Left atrial fibrosis predicts left ventricular ejection fraction response after atrial fibrillation ablation in heart failure patients: the Fibrosis-HF Study

Abstract

Aims

Atrial fibrillation (AF) and heart failure (HF) often coexist. Catheter ablation has been reported to restore left ventricular (LV) function but patients benefit differently. This study investigated the correlation between left atrial (LA) fibrosis extent and LV ejection fraction (LVEF) recovery after AF ablation.

Methods and results

In this study, 103 patients [64 years, 69% men, 79% persistent AF, LVEF 33% interquartile range (IQR) (25–38)] undergoing first time AF ablation were investigated. Identification of LA fibrosis and selection of ablation strategy were based on sinus rhythm voltage mapping. Continuous rhythm monitoring was used to assess ablation success. Improvement in post-ablation LVEF was measured as primary study endpoint. An absolute increase in post-ablation LVEF ≥10% was defined as ‘Super Response’. Left atrial fibrosis was present in 38% of patients. After ablation LVEF increased by absolute 15% (IQR 6–25) (P < 0.001). Left ventricular ejection fraction improvement was higher in patients without LA fibrosis [15% (IQR 10–25) vs. 10% (IQR 0–20), P < 0.001]. An inverse correlation between LVEF improvement and the extent of LA fibrosis was found (R² = 0.931). In multivariate analysis, the presence of LA fibrosis was the only independent predictor for failing LVEF improvement [odds ratio 7.2 (95% confidence interval 2.2–23.4), P < 0.001]. Echocardiographic ‘Super Response’ was observed in 55/64 (86%) patients without and 21/39 (54%) patients with LA fibrosis, respectively (P < 0.001).

Conclusion

Presence and extent of LA fibrosis predict LVEF response in HF patients undergoing AF ablation. The assessment of LA fibrosis may impact prognostic stratification and clinical management in HF patients with AF.

Open in new tabDownload slide

Introduction

Atrial fibrillation (AF) and heart failure (HF) are linked, share various cardiovascular risk factors, and increase mortality.^1–3 Atrial fibrillation is accepted to be a cause of tachycardia-induced cardiomyopathy, whereas HF is a powerful independent predictor for the development of AF.

Recently, randomized trials suggested that the clinical outcome of patients with systolic left ventricular (LV) dysfunction undergoing AF ablation might be superior to those treated by conventional drug therapy.^4–6 The restoration of sinus rhythm (SR) by AF ablation resulted in significant improvement in ventricular function and reduced mortality.^4,7–9

However, there is considerable variation in clinical response to AF ablation among the patients. The observation of LV ejection fraction (LVEF) responders and non-responders after successful rhythm control leads to further considerations on the pathophysiological interplay between AF and HF. As such fibrosis in the left atrium (LA) and LV is considered a substrate for the development of AF and HF—potentially linking both disease entities. Although data on LV fibrosis exist,^10–12 it is unclear whether LA fibrosis also has an impact on ventricular function and prognosis.

Therefore, the aim of this study was to investigate the correlation between the presence/extent of LA fibrosis and later LVEF recovery after successful rhythm control in HF patients with AF.

Methods

Study design

An investigator-initiated, prospective, non-randomized study was conducted at Heart Center Dresden, Germany between June 2013 and June 2017. The study was approved by the institutional ethical review board (EK 284092012), and all participants provided written informed consent. Data were collected, managed, and analysed at Heart Centre Dresden and the Steinbeis Research Institute ‘Rhythm and Heart’.

Inclusion and exclusion criteria

Patients were included if they (i) had a reduced LVEF ≤40% on transthoracic echocardiography (TTE) under either normal SR, atrial paced rhythm or rate-controlled AF <90 b.p.m., (ii) suffered from clinical HF, and (iii) suffered from symptomatic AF.

Patients were excluded if they (i) had undergone prior AF ablation, (ii) presented with other reversible causes of HF (e.g. coronary or valvular), (iii) suffered from recent or ongoing myocarditis or pericarditis, and (iv) received additional HF device therapy or cardiac transplantation.

Study endpoints

The primary study endpoint measured improvement in post-ablation LVEF >6 months after the ablation procedure. An absolute increase in post-ablation LVEF ≥10% was defined as ‘Super Response’.

The secondary study endpoint measured ablation success. Using continuous device monitoring with an AF detection ≥6 min a total post-procedural AF burden <0.1% was defined as successful rhythm control.⁷

Assessment of left atrium surface area and left atrium fibrosis

Assessment of LA surface area and LA fibrosis followed a clinical routine and were performed as described previously.^10,13 In brief, a bipolar voltage map was created simultaneously with LA surface reconstruction, guided by a three-dimensional electroanatomical mapping system (CARTO3, Biosense Webster, or Ensite Precision, Abbott) using a circular mapping catheter (Lasso, 4 mm interelectrode spacing, Biosense Webster or Advisor FL, 3 mm, interelectrode spacing, Abbott). All mapping points were taken in SR. Patients in AF received electrical cardioversion in the electrophysiology lab. For each mapping point, stable contact between the local atrial tissue and each pair of electrodes of the circular mapping catheter was required. Extra care was taken while collecting voltage points on the border of low voltage zones. Sufficient quality of the acquired voltage points was verified by the following criteria: (i) identical P-wave morphology, (ii) identical coronary sinus activation sequence, (iii) identical cycle length, (iv) identical local bipolar electrogram morphology for at least two beats, (v) consistent differences of local activation time and for at least two beats, and (vi) reproducibility.

Quantification of LA fibrosis was performed as described previously.^10,13 Predefined landmarks (see caption of Table 2) were used to separate the LA into five regions, i.e. septum, anterior, posterior, inferior, and lateral walls, omitting the LA appendage. Each region was further divided into nine equally sized blocks. In each predefined region at least 30 voltage mapping points were collected. Median voltage within each block was calculated and used for further offline analysis. Contiguous areas of bipolar voltage <0.5 mV were considered as an area of LA fibrosis. Total LA surface area was defined as the LA body area without the pulmonary vein (PV) antrum regions, LA appendage orifice, and mitral valve. Medians of the total LA surface area and area of each predefined region were measured offline on the three-dimensional reconstructed LA model. Localization and extent of LA fibrosis were measured within each region and block separately. Median values of LA fibrosis area were set in relation to the surface area of each region and the entire LA.

Atrial fibrillation ablation concept

Ablation was performed as described previously.^10,13 In brief, all patients received wide encircling PV isolation (PVI) with proven entrance and exit block. Additional extra PV ablation was individualized based on the voltage map where fibrotic substrates were targeted with (i) homogenization of small areas, (ii) linear lesions connecting fibrosis and anatomical obstacles (e.g. PVs or mitral annulus), and (iii) linear lesions isolating large fibrotic areas (e.g. posterior LA wall).

Post-procedural management and rhythm follow-up

Antiarrhythmic medication was routinely discontinued, and patients remained on β-blocker only. In case of recurrences, antiarrhythmic drugs were re-initiated upon individual clinical decision. Oral anticoagulation was continued for at least 3 months and thereafter according to CHA2DS2-VASc score with deviations upon patient and physician’s discretion.

Rhythm follow-up was primarily based on continuous monitoring provided through implanted cardiac devices. In patients without pacemakers/defibrillators containing intracardiac atrial lead information, implantable loop recorders were inserted subcutaneously. If patients declined, repetitive 4-day-Holters were performed. All recorded episodes were manually validated for AF by trained staff. Recurrence was defined as AF detection ≥6 min. Date, time, and length of each episode were added and set into relation of the entire follow-up period to calculate total AF burden. Atrial fibrillation burden ≤0.1% (≤10 min/week) was defined as subclinical AF.

Re-ablation for symptomatic drug-refractory recurrences of AF and atrial tachycardia was scheduled after at least 3 months from the index procedure. Re-ablation re-started follow-up assessment within this study.

Study-related investigations at baseline and during follow-up

Patients were seen at baseline, after 6 and 12 months, with further visits upon patient and physician’s discretion. During each follow-up visit continuous electrocardiogram (ECG) device monitoring data were retrieved, functional status and medication were assessed, and a 12-lead ECG and a two-dimensional TTE were recorded. Echocardiography was performed by one trained technician and validated by a specialized physician with more than 10 years cardiac imaging experience. Final LVEF values were obtained from an average of three single measurements using Simpson’s biplane method. Reasonable LVEF measurements during AF had to be performed while under optimal rate control (heart rate < 90 b.p.m.).

Statistics

Continuous variables were tested for normal distribution using the Shapiro–Wilk test. Continuous data with normal distribution are presented as mean ± standard deviation, otherwise as median and interquartile range (IQR) if not stated otherwise. Categorical data are reported as number and percentage of patients.

At baseline and at follow-up comparison of continuous data was performed using the Student’s t-test if normally distributed, otherwise with the Wilcoxon signed-rank test. The χ² test was used to compare categorical data.

Differences of continuous data between baseline and follow-up were tested with the sign test for paired samples.

To test for predictors of LVEF improvement, we used a logistic regression analysis and calculated odds ratios (ORs) with 95% confidence interval (CI), adjusting for the following variables in the model: age, gender, diabetes mellitus, arterial hypertension, underlying cardiac disease, LA fibrosis, and baseline LVEF.

Correlation between LA fibrosis and LVEF improvement as well as LA fibrosis and rhythm outcome were evaluated by regression analysis.

A P-value of lower than 0.05 was considered significant. The data were collected and managed by the investigators (S.N. and M.B.K.) and all statistics were performed using Stata version 15 (Stata Corporation, College Station, TX, USA).

Results

Study cohort

We included 103 consecutive patients [median age 64 (56–73) years, 69% male] with clinical HF [median New York Heart Association Class II (II; III)] and a severely reduced LVEF of 33% (25–38) who suffered from AF (79% persistent).

Patients showed LA and LV enlargement (LA 48 ± 5 mm, LV end-diastolic 56 ± 9 mm). They had a median CHA2DS2-VASc score of 3 (2–4). Prior rhythm control using antiarrhythmic drugs had been attempted in 24% of patients. Optimal medical HF therapy was established. Detailed baseline characteristics are shown in Table 1.

The 103 patients underwent 120 ablation procedures (1.2 procedures/patient). A second ablation procedure was performed in 14/103 (14%) and a maximum of three procedures in 3/103 (3%) patients, respectively. All patients have completed >6 months of follow-up after the last ablation procedure.

Assessment of LA fibrosis

Left atrial fibrosis was detected in 39/103 (38%) patients. All of them received additional ablation personalized to the localization of the individual fibrotic tissue areas. For LA fibrosis quantification a total of 183 ± 35 mapping points per entire LA were collected. Baseline characteristics of patients with and without atrial fibrosis are shown in Table 1. Patients with LA fibrosis were older and had a higher prevalence of diabetes. The extent of fibrotic surface areas measured a median of 16.1 cm² (IQR 9.6–27.0) and covered 19% (IQR 11–27) of the total LA surface of the affected patients. Tissue pathologies were most frequently found in the LA anterior wall (62% of the affected patients) with a median fibrosis extent of 7.4 cm² (IQR 4.1–12.8). Fibrosis was second most prevalent in the LA septum and the LA posterior wall (51% of the affected patients each), with a median fibrosis extent of 8.2 cm² (IQR 3.9–13.9) and 8.8 cm² (IQR 4.4–14.1), respectively. Examples of LA fibrosis burden and distribution are presented in Figure 1 and Table 2.

Ablation outcome

Mean follow-up after the last ablation procedure measured 11 ± 5 months. Rhythm follow-up data after the last ablation procedure are complete in 103/103 (100%) patients at 6 months and 79/103 (77%) patients at 12 months, respectively. Permanent continuous monitoring data are available in 87/103 (84%) patients.

After 1.2 procedures/patient freedom from AF recurrence was documented in 87/103 (85%) patients. With continuous device monitoring 73/87 (84%) patients recorded a total AF burden <0.1%. Freedom from any device detected arrhythmia measured 70/87 (80%).

Rhythm outcome did not differ between patients with and without LA fibrosis. Freedom from AF recurrence measured 33/39 (84%) in patients with and 54/64 (84%) in patients without LA fibrosis, respectively (P = 0.485). There was no significant association between LA fibrosis extent and AF burden during follow-up (P = 0.299).

Primary study endpoint

Follow-up measurements of LVEF in SR or atrial paced rhythm were available in 103/103 (100%) patients with no need of prior cardioversion. Echocardiographic outcome data are presented in Table 3. In comparison to baseline, LVEF improved from 33% (IQR 25–38) to 50% (IQR 35–55) (P < 0.001), which indicates an absolute and relative LVEF improvement of 15% (IQR 6–25) and 50% (IQR 25–71), respectively. Echocardiographic ‘Super Response’ with an absolute LVEF improvement ≥10% was observed in 76/103 (74%) patients.

Post-ablation LA and LV reverse remodelling were further observed. A significant LA diameter reduction from 48 ± 5 to 45 ± 8 mm (P = 0.002) was present in the overall cohort as well as both subgroups (see Table 3). This is of importance, since LA fibrosis positive patients had a larger LA diameter at baseline (50 ± 5 vs. 47 ± 6 mm, P = 0.047). There was a trend for reduction of the LV diameter from 56 ± 9 to 55 ± 8 mm (P = 0.36). In the overall cohort 36 patients (36%) presented with moderate to severe mitral regurgitation (MR). Most patients had mild MR [overall cohort: n = 63, 62%; LA fibrosis (−): n = 43, 67%; LA fibrosis (+): n = 20, 54%]. Numerically, moderate MR was more frequently found in patients with LA fibrosis (43% vs. 25%) and in patients who missed LVEF improvement (46% vs. 27%) but was not statistically significant (P = 0.222 and P = 0.221). At follow-up significant improvement of MR severity was observed in the overall cohort (P = 0.047), especially in patients without LA fibrosis (P = 0.017) but not in patients with LA fibrosis (P = 0.786) (also see Tables 1 and 3).

Primary study endpoint and LA fibrosis

Absolute LVEF improvement was significantly higher in patients without LA fibrosis [15% (IQR 10–25) vs. 10% (IQR 0–20), P < 0.001].

Echocardiographic ‘Super Response’ with an absolute LVEF improvement ≥10% was observed in 55/64 (86%) patients without and 21/39 (54%) patients with LA fibrosis, respectively (P < 0.001).

A significant inverse correlation between LVEF improvement and the extent of the fibrotic LA surface was found (R² = 0.931) (Figure 2). Lack of significant LVEF improvement was observed in patients with LA fibrosis covering more than 35–40% of the LA surface. The probability of not achieving echocardiographic ‘Super Response’ increased 1.06-fold per percent LA surface fibrosis.

In multivariate analysis adjusting for age, gender, diabetes mellitus, arterial hypertension, underlying cardiac disease, LA fibrosis, and baseline LVEF, only the presence of LA fibrosis was found as an independent variable predicting failing LVEF improvement [OR 7.2 (95% CI 2.2–23.4), P = 0.001].

Assuming an echocardiographic ‘Response’ criterion typically used in HF device therapy (relative LVEF improvement > 15%) patients with fibrosis affecting <20% of the LA surface benefitted from rhythm control.

Discussion

Main findings

In this cohort of 103 patients with HF and pre-dominantly persistent AF only 38% showed LA fibrosis during endocardial voltage mapping. Freedom from AF recurrence was achieved in a substantial proportion of HF patients (84%) irrespective of presence or absence of LA fibrosis when a tailored substrate-based ablation concept was applied.

Overall, rhythm control leads to a significant LVEF improvement from 33% to 50%. Left atrial fibrosis was the only independent predictor for LVEF response. Echocardiographic ‘Super Response’ was found significantly more often in patients without LA fibrosis. Lack of LVEF improvement was observed in patients with fibrosis covering more than 35–40% of the LA surface.

Atrial fibrillation and heart failure

Several studies have investigated patients with AF and HF. Some of these trials have raised scepticism on the benefit of rhythm control over rate control.^14,15 Others—mostly ablation trials—have shown improvement in soft clinical endpoints, LVEF and mortality in HF patients undergoing AF ablation ^{4–6,16–18}. Guidelines provide a Class IIb indication for AF ablation in patients, in whom AF is suspected to contribute to HF development.¹⁹ In clinical practice, HF patients respond differently to AF ablation. Discrepancies in underlying cardiac pathologies of AF and HF development have to be suspected.

The atrial fibrillation—heart failure interplay

Conceptually AF and HF are linked by three potential interactions: (i) AF causes HF, (ii) HF causes AF, and (iii) pre-existing structural heart disease causes secondary AF which subsequently deteriorates HF.

Separating these mechanisms in a clinical setting may be difficult. Amount and representation of these three patient groups in existing AF/HF studies are unclear.

From a clinical perspective such discrimination has a high yield of relevance not only to judge prognostic effects of rhythm control but also to guide post-ablation HF management in terms of medication, sudden cardiac death risk stratification, and HF device therapy.

According to this study, atrial fibrosis could be a strong parameter that may help to stratify HF patients towards an expected benefit of rhythm control and to guide usage of other HF therapies. Based on this study we adapted our in-house management for HF/AF patients according to Figure 3. Obviously, controlled prospective clinical endpoint assessment is needed.

Emerging role of LA fibrosis

Pathology

Today fibrotic degeneration of atrial histoarchitecture is considered to alter cellular coupling with subsequent conduction impairment. Affected myocardial regions are substrates for re-entry—a keystone in AF development. Sinus rhythm voltage mapping enables clinicians to map presence, localization, and extent of such fibrotic tissue pathologies within millimetres. Development of high-density mapping catheters has further expanded our understanding of atrial tissue pathologies. Historically, atrial voltage <0.5 mV became a surrogate for the presence of atrial fibrosis and showed association with clinical outcomes, comorbidities, as well as AF trigger sites. However, the exact nature of the threshold between histological fibrosis and low voltage has not been described. First data for correlation of atrial fibrosis detected by magnetic resonance imaging (MRI) and low voltage during SR as well as mapping during AF exist.²⁰ Whether MRI-detected LA fibrosis can serve as a reference for electrophysiological conduction properties and explain arrhythmia mechanisms is debatable. Probably both techniques will contribute to future AF management but a one to one convertibility is not established so far.

Clinical atrial fibrillation type

Discrepancies between severity of atrial fibrosis and clinical AF type have been described.¹¹ Similar findings were observed in this study. In patients with HF and pre-dominantly persistent AF one would expect severe structural disarrangement with extensive atrial fibrosis. The data did not confirm this assumption. Only 38% of patients had evidence of fibrosis. The remaining patients showed a surprisingly healthy representation of LA myocardium indicating an only limited atrial disease.

Left ventricular ejection fraction response

Post-ablation LVEF response was a further indicator that AF/HF patients without atrial fibrosis are different in pathophysiology and clinical outcome. The data showed a seven times higher likelihood of a positive LVEF response in patients without atrial fibrosis. Median absolute LVEF improvement was 5% higher in patients without atrial fibrosis. Echocardiographic ‘Super Response’ was significantly more frequent in patients without atrial fibrosis.

Ablation outcome

The observed differences in LVEF response in patients with and without atrial fibrosis were not attributable towards differences in ablation success.

Overall durable rhythm control was achieved in 84% of patients. The equal ablation outcome in patients with and without extensive LA fibrosis can be interpreted as an effect of an individualized ablation strategy.

Evolution of AF ablation is moving from standardized lesions to personalized treatment of individual arrhythmia substrates.^11,12,16,17 In that context SR voltage mapping can identify areas of LA fibrosis as ablation targets beyond PVI.

When applied to this cohort of HF patients with pre-dominantly persistent AF, such strategy led to brief interventions with sole PVI in roughly two-third of patients (62%), in whom LA fibrosis was not detected.

Limitations

This was a non-randomized study. Data only apply to HF patients undergoing ablation-based rhythm control. The small cohort size with confounding factors may decrease the statistical power. Catheter ablation was performed at a high-volume centre with experienced physicians using advanced individualized ablation concepts. We did not obtained tissue for histological analysis or performed comparison to MRI to confirm that LA fibrosis detected by electroanatomical mapping could reflect fibro-fatty infiltration or amyloid. Whether prior cardioversion or current technique settings of the employed bipolar voltage map could identify the entire LA fibrosis with appropriate clinical meanings is debatable. Potential limitations in terms of reproducibility due to semi-quantitative echocardiographic LVEF measurements cannot entirely be excluded.

Conclusion with clinical and scientific perspectives

First, this study may help to explain conflicting data on HF outcome after AF rhythm control by highlighting structural phenotypic differences between AF/HF patients that point towards different mechanisms of the AF/HF interplay.

Second, the data may have implications on clinical management. The observation that patients with more than 35–40% LA fibrosis did not benefit from LVEF improvement raises the question whether rhythm control is a valuable treatment strategy in such patients. In contrary, in patients in whom lack of atrial fibrosis leads to the expectation of an echocardiographic ‘Super Response’, the questions should be raised if, when and how other HF therapies—such as device therapies—are still indicated.

Third, the study highlights the need for strategic pre-interventional assessment of atrial fibrosis—ideally non-invasively—not only to plan the ablation but also to stratify patient management according to the individual clinical AF/HF scenario.

Acknowledgements

We thank the staff of our EP department for professional patient care and the staff of Steinbeis Research Institute ‘Rhythm and Heart’ for collaboration with rhythm monitoring data evaluation and analysis. We also thank Sven Resnjanskij for expert statistical advice.

Conflict of interest: C.P. has an upcoming assignment as Chief Medical Officer at Abbott EP and Heart Failure Division, starting July 2020. B.K. has received modest lecture honoraria and congress sponsoring from Abbott and Impulse Dynamics. All other authors declared no conflict of interest.

Data availability

Data are available on request.

References

Author notes