Elucidating the Clinical Implications and Pathophysiology of Pulmonary Hypertension in Heart Failure with Preserved Ejection Fraction: A Call to Action

Abstract

This statement focuses on the need to better understand the epidemiology, pathophysiology, and treatment of pulmonary hypertension (PH) in patients with heart failure with preserved ejection fraction. This clinical phenotype is important because it is common, strongly associated with adverse outcomes, and lacks evidence-based therapies. Our goal is to clarify key knowledge gaps around PH due to HF with preserved ejection fraction and to suggest specific, actionable scientific directions for addressing such gaps. Areas in need of additional investigation include refined disease definitions and interpretation of hemodynamics as well as greater insights on non-cardiac contributors to PH risk, optimized animal models, and further molecular studies in patients with combined pre- and post-capillary PH. We highlight translational approaches that may provide important biologic insight into pathophysiology and reveal new therapeutic targets. Finally, we discuss the current and future landscape of potential therapies for patients with heart failure with preserved ejection fraction and pulmonary vascular dysfunction including considerations of precision medicine, novel trial design, and device-based therapies among other considerations. This statement provides a synthesis of important knowledge gaps culminating in a collection of specific research priorities that we argue warrant investment from the scientific community.

Introduction and Call to Action

Heart failure with preserved ejection fraction (HFpEF) is one of the leading causes of pulmonary hypertension (PH) in the world.¹ The development of PH and particularly pulmonary vascular disease (which distinguishes functional pressure elevation from vascular dysfunction or remodeling) are among the strongest risk factors for adverse outcomes in HFpEF.² Despite this recognition, no evidence-based therapies exist for PH due to HFpEF (PH-HFpEF), in part because the pathophysiology is poorly understood. In this call to action we encourage the scientific community to prioritize the study of PH-HFpEF, which has implications for collaboration, data sharing, and clinical trial design, among other considerations. The goal of this statement is to clarify key knowledge gaps around PH-HFpEF and to suggest scientific directions for addressing such gaps, which we synthesize in Table 1.

Table 1:

Current Problems in PH-HFpEF and Potential Solutions

PH-HFpEF Definition, Prevalence, and Incidence
Problem	Approach
Bias among retrospective cohorts with missing data and potentially poorly generalizable populations	Prospective, multicenter studies with shared phenotyping protocols to understand PH prevalence and progression among patients with HFpEF (e.g., NIH Heart Share network)
The prevalence and incidence of PH-HFpEF are unknown due to lack of longitudinal data with standardized phenotyping	Data from well-phenotyped, longitudinal EHR cohorts (despite limitations of such data) can provide prevalence and incidence rates specific to patients seeking care
Diagnosis and Interpretation of Hemodynamic Data
Problem	Approach
Lack of standardization of provocative maneuver protocols and inconsistent interpretation of results	Standardize protocols and perform validation studies using provocative maneuvers
Need for reliable non-invasive evaluation of PH severity and etiology in HFpEF	Prospective, rigorous, non-invasive studies linked to invasive data to validate echo/exercise predictors of Ipc-PH and Cpc-PH
Unclear role of invasive exercise testing to guide therapy and monitor for progression of PH-HFpEF	Prospective outcomes studies to assess the importance of invasive exercise testing to guide therapy as well as monitor for progression of PH-HFpEF
Non-Cardiac Contributors to PH Risk in HFpEF
Problem	Approach
Need to understand demographic and clinical risk factors for PH in HFpEF – are observed demographic discrepancies based on biologic differences or related to differences in treatment/socioeconomic/environmental factors?	Studies focusing on the role of sex and socioeconomic/environmental exposures as biologic variables
High prevalence of non-cardiac comorbidities may contribute or even drive PH in some patients with HFpEF	•Studies with more detailed non-cardiac phenotyping (particularly disordered breathing and parenchymal lung disease) to understand drivers of overlapping PH etiologies. •Studies focused on understanding the impact of the obesity epidemic on PH-HFpEF risk and interventions (e.g., metformin, mobile health, bariatric surgery) that may reduce or mitigate PH risk in obesity.
Combined Pre-and Post-Capillary PH: Diagnosis and Pathophysiology
Problem	Approach
The prevalence, incidence, and clinical risk factors for Cpc-PH are unclear	Prospective studies with standardized clinical and molecular phenotyping can shed more light on the epidemiology of Cpc-PH
Lack of robust non-invasive markers to identify Cpc-PH	Future research should focus on determining non-invasive imaging or biomarker surrogates for PH-HFpEF and subphenotypes.
Pathophysiology of PH and Translational Approaches to Understand Vascular Remodeling in HFpEF
Problem	•Approach
Poor understanding of Cpc-PH pathophysiology due to lack of molecular data and relevant biospecimens	•Leverage existing data from PVDOMICS and future results of the HeartShare program. •Investment in innovative approaches to study vascular biology including harvesting PA endothelial cells, peripheral and transpulmonary blood samples, inducible stem cells, and efforts to collect lung samples from patients for molecular studies.
Animal Models of PH-HFpEF
Problem	Approach
Lack of a single model that fits all aspects of the disease due to phenotypic heterogeneity and multifactorial nature of the disease.	Selection of appropriate fit-for-purpose models and comprehensive characterization of multidimensional phenotypical readouts.
Lack of large animal models close to human translation	Development of large animal models of PH-HFpEF that can expedite human translation
Treatment Approaches
Problem	Approach
Heterogeneity in the diagnostic method and definition of PH in clinical trials	Developing a universal definition of PH-HFpEF for inclusion in clinical trials
Enrollment of heterogeneous patients with HFpEF in RCT’s and Challenges identifying eligible patients for trials	Precision medicine trials that target specific therapies to specific HFpEF subphenotypes, such as PH-HFpEFNovel RCT designs such as umbrella trials, bucket trials, and adaptive trialsIdentify non-invasive imaging or biomarker surrogates for PH-HFpEF and subphenotypes Create surrogate-based a priori sub-group analyses to enhance RCT interpretation.•Leverage EHR data and machine learning to identify eligible trial participants

PH-HFpEF Definition, Prevalence, and Incidence

The definition of HFpEF varies widely between societal statements, guidelines, and clinical trials. Equally varied is the PH-HFpEF definition, which is often based on echocardiographic data rather than the gold standard, right heart catheterization (RHC). ^3–5 The current consensus definition of different PH hemodynamic profiles is shown in Table 2 along with alternative definitions used in clinical trials and epidemiologic studies. Most data on HFpEF prevalence is derived from registries and electronic health record (EHR)-based studies. Consequently, important details on PH subgroups are unavailable or potentially inaccurate, since these sources rely primarily on international classification of disease codes and are associated with selection bias towards tertiary referral populations.^{15, 16} The fact that most patients with HFpEF are not referred for RHC also likely introduces bias in retrospective cohorts with invasive hemodynamics including potential enrichment with patients more challenging to manage. Invasive phenotyping of patients with suspected HF and/or PH is important because individuals with mean PA pressure (mPAP) ≥19 mmHg as well as those with mildly elevated pulmonary pressure and pulmonary vascular resistance (PVR) ≥2.2 WU are at particularly elevated mortality risk.¹⁷

Table 2.

Current Hemodynamic Definitions of Pulmonary Hypertension and PH-HFpEF

Hemodynamic Definition of Pulmonary Hypertension 3
Definitions	Hemodynamic Criteria	WHO Groups
Pre-capillary PH	mPAP >20mmHg PAWP ≤ 15mmHg PVR ≥3 WU	1,3,4,5
Isolated post-capillary PH	mPAP >20mmHg PAWP > 15mmHg PVR <3 WU	2,5
Combined pre- and post-capillary PH	mPAP >20mmHg PAWP > 15mmHg PVR ≥3 WU	2,5
Components of PH-HFpEF Definition in Clinical Trials and Epidemiologic studies
Modality	Methods or Criteria Used
Clinical	Elevated BNP or NT-pro-BNP Heart Failure Signs and Symptoms (Framingham Criteria) Heart Failure Hospitalization
Echocardiography	LVEF ≥ 40–55% and RVSP > 35–40mmHg or TRV > 2.9 m/s
Hemodynamic	mPAP≥ 25mmHg TPG ≥ 12mmHg PAWP ≥ 15–20mmHg or LVEDP >15 Confrontational Fluid Challenge 500mL of Normal Saline over 5–10minutes; PAWP>15mmHg 6, 7 Normal Saline at 7mL/kg over 5–10minutes; PAWP≥18mmHg8 Normal Saline at 100–200mL/min over two 7minute intervals; PAWP/saline slope ≥25 (±12) mmHg*L*min9 Normal Saline 10mL/kg; ≥21(±4)mmHg10 Exercise Upright cycle ergometer with continuous incremental ramp cycle (5–30W/min) at 60rpm; mPAP/CO >2–3mmHg/L/min 11, 12 Supine cycle ergometry at 60rpm increasing workload 10–30W every 3–5minutes; mPAP>30mmHg13 Weighted (4lb) Arm Adduction or Supine cycle ergometry 20W workload x 5 minutes then increasing 10W every 3 minutes; PAWP ≥ 25mmHg 10, 14

WHO: World Health Organizations; PH: pulmonary hypertension; mPAP: mean pulmonary artery pressure; PAWP: pulmonary artery wedge pressure; PVR: pulmonary vascular resistance; LVEF: left ventricular ejection fraction; RVSP: right ventricular systolic pressure; TRV: tricuspid regurgitant velocity; TPG: transpulmonary gradient; W:Watts

The reported prevalence of PH-HFpEF varies widely depending on the population (clinical trial participants vs. hospital-based cohorts), diagnostic approach (echocardiography or RHC) and the definition used (Table 3). For example, the prevalence of a tricuspid regurgitant velocity >2.9m/sec was 36% among TOPCAT echocardiography sub-study participants whereas the prevalence was 83% in a population-based cohort using a similar definition.^{2, 20} Notably TOPCAT was not powered specifically to assess PH-HFpEF and, thus, suboptimal for guiding information on prevalence. Longitudinal data on PH incidence and progression among well-phenotyped PH-HFpEF cohorts are also lacking.

Table 3. Prevalence of Pulmonary Hypertension in Heart Failure with Preserved Ejection Fraction.

There is significant variability in the definition of PH-HFpEF, modality of evaluation and prevalence, however, PH is present in a substantial proportion of the population, ranging from 36–83%.

Study	Years	Population	Diagnostics	Definition	Prevalence of PH	Severity
Lam et al 2	2003–2005	Olmsted County Heart Failure Surveillance Study	Echo estimated PASP	•Framingham criteria •LVEF ≥50% •Echo PASP >35 mm Hg	83%	PASP= 48 (37–56) mm Hg
Gerges et al 18	Retrospective Cohort 1996–2003	Medical University of Vienna	•Echo, RHC	•Heart Failure signs and symptoms •LVEF≥45% •mPAP≥25 mm Hg	54.4%	Cpc-PH mPAP 45.6±12.8 mm Hg Ipc-PH mPAP 36.4± 8.1 mm Hg
	Prospective Cohort 2012–2013				63%	Cpc-PH mPAP 44.2±13.2 mm Hg Ipc-PH mPAP 34.3±7.0 mm Hg
Leung et al 19	1996–2007	Dartmouth Dynamic Registry	LHC/RHC	•LVEDP >15 •LVEF ≥50% •mPAP >25	52.5%	mPAP 34.2±7.8 mm Hg
Shah et al20	2006–2012	TOPCAT- echocardiography cohort	Echo measure TRV	•LVEF ≥45% •HF hospitalization or elevated BNP/NT-proBNP •TRV >2.9m/s	36%	Mean TRV 3.28 (±0.33) m/s
Melenovsky et al 21	2005–2012	Mayo Clinic	Echo, RHC	•Framingham criteria •LVEF ≥50% •PAWP≥15mmHg •mPAP > 25 mm Hg	81%	mPAP 36±11 mm Hg
Mohammed et al 22	2003–2009	Mayo Clinic, Olmstead County HFpEF Cohort	Echo	•Framingham criteria •LVEF ≥50% •PASP>40mmHg	64.5%
Ho et al12	2006–2017	Massachusetts General Hospital	Invasive CPET	•EF ≥50% •mPAP/CO >3mmHg/L/min	41% (exercise PH)

PASP: pulmonary artery systolic pressure; RHC: right heart catheterization; Cpc-PH: combined pre- and post-capillary PH; Ipc-PH: isolated post-capillary PH; LVEDP: left ventricular end diastolic pressure; LVEF: left ventricular ejection fraction; HF: heart failure; BNP: brain natriuretic peptide; NT-proBNP: N-terminal pro-brain natriuretic peptide; TRV: tricuspid regurgitant velocity; PAWP: pulmonary artery wedge pressure; CPET: cardiopulmonary exercise test mPAP: mean pulmonary arterial pressure; CO; cardiac output

Diagnosis of PH-HFpEF and Interpretation of Hemodynamic Data

In patients with suspected PH-HFpEF, RHC usually reveals PH, elevated pulmonary artery wedge pressure (PAWP), and normal or elevated PVR. However, in some cases, PH may be present without significant elevation of PAWP (often in the setting of diuretic therapy). Provocative maneuvers with exercise or fluid challenge can unmask left ventricular (LV) diastolic dysfunction.²³ These maneuvers may improve diagnostic accuracy but suffer from lack of widespread feasibility and standardization and variability in interpretation. Practical limitations to the widespread use of provocative maneuvers include equipment constraints as well as the absence of standardized patient selection criteria or evidence-based guidelines informing the interpretation of test results.^{24, 25} Patient positioning in invasive cardiopulmonary exercise testing (CPET) varies between upright, supine, and semi-recumbent by center, as does workload protocol and even PH-HFpEF hemodynamic definitions (ie mPAP/cardiac output slope versus standard hemodynamic variables).^{12, 26} Fluid challenge may be simpler than exercise but criteria for PH-HFpEF diagnosis, association of fluid-challenge hemodynamics with outcomes, and implications for management remain to be defined.

When PH patients have a low resting PAWP that subsequently increases with provocation, they are generally categorized as PH-HFpEF or PH with “occult left heart disease.” It remains unclear if these patients are simply adequately diuresed or the resting PAWP are persistently low and they develop hemodynamic congestion primarily with exercise or other provocation. Interpretation of provocative maneuvers must incorporate pre-test probability of disease, an important factor which can be difficult to standardize.

There is a critical need for a non-invasive tool to fully characterize the presence of pulmonary vascular dysfunction, right ventricular (RV)-PA coupling, and intracardiac filling pressures in HFpEF. An ideal tool would differentiate HFpEF patients without PH from those with isolated post-capillary PH (Ipc-PH) versus combined pre- and post-capillary PH (Cpc-PH) non-invasively. One such measure may be non-invasive assessment of RV-PA coupling [tricuspid annular plane systolic excursion/PA systolic pressures (PASP)] which is associated with increased mortality risk, even when the estimated PASP is near normal.²⁷ Careful echocardiographic assessment of RV-PA coupling, estimated PVR, PA acceleration time, or composites of echocardiographic features may help stratify risk in PH-HFpEF patients and drive further testing. However, echocardiography, while widely available and non-invasive, has significant limitations beyond image quality. Two-dimensional echocardiography does not characterize the crescentic anatomy of the RV fully or reliably quantify right atrial pressure, and is associated with limited accuracy for estimating PASP at peak exercise. ²⁸ Although 3-D echocardiography may offset some of these limitations, it is not widely used in mainstream clinical practice.^{29, 30} Cardiac MRI (CMR) provides reliable and reproducible functional and volume assessment of the RV and ventricular-arterial coupling which is important in prognostication.^{31, 32} Beyond RV assessment, left atrial volume and septal angle by CMR (or cardiac CT) can help differentiate between PH-HFpEF from PAH.^33–35 Nevertheless, diagnosing and prognosticating PH patients appropriately requires invasive testing since at present key hemodynamic parameters including PAWP, PVR, and cardiac output are not valid measures using non-invasive imaging tools. Provocative maneuvers during echocardiography may influence the pre-test probability of PH on RHC and provide insight the into disease, particularly under circumstances in which valvular disease or left ventricular (LV) dysfunction are observed.

Non-Cardiac Contributors to PH Risk in HFpEF

The specific effects of age, sex, race, and comorbid conditions on PH risk in patients with HFpEF requires further study. Female sex and Black race were risk factors for PH in unselected cohorts referred for diagnostic testing, but it is unclear if this is driven by treatment differences and/or socioeconomic status.^{36, 37} Systematically collected data examining potential influences of other comorbidities (e.g. sleep disordered breathing, atrial arrhythmias, and lung disease) on PH risk or outcomes in HFpEF is lacking. There is evidence that obesity alone increases PA pressure (but not pulmonary vascular remodeling).³⁸ Obesity and metabolic syndrome co-exist in up to 50% of patients with PH due to left heart disease (PH-LHD) with some data suggesting that the metabolic syndrome and the associated inflammatory milieu may contribute to pulmonary vascular disease.^{39, 40}. Sleep-related breathing disorders are common in PH-HFpEF and suspected to contribute to PH via intermittent hypoxia with resultant cytokine and hormonal derangements that lead to pulmonary vascular remodeling.⁴¹ Implementation of positive airway pressure therapy is associated with a reduction mPAP and PVR, though a direct link pathophysiologic link between sleep-disordered breathing in pulmonary vascular disease per se has not been established.^{42, 43} More complete understanding of the interaction between obesity and PH is critical as the obese-HFpEF phenotype is so common and likely, pathophysiologically distinct.⁴⁴

Cancer and cancer therapies are associated with increased risk of cardiovascular disease, including HFpEF, through effects on diastolic and microvascular dysfunction among other mechanisms.^{45, 46} The potential downstream effects of these interactions on pulmonary pressure are unknown. Moreover, cancer therapies (most notably the tyroskine kinase inhitibor dasatinib) may also cause direct pulmonary vascular dysfunction.⁴⁷ These observations highlight the need for clinical and epidemiologic vigiliance with respect to PH-HFpEF risk as new cancer therapeutics emerge and cancer-related cardiovascular surveillance becomes standard-of-care.

Combined Pre-and Post-Capillary PH: Diagnosis and Pathophysiology

Combined pre-and post-capillary PH differs from Ipc-PH by the presence of PVR ≥3WU. The prevalence of Cpc-PH among retrospective heart failure referral populations ranges from 12–40%.^{18, 37, 48, 49} It is important to distinguish Cpc-PH from Ipc-PH as Cpc-PH is associated with worse outcomes and, in contrast to Ipc-PH, is a population currently enrolling in targeted clinical trials testing therapies for pulmonary vascular disease.^{18, 37} The hemodynamic definition of Cpc-PH has varied over time and should be standardized to facilitate clinical trials as different therapeutic approaches for Cpc-PH and Ipc-PH may be needed.³⁷

The pathophysiology underlying the Cpc-PH subgroup is poorly understood. Chronic, severe left atrial hypertension and left atrial dysfunction leading to vascular remodeling is typically cited as the primary driver of Cpc-PH, a notion based on observations described in patients with rheumatic mitral valve stenosis. However, this fixed obstructive model is dissimilar to the dynamic changes in congestion and loading that are seen with PH-HFpEF. Varying theories exist about the development of Cpc-PH in HFpEF. While generally regarded as pathologic, it is possible that pulmonary vascular changes in HFpEF may be adaptive, specifically working to protect the left heart from intolerable preload. ⁵⁰ Conversely, Cpc-PH may reflect maladaptive progression of Ipc-PH driven by persistent, severe hemodynamic congestion, or represent an intermediate hemodynamic pathophenotype with similarities to PAH. Venous remodeling occurs in Cpc-PH, potentially representing a subgroup of patients that develop disease akin to a pulmonary veno-occlusive forme fruste than PAH per se. There may be subtypes of Cpc-PH influenced by genetic variants or other molecular drivers that predispose individuals with left atrial hypertension to the development of pulmonary vascular disease. Single nucleotide polymorphisms shared by Cpc-PH and PAH have been identified, although the generalizability of these variants to other Cpc-PH cohorts is not known.³⁷ Acquired metabolic dysfunction (e.g. obesity, insulin resistance) may also increase risk of developing vascular remodeling in patients with PH-HFpEF. Worsening of the obesity epidemic would predict a rising prevalence of Cpc-PH. It is also possible that many patients with Ipc-PH at the time of diagnosis progress to Cpc-PH over time, though longitudinal data are lacking.

Pathophysiology of PH and Translational Approaches to Understand Vascular Remodeling in HFpEF

Little is known about pathophysiological processes and cellular/molecular mechanisms involved in the regulation of PH-HFpEF. What little we do know suggests that pulmonary vascular pathology in PH-HFpEF is multifactorial and involves complex systemic alterations. PH-HFpEF patients display global (veins, indeterminate vessels, and arteries) pulmonary vascular remodeling and the severity of PH correlates most strongly with intimal thickening in pulmonary veins and small indeterminate vessels.⁵¹ Of note, only a subset of HFpEF patients display more than 50% of venous intimal thickening, indicating that individual patients who develop the disease can vary markedly. These structural changes are observed regardless of LV systolic function but appear particularly striking in HFpEF. Lymphatic function and drainage are impaired in obesity, metabolic syndrome, and chronic inflammatory states, all recognized co-morbidities of HFpEF. There is emerging evidence of reduced lymphatic reserve in HFpEF which may play a role in PH-HFpEF and the impact on cardiac function and outcomes.⁵²

The role of right ventricular (RV) pathology and emerging evidence for unique RV pathophenotypes also warrants further study. For example, in patients with HFpEF and Cpc-PH, exercise can unmask impaired RV systolic reserve and enhanced interventricular interdependence that may not be evident at rest.⁵³ Longitudinal studies of RV structure and function are a priority for further study based on evidence that the RV may decline out of proportion to changes in LV structure and function in patients with HFpEF.⁵⁴ Recent data also suggest a potential contribution of atrial myopathy due to atrial fibrillation in promoting pulmonary vascular disease and RV dysfunction in patients with HFpEF.⁵⁵ Whether atrial fibrillation management (i.e. rhythm vs rate control) affects the natural history of PH and RV function is unknown.

Knowledge of pulmonary vascular structure is critical but histology is limited to research lung biopsies in HF patients undergoing thoracic surgery or post-mortem sampling.⁵¹ With advances in proteomics, transcriptomics, and digital spatial profiling in formalin fixed paraffin embedded samples, autopsy specimens now provide a means to study not only histopathology of PH-HFpEF but also molecular mechanisms.⁵¹ Innovative imaging solutions for assessing global pulmonary vascular structure are needed.

A significant shortcoming of PH-HFpEF research is the lack of human HF pulmonary vascular tissue to characterize molecular mechanisms. Culture and molecular profiling of PA endothelial cells from the catheter balloon tip⁵⁶ and related approaches⁵⁷ are underway and may provide a new avenue for characterizing the PA endothelial cell response to stressors or potential therapies. Molecular profiling of transpulmonary blood samples may lead to diagnostic tools to distinguish Ipc-PH and Cpc-PH and facilitate systems biology and -omics approaches to the pathophysiologic mechanisms driving pulmonary vascular remodeling.^58,59 The growing expansion of large de-identified databases with associated biobanks offers an important avenue of investigation to understand how genetic variation and plasma markers associate with PH-HFpEF risk.^{58, 60} These resources will be important adjuncts to mechanistic basic studies in animal models to verify or support causal pathways in the development of pulmonary vascular remodeling in HFpEF.

Animal Models of PH-HFpEF

Given the heterogeneity of disease phenotypes and diverse cardiac/non-cardiac contributing factors, relevant and reliable animal models need to be selected carefully as many models may only resemble a certain subtype of PH-HFpEF patients (Table 4). For example, single-hit aortic banding animal models may be useful for examining the contributions of cardiac factors without interference from additional comorbidities.⁶¹ Leptin-deficient (ob/ob) mice and high-fat diet (HFD)-exposed mice have been used to model metabolic syndrome-associated PH and/or HFpEF.⁶² The administration of HFD in a mouse prone to development of metabolic syndrome, the AKR/J mouse, recapitulates many features of PH-HFpEF, including a unique obese HFpEF-related RV phenotype seen in human HFpEF.⁶³ Note that not all mouse strains develop HFD-induced PH-HFpEF and the disease phenotype is mild in susceptible strains.⁶⁴

Table 4:

Animal Models of PH-HFpEF

	Experimental PH-HFpEF models/Comorbidities and disease modifiers	Type of animal species	Advantages	Limitations
Single-hit models	Aortic banding	Mouse Rat Cat Pig	•Reliable and commonly used model for examining cardiac factors without interference from additional comorbidities •Disease phenotypes have been extensively characterized	•The acute increase in afterload does not reflect the pathophysiology of human HFpEF •Prolonged banding (after ~4weeks) leads to LV dilation and systolic heart failure (HFrEF)
	Leptin-deficient ob/ob mouse	Mouse	•No surgery or additional treatment is needed	•Incomplete characterization of PH-HFpEF in this model
	High-fat diet (HFD)	Mouse	•Can be combined with specific genetic manipulations	•Not all mouse strains develop HFD-induced PH-HFpEF •Mild PH-HFpEF phenotypes in susceptible mouse strains
Multi-hit models	Supra-coronary banding + HFD + olanzapine	Rat	•Recapitulate many features of human disease •Permits omics analyses of specific vessel types	•Exercise intolerance, kidney dysfunction and skeletal muscle abnormalities have not been reported
	SU5416 + obese ZSF1 rat	Rat	•Recapitulate many of the key features of human disease, including comorbidities, systemic alterations, physical inactivity, and exercise intolerance •Reproducible •Develop exercise-induced PH-HFpEF during treadmill training	•Relatively expensive •Female obese ZSF1 rats are resistant to the development of hyperglycemia and proteinuria
Large-Animal Model	Banding of pulmonary veins	Pig	•Recapitulates global vascular remodeling observed in humans	•Requires advanced surgical expertise •May not model the effect of comorbidities on pulmonary hypertension (e.g., systemic hypertension)

Two promising multi-hit rat models of PH-HFpEF have been developed recently based on the obesity/metabolic phenotype, comorbidities, pulmonary vasculopathy and/or inflammation (Table 4). The supra-coronary aortic banding together with HFD and olanzapine (an antipsychotic associated with insulin resistance) model appears to recapitulate many of the key features of human disease, although exercise intolerance, kidney dysfunction and skeletal muscle abnormalities have not been reported.⁶⁵ The combination of SU5416 (vascular endothelial growth factor receptor-type A inhibitor known to induce lung endothelial injury and apoptosis) in obese ZSF1 rat represents a reproducible model that recapitulates the combination of systemic alterations, comorbidities, physical inactivity and exercise intolerance often found in human PH-HFpEF.⁶⁶ Notably, SU5416-treated obese ZSF1 rats develop exercise-induced PH (EIPH) during treadmill exercise.⁶⁷ As approximately 50–88% of HFpEF patients develop EIPH, even at low-level exercise, this model may serve as an important tool for exploring potential mechanisms and treatment options for EIPH in HFpEF.¹⁴

Finally, there is an unmet need to develop and use large animal models, which more closely model human physiology and thus may offer more relevant pathophysiologic insights. The use of pulmonary vein banding recapitulates the severe pulmonary arterial, pulmonary venous, and right ventricular remodeling that has been observed in HFpEF patients (Table 4).^{51, 68} The search for an ideal animal model that fits all aspects of the disease remains challenging due to phenotypic heterogeneity and the multifactorial nature of the disease. Through careful definition of specific questions, selection of appropriate fit-for-purpose models (preferably more than one) and comprehensive characterization of multidimensional phenotypical readouts, useful clinical insights may be obtained.

Treatment Approaches

The management of PH-HFpEF is challenging because of the lack of proven PH therapies in the setting of HFpEF. Conventional practice in treating secondary PH is to focus initial efforts on treating the underlying condition. Questions regarding treatment approaches include (1) Is PH simply a marker of disease severity or truly a target for therapy in PH-HFpEF? (2) Should Cpc-PH be considered separately from Ipc-PH in clinical trials? and (3) Should greater consideration be given to the role of RV dysfunction in HFpEF?

Limitations of prior PH-HFpEF studies

To date, randomized controlled trials (RCTs) of PH therapies in left heart disease have included both HFpEF and heart failure with reduced ejection fraction (HFrEF) patients, varied in method of diagnosis of PH, or applied varied definitions of PH making it more difficult to determine therapeutic response. Among HFpEF RCTs (Table 5), prospective evaluation for PH has been rare, which limits the interpretation of trial results with respect to PH responsiveness. Important challenges to the prospective evaluation of PH include (among others) the feasibility of RHC for all study participants and the reliability and reproducibility of non-invasive diagnostic criteria. Previously completed and ongoing phase 3 RCTs in HFpEF vary in phenotyping patients, ranging from simple clinical and biomarker phenotyping (e.g., EMPEROR-Preserved [SGLT2 inhibitors]) to Doppler echocardiography in a subset (e.g., PARAGON [sacubitril/valsartan]) to RHC with exercise in all participants (e.g., REDUCE LAP-HF II [interatrial shunt device]).^73–75 Development of a non-invasive diagnostic score that could reasonably differentiate Cpc-PH from Ipc-PH would be valuable. A validated non-invasive score would circumvent the limitation that invasive hemodynamic testing can impose on large, multi-site studies, particularly in resource-limited areas.

Table 5:

Selected Recent Clinical Trials Applicable to PH-HFpEF

Acronym/ Trial Number	Intervention	Design	Phase	Target HFpEF population	Primary Endpoint	Comments
Vasodilation
SERENADE NCT03153111	Macitentan 24–52 weeks	Multicenter Double-blind Randomized Placebo-controlled	2b	HFpEF (EF ≥ 40%) pulmonary vascular disease RV dysfunction	Change in NT-pro-BNP	Enrichment design Terminated early due to slow enrollment; open label extension; results pending
SOUTHPAW NCT03037580	Oral Treprostinil, 24 weeks	Multicenter Double-blind Randomized Placebo-controlled	3	HFpEF (EF ≥ 45%) RHC confirmed WHO Group 2 PH	Change in 6-minute walk distance	Enrichment design, Terminated early due to slow enrollment; open label extension
HELP-PH-HFpEF70	Intravenous Levosimendan, 6 weeks	Multicenter Double-blind Randomized Placebo-controlled	2	WHO Group 2 PH HFpEF (EF ≥ 40%) PAP ≥ 35 mm Hg PCWP ≥ 20 mm Hg	Change in PCWP with bicycle exercise	Enrichment design; Randomization: ≥ 4 mmHg ↓ PCWP from baseline exercise w/ ≤ 10% ↓ CI. Levosimendan did not reduce exercise-PCWP but did reduce PCWP incorporating data from rest and exercise and increased 6MWD
DYNAMIC NCT02744339	Riociguat 26 weeks	Multicenter Double-blind Randomized Placebo-controlled	2	WHO Group 2 PH HFpEF (EF ≥ 50%) mPAP ≥ 25 mm Hg PCWP > 15 mm Hg	Change in CO by RHC
BEAT HFpEF69	Inhaled Albuterol Acute intervention	Single center Randomized Placebo-controlled	2	HFpEF (EF ≥ 50%) PCWP >15	Change in PVR at 20-Watt exercise	Albuterol improved exercise PVR as compared with placebo (−0.6±0.5 versus +0.1±0.7 WU; P=0.003).
Nebivolol NCT02053246	Nebivolol (β3 agonist) 18 weeks	Single center study	4	HFpEF(EF ≥ 45%) mPAP ≥ 25 PCWP ≥ 15	Change in PVR	Low enrollment
Metabolic
Metformin for PH-HFpEF NCT03629340	Metformin 12 weeks	Multicenter Randomized Placebo-controlled Crossover	2	RHC confirmed PH-HFpEF mPAP ≥ 25 mm Hg PCWP ≥ 15 mm Hg TPG ≥ 12 mm hg Metabolic syndrome	Change in mPAP with submaximal exercise
EMPEROR-Preserved71	Empagliflozin, ~24 months	Multicenter Double-blind Randomized Placebo-controlled	3	HFpEF (EF ≥ 40%) Elevated NT-pro-BNP	Composite: CV death or HF hospitalization
DELIVER72	Dapagliflozin Event driven trial	International, Double-blind, Randomized, Placebo-Controlled	3	HFpEF (EF≥ 40%) Structural heart disease	Composite: CV death, HF hospitalization, or urgent HF visit
PRESERVED-HF NCT03030235	Dapagliflozin, 12 weeks	Multicenter Randomized Double-blind Placebo-controlled	4	HFpEF (EF≥ 45%) Elevated NT-pro-BNP or BNP	Change in heart failure related health status (KCCQ)
Device-based
REBALANCE-HF NCT04592445	Right greater splanchnic nerve ablation	Multicenter Double-blind Randomized Sham-control	Feasibility	HFpEF (EF ≥ 50%) PCWP ≥25 mm Hg with supine exercise	Change in mean PCWP at rest, exercise and with provocative maneuvers
ASPIRE PH NCT04555161	Implanted device in the central PA	Multicenter Open label	Feasibility	WHO Group 1 PAH, potential for applicability to Group 2 PH	Safety: Device or procedure-related serious adverse events	Improves central PA compliance
TROPHY II NCT03611270	PA Denervation via TIVUS system	Multicenter Open Label		Cpc-PH HFpEF or HFrEF	Procedural related adverse events up to 30 days post-procedure

Abbreviations: HFpEF – heart failure with preserved ejection fraction, HFrEF – heart failure with reduced ejection fraction, PAH – pulmonary arterial hypertension, PA – pulmonary artery, PAWP – pulmonary artery wedge pressure, mPAP – mean pulmonary artery pressure, TPG – transpulmonary gradient, PVR - pulmonary vascular resistance, and CO _ cardiac output, PAP – pulmonary artery pressure, KCCQ – Kansas City Cardiomyopathy Questionnaire, RV- right ventricle, EF – ejection fraction, RHC – right heart catheterization, WHO – World Health Organization, BNP – Brain natriuretic peptide, NT-proBNP – N-terminal pro brain natriuretic peptide, Cpc-PH – combined post-and pre-capillary PH.

Precision therapeutics and novel trial designs

The heterogeneity of the HFpEF syndrome is a prevailing reason for the disappointing track record of many prior RCTs, and has sparked calls for a phenotype-specific approach to HFpEF with precision medicine trials that tailor specific therapies to specific HFpEF subphenotypes such as PH-HFpEF.⁷⁶

Novel methods to identify patients with PH-HFpEF may improve efforts to target the specific PH subphenotypes and make trial enrollment more efficient. For example, machine learning using electronic health record data and/or electrocardiograms and echocardiograms have also been developed for the automated identification of patients with specific types of myocardial disease and could be applied to PH-HFpEF as well to identify patients in a high-throughput fashion.^{75, 77–80}

Successful RCTs in PH-HFpEF will likely require novel RCT designs. Examples include umbrella trials, bucket trials, and adaptive trials.⁸¹ An umbrella design would involve taking the heterogeneous group of HFpEF patients and performing phenotyping (e.g., biomarkers, echocardiography, invasive hemodynamics, exercise testing) and identifying more homogeneous subtypes such as Cpc-PH that would then be directed towards targeted RCTs. In this way the RCT is enriched (i.e. enrichment trial) for patients who are most likely to respond to the treatment being tested, which is an approach that has been used in multiple PH-HFpEF trials (Table 5). Bucket trials involve identifying patients who share a similar disease mechanism that could be ameliorated by the treatment being tested. For example, patients with Group 1, Group 2 (including PH-HFpEF), and Group 3 PH could be tested for a specific genetic variant or other molecular marker associated with pulmonary vascular remodeling; if present, they would then be enrolled in an RCT for a medication that specifically targets the molecular mechanism associated with that genetic variant. A similar approach could be used across all types of PH for drugs or devices that treat RV dysfunction. In these trials, the various types of disease that enter the bucket trial could have a unified outcome or varied outcome depending on the type of disease. Finally, in adaptive trials, pre-specified rules are incorporated to account for early information from intermediate endpoints, thereby increasing likelihood for success (by enhancing potential efficacy by homing in on the type of patient most likely to benefit, improving safety, or picking the best outcome that will most likely show a benefit). Additionally, defining appropriate clinical trial endpoints is important for the successful identification of therapeutic interventions. Primary endpoints for PH-HFpEF trials may benefit by learning from both prior HFpEF and PAH trials and utilize a combination of recurrent HF hospitalizations, six minute walk distance, and Kansas City Cardiomyopathy questionnaire (KCCQ) (Table 5).

Mechanistic targets in PH-HFpEF

PH-HFpEF is a multi-factorial syndrome with a range of disease entities (e.g. Ipc-PH, Cpc-PH, and likely others) which creates a conundrum on which disease mechanisms to target. Treatment of HFpEF to avoid progression and development of PH is one approach. Recent large-scale RCTs such as PARAGON-HF and EMPEROR-Preserved may assist with this goal (Table 5).^{71, 74} Additional mechanistic targets include pulmonary vasodilation, RV dysfunction, RV metabolism, PA compliance, RV-PA coupling, splanchnic vasodilation, and counteracting the genetic predisposition to maladaptation of the pulmonary vasculature and extracellular components, which lead to pulmonary vascular remodeling.^82–84 For example, inhaled albuterol (beta-agonist) was shown to improve RV-PA coupling, exercise PVR, and left heart filling and therefore may have utility in a PH-HFpEF population.⁶⁹ Treatment of pathological changes in the pulmonary veins and treatment of pulmonary endothelial dysfunction represent two additional targets.⁵¹

Role of patient reported outcomes

The 6^th World Symposium on PH highlighted the importance of patient perspectives and patient reported outcomes (PROs), specifically advocating for the inclusion of PROs as secondary endpoints in RCTs.⁸⁵ Given that there are no validated surrogate endpoints in PH-HFpEF, assessment of PROs such as quality of life, health status, functional status, or exercise capacity (cardiopulmonary exercise tests or 6MWD) in addition to hospitalizations and mortality is important. There are several validated quality of life measures for use in both PAH and HF with some validated in both populations.⁸⁶ Validation and incorporation of quality of life measures with functional measures such as 6MWD may help to identify therapies in PH-HFpEF that substantially improve patient QoL or health status, which are important to patients and predictors of prognosis.

Exercise and potential role of digital health

Supervised exercise programs (e.g. cardiac rehabilitation) are consistently associated with improvement in quality of life and cardiorespiratory fitness in patients with HFpEF or PH.^87–91 However, no data exist on potential role of exercise in patients with PH-HFpEF specifically. Moreover, the mechanisms by which activity interventions improve exercise capacity (e.g. skeletal muscle function, cardiopulmonary reserve, etc.) warrant further study. The widespread use of commercial activity monitors may facilitate remote, unsupervised interventions to increase physical activity. For example, text-based smartphone motivational interventions could be used to augment activity levels in PH-HFpEF as was recently shown to be feasible in a PAH population.⁹² Such interventions may be an attractive adjunct to medical care because many patients do not have easy access to rehabilitation facilities and many insurers do not cover costs of supervised exercise programs for HFpEF or PH.

Metabolic interventions

More than half of patients with PH due to left heart disease have metabolic syndrome.⁴⁰ Ranchoux and colleagues demonstrated a link between metabolic syndrome and the development of precapillary PH via activation of interlukin-6 (IL-6) associated pathways in animal models as well as increased expression of IL-6 in lung tissue of patients with PH due to left heart disease.⁶⁵ Metformin, anti-IL-6 antibodies, or other anti-inflammatory agents along with SGLT2-inhibitors and GLP1-receptor agonists may represent novel therapeutic interventions for the treatment of PH-HFpEF (Table 5).

Repurposing pulmonary vasodilators

The success of pulmonary vasodilator therapy in PAH has led to the investigation of these therapies in PH-HFpEF. Unlike RCTs of these drugs in all HFpEF, subsequent RCTs used a phenotype-specific approach to enrich the trials with patients deemed most likely to benefit (Table 5). Macitentan, an endothelin receptor antagonist, and oral treprostinil, a prostacyclin analogue, were tested in the SERENADE and SOUTHPAW trials, respectively.^{93, 94} Each of these trials (1) required a PH-HFpEF phenotype with either elevated PVR or RV dysfunction; and (2) incorporated a run-in phase to select out those with fluid retention or pulmonary edema. Both trials were terminated early for slow enrollment, the Achilles heel of precision medicine trials. Despite the desire to utilize and study pulmonary vasodilators in PH-HFpEF, given the poor track record of these drugs thus far, current guidelines/consensus statement strongly recommend against using Group 1 PH therapies in PH due to left heart disease outside of a clinical trial setting for HFpEF and Cpc-PH (30545974).

Augmenting RV function as a therapeutic target

The HELP-PH-HFpEF trial used a similar approach to SERENADE and SOUTHPAW but enrolled patients with invasive hemodynamic evidence of PH-HFpEF and also included a 24-hour run-in phase of intravenous levosimendan (a calcium sensitizer that has inotropic effects) to determine whether the drug reduced exercise PAWP by > 4 mmHg, which was required for subsequent randomization to levosimendan vs. placebo.^93–95 Although it did not meet its primary endpoint of lowering exercise PAWP, levosimendan was associated with a 30-meter placebo-corrected increase in 6MWD, and had favorable hemodynamic effects, thereby supporting the novel trial design as a potential blueprint for future RCTs.⁷⁰

Targeting the RV with treatments such as levosimendan are of potential utility in PH-HFpEF patients with RV dysfunction by increasing unstressed blood volume and thereby limiting excessive splanchnic vasoconstriction and stressed blood volume delivery to the sick right heart in these patients.⁹⁶ Myotropes (e.g., myosin activators) that do not increase myocardial oxygen demand are now available and may be of use in augmenting RV contractility. A fundamental question concerns why some patients with PH-HFpEF develop RV dysfunction and others do not. Molecular markers of RV compensation or that predict decompensation may identify patients more likely to respond to myotropes.

Device-based therapeutics in PH-HFpEF

Finally, novel device-based therapeutics may have a role in PH-HFpEF (Table 5). Splanchnic denervation to improve venous capacitance in patients with PH-HFpEF who frequently develop cardiorenal syndrome with high central venous pressures is currently in development.^{97, 98} Subgroup analysis of the results of the ongoing REBALANCE-HF trial could provide insight into patients with PH-HFpEF and Cpc-PH. Improving splanchnic venous capacitance increases unstressed blood volume and decreases stress blood volume. Increased stressed blood volume is an important pathophysiologic factor in HFpEF, particularly in obesity-related HFpEF, as it has been shown to impact RV-PA coupling and may provide insight into the progression from HFpEF to PH-HFpEF.⁹⁹ Another device with potential applicability to PH-HFpEF involves percutaneous mechanical unloading of the PA with a gas-filled balloon that inflates and deflates during each cardiac cycle thereby restoring central PA compliance.¹⁰⁰ This device, which is under development in PAH, may also be a novel therapeutic in PH-HFpEF patients in whom proximal PA stiffening is a major problem and more common than distal PA stiffening. Finally, pulmonary artery denervation improves PVR and increases 6MWD in patients with PAH and may also be effective in patients with PH-HFpEF.

Conclusion

PH-HFpEF is a growing epidemic with high morbidity, mortality, and no treatment. The clear unmet need and lethal nature of PH-HFpEF must be met with novel solutions at all levels of therapeutic development. We highlight the critical knowledge gaps in PH-HFpEF and offer scientific directions for closing these gaps (Table 1), with a hope to develop novel treatments for patients with PH-HFpEF in the near future.