ABSTRACT
Background
The utility of post‐procedural HFA‐PEFF reassessment for long‐term heart failure (HF) risk stratification after atrial fibrillation (AF) ablation remains uncertain.
Methods
In this single‐center retrospective cohort study, we evaluated pre‐procedural and post‐procedural HFA‐PEFF scores in 856 patients with preserved left ventricular ejection fraction (≥ 50%) undergoing index AF ablation. HF hospitalization from 1 year after ablation was compared across risk strata (low, 0–1; intermediate, 2–4; high, 5–6 points) for both scores. Discrimination was assessed using time‐dependent receiver operating characteristic analysis at 2–5 years.
Results
Post‐procedural HFA‐PEFF was assessed at a median of 91 (interquartile range, 87–97) days after ablation, and 92.8% of reassessments were obtained during sinus rhythm or atrial pacing. The HFA‐PEFF score decreased significantly from 3 (2–4) to 2 (1–4) (p < 0.001), driven primarily by changes in the biomarker domain. During a median follow‐up of 5.2 (3.5–7.6) years, 34 patients (4.0%) had HF hospitalization. Post‐procedural HFA‐PEFF showed higher discrimination than the pre‐procedural score, especially at 5 years (area under the curve [95% confidence interval], 0.791 [0.686–0.896] versus 0.686 [0.562–0.809]; p = 0.028). In subgroup analyses, discrimination for change‐based HFA‐PEFF improvement (Δ ≤ −1 versus Δ ≥ 0) was greater in the baseline high‐risk stratum (0.784 [0.635–0.934]) than in the low/intermediate‐risk stratum (0.647 [0.546–0.748]).
Conclusions
In patients with preserved left ventricular ejection fraction undergoing AF ablation, post‐procedural HFA‐PEFF reassessment may add value for long‐term HF hospitalization risk stratification, particularly in those with high baseline HFA‐PEFF scores.
Keywords: atrial fibrillation, catheter ablation, heart failure with preserved ejection fraction, HFA‐PEFF score
In this single‐center cohort of 856 patients with preserved left ventricular ejection fraction undergoing atrial fibrillation ablation, post‐procedural HFA‐PEFF reassessment provided better long‐term discrimination for heart failure hospitalization than pre‐procedural assessment over a median 5.2‐year follow‐up. The added value was most evident in patients with high baseline HFA‐PEFF scores (5–6 points), in whom comprehensive score change outperformed biomarker‐domain change alone, supporting selective post‐procedural reassessment rather than routine reassessment in all patients.
Abbreviations
- AF
atrial fibrillation
- HF
heart failure
- HFA‐PEFF
Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology
- IQR
interquartile range
- LVEF
left ventricular ejection fraction
- ROC
receiver operating characteristic
1. Introduction
The HFA‐PEFF (Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology) score is a guideline‐recommended diagnostic algorithm for heart failure with preserved ejection fraction (HFpEF) [1, 2, 3, 4], and increasing evidence suggests that its clinical relevance may extend beyond diagnosis to clinical event risk stratification [5].
In patients undergoing AF ablation, higher HFA‐PEFF scores, particularly scores of 5 to 6, are associated with increased filling pressures, even irrespective of a formal HFpEF diagnosis [4, 6]. Because AF ablation can change rhythm status and hemodynamics, HFA‐PEFF may also change dynamically from pre‐ to post‐procedure [7, 8]. However, whether post‐procedural HFA‐PEFF reassessment adds value to long‐term HF risk stratification after AF ablation remains unclear. Because HFA‐PEFF reassessment requires repeat echocardiography and natriuretic peptide testing, it is important to clarify whether reassessment should be performed routinely or selectively in patients for whom it is most informative.
Therefore, we compared the prognostic performance of pre‐procedural and post‐procedural HFA‐PEFF scores for long‐term HF hospitalization after AF ablation in patients with preserved left ventricular ejection fraction (LVEF) and explored the subgroup in which post‐procedural reassessment was most informative.
2. Methods
2.1. Study Setting and Population
This single‐center, retrospective cohort study included consecutive patients who underwent index catheter ablation for AF between October 2011 and March 2024. Patients were eligible if they had preserved left ventricular ejection fraction (LVEF) ≥ 50%, no history of underlying myocardial disease, and available echocardiographic and B‐type natriuretic peptide (BNP) data both before and within 1 year after the procedure. Underlying myocardial disease included cardiomyopathy, congenital heart disease, greater‐than‐moderate valvular heart disease, and prior valvular surgery. The study was approved by the Institutional Review Board of Yamaguchi University Hospital, which waived the requirement for written informed consent using an opt‐out approach.
2.2. Data Collection and HFA‐PEFF Assessment
Baseline clinical characteristics and procedural/follow‐up data, including redo procedures, arrhythmia recurrence, and rhythm status at reassessment and HF hospitalization, were extracted from medical records. Pre‐procedural echocardiography was generally performed within 1 month before ablation under clinically stable conditions. Post‐procedural echocardiography was generally scheduled approximately 3 months after the index or redo ablation procedure. When more than one echocardiographic examination was available within 1 year after the index procedure, results obtained during sinus rhythm or atrial pacing were preferentially analyzed. BNP was measured as part of routine clinical care, mostly at the same visit as echocardiography; therefore, rhythm status for biomarker‐domain scoring was determined from the corresponding echocardiographic/clinical assessment.
The HFA‐PEFF score was calculated according to the HFA‐PEFF diagnostic algorithm (Table S1) [2]. Points were assigned across the functional, morphological, and biomarker domains and summed to yield an overall score ranging from 0 to 6. Patients were categorized as Low‐risk (0–1 point), Intermediate‐risk (2–4 points), or High‐risk (5–6 points). When a component required for scoring was missing, the corresponding item was assigned 0 points.
2.3. Study Endpoints
The primary endpoint was HF hospitalization occurring from 1 year after the index procedure onward, ensuring temporal separation between post‐procedural reassessment and subsequent outcome ascertainment. HF hospitalization was defined as any unplanned admission requiring intravenous diuretic therapy and/or other intensified treatment for decompensated HF after discharge following the index ablation procedure, as previously described [9, 10]. The diagnosis was adjudicated by consultant cardiologists on the basis of symptoms together with BNP elevation and/or pulmonary congestion on chest radiography. Secondary endpoints included the shift in HFA‐PEFF score and risk category from pre‐ to post‐procedure and the comparative discrimination for subsequent HF hospitalization using pre‐ and post‐procedural HFA‐PEFF scores, including change‐based analyses within prespecified baseline high‐risk and low/intermediate‐risk strata.
2.4. Ablation Strategy and Follow‐Up
For the index ablation, pulmonary vein isolation was performed using a three‐dimensional electroanatomical mapping system regardless of the AF type. Ablation was performed by using radiofrequency energy or balloon‐based catheters, predominantly cryoballoons. Atrial substrate modification was not routinely performed during the index procedure and was reserved for repeat ablation when pulmonary vein isolation (and, when applicable, superior vena cava isolation) had already been confirmed to be durable.
Patients were typically discharged within a few days after the procedure. A standardized follow‐up visit at 3 months included a comprehensive clinical evaluation. Subsequent follow‐ups were generally scheduled every 6–12 months for up to 5 years or until the end of follow‐up, with additional assessments performed as clinically indicated. Arrhythmia recurrence was defined as any atrial tachyarrhythmia lasting > 30 s beyond a 3‐month blanking period.
2.5. Statistical Analysis
Continuous variables are summarized as mean with standard deviation or median with interquartile range, as appropriate. Categorical variables are reported as counts with percentages. Pre‐ to post‐procedural comparisons were performed using paired t‐tests, Wilcoxon signed‐rank tests, or McNemar's tests, as appropriate. The cumulative probability of HF hospitalization was estimated using the Kaplan–Meier method and compared across HFA‐PEFF risk categories using the log‐rank test. Incidence rates were calculated as events per 100 person‐years with 95% confidence intervals (CIs) and compared using a Poisson model.
Discriminative performance for subsequent HF hospitalization was evaluated using time‐dependent receiver operating characteristic analysis at 2, 3, 4, and 5 years after the index procedure. Areas under the curve (AUCs) with 95% CIs were estimated for pre‐procedural and post‐procedural HFA‐PEFF scores using inverse probability of censoring weighting to account for right censoring. Comparisons between paired time‐dependent AUCs were performed using influence‐function‐based variance estimates.
To evaluate the value of post‐procedural HFA‐PEFF reassessment, change‐based analyses were performed within prespecified baseline HFA‐PEFF strata: High‐risk and Low/Intermediate‐risk. ΔHFA‐PEFF and Δbiomarker‐domain score were each defined as the post‐procedural score minus the pre‐procedural score and categorized as Improved (Δ ≤ −1) or Not improved (Δ ≥ 0). Within each stratum, 5‐year time‐dependent AUCs were compared between ΔHFA‐PEFF and Δbiomarker‐domain score.
Because AF type may influence rhythm‐related changes in BNP levels and left atrial volume index after ablation, additional exploratory analyses were performed to compare pre‐ to post‐procedural changes in the HFA‐PEFF score and its components between patients with paroxysmal AF and those with persistent AF. Δ values were calculated as post‐procedural values minus pre‐procedural values. Between‐group differences in Δ were assessed using the Wilcoxon rank‐sum test.
To explore the potential influence of recurrent atrial tachyarrhythmia, we compared atrial tachyarrhythmia recurrence and rhythm status according to HF hospitalization and HFA‐PEFF improvement status using Fisher's exact test.
Multivariable Cox proportional hazards models were constructed to evaluate the association between post‐procedural HFA‐PEFF score and HF hospitalization after adjustment for selected clinically relevant covariates. Because HF hospitalization was assessed from 1 year after the index procedure onward, atrial tachyarrhythmia recurrence within 1 year after the index procedure was used as the recurrence variable in multivariable analyses to preserve temporal ordering. Given the limited number of HF hospitalization events, the models included a limited number of covariates. Model 4 additionally included baseline BNP to account for baseline biomarker burden; BNP was log2‐transformed before inclusion in the model.
Sensitivity analyses repeated the time‐dependent receiver operating characteristic analysis in patients with complete paired HFA‐PEFF component data and after accounting for all‐cause death as a competing event using an inverse probability of censoring weighting framework.
Subgroup, change‐based, and additional multivariable analyses were considered exploratory because of the limited number of HF hospitalization events. Two‐sided p < 0.05 was considered statistically significant. Analyses were performed using R version 4.3.2.
3. Results
3.1. Study Population
A total of 856 patients were included (Figure S1), and baseline characteristics are summarized in Table 1. During a median follow‐up of 5.2 (interquartile range, 3.5–7.6) years, 379 patients (44.2%) experienced arrhythmia recurrence after the index procedure, and 196 (22.9%) after the final procedure. Within 1 year after the index procedure, arrhythmia recurrence occurred in 235 patients (27.5%).
TABLE 1.
Patient characteristics.
| Characteristics | Total (n = 856) |
|---|---|
| Age (years), mean (SD) | 66 (12) |
| Female sex, n (%) | 277 (32.4) |
| Persistent AF, n (%) | 279 (32.6) |
| AF duration (months), median (IQR) | 12 (5–48) |
| BSA (m2), mean (SD) | 1.7 (0.2) |
| Systolic blood pressure (mmHg), mean (SD) | 126 (17) |
| Diastolic blood pressure (mmHg), mean (SD) | 75 (13) |
| Heart rate (/min), mean (SD) | 70 (15) |
| CHA2DS2‐VASc score, mean (SD) | 2.2 (1.5) |
| LAD (mm), mean (SD) | 41 (7) |
| LVEF (%), mean (SD) | 65 (7) |
| eGFR (mL/min/1.73 m2), mean (SD) | 64 (17) |
| Comorbidities | |
| Hypertension, n (%) | 502 (58.6) |
| Diabetes, n (%) | 142 (16.6) |
| Congestive HF, n (%) | 85 (9.9) |
| History of ischemic stroke/TIA, n (%) | 86 (10.0) |
| Atherosclerosis, n (%) | 70 (8.2) |
| Procedure‐related parameters | |
| Radiofrequency PVI, n (%) | 601 (70.2) |
| Balloon‐based PVI, n (%) | 255 (29.8) |
| CTI‐ablation, n (%) | 162 (18.9) |
| SVCI, n (%) | 237 (27.7) |
| Therapeutic agents | |
| DOACs, n (%) | 725 (84.7) |
| VKA, n (%) | 127 (14.8) |
| ACEI/ARB, n (%) | 339 (39.6) |
| ARNI, n (%) | 8 (0.9) |
| Beta‐blocker, n (%) | 446 (52.1) |
| MRA, n (%) | 60 (7.0) |
| Statin, n (%) | 249 (29.1) |
| SGLT2i, n (%) | 29 (3.4) |
| GLP‐1RA, n (%) | 1 (0.1) |
| Diuretics, n (%) | 115 (13.4) |
| Class I AAD, n (%) | 272 (31.8) |
| Class III AAD, n (%) | 86 (10.0) |
Note: Continuous data are expressed as mean (SD) or median (IQR; first‐third quartile), while categorical data are expressed as numbers (percentages).
Abbreviations: AAD, antiarrhythmic drug; ACEI, angiotensin‐converting enzyme inhibitor; AF, atrial fibrillation; ARB, angiotensin II receptor blocker; ARNI, angiotensin receptor‐neprilysin inhibitor; BSA, body surface area; CHA2DS2‐VASc, Congestive heart failure, Hypertension, Age ≥ 75 years, Diabetes, prior Stroke/TIA/thromboembolism, Vascular disease, Age 65–74 years, Sex category (female); CTI, cavotricuspid isthmus; DOAC, direct oral anticoagulant; eGFR, estimated glomerular filtration rate; GLP‐1RA, glucagon‐like peptide‐1 receptor agonist; IQR, interquartile range; LAD, left atrial diameter; LVEF, left ventricular ejection fraction; MRA, mineralocorticoid receptor antagonist; PVI, pulmonary vein isolation; SD, standard deviation; SGLT2i, sodium‐glucose cotransporter 2 inhibitor; SVCI, superior vena cava isolation; TIA, transient ischemic attack; VKA, vitamin K antagonist.
3.2. Post‐Procedural Assessment and Changes in HFA‐PEFF Score
In the overall cohort, post‐procedural echocardiography was performed at a median of 91 (87–97) days after the index procedure, and 92.8% of post‐procedural assessments were obtained during sinus rhythm or atrial pacing. In 46 patients (5.3%), post‐procedural reassessment was based on echocardiographic data obtained after redo ablation performed within 1 year after the index procedure; in these patients, the interval from the index procedure to the reassessment echocardiogram was a median of 218 (137–246) days. Among these patients, additional atrial substrate modification was performed in 5 patients (0.6% of the overall cohort).
Figure 1 illustrates the pre‐ to post‐procedural shifts in HFA‐PEFF risk classification and transitions in domain scores. Overall, the HFA‐PEFF score decreased significantly from 3 (2–4) pre‐procedure to 2 (1–4) post‐procedure (p < 0.001; Table 2). This improvement was primarily driven by a reduction in the biomarker domain score (from 1 [0–1] to 0 [0–1], p < 0.001; Table 2). Among the echocardiographic measures, the largest absolute change was observed in the left atrial volume index (p < 0.001). Missingness of individual HFA‐PEFF components was low overall and is shown in brackets in Table 2. The highest missing rate was observed for estimated pulmonary artery systolic pressure, which increased from 7.0% before the procedure to 13.0% after the procedure, whereas missingness was limited for E/e′, left atrial volume index, and BNP.
FIGURE 1.
Pre‐ to post‐procedural shifts in HFA‐PEFF after AF ablation. Sankey‐style flow diagrams show pre‐ to post‐procedural shifts in HFA‐PEFF risk classification (A) and domain scores (B). AF, atrial fibrillation; HFA‐PEFF, Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology.
TABLE 2.
Changes in HFA‐PEFF score and individual components from pre‐ to post‐procedure.
| Variables | Pre‐procedure | Post‐procedure | p |
|---|---|---|---|
| AF/AFL rhythm, n (%) | 305 (35.7) [0] | 62 (7.2) [0] | < 0.001 |
| HFA‐PEFF score, median (IQR) | 3 (2–4) [0] | 2 (1–4) [0] | < 0.001 |
| Functional domain, median (IQR) | 0 (0–1) [0] | 0 (0–1) [0] | 0.025 |
| E/e′, mean (SD) | 9.3 (3.5) [3.6] | 9.5 (3.6) [2.8] | 0.032 |
| ePASP (mmHg), mean (SD) | 26 (5) [7.0] | 26 (5) [13.0] | 0.162 |
| Morphological domain, median (IQR) | 2 (1–2) [0] | 2 (1–2) [0] | < 0.001 |
| LAVI (mL/m2), mean (SD) | 43 (14) [5.2] | 39 (12) [4.8] | < 0.001 |
| LVMI (g/m2), mean (SD) | 85 (18) [0] | 84 (18) [0] | 0.008 |
| RWT | 0.39 (0.11) [0] | 0.38 (0.06) [0] | 0.088 |
| Biomarker domain, median (IQR) | 1 (0–1) [0] | 0 (0–1) [0] | < 0.001 |
| BNP (pg/mL), median (IQR) | 67 (28–140) [0] | 35 (18–74) [2.1] | < 0.001 |
Note: Data are shown as mean (SD), median (IQR), or n (%), as appropriate. Missing data are shown in brackets as [missing rate (%)]. Pre‐ and post‐procedure comparisons were performed using paired analyses with available paired observations.
Abbreviations: AF, atrial fibrillation; AFL, atrial flutter; BNP, B‐type natriuretic peptide; ePASP, estimated pulmonary artery systolic pressure; HFA‐PEFF, Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology; IQR, interquartile range; LAVI, left atrial volume index; LVMI, left ventricular mass index; RWT, relative wall thickness; SD, standard deviation.
In exploratory analyses stratified by AF type, reductions in BNP and left atrial volume index were greater in patients with persistent AF than in those with paroxysmal AF (Table S3). In contrast, changes in the total HFA‐PEFF score and each domain score did not significantly differ between AF types. Functional indices, including E/e′ and estimated pulmonary artery systolic pressure, showed only modest changes in both groups.
3.3. HF Hospitalization
From 1 year after the index procedure onward, 34 patients experienced at least 1 HF hospitalization. Analyses were based on time to first HF hospitalization, although a total of 45 HF hospitalizations occurred during follow‐up. Among total HF hospitalizations, 29 events (64.4%) had AF or atrial flutter at admission. BNP at HF hospitalization had a median value of 446 (308–654) pg/mL. The median time to the first HF hospitalization was 5.1 (3.3–7.8) years. All‐cause death occurred in 46 patients at a median of 4.6 (2.5–6.8) years. The Kaplan–Meier curves are shown in Figure 2, and HF hospitalization was consistently stratified by the HFA‐PEFF risk category defined at both the pre‐ and post‐procedural assessments (Table S2), with the highest cumulative probability and incidence rate observed in the High‐risk category, particularly at the post‐procedural assessment.
FIGURE 2.
Cumulative probability of HF hospitalization by HFA‐PEFF risk category Kaplan–Meier curves show the cumulative probability of HF hospitalization from 1 year after the index ablation, stratified by pre‐procedural (A) and post‐procedural (B) HFA‐PEFF risk categories. CI, confidence interval; HF, heart failure; HFA‐PEFF, Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology.
Additional analyses related to atrial tachyarrhythmia recurrence are shown in Tables S4–S6. Atrial tachyarrhythmia recurrence within 1 year after the index procedure did not significantly differ between patients with and without HF hospitalization, whereas recurrence after the index or final procedure was more frequent in patients with HF hospitalization (Table S4). Atrial tachyarrhythmia recurrence within 1 year was not significantly different according to HFA‐PEFF improvement status in the overall cohort or in the baseline High‐risk stratum, although it was more frequent in patients without HFA‐PEFF improvement in the baseline Low/Intermediate‐risk stratum (Table S5). In multivariable Cox models, the post‐procedural HFA‐PEFF score remained associated with HF hospitalization after adjustment for selected clinical covariates, including atrial tachyarrhythmia recurrence within 1 year (Table S6).
3.4. Discrimination for HF Hospitalization
Time‐dependent ROC analyses at prespecified time points (2, 3, 4, and 5 years) generally showed higher AUCs for post‐procedural HFA‐PEFF scores than for pre‐procedural scores (Figure 3 and Table 3A). The separation was most pronounced at 5 years (AUC: 0.791 [95% CI, 0.686–0.896] versus 0.686 [0.562–0.809], p = 0.028). Sensitivity analyses in the complete paired cohort and in models accounting for all‐cause death as a competing event showed similar patterns (Tables 3B and 3C).
FIGURE 3.
Time‐dependent discrimination of HF hospitalization using pre‐ and post‐procedural HFA‐PEFF Time‐dependent receiver operating characteristic curves show discrimination for HF hospitalization by pre‐ and post‐procedural HFA‐PEFF scores at 2–5 years after the index procedure (A‐D). HF, heart failure; HFA‐PEFF, Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology; ROC, receiver operating characteristic.
TABLE 3A.
Time‐dependent ROC.
| Time (years) | Pre‐procedural HFA‐PEFF score, AUC (95% CI) | Post‐procedural HFA‐PEFF score, AUC (95% CI) | p |
|---|---|---|---|
| 2 | 0.738 (0.538–0.939) | 0.802 (0.675–0.929) | 0.538 |
| 3 | 0.589 (0.376–0.803) | 0.712 (0.563–0.862) | 0.056 |
| 4 | 0.700 (0.555–0.844) | 0.788 (0.665–0.911) | 0.125 |
| 5 | 0.686 (0.562–0.809) | 0.791 (0.686–0.896) | 0.028 |
Note: Data are shown as AUC (95% CI). p values compare pre‐ and post‐procedural HFA‐PEFF scores using paired time‐dependent AUC analysis with right‐censoring adjustment.
Abbreviations: AUC, area under the curve; CI, confidence interval; HFA‐PEFF, Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology; ROC, receiver operating characteristic.
TABLE 3B.
Time‐dependent ROC analyses in the complete paired cohort.
| Time (years) | Pre‐procedural HFA‐PEFF score, AUC (95% CI) | Post‐procedural HFA‐PEFF score, AUC (95% CI) | p |
|---|---|---|---|
| 2 | 0.624 (0.417–0.832) | 0.785 (0.607–0.964) | 0.113 |
| 3 | 0.589 (0.403–0.776) | 0.731 (0.560–0.902) | 0.018 |
| 4 | 0.669 (0.521–0.817) | 0.830 (0.706–0.954) | 0.002 |
| 5 | 0.682 (0.539–0.825) | 0.843 (0.727–0.961) | 0.001 |
Note: Data are shown as AUC (95% CI). p values compare pre‐ and post‐procedural HFA‐PEFF scores using paired time‐dependent AUC analysis.
Abbreviations: AUC, area under the curve; CI, confidence interval; HFA‐PEFF, Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology; ROC, receiver operating characteristic.
TABLE 3C.
Time‐dependent ROC with competing risks.
| Time (years) | Pre‐procedural HFA‐PEFF score, AUC (95% CI) | Post‐procedural HFA‐PEFF score, AUC (95% CI) | p |
|---|---|---|---|
| 2 | 0.739 (0.539–0.938) | 0.801 (0.674–0.927) | 0.541 |
| 3 | 0.589 (0.375–0.802) | 0.710 (0.560–0.860) | 0.057 |
| 4 | 0.698 (0.554–0.842) | 0.787 (0.663–0.910) | 0.125 |
| 5 | 0.688 (0.564–0.812) | 0.790 (0.685–0.895) | 0.032 |
Note: Data are shown as AUC (95% CI). AUCs were estimated with all‐cause death treated as a competing event; p values compare paired time‐dependent AUCs.
Abbreviations: AUC, area under the curve; CI, confidence interval; HFA‐PEFF, Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology; ROC, receiver operating characteristic.
3.5. Comparison of ΔHFA‐PEFF and Δbiomarker Domain Score
Figure 4 and Tables 4A and 4B compare changes in the HFA‐PEFF score (ΔHFA‐PEFF) with changes in the biomarker domain score alone (Δbiomarker). Within the High‐risk group (n = 126), patients with HFA‐PEFF improvement (Δ ≤ −1) had a lower incidence of HF hospitalization than those without improvement (Table 4A), and 5‐year discrimination was greater for ΔHFA‐PEFF than for Δbiomarker (0.784 [0.635–0.934] versus 0.647 [0.496–0.799]; p < 0.001; Table 4B). In contrast, within the Low/Intermediate‐risk group (n = 730), HF hospitalization rates were low regardless of improvement status (Table 4A), and discrimination was similar between the two metrics (0.647 [0.546–0.748] versus 0.643 [0.621–0.664]; p = 0.932; Table 4B).
FIGURE 4.
Stratification after ablation based on changes in HFA‐PEFF and the biomarker domain Kaplan–Meier curves show the cumulative probability of HF hospitalization by improvement status, defined as improved (Δ ≤ −1) or not improved (Δ ≥ 0), according to ΔHFA‐PEFF and Δbiomarker domain in baseline High‐risk (A, B) and Low/Intermediate‐risk (C, D) strata. Time‐dependent receiver operating characteristic curves compare discrimination by ΔHFA‐PEFF and Δbiomarker domain at 5 years in baseline High‐risk (E) and Low/Intermediate‐risk (F) strata. CI, confidence interval; HF, heart failure; HFA‐PEFF, Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology; ROC, receiver operating characteristic.
TABLE 4A.
Cumulative probability and incidence rate of HF hospitalization by ΔHFA‐PEFF and Δbiomarker domain within high‐risk and low/intermediate‐risk groups.
| Group | Cumulative probability at median follow‐up period (5.2 years), % (95% CI) | p | Incidence rate throughout follow‐up, per 100 person‐years (95% CI) | p |
|---|---|---|---|---|
| Stratified by ΔHFA‐PEFF score (within High‐risk group) | ||||
| Improved (Δ ≤ −1, n = 77) | 1.42 (0–4.16) | 0.026 | 1.75 (0.75–3.46) | 0.031 |
| Not improved (Δ ≥ 0, n = 49) | 18.8 (3.68–31.6) | 4.66 (2.40–8.14) | ||
| Stratified by Δbiomarker domain score (within High‐risk group) | ||||
| Improved (Δ ≤ −1, n = 48) | 2.27 (0–6.57) | 0.198 | 1.36 (0.37–3.48) | 0.043 |
| Not improved (Δ ≥ 0, n = 78) | 10.96 (2.15–18.98) | 3.82 (2.18–6.21) | ||
| Stratified by ΔHFA‐PEFF score (within Low/Intermediate‐risk group) | ||||
| Improved (Δ ≤ −1, n = 276) | 1.16 (0–2.79) | 0.578 | 0.55 (0.25–1.06) | 0.906 |
| Not improved (Δ ≥ 0, n = 454) | 2.51 (0.87–4.12) | 0.58 (0.33–0.95) | ||
| Stratified by Δbiomarker domain score (within Low/Intermediate‐risk group) | ||||
| Improved (Δ ≤ −1, n = 188) | 0.99 (0–2.90) | 0.890 | 0.71 (0.30–1.40) | 0.502 |
| Not improved (Δ ≥ 0, n = 542) | 2.35 (0.89–3.80) | 0.53 (0.31–0.85) | ||
Note: Data are shown as cumulative probability at the median follow‐up of 5.2 years and incidence rate per 100 person‐years, each with 95% CI. Risk strata were defined by pre‐procedural HFA‐PEFF score: High‐risk, 5–6 points; Low/Intermediate‐risk, 0–4 points.
Abbreviations: CI, confidence interval; HF, heart failure; HFA‐PEFF, Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology.
TABLE 4B.
Time‐dependent ROC at 5 years by ΔHFA‐PEFF and Δbiomarker domain, shown separately within high‐risk and low/intermediate‐risk groups.
| Group | ΔHFA‐PEFF score, AUC (95% CI) | Δbiomarker domain score, AUC (95% CI) | p |
|---|---|---|---|
| High‐risk | 0.784 (0.635–0.934) | 0.647 (0.496–0.799) | < 0.001 |
| Low/Intermediate‐risk | 0.647 (0.546–0.748) | 0.643 (0.621–0.664) | 0.932 |
Note: Data are shown as AUC (95% CI). AUCs compare Improved (Δ ≤ −1) versus Not improved (Δ ≥ 0) status within each baseline risk stratum.
Abbreviations: CI, confidence interval; HF, heart failure; HFA‐PEFF, Heart Failure Association Pretest assessment, Echocardiography and natriuretic peptide, Functional testing, and Final etiology.
4. Discussion
4.1. General Overview
The present study evaluated the clinical utility of the HFA‐PEFF score before and after AF ablation for long‐term HF risk stratification in a cohort with preserved LVEF. First, the overall HFA‐PEFF score decreased significantly from pre‐ to post‐procedure, with the largest contribution from a reduction in the biomarker domain. Second, although both pre‐ and post‐procedural HFA‐PEFF risk categories stratified HF hospitalization risk, time‐dependent ROC analyses showed higher AUCs for the post‐procedural score, with the largest separation at 5 years. Third, the apparent utility of post‐procedural reassessment differed by baseline risk stratum. In the High‐risk group, discrimination for change‐based improvement (Δ ≤ −1) was more apparent. Additionally, ΔHFA‐PEFF showed higher discrimination than the Δbiomarker‐domain score alone. In contrast, in the Low/Intermediate‐risk group, ΔHFA‐PEFF provided discrimination comparable to Δbiomarker‐domain score alone, suggesting limited incremental value of comprehensive reassessment.
4.2. The Change in HFA‐PEFF Score After the Ablation
While randomized trials of AF ablation in HFpEF have yet to demonstrate definitive benefits for hard outcomes [11, 12, 13], observational cohorts have reported post‐ablation improvements in natriuretic peptide levels and echocardiographic indices [7, 8, 14]. These changes may lower the HFA‐PEFF score, particularly through the biomarker and morphological domains. Notably, this reduction has been associated with baseline features consistent with a less advanced phenotype, including higher hemoglobin levels, smaller left atrial size, and lower filling pressure surrogates [7].
In this context, a post‐ablation decrease in the HFA‐PEFF score may reflect a shift in the post‐ablation clinical phenotype rather than definitive evidence of intrinsic improvement in HFpEF physiology. In the present exploratory analysis stratified by AF type, reductions in BNP and left atrial volume index were greater in patients with persistent AF, whereas changes in E/e′ and estimated pulmonary artery systolic pressure were limited, and changes in the total HFA‐PEFF score and domain scores did not significantly differ between AF types. These findings support the interpretation that post‐procedural HFA‐PEFF reassessment may capture rhythm‐related biomarker and atrial‐size changes, together with residual abnormalities after ablation. In patients with preserved systolic function, reverse remodeling is difficult to assess using LVEF improvement, as is commonly done in patients with systolic dysfunction [15]. Instead, reductions in natriuretic peptide levels and left atrial size may serve as alternative phenotypic correlates, although whether these changes primarily reflect ventricular versus atrial myocardial improvement remains uncertain [16, 17].
4.3. Clinical Significance of Post‐Procedural HFA‐PEFF Score
Thus, a post‐procedural HFA‐PEFF score may better reflect the updated post‐ablation physiologic status and provide additional information beyond the pre‐procedural score for subsequent risk stratification. However, the literature linking changes in HFA‐PEFF from pre‐ to post‐procedure with clinical outcomes remains sparse and, when available, has largely been limited to relatively short‐term follow‐up periods (e.g., up to 3 years) [8]. To help address this gap, we evaluated long‐term outcomes. Our cohort had a generally low absolute event risk, given preserved LVEF and the absence of underlying myocardial disease. Accordingly, when limited to events occurring from 1 year after the procedure, HF hospitalizations occurred after a prolonged period (median 5.1 years). Even in this context, post‐procedural HFA‐PEFF generally showed higher discrimination than the pre‐procedural score.
One possible explanation for the higher discrimination of the post‐procedural score is that reassessment may help distinguish baseline HFA‐PEFF elevation driven by transient AF‐related hemodynamic perturbations from elevation reflecting established diastolic abnormality or underlying atrial cardiomyopathy [18]. Because the high‐risk category (5–6 points) may include both transient and persistent contributors to score elevation, post‐procedural reassessment may be particularly informative in this group by separating these components after ablation‐mediated rhythm intervention.
Of note, several HFA‐PEFF components, including BNP, E/e′, and left atrial volume index, as well as post‐procedural HFA‐PEFF reassessment, have been associated with atrial tachyarrhythmia recurrence after AF ablation [19, 20, 21, 22]. These components may reflect shared substrates for both recurrent atrial tachyarrhythmias and HF hospitalization, such as atrial remodeling, elevated filling pressure, and biomarker abnormalities. In the present study, atrial tachyarrhythmia recurrence during follow‐up was also more frequent among patients with HF hospitalization. However, post‐procedural HFA‐PEFF remained associated with HF hospitalization after accounting for atrial tachyarrhythmia recurrence within 1 year after the index procedure, suggesting that post‐procedural reassessment may capture residual HF vulnerability beyond this recurrence status alone.
4.4. Clinical Implications
These findings suggest that pre‐ and post‐procedural HFA‐PEFF may have complementary roles in long‐term HF risk stratification after AF ablation. The pre‐procedural score may be useful as a gatekeeper to identify patients in whom post‐procedural reassessment is most informative, particularly those in the High‐risk category (5–6 points). In this subgroup, recalculating the full HFA‐PEFF score after ablation may provide updated risk information beyond baseline evaluation, and comprehensive score change appears more informative than biomarker‐domain change alone.
In contrast, in baseline Low/Intermediate‐risk patients (0–4 points), the absolute risk of HF hospitalization was low, and comprehensive change‐based reassessment did not outperform biomarker‐domain change alone. Therefore, the incremental value of systematic repeat echocardiography solely for HFA‐PEFF rescoring may be limited in this group.
4.5. Limitations
This study has several limitations. First, because this was a single‐center retrospective study, residual selection bias and limited generalizability cannot be excluded; therefore, the findings should be considered hypothesis‐generating and require external validation. Second, although HF hospitalization was adjudicated using predefined criteria, complete separation of AF‐ or atrial flutter‐related symptoms from HF decompensation remained difficult in some events. Third, the study period spanned more than a decade, during which ablation strategies, periprocedural management, HF therapy, and echocardiographic platforms evolved; these temporal changes and clinically reported measurements may have influenced arrhythmia recurrence, HF outcomes, and HFA‐PEFF components. Fourth, atrial substrate modification beyond pulmonary vein isolation was uncommon in our population, and uptake of contemporary HFpEF therapies, including sodium‐glucose cotransporter 2 inhibitors, was low; therefore, generalizability to patients treated with more extensive ablation strategies or contemporary HFpEF‐directed therapy remains uncertain. Fifth, post‐procedural reassessment was preferentially performed during sinus rhythm or atrial pacing to minimize rhythm‐related variability and to characterize the rhythm‐stabilized post‐ablation clinical phenotype. However, this approach may have biased post‐procedural HFA‐PEFF scores toward lower values, particularly through rhythm‐dependent changes in BNP and left atrial size. Therefore, a decrease in HFA‐PEFF score should not be interpreted as definitive evidence of intrinsic improvement in HFpEF physiology. Sixth, atrial tachyarrhythmia recurrence during follow‐up was more frequent among patients with HF hospitalization; however, recurrence after the index or final procedure may occur in temporal proximity to, or after, HF hospitalization. Therefore, these recurrence variables should be interpreted descriptively. To preserve temporal ordering, atrial tachyarrhythmia recurrence within 1 year after the index procedure was used in the multivariable analyses, but this approach may not fully capture the cumulative burden, late recurrence patterns, or hemodynamic impact of recurrent atrial tachyarrhythmias during long‐term follow‐up. Seventh, the number of HF hospitalization events was limited. Therefore, time‐dependent ROC analyses, subgroup analyses, change‐based comparisons between ΔHFA‐PEFF and Δbiomarker‐domain score, and additional multivariable models may be statistically unstable, particularly in smaller subgroups. To reduce the risk of overfitting, the multivariable models were limited to a small number of clinically relevant covariates. Accordingly, these analyses should be interpreted as exploratory and require confirmation in larger cohorts with more HF events. Finally, all procedures used conventional thermal ablation technologies, and the applicability of these findings to pulsed‐field ablation requires further study [23].
5. Conclusions
In this cohort with preserved LVEF undergoing AF ablation, post‐procedural HFA‐PEFF reassessment may provide additional value for long‐term HF hospitalization risk stratification. This value appeared most evident in patients with high baseline HFA‐PEFF scores.
Funding
This work was supported by the Daiwa Securities Health Foundation Research Grant Program, 52nd Research Grant, grant number 20.
Ethics Statement
The study protocol was approved by the Institutional Review Board of Yamaguchi University Hospital (H2019‐044‐2).
Consent
The requirement for written informed consent was waived, and an opt‐out approach was used.
Conflicts of Interest
Yasuhiro Yoshiga received lecture fees from Medtronic Inc. Akihiko Shimizu serves on the Editorial Board of Journal of Arrhythmia. The other authors declare no conflicts of interest.
Supporting information
Table S1: HFA‐PEFF scoring criteria.
Table S2: Cumulative probability and incidence rate of HF hospitalization by pre‐ and post‐procedural HFA‐PEFF risk categories.
Table S3: Changes in HFA‐PEFF score and components according to AF type.
Table S4: Atrial tachyarrhythmia recurrence and rhythm status according to HF hospitalization.
Table S5: Atrial tachyarrhythmia recurrence within 1 year according to HFA‐PEFF improvement status.
Table S6: Multivariable analyses of post‐procedural HFA‐PEFF score for HF hospitalization.
Figure S1: Study flowchart.
Acknowledgments
We thank Editage (www.editage.jp) for the English language editing. We gratefully acknowledge the Daiwa Securities Health Foundation for its support through the 52nd Research Grant Program.
Data Availability Statement
De‐identified data and analytical codes are available from the corresponding author upon reasonable request.
References
- 1. Bozkurt B., Coats A. J. S., Tsutsui H., et al., “Universal Definition and Classification of Heart Failure: A Report of the Heart Failure Society of America, Heart Failure Association of the European Society of Cardiology, Japanese Heart Failure Society and Writing Committee of the Universal Definition o,” European Journal of Heart Failure 23, no. 3 (2021): 352–380.33605000 [Google Scholar]
- 2. Pieske B., Tschöpe C., de Boer R. A., et al., “How to Diagnose Heart Failure With Preserved Ejection Fraction: The HFA–PEFF Diagnostic Algorithm: A Consensus Recommendation From the Heart Failure Association (HFA) of the European Society of Cardiology (ESC),” European Heart Journal 40, no. 40 (2019): 3297–3317. [DOI] [PubMed] [Google Scholar]
- 3. Heidenreich P. A., Bozkurt B., Aguilar D., et al., “2022 AHA/ACC/HFSA Guideline for the Management of Heart Failure: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines,” Circulation 145, no. 18 (2022): e895–e1032. [DOI] [PubMed] [Google Scholar]
- 4. Ariyaratnam J. P., Mishima R. S., Kadhim K., et al., “Utility and Validity of the HFA‐PEFF and H2FPEF Scores in Patients With Symptomatic Atrial Fibrillation,” JACC: Heart Failure 12, no. 6 (2024): 1015–1025. [DOI] [PubMed] [Google Scholar]
- 5. Sotomi Y., Iwakura K., Hikoso S., et al., “Prognostic Significance of the HFA‐PEFF Score in Patients With Heart Failure With Preserved Ejection Fraction,” ESC Heart Failure 8, no. 3 (2021): 2154–2164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Gawałko M., Habibi Z., Verhaert D. V. M., et al., “Discordance Between Left Atrial Pressure and Risk Scores in Assessment of Heart Failure With Preserved Ejection Fraction in Patients With Atrial Fibrillation Undergoing Catheter Ablation,” Europace 27, no. 10 (2025): euaf245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Egami Y., Kobayashi N., Sugino A., et al., “Predictors of Improvement in Heart Failure With Preserved Ejection Fraction After Catheter Ablation in Patients With Atrial Fibrillation,” Heart Rhythm 23, no. 1 (2026): 56–62. [DOI] [PubMed] [Google Scholar]
- 8. Sumiyoshi H., Tasaka H., Yoshida K., Yoshino M., and Kadota K., “Heart Failure Score and Outcomes in Patients With Preserved Ejection Fraction After Catheter Ablation for Atrial Fibrillation,” ESC Heart Failure 11, no. 5 (2024): 2986–2998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Ishiguchi H., Yoshiga Y., Shimizu A., et al., “Long‐Term Events Following Catheter‐Ablation for Atrial Fibrillation in Heart Failure With Preserved Ejection Fraction,” ESC Heart Failure 9, no. 5 (2022): 3505–3518. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Ishiguchi H., Yoshiga Y., Shimizu A., et al., “Novel Method for Risk Stratification of Major Adverse Clinical Events Using Pre‐ and Post‐Ablation Left Atrial Volume Index in Patients With Persistent Atrial Fibrillation,” Circulation Reports 6, no. 10 (2024): 415–423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Ichijo S., Miyazaki S., Kusa S., et al., “Impact of Catheter Ablation of Atrial Fibrillation on Long‐Term Clinical Outcomes in Patients With Heart Failure,” Journal of Cardiology 72, no. 3 (2018): 240–246. [DOI] [PubMed] [Google Scholar]
- 12. Parkash R., Wells G. A., Rouleau J., et al., “Randomized Ablation‐Based Rhythm‐Control Versus Rate‐Control Trial in Patients With Heart Failure and Atrial Fibrillation: Results From the RAFT‐AF Trial,” Circulation 145, no. 23 (2022): 1693–1704. [DOI] [PubMed] [Google Scholar]
- 13. Chieng D., Sugumar H., Segan L., et al., “Atrial Fibrillation Ablation for Heart Failure With Preserved Ejection Fraction,” JACC: Heart Failure 11, no. 6 (2023): 646–658. [DOI] [PubMed] [Google Scholar]
- 14. Fujimoto H., Doi N., Okayama S., et al., “Long‐Term Prognosis of Patients Undergoing Radiofrequency Catheter Ablation for Atrial Fibrillation: Comparison Between Heart Failure Subtypes Based on Left Ventricular Ejection Fraction,” Europace 24, no. 4 (2022): 576–586. [DOI] [PubMed] [Google Scholar]
- 15. Gopinathannair R., Etheridge S. P., Marchlinski F. E., Spinale F. G., Lakkireddy D., and Olshansky B., “Arrhythmia‐Induced Cardiomyopathies Mechanisms, Recognition, and Management,” Journal of the American College of Cardiology 66, no. 15 (2015): 1714–1728. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Prabhu S., Taylor A. J., Costello B. T., et al., “Catheter Ablation Versus Medical Rate Control in Atrial Fibrillation and Systolic Dysfunction: The CAMERA‐MRI Study,” Journal of the American College of Cardiology 70, no. 16 (2017): 1949–1961. [DOI] [PubMed] [Google Scholar]
- 17. Kuroda S., Mizukami A., Hiroki J., et al., “Clinical Impact of Serial Change in Brain Natriuretic Peptide Before and After Catheter Ablation in Patients With Atrial Fibrillation and Heart Failure,” Journal of Cardiology 77, no. 5 (2021): 517–524. [DOI] [PubMed] [Google Scholar]
- 18. Goette A., Corradi D., Dobrev D., et al., “Atrial Cardiomyopathy Revisited ‐ Evolution of a Concept. A Clinical Consensus Statement of the European Heart Rhythm Association (EHRA) of the ESC, the Heart Rhythm Society (HRS), the Asian Pacific Heart Rhythm Association (APHRS), and the Latin American Heart Rhythm Society (LAHRS),” Europace 26, no. 9 (2024): euae204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Jiang H., Wang W., Wang C., Xie X., and Hou Y., “Association of Pre‐Ablation Level of Potential Blood Markers With Atrial Fibrillation Recurrence After Catheter Ablation: A Meta‐Analysis,” Europace 19, no. 3 (2017): 392–400. [DOI] [PubMed] [Google Scholar]
- 20. Masuda M., Fujita M., Iida O., et al., “An E/E' Ratio on Echocardiography Predicts the Existence of Left Atrial Low‐Voltage Areas and Poor Outcomes After Catheter Ablation for Atrial Fibrillation,” Europace 20, no. 5 (2018): e60–e68. [DOI] [PubMed] [Google Scholar]
- 21. Njoku A., Kannabhiran M., Arora R., et al., “Left Atrial Volume Predicts Atrial Fibrillation Recurrence After Radiofrequency Ablation: A Meta‐Analysis,” Europace 20, no. 1 (2018): 33–42. [DOI] [PubMed] [Google Scholar]
- 22. Sumiyoshi H., Tasaka H., Yoshida K., et al., “Association Between Postprocedural Echocardiographic and Laboratory Data and Clinical Outcomes in Patients With Preserved Ejection Fraction After Catheter Ablation for Atrial Fibrillation,” Heart Rhythm 22, no. 4 (2025): 914–922. [DOI] [PubMed] [Google Scholar]
- 23. Verhaeghe L., L'Hoyes W., Vijgen J., et al., “Left Atrial Function After Atrial Fibrillation Ablation With Pulsed Field vs Radiofrequency Energy: A Comparative Study,” Heart Rhythm 23, no. 7 (2025): 1542–1550. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: HFA‐PEFF scoring criteria.
Table S2: Cumulative probability and incidence rate of HF hospitalization by pre‐ and post‐procedural HFA‐PEFF risk categories.
Table S3: Changes in HFA‐PEFF score and components according to AF type.
Table S4: Atrial tachyarrhythmia recurrence and rhythm status according to HF hospitalization.
Table S5: Atrial tachyarrhythmia recurrence within 1 year according to HFA‐PEFF improvement status.
Table S6: Multivariable analyses of post‐procedural HFA‐PEFF score for HF hospitalization.
Figure S1: Study flowchart.
Data Availability Statement
De‐identified data and analytical codes are available from the corresponding author upon reasonable request.