Introduction

Pulmonary hypertension (PH) associated with left heart diseases is the most common form, classified as post-capillary or Group 2 PH (1,2). It includes heart failure (HF), valvular, and congenital heart diseases. There is great interest in the pathophysiology, clinical evolution, and therapies of left-sided PH (2). Especially in the last decade, PH and right ventricular dysfunction (RVD) have been the focus of interest in heart failure with preserved ejection fraction (HFpEF) (3–6), a condition with similar clinical manifestations and event rate to heart failure with reduced ejection fraction (HFrEF) (7,8), but different in etiology, cardiac remodeling, pathophysiology, comorbid disorders, and response to therapy (9). In HFrEF, PH has long been investigated in advanced stages (i.e., in patients waiting for left ventricular assist device [LVAD] or heart transplantation) (8). However, in HFpEF, the role of PH is under intense scrutiny (10), and a thorough recognition of putative mechanisms is demanding, given the high epidemiological impact and the lack of effective management (2).

There are, therefore, intriguing questions that deserve specific attention on how, in the 2 conditions, the molecular pathways, the pathophysiology, and the clinical manifestations of PH may differ, despite similar hemodynamic profile and outcomes. This has obvious implications in terms of prevention and treatment.

We highlight the underlying differences between PH-HFrEF and PH-HFpEF with the aim of guiding research efforts for novel therapies. Specifically, we focus on the characteristic derangements in left heart driving forces biological pathways, hemodynamics, and clinical phenotypes observed in the 2 HF conditions (Central Illustration).

Definition, Characterization, and Epidemiology of Group 2 PH in HF

During the latest 2018 WHO conference, group 2 PH (2) was defined as a mean pulmonary arterial pressure (mPAP) >20 mm Hg and pulmonary arterial wedge pressure (PAWP) >15 mm Hg. Isolated post capillary (Ipc)-PH or combined pre- and post-capillary (Cpc)-PH were defined on a pulmonary vascular resistance (PVR) <3 WU or ≥3 WU, respectively.

This applies to group 2 PH at a whole, including valvular or congenital heart disease; however, PH-HFrEF and PH-HFpEF present with dissimilarities (11,12). Typically, patients with PH-HFpEF are elderly women with the features of metabolic syndrome (9,13), and when metabolic syndrome and obesity coexist, the worse PH phenotype is observed (13–15).

Additional triggers for PH-HFpEF include microvascular coronary disease, atrial fibrillation (AF), chronic kidney disease, and chronic obstructive pulmonary disease (4). PH-HFrEF is more common in middle-aged men with coronary heart disease or dilated cardiomyopathy (8).

Right heart catheterization (RHC) remains the gold standard for a correct PH diagnosis and for timing heart transplantation (HTx) or LVAD support (2).

RHC is also valuable for investigating the hemodynamic response during exercise, discriminating PH-HFpEF from PAH (2) and monitoring interventions (16,17). However, PH prevalence is derived from a nonuniform application of RHC diagnostic standards and observations obtained in different stages of the disease, explaining the wide variability (i.e., between 40% and 75% for PH-HFrEF [18] and 36% to 83% for PH-HFpEF [3]). Some reports also refer to old diagnostic criteria or to the use of pulmonary artery systolic pressure (PASP) estimated by echo Doppler (1). For example, in the recent subanalysis of the PARAGON-HF (Prospective Comparison of ARNI [Angiotensin Receptor-Neprilysin Inhibitor] with ARB [Angiotensin Receptor Blocker] Global Outcomes in HF with Preserved Ejection Fraction) trial, the rate of PH-HFpEF was 31% (6). Nonetheless, these numbers would consistently change based on the HFpEF phenogroup subdivision described by Shah et al. (15), where the true PH phenotype exhibited higher PVR, right atrial pressure, and combined kidney disease, as major PH contributory mechanisms.

There is also some variability in the expected Cpc-PH prevalence. In a large sample of PH-HFrEF, Miller et al. (8) documented a 55% Cpc-PH rate, using PVR criteria. In a retrospective analysis of 2,587 PH-HFpEF patients, the prevalence varied from 8.8% to 3.5% by using a diastolic pressure gradient (DPG) >7 mm Hg or a PVR >3 WU as hemodynamic criteria (19). Another report, merging retrospective and prospective data, found a Cpc-PH rate of 22.6% and 18.8%, for HFpEF and HFrEF, respectively, using a DPG >7 mm Hg (18). For a similar DPG cutoff, a higher Cpc-PH rate (38%) was observed in PH-HFpEF compared with PH-HFrEF (17%) (11).

Overall, these data suggest that Cpc-PH may be more frequent in HFpEF.

Left Heart Driving Forces and Hemodynamics

In both HFrEF and HFpEF, the primary hemodynamic driver of PH is an impaired LV relaxation and filling, which yields backward transmission of elevated left atrial pressure (LAP) to the venous system, capillaries, and arteries and, ultimately, to the right heart. LAP elevation results from the interaction between the type of LV remodeling and left atrial (LA) adaptation in dimensions and dynamics. Indeed, the type of LV myocyte hypertrophy and the amount of reactive and replacement fibrosis well separate HFpEF from HFrEF (20).

Impaired LV filling and cardiac remodeling

In HFpEF, LV concentric hypertrophy and increased diastolic stiffness are common in hypertension and obesity (14,21). Approximately one-third of patients may have diabetes and 10% to 15% (9) may experience infiltrative disease such as cardiac amyloidosis, which is, however, a condition that needs to be diagnosed and excluded from a typical HFpEF phenotype, as indicated by the recent HFA-PEFF score (22). Comorbid conditions exert negative effects on the myocardium, stimulating oxidative stress and hypertrophic pathways (8). On this background, Paulus and Tschope (23) proposed a unifying hypothesis based on the central role of proinflammatory pathways to alter passive tension myocyte properties by increasing collagen proliferation.

High wall stiffness is sustained by increased phosphorylation of titin protein, overexpression of growth factor ß signaling, reduced expression of elastases, high mitochondrial oxidative stress, and epigenetic changes definitively impairing cellular Ca⁺ homeostasis (9).

In HFrEF, LV morphology and myocyte adaptations are driven by excessive wall stress, in most cases due to ischemia and a myocyte genetic background favoring chamber dilation and eccentric hypertrophy. The cardiomyocyte loss alters the balance between collagen deposition and degradation, and patchy areas of fibrosis develop through replacement of dead cardiomyocytes (20). Alterations within the extracellular matrix further contribute to LV enlargement and promote impairment in LV compliance and late increase in stiffness. Abnormalities in LV filling correlate with symptoms better than LVEF and predict outcome (24). At variance with HFrEF, increased LV filling pressures with altered LV passive chamber properties develop early and are prominent in HFpEF (9).

An elevated arterial elastance is an additional pathogenetic substrate for LV stiffness in HFpEF related to aging (25) and menopause estrogen deficiency (26). Accordingly, PH-HFpEF may be exposed to a negative LV mismatch when receiving vasodilators (27).

An enhanced diastolic ventricular interaction is common in PH-HFrEF; RV enlargement and pericardial restraint further impair LV filling, a phenomenon observed in approximately 40% of HFrEF (1). In HFpEF, an unfavorable diastolic ventricular interaction is unmasked during exercise even in the earlier stages (28) and becomes a primary mechanism of PAWP increase in the obese phenotype (14).

Abnormalities in LA dynamics and their determinants

The LA is extremely sensitive to volume and pressure overload and remodels quite rapidly (29). An impaired LA dynamics is fundamental in PH, becoming the major contributor to the elevation in pulsatile loading (1). LA remodeling reflects abnormalities in LV filling pressure and diastolic function unless AF, high-output states, and mitral regurgitation (MR) coexist (30).

The HFrEF phenotype exhibits LA enlargement and eccentric remodeling secondary to severe MR, whereas in HFpEF, the LA becomes stiffer earlier, dilates more slowly, and shows a peculiar cellular substrate favoring AF (9).

Comorbidities and inflammatory mediators also play a main role in atrial dysfunction, rendering the atrium stiff, a condition that begets AF (31). The AF predominant HFpEF syndrome is a unique phenotype associated with MR, higher PVR, and worse outcome (31).

Irrespective of AF, MR coexistence represents the leading hemodynamic mechanism to PH (30,32). Interestingly, data from the ATTEND registry showed that HFpEF may develop MR at a similar rate compared with HFrEF, bearing similar prognostic impact (33). Nonetheless, the observed differences were a higher frequency of mild MR in HFpEF versus severe degree in HFrEF. At variance with HFrEF, in which MR develops on the imbalance between tethering and closing forces, HFpEF exhibits the so called atrial functional MR (34), i.e., mitral annulus (MA) dilatation induced by AF and/or, irrespective of AF, secondary to increased LV end-diastolic pressure and MA dilation (34). AF and HFpEF share pathophysiological grounds (31), making it problematic to differentiate what comes first.

Both HFrEF and HFpEF exhibit impaired LA dynamics with deranged LA reservoir and contractile function tightly connected with PH and RV dysfunction (35). Interestingly, in HFpEF, an abnormal reservoir dynamics combines with elevated PAWP, even when LA volume is normal and functional abnormalities in LA dynamics and hemodynamics are better characterized on exercise. In a study of HFrEF or HFpEF and of healthy control subjects, changes in LA strain and LA pump function (strain rate) were assessed at rest, during exercise, and during the recovery phase. Either HFpEF and HFrEF showed a markedly depressed response (Figures 1A and 1B), with a steep increase in the PASP/tricuspid annular plane systolic excursion relationship for lower LA strain, pointing to the role of the LA in the RV to pulmonary circulation (Pc) uncoupling (Figure 1C) (32).

Increased LAP and pulmonary hemodynamic phenotypes

Any increase in LAP, even mild, perturbates pulmonary hemodynamics and exacerbates symptoms (36). The longitudinal changes in pulmonary hemodynamics at different LAP stages are undefined; however, an analysis by Adir et al. (11) comparing the pulmonary hemodynamics in response to long-lasting elevation in LAP in PH-HFrEF and PH-HFpEF provided intriguing differences. For similar average PAWP, PH-HFpEF had higher DPG and PVR, similar arterial compliance and lower MR rate, indicating a more pronounced vascular disease under similar pulsatile loading and likely reflecting a worse lung microvascular pathology secondary to inflammation.

The thorough assessment of pulmonary hemodynamics during exercise defines distinct pathophysiological features that help to unmask peculiar hemodynamic patterns (37).

Although exercise-induced MR has been recently documented in HFpEF (30), it remains the main trigger of PH in HFrEF (32). There is established evidence that Cpc-PH combines with exercise ventilatory inefficiency similarly in PH-HFpEF and PH-HFrEF (37).

Pathobiology of Lung Vascular Disease and Remodeling

Irrespective of LVEF, endothelial injury is central to the microvascular dysfunction of PH (Figure 2). Studies on the lung vasculature in humans are limited, but preclinical evidence and genetic/epigenetic blood-omic analyses are generating a considerable amount of data (38).

Veins

Intriguing isolated findings have shown that veins undergo a remarkable remodeling process in both HFpEF and HFrEF (39), consisting of luminal narrowing with increased neointimal thickening and medial hypertrophy. Intimal thickening and PASP correlated even after adjusting for arterial medial and intimal thickening, and a loss in gas diffusion was reported. In few patients with sequential lung biopsy at the time of removal of the LVADs, high levels of urokinase plasminogen activator receptor expression have been detected, representing the only pathway so far identified in venous remodeling process (40).

Capillaries and small arteries

The raised backward pressure causes lung capillary and small artery stress failure, a barotrauma that breaks the endothelial layer and promotes fluid and protein swelling in the interstitium. Fluid triggers a cascade of molecular mediators of vascular remodeling, whose pathogenesis is attractive for new targeted therapeutic intervention (41). Edema activates inflammatory mediators, inhibits nitric oxide and natriuretic peptide activities, and increases endothelin (ET)-1 expression leading to fibroblasts/myofibroblast proliferation, luminal occlusion, and thickening of the alveolar septa (1,36). The transition from alveolar-capillary stress failure to remodeling is critical and, in clinical practice, is reflected by a reduced alveolar membrane gas diffusion and high levels of surfactant protein type B (36).

A genetic analysis of vascular remodeling in PH-HFrEF demonstrated an enrichment in genes related to cytoskeleton structure and immune function involving biologically plausible pathways including actin binding, extracellular matrix, basement membrane, transferase activity, pre-ribosome structure, and the major histocompatibility Class II protein complex (42).

A reduced gas diffusion in HFrEF has long been documented (1), and there has also been a recent focus on PH-HFpEF (43).

Evidence on abnormal insulin metabolism as a mediator of pulmonary vascular disease in HFpEF is increasing (44). Activation of macrophage and leukocyte adhesion (45) as well as an overexpression of the ET-1 pathway (46) have been observed in the capillaries of diabetic rats. Increased reactive oxygen species production in diabetic mice is also implicated in the loss of pulmonary endothelial permeability, also sustained by a higher caveolae number.

In models of metabolic syndrome and PH, interleukin (IL)-6 overexpression induces pulmonary arterial smooth muscle cells (SMCs) proliferation and remodeling through STAT3 activation (44).

In a human ex vivo study, high glucose treatment induced SMCs proliferation via reactive oxygen species generation (47), whereas in obese models, SMCs proliferation was found to depend on the imbalance of adiponectin versus proinflammatory adipokines.

The Right Heart and its Coupling with the Pulmonary Circulation

Adaptations of the RV to chronic pressure overload have been detailed elsewhere (36). Briefly, they consist of hypertrophy, progression to chamber dilatation, functional tricuspid incompetence, and RV failure. Padang et al. (48) recently investigated a cohort of 1,299 patients with newly diagnosed severe RV failure and different etiologies, to examine survival at mid- and long-term. Over a 2-year mean follow-up, severe RV failure was associated with an extremely high mortality rate (62%) and similar outcome for HFpEF and HFrEF. However, the determinants of RVD and the clinical manifestations over time may differ because in HFrEF, the RVD generally develops on an intrinsic myopathy and functional tricuspid regurgitation is independently associated with PH and more severe HF (49). Accordingly, in a large observational study, a substantial proportion of HFrEF patients had low tricuspid annular plane systolic excursion with normal pulmonary pressure, whereas RVD was negligible when pulmonary pressure was normal in HFpEF (12). In HFpEF, RV diastolic dysfunction and increased stiffness occur before systolic dysfunction with a slower progression to dilation and functional deterioration (50).

AF and RV pacing are common backgrounds that, combined with the RV inflammatory injury, well explain the RV diffuse fibrosis as peculiar of HFpEF (51). RV to Pc uncoupling may occur earlier in the PH-HFpEF evolving stages (5).

Hemodynamic Phenotyping and New Pathways for Targeted Treatments of PH

There is no established treatment for Group 2 PH, and guidelines warn against the use of pulmonary vasodilators (2). However, given the burden of PH in HF, the search for new targeted therapies is demanding and pharmacological trials are underway. Early experiences with PAH therapies in this setting were disappointing because they were carried out with the aim to reduce pulmonary pressures rather than targeting the biological vascular properties.

Hemodynamic phenotyping by machine learning is advocated by many as the most likely way to identify the subset of patients that may benefit of treatment (52).

Therefore, there appear to be 2 major unmet needs in group 2 PH treatment: first, the thorough identification of the hemodynamic phenotypes; and second, the definition of pathways to be targeted based on solid experimental evidence (36). A herewith hemodynamically oriented way is proposed referring to the pulmonary arterial compliance versus PVR exponential relationship. As exemplified in Figure 3A, there are some subjects (case A, Ipc-PH) who just benefit from moving pulmonary arterial compliance upward and rightward by reducing PAWP; other subjects (case B, Cpc-PH) get an advantage from moving leftward and upward by reducing PVR and shifting out of the flat portion of the curve.

Indeed, case A sees an advantage from pharmacological therapies, especially diuretic agents, valve replacement, mitral clip, or TAVR when appropriate, whereas case B benefits from the same approach for case A plus LVAD or transplantation, when indicated. There might be a benefit from a pulmonary vasodilator for PH-HFpEF candidates who generally are not recipients of LVAD or HTx. Tolerability and safety are important prerequisites, and considering the potential volume overload/preload sensitivity of the LV, pulmonary selectivity should be looked at. Case C (Figure 3B) is a responder without benefits at all by treatment, i.e., moving back along the flat curve, and case D is a nonresponder to pulmonary vasodilators. These phenotypes should be viewed as different entities in a continuum of iterative relationship between disease duration and response to therapy.

In a study of Cpc PH-HFpEF, sildenafil was effective in a long-term follow-up to improve hemodynamics and gas diffusion (17), whereas a study of Ipc-PH did not show any benefit (53). The PASSION trial is underway and will scrutinize the long-term effects of tadalafil in Cpc-PH-HFpEF patients on event/hospitalization. The ongoing SPHERE-HF (β3 Agonist Treatment in Chronic Pulmonary Hypertension Secondary to Heart Failure), a phase II randomized, double-blind clinical trial (54), will test the effects of a β3 adrenergic receptor agonist, mirabegron, on PVR and arteriolar remodeling based on pilot small number trials showing that albuterol may improve exercise pulmonary vascular reserve (55).

Along with more advanced molecular phenotyping in humans with Cpc-PH, increasingly preclinical reports are providing directions on specific pathways to target. Findings by Lai et al. (56) have paved the way to a connection between metabolic syndrome features and lung vasculopathy. More recently, Ranchoux et al. (44) have identified a role of inflammatory signaling through STAT3 and IL-6 in the microvascular remodeling of PH-HFpEF, demonstrating that the pulmonary vascular phenotype could be rescued by either IL-6 antibody or metformin.

PBI 40-50, a first in class antifibrotic agent, showed a modulating activity on mRNA protein expression of fibrogenic markers in lung tissue of post-MI PH-HFrEF (41). Fasudil, a rho kinase signaling inhibitor, reversed microvascular remodeling by inhibiting ET-1 and ETA expression (57). These compounds are waiting for phase II human studies, and trials are underway to test the effects of improving insulin sensitivity ( NCT03617458, NCT03069716, and NCT03629340).

Conclusions and Perspectives

Despite their similar hemodynamic categorization, there is a growing appreciation of evolving differences between PH-HFrEF and PH-HFpEF. Left hemodynamic drivers for PH-HFrEF are LV dilatation, secondary MR, and LA enlargement. Predominant mechanisms in PH-HFpEF are diastolic stiffness, atrial myopathy more susceptible to evolve to AF, and LA functional MR. Vascular stress failure and remodeling similarly affect veins, capillaries, and small arteries, but a main differential trait is the metabolic injury superimposed to the pressure-induced one, typical of HFpEF with metabolic syndrome. An intrinsic RV pathology is frequently observed in PH-HFrEF, making RV to Pc uncoupling an earlier occurrence.

Further definition in the trajectory of left-sided PH and identification of additional differential pathways for targeting vascular pathology are warranted.

Footnotes