Right Heart Cardiac Magnetic Resonance Profiles and Cardiopulmonary Reserve in Older Adults Undergoing Long‐Term Multicomponent Exercise: A 2‐Year Longitudinal Study

ABSTRACT

Right heart geometry and function are essential for preserving physical performance in healthy aging, yet evidence regarding how long‐term supervised multicomponent exercise influences right heart characteristics and cardiopulmonary reserve over time remains limited. To address this gap, this interim analysis of a prospective, non‐randomized, two‐armed cohort study included 155 adults aged ≥ 65 years assigned to supervised multicomponent training (n = 75) or education control group (n = 80). Cardiac magnetic resonance imaging and cardiopulmonary exercise testing were performed at baseline, 1‐year, and 2‐year follow‐ups, with right heart structure and function as the primary outcomes; post hoc analyses were stratified by training adherence. Among these participants, 92 (training: n = 54; control: n = 38) completed the 2‐year follow‐up and were included in the final analysis. No significant time × group interactions were detected for right or left heart parameters (all p > 0.05). After multivariable adjustment, the exercise group showed higher overall adjusted mean values for right atrial volumes and selected right ventricular functional measures, including stroke volume, cardiac index, and peak filling rate (all p < 0.05). In contrast, significant time × group interactions were identified for peak VO₂ (p = 0.003), and VO₂ at AT (p = 0.033), indicating distinct longitudinal cardiopulmonary trajectories between groups. Post hoc subgroup analyses showed that between‐group differences in right heart volumetric measures persisted through 2 years in the high‐adherence group, whereas those in the low‐adherence group were less consistent. Overall, long‐term multicomponent exercise was associated with higher levels of selected right heart volumetric and functional measures without detectable alterations in longitudinal cardiac remodeling trajectories in older adults, alongside favorable longitudinal changes in cardiopulmonary reserve. These findings suggest that exercise‐related benefits in healthy aging may be characterized primarily by functional adaptations rather than longitudinal structural cardiac remodeling.

Trial Registration: The original study was registered at https://www.chictr.org.cn/bin/userProject (ChiCTR2000032404, Registered on April 27, 2020)

Abbreviations

BSA

body surface area

CI

confidence interval

CMR

cardiac magnetic resonance

CPET

cardiopulmonary exercise testing

EAT

epicardial adipose tissue

LA

left atrial

LAEF

left atrial ejection fraction

LAVED

left atrial end‐diastolic volume

LAVES

left atrial end‐systolic volume

LV

left ventricular

LVCI

left ventricular cardiac index

LVEDV

left ventricular end‐diastolic volume

LVEF

left ventricular ejection fraction

LVESV

left ventricular end‐systolic volume

LVSV

left ventricular stroke volume

Peak VO₂

peak oxygen uptake

RA

right atrial

RAEF

right atrial ejection fraction

RAVED

right atrial volume at ventricular end‐diastolic

RAVES

right atrial volume at ventricular end‐systolic

RV

right ventricular

RVCI

right ventricular cardiac index

RVEDV

right ventricular end‐diastolic volume

RVSV

right ventricular stroke volume

SSFP

steady‐state free precession

VO₂ at AT

oxygen uptake at anaerobic threshold

1. Introduction

Cardiac aging not only increases the risk of cardiovascular disease [1, 2], but also significantly impairs exercise capacity and quality of life in older adults. With aging, the heart undergoes structural and functional changes, including changes in ventricular structure, chamber size, myocardial composition, and systolic and diastolic function [3, 4]. While these age‐related adaptations have been extensively characterized in the left ventricle, the right ventricle, a key determinant of cardiopulmonary function and exercise tolerance, has received comparatively limited attention. This gap is partly attributable to the right ventricle's complex geometry and the technical challenges associated with its accurate functional assessment. Recent advances in cardiac magnetic resonance (CMR) imaging and post‐processing techniques now allow precise and reproducible evaluation of right heart structure and function.

Regular exercise is known to induce adaptive changes in cardiac structure and function [5, 6]. Multicomponent exercise training, comprising resistance, balance, flexibility, and aerobic exercises, is widely recommended for promoting cardiovascular health in older adults [7]. Although prior studies have demonstrated short‐term improvements in cardiopulmonary fitness and physical performance following exercise interventions [8, 9, 10, 11], the specific long‐term structural adaptations of the right heart to multicomponent exercise in aging populations have received comparatively less research attention. Importantly, few studies have employed quantitative imaging to characterize longitudinal structural and functional adaptations of the right heart. Our previous work demonstrated that 1 year of supervised exercise training was associated with greater improvements in right heart structure and function, suggesting that the right ventricle may be particularly responsive to exercise‐based interventions in older adults [12].

Building on these findings, we conducted a 2‐year follow‐up of a supervised multicomponent exercise program in older adults, using CMR to evaluate resting right heart characteristics and their longitudinal trajectories, and to assess the impact of exercise adherence. These longitudinal imaging findings might contribute to addressing a potential gap in understanding right heart changes during healthy aging.

2. Methods

2.1. Study Design and Participants

This study represented an interim analysis of an ongoing, prospective, two‐armed, health education‐controlled cohort study (registration number: ChiCTR2000032404). The primary objective of the parent cohort was to evaluate the impact of an individualized, exercise‐based health maintenance program on clinical adverse events (mobility disability, hospitalization, and mortality) in older adults.

Participants were recruited from community‐dwelling adults aged ≥ 65 years. Inclusion criteria required participants to be non‐disabled, capable of independently completing a 400‐m walk within 15 min without assistive devices, and willing to participate in long‐term exercise training and annual follow‐up assessments. Key exclusion criteria were: (1) acute cardiovascular conditions (e.g., myocardial infarction, heart failure, or uncontrolled arrhythmias); (2) presence of implanted pacemakers or defibrillators; (3) conditions interfering with assessments, such as edema (affecting body composition measurement) or asthma; and (4) cognitive impairment, neurological diseases, or other conditions preventing cooperation with exercise and evaluation protocols.

Between February 2021 and December 2022, participants were recruited from the outpatient department of Fuzhou University Affiliated Provincial Hospital. Eligible participants were enrolled after screening and completion of baseline assessments. All participants provided written informed consent. The study was approved by the Ethics Committee of Fuzhou University Affiliated Provincial Hospital and conducted in accordance with the Declaration of Helsinki.

Participants were allocated to the intervention or control group through a shared decision‐making process involving physicians and participants. Allocation was based on participant preferences, lifestyle considerations, and the anticipated sustainability of the long‐term health maintenance plan. The intervention group received center‐based supervised multicomponent exercise in addition to health education, whereas the control group received health education alone. To ensure intervention fidelity, participants in the exercise group who missed four consecutive supervised sessions were excluded from the primary analysis according to the study protocol.

The study population comprised well‐functioning older adults capable of independently completing structured exercise training. The intervention was therefore designed as a preventive strategy to maintain or improve cardiorespiratory fitness and physical function, rather than a rehabilitative program targeting functionally impaired populations.

Post hoc subgroup analyses were performed to assess the dose–response relationship between exercise adherence and outcomes. Training adherence was defined as the percentage of completed sessions (prescribed as one supervised center‐based session and at least two home‐based sessions per week) over the intervention period. Participants in the exercise group were categorized into high adherence (≥ 75% of prescribed sessions) and low adherence (< 75%) subgroups for further comparison.

2.2. Exercise Training Group

Participants in the training group received individualized exercise prescriptions developed by a multidisciplinary team of geriatricians and rehabilitation specialists. The intervention followed a hybrid model consisting of one weekly supervised center‐based session and at least two home‐based sessions. During the initial four supervised sessions, participants were instructed in the complete training protocol and provided with exercise materials and equipment.

The exercise program included aerobic, resistance, flexibility, and balance components. Aerobic training intensity was prescribed at the anaerobic threshold determined by baseline cardiopulmonary exercise testing. Exercise duration was progressively increased from 15 to 20 min to a target of 35 min, excluding a 5–10‐min warm‐up and cool‐down period. Modalities included treadmill exercise or stationary cycling, with continuous monitoring of electrocardiogram, blood pressure, and heart rate.

Resistance training was performed 2–3 times per week and targeted major muscle groups. Initial intensity was set at 30%–40% of one‐repetition maximum (1‐RM) for upper body exercises and 50%–60% for lower body exercises. The protocol followed a progressive overload approach, increasing repetitions (8–12 per set) before incremental increases in resistance using resistance bands, dumbbells, or sandbags.

Flexibility and balance training were performed 2–3 times weekly. Static stretching of major muscle groups was held for 15–60 s, and balance exercises included tandem walking, single‐leg standing, and squatting.

Home‐based exercise consisted of brisk walking or jogging combined with resistance training using free weights or resistance bands. Exercise intensity was self‐monitored using the Borg Rating of Perceived Exertion scale (6–20), with a target intensity of 12–16. Adherence was monitored through supervised attendance records and detailed home‐exercise logs. Further details of the training protocol have been previously validated and described [12].

2.3. Health Education Control Group

Participants in the control group attended weekly educational seminars covering exercise principles and general health‐related topics. After the initial four sessions, participants continued with health education alone. They were permitted to engage in physical activity according to their usual lifestyle but were not provided with structured exercise programs, supervised training, or specific prescriptions regarding frequency, intensity, or duration.

2.4. Outcome Measures

All participants underwent assessments at baseline, 1‐year, and 2‐year follow‐up intervals, including standardized questionnaires, resting Cardiac Magnetic Resonance (CMR), and Cardiopulmonary Exercise Testing (CPET).

2.5. General Information

Trained investigators conducted face‐to‐face interviews to collect demographic and sociological data, including age, marital status, and place of residence. Medical history (e.g., coronary artery disease, hypertension, Type 2 diabetes, other major diseases, and history of falls), smoking status, and medication use were also recorded. Physical fitness and functional indicators were additionally assessed, including height, weight, body mass index, blood pressure, heart rate, 6‐min walk distance, 5‐times sit‐to‐stand test, and 4‐m gait speed. All measurements were obtained using standardized protocols.

2.6. Laboratory Tests

After an 8‐h overnight fast, venous blood samples and morning urine samples were collected. Blood samples were analyzed for triglycerides, total cholesterol, high‐density lipoprotein cholesterol, low‐density lipoprotein cholesterol, apolipoprotein A, apolipoprotein B, fasting blood glucose, uric acid, creatinine, and glycated hemoglobin (HbA1c). Urine samples were analyzed for creatinine and microalbumin, and the urinary albumin‐to‐creatinine ratio was calculated.

2.7. CMR

CMR was performed using a 3.0 T MRI scanner (Prisma, Siemens Healthcare, Erlangen, Germany) in the supine position. Cine images were acquired using balanced steady‐state free precession (SSFP) sequences during end‐expiratory breath‐holding. Short‐axis stacks and long‐axis (four‐chamber and two‐chamber) views were obtained to assess cardiac structure and function. T1 and T2 mapping were performed using modified Look‐Locker inversion recovery sequence and T2‐prepared SSFP at the basal, mid, and apical short‐axis levels.

An experienced radiologist (18 years of experience), blinded to clinical data, performed image analysis using CVI42 software (Circle Cardiovascular Imaging, Calgary, Canada). Ventricular mass and bi‐plane atrial/ventricular volumes were measured using semi‐automatic algorithms with manual contour correction. Myocardial T1 and T2 values were obtained by tracing regions of interest with 10% erosion at the endocardial and epicardial borders to reduce partial volume effects. Epicardial adipose tissue (EAT) volume was calculated by manually tracing the region between the myocardium and pericardium across the short‐axis stack. Further methodological details were provided in previous research [13].

2.8. CPET

CPET was performed using a cycle ergometer (Schiller, Switzerland) with a continuous ramp protocol to symptom limitation. All tests were conducted between 2 and 5 p.m. Pre‐test equipment calibration included temperature, humidity, volume, and gas calibration. Heart rate, blood pressure, and ECG were measured in both supine and sitting positions. CPET parameters obtained included peak oxygen uptake (peak VO₂) and oxygen uptake at anaerobic threshold (VO₂ at AT).

Detailed protocol and procedures have been described previously [12].

2.9. Outcomes for Current Interim Analysis

2.9.1. Primary Outcome

The primary outcome was resting right heart parameters obtained through CMR, including right atrial (RA) and right ventricular (RV) structural and functional parameters (e.g., volumes, stroke volumes and ejection fraction).

2.9.2. Secondary Outcomes

(1) Left heart morphology and function (e.g., left ventricular (LV)/left atrial (LA) volumes, LV mass, and peak filling rate); (2) Myocardial tissue characterization (T1/T2 mapping and EAT volume); and (3) CPET‐derived exercise capacity metrics.

2.10. Statistical Analysis

All data were analyzed using IBM SPSS (version 27.0). A two‐sided p value of < 0.05 was considered statistically significant. Continuous variables were presented as mean ± standard deviation, and categorical variables were presented as counts and percentages. Baseline continuous variables were compared among groups using one‐way analysis of variance (ANOVA), while categorical variables were compared using the chi‐square test or Fisher's exact test, as appropriate.

For primary analyses, missing baseline covariates were managed using mean imputation. To assess the robustness of this approach, we performed a multiple imputation sensitivity analysis using the Fully Conditional Specification method. Specifically, 20 imputed datasets were generated with 50 iterations, and convergence was confirmed by visual inspection of trace plots. The imputation model included all baseline covariates, group assignment, and outcome variables. Continuous and categorical variables were imputed using linear and logistic regression, respectively. Pooled estimates from the imputed datasets were derived using Rubin's rules. As a sensitivity analysis, baseline characteristics and between‐group comparisons were re‐evaluated using the multiply imputed datasets. Cumulative incidence of clinical adverse events based on the baseline cohort was estimated using the Kaplan–Meier method and compared via the log‐rank test. Cox proportional hazards regression models were constructed to evaluate the risk of adverse events associated with exercise training, reporting hazard ratios (HRs) and 95% confidence intervals (CIs) in both unadjusted models and models adjusted for age and sex.

Repeated‐measures outcomes over the 2‐year follow‐up were analyzed using ANCOVA‐style linear mixed‐effects models (LMM) with restricted maximum likelihood estimation. To evaluate the per‐protocol intervention effect under full adherence, a complete‐case framework was implemented including participants who completed the full training and follow‐up period. Specifically, the dependent variable comprised post‐baseline measurements at each follow‐up time point (T1, T2), and the LMM included the follow‐up time points, group allocation, and their interaction as fixed effects. To adjust for baseline status, the baseline measurement (T0) was incorporated into the model as a time‐invariant covariate, alongside prespecified baseline clinical confounders (including age, sex, coronary artery disease, hypertension, Type 2 diabetes, lipid‐lowering drugs, body mass index, 4‐m gait speed, 5‐times sit‐to‐stand test, low‐density lipoprotein cholesterol, urinary albumin‐to‐creatinine ratio, peak VO₂, and individual baseline values). A random intercept for each participant was included to account for within‐subject correlations. Model selection was guided by fit indices (Akaike/Bayesian Information Criteria) and covariance structure comparisons, identifying the unstructured covariance matrix as the optimal structure. When significant group main effects or time × group interactions were identified, post hoc pairwise comparisons between groups at each follow‐up time point were conducted based on model‐derived fixed‐effect parameter estimates.

3. Results

3.1. Clinical Safety and Baseline Characteristics

During the 2‐year follow‐up, composite clinical adverse events (including mobility disability, hospitalization, malignant arrhythmias, and mortality) occurred in 26.7% of the exercise group and 18.8% of controls (p = 0.239; Table S1), with all events adjudicated as intervention‐unrelated. Neither Kaplan–Meier survival analysis (log‐rank p = 0.558; Figure S1) nor Cox proportional hazards regression (adjusted HR = 1.280, 95% CI: 0.640–2.559, p = 0.485) demonstrated an increased clinical risk associated with exercise training.

Ultimately, 92 participants completed the 2‐year study and were included in the interim analysis (mean age 69.8 years, 44 males; Figure 1, Table 1). Baseline clinical and quantitative resting imaging metrics were balanced between groups, except for BMI and lipid‐lowering medication usage (p = 0.011 and p = 0.043; Tables 1 and 2). Sensitivity analyses using multiple imputation for missing baseline covariates yielded consistent descriptive and comparative results (Table S2). Baseline characteristics and CMR parameters of the entire cohort (n = 155) are provided in Tables S3 and S4. Comparisons between completers and those lost to follow‐up revealed minimal attrition bias, limited to BMI and right atrial ejection fraction in the training arm (Tables S5 and S6).

FIGURE 1.

Research flow chart.

TABLE 1.

Characteristics of the study participants at baseline.

Characteristics	Total (n = 92)	Training group (n = 54)	Control group (n = 38)	p
Age, year	69.79 ± 3.78	69.11 ± 2.67	70.76 ± 4.81	0.060
Systolic BP, mmHg	129.20 ± 18.01	128.33 ± 18.30	130.42 ± 17.76	0.587
Diastolic BP, mmHg	75.01 ± 10.84	74.80 ± 11.71	75.32 ± 9.61	0.822
Heart rate, bpm	71.00 ± 8.07	70.80 ± 8.30	71.29 ± 7.82	0.775
BMI, kg/m2	23.22 ± 2.56	22.66 ± 2.76	24.02 ± 2.01	0.011
6MWT, m	468.90 ± 45.30	476.56 ± 37.11	458.03 ± 53.55	0.070
4‐m gait speed, m/s	1.43 ± 0.24	1.46 ± 0.24	1.39 ± 0.24	0.141
5STS, s	8.17 ± 1.53	7.95 ± 1.45	8.48 ± 1.61	0.109
Max grip, kg	27.45 ± 9.11	28.41 ± 9.83	26.08 ± 7.90	0.213
Triglyceride, mmol/L	1.48 ± 0.78	1.40 ± 0.68	1.60 ± 0.89	0.239
Cholesterol, mmol/L	4.80 ± 1.10	4.79 ± 1.13	4.82 ± 1.05	0.896
LDL‐C, mmol/L	3.06 ± 0.83	3.06 ± 0.88	3.07 ± 0.76	0.944
HDL‐C, mmol/L	1.36 ± 0.36	1.39 ± 0.35	1.31 ± 0.36	0.271
ApoA, g/L	1.47 ± 0.27	1.50 ± 0.27	1.44 ± 0.27	0.271
ApoB, g/L	0.89 ± 0.23	0.87 ± 0.23	0.93 ± 0.23	0.279
Fasting plasma glucose, mmol/L	5.89 ± 0.86	5.78 ± 0.81	6.05 ± 0.91	0.134
Serum uric acid, μmol/L	364.83 ± 82.45	366.63 ± 89.60	362.26 ± 72.17	0.804
Serum creatinine, μmol/L	75.60 ± 18.34	77.94 ± 20.07	72.26 ± 15.18	0.144
Glycated hemoglobin, %	6.18 ± 0.73	6.08 ± 0.57	6.34 ± 0.90	0.118
UACR, mg/g	27.26 ± 57.65	26.98 ± 59.93	27.66 ± 55.05	0.956
Peak VO2, mL/kg/min	22.97 ± 4.64	23.41 ± 5.02	22.05 ± 4.08	0.172
Absolute peak VO2, L/min	1.40 ± 0.37	1.43 ± 0.38	1.36 ± 0.35	0.397
VO2 at AT, mL/kg/min	13.17 ± 2.32	13.22 ± 2.36	13.10 ± 2.29	0.812
Absolute VO2 at AT, L/min	0.81 ± 0.19	0.80 ± 0.18	0.81 ± 0.20	0.805
Male	44 (47.8)	28 (51.9)	16 (42.1)	0.357
Smoking	12 (13.0)	7 (13.0)	5 (13.2)	> 0.999
Coronary artery disease	5 (5.4)	4 (7.4)	1 (2.6)	0.400
Hypertension	47 (51.1)	23 (42.6)	24 (63.2)	0.052
Type 2 diabetes	33 (35.9)	17 (31.5)	16 (42.1)	0.296
β‐Adrenergic receptor blocker	20 (21.7)	9 (16.7)	11 (28.9)	0.160
ACEI/ARB	29 (31.5)	15 (27.8)	14 (36.8)	0.357
SGLT2i	2 (2.2)	2 (3.8)	0 (0.0)	0.507
Diuretic	6 (6.5)	2 (3.7)	4 (10.5)	0.226
Lipid‐lowering drugs	43 (46.7)	30 (55.6)	13 (34.2)	0.043

Note: Data are presented as mean ± standard deviation or N (%). Bold type indicates a p value < 0.05. p values were calculated by independent‐samples t‐test or chi‐square test or Fisher's exact test (two‐tailed).

Abbreviations: ACEI/ARB, Angiotensin‐converting enzyme inhibitor/Angiotensin II receptor blocker; ApoA, Apolipoprotein A; ApoB, Apolipoprotein B; BP, blood pressure; BMI, body mass index; 5STS, 5‐times sit‐to‐stand test; HDL‐C, high density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; Peak VO₂, peak oxygen uptake; 6MWT, 6‐min walk test; SGLT2i, sodium‐glucose cotransporter 2 inhibitors; UACR, urinary albumin‐to‐creatinine ratio; VO₂ at AT, oxygen uptake at anaerobic threshold.

TABLE 2.

Baseline cardiac magnetic resonance parameters of the study participants.

Characteristics	Total (n = 92)	Training group (n = 54)	Control group (n = 38)	p
RAVED, mL	32.11 ± 14.40	32.59 ± 14.03	31.43 ± 15.06	0.706
RAVED/BSA, mL/m2	19.13 ± 7.95	19.51 ± 7.91	18.59 ± 8.07	0.586
RAVES, mL	49.78 ± 19.80	48.69 ± 19.12	51.32 ± 20.88	0.533
RAVES/BSA, mL/m2	29.81 ± 11.38	29.33 ± 11.41	30.50 ± 11.46	0.629
RVEDV, mL	110.87 ± 25.52	110.92 ± 24.31	110.78 ± 27.49	0.979
RVEDV/BSA, mL/m2	66.34 ± 12.97	66.43 ± 12.23	66.21 ± 14.12	0.938
RVESV, mL	44.59 ± 13.83	44.67 ± 13.29	44.47 ± 14.75	0.944
RVESV/BSA, mL/m2	26.55 ± 7.07	26.66 ± 6.88	26.40 ± 7.42	0.863
RVSV, mL	66.28 ± 16.67	66.25 ± 15.10	66.31 ± 18.89	0.985
RVSV/BSA, mL/m2	39.79 ± 9.26	39.77 ± 8.12	39.81 ± 10.79	0.982
RVEF, %	59.88 ± 7.49	59.98 ± 6.30	59.73 ± 9.01	0.875
RVCI, L/min/m2	2.75 ± 0.72	2.70 ± 0.64	2.81 ± 0.83	0.487
RAEF, %	50.00 ± 9.59	48.80 ± 10.07	51.71 ± 8.71	0.153
RV peak ejection rate, mL/s	397.93 ± 111.40	394.90 ± 119.11	402.23 ± 100.83	0.758
RV peak filling rate, mL/s	310.09 ± 120.18	308.53 ± 117.28	312.30 ± 125.73	0.883
LAVED, mL	23.46 ± 8.87	22.21 ± 9.65	25.24 ± 7.38	0.107
LAVED/BSA, mL/m2	14.14 ± 5.26	13.39 ± 5.63	15.21 ± 4.54	0.102
LAVES, mL	49.77 ± 14.74	48.12 ± 14.73	52.10 ± 14.62	0.204
LAVES/BSA, mL/m2	30.03 ± 8.98	29.03 ± 8.87	31.45 ± 9.06	0.205
LVEDV, mL	106.78 ± 22.57	105.95 ± 20.54	107.95 ± 25.42	0.677
LVEDV/BSA, mL/m2	64.01 ± 11.63	63.58 ± 10.77	64.62 ± 12.89	0.673
LVESV, mL	38.82 ± 10.86	37.95 ± 8.02	40.04 ± 13.98	0.411
LVESV/BSA, mL/m2	23.24 ± 5.68	22.84 ± 4.43	23.81 ± 7.11	0.458
LVSV, mL	67.96 ± 15.35	67.99 ± 15.20	67.91 ± 15.76	0.981
LVSV/BSA, mL/m2	40.77 ± 8.35	40.74 ± 8.07	40.82 ± 8.86	0.967
LVEF, %	63.68 ± 6.04	63.95 ± 5.09	63.29 ± 7.24	0.626
LVCI, L/min/m2	2.82 ± 0.70	2.77 ± 0.67	2.90 ± 0.76	0.423
LAEF, %	58.49 ± 7.80	58.87 ± 8.17	57.95 ± 7.32	0.579
LV peak ejection rate, mL/s	395.37 ± 115.75	388.60 ± 111.44	404.98 ± 122.48	0.507
LV peak filling rate, mL/s	306.08 ± 103.33	308.04 ± 105.26	303.30 ± 101.84	0.830
Peak diastolic LV wall thickness, mm	10.18 ± 1.78	9.90 ± 1.69	10.57 ± 1.85	0.073
Peak LV wall thickness, mm	14.18 ± 2.07	13.93 ± 1.92	14.54 ± 2.25	0.164
MAPSE Lateral, mm	14.61 ± 2.74	14.78 ± 2.62	14.37 ± 2.92	0.488
MAPSE Septal, mm	11.13 ± 3.66	11.45 ± 3.74	10.67 ± 3.56	0.322
TAPSE, mm	18.12 ± 5.40	17.86 ± 5.20	18.49 ± 5.72	0.585
Peak radial strain, %	40.20 ± 30.18	42.21 ± 38.58	37.34 ± 9.74	0.449
Peak circumferential strain, %	−20.57 ± 2.19	−20.59 ± 1.84	−20.54 ± 2.63	0.931
Peak longitudinal strain, %	−17.11 ± 4.28	−16.68 ± 5.22	−17.72 ± 2.35	0.254
Native T1 a , ms	1218.21 ± 70.28	1221.06 ± 66.52	1214.15 ± 76.03	0.645
Native T2 a , ms	42.54 ± 5.04	42.74 ± 4.76	42.26 ± 5.46	0.651
Epicardial fat volume, mL	147.61 ± 42.55	143.30 ± 45.02	153.73 ± 38.52	0.249

Note: Data are presented as mean ± standard deviation. Peak diastolic LV wall thickness: maximum LV myocardial wall thickness across all LV slices during diastole. Peak LV wall thickness: maximum LV myocardial wall thickness across all LV slices and cardiac phases.

Abbreviations: BSA, body surface area; LA, left atrial; LAEF, left atrial ejection fraction; LAVED, left atrial volume at end‐diastolic; LAVES, left atrial volume at end‐systolic; LV, left ventricular; LVCI, left ventricular cardiac index; LVEDV, left ventricular end‐diastolic volume; LVEF, left ventricular ejection fraction; LVESV, left ventricular end‐systolic volume; LVSV, left ventricular stroke volume; MAPSE, mitral annular plane systolic excursion; RA, right atrial; RAEF, right atrial ejection fraction; RAVED, right atrial volume at ventricular end‐diastole; RAVES, right atrial volume at ventricular end‐systole; RV, right ventricular; RVCI, right ventricular cardiac index; RVEDV, right ventricular end‐diastolic volume; RVEF, right ventricular ejection fraction; RVSV, right ventricular stroke volume.

^a Native T1 and T2 values indicate T1 and T2 relaxation times measured by T1 and T2 mapping. All p values were calculated by independent‐samples t‐test.

3.2. Exercise‐Related Group Differences in Resting Right Heart Parameters

Detailed longitudinal CMR outcomes across the 2‐year follow‐up are summarized in Table 3 (with additional descriptive trajectories in Table S7). Across all right heart parameters, no significant time × group interactions were detected (all p > 0.05), indicating that longitudinal trajectories over the 2‐year follow‐up were comparable between groups. After multivariable adjustment (including baseline characteristics, clinical status, physical performance, laboratory parameters, and individual baseline values), significant group main effects were identified for right atrial volume at ventricular end‐diastole (RAVED, p = 0.003) and right atrial volume at ventricular end‐systole (RAVES, p < 0.001), as well as their body surface area (BSA)‐indexed metrics (Figure 2). Right heart functional parameters, including RV stroke volume (RVSV), BSA‐indexed RVSV, RV cardiac index (RVCI), and RV peak filling rate also demonstrated significant group main effects (p = 0.002, p = 0.006, p = 0.016, and p = 0.006, respectively; Figure 3).

TABLE 3.

Effects of time, group, and their interaction on outcomes from linear mixed‐effects models.

Characteristics	p
	Time effect	Group effect	Time × group interaction	1‐year between‐group	2‐year between‐group
RAVED, mL	0.013	0.003	0.473	0.029	0.002
RAVED/BSA, mL/m2	0.038	0.020	0.904	0.047	0.040
RAVES, mL	0.050	0.001	0.401	0.007	0.002
RAVES/BSA, mL/m2	0.048	< 0.001	0.297	0.014	0.002
RVEDV, mL	0.857	0.001	0.693	0.002	0.011
RVEDV/BSA, mL/m2	0.747	0.007	0.726	0.007	0.030
RVESV, mL	0.018	0.421	0.351
RVESV/BSA, mL/m2	0.011	0.410	0.339
RVSV, mL	0.172	0.002	0.294	0.001	0.029
RVSV/BSA, mL/m2	0.203	0.006	0.318	0.003	0.055
RVEF, %	0.009	0.158	0.117
RVCI, L/min/m2	0.260	0.016	0.157	0.006	0.094
RAEF, %	0.026	0.163	0.165
RV peak ejection rate, mL/s	0.664	0.092	0.747
RV peak filling rate, mL/s	0.287	0.006	0.955	0.020	0.021
LAVED, mL	0.046	0.446	0.727
LAVED/BSA, mL/m2	0.081	0.468	0.654
LAVES, mL	0.056	0.990	0.636
LAVES/BSA, mL/m2	0.090	0.990	0.745
LVEDV, mL	0.117	0.004	0.976	0.010	0.015
LVEDV/BSA, mL/m2	0.201	0.006	0.950	0.019	0.020
LVESV, mL	0.002	0.341	0.685
LVESV/BSA, mL/m2	0.003	0.370	0.637
LVSV, mL	0.280	0.004	0.854	0.011	0.009
LVSV/BSA, mL/m2	0.004	0.206	0.171
LVEF, %	0.006	0.063	0.381
LVCI, L/min/m2	0.299	0.012	0.846	0.023	0.022
LAEF, %	0.086	0.271	0.303
LV peak ejection rate, mL/s	0.309	0.018	0.422	0.150	0.016
LV peak filling rate, mL/s	0.201	0.014	0.194	0.219	0.002
Peak diastolic LV wall thickness, mm	0.302	0.063	0.371
Peak LV wall thickness, mm	0.029	0.906	0.639
MAPSE Lateral, mm	0.480	0.255	0.212
MAPSE Septal, mm	0.455	0.855	0.065
TAPSE, mm	0.492	0.473	0.928
Peak radial strain, %	0.360	0.847	0.238
Peak circumferential strain, %	0.462	0.419	0.213
Peak longitudinal strain, %	0.002	0.676	0.893
Native T1 a , ms	< 0.001	0.768	0.141
Native T2 a , ms	0.218	0.993	0.287
Epicardial fat volume, mL	0.083	0.028	0.049	0.002	0.416
Peak VO2, mL/kg/min	0.070	0.023	0.003	0.316	0.001
Absolute peak VO2, L/min	0.210	< 0.001	0.005	0.009	< 0.001
VO2 at AT, mL/kg/min	0.224	0.026	0.033	0.193	0.006
Absolute VO2 at AT, L/min	0.770	0.001	0.010	0.059	< 0.001

Note: Peak diastolic LV wall thickness: maximum LV myocardial wall thickness across all LV slices during diastole. Peak LV wall thickness: maximum LV myocardial wall thickness across all LV slices and cardiac phases.

Abbreviations: BSA, body surface area; LA, left atrial; LAEF, left atrial ejection fraction; LAVED, left atrial volume at end‐diastolic; LAVES, left atrial volume at end‐systolic; LV, left ventricular; LVCI, left ventricular cardiac index; LVEDV, left ventricular end‐diastolic volume; LVEF, left ventricular ejection fraction; LVESV, left ventricular end‐systolic volume; LVSV, left ventricular stroke volume; MAPSE, mitral annular plane systolic excursion; Peak VO₂, peak oxygen uptake; RA, right atrial; RAEF, right atrial ejection fraction; RAVED, right atrial volume at ventricular end‐diastole; RAVES, right atrial volume at ventricular end‐systole; RV, right ventricular; RVCI, right ventricular cardiac index; RVEDV, right ventricular end‐diastolic volume; RVEF, right ventricular ejection fraction; RVSV, right ventricular stroke volume; VO₂ at AT, oxygen uptake at anaerobic threshold.

^a Native T1 and T2 values indicate T1 and T2 relaxation times measured by T1 and T2 mapping. Bold type indicates a p value < 0.05. p values were adjusted for baseline age, sex, coronary artery disease, hypertension, Type 2 diabetes, lipid‐lowering drugs, body mass index, 4‐m gait speed, 5‐times sit‐to‐stand test, low‐density lipoprotein cholesterol, urinary albumin‐to‐creatinine ratio, individual baseline values and peak VO₂ (except for analyses of peak VO₂‐related parameters).

FIGURE 2.

Mean changes in right atrial structural parameters during follow‐up. (A) RAVED/BSA and (B) RAVES/BSA. Data points represent raw means and error bars represent 95% CI (n = 54 for the training group and n = 38 for the control group at each time point). p values were adjusted for baseline age, sex, coronary artery disease, hypertension, Type 2 diabetes, lipid‐lowering drugs, body mass index, 4‐m gait speed, 5‐times sit‐to‐stand test, low‐density lipoprotein cholesterol, urinary albumin‐to‐creatinine ratio, peak VO₂ and individual baseline values. BSA, body surface area; RAVED, right atrial volume at ventricular end‐diastolic; RAVES, right atrial volume at ventricular end‐systolic.

FIGURE 3.

Mean changes in right ventricular functional parameters during follow‐up. (A) RVSV/BSA, (B) RVCI, and (C) RV peak filling rate. Data points represent raw means and error bars represent 95% CI (n = 54 for the training group and n = 38 for the control group at each time point). p values were adjusted for baseline age, sex, coronary artery disease, hypertension, Type 2 diabetes, lipid‐lowering drugs, body mass index, 4‐m gait speed, 5‐times sit‐to‐stand test, low‐density lipoprotein cholesterol, urinary albumin‐to‐creatinine ratio, peak VO₂ and individual baseline values. BSA, body surface area; RVSV, right ventricular stroke volume; RVCI, right ventricular cardiac index; RV, right ventricle.

Post hoc pairwise assessments based on fixed‐effect parameter estimates confirmed that the exercise group had higher adjusted mean values of RAVED, RAVES, RVEDV, and RVSV at both 1‐ and 2‐year follow‐ups (all p < 0.05), whereas higher adjusted mean values for BSA‐indexed RVSV and RVCI were observed exclusively at the 1‐year visit (both p < 0.05).

Significant time effects were noted for RV end‐systolic volume (RVESV, p = 0.018) and RV ejection fraction (p = 0.009). In the absence of significant time × group interactions, these time effects reflect longitudinal shifts across the entire study cohort over the 2‐year period, regardless of exercise grouping.

3.3. Exercise‐Related Group Differences in Resting Left Heart Parameters

For LA parameters, no significant group main effects or time × group interactions were observed for left atrial parameters (all p > 0.05). For left ventricular (LV) structural parameters, LVEDV and BSA‐indexed LVEDV demonstrated significant group main effects (p = 0.004 and p = 0.006, respectively), whereas LVESV showed a significant time effect (p = 0.002). Regarding LV functional metrics, significant group main effects were observed for LVSV, LVCI, LV peak ejection rate, and LV peak filling rate (all p < 0.05; Figure 4). Regarding LV morphometric parameters, global systolic and deformation parameters, peak LV wall thickness, LVEF, and peak longitudinal strain all demonstrated significant time effects (p = 0.029, p = 0.006, and p = 0.002, respectively), whereas no significant group main effects or time × group interactions were observed for these metrics (all p > 0.05).

FIGURE 4.

Mean changes in left ventricular structural and functional parameters during follow‐up. (A) LVSV/BSA, (B) LVCI, (C) LV peak ejection rate, and (D) LV peak filling rate. Data points represent raw means and error bars represent 95% CI (n = 54 for the training group and n = 38 for the control group at each time point). p values were adjusted for baseline age, sex, coronary artery disease, hypertension, Type 2 diabetes, lipid‐lowering drugs, body mass index, 4‐m gait speed, 5‐times sit‐to‐stand test, low‐density lipoprotein cholesterol, urinary albumin‐to‐creatinine ratio, peak VO₂ and individual baseline values. BSA, body surface area; LVSV, left ventricular stroke volume; LVCI, left ventricular cardiac index.

3.4. Myocardial Tissue Characterization, Epicardial Adiposity, and Exercise Capacity

Native T1 exhibited a significant time effect (p < 0.001) without group main effect or time × group interaction (p > 0.05), indicating progressive age‐related myocardial tissue alterations across the cohort.

In contrast to static cardiac structural metrics, significant time × group interactions were detected for epicardial fat volume (p = 0.049), peak VO₂ (p = 0.003), and VO₂ at AT (p = 0.033, Figure 5), as well as for their absolute counterparts (absolute peak VO₂, p = 0.005; absolute VO₂ at AT, p = 0.010, Figure 6), demonstrating distinct longitudinal trajectories between the exercise and control groups over time. Significant group main effects were also observed for these outcomes (all p < 0.05). Post hoc between‐group comparisons showed that the adjusted mean epicardial fat volume was lower in the control group than in the training group at the 1‐year follow‐up (p = 0.002; model‐derived estimated marginal means: 126.74 mL vs. 147.29 mL, respectively). However, this between‐group difference was attenuated and was no longer statistically significant at the 2‐year follow‐up (p = 0.416). The exercise group had higher adjusted mean values of absolute peak VO₂ at both 1‐year and 2‐year follow‐ups (p = 0.009, p < 0.001, respectively). Similarly, significant between‐group differences in adjusted means for peak VO₂, absolute VO₂ at AT, and VO₂ at AT were observed at the 2‐year follow‐up (p = 0.001, p < 0.001 and p = 0.006, respectively).

FIGURE 5.

Mean changes in epicardial fat volume, weight‐normalized cardiopulmonary reserve and cardiopulmonary exercise variables during follow‐up. (A) Epicardial fat volume, (B) Peak VO₂, and (C) VO₂ at AT. Data points represent raw means and error bars represent 95% CI (n = 54 for the training group and n = 38 for the control group at each time point). p values were adjusted for baseline age, sex, coronary artery disease, hypertension, Type 2 diabetes, lipid‐lowering drugs, body mass index, 4‐m gait speed, 5‐times sit‐to‐stand test, low‐density lipoprotein cholesterol, urinary albumin‐to‐creatinine ratio, peak VO₂ and individual baseline values. Peak VO₂, peak oxygen uptake; VO₂ at AT, oxygen uptake at anaerobic threshold.

FIGURE 6.

Mean changes in absolute cardiopulmonary reserve and cardiopulmonary exercise variables during follow‐up. (A) Absolute peak VO₂ and (B) Absolute VO₂ at AT. Data points represent raw means and error bars represent 95% CI (n = 54 for the training group and n = 38 for the control group at each time point). p values were adjusted for baseline age, sex, coronary artery disease, hypertension, Type 2 diabetes, lipid‐lowering drugs, body mass index, 4‐m gait speed, 5‐times sit‐to‐stand test, low‐density lipoprotein cholesterol, urinary albumin‐to‐creatinine ratio, individual baseline values and peak VO₂ (except for analysis of absolute peak VO₂). Peak VO₂, peak oxygen uptake; VO₂ at AT, oxygen uptake at anaerobic threshold.

3.5. Post Hoc Analysis According to Exercise Adherence

Participants were stratified by exercise adherence into high‐adherence, low‐adherence, and control groups (baseline characteristics and initial imaging parameters detailed in Tables S8 and S9; longitudinal right heart outcomes in Table 4 and Table S10). In this three‐group linear mixed‐effects model, no significant time × group interactions were observed for any right heart parameters (all p > 0.05). However, significant group main effects were identified for RAVED, RAVES, RVEDV, RVSV, RVCI, and RV peak filling rate (all p < 0.05; Figure S2). Pairwise comparisons demonstrated that higher adjusted mean values of RAVED, RAVES, RVEDV, and RVSV in the high‐adherence group relative to controls were statistically significant at both 1‐ and 2‐year follow‐ups (p < 0.05). In contrast, differences between the low‐adherence and control groups followed a more variable trajectory where RVEDV and RVSV diverged significantly only at 1 year, RAVED only at 2 years, and RAVES at both time points (p < 0.05).

TABLE 4.

Linear Mixed‐Effects Model analyses of right heart parameters of high‐adherence training, low‐adherence training, and control groups.

Characteristics	p
	Group effect	Time × group interaction	High versus control (1 year)	Low versus control (1 year)	High versus control (2 years)	Low versus control (2 years)
RAVED, mL	0.009	0.697	0.020	0.230	0.002	0.024
RAVED/BSA, mL/m2	0.036	0.992	0.026	0.321	0.024	0.276
RAVES, mL	0.002	0.449	0.023	0.016	0.002	0.040
RAVES/BSA, mL/m2	0.002	0.323	0.029	0.035	< 0.001	0.065
RVEDV, mL	0.006	0.574	0.007	0.006	0.013	0.100
RVEDV/BSA, mL/m2	0.025	0.693	0.018	0.021	0.030	0.170
RVESV, mL	0.696	0.647
RVESV/BSA, mL/m2	0.645	0.632
RVSV, mL	0.009	0.254	0.012	0.001	0.039	0.137
RVSV/BSA, mL/m2	0.023	0.353	0.018	0.006	0.061	0.217
RVEF, %	0.368	0.210
RVCI, L/min/m2	0.046	0.245	0.021	0.014	0.092	0.288
RAEF, %	0.327	0.335
RV peak ejection rate, mL/s	0.245	0.703
RV peak filling rate, mL/s	0.021	0.473	0.058	0.030	0.015	0.216

Note: Bold type indicates a p value < 0.05. p values were adjusted for baseline age, sex, coronary artery disease, hypertension, Type 2 diabetes, lipid‐lowering drugs, body mass index, 4‐m gait speed, 5‐times sit‐to‐stand test, low‐density lipoprotein cholesterol, urinary albumin‐to‐creatinine ratio, individual baseline values and peak VO₂.

Abbreviations: BSA, body surface area; RA, right atrial; RAEF, right atrial ejection fraction; RAVED, right atrial volume at ventricular end‐diastole; RAVES, right atrial volume at ventricular end‐systole; RV, right ventricular; RVCI, right ventricular cardiac index; RVEDV, right ventricular end‐diastolic volume; RVEF, right ventricular ejection fraction; RVSV, right ventricular stroke volume.

4. Discussion

This 2‐year longitudinal study evaluated the impacts of a supervised multicomponent exercise program on cardiac remodeling, tissue characterization, and exercise capacity in older adults within a reassuring clinical safety profile. For resting CMR parameters, participants in the exercise group maintained higher overall adjusted mean values across multiple right and selected left volumetric and functional metrics. However, the absence of time × group interactions indicates that these differences represent parallel, level‐shift differences between cohorts rather than exercise‐induced alterations in longitudinal remodeling trajectories, suggesting that structural adaptations of the aging heart to exercise may be physiologically constrained. Furthermore, observed time effects for parameters such as Native T1, peak LV wall thickness, LVEF, and strain reflect overall age‐dependent shifts across the total cohort rather than intervention‐specific responses. In contrast, significant time × group interactions were identified for cardiopulmonary reserve metrics (peak VO₂ and VO₂ at AT). These findings demonstrate distinct longitudinal trajectories between groups, suggesting that the benefits observed with long‐term multicomponent exercise may be linked to favorable adaptations in functional aerobic reserve, which occur independently of structural cardiac remodeling.

Previous echocardiographic studies have indicated that exercise‐induced cardiac adaptations in athletes are primarily manifested as structural remodeling rather than alterations in functional parameters [14]. Building upon our earlier work, which demonstrated that 1 year of exercise training predominantly affects right heart morphology in older adults [12], the present 2‐year analysis observed that the exercise group maintained higher adjusted mean values for RA volumes, RVEDV, and RVSV at both follow‐up time points. Despite the absence of diverging trajectories, it suggests that exercise training may induce an early adaptive response that stabilizes at a higher functional plateau, rather than triggering progressive chamber enlargement. The maintained larger right heart volumes and stroke volume likely contribute to a clinically meaningful preload reserve, which may help the older heart cope with daily circulatory demands. Chronic adaptations such as increased venous return and improved atrial compliance have been proposed to support this phenotype [15, 16, 17], although these mechanisms were not directly measured in our study.

The post hoc analysis stratified by exercise adherence showed that sustained higher right heart volumes (RAVED, RAVES, RVEDV, and RVSV) were primarily evident in the high‐adherence group throughout follow‐up. Although the time × group interaction was not significant in the three‐group model, the pattern suggests that differences in right heart volumes were more consistently maintained over 2 years in participants with higher exercise volumes. These observations are in line with the existing literature reporting an association between long‐term endurance training and larger RV cavity dimensions and stroke volume [18, 19, 20, 21]. While these findings raise the possibility that continued higher‐volume exercise contributes to preserving right heart dimensions over time, the non‐randomized nature of the adherence subgroups precludes definitive causal inferences, and this hypothesis warrants prospective evaluation.

Compared with the right heart, fewer left heart parameters showed group differences, and none exhibited time × group interactions. Lifelong exercise helps preserve cardiac compliance. However, older adults starting an exercise program rarely exhibit the strong left ventricular remodeling seen in younger people [22, 23]. In advanced age, these adaptations are often blunted or require much longer training [24]. Thus, without significant time × group interactions, our results likely reflect long‐term baseline differences between groups rather than intervention‐induced changes. Whether longer or more intense exercise can overcome this age‐related resistance warrants further research.

Native T1 showed only a time effect, reflecting age‐related diffuse myocardial changes across the entire cohort that exercise did not attenuate. Previous evidence regarding the effect of exercise on epicardial adipose tissue has been inconsistent. While an earlier meta‐analysis found insufficient evidence for exercise‐induced epicardial fat volume reduction [25], a later meta‐analysis including more exercise trials reported a significant benefit of exercise over control conditions [26]. Notably, previous studies predominantly involved participants with overweight or obesity, whereas our older participants were relatively lean. This difference in baseline adiposity may partly explain the absence of an additional exercise‐induced reduction in epicardial fat volume in our study. The decline observed in controls may reflect spontaneous longitudinal variation in epicardial fat volume, which has been shown to be associated with changes in body weight even without structured intervention [27].

Functional capacity data provided the strongest evidence of a training effect. Significant time × group interactions for peak VO₂ and VO₂ at AT, with between‐group differences reaching significance at 2 years, suggest that long‐term exercise helps preserve or support cardiopulmonary functional capacity over time, consistent with the established role of endurance training in enhancing the central and peripheral physiological determinants of peak VO₂ adaptation [28]. Higher training volumes were associated with greater gains in exercise capacity [29, 30, 31, 32], though given the non‐randomized nature of adherence groups, these dose–response observations should be interpreted with caution. Importantly, this functional improvement occurred independently of resting chamber remodeling, further supporting that the core mechanisms of exercise benefit in older adults do not depend on structural cardiac enlargement.

5. Strengths and Limitations

The current study compared older adults who received health education with those who participated in long‐term, supervised, center‐based exercise training. A key strength is the 1‐ and 2‐year follow‐up analysis of RA and RV remodeling induced by multicomponent exercise, addressing a gap in research that has primarily focused on the left heart. The use of CMR provided precise, observer‐independent measurements of cardiac structure and function, particularly valuable for assessing the right heart. Additionally, stratification into high‐ and low‐adherence groups allowed evaluation of dose‐related associations between training exposure and cardiovascular adaptations.

However, several limitations should be acknowledged. First, the study population consisted of relatively healthy community‐dwelling older adults capable of completing structured exercise training, which may limit the generalizability of the findings to frail individuals or those with significant comorbidities. In addition, the sample size was relatively limited due to the demanding nature of long‐term supervised exercise interventions, which may affect statistical power. Second, the non‐randomized design, together with the relatively high attrition rate and the per‐protocol nature of the primary analysis, may have introduced selection and survivorship (healthy adherer) bias. In addition, baseline imbalances were observed in BMI and the use of lipid‐lowering medications between the two groups. Although these variables, together with other relevant baseline covariates, were adjusted for in the statistical analyses, residual confounding from measured and unmeasured factors cannot be completely excluded. Third, missing baseline covariates were managed using mean imputation in primary analyses; however, sensitivity analyses with multiple imputation yielded consistent results, supporting the robustness of the findings to the missing‐data handling method. Finally, the CMR protocol did not include late gadolinium enhancement (LGE), and therefore focal myocardial fibrosis could not be directly assessed. Cardiac tissue characterization was based on Native T1 mapping, reflecting diffuse myocardial properties, though its interpretation is also influenced by factors such as hydration status, hematocrit, edema, and scanner/sequence variations. These factors should be considered when interpreting clinical outcomes, particularly in relation to potential arrhythmia‐related events.

6. Conclusion

Over a 2‐year follow‐up, long‐term supervised multicomponent exercise was associated with higher overall resting right heart volumetric and functional profiles in older adults, with high adherence linked to greater right heart volumes. However, exercise did not alter the longitudinal trajectories of right heart structural remodeling or left heart parameters. Notably, exercise was associated with favorable cardiopulmonary reserve profiles. These findings suggest that the cardiovascular benefits of exercise in healthy aging are primarily mediated through functional adaptations, occurring independently of structural cardiac remodeling.

7. Perspective

Future research should extend this paradigm to more vulnerable populations, such as frail or advanced‐age older adults, to determine whether long‐term structured exercise elicits distinct cardiac structural and functional adaptations in individuals with compromised baseline reserves. In addition, rigorously controlled randomized trials are needed to minimize potential selection, attrition, and healthy‐adherer biases associated with non‐randomized designs and adherence‐based analyses, thereby strengthening causal inference regarding exercise‐related cardiac remodeling and functional benefits and clarifying whether these adaptations vary according to exercise dose.

Future investigations should also explore whether non‐macroscopic cardiac adaptations, including myocardial tissue alterations and microcirculatory remodeling, contribute to improvements in cardiopulmonary reserve following structured exercise in older adults. Methodologically, integrating late gadolinium enhancement imaging for detecting focal myocardial fibrosis with advanced assessments of coronary microcirculatory function, such as coronary flow velocity reserve and myocardial perfusion imaging, could provide a more comprehensive understanding of the mechanisms underlying improvements in cardiopulmonary reserve.

Funding

This research was supported by the Major Scientific Research Special Project (grant number: 2022ZD01006) from Fujian Health Committee, Startup Fund for Scientific Research from Fujian Medical University (grant number: 2020QH1143), Fujian Natural Science Foundation (grant number: 2026J001154), Fujian Provincial Science and Technology Innovation Joint Fund Project (grant number: 2023Y9276), Fujian Provincial Innovation Strategy Joint Project (grant number: 2023R0123) and Central Leading Local Science and Technology Development Special Fund Project (grant number: 2023L3018) from Fujian Provincial Department of Science and Technology.

Ethics Statement

This study was approved by the Ethics Committee of Fuzhou University Affiliated Provincial Hospital (approval number: K2020‐04‐038).

Consent

All participants provided written informed consent.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Kaplan–Meier curves of the cumulative incidence of adverse events in the training group versus the control group.

Figure S2: Mean changes in right heart structural and functional parameters according to exercise adherence. (A) RAVED/BSA, (B) RAVES/BSA, (C) RVSV/BSA D. RVCI. Data points represent raw means and error bars represent 95% CI (n = 35 for the high‐adherence training group, n = 19 for the low‐adherence training group and n = 38 for the control group at each time point). p values were adjusted for baseline age, sex, coronary artery disease, hypertension, Type 2 diabetes, lipid‐lowering drugs, body mass index, 4‐m gait speed, 5‐times sit‐to‐stand test, low‐density lipoprotein cholesterol, urinary albumin‐to‐creatinine ratio, peak VO₂ and individual baseline values. BSA, body surface area; RAVED, right atrial volume at ventricular end‐diastolic; RAVES, right atrial volume at ventricular end‐systolic; RVSV, right ventricular stroke volume; RVCI, right ventricular cardiac index.

Table S1: Cumulative clinical adverse events during the 2‐year follow‐up.

Table S2: Characteristics of the study participants after multiple imputation of missing baseline data at baseline.

Table S3: Baseline characteristics of the initial enrolled participants.

Table S4: Baseline cardiac magnetic resonance parameters of the initial enrolled participants.

Table S5: Baseline characteristics of included participants and those lost to follow‐up in the training and control groups.

Table S6: Baseline cardiac magnetic resonance parameters of the included participants and those lost to follow‐up in training and control groups.

Table S7: Cardiac magnetic resonance parameters of the study participants at 1‐year and 2‐year follow‐up.

Table S8: Baseline characteristics of the high‐adherence training, low‐adherence training, and control groups.

Table S9: Baseline cardiac magnetic resonance parameters of the high‐adherence training, low‐adherence training, and control groups.

Table S10: Right heart cardiac magnetic resonance parameters of the high‐adherence training, low‐adherence training, and control groups at 1‐year and 2‐year follow‐up.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy.