Left ventricular diastolic dysfunction and exercise intolerance in obese heart failure with preserved ejection fraction

Abstract

This study tested the hypothesis that early left ventricular (LV) relaxation is impaired in older obese patients with heart failure with preserved ejection fraction (HFpEF), and related to decreased peak exercise oxygen uptake (peak V̇o₂). LV strain and strain rate were measured by feature tracking of magnetic resonance cine images in 79 older obese patients with HFpEF (mean age: 66 yr; mean body mass index: 38 kg/m²) and 54 healthy control participants. LV diastolic strain rates were indexed to cardiac preload as estimated by echocardiography derived diastolic filling pressures (E/e′), and correlated to peak V̇o₂. LV circumferential early diastolic strain rate was impaired in HFpEF compared with controls (0.93 ± 0.05/s vs. 1.20 ± 0.07/s, P = 0.014); however, we observed no group differences in early LV radial or longitudinal diastolic strain rates. Isolating myocardial relaxation by indexing all three early LV diastolic strain rates (i.e. circumferential, radial, and longitudinal) to E/e′ amplified the group difference in early LV diastolic circumferential strain rate (0.08 ± 0.03 vs. 0.13 ± 0.05, P < 0.0001), and unmasked differences in early radial and longitudinal diastolic strain rate. Moreover, when indexing to E/e′, early LV diastolic strain rates from all three principal strains, were modestly related with peak V̇o₂ (R = 0.36, −0.27, and 0.35, respectively, all P < 0.01); this response, however, was almost entirely driven by E/e′ itself, (R = −0.52, P < 0.001). Taken together, we found that although LV relaxation is impaired in older obese patients with HFpEF, and modestly correlates with their severely reduced peak exercise V̇o₂, LV filling pressures appear to play a much more important role in determining exercise intolerance.

NEW & NOTEWORTHY Using a multimodal imaging approach to uncouple tissue deformation from atrial pressure, we found that left ventricular (LV) relaxation is impaired in older obese patients with HFpEF, but only modestly correlates with their severely reduced peak V̇o₂. In contrast, the data show a much stronger relationship between elevated LV filling pressures and exercise intolerance, refocusing future therapeutic priorities.

INTRODUCTION

Heart failure (HF) with preserved ejection fraction (HFpEF) is the fastest growing form of HF and is associated with a 65% five year mortality rate (1, 2). The primary chronic symptom in HFpEF even when clinically stable and nonedematous is decreased exercise tolerance, measured objectively as decreased peak oxygen uptake (peak V̇o₂) (3–9). The physiological mechanisms underpinning the decreased peak V̇o₂ in older obese patients with HFpEF has not been well characterized, however may be due in part, to impaired left ventricular (LV) diastolic function (9–12).

Diastolic dysfunction is often defined by an elevated LV end-diastolic pressure; however, this static measure—obtained from a single time-point in the cardiac cycle only at end–diastole— fails to provide insight into dynamic LV tissue relaxation during early diastole. Various imaging techniques, including Doppler ultrasound and cardiac magnetic resonance imaging (CMRI), are commonly used to assess this early period of diastole; however, they are often insensitive to detect HFpEF in outpatients with unexplained dyspnea in an early stage of the disease (13, 14); potentially leaving a large proportion of patients undetected in daily clinical practice. This may be explained by the fact that both Doppler echocardiography and CMRI measures of diastolic function are heavily influenced by loading conditions. Indeed, data from a prior CMRI study highlight the importance of controlling for preload (15). However, this proof-of-concept study used CMRI-derived volume-time profiles, which is incredibly time consuming, operator dependent, and clinically infeasible.

Myocardial feature tracking of conventional CMRI cine images alleviates many of the aforementioned limitations by reducing operator dependence and postprocessing time, and provides detailed quantitative information about myocardial relaxation (i.e., diastolic strain rate). Strain and strain rate are markers of deformation, and have been proven to be a powerful diagnostic and prognostic indicators for global and regional LV function. Although diastolic strain rate is increasingly recognized as a powerful discriminator of diastolic dysfunction, and predictive of cardiovascular events (16–18), like other imaging approaches, it remains preload dependent (19). We therefore sought to isolate diastolic relaxation from the hemodynamic driving pressure by correcting the diastolic strain rate for LV preload (estimated with Doppler ultrasound as the ratio of early diastolic mitral inflow velocity to early diastolic mitral annulus velocity, E/e′ ratio). Using this metric, we hypothesized that LV early diastolic strain rate indexed to LV filling pressure would be impaired in patients with HFpEF compared with controls, and be associated with peak V̇o₂.

METHODS

Study Participants

All patients with HFpEF met the previously described inclusion criteria (6): ≥60 yr of age; body mass index ≥30 kg/m²; signs and symptoms of HF defined by the National Health and Nutrition Examination Survey score ≥3 (20), or the previously published criteria (21) or both; LV ejection fraction ≥50%; no segmental wall motion abnormalities; no contraindications to CMRI; and no significant ischemic or valvular heart disease, pulmonary disease, anemia, or other disorders that could explain the patients’ symptoms (4, 22, 23). Patients with significant pericardial and valvular disease were also excluded from the study. Healthy controls were recruited and screened, and excluded if they had any chronic medical illness, were on any chronic medication, had current complaints or an abnormal physical examination (including blood pressure ≥140/90 mmHg), had abnormal results on the screening tests (echocardiogram, electrocardiogram, and cardiopulmonary exercise testing), or regularly undertook vigorous exercise (4, 23). All control subjects had a normal LV filling pattern (24). The study was approved by the Wake Forest University Health Sciences Institutional Review Board. All participants provided written informed consent.

Each participant was studied on two separate occasions (8 ± 9 days apart for HFpEF, 14 ± 20 days for controls: on the first visit, the participants completed an echocardiogram performed after ≥15 min of supine rest, followed by an incremental cardiopulmonary exercise test. On the second visit, participants completed a resting CMRI study. Before study visits, all participants were instructed to refrain from vigorous exercise for 24 h leading up to the laboratory visit.

Echocardiography

Echocardiography was performed with all participants resting in a semirecumbent position for ≥15 min. Doppler echocardiography was used to assess LV filling patterns, mitral septal annular velocity, and pulse-wave velocity in accordance with the American Society of Echocardiography recommendations (25), as previously described (6). All Doppler values represent the average of three cardiac cycles. The E/e′ ratio was calculated as a surrogate measure of LV filling pressures (25). Left atrial diameter was measured at the widest region of the left atrium in the 4-chamber view in end-diastole.

Cardiopulmonary Exercise Testing

Cardiopulmonary exercise testing was performed on a motorized treadmill using the modified Naughton protocol, or the Modified Bruce in control subjects with higher self-reported physical activity level, as previously described (6, 26). Participants were given detailed, standardized instructions before performing symptom-limited exhaustive test to volitional fatigue. Metabolic gas exchange was measured continuously during exercise and averaged over 15-s intervals (Medgraphics Ultima, Medical Graphics Corp., St. Paul, Minnesota) (6, 23, 26). Peak V̇o₂ was calculated as the average of measures from the last 30 s during peak exercise (6, 23, 26).

Left Ventricular Morphology and Deformation by CMRI

LV mass and volumes were assessed by CMRI (1.5 T Avanto, Siemens Healthineers) from a series of multislice, multiphase gradient-echo sequences positioned perpendicular to the long-axis of the ventricles (short-axis), spanning apex to base (6). Typical CMRI imaging parameters included: flip angle 76°; repetition/echo time: 40–50/1.1–1.2 ms; slice thickness: 7–8 mm; with 25 cardiac phases. From these images, the epi- and endocardial borders of each slice were traced manually at end-diastole and end-systole, with volumes and mass derived via the method of disks. LV stroke volume and ejection fraction were calculated accordingly. All mass and volume data were expressed as absolute values and indexed to body surface area. LV concentricity was calculated by dividing LV mass by LV end-diastolic volume.

The primary end point of this investigation was LV early diastolic strain rate by myocardial feature tracking. All LV strain data was measured offline using a commercially available software package (Cvi⁴²; v 5.3.0, Circle Cardiovascular Imaging Inc., Calgary, AB, Canada). This approach has indeed been previously validated (27–29), with excellent agreement against gold-standard CMRI tissue tagging reported by our group (30). Briefly, the endo- and epicardial borders of the previously described cine images were manually delineated at end-diastole before the feature tracking algorithm was applied. Caution was taken to avoid including slices which included the LV outflow tract and/or left atrium, apical slices without clear delineation of the LV lumen at end-systole (a minimum of 2 cm proximal to luminal obliteration), and insufficient tracking quality. Early diastolic strain rate is the slope (i.e., speed) of myocardial deformation during early diastole, reflective of early LV relaxation and diastolic function (31). Early diastolic strain rates were calculated in the longitudinal, circumferential, and radial planes. Horizontal long-axis cine images were tracked to derive longitudinal strain and strain rate, whereas short-axis cine images were used to derive circumferential and radial strain and strain rate. Strain and strain rates were obtained for each segment and the global values were defined as the mean of all segmental values. All of the data were analyzed by a single observer (T.J.S.), who was blinded to each subject’s medical history and/or group allocation. Our in-laboratory intra-rater reliability in 15 randomly selected subjects, expressed as a coefficient of variation, for each of the primary endpoints is as follows (means ± standard deviation): LV circumferential strain, 3.1% ± 2.3%; LV early diastolic circumferential strain rate, 4.9% ± 4.6%; LV radial strain, 3.6% ± 2.9%; LV early diastolic radial strain rate, 6.7% ± 7.1%; LV longitudinal strain, 4.4% ± 3.5%; and LV early diastolic longitudinal strain rate, 4.6% ± 4.4%. Moreover, our in-laboratory interrater reliability, expressed as a coefficient of variation, for each of the primary endpoints is as follows (means ± standard deviation): LV circumferential strain, 8.4% ± 9.5%; LV early diastolic circumferential strain rate, 8.8% ± 7.5%; LV longitudinal strain, 8.1% ± 6.1%; and LV early diastolic longitudinal strain rate, 10.2% ± 7.9%.

Statistical Analysis

All variables were tested for normality with histograms and quantile-quantile plots. Participant characteristics were compared between HFpEF and healthy control groups with t tests for continuous variables and chi-squared tests for categorical variables. Strain, strain rates, and strain rates indexed to E/e′ were also analyzed using analysis of covariance to adjust for group differences of age, sex, body mass index (BMI), and race. Pearson correlation coefficients were calculated on HFpEF and healthy control groups combined to analyze univariate relationships between E/e′, strain measures, and peak V̇o₂. Multivariate linear regression was used to evaluate associations between strain measures and peak V̇o₂ adjusted for age, sex, race, and BMI. To assess the relative contribution of E/e′ in strain rates indexed to E/e′, univariate and multivariate regression was performed with E/e′ and circumferential, radial, and longitudinal strain rates. Two-sided P < 0.05 was considered significant for all statistical tests. All statistical analyses were performed in SAS Enterprise Guide v 7.11.

RESULTS

Patient Characteristics

Patients with HFpEF were clinically stable (NYHA Class II and III) with typical characteristics of HFpEF. Relative to control participants, patients with HFpEF were more predominantly female, less frequently of white race, had increased BMI, higher resting systolic blood pressure, increased Doppler derived LV E/e′ and impaired filling patterns, and lower absolute and relative peak V̇o₂ (Table 1).

Table 1.

Patient characteristics

Variable	HFpEF	Healthy Controls	P Value
Participant characteristics
n	79	54
Age, yr	66.4 ± 5.2	69.1 ± 7.3	0.027
Women, n (%)	66 (84)	33 (61)	0.004
White, n (%)	42 (53)	51 (94)	<0.001
Body weight, kg	101.3 ± 14.5	73.9 ± 14.8	<0.001
BSA, m2	2.0 ± 0.2	1.8 ± 0.2	<0.001
BMI, kg/m2	38.4 ± 5.1	25.9 ± 4.5	<0.001
NYHA Class II, n (%)	52 (66)	0 (0)	<0.001
NYHA Class III, n (%)	27 (34)	0 (0)	<0.001
Resting HR, beats/min	73 ± 11	65 ± 10	<0.001
Echocardiogram measures
Ejection fraction, %	61.0 ± 6.1	58.9 ± 4.8	0.036
Relative wall thickness, mm	0.57 ± 0.12	0.38 ± 0.05	<0.001
Diastolic filling pattern†
Normal, n (%)	2 (3)	54 (100)	<0.001
Impaired relaxation, n (%)	69 (87)	0 (0)	<0.001
Pseudonormal, n (%)	8 (10)	0 (0)	0.016
Restrictive, n (%)	0 (0)	0 (0)	0.99
E/A ratio	0.9 ± 0.2	0.9 ± 0.2	0.19
Septal e′, cm/s	6.1 ± 1.5	7.9 ± 1.6	<0.001
Lateral e′, cm/s	7.3 ± 1.8	9.6 ± 2.0	<0.001
E/e′ ratio	13.0 ± 3.6	9.3 ± 2.0	<0.001
LA diameter, cm	4.0 ± 0.5	3.4 ± 0.5	<0.001
CMRI measures
LV EDV, mL	116.2 ± 31.0	114.1 ± 25.2	0.69
LV EDV index, mL/m2	56.7 ± 14.0	61.9 ± 11.6	0.027
LV mass, g	88.4 ± 22.0	81.7 ± 18.3	0.074
LV mass index, g/m2	43.1 ± 9.6	44.1 ± 7.3	0.50
LV concentricity, g/mL	0.78 ± 0.17	0.74 ± 0.19	0.16
Medical history
Current atrial fibrillation, n (%)	2 (3)	0 (0)	-
History of diabetes mellitus, n (%)	51 (65)	0 (0)	-
History of hypertension, n (%)	75 (95)	0 (0)	-
Systolic blood pressure, mmHg	133 ± 13	124 ± 11	<0.001
Diastolic blood pressure, mmHg	77 ± 8	75 ± 6	0.12
Current medications
ACE-inhibitors, n (%)	28 (36)	0 (0)	-
Diuretics, n (%)	56 (72)	0 (0)	-
β-blockers, n (%)	30 (38)	0 (0)	-
Calcium antagonists, n (%)	25 (32)	0 (0)	-
Nitrates, n (%)	6 (8)	0 (0)	-
ARBs, n (%)	26 (33)	0 (0)	-
Peak V̇O2, mL·kg−1·min−1	14.9 ± 0.98	25.3 ± 0.29	<0.001
Peak V̇O2, mL/min	1,506 ± 36	1,865 ± 83	<0.001

Values are means ± SD, or n (%); †diastolic filling pattern determined according to the American Society of Echocardiography Criteria; ARB, angiotensin receptor blocker; ACE, angiotensin-converting enzyme; BMI, body mass index; BSA, body surface area; CMRI, cardiac magnetic resonance imaging; E, E-wave velocity; e′, early mitral annulus velocity (septal); HFpEF, heart failure with preserved ejection fraction; HR, heart rate; LA, left atrial; LV, left ventricle; NYHA, New York Heart Association. P values derived from t tests and χ² tests, where appropriate.

Myocardial Feature Tracking

Early LV diastolic circumferential strain rate was significantly reduced in HFpEF compared to controls; however, we observed no group differences in early LV radial or longitudinal diastolic strain rates (Table 2). We also observed no significant relationships between early LV diastolic strain rate and peak V̇o₂ (Fig. 1).

Table 2.

Left ventricular strain and strain rates in HFpEF and healthy controls

Variable	Raw Means ± SD			LS Means ± SE
	HFpEF	Control	P value	HFpEF	Control	P value
Radial strain, %	45.8 ± 1.2	49.3 ± 1.6	0.071	45.4 ± 1.6	49.9 ± 2.1	0.16
Systolic radial SR, s-1	3.0 ± 0.1	2.6 ± 0.1	0.013	2.9 ± 0.1	2.7 ± 0.2	0.31
Early diastolic radial SR, s-1	−2.5 ± 0.1	−2.9 ± 0.2	0.064	−2.5 ± 0.2	−2.9 ± 0.2	0.18
Late diastolic radial SR, s-1	−1.2 ± 0.1	−1.0 ± 0.04	0.004	−1.23 ± 0.09	−0.92 ± 0.11	0.073
Circumferential strain, %	−22.1 ± 0.4	−23.4 ± 0.4	0.021	−21.8 ± 0.5	−23.8 ± 0.6	0.042
Systolic circumferential SR, s-1	−1.23 ± 0.03	−1.16 ± 0.03	0.095	−1.25 ± 0.04	−1.13 ± 0.05	0.12
Early diastolic circumferential SR, s-1	0.95 ± 0.04	1.18 ± 0.06	0.001	0.93 ± 0.05	1.20 ± 0.07	0.014
Late diastolic circumferential SR, s-1	0.93 ± 0.05	0.83 ± 0.04	0.11	0.95 ± 0.05	0.80 ± 0.07	0.16
Longitudinal strain, %	−20.0 ± 0.4	−18.4 ± 0.5	0.025	−19.6 ± 0.5	−19.1 ± 0.7	0.64
Systolic longitudinal SR, s-1	−1.16 ± 0.03	−1.04 ± 0.03	0.005	−1.16 ± 0.03	−1.04 ± 0.05	0.086
Early diastolic longitudinal SR, s-1	0.91 ± 0.04	0.91 ± 0.03	0.92	0.93 ± 0.05	0.88 ± 0.07	0.61
Late diastolic longitudinal SR, s-1	1.08 ± 0.04	0.79 ± 0.03	<0.001	1.03 ± 0.05	0.86 ± 0.07	0.079

Raw data are means ± SD; LS values are means ± SE, adjusted for age, sex, BMI, and race. HFpEF, heart failure with preserved ejection fraction; SR, strain rate. P values derived from t tests and analysis of covariance to adjust for group differences in age, sex, BMI, and race. Boldface indicates significance.

Figure 1.

Indexing measures of early diastolic radial, circumferential, and longitudinal (A–C) strain rate to E/e′ (blue) unmasks significant relationships with peak V̇O₂, not observed before indexing (black). These data highlight the importance of controlling for preload when interpreting measures of early diastolic relaxation. Data are presented as linear regression and 95% confidence intervals.

Indexing all three early LV diastolic strain rates (i.e., circumferential, radial, and longitudinal) to E/e′ exaggerated the group difference in early circumferential diastolic strain rate, and revealed significant group differences in early radial and longitudinal diastolic strain rates (Table 3). Moreover, indexing these early LV diastolic strain rates to E/e′ unmasked modest relationships between early LV relaxation and peak V̇o₂, which were not otherwise observed (Fig. 1). However, this relationship appears to be driven almost entirely by E/e′, which by itself was more closely related to peak V̇o₂ (Fig. 2). Indeed, in univariate regression modeling, E/e′ was significantly predictive of peak V̇o₂ (β = −1.05 ± 0.15, P < 0.001), whereas early diastolic circumferential, radial, longitudinal strain rate were not. Moreover, E/e′ retained a significant relationship with peak V̇o₂ in separate multivariate models that included early diastolic circumferential (β = −1.08 ± 0.17, P < 0.001), radial (β = −1.07 ± 0.17, P < 0.001), and longitudinal strain rate (β = −1.06 ± 0.15, P < 0.001), and when adjusted for age, sex, BMI, and race (Table 4).

Table 3.

Left ventricular early diastolic strain rates indexed to E/e′

	HFpEF	Healthy Controls	P Value
n	79	54
Early diastolic circumferential SR / E/e′	0.08 ± 0.03	0.13 ± 0.05	<0.001
Early diastolic radial SR / E/e′	−0.21 ± 0.09	−0.32 ± 0.13	<0.001
Early diastolic longitudinal SR / E/e′	0.08 ± 0.03	0.10 ± 0.04	<0.001
Adjusted for age, sex, BMI, and race
Early diastolic circumferential SR / E/e′	0.08 ± 0.01	0.13 ± 0.01	<0.001
Early diastolic radial SR / E/e′	−0.21 ± 0.02	−0.31 ± 0.02	0.003
Early diastolic longitudinal SR / E/e′	0.08 ± 0.01	0.09 ± 0.01	0.19

Values are means ± SE; E/e′, early mitral inflow-to-early septal annular velocity; HFpEF, heart failure with preserved ejection fraction; SR, strain rate. P values derived from t tests.

Figure 2.

Relationship between left ventricular early mitral inflow velocity-to-early annular tissue velocity (E/e′) and peak oxygen consumption (V̇O₂). Data are presented as linear regression and 95% confidence intervals.

Table 4.

Multivariate relationships between left ventricular early relaxation and peak V̇O₂

Multivariate Regressions
Adjusted for age, sex, race, and BMI	Parameter Estimates	P Value
E/e′	−0.34 ± 0.11	0.002
Early diastolic radial SR	0.70 ± 0.35	0.05
Early diastolic circumferential SR	−1.72 ± 1.05	0.10
Early diastolic longitudinal SR	−1.15 ± 1.09	0.29
Indexed data
Early diastolic radial SR / E/e′	0.87 ± 3.52	0.80
Early diastolic circumferential SR / E/e′	3.96 ± 9.40	0.67
Early diastolic longitudinal SR / E/e′	13.36 ± 10.41	0.20

Values are means ± SE. E/e′, Doppler early mitral inflow-to-early septal annular velocity; SR, strain rate. Data analyzed using multivariate linear regression.

LV systolic function by CMRI feature tracking are also detailed in Table 2. We observed a marked impairment in LV circumferential strain in patients with HFpEF compared with controls, while LV longitudinal strain was significantly higher. After adjustment for age, sex, BMI, and race, however, differences in longitudinal systolic strain were annulled.

DISCUSSION

The data herein provide novel insight into the importance of controlling for cardiac preload when interpreting measures of early diastolic relaxation, and the mechanisms contributing to impaired exercise tolerance in obese HFpEF. Isolating diastolic relaxation from the hemodynamic driving pressure, by indexing measures of early LV diastolic strain rate to E/e′, unmasked impaired early LV relaxation in our well-phenotyped cohort of obese patients with HFpEF, which was modestly related to peak V̇o₂. This relationship, however, appears to be driven almost entirely by preload itself, with E/e′ being more closely related to peak V̇o₂.

Impaired Diastolic Function in HFpEF

LV diastolic dysfunction is a common feature of HFpEF and is often confirmed using invasively measured end-diastolic pressures (32–34). However, this invasive measure is not always clinically feasible, and only provides insight into a single period of the cardiac cycle. Although noninvasive surrogate measures like E/e′ have been widely adopted, they are often insensitive to detect HFpEF in outpatients with unexplained dyspnea in an early stage of the disease (13, 14). Diastolic strain imaging has emerged as a powerful noninvasive tool for identifying diastolic dysfunction due to its excellent sensitivity, specificity, and ability to predict major cardiovascular events and mortality (35, 36). Indeed, adding diastolic strain rate imaging to conventional risk factors significantly improved discrimination and reclassification for HF and atrial fibrillation in a substudy of the multiethnic study of atherosclerosis (16). Together, this underscores our findings of impaired early diastolic circumferential strain rates in HFpEF compared with controls.

Deformation of the LV is strongly dependent on its geometry and on the preload and afterload to which it is subjected. A major limitation of both Doppler and diastolic strain rate imaging however, is that it they are both influenced by loading conditions. To address this limitation, we adapted a technique recently described by Heida et al. (15), and indexed our CMRI derived measures of LV relaxation to the driving force of blood entering the ventricle; thereby isolating myocardial relaxation from the hemodynamic driving pressure. This technique, 1) modestly amplified the group difference previously observed with early diastolic circumferential strain rate and 2) unmasked new group differences in early longitudinal and radial diastolic strain rates. This novel approach provides important additive insight into the interpretation, and prognostic value, of diastolic strain rates in HFpEF and other analogous conditions presenting with diastolic dysfunction.

The relationship between changes in passive and active biophysical properties of the myocardium and the resulting LV displacement and strain fields is complex. We speculate that the impairment in early LV relaxation observed herein, is likely the product of: 1) reduced titin-mediated elastic recoil and early diastolic suction (37, 38); 2) impaired ventricular interaction, such that pericardial constraint and/or pulmonary hypertension may limit diastolic mechanics (39); 3) large artery stiffness and impaired ventricular-arterial coupling, affecting calcium handling and actin-myosin cross-bridge cycling (40), and/or 4) cardiometabolic derangements, leading to ectopic myocardial fat deposition and diastolic dysfunction (9, 41). More work is needed, however, to define the specific mechanism(s) contributing to the present results.

Relationship between LV Early Relaxation and Exercise Capacity

After controlling for cardiac preload, all three early LV diastolic strain rates (radial, circumferential, and longitudinal) modestly correlated with peak V̇o₂. It must be acknowledged, however, that combining early diastolic strain rates with E/e′ adds variance to the relationship with peak V̇o₂, compared to E/e′ alone, redirecting the focus of future work to cardiac filling pressures and the remaining ∼73% of variance contributing to peak V̇o₂ and exercise intolerance in HFpEF. Moreover, that the relationships were no longer preserved following adjustment for age, sex, BMI, and race, strongly suggests that each of these risk factors are themselves contributing to impaired early LV relaxation. That E/e′ appears to be a superior, independent, predictor of peak V̇o₂, compared to early LV relaxation, is consistent with recent observations in obese patients with HFpEF describing a relationship between severely impaired peak V̇o₂ and increased pulmonary capillary wedge pressure and pericardial constraint (42). We extend these observations, by defining the contribution of early LV relaxation to peak V̇o₂. That early LV relaxation (indexed to E/e′) was only weakly related to peak V̇o₂, and largely dependent on E/e′, adds novel mechanistic insight into the determinants of exercise intolerance in this at risk population. Specifically, we interpret these data to suggest that early LV relaxation contributes only minimally to exercise intolerance in HFpEF, whereas elevated cardiac filling pressures play a more dominant role. These data are particularly timely, given the current focus on interventions aimed at reducing cardiac filling pressures both acutely (e.g., NCT04068844) and chronically (e.g., NCT03088033).

Impaired Systolic Function in HFpEF

Patients with HFpEF often present with subclinical systolic dysfunction (10, 43). Here, we report a significant impairment in LV systolic circumferential strain [an important determinant of LV ejection fraction (44)] in HFpEF compared with controls. This is entirely consistent with previous observations, suggesting that LV circumferential strain is ∼25% lower in HFpEF than healthy age-matched control subjects (45–49), with the level of impairment related to the severity of HF diagnosis (46). In contrast to prior observations however (45–51), we found that LV longitudinal strain was greater in HFpEF than controls. Although several reasons may exist to explain these desperate results, including differences in data collection and analysis (i.e., CMRI vs. echocardiography, feature tracking vs. speckle tracking), we believe the most likely explanation relates to the fact that the group differences in longitudinal strain were abrogated once adjusted for age, sex, BMI, and race. Indeed, age, sex, and obesity are known to affect LV mechanics (52–54), and therefore likely played a key role driving the differences in systolic longitudinal deformation observed herein. That we were able to identify an altered balance between global circumferential and longitudinal shortening, despite normal LV ejection fraction, is consistent with other disease states, including Fabry disease (55).

Limitations

The study sample size was moderate, limiting the inference of these results to a wider population. Although there were some intergroup differences in participant characteristics, data were adjusted for age, sex, BMI, and race. Although the BMI of our control group was lower than that of the HFpEF cohort, we believe that a comparison with a true healthy, successful aging control group, free of other cardiovascular risk factors is important. Our analyses were also limited by the lack of invasive hemodynamic data regarding LV filling pressures. Nonetheless, the noninvasive measurement of LV filling pressure used herein is currently recommended by the consensus of the American Society of Echocardiography and European Association of Cardiovascular Imaging (25). Relatedly, our estimate of LV filling pressure could be criticized for sharing some of the same properties as our measure of early LV relaxation (namely e′ and early diastolic strain rate). Future studies that include direct pressure measurements, preferably collected simultaneously, would help to strengthen the interpretation of results. Although the CMRI and echo-Doppler were not simultaneous, they were in close proximity and there was no change in patient condition (i.e., clinically stable) between the two exams. The fact that we observed significant negative relationships between resting LV diastolic function and peak V̇o₂ is a significant addition to this field. However, future work should seek to confirm and extend this work by examining the relationship between LV diastolic function and V̇o₂ during maximal and sub-maximal exercise. Although the present data focused on early diastolic strain rates from CMRI, the fundamental concept proposed herein can be applied to a variety of alternative imaging modalities (e.g., speckle tracking echocardiography). In this way, the purpose of the present investigation is less about the integration of multimodel imaging, as much as it provides a unique opportunity to advance our pathophysiological understanding of exercise intolerance in HFpEF. Finally, the observational nature of our study precludes inferences of attribution.

CONCLUSIONS

Using a multimodel imaging approach, we found that uncoupling tissue deformation from cardiac preload unmasks impairments in LV relaxation in older obese patients with HFpEF that otherwise was not seen before indexing to E/e′. The data further show that impairments in early diastolic relaxation (indexed to E/e′) modestly correlate with the severely reduced peak exercise V̇o₂ seen in older, obese, HFpEF. Although these data highlight the importance of controlling for cardiac preload when assessing LV relaxation, the stronger relationship observed between E/e′ and peak V̇o₂ redirects future investigation and therapeutic focus toward cardiac hemodynamics.

GRANTS

This study was supported in part by National Institutes of Health Grants R01HL136601, P01HL137630, R01AG18917, R01AG045551, and R01HL107257; P30-AG21331 and U24 AG05964, and by the Kermit G. Phillips II Chair in Cardiovascular Medicine at the Wake Forest School of Medicine (to. D. W. Kitzman), and American Heart Association Grant 18PRE33960358 (to T. J. Samuel). B. Upadhya has received research funding from Novartis and Corvia Medical.

DISCLOSURES

Dr. Kitzman has been a consultant for Relypsa, Abbvie, Regeneron, GlaxoSmithKline, Merck, Corvia Medical, Bayer, and St. Luke’s Medical Center in Kansas City, Kansas, receiving grant support from Novartis, Bayer, and St. Luke’s Medical Center in Kansas City, Kansas, and owning stock in Gilead Sciences. Dr. Brubaker has been a consultant for Merck, Corvia Medical, and Boehringer Ingelheim. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

T.J.S., D.W.K., and M.D.N. conceived and designed research; B.U., P.B., and W.G.H. performed experiments; T.J.S., M.B.N., and M.D.N. analyzed data; T.J.S., D.W.K., M.J.H., B.U., P.B., W.G.H., and M.D.N. interpreted results of experiments; T.J.S. prepared figures; T.J.S. and M.D.N. drafted manuscript; T.J.S., D.W.K., M.J.H., B.U., P.B., M.B.N., W.G.H., and M.D.N. edited and revised manuscript; T.J.S., D.W.K., M.J.H., B.U., P.B., M.B.N., W.G.H., and M.D.N. approved final version of manuscript.