Differential increase in peak oxygen uptake with cardiac rehabilitation among patients with stage B versus C heart failure

Abstract

BACKGROUND

Although cardiac rehabilitation (CR) improves the exercise capacity of patients with cardiovascular disease, whether the magnitude of the benefit differs between patients with stage B versus C heart failure (HF) remains unclear.

AIM

To evaluate the impact of outpatient CR on the exercise capacity of patients with stage B versus C HF.

DESIGN

Multicenter retrospective observational study.

SETTING

Patients with stage B and stage C HF.

POPULATION

Outpatients underwent cardiac rehabilitation.

INTERVENTION

Not applicable.

METHODS

This multicenter retrospective study included 324 outpatients who underwent CR. Patients were classified by HF stage (B or C), and propensity score matching was performed. The primary endpoint was the change in peak oxygen uptake (peak VO₂) from baseline to follow-up. Secondary outcomes were to identify factors associated with an increase in peak VO₂ in patients with stage B and C HF.

RESULTS

Of the 324 outpatients who underwent CR, patients were classified by HF stage (B or C), and propensity score matching was performed. Among the 254 matched patients, the peak VO₂ change was greater in those with stage B versus C HF (0.36 [0.00 to 0.89] vs. 0.22 [-0.12 to 0.22] mL/kg/min per month; P=0.04). Among patients with stage B HF, older age, higher Geriatric Nutritional Risk Index, lower skeletal muscle mass, and higher baseline peak VO₂ were associated with poor increase in peak VO₂. Among patients with stage C HF, the presence of diabetes, lower handgrip strength, and higher Short Physical Performance Battery were associated with poor increases.

CONCLUSIONS

Peak VO₂ increased to a greater extent in patients with stage B versus C HF who underwent CR. Moreover, the clinical factors associated with exercise capacity gains differed by HF stage, underscoring the need to tailor CR strategies to each patient’s disease stage and phenotype.

CLINICAL REHABILITATION IMPACT

The identification of stage-specific predictors of cardiac rehabilitation response underscores the importance of personalized rehabilitation strategies tailored to heart failure stage and patient phenotype, rather than relying on uniform protocols.

Cardiac rehabilitation (CR) is a comprehensive secondary prevention program designed for patients with cardiovascular disease that integrates structured exercise training with nutritional counseling, pharmacological management, psychosocial support, and lifestyle modifications.^{1, 2} This multidisciplinary approach improves patient exercise capacity, enhances quality of life, and reduces cardiovascular event and mortality risks.³ Recent clinical guidelines strongly recommend CR as a standard component of care following major cardiovascular events or interventions, including myocardial infarction, heart failure (HF), and revascularization procedures.^{1, 2} Meta-analyses have demonstrated that CR participation among patients with various cardiovascular conditions is associated with significant improvements in their functional status and long-term prognosis.^{4, 5}

In recent years, the staging classification proposed by the American College of Cardiology Foundation and American Heart Association has been widely adopted in clinical practice to define HF progression.⁶ This system categorizes patients into stages A through D based on the presence of structural heart abnormalities and clinical symptoms. Stage B HF is defined by the presence of structural heart disease or impaired cardiac function without overt symptoms, whereas stage C HF is characterized by structural abnormalities accompanied by clinical signs or symptoms of HF. CR has been primarily directed at patients with symptomatic HF, particularly those classified as stage C or above. However, growing attention has recently been directed toward the potential benefits of early intervention in patients with stage B HF.⁷ Although asymptomatic, individuals with stage B HF — such as those with left ventricular hypertrophy, mild systolic dysfunction, valvular heart disease, or a history of myocardial infarction — are at significantly increased risk of progressing to symptomatic HF. Therefore, preventive strategies for this population may be critical for delaying HF progression and improving their long-term outcomes.⁸

Patients with stage B versus C HF differ substantially in clinical background, including systemic condition, underlying pathophysiology, and physical function.⁹ These differences may influence their responsiveness to CR and suggest the presence of distinct subgroups for whom CR may be more or less effective. However, in routine clinical practice, individualized CR programs tailored separately for patients with stages B and C HF have not been implemented consistently. Instead, standardized and uniform CR protocols are commonly applied regardless of the HF stage. One potential reason for this is that the differential responsiveness to CR and the characteristics of potential non-responders in stage B and stage C HF have not yet been adequately investigated. Clarifying how CR responsiveness varies across HF stages and identifying subgroups at risk of poor response may raise awareness of the need for stage-specific personalized CR strategies and support the development of tailored programs that better reflect patient phenotypes and disease stages. Accordingly, this study aimed to evaluate the impact of outpatient CR on the exercise capacity of patients with stage B versus C HF. We classified patients undergoing outpatient CR by HF stage and compared the changes in exercise capacity between them. We also analyzed the clinical factors associated with an increase in exercise capacity following CR with the goal of identifying patient characteristics that may help optimize CR program design and implementation.

Materials and methods

Study design and population

This retrospective multicenter observational study was conducted at five institutions. Eligible patients were those diagnosed with cardiovascular diseases (angina pectoris, myocardial infarction, atrial fibrillation, hypertrophic cardiomyopathy, or HF) who participated in outpatient CR between October 2015 and January 2025. Patients were included if they underwent cardiopulmonary exercise testing (CPX) at the initiation of CR (baseline) and at a follow-up 3-9 months later. Patients were excluded if any of the following were present: inability to perform symptom-limited exercise during CPX; inability to ambulate independently; presence of advanced atrioventricular block; implantation of a permanent pacemaker, implantable cardioverter-defibrillator, or cardiac resynchronization therapy device; decompensated HF; severe valvular heart disease; or a history of cardiac transplantation. Based on their baseline clinical data, patients were classified into two groups: those with a history of hospitalization for HF or receiving loop diuretics were defined as having stage C HF, whereas all others were classified as having stage B HF. Due to the retrospective cohort nature of this study, detailed symptom severity scales (such as the NYHA classification) were not consistently recorded for all patients. Therefore, as a pragmatic surrogate reflecting symptomatic HF in real-world clinical practice, we used the presence of prior hospitalization for HF and the use of loop diuretics as supplementary staging criteria. Patients who met both criteria were classified as having stage C HF, whereas those who met neither criterion were classified as having stage B HF. This approach is consistent with the methods used in large cohort studies such as CHART-2 and recent epidemiological investigations that have employed similar practical indicators for HF staging.^{10, 11} Upon the stratification by HF stage, propensity score matching was applied to minimize the influence of baseline confounders and ensure intergroup comparability. This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of our Institute. Informed consent was obtained from all participants using an opt-out procedure.

Cardiac rehabilitation

CR was conducted under physical therapist supervision in accordance with European Association of Preventive Cardiology guidelines.¹² Each session consisted of 30-45 min of aerobic exercise, followed by 10-15 min of resistance training. The patients participated in one to two sessions per week. At the initial visit, CPX was performed to determine the patient’s anaerobic threshold; this information was then used to individualize their exercise intensity. Training intensity was adjusted based on the target heart rate corresponding to the anaerobic threshold or a perceived exertion level of 11-13 on the Borg scale and progressively increased under supervision. In addition to the supervised sessions, the patients were encouraged to perform at least 30 min of moderate physical activity at home daily. The CR participants also received education on healthy nutrition, stress management, medication adherence, cardiovascular risk factor control, and proper sleep hygiene.

To confirm adherence, only patients who consistently participated in supervised center-based CR throughout the intervention period and completed both baseline and follow-up CPX assessments were eligible for inclusion. This approach ensured that all participants received the intended intervention at a sufficient dosage and continuity. A standardized CR framework was implemented to minimize variability across centers. All patients underwent CPX before program initiation, and exercise prescriptions, including aerobic training intensity and resistance exercise progression, were based on CPX-derived physiological thresholds. The core CR model (aerobic exercise followed by resistance training under the supervision of a physical therapist) was uniformly applied at all institutions. Although minor procedural differences existed across the centers, the standardized prescription approach using CPX and the requirement for supervised delivery ensured a high level of inter-site consistency. Furthermore, as all participants successfully completed maximal CPX testing at both time points, the cohort represented a functionally comparable population with sufficient exercise tolerance to adhere to the prescribed program.

Outcome measures

The primary outcome was the change in peak oxygen uptake (peak VO₂) per month from baseline to follow-up, compared between groups. To account for the variability in the duration of CR exposure, the change in peak VO₂ was normalized per month by dividing the absolute change between baseline and follow-up. This approach was applied to reduce the impact of heterogeneous follow-up periods among the participants and ensure comparability across individuals. In addition, analyses were performed using change-per-month values rather than raw absolute changes, allowing for adjustment for differences in CR duration. Secondary outcomes were to identify factors associated with an increase in peak VO₂ in patients with stage B and C HF.

Data collection

At the time of the baseline CPX, data were collected on age, sex, body mass index, vital signs, medical history, laboratory parameters, current medications, and left ventricular ejection fraction (LVEF) on echocardiography. Physical parameters including handgrip strength, appendicular skeletal muscle mass, Short Physical Performance Battery (SPPB), and CPX data were also recorded. At the time of the follow-up CPX, laboratory values, LVEF, and physical parameters were collected. Geriatric Nutritional Risk Index (GNRI) was calculated using the following formula: 14.89 × serum albumin (g/dL) + 41.7 × Body Mass Index (kg/m²)/22.

CPX testing

CPX was performed using a cycle ergometer and ramp protocol. During the test, the patient’s heart rate, blood pressure, 12-lead electrocardiogram findings, and respiratory gas exchange were continuously monitored. Maximal effort was defined as a respiratory exchange ratio ≥1.1 or achievement of ≥85% of the age-predicted maximum heart rate. Peak VO₂ was defined as the average oxygen consumption during the final 30 s of exercise.

Statistical analysis

To optimize potential confounders, propensity score matching was performed using age and sex as adjustment variables. Continuous variables are expressed as median [interquartile range], while categorical variables are presented as numbers and percentages. Intragroup changes from baseline to follow-up were assessed using Wilcoxon’s signed-rank test, while intergroup comparisons of changes were performed using the Mann-Whitney U-Test. The primary outcome, the monthly change in peak VO₂ from baseline to follow-up, was compared between groups using the Mann-Whitney U-Test. Based on previous reports, a 10% increase in peak VO₂ from baseline to follow-up was defined as “increase in peak VO₂.”^{13, 14} A multivariable logistic regression analysis was performed to identify clinical factors associated with peak VO₂ increases. Covariates were selected based on prognostic factors affecting the outcomes of exercise therapy.^{15, 16} Statistical significance was set at P<0.05. All statistical analyses were performed using JMP Pro version 18.1.0 (SAS Institute Inc., Cary, NC, USA).

Results

Patients

Of the 324 enrolled patients, 135 had stage B HF versus 189 with stage C HF (Supplementary Digital Material 1: Supplementary Figure 1). After 1:1 propensity score matching, 254 patients (127 in each group) were included in the final analysis. The median age of the total cohort was 72 years (range, 62-78 years), with 70% being male, and the median follow-up period was 180 days (range, 98-194 days).

Patients with stage C versus B HF had a significantly higher prevalence of atrial fibrillation and significantly higher levels of serum creatinine, high-sensitivity C-reactive protein, and B-type natriuretic peptide (BNP) or N-terminal BNP. Conversely, patients in the stage C HF group had significantly lower systolic blood pressure, serum albumin levels, and GNRI scores (Table I). The use of antiplatelet agents and statins was significantly more common in the stage B HF group, whereas the use of mineralocorticoid receptor antagonists was significantly more common in the stage C HF group. The use of beta-blockers, renin-angiotensin system inhibitors, and sodium-glucose cotransporter 2 inhibitors was comparable between groups.

Table I. —Patients’ baseline characteristics.

Characteristics	Stage B HF (N.=127)	Stage C HF (N.=127)	P value
Age (years)	72 [60-77]	73 [62-78]	0.58
Male	90 (71%)	87 (69%)	0.68
Body Mass Index (kg/m2)	23.8 [21.4-26.0]	22.7 [20.4-25.0]	0.06
Systolic blood pressure (mmHg)	126 [111-138]	119 [109-129]	0.01
Heart rate (bpm)	71 [62-79]	70 [62-80]	0.64
Disease
Hypertension	115 (91%)	102 (80%)	0.02
Diabetes mellitus	33 (26%)	44 (35%)	0.13
Atrial fibrillation	11 (9%)	57 (45%)	<0.0001
Prior myocardial infarction	73 (58%)	40 (32%)	<0.0001
Laboratory parameters
Hemoglobin (g/dL)	12.3 [11.5-13.2]	11.8 [10.8-13.2]	0.78
Serum creatinine (mg/dL)	0.91 [0.76-1.10]	1.00 [0.82-1.30]	<0.001
Serum albumin (g/dL)	4.2 [3.9-4.4] (N.=124)	3.9 [3.6-4.2] (N.=120)	<0.0001
High-sensitive CRP (mg/dL)	0.07 [0.04-0.16]	0.18 [0.05-0.47]	<0.0001
BNP ≥100 or NT-pro BNP ≥400 pg/mL	18 (17) (N.=108)	75 (64) (N.=118)	<0.0001
Medication
Antiplatelet therapy	100 (79%)	52 (41%)	<0.0001
Statin	98 (77%)	60 (47%)	<0.0001
Beta-blocker	93 (73%)	94 (74%)	0.89
ACEi/ARB/ARNI	87 (69%)	100 (79%)	0.89
MRA	19 (15%)	73 (58%)	<0.0001
SGLT2i	23 (18%)	28 (22%)	0.43
Loop diuretics	0 (0%)	35 (28%)	<0.0001
GNRI	108 [101-113]	101 [94-108]	<0.0001
LVEF (%)	60 [50-66]	48 [36-67]	<0.0001
LVEF <50%	30 (24%)	66 (52%)	<0.0001

Data are presented as median [25^th-75^th percentile] or number (%). ACEi: angiotensin-converting enzyme inhibitor; ARB: angiotensin receptor blocker; ARNI: angiotensin receptor neprilysin inhibitor; BNP: B-type natriuretic peptide; CRP: C-reactive protein; GNRI: Geriatric Nutritional Risk Index; HF: heart failure; LVEF: left ventricular ejection fraction; MRA: mineralocorticoid receptor antagonist; SGLT2i: sodium-glucose cotransporter 2 inhibitor.

Clinical parameters at baseline versus follow-up by group

Changes in peak VO₂ from baseline to follow-up are shown in Figure 1. In patients with stage B HF, peak VO₂ significantly increased from 17.5 (range, 14.2-21.0) to 18.7 (16.1-22.9) mL/kg/min (P<0.001; Figure 1A). Similarly, in patients with stage C HF, peak VO₂ increased significantly from 14.6 (range, 12.7-18.5) to 16.1 (range, 13.7-19.8) mL/kg/min (P<0.001; Figure 1B).

Figure 1.

—Intragroup changes in peak VO₂ from baseline to follow-up: A) stage B HF; B) stage C HF. HF: heart failure; VO₂: oxygen uptake.

Other physical parameters, including handgrip strength, peak metabolic equivalents, minute ventilation/carbon dioxide production (VE/VCO₂) slope, and anaerobic threshold, improved significantly from baseline to follow-up in both groups (Supplementary Digital Material 2: Supplementary Table I). Appendicular skeletal muscle mass index significantly increased only in patients with stage C HF. Among those with stage B HF, most laboratory parameters (excluding renal function) and LVEF improved significantly at follow-up versus baseline (Supplementary Digital Material 3: Supplementary Table II), while among those with stage C HF, serum albumin and low-density lipoprotein cholesterol levels improved significantly.

Primary outcome

Changes in peak VO₂ from baseline to follow-up were compared between groups (Figure 2). The increase in peak VO₂ per month was significantly greater in the stage B versus C HF group (0.36 [range, 0.00 to 0.89] mL/kg/min vs. 0.22 [range, -0.12 to 0.68] mL/kg/min, P=0.043).

Figure 2.

—Changes in peak VO₂/month from baseline to follow-up of stage B versus C HF. HF: heart failure; VO₂: oxygen uptake.

Changes in the physical parameters from baseline to follow-up between group

Changes in the physical parameters and CPX results from baseline to follow-up were compared between groups (Table II). No significant intergroup differences were noted in handgrip strength, peak metabolic equivalents, or VE/VCO₂ slope. The anaerobic threshold and peak respiratory exchange ratio significantly improved in the stage B HF group, whereas the appendicular skeletal muscle mass index significantly improved in the stage C HF group.

Table II. —Changes in patients’ physical parameters and cardiopulmonary exercise test results between baseline and follow-up by study group.

Parameter	Stage B HF	Stage C HF	P value
Handgrip strength (kg)	1.1 [-0.7 to 3.4]	1.6 [-0.6 to 0.6]	0.52
ASMI (kg/m2)	0.0 [-0.1 to 0.3]	0.2 [-0.1 to 0.4]	0.03
SPPB	0 [0 to 0]	0 [0 to 0]	0.83
Cardiopulmonary exercise test
Peak METs/month	0.3 [0.0 to 0.9]	0.2 [-0.1 to 0.7]	0.06
VE/VCO2 slope/month	-0.4 [-1.1 to 0.4]	-0.3 [-1.3 to 0.4]	0.82
Peak RER/month	0.01 [-0.01 to 0.02]	-0.01 [-0.02 to 0.01]	<0.01
AT/month (mL/kg/min)	0.2 [0.0 to 0.6]	0.1 [-0.2 to 0.4]	<0.01

Data are presented as median [25^th-75^th percentile] or number (%). ASMI: appendicular skeletal muscle mass index; AT: anaerobic threshold; HF: heart failure; METs: metabolic equivalents; RER: respiratory exchange ratio; SPPB: Short Physical Performance Battery; VCO₂, carbon dioxide production; VE: minute ventilation; VO₂: oxygen uptake.

Secondary outcomes

Factors associated with increase in peak VO₂ were explored using a multivariable logistic regression analysis (Figure 3). In the stage B HF group, age, GNRI, appendicular skeletal muscle mass index, and baseline peak VO₂ were significantly associated with increase in peak VO₂. In stage C HF, diabetes mellitus, handgrip strength, and SPPB were significantly correlated with increase in peak VO₂.

Figure 3.

—Factors associated with increase in peak VO₂ on multivariable logistic regression analysis. ASMI: appendicular skeletal muscle mass index; BNP: B-type natriuretic peptide; CI: confidence interval; CRP: C-reactive protein; GNRI: Geriatric Nutritional Risk Index; HF: heart failure; LVEF: left ventricular ejection fraction; OR: odds ratio; SPPB: Short Physical Performance Battery; VCO₂: carbon dioxide production; VE: minute ventilation; VO₂: oxygen uptake. *Adjusted for all covariates.

Discussion

This multicenter retrospective observational study investigated how responsiveness to improvements in exercise capacity varied across HF stages in patients undergoing outpatient CR. Patients were stratified into stages B and C HF based on their clinical history, and their outcomes were compared using propensity score matching. As a result, patients with stage B HF demonstrated significantly greater increases in peak VO₂ compared to those with stage C HF. Notably, this difference suggests that exercise responsiveness is not uniform across the HF continuum and that CR effectiveness may be stage-dependent. To the best of our knowledge, this is the first study to directly compare the CR magnitude between stage B and C HF and to identify stage-specific predictors of improvement. Clinical factors related to increase in peak VO₂ in stage B and C HF revealed that distinct factors influenced the enhancement of exercise capacity for each.

Our findings indicate that the magnitude of the response may be influenced by residual physiological reserve. Stage B HF represents a phase with structural heart disease but without overt symptoms of HF, in which early intervention may prevent progression and improve outcomes. Exercise capacity is an important physiological marker that predicts prognosis independent of cardiac dysfunction severity.¹⁷ Our findings suggest that initiating CR proactively in stage B HF may lead to superior increase in exercise capacity. Moreover, the observed differences in factors associated with increase in peak VO₂ between stage B and C HF highlight the critical importance of personalized medicine approaches. CR efficacy is influenced by a variety of factors, including age, sex, comorbidities, pre-exercise physical activity levels, and nutritional status.¹⁸ Therefore, it is essential to implement CR and establish tailored intervention strategies and follow-up systems that consider individual patient backgrounds to optimize their exercise capacity gains and improve their long-term prognoses. In our study, patients with stage B HF tended to have a higher baseline peak VO₂ and better nutritional status, which may limit the measurable room for improvement and introduce ceiling effects. Conversely, in stage C HF, severely impaired patients may show floor effects, with a limited capacity to increase exercise performance despite rehabilitation. Thus, part of the stage-related differences in changes in peak VO₂ may be explained by the baseline status rather than the HF stage.

Our study suggests that there may be several subgroups of patients with a lower responsiveness to CR. In stage B HF, older patients may experience physiological age-related changes such as reduced cardiac output reserve, mitochondrial dysfunction in skeletal muscle, and diminished regenerative capacity, all of which can hinder improvements in exercise tolerance.¹⁹ Additionally, loss of skeletal muscle mass, including sarcopenia, is a key limiting factor for exercise capacity and may not be adequately reversed by conventional aerobic-based CR programs.²⁰ Interestingly, even among patients with higher GNRI, indicating relatively preserved nutritional status, the increase in exercise capacity was modest. This may reflect a ceiling effect, where the preserved nutritional and physical reserves limit the potential for further gains through CR.²¹ Similarly, patients with higher baseline peak VO₂ may already have sufficient exercise capacity, leaving limited room for further improvement.²¹ In stage C HF, the presence of diabetes is known to impair muscle function and oxygen utilization during exercise due to insulin resistance, microvascular dysfunction, and alterations in muscle fiber composition.²² Diabetic patients are also more likely to have peripheral and autonomic neuropathies, which can compromise the cardiovascular and muscular responses to exercise and thereby limit CR efficacy.²² Reduced handgrip strength, a known indicator of global muscle function and sarcopenia, is also a poor prognostic factor in HF.²³ Patients with low handgrip strength may have limited adaptability to aerobic or resistance training during CR, resulting in difficulty adjusting exercise intensity and maintaining training continuity.²⁰ Beyond muscle mass, intrinsic deficits in neuromuscular junction function and muscle force generation may also contribute.²⁰ Paradoxically, we observed that patients with higher SPPB, indicating better physical function, showed less increase in exercise capacity. This too may reflect a ceiling effect, whereby those with high baseline performance have limited potential for measurable improvement through CR.²⁴

Although CR is a standard intervention, its efficacy varies and a substantial proportion of patients exhibit limited improvement in exercise capacity despite receiving standardized CR. Our findings and prior evidence suggest that specific baseline characteristics may help identify patients with reduced responsiveness to CR.²⁵ In stage B HF, structural abnormalities exist but overt symptoms are absent, suggesting that metabolic reserve and adaptability may still be partially preserved. However, our results indicated that individuals with frailty, sarcopenia, or malnutrition may respond less favorably to CR. This highlights the need for a comprehensive baseline assessment of muscle strength, skeletal muscle mass, and nutritional status. The incorporation of resistance training, nutritional support, and behavioral strategies, may improve the responsiveness of such patients. Advanced cardiac dysfunction, malnutrition, sarcopenia, and cachexia are common in patients with stage C HF. These patients may benefit from more intensive or multimodal strategies, including electromyostimulation, high-intensity interval training, and pharmacological therapies targeting skeletal muscle metabolism such as SGLT2 inhibitors and anabolic agents.^{26, 27} Moreover, psychological, cognitive, and social factors, including depression, cognitive impairment, social isolation, and low health literacy, can further limit CR participation and benefits. A multidisciplinary team-based approach is essential for addressing these non-physiological barriers. A uniform CR model may be insufficient for patients who are predicted to have limited responsiveness. Personalized CR strategies and predictive models integrating the clinical, functional, and psychosocial domains may help optimize patient selection and maximize therapeutic benefits.

A clinically notable finding of our study was that patients with stage B HF, traditionally considered asymptomatic or mildly symptomatic, demonstrated substantial responsiveness to CR. Given that CR has often been underutilized in this versus the stage C HF population, our results suggest that an earlier initiation of CR may contribute to improved long-term outcomes.²⁸ This finding aligns with the growing emphasis on preventive strategies in HF management and underscores the need to reconsider the role of CR as a primary and secondary preventive measure.²⁹ Furthermore, our findings emphasize the importance of stage-specific targeting of CR interventions. As HF progresses from stage B to C, uniform CR protocols may become insufficient. Instead, risk stratification and individualized multidisciplinary approaches become increasingly essential. To enhance the clinical applicability of our findings, we have summarized the stage-specific profiles, restricting mechanisms, and suggested CR strategies for stage B and C HF in a practical comparative table (Supplementary Digital Material 4: Supplementary Table III). This framework may help clinicians identify patients at risk of poor response and tailor CR prescriptions according to the HF stage and patient phenotype. The magnitude of improvement observed in stage B HF is not only statistically significant, but may also be clinically relevant. Previous studies have demonstrated that even modest increases in peak VO₂ (approximately +1.0-2.0 mL/kg/min) are associated with meaningful reductions in all-cause mortality and HF hospitalization.^{11, 12} In this context, the average trajectory observed in our cohort (+0.36 mL/kg/min per month) suggests that patients with stage B HF may reach this prognostically relevant threshold within a relatively short training duration. In contrast, patients with stage C HF appear to improve more slowly, requiring longer or more intensive CR strategies to achieve comparable clinical benefits. Our findings should be regarded as hypothesis-generating rather than as definitive evidence that stage B HF is universally more responsive to CR. Prospective studies that stratify patients according to their baseline functional capacity and HF stage are needed to confirm whether early-stage HF confers greater trainability.

Limitations of the study

This study has several limitations. First, its retrospective design limited causal inferences and may have introduced selection and information bias. Second, HF staging based on loop diuretic prescription and hospitalization history may have resulted in misclassification, because these variables can be influenced by non-HF indications, physician practice patterns, or timing of care rather than the true symptomatic HF status. Thus, the staging system used in this study should be interpreted as pragmatic rather than definitive. Third, although propensity score matching was applied, matching was performed using only age and sex, which may have resulted in a residual imbalance in the clinical characteristics and treatment exposure between the groups. Fourth, although a standardized CR framework was implemented, program adherence, exact exercise dose, and protocol fidelity across centers were not fully captured, and minor differences in exercise prescriptions or educational content may have influenced the outcomes. Fifth, although the peak VO₂ change was normalized per month to address variability in follow-up intervals, residual heterogeneity may remain because differences in temporal recovery dynamics or nonlinear improvement patterns cannot be fully accounted for by this adjustment. Therefore, the results should be interpreted with caution and future studies with fixed-interval assessments are warranted. Sixth, exercise capacity was evaluated solely using peak VO₂; therefore, incorporating additional measures such as six-minute walk distance, ventilatory efficiency indices, or patient-reported outcomes may have enabled a more comprehensive assessment of the functional response. Finally, the high responsiveness of patients with stage B HF to CR should be interpreted cautiously as ceiling or floor effects related to baseline peak VO₂ and functional status may have influenced the change in values, and regression to the mean cannot be fully excluded. Despite these limitations, this study is the first to directly compare CR responsiveness and its predictors across stage B and stage C HF, offering hypothesis-generating evidence to support prospective mechanistic studies and stage-specific CR strategies in future research.

Conclusions

In this multicenter retrospective study, we demonstrated that patients with stage B HF who underwent CR exhibited significantly greater increases in peak VO₂ compared to those with stage C HF. However, due to the retrospective design, potential stage misclassification, reliance on peak VO₂ alone, and the possibility of ceiling or floor effects, these findings should be interpreted as hypothesis-generating rather than definitive evidence of stage-specific responsiveness. Future prospective studies incorporating additional functional assessments and patient-reported outcomes are warranted for further validation and to guide individualized rehabilitation strategies across the HF continuum.

Supplementary Digital Material 1

Supplementary Figure 1

Study flowchart.

Supplementary Digital Material 2

Supplementary Table I

Patients' physical parameters at baseline versus follow-up by study group.

Supplementary Digital Material 3

Supplementary Table II

Laboratory parameters and LVEF at baseline and at follow-up.

Supplementary Digital Material 4

Supplementary Table III

Stage-specific profiles, restricting mechanisms, and suggested CR strategies in stage B vs. stage C HF.