. Figure 4. Mice with portal hypertension exhibit mild pulmonary hypertension and reduced levels of circulating BMP9. Male C57BL/6 mice aged 6-8 weeks (n=9) were treated with phenobarbital (0.35 g/L) via drinking water and CCl4 in olive oil (1:5, 0.4 ml/kg body weight, i.p. three times per week) or vehicle (n=10) for 16 weeks at the end of which they underwent right heart catheterization. Two groups were compared using unpaired T-test. (A) CCl4-treated mice develop splenomegaly due to portal hypertension based on elevated spleen to body weight ratio vs. vehicle-treated animals. (B) Cardiac ultrasound demonstrates presence of pulmonary hypertension in CCl4-treated mice based on a reduced ratio of pulmonary artery acceleration time (PAT) to pulmonary ejection time (PET). (C) CCl4-treated mice develop mild pulmonary hypertension based on a mild but significant increase in right ventricular systolic pressure (RVSP) compared to controls. (D) There was no difference in right ventricular hypertrophy between control and CCl4-treated mice. (E, F) Quantification and representative photomicrographs of immunofluorescence of lung sections for von Willebrand's factor (vWF) and smooth muscle -actin (SM-actin) revealed increased muscularization of small (50 µm) arterioles in CCl4-treated mice (bar=100 µm). (G) Levels of circulating BMP9 were significantly lower in mice treated with CCl4 for 20 weeks (n=14) than control mice (n=6) (one-way ANOVA with Dunnett multiple comparison), comparable to decreased levels observed in PoPH patients. Compared to control mice, levels of BMP9 were unchanged in mice treated with SUGEN/hypoxia (Su-Hx, n=8). Similarly, compared to control rats, BMP9 levels were unchanged in rats treated with monocrotaline (MCT, n=8), or Su-Hx (n=5) (one-way ANOVA with Dunnett multiple comparison). Data are expressed as mean±SEM, with significance **p<0.01; ***p<0.001; or p as indicated. Figure 5. Administering BMP9 ligand trap ALK1- Fc induces severe pulmonary hypertension and pulmonary vascular remodeling in mice treated with hypoxia. (A) Administration of ALK1-Fc (5 mg/kg i.p. single injection) to adult C57BL/6 mice reduced plasma BMP9 levels at 3, 24 and 48 hours (p<0.01, repeated measures one-way ANOVA with Dunnett’s multiple comparison). (B-G) Adult C57BL/6 mice received either vehicle or ALK1-Fc (5 mg/kg i.p. twice in 3 week) under normoxic or hypoxic (FIO2=0.10) conditions for a total of 3 weeks. (B) Exposure to hypoxia alone (n=7) resulted in moderately increased RVSP vs. controls (n=5), while mice receiving ALK1-Fc while exposed to hypoxia (n=8) exhibited markedly increased RVSP as compared to hypoxia, normoxia, or ALK1-Fc mice treated under normoxia groups. (C) Exposure to hypoxia resulted in significantly increased RVH based on Fulton’s ratio (RV/LV+S), which was not significantly increased by the addition of ALK1-Fc. (D) Treatment with ALK1-Fc under normoxic or hypoxic conditions, or exposure to hypoxia alone all decreased PAT/PET. (E) ALK1- Fc treatment does not alter cardiac function measured by left ventricular ejection fraction (LVEF) and fractional shortening measured by echocardiography. (F) Exposure to hypoxia increased the proportion of muscularized small arterioles (50 µm) based on SM-actin expression, which was further increased in mice receiving ALK1-Fc under hypoxia, shown (G) by co-staining for SM-actin and vWF (bar=50 µm). Data are expressed as mean±SEM, NS signifies p>0.05, *p<0.05; **p<0.01 and ***p<0.001, groups compared using two-way ANOVA with Sidak’s multiple comparisons tests. TABLES Derivation cohort Validation cohort WHO Group 2 PH WHO Group 3 PH WHO Group 1 PAH PoPH Controls WHO Group 1 PAH PoPH Controls Total sample size, n 90 11 49 63 17 11 20 21 Gender, female subjects, n (%) 72 (80.0)* 6 (54.5)† 27 (55.1) 38 (60.3)† 7 (41.2)† 8 (72.7) 14 (70.0)† 10 (47.6)† BMI, mean ± SEM 30.2 ± 0.87‡ 38.7 ± 2.8*** 27.1 ± 0.89 28.9 ± 0.84‡ 29.3 ± 1.4‡ n/a 35.8 ± 2.6** 31.1 ± 1.9‡ Age, years ± SEM 49.8 ± 1.5‡ 53.9 ± 4.2‡ 50.1 ± 1.7 56.3 ± 1.6‡ 51.7 ± 2.7‡ 49.3 ± 5.1 67.0 ± 2.8*** 63.4 ± 2.9*** PAH etiology, n (%) IPAH HPAH APAH-CTD APAH-STIM APAH-CHD APAH-HIV PVOD PoPH 41 (45.6) 2 (2.2) 20 (22.2) 9 (10.0) 3 (3.3) 2 (2.2) 2 (2.2) 11 (12.2) 27 (42.9) 1 (1.6) 14 (22.2) 1 (1.6) 3 (4.7) 17 (27.0) PAH: pulmonary arterial hypertension; PH: pulmonary hypertension; WHO: World Health Organization; BMI: body mass index; SD: standard deviation; IPAH: idiopathic PAH; HPAH: heritable PAH; APAH: associated PAH; CTD: connective tissue disease; CHD: congenital heart disease; STIM: toxin or stimulant drug exposure; APAH-HIV: human immunodeficiency virus, PVOD: pulmonary venoocclusive disease; PoPH: portopulmonary hypertension; n/a: not available. *p=0.01 vs. cohort matched controls, Fisher’s exact test; †p=NS vs. cohort matched controls, Fisher’s exact test; **p=0.01 vs. cohort matched controls, one-way ANOVA with Dunnett’s test for multiple comparisons; ***p≤0.001 vs. cohort matched controls, one-way ANOVA with Dunnett’s test for multiple comparisons. ‡p=NS vs. cohort matched controls, one-way ANOVA with Dunnett’s test for multiple comparisons. Table 1: Baseline demographic characteristics of pulmonary hypertension patients and healthy controls WHO Group 1 PAH WHO Group 2 PH WHO Group 3 PH Derivation cohort Validation cohort NYHA functional class, n (%)† Class I Class II Class III Class IV 7 (9.9) 35 (49.3) 28 (39.4) 1 (1.4) 3 (5.0) 26 (44.1) 26 (44.1) 4 (6.8) 0 (0) 9 (60) 5 (33.3) 1 (6.7) 3 (15) 6 (30) 11 (55) 0 (0) 6MWD (m), mean ± SEM (n)‡ 440 ± 21 (35) 379 ± 22 (50) 395 ± 14 (3) 324 ± 43 (9) RHC measurements, mean ± SEM (n) RAP (mmHg) mPAP (mmHg) PCWP (mmHg) CO (L/min) CI (L/min/m2) PVR (Wood units) 9 ± 1 (32) 49 ± 2 (42) 10 ± 1 (33) 5.0 ± 0.3 (40) 2.6 ± 0.1 (40) 8.8 ± 0.7 (41) 9 ± 1 (62) 48 ± 1 (62) 10 ± 0.4 (62) 4.5 ± 0.1 (62) 2.4 ± 0.1 (61) 9.6 ± 0.6 (62) 13 ± 2 (17) 43 ± 2 (19) 25 ± 1 (19) 5.7 ± 0.4 (18) 2.7 ± 0.2 (18) 3.6 ± 0.5 (18) 8 ± 1 (21) 33 ± 2 (21) 12 ± 1 (21) 5.4 ± 0.3 (21) 2.8 ± 0.2 (21) 4.4 ± 0.6 (21) TTE RVSP (mmHg), mean ± SEM (n)‡ 72 ± 4 (47) 70 ± 3 (46) 68 ± 4 (16) 61 ± 5 (13) PAH-therapy, % (n) Prostacyclin analogue Endothelin antagonist PDE-5 inhibitor 24 (47.1) 21 (43.1) 38 (74.5) N/A N/A N/A 0 (0) 2 (14.3) 5 (35.7) 0 2 (15.4) 4 (33.3) PAH: pulmonary arterial hypertension; PH: pulmonary hypertension; WHO: World Health Organization; NYHA: New York Heart Association; 6MWD: 6 minute walk distance; SD: standard deviation; RHC: right heart catheterization; RAP: right atrial pressure; mPAP: mean pulmonary artery pressure; PCWP: pulmonary capillary wedge pressure; PVR: pulmonary vascular resistance; CO: cardiac output; CI: cardiac index; TTE: trans thoracic echocardiogram; RVSP: right ventricular systolic pressure; PDE: phosphodiesterase; N/A: data not available. †p=NS differences in distribution of each group versus derivation cohort, Chi-square test; ‡p=NS differences versus derivation cohort, one-way ANOVA with Dunnett’s test for multiple comparisons. Table 2: Clinical characteristics of all PH patients Variable Cox hazard ratio (95% confidence interval) p value Age (for every 10 year increase) 1.42 (1.10 -1.82) 0.007 NYHA (for each class increase) 2.33 (1.41-3.84) 0.001 6 minute walk distance (for every 50 meter increase) 0.82 (0.71-0.94) 0.004 Right atrial pressure (for every 5 mmHg increase) 1.50 (1.17-1.91) 0.001 Cardiac index (for every 1 L/min/m2 increase) 0.66 (0.39-1.11) 0.116 Pulmonary vascular resistance (for every 1 WU increase) 1.06 (1.00-1.11) 0.057 BNP (for every 100 pg/mL increase) 1.20 (1.11-1.30) 0.001 BMP9 (for every 50 pg/mL increase) 0.66 (0.54-0.81) <0.001 sEng (for every 1 nl/mL increase) 1.08 (0.95-1.24) 0.242 NYHA: New York Heart Association; WU: Woods units; BNP: brain natriuretic peptide; BMP: Bone morphogenetic protein; sEng: soluble endoglin. Table 3. Univariate predictors of transplant-free survival in pulmonary arterial hypertension Liver disease no PAH PoPH Total sample size, n 42 28 Etiology of liver disease (n) HCV Alcohol HCV and alcohol HPS and HCV HPS and alcoholic hepatitis Autoimmune hepatitis NAFLD Unknown etiology Other* 33 3 2 1 1 2 9 4 2 0 0 0 1 7 5 Age, years ± SEM 53.1 ± 1.9† 52.6 ± 2.3† Gender, female subjects, n (%) 14 (33.3)‡ 14 (50.0)‡ MELD score, median (IQR) 7.5 (6.4-9.9)# 12.5 (10.2-15.0)# PAH: pulmonary arterial hypertension; PoPH: portopulmonary hypertension; HCV: Hepatitis C Virus; HPS: Hepatopulmonary syndrome; MELD: Model for End-Stage Liver Disease, NAFLD: Nonalcoholic fatty liver disease. *Other includes: biliary atresia (n=1), biliary cirrhosis (n=1), Hepatitis B (n=1), Budd-Chiari syndrome (n=1), Abernathy malformation (n=1). †p=NS, Student’s T-test; ‡p=NS Fisher’s exact test; #p<0.0001 Mann-Whitney test. Table 4: Baseline demographic characteristics of patients with liver disease without PAH vs. patients with portopulmonary hypertension 1