Can Exercise Training Enhance the Functional Capacity and Diastolic Function in Heart Failure With Preserved Ejection Fraction? A Systematic Review and Meta-analysis in the East Asian Population

Abstract

Background and Objectives

Recent research indicated that the physiological mechanisms in heart failure with preserved ejection fraction (HFpEF) patients across various ethnicities/races exhibited distinct findings. Published research predominantly derives from Western populations and lacks evidence concerning the effects of exercise training on East Asian individuals. This study aimed to examine the impact of exercise training interventions on peak oxygen consumption (VO₂) and diastolic function in the East Asian population with HFpEF.

Methods

Articles were retrieved utilizing Medical Subject Headings phrases across 3 databases. Our research, documented in the PROSPERO database, encompassed trials and prospective studies that investigated peak VO₂ and diastolic function in East Asian patients with HFpEF who participated in an exercise training program. The random-effects model was utilized to evaluate the aggregated effect, represented as a mean difference (MD). The leave-one-out strategy was utilized to exclude research to evaluate the robustness of the results.

Results

Six studies were examined; however, only 4 were appropriate for quantitative analysis. Exercise training for HFpEF patients successfully elevated the average peak VO₂ (MD, 1.99; 95% confidence interval [CI], 0.57, 3.40; p=0.006; I²=57%). However, there was no alteration in diastolic function (E/A [MD, −0.17; 95% CI, −0.57, 0.23; p=0.41; I²=89%] and E/e’ [MD, −0.92; 95% CI, −3.58, 1.75; p=0.50; I²=43%]) after exercise intervention.

Conclusions

Exercise training for a minimum of 12 weeks, irrespective of the exercise modality, is associated with improvements in exercise capacity among the selected East Asian patients, particularly in peak VO₂, while observing minimal influence on diastolic function.

Trial Registration

International Prospective Register of Systematic Reviews (PROSPERO) Identifier: CRD420251040914

Graphical Abstract

INTRODUCTION

The incidence of heart failure with preserved ejection fraction (HFpEF) among all heart failure (HF) types is on the rise.¹⁾ In the general population, almost 50% of individuals diagnosed with HF demonstrate intact ejection fraction (EF).¹⁾ The prevalence of HFpEF among HF cases varies from 44% to 72%, suggesting a possible cyclical rise in HFpEF occurrences in recent years.²⁾ HFpEF affects over 32 million people worldwide.²⁾ Moreover, patients with HFpEF are hospitalized approximately 1.4 times per year and demonstrate an annual death rate of over 15%.³⁾ A recent study by Edwards and O’Driscoll⁴⁾ demonstrated that exercise training markedly decreased morbidity and mortality rates by improving patients’ quality of life, as evaluated through the Minnesota Living with Heart Failure Questionnaire and Kansas City Cardiomyopathy Questionnaire.

Exercise training is fundamental in the treatment of HF. In the 2021 HF guidelines of the European Society of Cardiology, exercise is classified as a class 1A recommendation for all categories of HF.⁵⁾ Although there are no specific guidelines for exercise training in HFpEF, research repeatedly evidences its benefits. Multiple studies illustrate the benefits of exercise on functional adaptation, encompassing improved cardiac output (CO), greater peripheral vascular function, and modifications in skeletal muscle (increased oxidative muscle fibers and reduced muscular atrophy).⁶,⁷⁾ A key feature of HFpEF is reduced exercise tolerance, indicated by decreased peak oxygen consumption (VO₂) due to impaired cardiac, vascular, and skeletal muscle function. Consequently, 2 measures, namely peak VO₂ and diastolic function, are essential in characterizing the progression of HFpEF. A meta-analysis conducted by Baral et al.⁸⁾ demonstrated that exercise training, regardless of the intervention type, improves peak VO₂ and quality of life. Furthermore, multiple studies demonstrate that the ventilatory anaerobic threshold and the 6-minute walk test are improved.⁹,¹⁰⁾ Exercise-induced modifications in diastolic function have been thoroughly investigated, resulting in incongruous outcomes: certain investigations indicated no alterations in diastolic function, as assessed by E/e’ medial, e’ medial, and left atrial (LA) volume index; in contrast, aerobic interval training (AIT) reduced E/e’ and e’ in randomized controlled trials (RCTs) for HFpEF.¹¹,¹²⁾

Current guidelines derived from published research predominantly stem from Western populations and lack information about the effects of exercise training interventions on cardiac function, hemodynamic responses during exercise, and exercise capacity in East Asian patients with HFpEF. Research in Japan reveals that the median age of HFpEF patients is 80 years, with a mean body mass index of 23.9±4.7 kg/m², indicating a significantly distinct cardiovascular phenotype relative to Western populations.¹³⁾ Moreover, the study conducted by Fu et al.¹²⁾ exclusively within a Taiwanese cohort revealed that the mean age of the patients was 60.5 years, perhaps younger than the global average age of individuals with HFpEF. The differences in features require additional examination to determine the effects of exercise training on East Asian people with HFpEF. Furthermore, a meta-analysis evaluating the impact of exercise training on parameters such as peak VO₂ and diastolic function, shown by E/e’ and E/A ratios from echocardiography (the principal characteristics of HFpEF), is currently lacking for the East Asian population. This article seeks to examine the impact of exercise training interventions on peak VO₂ and diastolic function in the East Asian population with HFpEF.

METHODS

The authors registered this work with the International Prospective Register of Systematic Reviews (PROSPERO) under registration number CRD420251040914. Additionally, the methodologies employed in this investigation adhered to the standards set forth by the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) (Supplementary Table 1).

Search strategy

We performed an exhaustive automated search to identify all published controlled trials. Two authors conducted independent searches for articles on PubMed, ScienceDirect, and EuropePMC from September to November 2024. The authors employed diverse Medical Subject Headings and free-text terms to identify relevant articles via keywords utilizing Boolean operators, specifically: (aerobic exercise OR endurance exercise OR resistance training) AND (Asian people OR Asian) AND (Heart Failure Preserved Ejection Fraction OR HFpEF) AND (PeakVO₂ OR VO₂ max) AND (Diastolic Function OR E/A OR E/e’) in the PubMed database. In both ScienceDirect and EuropePMC, the authors employed the keyword combination: (aerobic exercise OR endurance exercise OR resistance training) AND (Asian people OR Asian) AND (Heart Failure Preserved Ejection Fraction OR HFpEF). The inclusion and exclusion criteria were defined before the search was conducted, adhering to the PICO (Table 1). There are no restrictions on language utilisation during the data retrieval procedure. Furthermore, the writers compiled a comprehensive collection of references derived from the specified articles. The findings were subsequently cleansed of duplicates and verified against the qualifying criteria.

Table 1. PICO framework.

PICO	Keywords
Patient	East Asian ≥18 years old with HFpEF
Intervention	Exercise training intervention
Comparison	Standard medical care only
Outcome	Peak VO2 and diastolic function (E/e’ and E/A)

PICO = Patient/Problem, Intervention, Comparison, and Outcome; HFpEF = heart failure with preserved ejection fraction; VO₂ = oxygen consumption.

Eligibility criteria

Two reviewers (Desandri DR, Qhabibi FR) independently reviewed and chose the papers, while the third reviewer (Papadakis M) conducted a thorough examination to ascertain eligibility. The research underwent a stringent evaluation process based on established inclusion and exclusion criteria. The inclusion criteria were: (a) publication within the last 10 years; (b) study design as a human-controlled trial; (c) East Asian patients aged over 18 years diagnosed with HFpEF; (d) intervention involving aerobic or endurance exercise, resistance training, high-intensity interval training (HIIT), inspiratory muscle training (IMT), or a combination thereof; (e) control group receiving only standard medical care; (f) assessment of at least peak VO₂ and diastolic function (E/A, E/e’); and (g) publication in English. The exclusion criteria included: (a) inaccessible papers; (b) review articles; (c) non-human clinical trials; and (d) research lacking reported specified outcomes.

Research was gathered by manual and snowball sampling methods to uncover other pertinent studies on the subject. The author autonomously conducted a search for literature and gathered all pertinent titles and abstracts utilizing the Rayyan intelligent systematic review platform. Studies will be omitted if: (a) duplicates could not be found, (b) they lacked adequate data regarding the outcomes.

Study selection

The 2 authors (Desandri DR and Qhabibi FR) independently evaluated the content of the papers based on the inclusion criteria. All issues were addressed through consensus among the authors. Only studies that met the specified criteria were included; studies that did not were excluded. The precise reasons for exclusion are defined in the PRISMA flow chart.

Risk of bias (RoB) assessment

This study employs 2 methods for RoB evaluation. The initial category pertains to articles employing RCT study designs, which will be evaluated utilizing the Modified Jadad Scale. For observational research, including both prospective and retrospective cohorts, the Newcastle-Ottawa Scale (NOS) will be employed. The Modified Jadad Scale ranges from 0 to 8.¹⁴⁾ In the RoB assessment with the Modified Jadad Scale, a study is classified as high quality if it scores above 4, medium/moderate quality if it scores between 3 and 4, and low quality if it scores below 3.¹⁵⁾ The assessment system utilizing the NOS distinguishes studies as high quality if they achieve 3 or 4 stars in the selection domain, a minimum of 2 stars in the outcome/exposure domain, and at least 1 star in the comparability domain.¹⁶⁾

Quality assessment

The evaluation of the certainty of evidence in this study was performed using the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) tool. The level of trust or certainty was classified as very low, low, moderate, or high. Data from RCTs and trials provide evidence of high certainty, whereas observational cohort data indicate moderate certainty. Factors such as publication bias, indirectness, imprecision, inconsistency, and RoB can reduce certainty. Conversely, the dose-response relationship and substantial effects can increase certainty. We employed GRADEpro GDT (https://gdt.gradepro.org/app/) to conduct the assessment.

Data extraction

All data were acquired using a designated reporting technique. The authors (Desandri DR, Qhabibi FR) conducted independent assessments of the titles and abstracts from the research, and any discrepancies were addressed through discussion. The chosen studies were organized in a Google Sheet and later evaluated for their suitability through a discussion process until consensus was reached. The data about the outcome variables were thoroughly assessed. The information extracted from eligible studies comprises (a) the first author’s name; (b) the year of publication; (c) the study design; (d) the country or location of the studies; (e) the number of samples along with their characteristics; (f) session time, frequency, intensity, and duration of the exercise intervention; (g) modalities of the control group; and (h) outcomes. In cases with incomplete data, the corresponding authors were contacted. If the researchers opt not to respond, the study will be omitted from the analysis.

Statistical analysis/quantitative assessment

This study conducted a quantitative analysis of secondary data obtained from various publications published in the last decade. The quantitative study employed RevMan 5.4 (Cochrane, Copenhagen, Denmark). The meta-analysis employed inverse variance (IV) to determine the mean difference (MD) for each outcome. The standard deviation for each endpoint further employed the IV methodology. We employed the Q test to assess data heterogeneity and quantify the extent of heterogeneity (I²). Heterogeneity was deemed minor if the I² value fell below 40%. Additionally, where heterogeneity was identified in pooled research, a random-effects model was employed to address variances in study design and the extensive range of potential treatment effects among the included studies. If no substantial heterogeneity was detected (<40%), a fixed-effects model was used. The Galbraith plot was employed to provide the precision, assess heterogeneity among the effect sizes, and detect the influential outlier studies. Studies identified as outliers through this graphical Galbraith plot were subsequently subjected to sensitivity analysis. This study employed the leave-one-out strategy in the meta-analysis sensitivity analysis to evaluate the statistical significance of each pooled endpoint study upon exclusion of an individual study. This approach enables us to pinpoint each individual study contributing to heterogeneity and evaluate the robustness of our results. We employed Stata/MP 17 for Mac (StataCorp, College Station, TX, USA) to perform the Galbraith plot and sensitivity analysis.

Publication bias

Publication bias was quantitatively assessed utilizing Egger’s regression test, due to the restricted number of studies, with p<0.05 being statistically significant. The Egger regression test was conducted utilizing Stata/MP 17 for Mac (StataCorp). The results of the Egger’s test demonstrated statistical significance (p<0.05) for the existence of publication bias, but the lack of publication bias was revealed otherwise.

RESULTS

Study selection

Figure 1 depicts the PRISMA research flowchart. The literature search approach first identified 381 titles and abstracts from available databases published in the last 10 years. The results were derived from 3 databases and one additional source uncovered by snowball sampling methods. A total of 30 articles were eliminated due to duplicate entries, while 224 articles were deleted based on inappropriate titles or abstracts. Out of 127 identified articles for retrieval, 57 were discarded due to the unavailability of the report and the inaccessibility of the complete article. After a comprehensive review, the authors chose 70 studies for subsequent eligibility screening. In the course of this screening process, 32 papers were excluded due to inappropriate study design, 19 articles for involving non-Asian populations, 12 articles for incorrect interventions, and 5 articles for irrelevant results. The authors incorporated 2 studies together with 4 studies identified using snowball sampling, culminating in a total of 6 studies, owing to insufficient data and/or the unavailability of further study. This study comprised 6 research: 3 RCTs, 1 non-RCT, and 2 prospective cohort studies. This study included all enrolled cases.

Figure 1. Diagram of study selection using PRISMA flowchart.

PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-analyses.

RoB and quality assessment

Table 2 presents the outcomes from the RoB assessment, illustrating the quality of the papers incorporated in this meta-analysis. Additionally, Figure 2 illustrates the specifics of each evaluation point utilizing the Modified Jadad Scale and the NOS in every study. One study¹⁷⁾ is classified as high quality, while 2 studies¹²,¹⁸⁾ are deemed intermediate quality based on the Modified Jadad Score. Nevertheless, 3 research¹⁹,²⁰,²¹⁾ assessed with the NOS exhibit high quality. Furthermore, as seen in Table 3, the GRADE assessment revealed moderate certainty for functional capacity (peak VO₂) and diastolic function (E/e’), while evidence for diastolic function (E/A) was adjudicated as low certainty. These determinations were influenced by the inclusion of a non-RCT study²¹⁾ within pooled analysis across all 3 outcomes. Furthermore, following a rigorous evaluation of statistical heterogeneity and the range of 95% confidence interval (CI), the evidence was downgraded for inconsistency aspect (peak VO₂ and E/A), and for imprecision aspect (peak VO₂ and E/e’).

Table 2. Characteristics of studies.

Study	Year	Study design	Country	Intervention sessions attended	Participants characteristics	Exercise training program	Control modalities	Outcomes	Study quality
Fu et al.12)	2016	RCT	Taiwan	100%	Total: 59	3-minute intervals of AIT at 40% and 80% peak VO2 for 30 minutes a day, 3 days per week, for 12 weeks	General health care	Systolic and diastolic function; peak VO2; VE/VCO2; HR; MLHFQ; SBP, DBP, MAP; quality of life (SF36)	Moderate
					Intervention: 30 (M: 20/F: 10), mean age: 60.5 years
					Control: 29 (M: 18/F: 11), mean age: 62.4 years
Kinugasa et al.18)	2020	RCT	Japan	100%	Total: 20	Home-based training (inspiratory muscle training for 20 minutes, using a threshold inspiratory muscle trainer), at a resistance equal to 30% of their maximal MIP	Usual care	Improvement of % MIP; peak VO2; anaerobic threshold; VE/VCO2 slope	Moderate
					Intervention: 8
					Control: 12, mean age: 76 years
Liu et al.17)	2023	RCT	China	Intervention group: 85%	Total: 60	Home-based exercise (filmed and given to the patients) for 12 weeks	Conventional HF drugs treatment	Peak VO2; 6MWT; blood pressure; quality of sleep (PSQI); quality of life (SF36)	High
				Control group: 100%	Intervention: 17 (M: 12/ F: 5), mean age: 64.94 years
					Control: 20 (M: 12/ F: 8), mean age: 67.95 years
Sugita et al.21)	2023	Non-RCT	Japan	100%	Total: 99	Supervised exercise training for 5 months (5 minutes of warm-up, 30 minutes of aerobic exercise, 20 minutes of resistance training, 5 minutes of cooldown)	General health care	Peak VO2; VE/VCO2 slope; peak CO index; arteriovenous oxygen difference; diastolic function	Good
					Intervention: 49 (M: 24/ F: 25), mean age: 73 years
					Control: 50 (M: 24/ F: 26), mean age: 74 years
Hsu et al.19)	2019	Prospective-controlled study	Taiwan	100%	Total: 202	Supervised exercise training on a bicycle ergometer (HIIT, alternative 3 minutes 80% peak VO2 and 3 minutes 40% peak VO2), total 36 sessions	Multidisciplinary disease management program	Peak VO2; peak HR; peak O2 pulse; quality of life (SF36); LV geometry; 5-year survival rate	Good
					Intervention: 29, mean age: 61.5 years
					Control: 30, mean age: 62.8 years
Hsu et al.20)	2024	Prospective-controlled study	Taiwan	100%	Total: 214	Supervised bicycle ergometer (HIIT, exercised alternatively 3 minutes 80% peak VO2 and 3 minutes 40% peak VO2) for 30 minutes per session, total 36 sessions	GDMT	10-year survival; peak VO2; LV geometry	Good
					Intervention: 14, mean age: 59.5 years
					Control: 26, mean age: 59.8 years

RCT = randomized controlled trial; M = male; F = female; AIT = aerobic interval training; VO₂ = oxygen consumption; VE = minute ventilation; HR = heart rate; MLHFQ = The Minnesota Living with Heart Failure Questionnaire; SBP = systolic blood pressure; DBP = diastolic blood pressure; MAP = mean arterial pressure; SF36 = The 36-Item Short Form Survey; MIP = maximum inspiratory muscle pressure; HF = heart failure; 6MWT = 6-minute walk test; PSQI = Pittsburgh Sleep Quality Index; CO = cardiac output; HIIT = high-intensity interval training; LV = left ventricular; GDMT = guideline-directed medical therapy.

Figure 2. Risk of bias assessment using the Modified Jadad Scale and the Newcastle-Ottawa Scale.

RCT = randomized controlled trial.

Table 3. GRADE assessment.

Certainty assessment								Number of patients		Effect		Certainty	Importance
Number of studies		Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Exercise + GDMT	GDMT only	Relative (95% CI)	Absolute (95% CI)
Functional capacity [Peak VO2] (follow-up: range 12 weeks to 20 weeks)
	4	Non-randomized studies	Not serious	Serious*	Not serious	Serious†	Strong association	105	111	-	MD 1.99 more	⊕⊕⊕◯	CRITICAL
							all plausible residual confounding would reduce the demonstrated effect dose response gradient				(0.57 more to 3.4 more)	Moderate*,†
Diastolic function [E/e’] (follow-up: range 12 weeks to 20 weeks)
	2	Non-randomized studies	Not serious	Not serious	Not serious	Serious†	All plausible residual confounding would reduce the demonstrated effect dose response gradient	80	79	-	MD 0.92 lower	⊕⊕⊕◯	IMPORTANT
											(3.58 lower to 1.75 higher)	Moderate†
Diastolic function [E/A] (follow-up: range 12 weeks to 20 weeks)
	2	Non-randomized studies	Not serious	Serious*	Not serious	Not serious	All plausible residual confounding would reduce the demonstrated effect	80	79	-	MD 0.17 lower	⊕⊕◯◯	IMPORTANT
											(0.57 lower to 0.23 higher)	Low*

GRADE = Grades of Recommendation, Assessment, Development, and Evaluation; GDMT = guideline-directed medical therapy; CI = confidence interval; VO₂ = oxygen consumption; MD = mean difference.

^*Presented a substantial heterogeneity within studies; ^†One of studies revealed a wide range of 95% CI (high variability).

Characteristics of included studies

Table 2 presents the characteristics of the considered study attributes. The study comprised 3 RCTs, 1 non-RCT, and 2 prospective controlled studies, involving a total of 654 participants. The authors performed a quantitative analysis of 4 of the 6 trials. The research included East Asian patients from Japan, Taiwan, and China diagnosed with HFpEF (left ventricular [LV] EF over 50%). The intervention group participated in at least 12 weeks of exercise training, which encompassed AIT, home-based training, IMT, maximal inspiratory muscle pressure, resistance training, HIIT, or various combinations of these exercises, in conjunction with standard pharmacological therapy for HFpEF. In contrast, all studies administered solely standard pharmacological treatment to the control group. The follow-up duration differs across research, ranging from around 12 weeks to 5 months.

Primary outcomes

Peak VO₂

Various outcomes were evaluated about the differences in peak VO₂ among HFpEF patients who underwent an exercise training intervention alongside standard preserved HF therapy, compared to a control group receiving solely standard medical treatment. Figure 3A illustrates the effect of peak VO₂ variation among groups from 4 studies¹²,¹⁷,¹⁸,²¹⁾ involving 216 participants, indicating that exercise in HFpEF patients significantly enhanced average peak VO₂ compared to those receiving only standard medication therapy (MD, 1.99; 95% CI, 0.57, 3.40; p=0.006; I²=57%). The exercise treatments exhibit considerable diversity; however, the intervention could not be tailored to a specific exercise modality or modality subgroup analysis due to a limited number of eligible studies for each exercise modality.

Figure 3. The effect of exercise training on HFpEF patients toward the peak VO₂ alteration. (A) Forest plot of pooled estimation. (B) Forest plot of leave-one-out sensitivity analysis. (C) The Galbraith plot of peak VO₂ alteration endpoint.

HFpEF = heart failure with preserved ejection fraction; VO₂ = oxygen consumption; GDMT = guideline-directed medical therapy; SD = standard deviation; IV = inverse variance; CI = confidence interval.

Our study reveals that the peak VO₂ endpoint is substantially influenced by 3 studies, as indicated by sensitivity analysis employing the leave-one-out method; excluding any of these studies results in an inconsequential outcome, with a p-value beyond 0.05. Despite the exclusion of Sugita et al.’s paper²¹⁾ from the analysis, the results remained significant, as illustrated in Figure 3B. This indicates that the study enhances the reliability of the peak VO₂ endpoint analysis. Furthermore, according to the Galbraith plot in Figure 3C, no studies that are outliers showed by the entire dots that represented the included studies lie within the 95% limits area.

Diastolic function (E/e’)

Some studies further evaluated the diastolic function attributes of HFpEF patients engaged in an exercise regimen versus those treated exclusively with medicines. The study’s analysis reveals no significant difference in the E/e’ ratio between participants in the exercise program and those in the control group (MD, −0.92; 95% CI, −3.58, 1.75; p=0.50; I²=43%), as illustrated in Figure 4A. Given that these secondary indices were assessed in only 2 studies (insufficient evidence), the current lack of statistical significance likely reflects a paucity of data rather than a lack of physiological effect. The sensitivity analysis regarding the E/e’ diastolic function endpoint indicated that one of the 2 studies significantly influenced the meta-analysis results, as illustrated in Figure 4B. Visual inspection of the Galbraith plot in Figure 4C, confirms that no individual study that are outliers showed by all constituent studies residing within the 95% limits area.

Figure 4. The impact of exercise training on diastolic function, as shown by the E/e’ ratio, in individuals with HFpEF. (A) Forest plot of pooled estimation. (B) Forest plot of leave-one-out sensitivity analysis. (C) The Galbraith plot of E/e’ ratio endpoint.

HFpEF = heart failure with preserved ejection fraction; GDMT = guideline-directed medical therapy; SD = standard deviation; IV = inverse variance; CI = confidence interval.

Diastolic function (E/A)

Two studies also evaluated another diastolic function measure using echocardiographic metrics. The E/A ratio demonstrates that exercise regimens for HFpEF patients do not markedly influence diastolic function in comparison to those undergoing standard pharmaceutical treatment (MD, −0.17; 95% CI, −0.57, 0.23; p=0.41; I²=89%), as illustrated in Figure 5A. Owing to the inclusion of only 2 trials reporting these secondary parameters, the results should be interpreted with caution; the non-significant findings are likely a function of inadequate statistical power rather than an absence of physiological effect.

Figure 5. The impact of exercise training on diastolic function, as shown by the E/A ratio, in individuals with HFpEF. (A) Forest plot of pooled estimation. (B) Forest plot of leave-one-out sensitivity analysis. (C) The Galbraith plot of E/A ratio endpoint.

HFpEF = heart failure with preserved ejection fraction; GDMT = guideline-directed medical therapy; SD = standard deviation; IV = inverse variance; CI = confidence interval.

The results of the E/A ratio endpoint analysis were substantiated by a leave-one-out sensitivity analysis (Figure 5B), which revealed that excluding any of the 2 studies significantly influenced the outcomes. The sensitivity analysis results confirmed the robustness of the conclusion that exercise training did not significantly influence the E/A ratio in the diastolic function of HFpEF patients. Moreover, according to the Galbraith plot in Figure 5C, no studies that are outliers showed by all constituent studies lie within the 95% limits area.

Publication bias

The examination of the Egger regression test for most outcomes indicated no significant publication bias, yielding trivial findings with p>0.05 for all outcomes. The assessment of publication bias in this meta-analysis was constrained by the limited number of eligible studies. While Egger’s regression test was performed to evaluate small-study effects, the results should be interpreted with caution. Funnel plot asymmetry was not assessed as the low study volume precludes reliable visual inference. Furthermore, statistical test for publication bias by employed an Egger’s test are frequently underpowered when fewer than 10 studies are included, potentially limiting the sensitivity of analysis.

DISCUSSION

The most enduring symptom in HFpEF patients is considerable exercise intolerance, resulting in diminished quality of life. Previous studies suggested diminished aerobic capacity, evidenced by reduced peak VO₂ in HFpEF patients compared to age- and gender-matched controls.¹¹⁾ The lower aerobic capacity observed in HFpEF may stem from impaired convective and diffusive oxygen transfer, namely diminished CO and arteriovenous oxygen difference. An exercise training program may be advantageous in restoring physiological function to improve aerobic capacity and elevate the quality of life in patients with HF. Nonetheless, numerous studies and publications revealed that the physiological mechanisms and treatment benefits among different ethnicities or races of HFpEF patients demonstrated varied results.¹³,²²,²³⁾ The disparity in baseline physiological characteristics across various ethnicities substantially influences treatment outcomes in exercise programs for HFpEF patients. Furthermore, the scientific gap shown in this paper indicates that most guidelines and clinical decision-making consensus are based on evidence from trials conducted in Western populations. Therefore, modifications are required in its use owing to differences in underlying physiology and homeostatic conditions among individuals of various ethnic backgrounds. This research will augment evidence-based medicine about exercise prescription for the East Asian population diagnosed with HFpEF receiving standard therapy.

This meta-analysis concerning the selected East Asian population produces 2 primary results. In individuals with HFpEF, exercise training enhances exercise capacity, evidenced by an increase in peak VO₂ (MD, 1.99; 95% CI, 0.57, 3.40; p=0.006; I²=57%). Secondly, there is no modification in diastolic function (E/A [MD, −0.17; 95% CI, −0.57, 0.23; p=0.41; I²=89%] and E/e’ [MD, −0.92; 95% CI, −3.58, 1.75; p=0.50; I²=43%]) following an intervention. The data suggest that exercise training may enhance functional capacity in HFpEF through mechanisms not only linked to LV function. Patients with HFpEF, while frequently asymptomatic at rest, exhibit diminished exercise tolerance and hence a lower VO₂ max. HFpEF patients exhibit elevated filling pressures even at rest. Nonetheless, when the illness advances chronically, physiological alterations occur, affecting both the hemodynamic physiology of HFpEF and the skeletal system. Abudiab et al.²⁴⁾ examined 109 participants with HFpEF and 73 controls, analyzing hemodynamic alterations via right cardiac catheterization post-exercise. Although exhibiting comparable EF, stroke volume, CO, and systolic blood pressure at rest, the HFpEF cohort demonstrates significantly elevated right and left heart-filling pressures, pulmonary artery pressure, pulmonary vascular resistance, and pulse pressure. Post-exercise, pulmonary capillary wedge pressure increased from 16 to 33 mmHg, in contrast to the control, which ranged from 9 to 14 mmHg, while the increase in CO was attenuated.²⁴⁾ HFpEF is also thought to have lower aortic and carotid artery distensibility, which are correlated with their level of exercise intolerance and peak VO₂ during exercise. The diminished workout blood flow reserve may result from compromised peripheral artery endothelial functioning. Fundamental abnormalities in skeletal muscle are regarded as a main cause of exercise intolerance. Leg lean mass is decreased and intermuscular fat is elevated in HFpEF. A pro-inflammatory response, insulin resistance, altered oxygen transport, and, on a microscopic level, a decreased fraction of type I skeletal muscle fibers are just a few of the ways that increased adipose tissue can impact muscle performance.

While exercise seems to improve LV systolic and diastolic function in heart failure with reduced ejection fraction (HFrEF) patients according to several research, this investigation did not demonstrate a comparable enhancement in echocardiographic diastolic parameters. This may be attributed to the limited sample size (just 2 studies examined diastolic function) or the particular study population. Elevated arterial stiffness and endothelial dysfunction have been recognized as contributors to exercise intolerance in HFpEF. Nevertheless, numerous studies revealed negligible enhancement in these metrics following exercise training.²⁵,²⁶⁾ Exercise may enhance exercise tolerance via peripheral mechanisms, leading to enhanced oxygen utilization by skeletal muscles. The phenomenon may be attributed to skeletal muscle's greater plasticity and its significant functional enhancements following even a short duration of activity, in contrast to cardiac muscle.

The properties of skeletal muscle are predominantly influenced by race or ethnicity, perhaps elucidating racial inequalities in physical performance. Other factors that may influence VO₂ max include hemoglobin concentration, physical activity, food, body composition, and socioeconomic status. In the example case, predicted equations for VO₂ max were based on the Caucasian population and overpredicted VO₂ max in the Asian population.²⁷⁾ Further, Tromp et al.²⁸⁾ delineated the characteristics of HFpEF in Asia. The majority of patients are comparatively younger, with 37% of the population being under 65 years old and possessing 2 or more comorbidities.²⁹⁾

Our meta-analysis findings for the selected East Asian population are consistent with other meta-analyses including larger cohorts, primarily comprising Caucasian persons. Baral et al.⁸⁾ reported that exercise correlated with an enhancement in peak VO₂ (MD, 1.96 mL/kg/min; 95% CI, 1.25, 2.68; p<0.00001) and a reduction in the Minnesota Living with Heart Failure score for quality of life (MD, −12.06; 95% CI, −17.11, −7.01; p<0.00001), but found no significant correlation with diastolic function (E/A). Numerous research have examined diastolic function; nonetheless, the conclusions remain inconsistent, despite many producing significant results. A meta-analysis by Edwards and O’Driscoll ⁴⁾ and Pearson et al.²⁹⁾ demonstrated a significant improvement in E/e’ (MD, −1.709; 95% CI, −2.91, 0.51; p=0.005 and MD, −2.38; 95% CI, −3.47, −1.28; p<0.0001, respectively), but our meta-analysis indicated no alterations. Conversely, a study conducted by Pandey et al.⁷⁾ observed no notable alterations in diastolic E/A (MD, 0.08; 95% CI, −0.01, 0.16). The most recognized parameter of diastolic dysfunction is E/e’; however, the data indicated that E/e’ only exhibited a modest correlation with invasive filling pressures and outcomes in HFpEF. At the population level, the E/A and E/e’ ratios are well-established predictors of all-cause mortality, cardiovascular death, and HF hospitalizations across diverse clinical cohorts, including patients with preserved (HFpEF) and reduced (HFrEF) EFs. Although E/A and E/e’ ratios can be very helpful clinically in diagnosing HF, predicting the risk of HF, and determining the prognosis, these tests are not perfect and should be examined within the context of the rest of the echocardiogram and clinical scenarios.²⁹⁾ The utility of E/e’ for estimating LV filling pressure remains a subject of ongoing debate; notably, it has demonstrated poor correlation with direct LA pressure measurements in patients with hypertrophic cardiomyopathy. Furthermore, in the setting of atrial fibrillation, beat-to-beat variability and the absence of the ‘A’ wave significantly complicate the assessment of diastolic function. Given these inherent limitations, E/A and E/e’ ratios should not be analyzed in isolation. Despite these challenges, diastolic parameters—particularly E/e’ and E/A—continue to serve as common surrogate endpoints in clinical trial studies.²⁸,²⁹⁾ Diastolic function in this situation necessitates supplementary examinations, prolonged follow-ups, and additional investigations. These data emphasize the robust correlation between exercise training and enhanced exercise capacity, regardless of ethnicity. Although it does not substantially impact diastolic function, pharmacological therapy should be the primary focus in the treatment regimen for HFpEF.

This study’s limitations include, firstly, that the populations analyzed do not solely represent East Asians, as the research is restricted to only 3 countries: Japan, Taiwan, and China. It must be acknowledged that the present study focuses on a selected East Asian cohort rather than general population of the region. As such the findings may not be fully generalizable across the diverse demographic and clinical landscapes of East Asia. This limitation has been noted, and our conclusion has been narrowed to accurately reflect the specific cohort analyzed. Secondly, most studies were performed on restricted trial populations and were unblinded owing to the nature of the exercise intervention. Thirdly, not all studies evaluated every outcome, hindering our ability to do thorough meta-analyses encompassing all data. The heterogeneity of each meta-analysis is considerable, reflecting a high variety among the investigations. Ultimately, exercise treatments exhibit considerable diversity. The intervention could not be tailored to a specific exercise modality or modality subgroup analyses due to a limited number of eligible studies. Consequently, additional extensive studies with more diverse Asian populations from other countries and locations are necessary to yield more thorough and representative information. Multi-center research directly comparing various ethnicities in Asia may also be explored. Furthermore, research incorporating a broader evaluation of diastolic function characteristics is necessary.

In the selected East Asian cohort, engaging in HFpEF standard medical therapy along with exercise training for a minimum of 12 weeks, irrespective of the exercise modality, is associated with improvements in exercise capacity, as evidenced by increased peak VO₂. Conversely, no significant differences were observed in diastolic function parameters compared to medical therapy solely. Given that these secondary indices were assessed in only 2 studies (insufficient evidence), the current lack of statistical significance likely reflects a paucity of data rather than a lack of physiological effect, highlighting a critical gap for future longitudinal research.

SUPPLEMENTARY MATERIAL

Supplementary Table 1

PRISMA checklist