Heart Failure with Normal Natriuretic Peptide Levels and Preserved Ejection Fraction: A Prospective Clinical and Cardiac MRI Study

Abstract

Purpose

To describe the clinical presentation, comprehensive cardiac MRI characteristics, and prognosis of individuals with predisposed heart failure with preserved ejection fraction (HFpEF).

Materials and Methods

This prospective cohort study (part of MISSION-HFpEF [Multimodality Imaging in the Screening, Diagnosis, and Risk Stratification of HFpEF]; NCT04603404) was conducted from January 1, 2019, to September 30, 2021, and included individuals with suspected HFpEF who underwent cardiac MRI. Participants who had primary cardiomyopathy and primary valvular heart disease were excluded. Participants were split into a predisposed HFpEF group, defined as HFpEF with normal natriuretic peptide levels based on an HFA-PEFF (Heart Failure Association Pretest Assessment, Echocardiography and Natriuretic Peptide, Functional Testing, and Final Etiology) score of 4 from the latest European Society of Cardiology guidelines, and an HFpEF group (HFA-PEFF score of ≥ 5). An asymptomatic control group without heart failure was also included. Clinical and cardiac MRI-based characteristics and outcomes were compared between groups. The primary end points were death, heart failure hospitalization, or stroke.

Results

A total of 213 participants with HFpEF, 151 participants with predisposed HFpEF, and 100 participants in the control group were analyzed. Compared with the control group, participants with predisposed HFpEF had worse left ventricular remodeling and function and higher systemic inflammation. Compared with participants with HFpEF, those with predisposed HFpEF, whether obese or not, were younger and had higher plasma volume, lower prevalence of atrial fibrillation, lower left atrial volume index, and less impaired left ventricular global longitudinal strain (-12.2% ± 2.8 vs -13.9% ± 3.1; P < .001) and early-diastolic global longitudinal strain rate (eGLSR, 0.52/sec ± 0.20 vs 0.57/sec ± 0.15; P = .03) but similar prognosis. Atrial fibrillation occurrence (hazard ratio [HR] = 3.90; P = .009), hemoglobin level (HR = 0.94; P = .001), and eGLSR (per 0.2-per-second increase, HR = 0.28; P = .002) were independently associated with occurrence of primary end points in participants with predisposed HFpEF.

Conclusion

Participants with predisposed HFpEF showed relatively unique clinical and cardiac MRI features, warranting greater clinical attention. eGLSR should be considered as a prognostic factor in participants with predisposed HFpEF.

Keywords: Heart Failure with Preserved Ejection Fraction, Normal Natriuretic Peptide Levels, Cardiovascular Magnetic Resonance, Myocardial Strain, Prognosis

Clinical trial registration no. NCT04603404

Supplemental material is available for this article.

© RSNA, 2024

Summary

Participants with predisposed heart failure with preserved ejection fraction (HFpEF) had higher plasma volume levels, lower left atrial volume index, and less impaired strain parameters but similar prognosis compared with participants with HFpEF. Early-diastolic global longitudinal strain rate was independently associated with adverse outcomes.

Key Points

• ■ Cardiac MRI–derived early-diastolic global longitudinal strain rate (eGLSR) could help demonstrate cardiac function of participants with predisposed heart failure with preserved ejection fraction (HFpEF; 0.57/sec ± 0.15) versus participants with HFpEF (0.52/sec ± 0.20) (P = .03).

• ■ eGLSR should be considered as a prognostic factor (per 0.2-per-second increase, hazard ratio = 0.28; P = .002) in participants with predisposed HFpEF.

Introduction

Heart failure with preserved ejection fraction (HFpEF) comprises up to 50% of HF cases and is associated with considerable adverse outcomes and few efficient therapeutic strategies due to its multiple phenotypes (1,2). Improved phenotyping of HFpEF could increase understanding of its pathophysiology and guide phenotype-targeted therapies. To achieve this, our group launched a large-scale project, MISSION-HFpEF (Multimodality Imaging in the Screening, Diagnosis, and Risk Stratification of HFpEF; NCT04603404), which aimed to investigate the phenotypes, diagnosis, and risk stratification of participants with HFpEF. To date, we have illustrated the clinical presentation and distinct imaging features in participants with HFpEF and essential hypertension (3) as well as participants with HFpEF and obesity (4). With further investigation, we found that some participants with HFpEF may have normal natriuretic peptide (NP) levels, which makes diagnosis challenging (5,6). Recent research has shown that participants with HFpEF and normal NP levels, diagnosed by cardiac catheterization, had mild diastolic dysfunction and preserved cardiac output reserve during exercise; furthermore, these participants had an intermediate risk of all-cause mortality or HF hospitalization when compared with participants with HFpEF with high NP and those without HF. Therefore, it is clinically important to identify and classify this HFpEF phenotype with normal NP levels (6). Because the risks associated with invasive cardiac catheterization have limited its wide use in the clinical setting, diagnosing HFpEF based on the latest European Society of Cardiology HFA-PEFF (Heart Failure Association Pretest Assessment, Echocardiography and Natriuretic Peptide, Functional Testing, and Final Etiology) score would be more accessible and easily implemented (2,7). Using this score, we proposed an innovative and clinical-based classification of a predisposed HFpEF cohort—participants with HF with preserved ejection fraction and normal NP levels who showed diastolic dysfunction and cardiac remodeling (reaching four points by HFA-PEFF score)—and attempted to illustrate the subclinical cardiac function and myocardial tissue characteristics of this scarcely demonstrated predisposed HFpEF cohort.

Cardiac MRI is the reference standard for cardiac morphology and function (8). In particular, cardiac MRI feature tracking and extracellular volume (ECV) fraction provide advantages in identifying subtle dysfunction and myocardial fibrosis (3,9). Previous publications have also confirmed reliability and prognostic value of strains and ECV fraction in the HFpEF spectrum (10–12). Hence, this study aimed to use a feasible and accessible approach to describe the clinical features and cardiac MRI–derived left ventricular (LV) remodeling, subtle dysfunction, and tissue characteristics in individuals with predisposed HFpEF compared with definite HFpEF participants and the control group and assess short-term clinical follow-up.

Materials and Methods

Study Participants

As part of the MISSION-HFpEF project, this was a prospective single-center cohort study approved by the Ethics Review Board of Fuwai Hospital (Ref. 2019–1307). Written informed consent was obtained from all study participants. Consecutive participants with suspected HFpEF who underwent cardiac MRI at Fuwai Hospital between January 1, 2019, and September 30, 2021, were enrolled. The first group included participants with HFpEF, defined as an echocardiographic and natriuretic peptide score of 5 or more points in the latest European Society of Cardiology guidelines (2,7), which requires the following: (a) symptoms and/or signs of HF; (b) LV ejection fraction of 50% or more; (c) elevated N-terminal pro–brain NP (NT-proBNP) level: NT-proBNP level of more than 125 pg/mL or B-type NP (BNP) level of more than 35 pg/mL in sinus rhythm or NT-proBNP level of more than 365 pg/mL or BNP level of more than 105 pg/mL in atrial fibrillation (AF) (to convert pg/mL to ng/L, multiply by 1.0); and (d) relevant structural heart disease and LV diastolic dysfunction. The second group included participants with predisposed HFpEF and normal NP levels (NT-proBNP level of 125 pg/mL or less or BNP level of 35 pg/mL or less in sinus rhythm), defined as an echocardiographic and natriuretic peptide score of 4 points and manifested as relevant structural heart disease and LV diastolic dysfunction (2,7). Participants with predisposed HFpEF were not confirmed by cardiac catheterization as in the referenced literature (6). The third group comprised asymptomatic control individuals without HF, recruited from the community and the hospital, who had an associated imaging examination proving no abnormality. In all three groups, individuals with primary cardiomyopathy, primary valvular heart disease, significant coronary artery disease, idiopathic pulmonary artery hypertension, pericardial constriction, severe arrhythmia, severe renal dysfunction (glomerular filtration rate of less than 30 mL/min/1.73 m² [to convert to mL/sec/1.73 m², multiply by 0.0167]), and poor image quality were excluded. Plasma volume was estimated by (1 − hematocrit) × (a + [b × weight in kilograms]), where a = 1530 for men and 864 for women, and b = 41 for men and 47.9 for women (13).

Echocardiography Protocol and Analysis

Echocardiography with conventional and tissue Doppler imaging was performed using a commercial US system (EPIQ 7; Philips Healthcare). All images were digitally stored for offline analysis (W.W., with more than 20 years of experience in echocardiography) on a dedicated workstation (QLAB 13.0; Philips Healthcare) with clinical data blinded. Mitral inflow and mitral annulus parameters were measured at rest from the apical four-chamber view by pulse and tissue Doppler imaging.

Cardiac MRI Protocol and Image Analysis

All participants underwent cardiac MRI examinations using 3-T scanners (MR750; General Electric HealthCare or Ingenia; Philips Healthcare or Magnetom Skyra; Siemens Healthcare) during sinus rhythm with retrospective electrocardiographic gating and an eight-channel cardiac coil array. Steady-state, free-precession, breath-hold cine images were obtained in three long-axis planes and in short-axis sections. A modified Look-Locker inversion recovery sequence was used for T1 mapping, as has been previously described (14). Late gadolinium enhancement (LGE) image acquisition started at 10 minutes after administration of gadolinium-based contrast agent. The detailed imaging sequences are presented in Appendix S1.

All cardiac MR images were transferred to an offline workstation and processed with commercial processing software packages by two authors (J.H. and W.Y., each with 4 years of experience in cardiac MRI) while blinded to echocardiography results and clinical data. Structural and functional parameters were measured by CVI42 (Circle Cardiovascular Imaging) (15). LV end-diastolic volume and LV end-systolic volume and mass were indexed by body surface area. We used methods previously described to calculate the ECV fraction with QMap (Medis QMap) (14). Hematocrit level was determined for each participant from a venous blood sample drawn less than 24 hours before cardiac MRI examination. LGE was semiautomatically quantified using the full-width half-maximum method with manual correction using QMass (Medis QMass). Any obvious blood pool or pericardial partial volume artifacts were manually corrected. For strain analysis, the endocardial and epicardial contours were automatically detected with manual correction in end systole and end diastole using CVI42. Papillary muscles were excluded from the endocardial contour (3). Subsequently, global longitudinal strain (GLS), systolic global longitudinal strain rate, and early-diastolic global longitudinal strain rate (eGLSR) were derived from three long-axis cine images; global circumferential strain (GCS), global radial strain, systolic global circumferential strain rate, systolic global radial strain rate, early-diastolic global circumferential strain rate, and early-diastolic global radial strain rate were derived from the short-axis stack. Representative left atrial volume, myocardial strains, and ECV fraction are plotted in Figure 1.

Figure 1:

Measurement of left atrial volume, global longitudinal strain (GLS), and extracellular volume (ECV) fraction. Representative (A) two-chamber and (B) four-chamber long-axis images of left atrial volume. (A–C) Left ventricular (LV) long-axis images of GLS and rate. (E–G) LV short-axis images of global radial strain and global circumferential strain and strain rate. (I–K) LV short-axis images of ECV fraction mapping. (D, L) Bull’s-eye diagram of the corresponding GLS and ECV fraction values, respectively. (H) Representation of the values of GLS, early-diastolic GLS rate, and systolic global longitudinal strain rate in the line graph. eGLSR = early-diastolic global longitudinal strain rate, GLSR = global longitudinal strain rate, sGLSR = systolic global longitudinal strain rate.

Clinical Outcomes

The primary end points included all-cause death, HF hospitalization, or stroke (16), analyzed as time to the first event, and were determined by assessing hospital charts and follow-ups. HF hospitalization was defined as admission to the hospital with signs and symptoms of HF as the major reason for hospitalization and need for diuretic treatment (17). All suspected outcome events were reviewed by two independent investigators blinded to cardiac MRI data.

Statistical Analysis

Continuous data are reported as means ± SDs if normally distributed or medians (IQRs) if nonnormally distributed, and categorical data are reported as numbers (percentages). Comparisons among the HFpEF cohort, predisposed HFpEF cohort, and control cohort were made using one-way analysis of variance, Kruskal-Wallis tests, or χ² tests, with post hoc least significant difference analysis as appropriate. Sensitivity analyses were also performed in subgroups of participants with HFpEF and predisposed HFpEF without obesity or coronary artery disease. Two-sample independent t tests, Wilcoxon rank sum tests, or χ² tests were used where appropriate to assess the difference between the participants with predisposed HFpEF with and without end points. Univariable Cox regression was initially performed to identify variables associated with outcomes. To avoid model overfitting, particular attention was paid to variables that have previously reported prognostic importance (11,18–20). All the covariates associated with end points at P < .05 were then chosen for subsequent multivariable Cox analyses to identify independent predictors using forward stepwise likelihood ratio test. Intraobserver and interobserver variability in measurement of cardiac MRI parameters were analyzed in terms of intraclass or interclass correlation coefficient and Bland-Altman plots. Two-sided P < .05 was considered statistically significant. All analyses were performed using SPSS Statistics for Windows, version 26.0 (IBM); OriginPro 2021, version 9.8 (OriginLab); and R software, version 4.1.2 (R Project for Statistical Computing).

Results

Participant Characteristics

A total of 464 participants were enrolled in this study, including 213 with HFpEF (mean age, 59 years ± 14 [SD]; 139 [65.3%] men and 74 [34.7%] women), 151 with predisposed HFpEF (mean age, 50 years ± 14; 114 [75.5%] men and 37 [24.5%] women), and 100 in the control group (mean age, 45 years ± 11; 45 [45.0%] men and 55 [55.0%] women) (Fig 2). Among the overall HF cohort, participants with predisposed HFpEF and normal NP levels were common, accounting for 41.5% (151 of 364). Participants in both the HFpEF and predisposed HFpEF groups were older and less likely to be women and had greater body mass index (calculated as weight in kilograms divided by height in meters squared), higher plasma volume, and more obvious systemic inflammation than control participants (Table 1). Participants with predisposed HFpEF were younger and had higher body mass index, higher plasma volume levels, and a higher prevalence of hypertension (91.4% [138 of 151] vs 81.2% [173 of 213]; P = .01) and obesity (55.0% [83 of 151] vs 27.2% [58 of 213]; P < .001) yet a lower prevalence of AF (12.6% [19 of 151] vs 39.4% [84 of 213]; P < .001) compared with those with HFpEF (Table 1). Moreover, participants with predisposed HFpEF displayed higher hemoglobin levels and glomerular filtration rate but lower NT-proBNP levels and less use of loop diuretics compared with participants with HFpEF (Fig 3). There was no evidence of differences in sex or New York Heart Association functional class between the HFpEF and predisposed HFpEF groups. In the subgroups of participants with HF without obesity, the same significant differences were found for the majority of parameters as found in the overall HF cohort (Table S1).

Figure 2:

Study participant flowchart. BNP = B-type natriuretic peptide, CTEPH = chronic thromboembolic pulmonary hypertension, HFA-PEFF = Heart Failure Association Pretest Assessment, Echocardiography and Natriuretic Peptide, Functional Testing, and Final Etiology, HFpEF = heart failure with preserved ejection fraction, LVEF = left ventricular ejection fraction.

Table 1:

Baseline Characteristics of Study Participants

Figure 3:

Scatterplots compare clinical and laboratory markers in heart failure with preserved ejection fraction (HFpEF), predisposed HFpEF, and control groups. GFR = glomerular filtration rate, NT-proBNP = N-terminal pro–brain natriuretic peptide. * P < .05 versus predisposed HFpEF. ^† P < .05 versus control group.

Cardiac Structure and Function

At echocardiography, no evidence of a difference was observed between participants with predisposed HFpEF and HFpEF, except for the early-diastolic mitral inflow velocity (Table S2). Table 2 shows the detailed cardiac MRI data for all enrolled participants. Both participants with HFpEF and predisposed HFpEF showed slightly lower LV ejection fraction (but still within the HFpEF range) and higher LV end-diastolic volume index, LV end-systolic volume index, and LV mass index when compared with controls. There was no evidence of differences in heart rate, LV end-diastolic diameter index, LV mass, and LV volume between participants with HFpEF or predisposed HFpEF. Compared with participants with HFpEF, those with predisposed HFpEF showed a lower left atrial volume index (52.0 mL/m² ± 23.1 vs 38.3 mL/m² ± 11.2; P < .001) (Fig 4).

Table 2:

Cardiac MRI Characteristics of Study Participants

Figure 4:

Graphs of distribution of cardiac MRI markers in heart failure with preserved ejection fraction (HFpEF), predisposed HFpEF, and control groups. ECV = extracellular volume, eGLSR = early-diastolic global longitudinal strain rate, GLS = global longitudinal strain, LAVi = left atrial volume index. * P < .05 versus predisposed HFpEF. ^† P < .05 versus control group.

Myocardial Strain and Tissue Characterization

Regarding subclinical cardiac dysfunction, both participants with HFpEF and predisposed HFpEF presented with abnormal myocardial strain versus the control group (all P < .05; Table 2). Compared with participants with HFpEF, those with predisposed HFpEF showed less impaired GLS (-12.2% ± 2.8 vs -13.9% ± 3.1; P < .001), systolic global longitudinal strain rate (-0.65/sec ± 0.15 vs -0.69/sec ± 0.15; P = .006), systolic global circumferential strain rate (-0.80/sec ± 0.20 vs -0.85/sec ± 0.19; P = .02), and eGLSR (0.52/sec ± 0.20 vs 0.57/sec ± 0.15; P = .03). There was no evidence of differences in other strain parameters between the HFpEF and predisposed HFpEF groups, despite the numerically incremental trend. In the subgroups of participants with HF without obesity, the same significant differences were found for most parameters as found in the overall HF cohort (Table S3).

Of the 464 total participants, 256 (55.2%) underwent contrast-enhanced cardiac MRI. LGE was positive in 45 of 147 (30.6%) participants with HFpEF and 27 of 99 (27.3%) participants with predisposed HFpEF. There was no evidence of a difference in LGE extent between the HFpEF and predisposed HFpEF groups. In addition, 50.7% (108 of 213) of participants with HFpEF, 47.7% (72 of 151) of participants with predisposed HFpEF, and 8.0% (eight of 100) of control participants completed T1 mapping examination. Participants with HFpEF showed no evidence of a difference in ECV fraction values compared with participants with predisposed HFpEF.

Clinical Outcomes in Participants with HF

A total of 335 participants (191 with HFpEF and 144 with predisposed HFpEF) were subsequently evaluated for clinical outcomes, of which 22 with HFpEF and seven with predisposed HFpEF were lost to follow-up. With the median follow-up of 771 days (IQR: 580–1138 days), there were 29 primary end points in participants with HFpEF and 20 primary end points in participants with predisposed HFpEF. No evidence of a difference was observed between the HFpEF and predisposed HFpEF groups in Kaplan-Meier analysis (P = .79) (Fig 5) or multivariable Cox analysis (Table S4). In participants with predisposed HFpEF, significant differences were observed in AF occurrence, hematocrit levels, hemoglobin levels, and eGLSR between those with and without end points (Table 3). Plasma volume levels, AF occurrence, hematocrit levels, hemoglobin levels, NT-proBNP levels, and eGLSR (per 0.2-per-second increase) were associated with primary outcomes in univariable Cox analysis (Table 4). However, only AF occurrence (hazard ratio [HR] = 3.90; P = .009), hemoglobin levels (HR = 0.94; P = .001), and eGLSR (per 0.2-per-second increase, HR = 0.28; P = .002) were independently associated with primary outcomes in multivariable Cox analysis. In the subgroups of participants with predisposed HFpEF without coronary artery disease, the majority of parameters, especially eGLSR, were still significantly associated with primary outcomes in multivariable Cox analysis (Table S5). eGLSR was also independently associated with all-cause death and HF hospitalization (Table S6).

Figure 5:

Graph shows Kaplan-Meier analysis of event-free survival for primary end points in participants with heart failure with preserved ejection fraction (HFpEF) and participants with predisposed HFpEF. Primary end points include all-cause death, heart failure hospitalization, or stroke.

Table 3:

Clinical and Cardiac MRI Characteristics of Participants with Predisposed HFpEF

Table 4:

Univariable and Multivariable Cox Proportional Regression for Prediction of Primary End Points in Participants with Predisposed HFpEF

Intraobserver and Interobserver Variabilities

Intraclass or interclass correlation coefficient and Bland-Altman plots of myocardial strains and ECV fraction parameters were performed in 20 randomly chosen participants. Measurements for all strain parameters and ECV fraction were reproducible at the intraobserver and interobserver level, with intraclass or interclass correlation coefficient greater than 0.80 for all parameters except systolic global longitudinal strain rate (Table S7). GLS and ECV fraction had an intraobserver variability of -0.05% ± 0.13 and -0.38% ± 0.12 and an interobserver variability of -0.01% ± 0.15 and -0.56% ± 0.18, respectively (Fig S1).

Discussion

This relatively large-scale prospective observational and prognostic study assessing cardiac MRI features and clinical outcomes in individuals with HFpEF demonstrated several important findings: (a) participants with predisposed HFpEF and normal NP levels accounted for a significant proportion of the overall HF cohort; (b) compared with the control group, participants with predisposed HFpEF showed higher volume load, higher systemic inflammation, and worse cardiac remodeling and function; (c) compared with participants with typical HFpEF, those with predisposed HFpEF, whether obese or not, were younger and showed lower left atrial volume index, GLS, and early-diastolic longitudinal strain rate, but no evidence of a difference in primary outcomes was observed between the two groups; and (d) in participants with predisposed HFpEF, AF occurrence, hemoglobin levels, and eGLSR were independently associated with death, HF hospitalization, or stroke, warranting special attention.

Although the reference standard for diagnosing HFpEF is cardiac catheterization, its main use is limited to the research setting due to invasiveness and radiation (2). To address this limitation, we introduced a new and clinical-based cohort of individuals with predisposed HFpEF and normal NP levels based on the noninvasive echocardiographic and natriuretic peptide score in the latest European Society of Cardiology guidelines (2,7). This predisposed HFpEF cohort satisfied the requirement for cardiac diastolic dysfunction and cardiac remodeling, reaching 4 points in HFA-PEFF score. We found that in the clinical setting, there was a higher proportion of individuals with predisposed HFpEF and normal NP across the HF spectrum (42%) as compared with the general proportion of individuals with catheterization-proven HFpEF (up to 20%) (2). The difference in diagnostic criteria and participant inclusion in our study, which was more accessible to implement in a clinical setting, may have accounted for this result. Consistent with previous studies (6), we found that the predisposed HFpEF group was older and had a lower proportion of women, higher body mass index, higher volume load, more obvious systemic inflammation, and systolic and diastolic dysfunction compared with the control group, which are common pathophysiologic features in participants with HFpEF (4).

To date, only one study systemically demonstrated normal NP levels in participants with HFpEF using cardiac catheterization and showed that participants with HFpEF and normal NP levels had relatively worse outcomes than participants with HFpEF (6). However, the myocardial strain and tissue features were not investigated, and the risks associated with catheterization limit the study’s clinical generalizability. In contrast, our study proposed a new clinical-based classification of HFpEF from 2021 European Society of Cardiology guidelines and considered the cardiac MRI features of this cohort. In comparison to the HFpEF group, participants with predisposed HFpEF were younger and had better renal function, a higher prevalence of hypertension and obesity, and a lower prevalence of AF. These results may partly explain the phenomenon of normal NP levels in participants with predisposed HFpEF (21). Regarding the structural and functional alterations, cardiac MRI has the advantage of demonstrating subclinical dysfunction and myocardial tissue characteristics through myocardial strain and ECV fraction (3,9). In our study, although the participants with predisposed HFpEF showed similar LV remodeling compared with participants with HFpEF, the results showed that those with predisposed HFpEF, whether obese or not, had less impaired left atrial volume index, GLS, systolic global longitudinal strain rate, systolic global circumferential strain rate, and eGLSR. Before traditional structural and functional parameters, these parameters reflected the relatively preserved left atrial and ventricular function in participants with predisposed HFpEF.

Despite less impaired, subtle LV dysfunction first elucidated in this study, our data showed no evidence of a difference in primary outcomes between participants with predisposed HFpEF and those with HFpEF. This result warrants special clinical attention for participants with predisposed HFpEF, notwithstanding normal NP levels. Further research is needed to determine which markers may be associated with the prognosis of participants with predisposed HFpEF. These participants may represent an independent HFpEF subtype with adverse prognosis, and we further evaluated their prognosis in this study. The AF occurrence, hemoglobin levels, and eGLSR (per 0.2-per-second increase) were independently associated with primary outcomes. In keeping with previous studies, AF occurrence was independently associated with adverse prognosis in participants with HFpEF (22,23), and this study extended and proved the prognostic role of AF occurrence in these participants. Furthermore, studies have shown that participants with HFpEF commonly had both diastolic and systolic dysfunction, and LV diastolic dysfunction plays a central role in the pathophysiology of HFpEF (24). Although GLS was the more sensitive marker to identify subtle impairments in systolic function, eGLSR was the better and earlier marker to directly demonstrate diastolic function. The eGLSR occurs during the period of early diastolic filling. Therefore, it is directly affected by ongoing LV relaxation and LV stiffness (25) and represents diastolic dysfunction before left atrial and LV strain parameters. Previous studies also showed that eGLSR was a sensitive and direct cardiac MRI marker to reflect LV diastolic function (26) and was associated with adverse prognosis in multiple cardiovascular disease (11,27). Our study further demonstrated that a 0.2-per-second increase of eGLSR was associated with a 72% decreased risk of primary end points in participants with predisposed HFpEF. Moreover, we evaluated the prognostic value of eGLSR in a subgroup of participants with predisposed HFpEF without coronary artery disease and found that the majority of parameters, especially eGLSR, were still independently associated with primary outcomes. The results of the present study support the hypothesis that predisposed HFpEF may relate to a separate subtype with less impaired subtle myocardial function while showing similar adverse prognosis, and myocardial strain rate—eGLSR—would be an objective marker to assist clinical intervention. Accordingly, large-scale investigations are warranted to validate this assumption. Future studies should give greater attention to participants with predisposed HFpEF, validate the clinical presentations and cardiac MRI features, compare the prognosis of catheterization-diagnosed and clinically defined participants with predisposed HFpEF, confirm the prognostic value of strain parameters in the large-scale study, and implement these results into the clinical practice.

There were several limitations to our study. First, the diagnosis of HFpEF was based on the latest European Society of Cardiology guidelines without catheterization (7). However, the invasiveness of catheterization limits its clinical generalizability, and we comprehensively considered the existing accurate and accessible diagnostic methods to diagnose HFpEF (28,29). Second, this was a cross-sectional study with short-term follow-up in a single center. This limitation—as well as the limited number of end points and a more distinct population of HFpEF or predisposed HFpEF with a much smaller body mass index, lower NP levels, and less frequent diuretic use, representing an obvious limitation (including admission rate bias)—allows us to only suggest, rather than to affirm, the association of cardiac MRI parameters with outcomes. A large-scale, multicenter prospective study is now being conducted to further validate the results in a much broader population and ascertain long-term prognostic value of these imaging parameters. Third, as a tertiary referral hospital, participants may receive preliminary treatment to alleviate the symptoms before visiting our hospital, so the normal NP levels in the predisposed HFpEF cohort may relate to their previous treatment to some extent. Fourth, the intraobserver and interobserver reproducibility intraclass or interclass correlation coefficient for eGLSR was slightly above 0.80 and was much lower than that for GLS. However, eGLSR still showed acceptable reproducibility in the clinical setting. With further advances in cardiac MRI, the clinical setting will witness the increased reproducibility and use of eGLSR. Finally, the limited sample size of control participants (n = 8) who completed T1 mapping examination may account for the observed lack of significant differences in ECV fraction values among HFpEF, predisposed HFpEF, and control groups.

In summary, this prospective study evaluating a clearly defined cohort of individuals with HFpEF showed that those with predisposed HFpEF and normal NP levels had higher plasma volume levels, relatively lower left atrial volume index, and less impaired GLS and eGLSR compared with those with HFpEF; however, both groups had similar prognosis, indicating that greater clinical attention should be given to participants with predisposed HFpEF. Additionally, AF occurrence and eGLSR were independently associated with an adverse prognosis in individuals with predisposed HFpEF, warranting further large-scale prognostic studies.