Bringing doctors together

or

A free, private network for verified physicians

Leading voices

Dr. Shashank Sinha

Advanced Heart Failure · Inova Fairfax Hospital

View

Dr. Javed Butler

Advanced Heart Failure · Baylor Health Care System

View

Dr. Andrew Sauer

Advanced Heart Failure · Saint Luke's

View

Dr. Anu Lala

Advanced Heart Failure · Mount Sinai

View

Dr. Robert Mentz

Advanced Heart Failure · Duke University

View

Dr. Sunny Varshney

Advanced Heart Failure · Stanford

View

Dr. Stephanie Hsiao

Advanced Heart Failure · Stanford

View

Dr. Michelle Kittleson

Advanced Heart Failure · Cedars-Sinai

View

Dr. Adrian Hernandez

Advanced Heart Failure · Duke

View

Dr. Dorairaj Prabhakaran

General Cardiology

View
  • Discover your community

  • Stay current

  • A profile that represents you

  • Find your people

About RoonPrivacyTerms

    Overview

    Burden of Potentially Undiagnosed Heart Failure With Preserved Ejection Fraction in Atrial Fibrillation and Effects of Catheter Ablation: Insights From CABANA.

    Feb 11, 2026DOI: 10.1161/jaha.125.047295Journal ArticleAHA

    Written by: Yogesh N V ReddyNicholas BergeronRickey E CarterMargaret M RedfieldBarry A Borlaug

    Find these authors on Roon

    YR
    Dr. Yogesh Reddy
    Advanced Heart Failure · Mayo Clinic College of Medicine

    METHODS

    This post hoc analysis was conducted using deidentified data from the National Heart, Lung, and Blood Institute funded CABANA trial that was accessed through Biolincc following local institutional review board approval at Mayo Clinic and regulatory approval. The data supporting the current analysis are publicly available through the National Institutes of Health Biolincc program. The CABANA study enrolled participants with AF from 126 centers in 10 countries between November 2009 to April 2016 and randomized them to a strategy of catheter ablation or medical therapy.16 All participants provided written informed consent. Patients were eligible for inclusion if they were >65 years old, or <65 with at least 1 stroke risk factor, had experienced at least 2 episodes of paroxysmal AF or 1 episode of persistent AF within the previous 6 months, and were suitable candidates for either catheter ablation or medical therapy.16, 17To identify individuals with probable undiagnosed HFpEF, we excluded patients with a baseline EF of <50% and those with investigator reported HFpEF to focus on the large cohort with AF but without diagnosed HF at baseline. Among the remaining patients, we identified those who had self‐reported dyspnea based on the shortness of breath question recorded on their baseline Mayo Atrial Fibrillation‐Specific Symptom Inventory (MAFSI) questionnaire.18 Patients could respond to the shortness of breath question on the MAFSI as either having no dyspnea, or having dyspnea quantified as mild, moderate, or extreme. Patients with missing response to this question were assumed to have no dyspnea for this analysis. Among those patients with MAFSI self‐reported dyspnea (mild, moderate, or extreme dyspnea) and no physician‐reported HF, we used the validated HFpEF‐ABA algorithm to estimate the likelihood of undiagnosed HFpEF in this subset. The HFpEF‐ABA score uses age, AF status, and body mass index and can quantify the estimated probability of undiagnosed HFpEF in patients with dyspnea.15 A continuum of HFpEF probability is obtained first by computing log odds of HFpEF by the ABA algorithm (log odds=−7.789+0.063age in years +0.135 body mass index in kg/m2+2.041*AF status [yes=1, no=0 with all patients assigned 1 due to history of AF]), followed by transformation of log odds to probability (probability=1/[(1+e−log odds)]). Among such patients with dyspnea, the score has excellent discrimination (area under the cruve 0.839) and calibration compared with a cohort with invasively proven HFpEF defined by elevation in left heart filling pressures during rest or exertion. The score provides comparable discrimination to the echocardiography dependent H2FPEF score,15 and given the lack of necessary echocardiographic data in the CABANA data set (E/e′ and echocardiographic pulmonary artery systolic pressure),14 the HFpEF‐ABA algorithm was used in this analysis.
    Patients with self‐reported dyspnea and HFpEF‐ABA probability rule‐in ≥75% are referred to as having a high probability of potentially undiagnosed HFpEF or probable HFpEF.15, 19 The remaining patients are defined as the group without HFpEF. We additionally performed sensitivity analyses with more stringent case–control assignment, where probable HFpEF was defined by dyspnea with HFpEF‐ABA probability ≥90%.
    We evaluated longitudinal change in QOL indices that were available from Biolincc, namely the MAFSI severity and frequency scores (lower better) and EuroQol‐5 Dimension‐3 Level (EQ5D) (higher better), between the randomized arms, stratified by the presence of probable HFpEF versus non‐HFpEF. QOL end points were evaluated at baseline, 3 months, 6 months, 12 months, and annually thereafter up to 60 months (additional details of QOL scales in online supplement).
    The MAFSI scoring system consists of a 10‐item questionnaire evaluating the frequency and severity of AF symptoms. Frequency responses range from 0 (never) to 4 (always), resulting in a composite frequency score of 0 to 40, with higher scores indicating greater symptom frequency. Severity responses range from 0 (none) to 3 (severe), producing a composite severity score of 0 to 30, with higher scores indicating greater symptom severity.18 A change in MAFSI frequency score of 1.6 and severity score of 1.3 are considered clinically significant.20 Questionnaires with ≥5missing values were excluded from analysis.
    The EQ5D is a 5‐item questionnaire measuring healthy status in 5 domains, including mobility, self‐care, usual activities, pain, and anxiety/depression. Each subcategory is scored from 1 (no problems) to 3 (extreme problems) with a composite summary score calculated using a weighted algorithm (range 0–100, higher better).21 Questionnaires with any missing subcategory values were excluded from composite score calculations.
    The trial primary end point of CABANA was a composite of all‐cause mortality, serious bleed, cardiac arrest, or disabling stroke. We evaluated additional post hoc end points that are used in HF outcome trials, including (1) time to first HF hospitalization, (2) all‐cause mortality, and (3) time to first cardiovascular hospitalization.
    Baseline comparisons between the groups with probable HFpEF and without HFpEF were done using chi‐square test, t test, and Wilcoxon rank‐sum test as appropriate. To determine longitudinal changes in QOL indices and to account for the correlated nature of repeated measurements in the same individual we used a mixed model approach to test the effect of randomization to ablation on QOL by the intention to treat. Separate models were fit for MAFSI severity, MAFSI frequency, and EQ5D composite score with the baseline, 3‐, 6‐, 12‐, 24‐, 36‐, 48‐, and 60‐month responses used as the outcome variable. To fit this repeated measure mixed model, we used a compound symmetry covariance structure using participant ID as a random effect, with HFpEF group, time, randomization arm, and the full factorial of interaction by group with HFpEF, time, and randomized treatment arm as fixed effects. We additionally used logistic regression to determine the association between baseline group with HFpEF and functional impairment in each individual component of the MAFSI and EQ5D. We also used logistic regression to explore the effect of ablation in the subgroup with probable HFpEF to improve each individual MAFSI and EQ5D component between the baseline and 12‐month follow‐up period. For this analysis improvement was defined as a favorable change in the score or no worsening compared with a worsening score from baseline to 12 months. Because the HFpEF‐ABA score has been calibrated against the true prevalence of HFpEF in cohorts with dyspnea undergoing gold‐standard invasive testing,15 we used the average HFpEF‐ABA score to estimate the potential prevalence of undiagnosed HFpEF among patients with dyspnea. Cox models were to compare time to event outcomes between groups.

    RESULTS

    Out of the total CABANA trial population of 2204 participants with AF, 197 (8.9%) had EF<50%, and 260 (11.8%) patients had prior diagnosis of HFpEF leaving 1747 (79.3%) at risk for undiagnosed HFpEF, forming the population for this study (Figure S1). Of the study cohort, 1225 (70.1%) had probable HFpEF, and the remaining 522 (29.9%) formed the group without HFpEF. Of the 522 individuals in the group without HFpEF, the majority 89% (n=463) denied any symptoms of dyspnea, whereas 11% (n=59) reported dyspnea with HFpEF‐ABA probability<75%.
    Overall, 74% of participants (n=1284) complained of dyspnea of whom 97% had high HFpEF‐ABA probability. The median HFpEF‐ABA probability in these 1284 patients with dyspnea and no documented HFpEF was 0.928, which would translate to an estimate of 1192 (92.8% of 1284) of these individuals potentially having undiagnosed HFpEF, which is 68% (1192/1747) of the entire CABANA population without diagnosed HFpEF or HF with reduced EF (Figure 1).
    The group with probable HFpEF included only a minority of patients with known HFpEF (17%), with most (83%) representing patients with dyspnea and high HFpEF probability without diagnosed HFpEF. Physician adjudicated NYHA class did not reflect patient reported symptom burden in patients with self‐reported dyspnea with nearly half having reported NYHA class I despite patient reported dyspnea on questionnaire (A). Patients with probable HFpEF had more symptoms of dyspnea and effort intolerance during ordinary activities consistent with probable undiagnosed HFpEF (B). We estimated that ∼93% of patients with AF, dyspnea, and no known HF in CABANA likely had undiagnosed HFpEF (C). This translated into 68% of the overall CABANA population with no known HF potentially having undiagnosed HFpEF. AF, indicates atrial fibrillation; CABANA, Catheter Ablation vs. Antiarrhythmic Drug Therapy; HFpEF, heart failure with preserved ejection fraction; and NYHA, New York Heart Association.
    As expected, patients with probable HFpEF were older, with modestly higher body mass index compared with the group without HFpEF (Table 1). Persistent and long‐standing persistent AF was more common in the group with probable HFpEF. Patients with probable HFpEF were more likely to be female, with greater comorbidity burden including hypertension. Patients with probable HFpEF had higher resting heart rates but had comparable rate and rhythm control strategies to those without HFpEF. Echocardiographic site‐based diagnosis of diastolic dysfunction was high and comparable in both groups (79% versus 78%). Left atrium size and mitral regurgitation prevalence were higher in those with probable HFpEF (Table 1).
    Baseline Characteristics
    HFpEF indicates heart failure with preserved ejection fraction; and LV, left ventricular.
    Other refers to patients not self‐reported as White or Black.
    Although patients with probable HFpEF all self‐reported symptoms of dyspnea, site investigators assigned 48% of them to New York Heart Association class I (asymptomatic, Figure 1A; Table 2). Canadian Cardiovascular Society Severity of Atrial Fibrillation Class was worse, with greater symptom burden in probable HFpEF. The EQ5D QOL score was more impaired in probable HFpEF compared with the group without HFpEF (67.8±16.9 versus 75.5±16.7, P<0.0001) related to greater anxiety, pain, and problems with mobility, self‐care, and usual activities (Table 2; Figure 1).
    Baseline Symptoms and Quality of Life by EQ5D
    EQ5D indicates EuroQuality of Life‐5 DimensionD‐3 Level; and HFpEF, heart failure with preserved ejection fraction.
    Baseline MAFSI data were available in 96.7% of participants, with dyspnea information missing in 3.2%. Patients with probable HFpEF demonstrated worse symptoms on both the frequency and severity summary scales, including more palpitations, lightheaded episodes, and feelings of warm or being flushed (Table S1). There were also more frequent perceptions of slow heartbeat and syncopal events in the group with probable HFpEF. Symptoms of effort intolerance (chest pain, inability to exercise, fatigue/lack of energy, and weakness) were greater in frequency and severity in the group with probable HFpEF (Figures 1 and 2; Table 3). Overall, the presence of probable HFpEF was strongly associated with greater odds of worse symptom scores across all individual components of the MAFSI and EQ5D scales (Table S2).
    Probable HFpEF was associated with worse frequency and severity of tiredness/lack of energy (A) and inability to exercise (B). Numbers represent percentages. AF indicates atrial fibrillation; and HFpEF, heart failure with preserved ejection fraction.
    Baseline MAFSI Effort Intolerance Related Symptoms
    MAFSI indicates Mayo Atrial Fibrillation‐Specific Symptom Inventory.
    Following randomization, participants with probable HFpEF continued to have worse EQ5D QOL scores (lower) and MAFSI frequency and severity scores (higher) across all time points (group P<0.0001, Figure 3). Despite this, ablation resulted in greater improvement in QOL by all scales (P=0.019, P<0.0001 and P<0.0001), with greater improvement seen in the group with probable HFpEF compared with the group without HFpEF (interaction P=0.005, P=0.005 and P=0.04 for EQ5D, MAFSI frequency, and MAFSI severity scales respectively).
    Probable HFpEF was associated with worse symptoms and quality of life at baseline (as assessed by EQ5D [A] and MAFSI frequency [B] and severity [C]) and follow‐up but was associated with the greatest benefit from catheter ablation compared with medical therapy. EQ5D indicates EuroQol‐5 DimensionD‐3 Level; EuroQol, European Quality of Life; HFpEF, heart failure with preserved ejection fraction; and MAFSI, Mayo Atrial Fibrillation‐Specific Symptom Inventory.
    There was overall a favorable change in EQ5D with ablation averaged across all phases of +3.65 (95% CI, 2.16–5.14, P<0.0001) in the group with probable HFpEF, which was not observed in the group without HFpEF (−0.34 [95% CI, −2.66 to +1.99], P=0.78; Figure 4). Similarly, there were significant improvements with ablation in MAFSI frequency and severity averaged over all phases in the group with probable HFpEF (−1.76 [95% CI, −2.29 to −1.24], P<0.0001 for frequency and −1.30 [95% CI, −1.72 to −0.88], P<0.0001 for severity) but no evidence for improvement in the group without HFpEF (−0.37 [95% CI, −1.19 to +0.45], P=0.37 for frequency and −0.48 [95% CI, −1.14 to +0.19], P=0.16 for severity; Figure 4).
    (A‐C) The magnitude of benefit of ablation was greatest on symptoms and quality of life for those with probable HFpEF at baseline compared with those without HFpEF. EQ5D indicates EuroQol‐5 DimensionD‐3 Level; EuroQol, European Quality of Life; HFpEF, heart failure with preserved ejection fraction; MAFSI, Mayo Atrial Fibrillation‐Specific Symptom Inventory; and treatment (Rx).
    Among individual components of the EQ5D and MAFSI scores, ablation was associated with an improvement at 12 months in symptoms of lightheadedness, slow heartbeat, chest pain, inability to exercise, and weakness (Table S2). After adjustment for baseline AF burden and baseline AF symptoms by Canadian Cardiovascular Society Severity of Atrial Fibrillation Class, there still remained a significant interaction between ablation and the group with HFpEF on QOL response (EQ5D interaction P=0.003; MAFSI frequency interaction P=0.01 and MAFSI severity interaction P=0.04).
    Use of an unstructured instead of compound symmetry covariance structure in the mixed models resulted in consistent evidence of greater ablation efficacy in probable HFpEF on EQ5D (groupablation interaction P=0.007), MAFSI frequency (groupablation interaction P=0.012) and MAFSI severity (groupablation interaction P=0.048). Using the change relative to baseline in EQ5D, MAFSI frequency, and MAFSI severity as the outcome variable with adjustment for the baseline value as a covariate produced similar results (interaction P=0.004, 0.007, and 0.07 respectively) (Figure S2). Subgroup analyses demonstrated consistently greater effect of ablation in the group with probable HFpEF by age, race, sex, and hypertension history (Figure S3). We then performed sensitivity analyses defining probable HFpEF more strictly as patients with dyspnea and very high HFpEF‐ABA probability (≥90%), with absence of HFpEF defined by absence of dyspnea. There were similar differences in baseline characteristics and worse QOL in the group with probable HFpEF (Tables S3 and S4). Like the primary analysis, ablation remained associated with greater efficacy in the group with probable HFpEF for MAFSI frequency (groupablation interaction P=0.008) and EQ5D (groupablation interaction P=0.013), with a directionally similar effect for MAFSI severity (groupablation interaction P=0.07).
    Patients with high probability of HFpEF demonstrated an increased risk of HF hospitalization (hazard ratio [HR], 2.19 [95% CI, 1.11–4.31], P=0.01) and cardiovascular hospitalization (HR, 1.20 [95% CI, 1.03–1.40], P=0.02) compared with the group without HFpEF.
    In patients with probable HFpEF, catheter ablation had no significant effect on the risk of HF hospitalization (HR, 0.94 [95% CI, 0.55–1.64], P=0.84). There was also no significant effect on the primary end point or all‐cause mortality in patients with probable HFpEF (Table S5). Ablation was associated with lower risk of cardiovascular hospitalization in the group with probable HFpEF (HR, 0.78 [95% CI, 0.66–0.92], P=0.003) with greater efficacy compared with the group without HFpEF (interaction P=0.03).

    DISCUSSION

    In this study we have estimated the prevalence and therapeutic implications of potentially undiagnosed HFpEF among patients with AF using a validated algorithm15 in the largest randomized trial testing the effects of catheter ablation on QOL. Among this large cohort of patients with AF considered for catheter ablation, approximately 7 out of 10 with no known HF complained of dyspnea and had high HFpEF probabilities that should trigger consideration of diagnostic evaluation for HFpEF. There was substantial underappreciation of dyspnea by physician site investigators, as nearly half of patients with dyspnea were regarded as asymptomatic by site investigators. In contrast to patients without probable HFpEF, those with probable HFpEF demonstrated worse symptoms of effort intolerance, greater risk of HF hospitalization and cardiovascular hospitalization, and the poorest overall QOL. However, patients with probable HFpEF demonstrated greater benefit from catheter ablation on QOL, along with evidence for reduction in risk for cardiovascular hospitalization, despite no effect on HF hospitalization. Although symptoms improved with ablation, residual QOL impairment and risk of HF hospitalization persisted in those with probable HFpEF, both of which could potentially be targeted with foundational HFpEF therapies if diagnostic testing were performed to confirm or refute the diagnosis of HFpEF. These data suggest a high burden of undiagnosed HFpEF among patients with AF who are being considered for ablation, calling for more systematic evaluation for HFpEF among patients with AF presenting for consideration of catheter ablation and further study into the mechanisms underlying potentially differential effects of catheter ablation in those with high probability of underlying HFpEF.
    Progressive left atrial dysfunction2, 22, 23 and resulting left atrial hypertension22, 24, 25 are central to the pathogenesis of HFpEF. It is therefore not surprising that AF is a common occurrence in HFpEF.26 However, among patients with prevalent AF, the noninvasive diagnosis of HFpEF can be challenging13, 14, 15 resulting in substantial potential for underdiagnosis. Multiple small series of patients with AF and dyspnea undergoing gold‐standard invasive testing have suggested a high prevalence of undiagnosed HFpEF (>80%).5, 11, 12, 13, 27
    In the current study using a large cohort with AF, we estimate a similarly high prevalence of potentially undiagnosed HFpEF (∼70%), with notable underappreciation of dyspnea burden by physicians. These data call for further study of optimization of screening approaches for unrecognized HFpEF in patients with AF and symptoms, including consideration of objective dyspnea questionnaires to enhance sensitivity to diagnose coexisting HFpEF. On an individual patient level, the presence of dyspnea in the setting of AF appears to be consistently associated with a high likelihood of undiagnosed HFpEF,5, 12, 13 such that further evaluation for HFpEF in AF should be considered merely on the basis of patient reported dyspnea alone. Codiagnosis of HFpEF not only identifies those who potentially may derive greater symptom benefit from AF ablation but would also allow simultaneous initiation of therapies such as sodium glucose cotransporter 2 inhibitors, mineralocorticoid receptor antagonists, diuretics, and for those with obesity, incretins, to maximize improvement in QOL, decrease risk of future HF hospitalization,28, 29, 30, 31, 32, 33 and potentially even decrease AF progression.34, 35, 36 Even if the prevalence of true HFpEF is 50% lower than that predicted by the HFpEF‐ABA probabilities, this would still represent a major gap in diagnostic diligence and opportunity to improve clinical outcomes.
    Prior randomized trials testing catheter ablation have included small and underpowered HFpEF subsets, without definitive symptom benefit.8, 9 A small randomized trial of 31 patients with invasively confirmed HFpEF found that ablation was associated with improvement in exercise filling pressure and QOL supporting possible functional benefit in HFpEF.11 Prior analysis of HF from CABANA focused on the smaller imprecisely defined subset of investigator reported patients with HF based upon case report forms, used site‐reported echocardiographic “diastolic dysfunction” as an estimate of HFpEF likelihood, and also combined analyses with the subset with HF with reduced EF (where there is established benefit to ablation).10, 37 The present analysis focuses on the larger cohort with no established diagnosis of HF, providing hypothesis‐generating findings that ablation might indeed improve QOL in patients with HFpEF, potentially to an even greater extent compared with those without HFpEF. There was notably no HF hospitalization benefit observed with a modest number of events, and the current findings call for an adequately powered randomized outcome trial testing ablation in patients with AF and aggressively treated HFpEF with contemporary medical therapy.
    This study was a post hoc analysis that clearly requires confirmation in a dedicated randomized trial of contemporary ablation technologies in patients with appropriately diagnosed and treated HFpEF. There is no single noninvasive gold standard to determine the diagnosis of HFpEF in a patient with AF and dyspnea, so there is potential for misclassification of individual patients based on HFpEF probabilities alone that have not been specifically calibrated among patients with AF. However, the HFpEF‐ABA algorithm has been validated and calibrated against the invasive gold standard and therefore provides reliable group estimates of HFpEF prevalence, and current guidelines consider a high noninvasive HFpEF probability to be sufficient for HFpEF diagnosis.38 Although there was no sham‐control in CABANA, the preservation of functional benefit out to 60 months, and previous demonstration of benefit of AF ablation on QOL with a sham‐control,39 support the hypothesis that the observed QOL benefit is unlikely to be merely a placebo effect.

    CONCLUSIONS

    Roughly 7 out of 10 of participants with AF referred for possible catheter ablation as part of the CABANA trial reported dyspnea and were found to display high probability for underlying HFpEF, with poorer QOL and increased risk of HF hospitalization than those without high probability of HFpEF. Although catheter ablation in probable HFpEF was associated with greater improvement in QOL, there remained high residual QOL impairment and risk of HF hospitalization despite ablation, raising the priority for more aggressive diagnostic evaluation to enable initiation of foundational treatments for HFpEF.

    Sources of Funding

    Dr Reddy is supported by National Institutes of Health grant K23HL164901. Dr Borlaug is supported by R01 HL128526, R01 HL162828, and U01 HL160226, from the National Institutes of Health/National Heart, Lung, and Blood Institute, and W81XWH2210245 from the Congressionally Directed Medical Research Programs, US Department of Defense.

    Disclosures

    Dr Reddy receives research grants from the National Institutes of Health, Sleep Number, Bayer Accelerated Pulmonary Hypertension Award, United Jenesis Award, Merck, and the Earl Wood Career development award from Mayo Clinic. Dr Borlaug receives research support from the National Institutes of Health and the US Department of Defense, as well as research grant funding from AstraZeneca, Axon, GlaxoSmithKline, Medtronic, Mesoblast, Novo Nordisk, and Tenax Therapeutics. Dr Borlaug has served as a consultant for Actelion, Amgen, Aria, Axon Therapies, BD, Boehringer Ingelheim, Cytokinetics, Edwards Lifesciences, Eli Lilly, Imbria, Janssen, Merck, Novo Nordisk, NGM, NXT, and VADovations, and is named inventor (US Patent no. 10307179) for the tools and approach for a minimally invasive pericardial modification procedure to treat heart failure.

    Supporting information

    Tables S1–S5
    Figures S1–S3
    CONSORT 2025 checklist
    YRDr. Yogesh Reddy

    Dr. Yogesh Reddy

    Cardiology · Advanced Heart Failure & Transplantation
    Mayo Clinic
    yogesh