Ventricular Stiffening and Chamber Contracture in Heart Failure with Higher Ejection Fraction

Abstract

Background:

Ancillary analyses from clinical trials have suggested reduced efficacy for neurohormonal antagonists among patients with heart failure and preserved ejection fraction (HFpEF) and higher range EF.

Methods:

621 patients with HFpEF were grouped into those with low-normal LVEF (HFpEF_<65%, n=319, 50%≤LVEF<65%) or HFpEF_≥65% (n=302, LVEF≥65%), and compared with 149 age-matched controls undergoing comprehensive echocardiography and invasive cardiopulmonary exercise testing. A sensitivity analysis was performed in a second noninvasive community-based cohort of patients with HFpEF (n=244) and healthy controls without cardiovascular disease (n=617).

Results:

Patients with HFpEF_≥65% had smaller LV end diastolic volume (LVEDV) than HFpEF_<65%, but LV systolic function assessed by preload recruitable stroke work and stroke work/EDV was similarly impaired. Patients with HFpEF_≥65% displayed an end diastolic pressure volume relationship (EDPVR) that was shifted leftward, with increased LV diastolic stiffness constant β in both invasive and community-based cohorts. Cardiac filling pressures and pulmonary artery pressures at rest and during exercise were similarly abnormal in all EF subgroups. While patients HFpEF_≥57% displayed leftward shifted EDPVR, those with HFpEF_<57% had a rightward shifted EDPVR more typical of HFrEF.

Conclusion:

Most pathophysiologic differences in patients with HFpEF and higher EF are related to smaller heart size, increased LV diastolic stiffness, and leftward shift in the EDPVR. These findings may help to explain the absence of efficacy for neurohormonal antagonists in this group and raise a new hypothesis, that interventions to stimulate eccentric LV remodeling and enhance diastolic capacitance may be beneficial for patients with HFpEF and EF in the higher range.

GRAPHICAL ABSTRACT

Introduction

Heart failure (HF) with preserved ejection fraction (EF, HFpEF) affects over half of all patients with HF but there are few effective treatments.^{1, 2} Secondary analyses from clinical trials have shown that within the broader spectrum of HFpEF, patients with EF falling in the lower range appear to respond more favorably to neurohormonal antagonists compared to those with EF falling within the higher range.^3–7 This has led to questions about where the border between HFpEF and HF with reduced EF should be drawn.^{8, 9} A recent study observed that patients with HFpEF and EF>60% had smaller hearts, increased left ventricular (LV) chamber stiffness during systole and diastole, and differential responses to acute load alterations compared to those with EF 50–59%.¹⁰ The latter study utilized robust LV pressure-volume relationships, but was small, highly selected, and importantly lacked a control group free of HF, and there was no assessments of aerobic capacity or cardiac reserve during dynamic exercise.

To address these knowledge gaps, we performed a detailed, multimodality comparison of cardiovascular structure and function, hemodynamics, and aerobic capacity in patients with HFpEF separated into those with EF falling into the higher and lower normal ranges, along with HF-free controls included to provide additional context. We hypothesized that patients with HFpEF and higher EF would display smaller LV chamber volumes resulting in an effective increase in diastolic chamber stiffness as compared to HFpEF_<65%, and that this may explain many of the pathophysiologic features in this cohort.

Methods

Study Population

The data, analytic methods, and study materials will be made available to other researchers based upon reasonable request for purposes of reproducing the results. This is a retrospective analysis of consecutive patients referred for assessment at the Mayo Clinic for the evaluation of exertional dyspnea. HFpEF was defined by the presence of clinical HF (dyspnea and fatigue), LVEF ≥50% with elevated LV filling pressures [pulmonary capillary wedge pressure (PCWP) ≥15 mmHg at rest or ≥25 mmHg with exercise). Patients with HFpEF were divided in two groups according to LVEF for the primary analysis: 1) HFpEF_<65% (50%≤EF<65%) or 2) HFpEF_≥65%. A convenience sample of consecutively evaluated patients without HF (normal rest/exercise PCWP) with dyspnea related to deconditioning or psychogenic causes were used as controls. Patients with unstable coronary artery disease, significant valvular heart disease (greater than moderate mitral regurgitation, greater than mild stenosis), non-Group 2 pulmonary hypertension, constrictive pericarditis, and hypertrophic or amyloid cardiomyopathy were excluded. The Mayo Clinic Institutional Review Board approved this study, and informed consent was provided by all subjects to use their data for research purposes.

To complement the main analysis from the invasive cohort and include a healthier control group without cardiovascular disease, we also analyzed a separate cohort of patients with and without HFpEF from a community-based sample, in which most of patients with HFpEF had been previously hospitalized, while controls were healthy volunteers recruited by random sampling from the community who also underwent echocardiographic evaluation but no invasive assessment. Some data from this cohort has been published previously^{11, 12} but not examining the impact of different EF cohorts. This cohort underwent additional evaluation of LV systolic properties, including end systolic elastance (Ees) using the single-beat technique, and determination of stress-corrected midwall fractional shortening (sc-MFS), an independent measure of LV myocardial contractility, as previously described.^{11, 12}

Assessment of Cardiac Structure, Function and Hemodynamics

Right heart catheterization was carried out using high fidelity micromanometer catheters at rest and during supine exercise to the point of volitional exhaustion as previously described.¹³ Full details are provided in the Supplementary Material.

Comprehensive 2-D, M-mode, Doppler and tissue Doppler echocardiography was performed according to the current guidelines.¹⁴ Left atrial volume was indexed to height as recently recommended.¹⁵ Left ventricular volumes at end systole and end diastolic (LVEDV) and LVEF were determined using Simpson’s method, and were also indexed to body surface area. In the community-based cohort LVEDV was assessed using the Doppler method assuming absent mitral regurgitation, as previously described.^{11, 12} Myocardial deformation analyses were performed offline from 2D images, using commercially available software (Image Arena, TomTec Imaging Systems, Unterschleissheim, Germany). LV global longitudinal strain (GLS), global longitudinal peak systolic strain rate and global longitudinal early diastolic strain rate were measured using 2D speckle tracking, determined as the average of the two apical views (4- and 2-chamber views).¹⁶

Because measures of strain and myocardial velocities vary with heart size, geometry, and loading conditions, LV contractility was assessed separately using load-independent measures that also account for afterload and preload, including preload recruitable stroke work (PRSW) and LV stroke work (LVSW). PRSW was determined according to single beat method validated by Lee and colleagues: PRSW = SBM_w*(LVEDV–V_w), where SBM_w presents slope of the single beat preload recruitable stroke work relationship [SBM_w=SW/(LVEDV – k*LVEDV + (1-k) * LVM/1.05)], with k defined by the ratio of epicardial shell volumes corresponding to the volume-axis intercept and baseline LVEDV, calculated as k=[0.0004*LVM]+0.6408, with V_w presenting volume within the epicardial shell [V_w=k*(LVEDV+LVM/1.05)–LVM/1.05].^{12, 17} LVSW was calculated as [(mean arterial pressure – PCWP)*SV*0.0136]¹⁸ and then divided by LVEDV to minimize preload sensitivity.¹⁹

Total blood volume (TBV) and plasma volume (PV) were estimated using validated formulas as PV = (1−hematocrit) × (a+[b×weight in kg]), where a=1530 for men and 864 for women, and b=41 for men and 47.9 for women.²⁰ TBV was then calculated as =PV/(1-hematocrit). The probability of underlying undiagnosed cardiac amyloid in patients was assessed using the ATTR-CM score (range −1 to 10), where higher scores indicate greater likelihood of amyloid, and scores ≥6 are considered high risk and warrant further testing.²¹

End-diastolic pressure/volume relationship

The LV end-diastolic pressure/volume relationship (EDPVR) is described by the following equation: P = αV^β. The values of α and β were calculated according to the single-beat approach developed and validated by Klotz et al, where β reflects LV diastolic stiffness.²² This method allows estimation of the entire EDPVR from a single set of pressure–volume coordinates based on the premise that volume-normalized EDPVRs share a common shape. Because α and β are covariant and derived from a highly nonlinear equation, group differences in the EDPVR were compared using the LVEDV at a common LVEDP of 15 mmHg and 30 mmHg (V₁₅ and V₃₀, respectively).²²

Statistical Analysis

Data are presented as median (interquartile range) or mean (standard deviation). Categorical data are expressed as numbers and percentages. The Chi-square test assessed differences in categorical data. Analysis of variance (one way ANOVA, ANCOVA and Kruskal Wallis) was used to assess differences in key variables between study groups, with post hoc comparisons performed using Bonferroni’s method. Linear regression was used to adjust for baseline group differences (sex and BMI) that may influence cardiac structure and function, except for cardiac structural measures that were separately indexed for body size (BSA-normalized LVEDV, V₁₅ and V₃₀, and height-normalized LV mass and LA volume), which were only adjusted for sex in linear regression analyses to avoid overadjustment. SPSS 28.0 (IBM, Armonk, New York) was used for all analyses. All tests with a 2-sided p-value <0.05 were considered statistically significant.

RESULTS

Of 621 patients with HFpEF, 319 (51%) had 50%≤EF<65% (HFpEF_<65%), and 302 (49%) had EF≥65% (HFpEF_≥65%). Compared with age-matched controls (n=149), both HFpEF groups had higher BMI and NTproBNP levels, poorer kidney function, lower hemoglobin, and were more likely to have hypertension and atrial fibrillation (Table 1). Patients with HFpEF_≥65% were more likely to be women and hypertensive as compared to HFpEF_<65%, but less likely to have atrial fibrillation. There were no differences in the prevalence of obesity, diabetes, or coronary disease, and no differences in renal function, hemoglobin, or NT-proBNP levels. ATTR-CM score was lower in patients with HFpEF_≥65%, suggesting lower probability of undiagnosed cardiac amyloid, though the prevalence of high risk scores was low in all groups. Estimated blood and plasma volumes were higher in both HFpEF groups compared with controls, but there were no differences between HFpEF_≥65% and HFpEF_<65%. (Table 1)

Table 1:

Baseline Characteristics

	Controls (n=149)	HFpEF<65% (n=319)	HFpEF≥65% (n=302)	P value
Age, y	67 ± 8	68 ± 11	68 ± 11	0.372
Women, n (%)	75 (50)	153 (48)	196 (65)1, 2	<0.001
BMI, kg/m2	29.4 ± 5.8	33.2 ± 7.41	34.0 ± 7.61	<0.001
Estimated plasma volume, l	2.9 ± 0.6	3.3 ± 0.71	3.4 ± 0.71	0.001
Estimated blood volume, l	4.9 ± 1.0	5.5 ± 1.11	5.4 ± 1.01	0.003
Comorbidities
Diabetes, n (%)	29 (20)	78 (24)	82 (27)	0.203
Hypertension, n (%)	115 (77)	259 (81)	280 (93)1, 2	<0.001
Coronary disease, n (%)	52 (35)	107 (34)	86 (28)	0.267
Atrial fibrillation, n (%)	8 (5)	132 (41)1	94 (31)1, 2	<0.001
Medications
Beta blockers, n (%)	53 (36)	174 (54)1	150 (50)1	0.004
ACE-inhibitor or ARB, n (%)	51 (34)	142 (44)1	153 (51)1	0.004
CCB, n (%)	27 (18)	58 (18)	57 (19)	0.980
Diuretic, n (%)	42 (28)	182 (57)1	169 (56)1	<0.001
Laboratories
NT-proBNP, pg/ml	80 (40,193)	257 (106, 987)1	272 (74, 763)1	<0.001
eGFR, ml/min/1.73 m2	68 ± 16	64 ± 191	63 ± 191	0.019
Hemoglobin, g/dL	13.0 ± 1.4	12.5 ± 1.51	12.3 ± 1.51	<0.001
ATTR-CM score ≥ 6, n (%)	6 (4)	22 (7)	4 (1)	0.129
ATTR-CM score	3 (1,4)	3 (2,5)1	2 (1,4)2	<0.001

ACE = angiotensin converting enzyme, ARB = angiotensin II receptor blocker, ATTR-CM = transthyretin amyloid cardiomyopathy, where higher scores indicate greater probability of undiagnosed cardiac amyloid, and scores ≥6 are considered “high risk” patients that warrant additional testing; BMI = body mass index, CCB = calcium channel blocker, eGFR = glomerular filtration rate, NT-pro-BNP = N-terminal pro-B type natriuretic peptide. Values are expressed as mean ± SD, median (IQR) or numbers (%). Final column shows the p value by analysis of variance for any difference between groups. Significant individual group differences (corrected for multiple hypothesis testing) are then shown by superscripts

¹(p<0.05 vs. Controls)

²(p<0.05 vs. HFpEF_<65%)_.

Cardiac Structure

LV diastolic dimension and LVEDV were smaller in HFpEF_≥65% than HFpEF_<65%, but LV volumes did not significantly differ between HFpEF_≥65% and controls (Table 2). LV mass indexed for height was greater in both HFpEF groups than in controls but did not differ between HFpEF_≥65% and HFpEF_<65%. Left atrial (LA) volume was greater in both HFpEF groups than controls, but LA volume was lower in HFpEF_≥65% compared with HFpEF_<65% (Table 2). Each of these differences remained significant after adjusting for sex.

Table 2:

Cardiac Structure and Function

	Controls (n=149)	HFpEF<65% (n=319)	HFpEF≥65% (n=302)	P valuea
LV end diastolic diameter, mm	48 ± 5	50 ± 5	48 ± 52	0.003
LV end diastolic volume, ml	107 ± 29	118 ± 321	109 ± 262	0.003
LV end-diastolic volume index, ml/m2	54 ± 12	57 ± 13	53 ± 112	0.001
LV mass, g	174 ± 53	204 ± 611	186 ± 522	<0.001
LV mass index, g/m2 .7	41 ± 11	48 ± 131	46 ± 131	0.002
Relative wall thickness	0.41 ± 0.07	0.42 ± 0.07	0.42 ± 0.07	0.148
LA volume, ml	56 ± 14	82± 381	72± 261, 2	<0.001
LA volume index, ml/m2.7	13 ± 3	19 ± 91	18 ± 71,2	<0.001
TR velocity, m/s	2.5 ± 0.3	2.7 ± 0.51	2.7 ± 0.51	<0.001
LV GLS, %	15.9 ± 2.2	14.0 ± 3.31	15.6 ± 2.92	<0.001
LV systolic strain rate, s−1	0.66 ± 0.10	0.57 ± 0.151	0.66 ± 0.132	<0.001
LV diastolic strain rate, s−1	0.53 ± 0.14	0.49 ± 0.15	0.55 ± 0.152	<0.001
LV s’, cm/s	8.2 ± 1.5	6.9 ± 1.61	7.3 ± 1.41	<0.001
LV E/e’	8.4 ± 3.1	12.8 ± 6.31	12.7 ± 4.21	<0.001
RV s’, cm/s	11.8 ± 3.2	10.0 ± 3.21	10.6 ± 2.21	0.002
TAPSE, mm	21 ± 5	17 ± 51	19 ± 5	0.005

LA=left atrial; LV=left ventricle; LVGLS=left ventricular global longitudinal strain; s’=annular systolic velocity; RV=right ventricle; TAPSE=tricuspid annular plain systolic excursion; TR=tricuspid regurgitation.

Values are expressed as mean ± SD, median (IQR) or numbers (%). Final column shows the p value by analysis of variance for any difference between groups.

Significant individual group differences (corrected for multiple hypothesis testing) are then shown by superscripts 1 (p<0.05 vs. Controls)

²(p<0.05 vs. HFpEF_<65%)_.

^aAdjusted for sex and BMI, except for LV end diastolic volume index, LV mass index, and LA volume index, which are only adjusted for sex

Cardiac Function and Functional Reserve

LV function assessed by global longitudinal strain, systolic strain rate, and diastolic strain rate was reduced in HFpEF_<65% as compared to both HFpEF_≥65% and controls (Table 2). None of the strain-based measures were significantly different between HFpEF_≥65% and controls. Tissue Doppler LV systolic tissue velocity was lower in both HFpEF groups compared with controls, while E/e’ ratio was higher in the HFpEF groups compared with controls, with no between-group difference in either tissue-Doppler based measure of LV function between HFpEF_≥65% and HFpEF_<65%.

As compared to controls, both HFpEF_≥65% and HFpEF_<65% displayed chamber-level LV systolic dysfunction, with lower preload recruitable stroke work and stroke work/EDV, but there were no differences between HFpEF_≥65% and HFpEF_<65% in systolic performance (Table 3), despite poorer systolic function in HFpEF_<65% when assessed by speckle-tracking strain and EF (Table 2).

Table 3:

Resting Hemodynamics

	Controls (n=149)	HFpEF<65% (n=319)	HFpEF≥65% (n=302)	P valuea
Heart rate, beats/min	70 ± 11	70 ± 12	70 ± 13	0.799
Mean arterial pressure, mmHg	96 ± 13	99 ± 14	99 ± 15	0.474
Cardiac output, l/min	5.4 ± 1.5	5.3 ± 1.61	5.3 ± 1.61	0.010
AVO2 diff, ml/dl	4.3 ± 0.8	4.7 ± 1.01	4.5 ± 1.02	0.001
Central pressures
RA pressure, mmHg	5 ± 3	11 ± 51	10 ± 51,2	<0.001
PA systolic pressure, mm Hg	28 ± 6	42 ± 141	41 ± 141	<0.001
PA mean pressure, mmHg	17 ± 4	28 ± 91	27 ± 91	<0.001
PCWP, mmHg	9 ± 3	17 ± 51	17 ± 61	<0.001
Vascular function
PVR, dyn·s·cm−5	125 ± 57	183 ± 1281	171 ± 1241	<0.001
PA compliance, ml/mmHg	4.5 ± 1.7	3.7 ± 1.81	3.7 ± 1.81	<0.001
SVR, dyn·s·cm−5	1446 ± 430	1452 ± 460	1460 ± 502	0.271
TAC, ml/mm Hg	1.2 ± 0.5	1.1 ± 0.51	1.1 ± 0.51	0.001
Ea, mm Hg/ml	1.8 (1.6, 2.3)	2.0 (1.6, 2.4)1	2.0 (1.5, 2.6)1	0.001
LV EDPVR
V15, ml	110 (96, 131)	113 (94, 134)	102 (90, 121)1,2	0.001
V15/BSA, ml/m2	57 (49, 65)	53 (46, 62)1	50 (44, 57)1,2	<0.001
V30, ml	124 (108, 146)	128 (106, 150)	115 (101, 136)1,2	0.001
V30/BSA, ml/m2	64 (55, 73)	60 (52, 69)1	56 (49, 64)1,2	<0.001
LV stiffness β	5.90 (5.86, 5.95)	6.08 (5.93, 6.52)1	6.15 (5.95, 6.52)1,2	<0.001
LV systolic function
PRSW, mmHg	94 (77, 117)	89 (65, 115)1	89 (69, 114)1	<0.001
LVSW/EDV (g/ml)	0.85 (0.69, 1.07)	0.70 (0.54, 0.95)1	0.70 (0.59, 0.99)1	<0.001

AVO₂ diff = arterial-venous oxygen content difference; BSA=body surface area; Ea=systemic effective arterial elastance; EDV=LV end diastolic volume; LVSW=left ventricular stroke work; PA=pulmonary artery; PCWP=pulmonary capillary wedge pressure; PRSW=pre-load recruitable stroke work; PVR=pulmonary vascular resistance; RA=right atrial; SVR=systemic vascular resistance; TAC=total arterial compliance; V₁₅=LVEDV at a common LV end-diastolic pressure of 15 mmHg; V₃₀=LVEDV at a common LV end-diastolic pressure of 30 mmHg. Values are expressed as mean ± SD or median (IQR). Final column shows the p value by analysis of variance for any difference between groups.

Significant individual group differences (corrected for multiple hypothesis testing) are then shown by superscripts 1 (p<0.05 vs. Controls)

²(p<0.05 vs. HFpEF_<65%)_.

^aAdjusted for sex and BMI, except V₁₅/BSA and V₁₅/BSA, which are only adjusted for sex

Rest Hemodynamics and Pressure-Volume Relationships

There were no differences in heart rate or blood pressure at rest among the 3 groups (Table 3). As compared with controls, both HFpEF groups displayed higher resting right and left heart filling pressures, higher pulmonary artery (PA) pressures, pulmonary vascular resistance (PVR) and systemic arterial elastance, and lower PA and total arterial compliance, with lower cardiac output (CO). However, there were no differences in resting hemodynamics between patients with HFpEF_≥65% and HFpEF_<65% (Table 3).

As compared to patients with HFpEF_<65% and controls, those with HFpEF_≥65% displayed an EDPVR that was shifted to the left, along with the highest LV stiffness coefficient β, indicating increased LV diastolic chamber stiffness, with lower LVEDV at common LVEDPs of 15 mmHg and 30 mmHg (V₁₅ and V₃₀) (Figure 1, Table 3).

Figure 1:

Mean end diastolic pressure volume relationships (EDPVR) for patients with HFsnEF (blue triangles), HFlnEF (red circles), and controls (black squares). The dashed lines denote mean resting left ventricular filling pressures measured at cardiac catheterization in the two HFpEF groups (upper horizontal line, 17 mmHg) and in the control patients (bottom horizontal line, 9 mmHg). LVEDV, left ventricular end diastolic volume; LVEDP, left ventricular end diastolic pressure.

V₁₅ and V₃₀ did not differ in HFpEF_<65% and controls. Using V₁₅ and V₃₀ indexed to BSA, both HFpEF_<65% and HFpEF_≥65% displayed leftward shifted EDPVR as compared to controls, and patients with HFpEF_≥65% again displayed increased stiffness compared with HFpEF_<65%. V₁₅ had a modest but significant inverse relationship with EF (r= -0.30, p<0.0001) that did not differ between groups.

Exercise Hemodynamics and Cardiac Function

Increases in mean arterial pressure with exercise were similar in all groups, but patients with HFpEF had lower peak heart rate (Table 4). As compared with controls, patients with HFpEF_≥65% and HFpEF_<65% displayed higher exercise right and left heart filling pressures, higher PA pressures, PVR, and systemic vascular resistance (SVR), lower PA compliance, and lower CO. However, there were no significant differences in the severity of exercise hemodynamic abnormalities between patients with HFpEF_≥65% and HFpEF_<65% (Table 4). Comparisons during exercise were similar at matched submaximal workload (20 W, Table S1). Peak VO₂ was similarly impaired in HFpEF_≥65% and HFpEF_<65% compared with controls, with no difference between the two HFpEF groups (Table 4).

Table 4:

Exercise Hemodynamics

	Controls (n=149)	HFpEF<65% (n=319)	HFpEF≥65% (n=302)	P valuea
Heart rate, beats/min	107 ± 22	102 ± 191	92 ± 231	<0.001
Mean BP, mmHg	113 ± 17	115 ± 20	117 ± 22	0.954
CO, l/min	10.2 ± 2.8	8.8 ± 3.31	9.1 ± 3.21	<0.001
AVO2 diff, ml/dl	9.3 ± 2.9	10.0 ± 2.71	9.3 ± 2.92	0.002
VO2, ml/min/kg	11.8 ± 4.0	9.5 ± 3.51	9.2 ± 3.31	<0.001
RER	1.0 ± 0.1	1.0 ± 0.1	1.0 ± 0.1	0.508
Lactate, mmol/L	4.3 ± 2.1	3.7 ± 2.2	3.3 ± 1.9	0.310
Central pressures
RA pressure, mmHg	10 ± 5	20 ± 71	20 ± 71	<0.001
PA systolic pressure, mmHg	47 ± 11	66 ± 161	66 ± 171	<0.001
PA mean pressure, mmHg	31 ± 8	46 ± 101	46 ± 111	<0.001
PCWP, mmHg	18 ± 6	31 ± 61	31 ± 71	<0.001
Vascular function
PVR, dyn·s·cm−5	110 ± 53	165 ± 1601	159 ± 1291	<0.001
PA compliance, ml/mmHg	3.3 ± 1.3	2.8 ± 1.51	2.9 ± 1.51	<0.001
SVR, dyn·s·cm−5	903 ± 324	948 ± 3241	944 ± 3981	0.005
TAC, ml/mm Hg	1.0 ± 0.8	1.0 ± 0.6	1.0 ± 0.5	0.223
Ea, mm Hg/ml	1.8 (1.4, 2.3)	1.8 (1.5, 2.4)	1.8 (1.4, 2.4)	0.083

AVO₂ diff = arterial-venous oxygen content difference; CO=cardiac output; Ea=systemic effective arterial elastance. PA=pulmonary artery; PCWP=pulmonary capillary wedge pressure; PVR=pulmonary vascular resistance; RA=right atrial; RER = respiratory exchange ratio; SVR=systemic vascular resistance; TAC=total arterial compliance; VO₂=Oxygen consumption;

Values are expressed as mean ± SD or median (IQR). Final column shows the p value by analysis of variance for any difference between groups.

Significant individual group differences (corrected for multiple hypothesis testing) are then shown by superscripts 1 (p<0.05 vs. Controls)

²(p<0.05 vs. HFpEF_<65%)_.

^aAdjusted for sex and BMI

A subgroup of patients underwent exercise echocardiography simultaneous with the invasive hemodynamic exercise testing (99/319 HFpEF_<65%, 40/302 HFpEF_≥65% and 38/149 controls). LV and right ventricular (RV) systolic velocities were impaired at rest and during exercise in both HFpEF groups compared with controls, while E/e’ ratio was higher in the two HFpEF groups (Table S2). There were no significant differences between HFpEF_≥65% and HFpEF_<65% in biventricular function during exercise.

Sensitivity Analyses

Sensitivity analyses were conducted to explore differing LVEF cutpoints (50–57%, 58–69%, and ≥70%) in a more granular fashion, and to evaluate women and men separately, given the well-known differences in heart size and EF in the sexes (Tables S3–S8). Consistent with the primary analyses, LV dimension and LVEDV were lower in both women and men with HFpEF with higher LVEF (Tables S3, S6). Men with HFpEF in the highest LVEF category had lower LV mass, along with smaller LVEDV, and greater RWT than men with HFpEF and LVEF in the lower ranges. LV strain was higher in both men and women in the higher LVEF ranges compared to lower LVEF ranges, and the EDPVR was shifted to the left in both women and men when evaluated separately and at these more granular LVEF cutpoints above (Tables S3–S8).

Similar to the primary analyses, there were no differences in rest or exercise hemodynamics in men and women with HFpEF in differing EF categories. Differences in the LV EDPVR were more evident using comparisons across HFpEF groups defined in this way, where patients with HF and EF≤57% displayed an EDPVR that was shifted to the right vs controls, while patients with HFpEF and increasing EF displayed EDPVR that was shifted progressively more leftward (Figure 2). These results were apparent in men and women when evaluated separately (Figures S1–S2).

Figure 2:

Mean end diastolic pressure volume relationships (EDPVR) for patients with for patients with HFpEF separated into those with lower EF (≤57%, red circles), middle EF (58–69%, blue triangles), and the highest EF (≥70%, pink triangles) as compared to controls (black squares). Abbreviations as in Figure 1.

Community-Based Cohort

In sensitivity analyses where a separate cohort of patients with HFpEF (n=244) were compared with healthy controls without cardiovascular disease (n=617) from a community-based cohort, we again observed lower V₁₅ and V₃₀, and increased LV stiffness constant β in HFpEF_≥65% as compared with both HFpEF_<65% and controls (Table 5, Figure 2). Left ventricular EDV was significantly lower in HFpEF_≥65% compared with both HFpEF_<65% and healthy controls. Ees was higher in HFpEF_≥65% compared to both HFpEF_<65% and controls, but sc-MFS was lower in both HFpEF groups compared to controls, being lowest in HFpEF_<65% (Table 5, Figure 3).

Table 5.

Cardiac Structure and Function in the Community-Based Sample

	Controls (n=617)	HFpEF<65% (n=156)	HFpEF≥65% (n=88)	p
LVEDV (ml)	134±31	141±40	119±301,2	<0.0001
LVEDVI (ml/m2)	73.1±13.0	72.6±17.6	64.5±14.31,2	<0.0001
LV mass (g/m2.7)	39.6±7.7	50.4±15.71	53.7±18.51	<0.0001
LVEF (%)	63±5	59±41	69±41,2	<0.0001
Ees (mmHg/ml)	1.99±0.59	2.18±0.761	2.72±0.921,2	<0.0001
Sc-mFS (%)	100±10	89±141	97±91,2	<0.0001
LV stiffness β	6.03 (6.01, 6.09)	6.06 (5.94, 6.57)	6.31 (6.01, 6.95)1,2	<0.0001
V15 (ml)	108±23	108±33	93±251,2	<0.0001
V15/BSA (ml/m2)	59.2±10.9	55.3±14.81	50.8±12.11,2	<0.0001
V30 (ml)	121±26	121±37	105±281,2	<0.0001
V30/BSA (ml/m2)	66.5±12.2	62.3±16.61	57.2±13.61,2	<0.0001

BSA = body surface area; EF = ejection fraction; Ees = LV end-systolic elastance; LV = left ventricular; LVEDV = LV end diastolic volume; LVEDVI = left ventricular volume index; LVM = Left ventricular mass; Sc-mFS = stress corrected mid-wall fractional shortening; V₁₅ = LVEDV at a common LV end-diastolic pressure of 15 mmHg; V₃₀ = LVEDV at a common LV end-diastolic pressure of 30 mmHg.

Values are expressed as mean ± SD or median (IQR). Final column shows the p value by analysis of variance for any difference between groups.

Significant individual group differences by the Tukey HSD test are then shown by superscripts 1 (p<0.05 vs. Controls)

²(p<0.05 vs. HFpEF_<65%)_.

Figure 3:

Left ventricular structure and function in the community-based cohort. Patients with HFpEF_≥65% displayed lower left ventricular (LV) end diastolic volume index (LVEDVI) compared with HFpEF_<65% and healthy controls. Estimated LV volume indexed to body surface area at an EDP of 15 mmHg (V₁₅ index) was lower in HFpEF_<65% than controls, and lower in HFpEF_≥65% than both HFpEF_<65% and controls. LV end systolic elastance (Ees) increased from controls to HFpEF_≥65% and HFpEF_≥65%, but LV myocardial contractility assessed by stress-corrected midwall fractional shortening (Sc-MFS) was lower in both HFpEF groups than controls.

Additional sensitivity analyses were conducted to explore differing LVEF cutpoints used in the literature (50–59% and ≥60%, Tables S9–S11). Consistent with the primary analyses, LV dimension and LVEDV were lower in HFpEF with higher LVEF and leftward shifted EDPVR (lower V₁₅ and V₃₀).

Discussion

We performed a detailed evaluation of cardiac structure, function, and hemodynamics in patients with HFpEF with higher EF (≥65%) and lower EF (50–64%), comparing these groups to one another and to controls without HF. As compared to HFpEF_<65%, patients with HFpEF_≥65% were more likely to be women and hypertensive, and less likely to have atrial fibrillation. Left ventricular and LA volumes were lower in HFpEF_≥65% than HFpEF_<65%, even after accounting for sex and BMI. LV volumes in HFpEF_≥65% did not differ from controls in the invasive cohort, but LVEDV was lower in HFpEF_≥65% than controls in the community-based cohort. Strain-based measures of LV deformation were more impaired in HFpEF_<65% than HFpEF_≥65% and controls, but LV systolic function assessed by load-independent measures was impaired in both HFpEF_≥65% and HFpEF_<65%, despite higher Ees. The EDPVR was leftward shifted and LV β was higher in HFpEF_≥65% compared to both HFpEF_<65% and controls, indicating greater LV diastolic stiffness. Despite these differences, patients with HFpEF_<65% and HFpEF_≥65% displayed similar hemodynamic derangements at rest and with exercise, with similar impairments in aerobic capacity. Collectively, these data indicate that the most consistent and fundamental defining feature of patients with HF and EF in the higher range is a smaller, stiffer LV chamber with both diastolic and systolic myocardial dysfunction. The reduction in LV chamber size in HFpEF_≥65% may explain the less impressive benefit from neurohormonal antagonists in these patients observed in secondary analyses from trial data, and suggests a new hypothesis, that therapies that promote LV eccentric remodeling and improved diastolic capacitance, such as exercise training or interventions to reduce external LV constraint, might be effective to improve outcomes for patients with HF and EF falling in the higher range.

Systolic Properties in HFpEF with Higher EF

Nearly 40 years ago, Topol and colleagues described a form of HF characterized by LV hypertrophy, supranormal EF, and diastolic dysfunction.²³ More recently, Rosch et al. observed that LV end-systolic stiffness (elastance, Ees) was higher in HFpEF with EF>60% compared with EF 50–60%.¹⁰ However, increases in both EF and Ees should not necessarily be interpreted as evidence of heightened contractility. Ees is a load-independent measure of systolic properties, but it is also influenced by diastolic LV stiffness and chamber size. ²⁴ Indeed, Ees is characteristically elevated in patients with HFpEF²⁵ even in the presence of significant contractile dysfunction.¹²

Here, utilizing measures of LV systolic function that are independent of diastolic stiffening, loading conditions, and that account for chamber volume, such as PRSW and LVSW/EDV, we found that patients with HFpEF_≥65% and HFpEF_<65% have similar impairment in systolic function. The impairments in contractility by LV systolic velocities, PRSW, SW/EDV, and Sc-MFS in HFpEF_≥65% were relatively modest at rest, but prior studies have shown that even mild impairments in systolic function may lead to marked impairments during exercise, and the lower contractility observed in HFpEF_≥65% compared with controls does not support the idea that these patients present with a hypercontractile state, as recently stated, which would make negative inotropic interventions such as cardiac myosin inhibitors less likely to be effective.

Remodeling and Chamber Stiffening in HFpEF with Higher EF

The most conspicuous feature we observed in HFpEF_≥65% was a smaller, stiffer LV as compared to HFpEF_<65%. This difference is partly related to the higher proportion of women in HFpEF_≥65%, as noted in other studies,^{3, 10, 23} and is consistent with the fact that women have smaller hearts than men.²⁶ However, both LVEDV and LVEDVI remained lower in HFpEF_≥65% compared with HFpEF_<65% even after adjusting for sex, and LA volumes were also lower, further evidence for a reduction in cardiac size in HFpEF_≥65%. Lower LA volume may be a consquence of the same chamber contracture and stiffening processes that affect the LV, potentially leading to a greater increase in pulmonary vascular pressures relative to LVEDP due to impairments in LA reservoir function. In the community-based sample, LV volumes were lower in HFpEF_≥65% compared to both HFpEF_≥65% and controls, whereas there was no difference in heart volumes compared to controls in the invasive cohort. This may be due to the fact that controls in the invasive arm were not truly normal, as they presented with exertional dyspnea, and had other indicators of preclinical HFpEF, such as reduced GLS.^{27, 28}

In agreement with Rosch et al,¹⁰ we also observed that patients with HFpEF_≥65% displayed a leftward shifted EDPVR compared to patients with lower EF. However, the study from Rosch et al. did not include a control group without HFpEF. Here, we observed that patients with HFpEF_≥65% are characterized by a small, stiffer LV chamber filling on the steeper portion of its EDPVR (Figures 1 and 2). Conversely, the EDPVR was shifted to the right in patients with HFpEF and the lowest EF as compared to HFpEF with higher EF and controls, indicating eccentric remodeling (Figure 2).

Rosch et al. identified a greater increase in LVEDP in those with higher EF during isometric handgrip, despite less ability to augment LVEDV, resulting in a blunted increase in cardiac output.¹⁰ In the present study we observed that PCWP was similarly elevated in HFpEF_≥65% and HFpEF_<65%, both at rest and with exercise. Impairments in cardiac output reserve were also similarly abnormal, as were exercise-induced pulmonary hypertensive responses, right heart filling pressures, LV and RV functional reserve by simultaneous stress echocardiography (Table S2) and peak VO₂. The reasons for the discrepant results may relate to the types of stress tested. Rosch and colleagues evaluated isometric handgrip, which increases blood pressure and heart rate, but induces much less dramatic changes in cardiac preload and afterload when compared with dynamic cycle ergometry.

The present data may help reconcile discrepant results from prior studies, which have variably shown that LV volumes are increased,^{29, 30} similar to,¹¹ or even decreased³¹ in HFpEF compared to controls. Similar discrepancies have been reported in the EDPVR in prior studies, where patients with HFpEF have been shown to display a leftward shift in some,^{11, 32} as here in HFpEF_≥65%, but not others.^{25, 33} The present data suggest that these between-study differences may relate at least partly to degree of inclusion of HFpEF patients with smaller and larger hearts and EF in these prior studies. Patients with HFpEF and lower EF did not display leftward shift in the EDPVR, and in the group with EF≤57% the EDPVR was rightward shifted (Figure 2), consistent with remodeling rather than volume contracture. This finding is more typical of HFrEF and is consistent with the observation that neurohormonal antagonists appear to have greater efficacy in HFpEF with lower EF.^3–5

Mechanisms for Lack of Benefit with Neurohormonal Antagonists

In HFrEF the LV is dilated, and treatment with neurohormonal antagonists leads to reverse remodeling.³⁴ Here, we show that in HFpEF_≥65%, the ventricle is contracted, which would suggest that therapies that work in large part through reverse remodeling would be less efficacious. There could also be safety concerns that might blunt benefits from some neurohormonal antagonists in HFpEF_≥65%. Vasodilators may be poorly tolerated in these patients,²³ and are known to produce greater hypotensive effects in patients with higher EF, due in part to excessive reduction in preload from venodilation.^{10, 35} The lack of benefit from reverse remodeling and poor tolerability to preload reduction may partly explain a lack of benefit in some patients.

Potential New Treatments for Patients with Higher EF

More importantly, the present data suggest a new hypothesis, that perhaps LV volume is too small in many patients with HFpEF ang higher EF, and it may be that these patients might benefit from modest eccentric remodeling—the opposite of what is desired with conventional neurohormonal antagonists.³⁴

One intervention that elicits this type of physiologic eccentric remodeling is exercise training. There is shrinkage in LVEDV with normal aging in humans, which this is associated with compensatory increases in EF.³⁶ Age-associated LVEDV contraction is greatest among individuals with increasing resistive and pulsatile arterial loading,³⁷ vascular abnormalities that are also common in patients with HFpEF.³⁸ Sedentary behavior is a strong risk factor for HFpEF³⁹ which may also promote volume contraction. Howden et al. showed that sustained increases in activity levels increase LV volume and diastolic capacitance (higher EDV with similar PCWP), changes that would be expected to reduce risk for HFpEF.⁴⁰ Changes in EF were not reported with training in this study, but proportional increases in EDV were more than two-fold greater than increases in stroke volume, which would imply a decrease in EF with training.

In contrast, among patients with HFpEF, exercise training has not been shown to promote eccentric remodeling,⁴¹ which suggests a potential role for other approaches. In this light, pericardiotomy may be another potential treatment for HFpEF_≥65%. We did not observe that pericardial restraint was greater in HFpEF patients with higher EF, but because pericardiectomy or pericardiotomy leads to eccentric remodeling and rightward shift in animals and humans,^42–44 along with improvements in LV diastolic reserve with volume loading,^{45, 46} we hypothesize that patients with HFpEF_≥65% may be positioned to derive greater benefit from this intervention based upon the present data. Increases in LVEDV during exercise are often blunted in HFpEF, despite marked elevation in PCWP, suggesting greater pericardial restraint on the heart limiting filling but augmenting LVEDP.^{47, 48}

In adults undergoing pericardiotomy for cardiac surgery, there is a mild, balanced increase in LV volume and mass that develops in the weeks following surgery, very similar to the morphologic changes observed with exercise training.⁴³ Preliminary studies in animal models and patients at risk for HFpEF indicate that increases in LV filling pressure with volume loading are markedly attenuated through pericardiotomy,^{45, 46} suggesting that targeting this external constraint could improve diastolic reserve. In tandem with these studies, the present data suggest that even modest increases in LVEDV and reductions in external constraint may be even more helpful in patients with HFpEF_≥65.

Limitations

The present data come from a single center leading to selection and referral bias. The observational nature of these data does not permit ascription of causality. The invasive control group were not completely healthy controls as they were referred for invasive testing because of exertional dyspnea, and they displayed other findings of cardiac dysfunction such as reduced GLS. However, the main results were confirmed in the sensitivity analysis using healthy controls free of cardiovascular disease drawn from a community-based sample (Table 5).^{11, 12} While LV filling pressures were measured using high fidelity micromanometers, LV mass and volume were measured by transthoracic echocardiography, which is less precise compared to cardiac MRI. However, this limitation would only be expected to reduce sensitivity to detect group differences rather than the specificity of the findings, and the limited precision related to echocardiography assessments applies equally to cases and controls, so there is no bias.

Conclusion

Patients with HFpEF and LVEF falling in the higher range display similar impairments in hemodynamics and functional capacity as those with HFpEF and lower EF. Left ventricular volume is lower and chamber stiffness is higher in HFpEF_≥65% than HFpEF_<65% and healthy controls without cardiovascular disease. Collectively, these data indicate that small left ventricular size and increased stiffness are defining features of HFpEF and higher EF. This may partially explain the differential responses to a number of treatments observed in clinical trials to date, and also raises a new hypothesis that interventions to increase LV cavity volume and improve diastolic capacitance may be effective to improve clinical status for patients with HFpEF and higher EF.

Abbreviations

HFpEF

heart failure of preserved ejection fraction

HFpEF_<65%

heart failure and low-normal ejection fraction (50%≤LVEF<65%)

HFpEF_≥65%

heart failure of higher ejection fraction (LVEF≥65%)

LA

left atrium

LV

left ventricle

LVEDP

left ventricular end-diastolic pressure

LVEDV

left ventricular end-diastolic volume

EDPVR

end-diastolic pressure/volume relationship

PA

pulmonary artery

PCWP

pulmonary capillary wedge pressure

PRSW

single-beat pre-load recruitable stroke work

V₁₅

LVEDV at a common LVEDP of 15 mmHg

V₃₀

LVEDV at a common LVEDP of 30 mmHg