Are the atrial natriuretic peptides a missing link predicting low‐voltage areas in atrial fibrillation? Introducing the novel biomarker‐based atrial fibrillation substrate prediction (ANP) score

Abstract Background In patients with atrial fibrillation (AF), left atrial (LA) enlargement, and the presence of low‐voltage areas (LVAs) indicate an advanced disease stage. NT‐proANP is a biomarker, which is significantly higher in both phenotypes. Prediction of LVAs before catheter ablation could impact the prognosis and therapeutical management in AF patients. Objective The aim of this study was to (a) analyze the predictive value of a novel biomarker‐based AF substrate prediction score, and (b) compare it with DR‐FLASH and APPLE scores. Methods Patients undergoing first AF catheter ablation were included. LA volume (LAV) was analyzed prior to ablation using cardiovascular magnetic resonance imaging (CMR). Blood plasma samples from the femoral vein were collected before AF ablation. NT‐proANP was analyzed using commercially available assays. LVAs were determined using high‐density maps during catheter ablation and defined as <0.5 mV. The novel ANP score (one point for Age ≥ 65 years, NT‐proANP > 17 ng/mL, and Persistent AF) was calculated at baseline. Results The study population included 156 AF patients (64 ± 10 years, 65% males, 61% persistent AF, 28% LVAs). The cut‐off ANP score ≥ 2 demonstrated 77% sensitivity and 70% specificity. On logistic regression (odds ratio [OR] 3.469) and receiver operating characteristic (ROC) analysis (area under the curve [AUC] 0.778, P < .001), the ANP score significantly predicted LVAs presence. There were no differences between novel ANP score – which is a new one ‐ is described in the Abstract; with APPLE (AUC 0.718, P = .378) and DR‐FLASH (AUC 0.766, P = .856) scores. Conclusions The novel biomarker‐based ANP score demonstrates good prediction of LVAs.


| INTRODUCTION
Atrial fibrillation (AF) is the most common sustained arrhythmia worldwide and is associated with relevant adverse events, impaired quality of life, intensive healthcare costs, and mortality. Interventional AF ablation is a key treatment for highly symptomatic AF patients. 1 Several clinical variables, blood biomarkers, and imaging parameters are associated with AF progression. The presence of low-voltage areas (LVAs) is a surrogate parameter of advanced electro-anatomical remodeling in the left atrium (LA) and is associated with significantly worse outcomes after pulmonary vein isolation (PVI). 2 LVAs are present in approximately 20% to 25% of all AF patients and may require additional ablation strategies, continuation of antiarrhythmic drug therapy, intensive follow-up, 2 resulting in higher healthcare costs.
Pathophysiological, electrical, and structural atrial remodeling plays an important role in AF pathogenesis, 3,4 and is associated with endothelial damage, inflammation, and fibrosis. 5 Blood biomarkers are widely used in epidemiological and clinical studies to improve the clinical assessment of AF. 6 There are several biomarkers such as galectin, GDF-15, or GDF-23, which are markers of pro-fibrotic changes and may play a role in the prediction for LVAs. However, recent studies analyzing prediction of LVAs using pro-fibrotic biomarkers are inconclusive. [7][8][9] Natriuretic peptides (NP) as NT-proANP, are predominantly released in the atrial myocardial wall as a response to atrial wall stretch. 10 Underlying diseases, such as hypertension or heart failure, increase atrial stretch leading to LA dilatation, and initiation of AF. Consequently, elevated NP levels were found to be associated with an increased risk of AF initiation, and AF recurrence after intervention. 11 The LVAs prediction remains challenging and is not widely available. 12 It requires either special algorithms of the cardiovascular magnetic resonance imaging (CMR) 12 or invasive electroanatomical mapping. 2 LVAs prediction before PVI would be very helpful, because in the case of high probability of LVAs presence, the interventionalist could choose RF ablation instead of cryoballoon ablation, where only PVI without individualized linear ablation is possible. There are only few clinical scores aimed to predict LVAs in AF patients. 13,14 While the DR-FLASH score was introduced to predict LVAs first, 14 the APPLE score was developed for the prediction of AF recurrences, 13,15 and later had been also proven as the LVAs prediction score. 16 Therefore, the aim of this study was to (a) analyze the predictive value of novel biomarker-based AF substrate prediction score, and (b) compare it with APPLE and DR-FLASH scores.

| METHODS
In the present study, we included 156 patients who underwent AF catheter ablation between October 2015 and April 2017 at the Heart Center Leipzig (Germany). Inclusion criterion was a highly symptomatic and refractory to antiarrhythmic treatment AF. Exclusion criteria were pregnancy, age <18 or >75, valvular AF, cancer, acute, or systemic inflammatory diseases. The studies were approved by the local Ethical Committee (Medical Faculty, University of Leipzig) and patients provided written informed consent for participation. Paroxysmal and persistent AF were defined according to the current guidelines. 17 Paroxysmal AF was defined as self-terminating within 7 days after onset. Persistent AF lasted longer than 7 days or required drugs or direct current cardioversion for termination.

| Cardiovascular magnetic resonance imaging
Prior to AF ablation, patients underwent 1.5 T CMR (Ingenia, Philips Medical) 1 to 2 days before the intervention as previously described. 18 Briefly, a contrast-enhanced MR angiography of the left atrium and the pulmonary veins was acquired during breath-holding without electrocardiogram (ECG) gating using real-time bolus tracking. CMR data were reviewed and total LA volume (LAV) was determined after exclusion of the atrial appendage (LAA) and the pulmonary veins (PV).

| Catheter ablation
The electro-anatomical mapping was performed as previously described. 19 Briefly, two three-dimensional (3D) mapping systems

| Blood samples
Blood samples were obtained in ethylenediaminetetraacetic acid (EDTA) test tubes in fasting state prior ablation procedure from the femoral vein and processed within 1 hour of collection. Blood plasma was prepared (1000×g for 10 minutes at 20 C) and aliquots were stored at −70 C for subsequent analysis. NT-proANP levels were studied using Luminex Screening Assay (R&D/bio-techne, Minneapolis, Minnesota).

| Scores
The APPLE score (one point for Age ≥ 65 years, Persistent AF, imPaired eGFR<60 mL/min/1.73 m 2 , Left atrial (LA) diameter ≥ 43 mm, EF < 50%) and the DR-FLASH score (one point for Diabetes mellitus, Renal dysfunction, persistent Form of AF, LA diameter > 45 mm, Age > 65 years, female Sex, and Hypertension), which was developed for the prediction of LVAs, were calculated before catheter ablation using baseline patients' characteristics. 13,14 The novel ANP score (one point for Age ≥ 65 years, NT-proANP over 75th percentile in peripheral circulation [≥17 ng/mL], and Persistent AF type) was built at baseline prior to ablation. The cut-off ≥2 had been chosen to facilitate comparison of all scores and enable interpretation of previous and current results.

| Statistics
Data are presented as mean and SD for normally distributed or median (interquartile range) for skewed continuous variables, and as proportions for categorical variables. The differences between continuous values were assessed using an unpaired t-test or the Mann-Whitney, and a χ² test for categorical variables.
Logistic regression analysis was used to identify factors associated with LVAs. Multivariable analysis, which included variables with a P-value <.05 found on univariable analysis, was performed to identify independent predictors for the presence of LVAs.
Receiver operating characteristic (ROC) curves were generated for a graphical illustration of ANP, DR-FLASH, and APPLE scores' performance in predicting LVAs, with the area under the curve (AUC) being equivalent to the c-index for determining the predictive value for a score. The c indices (ie, areas under the ROC curves) for the two scores were compared by using DeLong's method. 20 A P-value <.05 was considered statistically significant, and all analyses were performed with SPSS statistical software version 25 (SPSS Inc., Chicago, Illinois).
These variables were used to build the novel biomarker-based AF substrate prediction-the ANP score.

| Scores for LVAs prediction
All scores were significantly associated with LVAs ( Figure 2 the scores (P = .378 between ANP and APPLE scores, P = .856 between ANP and DR-FLASH scores). The novel ANP score ≥ 2 demonstrated a good sensitivity and specificity for the LVAs presence (72% and 70%, respectively). The sensitivity/specificity for APPLE and DR-FLASH scores >2 were 73/63% and 93/65%, respectively.
The predictive value of the scores was similar in the subgroup of patients with persistent AF (Table 3(b)).

| DISCUSSION
To the best of our knowledge, this is the first study demonstrating the predictive value of the novel biomarker-based score for prediction of electro-anatomical remodeling (LVAs) prior to catheter ablation. We found that LVAs prediction using the novel ANP score was non-inferior or even better than the DR-FLASH and APPLE scores. Compared to other scores, the ANP score demonstrated better specificity.
Importantly, using only three variables mirroring atrial myopathy with its structural changes, it was possible to build an LVAs score with a good predictive value. and LVAs presence. 25 Furthermore, current AF management guidelines list "age-related fibrosis" as an etiological factor underlying AF. 17 Renal dysfunction is another controversial clinical factor associated with LVAs. Compared to other observations, where impaired renal function was associated with atrial fibrosis, 26 we did not find this association in our current analysis. This could be explained by only modest eGFR reduction in the LVAs group compared to the non-LVAs group. Also, our study cohort was relatively young and healthy, and only few patients had renal impairment accordingly to stage III, while none of our patients had stages IV (pre-dialysis) and V (dialysis).

| Prediction of electro-anatomical substrate using clinical variables
LA enlargement had been used as a marker of advanced electroanatomical remodeling because of its strong association with AF progression and related success of different AF management strategies. [27][28][29] The DECAAF study demonstrated atrial fibrosis detection using delayed enhancement CMR. 12 However, beside high-volume tertiary centers, CMR is not available for many EP labs worldwide.
Recently, we demonstrated the importance of CMR measurements assessing different LA parameters and their association with LVAs. 18 We found that CMR derived LA volume was superior to different monoplane LA diameters. In addition to "passive" LAV measurements, we could also show that the "active" LA function was associated with LVAs. 25 However, in the present study, the association between LAV and LVAs was not significant.

| Blood biomarkers and electro-anatomical remodeling
Besides NT-proANP in our current analysis and in the previously publi- In current study, NT-proANP ≥17 ng/mL was associated with almost 3-fold risk for the LVAs presence. This finding is in line with a previous study, which proved an association of NT-proANP levels with LVAs and persistent AF. 19 Because of its secretion from the atria as a response to atrial stress, NT-proANP might be considered as a surrogate parameter for atrial myopathy. However, although recent epidemiological studies found that among natriuretic peptides, NT-proBNP was strongly associated with incident AF, 32

| Prediction of electro-anatomical substrate using scores
There is a vast majority of different scores for the prediction of AF incidence, thromboembolic complications, AF recurrences, or mortality associated with AF. 36  In contrast, the ANP score included only variables, which pathophysiological are very specific to describe atrial myopathy and are associated with AF initiation and progression. The only limiting factor might be NT-proANP measurement, as it is not widely available, yet. However, our previous and current results demonstrate that NT-proANP seems to be a specific blood marker of "stressed" atria, 19,34 paving the way toward larger studies proving this hypothesis.

| Strengths and limitations
The strength of the present study is a deeply phenotyped AF cohort with available clinical (age, sex, persistent AF, BMI), imaging (echocardiography, CMR), peri-interventional (electro-anatomical mapping) data, and blood samples of patients undergoing AF ablation. Despite the advantage of such unique phenotyping, this could be considered as a disadvantage, because validation of our results in an external cohort is problematicnot all ablation centers analyze electro-anatomical mapping or have available blood biobanking. Therefore, missing validation of the novel score is the major limitation of the present study.
Also, since quantitative measurements (eg, LVAs area or indexed LVAs area) were not performed in this cohort, a correlation analyses between LVAs areas' size and NT-proANP concentration was not possible.
Another limitation is not consistently obtained follow-up data as many patients were followed by general practitioners or cardiologists in the outpatient setting. Also, compared to our previous studies, there were significant differences in follow-up management among those patients, who presented to our clinic after 2014. 37 However, rhythm outcomes follow-up is of clinical interest and should be addressed in future randomized trials.

| CONCLUSION
The novel biomarker-based ANP score demonstrated good prediction of LVAs. Compared to other established AF substrate scores, the ANP score includes parameters associated with atrial myopathy and demonstrates better specificity.