The neurohormonal basis of pulmonary hypertension in heart failure with preserved ejection fraction

Abstract

Aims

Pulmonary hypertension (PH) represents an important phenotype among the broader spectrum of patients with heart failure with preserved ejection fraction (HFpEF), but its mechanistic basis remains unclear. We hypothesized that activation of endothelin and adrenomedullin, two counterregulatory pathways important in the pathophysiology of PH, would be greater in HFpEF patients with worsening PH, and would correlate with the severity of haemodynamic derangements and limitations in aerobic capacity and cardiopulmonary reserve.

Methods and results

Plasma levels of C-terminal pro-endothelin-1 (CT-proET-1) and mid-regional pro-adrenomedullin (MR-proADM), central haemodynamics, echocardiography, and oxygen consumption (VO₂) were measured at rest and during exercise in subjects with invasively-verified HFpEF (n = 38) and controls free of HF (n = 20) as part of a prospective study. Plasma levels of CT-proET-1 and MR-proADM were highly correlated with one another (r = 0.89, P < 0.0001), and compared to controls, subjects with HFpEF displayed higher levels of each neurohormone at rest and during exercise. C-terminal pro-endothelin-1 and MR-proADM levels were strongly correlated with mean pulmonary artery (PA) pressure (r = 0.73 and 0.65, both P < 0.0001) and pulmonary capillary wedge pressure (r = 0.67 and r = 0.62, both P < 0.0001) and inversely correlated with PA compliance (r = −0.52 and −0.43, both P < 0.001). As compared to controls, subjects with HFpEF displayed right ventricular (RV) reserve limitation, evidenced by less increases in RV s′ and e′ tissue velocities, during exercise. Baseline CT-proET-1 and MR-proADM levels were correlated with worse RV diastolic reserve (ΔRV e′, r = −0.59 and −0.67, both P < 0.001), reduced cardiac output responses to exercise (r = −0.59 and −0.61, both P < 0.0001), and more severely impaired peak VO₂ (r = −0.60 and −0.67, both P < 0.0001).

Conclusion

Subjects with HFpEF display activation of the endothelin and adrenomedullin neurohormonal pathways, the magnitude of which is associated with pulmonary haemodynamic derangements, limitations in RV functional reserve, reduced cardiac output, and more profoundly impaired exercise capacity in HFpEF. Further study is required to evaluate for causal relationships and determine if therapies targeting these counterregulatory pathways can improve outcomes in patients with the HFpEF-PH phenotype.

Clinical trial registration

NCT01418248; https://clinicaltrials.gov/ct2/results? term=NCT01418248&Search=Search

Introduction

Pulmonary hypertension (PH) is common and associated with increased morbidity and mortality in heart failure (HF) with preserved ejection fraction (HFpEF).^1–3 While PH in HF is in large part caused by passive elevation in downstream left heart filling pressure, recent studies have shown that many patients with HFpEF also display coexisting pulmonary vascular disease (PVD).^2–6 Patients with HFpEF and PH display worse pulmonary vascular load with impaired right ventricular (RV) reserve, which is associated with poorer exercise capacity and adverse outcomes.^4,6,7 In patients with earlier stages of HFpEF, this may become manifest exclusively by abnormal pulmonary vasodilation during exercise.^7,8 It has been proposed that these patients with HFpEF and PH represent a unique HFpEF phenotype,^3,4,8 which may share biological overlap with pulmonary arterial hypertension (PAH).^9,10 However, the potential neurohormonal basis driving abnormal right ventricular-pulmonary artery (RV-PA) coupling and HF in this phenotype remains unknown.

Plasma levels of endothelin-1 (ET-1) and adrenomedullin (ADM) are increased in patients with HF^11–14 as well as other forms of PH,^15,16 including PH due to HFrEF.¹⁷ Endothelin plays a crucial role in the development of PAH by causing pulmonary vasoconstriction, smooth muscle cell proliferation, and pulmonary vascular remodelling.¹⁴ In contrast, ADM is known to be a cardioprotective peptide with potentially important counterregulatory effects in the pulmonary circulation, including vasodilation and inhibition of vascular structural remodelling.^18–20 In animal models of PH, ET-1 and ADM are both up-regulated in tandem prior to the development of structural remodelling and worsening PH.²¹ Given this background, we hypothesized that C-terminal pro-endothelin-1 (CT-proET-1) and mid-regional pro-adrenomedullin (MR-proADM), would be elevated in HFpEF patients compared to controls at both rest and exercise, and that the magnitude of elevation would correlate with severity of PH, abnormalities in RV-PA coupling, and impairments in exercise capacity and cardiopulmonary reserve in HFpEF.

Methods

Subjects referred to the Mayo Clinic catheterization laboratory for invasive exercise right heart catheterization were enrolled in this prospective study. Written informed consent was provided by all patients prior to participation in study-related procedures. The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written. The study was approved by the Mayo Clinic Institutional Review Board and the study was registered (NCT01418248).

Study population

The diagnosis of HFpEF was determined using invasive exercise testing in all participants. Heart failure with preserved ejection fraction was confirmed by typical clinical symptoms (dyspnoea, fatigue), normal left ventricular (LV) ejection fraction (≥50%), and elevated pulmonary capillary wedge pressure (PCWP) at rest (>15 mmHg) and/or with exercise (≥25 mmHg).^22,23 After excluding significant valvular disease, infiltrative, restrictive or hypertrophic cardiomyopathy, constrictive pericarditis, pulmonary embolism, and RV myopathies, 38 met criteria for HFpEF. Control subjects (n = 20) were being evaluated for exertional dyspnoea according to the same invasive exercise evaluation but were found to display no evidence of HF or cardiac pathology after thorough clinical evaluation, imaging and invasive assessment, including normal rest and exercise PCWP (criteria above) and PA pressures.

Haemodynamic assessment

Patients were studied on their long-term medications in the fasted state after minimal sedation, as previous described.^4,23–25 Right heart catheterization was performed in the supine position through a 9-Fr sheath via the right internal jugular vein. Right atrial pressure (RAP), PA pressures, and PCWP were measured at end expiration (mean of ≥3 beats) using 2 Fr high-fidelity micromanometer-tipped catheters (Millar Instruments, Houston, TX, USA) advanced through the lumen of a 7 Fr fluid-filled catheter (Balloon wedge, Arrow). Pressure tracings were digitized (240 Hz) and stored for offline analysis.

Oxygen consumption (VO₂) was measured using expired gas analysis at rest and throughout exercise (MedGraphics, St. Paul, MN, USA). Arterial-venous O₂ content difference (AVO₂diff) was measured directly as the difference between systemic arterial and PA O₂ content. Cardiac output (Qp) was determined by the direct Fick method (Qp = VO₂/AVO₂diff). Pulmonary vascular function were assessed by pulmonary vascular resistance [PVR = (mean PA-PCWP)/Qp] and PA compliance [stroke volume/(PA pulse pressure)]. Abnormal exercise PVR was categorically defined as >1.58 WU, and abnormal exercise PA compliance as <2.1 mL/mmHg, with cut-offs taken as 2 SDs from the mean in normal older adults during exercise.^26,27 Following baseline assessments, all haemodynamic measurements were repeated during the first stage of exercise (20 W) followed by graded 10 W increments in workload (3-min stages) to subject-reported exhaustion.^4,23–25

Ventricular function assessment

Two-dimensional, Doppler, and tissue Doppler echocardiography were performed simultaneously with invasive haemodynamic assessment at rest and during all stages of exercise by experienced sonographers. Right ventricular systolic and diastolic function were assessed using lateral tricuspid annular tissue velocities.^25,28 Left ventricular systolic and diastolic function were evaluated at rest only using systolic and diastolic mitral annular tissue Doppler velocities (s′ and e′ velocities) and LV longitudinal strain. Speckle tracking analyses were performed offline with commercially available software (Syngo, Siemens Medical Solutions, Munich, Germany). Left ventricular longitudinal strain was measured from 2 apical views as previously described.²⁹ Strain values represent the mean of 3 beats and are expressed as absolute value.

Neurohormone assessment

Blood samples were obtained directly from the central circulation (superior vena cava) at rest, after 4 min of exercise at 20 W, and again at peak exercise. Plasma levels of CT-proET-1 [interassay coefficient of variations (CVs), 5.6% at 72.9 pmol/L and 3.8% at 198 pmol/L] and MR-proADM (CVs, 4.6% at 0.7 nmol/L and 4.3% at 4.3 nmol/L) were measured using time-resolved amplified cryptate emission assays (Thermo Scientific B·R·A·H·M·S Biomarkers, Waltham, MA, USA).

Statistical analysis

Data are reported as mean and standard deviation for normally distributed data, median and interquartile range for non-normally distributed data, or number (%) unless otherwise specified. Between-group differences were compared using an unpaired t-test, Wilcoxon rank-sum test, χ² or Fisher’s exact test, or two-way repeated measures analysis of variance (ANOVA), as appropriate. Linear correlations between continuous variables were assessed using Pearson’s correlation coefficient, where non-normally distributed variables were first log transformed. Comparison of correlation coefficients was determined using Meng’s test. Multiple linear regression analysis was used to adjust for age, renal function, and body mass index (BMI). Receiver-operating characteristic curve analysis based upon continuous biomarker levels was used to assess diagnostic accuracy for CT-proET-1 and MR-proADM to detect patients with abnormal pulmonary vascular responses to exercise. Sensitivity and specificity were calculated based upon the optimal cut-off value according to Youden’s index. All tests were two-sided, with a value of P < 0.05 considered significant.

Results

Subject characteristics

Compared to controls, subjects with HFpEF were older, had higher BMI and were more likely to be treated with beta-blockers (Table 1). There were no significant differences in sex, comorbidities, or other medication use. Heart failure with preserved ejection fraction subjects had higher NT-proBNP and creatinine levels and lower haemoglobin levels than controls. Left ventricular size, mass, and ejection fraction were similar in cases and controls. Heart failure with preserved ejection fraction subjects displayed lower systolic and diastolic mitral tissue velocities, reduced LV longitudinal strain, and higher mean E/e′ ratio and left atrial volume as compared to controls. Right ventricular size was similar between the groups (Table 1).

Table 1.

Baseline characteristics

	Control (n = 20)	HFpEF (n = 38)	P-value
Age (years)	62 ± 11	69 ± 11	0.01
Female sex (%)	50	50	1.0
Body mass index (kg/m2)	27.5 ± 4.8	34.1 ± 6.8	0.0003
Comorbidities
Hypertension (%)	90	92	1.0
Coronary disease (%)	20	39	0.4
Diabetes (%)	20	39	0.4
Atrial fibrillation (%)	10	18	0.5
Laboratories
Haemoglobin (g/dL)	13.9 ± 1.3	12.5 ± 1.4	0.0003
Creatinine	1.0 (0.8–1.1)	1.15 (0.9–1.5)	0.02
NT-proBNP (pg/mL)	100 (54–236)	437 (147–978)	0.0002
Medications
ACEi/ARBs (%)	50	71	0.2
Beta blockers (%)	40	74	0.02
Loop diuretics (%)	15	39	0.1
Echocardiography
LV diastolic dimension (mm)	48 ± 4	48 ± 5	0.7
LV mass index (g/m2)	88 ± 20	86 ± 22	0.8
LV ejection fraction (%)	62 ± 8	61 ± 8	0.9
E/e′ ratio	8 ± 2	14 ± 6	<0.0001
Medial mitral e′ velocity (cm/s)	8 ± 2	6 ± 2	0.005
Medial mitral s′ velocity (cm/s)	8 ± 2	6 ± 2	0.006
Lateral mitral e′ velocity (cm/s)	9 ± 2	7 ± 3	0.007
Lateral mitral s′ velocity (cm/s)	8 ± 2	7 ± 2	0.01
LV longitudinal strain (%)	16%	13%	0.04
LA volume index (mL/m2)	29 (20–41)	39 (33–57)	0.01
RV basal diameter (mm)	34 ± 8	38 ± 6	0.1
RV mid diameter (mm)	30 ± 7	31 ± 7	0.5

ACEI, angiotensin-converting enzyme inhibitors; ARB, angiotensin-receptor blockers; E, early mitral diastolic inflow velocity; e′, early diastolic mitral annular velocity; HFpEF, heart failure with preserved ejection fraction; LA, left atrial; LV, left ventricular; NT-proBNP, N-terminal pro B-type natriuretic peptide; RV, right ventricular; s′, systolic mitral annular velocity.

Baseline haemodynamics, right ventricular function, and markers of neurohormonal activation

Compared to controls, HFpEF subjects displayed higher right and left heart filling pressures, higher PA pressures, and lower PA compliance and PA oxygen saturation at rest, whereas blood pressures, heart rate, Qp, and PVR were similar between the groups (Table 2). Baseline RV systolic function (RV s′ tissue velocity) was impaired in subjects with HFpEF compared to controls, whereas RV diastolic function (RV e′ tissue velocity) was similar between the groups at rest.

Table 2.

Rest and exercise haemodynamics and RV reserve

	Control (n = 20)	HFpEF (n = 38)	P-value
Baseline
Haemodynamics
Heart rate (min−1)	69 ± 13	66 ± 11	0.3
Systolic blood pressure (mmHg)	142 ± 24	150 ± 22	0.2
Right atrial pressure (mmHg)	4 ± 2	11 ± 4	<0.0001
PA systolic pressure (mmHg)	28 ± 6	43 ± 12	<0.0001
PA mean pressure (mmHg)	17 ± 4	28 ± 8	<0.0001
PCWP (mmHg)	8 ± 3	18 ± 6	<0.0001
PVR (WU)	1.8 ± 0.8	2.0 ± 1.1	0.4
PAC (mL/mmHg)	4.2 ± 1.1	3.5 ± 1.3	0.02
Qp (L/min)	5.3 ± 1.9	5.0 ± 1.2	0.6
PA saturation (%)	73 ± 6	64 ± 5	<0.0001
RV function
RV e′ velocity (cm/s)	9 ± 3	8 ± 3	0.5
RV s′ velocity (cm/s)	12 ± 3	10 ± 2	0.01
Markers of neurohormonal activation
CT-proET-1 (pmol/L)	69 (56–74)	105 (83–130)	<0.0001
MR-proADM (nmol/L)	0.6 (0.5–0.7)	1.0 (0.8–1.4)	<0.0001
20 W exercise
Haemodynamics
Heart rate (min−1)	92 ± 14	88 ± 16	0.4
Systolic blood pressure (mmHg)	173 ± 26	175 ± 29	0.8
Right atrial pressure (mmHg)	9 ± 3	21 ± 5	<0.0001
PA systolic pressure (mmHg)	38 ± 10	68 ± 12	<0.0001
PA mean pressure (mmHg)	25 ± 6	49 ± 11	<0.0001
PCWP (mmHg)	15 ± 5	32 ± 5	<0.0001
PVR (WU)	1.4 ± 0.7	2.5 ± 1.1	<0.0001
PAC (mL/mmHg)	4.6 ± 2.1	2.3 ± 0.8	0.0001
Qp (L/min)	8.2 ± 1.9	6.6 ± 2.0	0.009
PA saturation (%)	53 ± 9	36 ± 9	<0.0001
RV function
RV e′ velocity (cm/s)	14 ± 5	11 ± 5	0.1
RV s′ velocity (cm/s)	14 ± 3	10 ± 3	<0.0001
Markers of neurohormonal activation
CT-proET-1 (pmol/L)	69 (60–80)	114 (88–139)	<0.0001
MR-proADM (nmol/L)	0.6 (0.5–0.7)	1.1 (0.8–1.4)	<0.0001
Peak exercise
Haemodynamics
Heart rate (min−1)	120 ± 16	98 ± 16	<0.0001
Systolic blood pressure (mmHg)	187 ± 26	185 ± 34	0.8
Right atrial pressure (mmHg)	8 ± 4	23 ± 6	<0.0001
PA systolic pressure (mmHg)	42 ± 9	72 ± 12	<0.0001
PA mean pressure (mmHg)	27 ± 7	50 ± 8	<0.0001
PCWP (mmHg)	14 ± 5	34 ± 6	<0.0001
PVR (WU)	1.2 ± 0.6	2.2 ± 1.1	0.0001
PAC (mL/mmHg)	4.3 ± 1.5	2.3 ± 0.9	<0.0001
Qp (L/min)	11.8 ± 3.6	7.7 ± 2.4	<0.0001
PA saturation (%)	44 ± 10	31 ± 10	<0.0001
RV function
RV e′ velocity (cm/s)	22 ± 9	12 ± 6	0.0003
RV s′ velocity (cm/s)	15 ± 3	11 ± 3	<0.0001
Markers of neurohormonal activation
CT-proET-1 (pmol/L)	73 (62–85)	118 (87–138)	<0.0001
MR-proADM (nmol/L)	0.6 (0.5–0.7)	1.0 (0.8–1.3)	<0.0001

CT-proET-1, C-terminal pro-endothelin-1; HFpEF, heart failure with preserved ejection fraction; MR-proADM, mid-regional pro-adrenomedullin; PA, pulmonary artery; PAC, pulmonary artery compliance; PCWP, pulmonary capillary wedge pressure; PVR, pulmonary vascular resistance; Qp, pulmonary blood flow; TV e′, early diastolic tissue Doppler velocity at lateral tricuspid annulus; RV s′, systolic tissue Doppler velocity at lateral tricuspid annulus.

Levels of CT-proET-1 was highly correlated with MR-proADM (r = 0.89, P < 0.0001). The correlation of CT-proET-1 with MR-proADM was significantly greater than its correlation with NT-proBNP (r = 0.61, P < 0.0001 vs. MR-proADM). Heart failure with preserved ejection fraction subjects displayed higher levels of both neurohormones compared to controls (Figure 1A and B). These differences remained significant after adjusting for age, creatinine levels, and BMI (all P < 0.05). Plasma levels of CT-proET-1 and MR-proADM were highly correlated with RAP (r = 0.66 and r = 0.66, both P < 0.0001), PA mean pressures (Take home figure), PA systolic pressures (r = 0.71 and r = 0.59, both P < 0.0001), and PCWP (r = 0.67 and r = 0.62, both P < 0.0001). In addition, there were correlations observed with PVR (r = 0.32, P = 0.02 and r = 0.26, P = 0.05), PA compliance (Take home figure), and PA oxygen saturation (r = −0.53, P < 0.0001 and r = −0.52, P < 0.0001, Supplementary material online, Figure S1).

Figure 1.

(A and B) Compared with controls, subjects with heart failure with preserved ejection fraction displayed higher C-terminal pro-endothelin-1 and mid-regional pro-adrenomedullin levels at rest. CT-proET-1, C-terminal pro-endothelin-1; HFpEF, heart failure with preserved ejection fraction; MR-proADM, mid-regional pro-adrenomedullin.

Take home figure.

Levels of C-terminal pro-endothelin-1 and mid-regional pro-adrenomedullin were highly correlated with mean pulmonary artery pressure (A and B) and pulmonary artery compliance (C and D). CT-proET-1, C-terminal pro-endothelin-1; MR-proADM, mid-regional pro-adrenomedullin; PA mean, mean pulmonary artery pressure; PAC, pulmonary artery compliance.

When dividing HFpEF patients based on the median value of CT-proET-1 (105 pmol/L), those with higher CT-proET-1 displayed worse RV-PA coupling, manifest by lower RV s′ velocity (11 ± 2 cm/s vs. 9 ± 2 cm/s, P = 0.008) and PA compliance (4.0 ± 1.3 mL/mmHg vs. 3.1 ± 1.0 mL/mmHg, P = 0.03) and higher PA mean pressure (24 ± 6 mmHg vs. 31 ± 8 mmHg, P = 0.005) than those with lower levels.

Exercise haemodynamics, right ventricular function, and markers of neurohormonal activation

With low level (20 W) and peak exercise, the differences between HFpEF and controls in cardiac filling pressures, PA pressures, PA compliance, and PA oxygen saturation increased further (Table 2). Subjects with HFpEF displayed lower Qp and higher PVR at 20 W and peak exercise, and lower heart rate at peak exercise as compared to controls. Plasma levels of CT-proET-1 and MR-proADM did not increase acutely with exercise in either HFpEF subjects or controls, with no difference between the groups in the exercise change (P for two-way repeated measures ANOVA 0.2 and 0.5, respectively). However, both CT-proET-1 and MR-proADM levels remained higher in HFpEF patients as compared to controls at all stages of exercise. Similar to baseline, there were significant correlations between neurohormone levels and PA pressures, resistance, and compliance during exercise (Supplementary material online, Table S1).

As compared to controls, subjects with HFpEF reached a lower workload (37 ± 16 W vs. 73 ± 28 W; P < 0.0001) at peak exercise, with markedly reduced peak VO₂ (7.8 ± 2.4 mL/min·kg vs. 14.2 ± 4.1 mL/min·kg; P < 0.0001). During low-level and peak exercise, subjects with HFpEF displayed RV reserve limitation, evidenced by less increases in RV s′ and e′ tissue velocities (Figure 2A). Baseline CT-proET-1 and MR-proADM levels were correlated with poorer RV diastolic reserve (Figure 2B and C), depressed RV systolic reserve (r = −0.34, P = 0.04 and r = −0.34, P = 0.04), lower Qp reserve (Figure 3A and B), and reduced exercise capacity (Figure 3C and D).

Figure 2.

(A) Compared with controls, subjects with heart failure with preserved ejection fraction displayed less increase in diastolic (RV e′) and systolic (RV s′) tissue Doppler velocities at lateral tricuspid annulus during peak exercise. (B and C) Baseline C-terminal pro-endothelin-1 and mid-regional pro-adrenomedullin levels were associated with impaired left ventricular diastolic reserve (ΔRV e′) with exercise. CT-proET-1, C-terminal pro-endothelin-1; HFpEF, heart failure with preserved ejection fraction; MR-proADM, mid-regional pro-adrenomedullin.

Figure 3.

Baseline C-terminal pro-endothelin-1 and mid-regional pro-adrenomedullin levels were highly correlated with limitations in cardiac output reserve with exercise and peak oxygen consumption (VO₂). CT-proET-1, C-terminal pro-endothelin-1; HFpEF, heart failure with preserved ejection fraction; MR-proADM, mid-regional pro-adrenomedullin.

Identification of the abnormal pulmonary vascular reserve phenotype

We next evaluated diagnostic abilities of neurohormonal biomarkers to identify abnormal pulmonary vascular reserve during exercise. Plasma levels of CT-proET-1 and MR-proADM measured at rest identified patients with abnormal PVR during exercise [area under the curve (AUC) 0.73–0.80] as well as impairments in exercise PA compliance (AUC 0.81–0.85) with high sensitivities (Table 3).

Table 3.

Diagnostic accuracy of neurohormonal biomarkers for detecting abnormal pulmonary vascular reserve

	PVR during peak exercise >1.58 WU
	Cut-off	AUC (95% CI)	P-value	Sensitivity (%)	Specificity (%)
CT-proET-1	102 pmol/L	0.80	<0.0001	70	92
MR-proADM	0.78 nmol/L	0.73	0.002	77	68

	PAC during peak exercise <2.1 mL/mmHg
	Cut-off	AUC (95% CI)	P-value	Sensitivity (%)	Specificity (%)
CT-proET-1	102 pmol/L	0.85	<0.0001	86	85
MR-proADM	0.78 nmol/L	0.81	0.001	100	71

AUC, area under the curve; CI, confidential interval; CT-proET-1, C-terminal pro-endothelin-1; MR-proADM, mid-regional pro-adrenomedullin; PAC, pulmonary artery compliance; PVR, pulmonary vascular resistance.

Discussion

We demonstrate for the first time robust relationships between endothelin activation and the counterregulatory hormone adrenomedullin with left-sided filling cardiac pressures, pulmonary vascular load, PA hypoxia, RV function, and exercise capacity, at rest and during exercise in HFpEF. As compared to control subjects, CT-proET-1 and MR-proADM levels at rest and during exercise were markedly elevated in HFpEF subjects. Levels of CT-proET-1 and MR-proADM were correlated with PA mean pressures, PA compliance, PVR, PA saturation, and PCWP at rest, suggesting that the activation of these neurohormones reflect PVD and filling pressures in HFpEF. C-terminal pro-endothelin-1 and MR-proADM levels were also associated with worse haemodynamics, limitations in RV systolic and diastolic reserve, and blunted ability to augment cardiac output during exercise, which led to marked impairment in exercise capacity. These data suggest that activation of the endothelin axis may play a role in the pathogenesis of PVD in HFpEF, while activation of adrenomedullin signalling may serve as a counterregulatory response that mitigates adverse effects from endothelin activation.

Pulmonary vascular disease in heart failure with preserved ejection fraction

Heart failure with preserved ejection fraction is a heterogeneous syndrome, and this heterogeneity is considered to be one of the likely explanations for the failure of clinical trials published in HFpEF to date.⁸ Accordingly, there has been an increasing interest to categorize different phenotypes within the broader spectrum of HFpEF into more pathophysiologically homogenous groups to develop phenotype-specific treatments in the future.^2,3,8,24,30

Pulmonary hypertension is common in patients with HFpEF and leads to symptoms of dyspnoea, ventilatory impairments, reductions in aerobic capacity, and increased risk for hospitalization and death.^31–34 In addition to left atrial hypertension, a substantial number of patients with HFpEF also develop PVD.^2–6 Recent studies have shown that HFpEF patients with PVD demonstrate unique pathophysiologic features, including worse pulmonary haemodynamics and impaired RV systolic reserve with exercise, reduced exercise capacity, and worse outcomes.^4,7 Pulmonary vasomotor dysfunction can be seen even in the earliest stages of disease, before left-sided filling pressures become elevated at rest.²⁵ Collectively, these data suggest that both left atrial hypertension and PVD are important therapeutic targets in HFpEF. The ostensible neurohormonal drivers of left atrial hypertension and abnormalities in the pulmonary circulation have remained unknown in this cohort but are important to identify in order to design new trials targeting this pathophysiology.

Neurohormonal activation in heart failure with preserved ejection fraction with pulmonary vascular disease

Biomarkers may help to characterize underlying pathophysiology of HFpEF.³⁵ Endothelin-1 plays a central role in the development of PAH by inducing pulmonary vasoconstriction, smooth muscle cell proliferation, and pulmonary vascular structural remodelling.³⁶ In contrast, ADM is a cardioprotective peptide and has direct effects on pulmonary circulation, including pulmonary vasodilation and inhibition of vascular structural remodelling.^18–20 ADM is known to be co-upregulated with ET-1 as a counterregulatory response in animal models of PAH.²¹ Accordingly, we hypothesized that both CT-proET-1 and MR-proADM may be associated with greater elevation in filling pressures and PVD in HFpEF.

In the current study, CT-proET-1 and MR-proADM were elevated in subjects with HFpEF compared to control subjects, even after adjusting for age, renal function, and BMI. The two opposing neurohormones were highly correlated with one another, with greater strength of association compared to other counterregulatory hormones such as NT-proBNP. Elevations in CT-proET-1 and MR-proADM were not merely related to left-sided filling pressures, as they were independently associated with pulsatile (pulmonary artery compliance) and steady-state pulmonary vascular load (PVR), which have also been related to adverse outcomes.^6,37 We also observed an inverse correlation between endothelin levels and PA oxygen saturation. These data are consistent with recent experimental studies, as well as a human HFpEF study.²⁵ Hypoxia induces expression of both ET-1 and ADM, and up-regulation of the pulmonary endothelin pathway contributes to the transition from isolated post-capillary PH (Ipc-PH) to combined pre- and post-capillary PH (Cpc-PH) characterized by structural and functional vascular remodelling.^21,38

Heart failure with preserved ejection fraction is a progressive syndrome in which many patients display abnormal pulmonary vasodilation with exercise in the earlier stages that is associated with reduced aerobic capacity²⁵ and adverse outcomes,³⁹ and may be targeted therapeutically.⁴⁰ Elevated levels of both CT-proET-1 and MR-proADM identified patients with this abnormal pulmonary vascular haemodynamics during exercise with reasonably high diagnostic accuracy. These data suggest that activation of the endothelin and adrenomedullin pathways play an important role even in the earlier phase of PVD, and may be a useful non-invasive tool for the screening of HFpEF patients with early-stage PVD, who may respond to treatments to mitigate or prevent development of more advanced pulmonary vascular remodelling and RV dysfunction.

The right ventricle is the upstream organ most affected by PH in HFpEF. Right ventricular dysfunction is common in HFpEF and associated with impaired functional capacity and adverse clinical outcomes.^25,41–44 Even if RV function is preserved at rest in HFpEF, it can worsen during the stress of exercise.^4,25 In the current study, RV systolic and particularly RV diastolic reserve with exercise was impaired in HFpEF. We further demonstrated that the magnitude of elevation in both CT-proET-1 and MR-proADM was associated with the limitations in RV function, as well as blunted ability to augment cardiac output and reduced exercise capacity. Based on previous studies,^13,14,17,20 we speculate that activation of the endothelin pathway in HFpEF might contribute to impairments in RV function by limiting the ability to eject blood through the lungs due to vasoconstriction and vascular remodelling. This may eventually lead to profound limitations in exercise capacity as observed in the current study. Adrenomedullin may serve to counterbalance this vasoconstriction, likely accounting for the observed elevations in MR-proADM and its specific correlation to CT-proET-1.^18–20

Therapeutic implications

While there is some evidence that pulmonary vasodilators might be of benefit in some patients with HFpEF and PVD,⁴⁵ there is concern that increased pulmonary blood flow in the setting of LV diastolic dysfunction may worsen lung congestion.⁴⁶ Clinical trials have demonstrated systemic fluid retention and increased risk of HF hospitalization with endothelin receptor antagonists in HF, leading to neutral or even negative results.^47–50 However, the largest trial of endothelin-A receptor blockade in HFpEF demonstrated an improvement in exercise time.⁵¹ Further study testing endothelin-A receptor blockade appears warranted, particularly since fluid retention (a known side effect of endothelin blockade) may not as prominent in patients with earlier stage HFpEF where resting volume overload is often absent. One multicentre trial to evaluate the efficacy of endothelin receptor antagonist macitentan is currently being tested in the HFpEF-PVD, which includes a run-in period to exclude patients that develop excessive volume overload due to endothelin blockade (NCT03153111).

In contrast to endothelin, adrenomedullin has multiple cardioprotective effects, including pulmonary vasodilation, inhibition of vascular structural remodelling, preservation of vascular integrity, and suppression of myocardial fibrosis and oxidative stress.^18–20 Given the coexpression of endothelin and adrenomedullin,²¹ and the tight correlations between both neurohormones in the current study, we speculate that ADM levels are elevated in HFpEF as a counterregulatory response. Previous small studies examining the effects of intravenous administration of adrenomedullin have demonstrated promising results, including a reduction in PVR and an increase in Qp, without an increase in PCWP.^18,52 The half-life of adrenomedullin in the blood is very short; making long-term treatment through agonist administration problematic, but because adrenomedullin may be cleared from the circulation through neprilysin, treatment with the neprilysin inhibitor sacubitril may also enhance ADM signalling pathways.⁵³

Limitations

This is a single-centre study from a tertiary referral centre and as such has inherent flaws relating to selection and referral bias. However, the invasive characterization of patients both at rest and with exercise at the time of biomarker sampling is a strength of the current study. As this was a mechanistic study, there was no adjustment made for multiple hypothesis testing. The sample size is relatively small, and this limits subgroup analyses and increases the risk for failing to detect a significant group difference that might be apparent in a larger same (Type II error). The cross-sectional design of this study does not permit determination of whether elevated CT-proET-1 and MR-proADM caused pulmonary haemodynamic perturbations, or whether increased LV filling pressure secondary to volume overload caused elevations in CT-proET-1 and MR-proADM to lead to pulmonary vasoconstriction and remodelling (or some combination of both). Future interventional studies with appropriate controls will be necessary to differentiate these possibilities. The control group was not truly normal in that they had prevalent comorbidities, and by the fact that they were referred to invasive exercise stress testing because of exertional dyspnoea. However, this would only be expected to bias the current results towards the null, as completely normal patients would be expected to display even lower levels of neurohormones. As expected based upon the presence of HFpEF, there were baseline differences in age, BMI, and renal function as compared to controls, but all differences remained significant after adjusting for these confounders. We measured MR-proADM, which is a stable non-functional fragment of the adrenomedullin pro-peptide. This does not distinguish between the biologically-active mature adrenomedullin and the non-functional adrenomedullin variant, but MR-proADM has been used in most studies and its associations with haemodynamics and prognosis have been established. We did not measure markers of fibrosis in this study and we cannot evaluate for any potential relationship with measures of PVD.

Conclusions

Subjects with HFpEF, especially those with significant PH due to left heart congestion and PVD, display elevations in CT-proET-1 and MR-proADM levels reflective of neurohormonal activation and pulmonary circulatory abnormalities at rest and during exercise. The increases in CT-proET-1 and MR-proADM levels are associated with pulmonary haemodynamic perturbations, limitations in RV reserve, impaired cardiac output reserve with exercise, and reduced exercise ability in HFpEF. Further study is required to determine if therapies targeting these pathways can improve outcomes in patients with the HFpEF-PH phenotype.

Funding

This research was supported by a competitive prospective grants award from the Department of Cardiovascular Medicine at Mayo Clinic. B.A.B. was supported by RO1 HL128526 and U10 HL110262. V.M. was supported by the Czech Healthcare Research Grant agency (AZV): 17-28784A. Y.N.V.R. was supported by T32 HL007111. M.O. was supported by a research fellowship from the Uehara Memorial Foundation, Japan.

Conflict of interest: none declared.

See page 3718 for the editorial comment on this article (doi: 10.1093/eurheartj/ehz765)