ORIGINAL RESEARCH published: 29 November 2019 doi: 10.3389/fneur.2019.01226 N-Terminal Pro B-Type Natriuretic Peptide’s Usefulness for Paroxysmal Atrial Fibrillation Detection Among Populations Carrying Cardiovascular Risk Factors Elena Palà 1 , Alejandro Bustamante 1,2*, Josep Lluis Clúa-Espuny 3,4 , Juan Acosta 5 , Felipe Gonzalez-Loyola 4,6 , Juan Ballesta-Ors 3,4 , Natalia Gill 1 , Andrea Caballero 7 , Jorge Pagola 2 , Alonso Pedrote 5 , Miguel Angel Muñoz 4,6 and Joan Montaner 1 1 Neurovascular Research Laboratory, Vall d’Hebron Institute of Research (VHIR)-Universitat Autónoma de Barcelona, Barcelona, Spain, 2 Department of Neurology, Hospital Universitari Vall d’Hebron, Barcelona, Spain, 3 Equip d’Atenció Primària Tortosa Est, SAP Terres de l’Ebre, Institut Català de la Salut, Tortosa, Spain, 4 Unitat de Suport a la Recerca de Barcelona, Institut d’Investigació en Atenció Primària IDIAP Jordi Gol, Barcelona, Spain, 5 Department of Cardiology, Hospital Universitario Virgen del Rocio, Sevilla, Spain, 6 Gerència Atenció Primària de Barcelona, Institut Català de la Salut, Barcelona, Spain, 7 Biochemical Department, Hospital Universitari Vall d’Hebron, Barcelona, Spain Edited by: Tatjana Rundek, University of Miami, United States Background: Atrial fibrillation (AF) systematic screening studies have not shown a clear Reviewed by: usefulness in stroke prevention, as AF might present as paroxysmal and asymptomatic. Dana Leifer, This study aims to determine the usefulness of some blood-biomarkers to identify Cornell University, United States paroxysmal atrial fibrillation in the context of a screening programme. Aristeidis H. Katsanos, McMaster University, Canada Methods: A total of 100 subjects aged 65–75 years with hypertension and diabetes *Correspondence: were randomly selected. AF was assessed by conventional electrocardiogram (ECG) Alejandro Bustamante alejandro.bustamante@vhir.org and 4 weeks monitoring with a wearable Holter device (NuuboTM ). N-terminal pro B-type natriuretic peptide (NT-proBNP), apolipoprotein CIII (ApoC-III), von Willebrand Specialty section: factor (vWF), ADAMTS13, urokinase plasminogen activator surface receptor (uPAR), and This article was submitted to Stroke, urokinase plasminogen activator (uPA) were determined in serum/plasma samples and a section of the journal the levels were compared depending on AF presence and mode of detection. Frontiers in Neurology Results: The AF prevalence in the studied population was found to be 20%. In Received: 16 May 2019 Accepted: 04 November 2019 seven subjects, AF was only detected after 1 month of Holter monitoring (hAF group). Published: 29 November 2019 NT-proBNP levels were higher in subjects with AF compared with subjects with no Citation: AF (p < 0.0001), even when only taking into account the hAF group (p = 0.031). No Palà E, Bustamante A, significant differences were found in the other biomarkers. The NT-proBNP >95 pg/ml Clúa-Espuny JL, Acosta J, Gonzalez-Loyola F, Ballesta-Ors J, cut-off showed high sensitivity and specificity to detect AF (95%, 66.2%) or hAF (85.72%, Gill N, Caballero A, Pagola J, 66.2%) and was found to be an independent predictor of AF and hAF in a logistic Pedrote A, Muñoz MA and Montaner J (2019) N-Terminal Pro regression analysis. NT-proBNP correlated with AF burden (r = 0.597, p = 0.024). B-Type Natriuretic Peptide’s Conclusion: NT-proBNP was elevated in AF cases not identified by ECG; thus, it may be Usefulness for Paroxysmal Atrial Fibrillation Detection Among used as a screening biomarker in asymptomatic high-risk populations, with a promising Populations Carrying Cardiovascular cut-off point of 95 pg/ml that requires further validation. Risk Factors. Front. Neurol. 10:1226. doi: 10.3389/fneur.2019.01226 Keywords: stroke, atrial fibrillation, biomarker, screening, NT-proBNP Frontiers in Neurology | www.frontiersin.org 1 November 2019 | Volume 10 | Article 1226 Palà et al. NT-proBNP in Paroxysmal Atrial Fibrillation INTRODUCTION de l’Ebre]. Physicians from these primary care centers were informed about the study protocol, and patients were invited by Atrial fibrillation (AF) is one of the fastest-growing them to participate in the study until 50 patients were recruited cardiovascular epidemics of the twenty-first century (1). in each catchment. Participants were ambulatory patients in a This heart rhythm disorder multiplies by five the risk of ischemic basal situation, meaning they were contacted at home to arrange stroke (2), and its prevalence in the population over 60 years has a visit in their primary care center specifically for the study. been quantified at 10.9% (3). Anticoagulation may reduce stroke During the appointments, patients received a comprehensive risk in AF individuals by ∼60% (4), but unfortunately, given assessment consisting of clinical characteristics (demographic its paroxysmal and asymptomatic nature at earlier stages, AF factors, vascular risk factors, medications, comorbidities, and is often undiagnosed and/or undertreated when stroke occurs. vitals), electrographic assessment and a blood sample collection. In a previous study, the absolute prevalence of undiagnosed AF Comorbid conditions were defined using standard outpatient in individuals over 60 years was 3.4%, representing up to 31% and inpatient ICD-10 codes by electronic data capture, including of the overall AF prevalence in our area (5, 6). International pharmacy records, laboratory data, and emergency room collaborations (7, 8) recommend opportunistic screening for and hospitalization diagnoses across all primary care centers AF in patients >65 years as a strategy to reduce stroke and and hospitals. death, but systematic screening programmes have failed to Individuals with chronic inflammatory diseases, cancer or show a clear benefit and those strategies are lacking in high-risk dementia were excluded. AF was assessed by conventional populations (9–11). electrocardiogram (ECG). Moreover, patients were monitored Blood biomarkers might represent an alternative tool for AF for 4 weeks with a wearable Holter device (NuuboTM ) as screening, because some of them might have elevated circulating described previously (21). The Holter device was composed of levels after AF paroxysms, a kind of “biological memory” offering a sensor that captured the electrocardiographic signal by non- a temporal advantage over electrical detection of the paroxysm invasive textile electrodes and a recorder of the signal that stored itself. N-terminal pro B-type natriuretic peptide (NT-proBNP) the information for subsequent detailed analysis. Participants has been found to be elevated in cardioembolic (CE) stroke were instructed by local trained researchers to wear the Holter (12) and to be associated with AF (13–16). Von Willebrand for 23 h and recharge it for 1 h daily. After 4 weeks, individuals factor (vWF) was found to be elevated in non-valvular AF (17) brought the Holter to their primary care center, and data were and, together with apolipoprotein CIII (ApoC-III), was found collected. Holter records, anonymized and encrypted, were sent to be predictor of stroke CE etiology [unpublished results from together with the ECG for blinded reading to the Rhythm the Stroke-Chip study (18)]. ADAMTS-13 is the key regulator Disorders Unit at the Cardiology Department of Hospital Virgen of vWF, and it has been associated with AF recurrence after del Rocio in Seville, to verify AF episodes. AF was defined cardioversion (19). Urokinase plasminogen activator surface following AHA guidelines as irregular R-R intervals without a receptor (uPAR) has been associated with AF prevalence (20) P wave signal, lasting for more than 60 s (22). Individuals with while its ligand, urokinase plasminogen activator (uPA), has not previous AF were not excluded at this step, but Holter monitoring been studied in AF patients. was optional in this subgroup. This study aims to determine the usefulness of the Patients were classified and compared in five groups defined aforementioned biomarkers to identify AF individuals in the by past medical history (PMH) for AF, ECG findings and context of a screening programme, especially in cases of Holter AF detection as follows: no AF, PMH-ECG-Holter+ paroxysmal AF that would not be detected using a regular 12- [Holter-detected AF (hAF)], PMH-ECG+, PMH+ECG-, and lead ECG. PMH+ECG+. Specific comparisons were done between the hAF and no AF groups. MATERIALS AND METHODS The AFRICAT study protocol was approved by the clinical research ethics committees of IDIAP Jordi Gol (P15/047) and Patients and Protocol Hospital Universitari Vall d’Hebron [PR (AG) 133-2015]. All AFRICAT (Atrial Fibrillation Research in CATalonia; participants signed informed consent before inclusion. The NCT03188484) is an observational, multicenter, population- study protocol conformed to the ethical guidelines of the 1975 based study with the aim of developing a screening programme Declaration of Helsinki. to detect new cases of AF in high-risk individuals in primary care centers. In Phase I of the project, 100 subjects aged 65–75 Biomarker Measurement with hypertension and diabetes were selected from primary care Blood was collected into EDTA and serum tubes at the registers (e-cap), which include the clinical records from most time of inclusion. After centrifugation at 1,500 g and 4◦ C for Catalonia inhabitants between January 2016 and December 2017. 15 min, plasma and serum aliquots were frozen at −80◦ C until In two different health service areas, Servei d’Atenció Primària biomarker determination. (SAP) Muntanya and SAP Terres de l’Ebre, patients fulfilling the Plasma NT-proBNP levels were determined by automated inclusion criteria were identified in primary care centers from immunoassay in a COBAS c8000 (Roche Diagnostics); serum each catchment [Centre d’Atenció Primària (CAP) Sant Rafael ApoC-III, plasma uPAR and plasma uPA by ELISA (Abnova, and CAP Trinitat Vella in SAP Muntanya; Equip d’Atenció R&D Systems, and Cloud-Clone); and plasma vWF and Primària (EAP) Tortosa Est and EAP Tortosa Oest in SAP Terres ADAMTS13 by a magnetic Luminex R assay (R&D Systems). Frontiers in Neurology | www.frontiersin.org 2 November 2019 | Volume 10 | Article 1226 Palà et al. NT-proBNP in Paroxysmal Atrial Fibrillation All assays were performed blinded to clinical information and (p < 0.0001) were more common in AF individuals. Of according to the manufacturer’s instructions. All samples were the 9 anticoagulated individuals, only one was anticoagulated tested in duplicate and inter-assay variation was determined by a for reasons other than AF (ischemic cardiopathy and deep commercial control (Human Serum, male AB, USA origin from venous thrombosis). The CHA2DS2-VASc score was higher in clotted, SIGMA, ref number H16914; Human plasma K2 EDTA, AF individuals (p = 0.035). Coronary heart disease ischemia Innovative Research, ref number IPLA-N) tested in duplicate was more common in hAF individuals compared to no AF in each plate. When inter-assay variation was >20% biomarker (p = 0.032). No cases of heart failure, valvular disease or levels were standardized using the Z-score. anticoagulation were present in this subgroup. Of the included 100 individuals, 96 were monitored with Statistical Analysis ECG Holter. All patients with negative ECG were monitored. Statistical analysis was conducted with SPSS version 20. Graphs Median monitoring time was 457 h. Six cutaneous adverse events were elaborated with GraphPad Prism 6. Data are expressed related to wearing the Holter were noted; these consisted of a as number (%) for categorical variables and as mean ± SD or rash that normally disappeared with topical steroids. However, median (interquartile range) for continuous variables, depending in two subjects, these adverse events prevented the monitoring on the data distribution. For univariate analysis, the Mann– being continued. Whitney U-test or Student’s t-test was used for continuous Of the 100 selected subjects, AF was present in 20 (11 newly variables, and the χ2 test was used for categorical variables. detected within the present study) classified as follows: 7 PMH- ANOVA or the Kruskal–Wallis test was used to compare ECG-Holter+ (hAF), 4 PMH-ECG+, 2 PMH+ECG-, and 7 >2 variables depending on variable distribution. Receiver PMH+ECG+ (Figure 1). operator characteristic (ROC) curves were configured to calculate In 14 individuals (11 newly detected and 3 PMH+ECG+ sensitivity, specificity, and cut-off values. The Spearman test was were monitored), AF was detected during Holter monitoring and used for correlations. Comparisons were first performed between AF burden was calculated. AF burden was defined as minutes AF patients and no AF patients and second between hAF and being in AF divided by the total minutes of readable records no AF patients (excluding other subtypes of AF). Binary logistic and is expressed as a percentage. The median AF burden was regression analysis was performed including variables associated 10.35% (1.00–56.50). with AF or hAF in univariate analysis. RESULTS Biomarker Analysis NT-proBNP, ApoC-III, sUPAR, vWF, and ADAMTS13 inter- and Patient Characteristics and AF Detection intra-assay variation was acceptable (coefficient of variation<20); Clinical characteristics of the cohort can be found in Table 1. hence, all the samples were included in the analysis. uPA levels Coronary heart disease (p = 0.008), heart failure (p = 0.007), were standardized because the inter-assay variation was higher valvular diseases (p = 0.024), and use of anticoagulants than 20%, and 9 samples were eliminated from statistical analysis TABLE 1 | Clinical characteristics of the cohort and comparison according to atrial fibrillation diagnosis. All patients (100) AF (20) No AF (80) hAF (7) P-valuea P-valueb Age 70 (68–73) 69 (66–71.5) 70 (68–73) 70 (65–72) 0.273 0.655 Sex (% female) 33 (33) 7 (35) 26 (32.5) 3 (42.9) 0.832 0.682 Tobacco 20 (20) 4 (20) 16 (20.3) 1 (14.3) 1.000 1.000 Alcohol 11 (11) 1 (5) 10 (12.5) 0 (0) 0.456 1.000 Dyslipidaemia 81 (81) 16 (80) 65 (81.3) 5 (71.4) 1.000 0.619 Coronary heart disease 18 (18) 8 (40) 10 (12.5) 3 (42.9) 0.008* 0.032* Heart failure 3 (3) 3 (15) 0 (0) 0 (0) 0.007* 1.000 Valvular disease 4 (4) 3 (15) 1 (1.3) 0 (0) 0.024* 0.767 Previous stroke 6 (6) 2 (10) 4 (5) 0 (0) 0.597 1.000 Anticoagulation 9 (9) 8 (40) 1 (1.3) 0 (0) 0.000* 0.920 Antiplatelets 50 (50) 7 (35) 43 (53.8) 3 (42.9) 0.134 0.702 Familiar history FA 5 (5) 2 (10) 3 (3.9) 0 (0) 0.273 1.000 SBP, mm Hg 143.50 (134–153) 140.5 (127.5–162.5) 144 (134–151.75) 140 (124–168) 0.973 0.773 DBP, mm Hg 78.95 ± 10.09 80.95 ± 10.85 78.45 ± 9.90 79.14 ± 6.66 0.332 0.857 Heart rate 77.94 ± 15.70 77.25 ± 16.94 78.11 ± 15.48 73 ± 22.14 0.724 0.421 CHA2 DS2 -VASc 4 (3–4) 4 (3–4) 4 (3–5) 4 (3–5) 0.035* 0.284 a P-value comparison AF vs. no AF. b P-value comparison hAF vs no AF. *P < 0.05. AF, atrial fibrillation; DBP, diastolic blood pressure; hAF, Holter-detected atrial fibrillation; SBP, systolic blood pressure. Frontiers in Neurology | www.frontiersin.org 3 November 2019 | Volume 10 | Article 1226 Palà et al. NT-proBNP in Paroxysmal Atrial Fibrillation FIGURE 1 | Subject classification. Individuals were classified depending on AF presence and subgroups were made depending on AF detection with different methods and previous medical history. TABLE 2 | Biomarker levels and comparisons between different groups. AF (20) No AF (80) hAF (7) P-valuea P-valueb NT-proBNP (pg/ml) 643.65 64.28 (37.08–133.10) 128.30 <0.0001* 0.03* (155.72–1339.25) (97.69–191.8) ApoC-III (ng/ml) 107,594 104,450 (766,64–130,393) 105,352 0.496 0.574 (92,501–129262.06) (91,956–132,268) ADAMTS13 (ng/ml) 1131.24 ± 471.419 1221.41 ± 459.80 1271.68 ± 561.01 0.437 0.786 vWF (ng/ml) 0.139 ± 0.052 0.139 ± 0.706 0.146 ± 0.060 0.987 0.802 uPA (std)c 2.01 (1.41–2.39) 1.60 (1.38–2.35) 1.68 (1.29–2.35) 0.720 0.868 uPAR (pg/ml) 2220.80 2267.54 (1518.33–2763.05) 2246.12 0.689 0.918 (557.74–2810.52) (1926.03–259.23) a P-value comparison AF vs. no AF. b P-value comparison hAF vs. no AF. c Ninety-one individuals included in the uPA analysis (73 no AF and 18 AF, of whom 6 were hAF). * P < 0.05. AF, atrial fibrillation; ApoC-III, Apolipoprotein C-III; hAF, Holter-detected group; NT-proBNP, N-terminal pro B-type natriuretic peptide; vWF, von Willebrand factor; uPA, urokinase plasminogen activator; uPAR, urokinase plasminogen activator surface receptor. because their duplicates showed CV > 20%. Median levels and regression analysis performed in the whole cohort. CHA2DS2- comparisons are shown in Table 2. VASc was an independent AF predictor when considered alone in the regression analysis (OR = 2.29; 95% CI 1.16–4.44; p = 0.017). However, after the addition of NT-proBNP>95 pg/ml, AF vs. No AF CHA2DS2-VASc was not significant in the model. NT-proBNP was significantly higher in individuals with AF Alternatively, the cut-off point of NT-proBNP of >125 pg/ml, compared with no AF [643.65 pg/ml (IQR 155.72–1339.25) which is recommended for heart failure diagnosis in the non- vs. 64.28 pg/ml (IQR 37.08–133.10), p < 0.0001; Figure 2A], acute setting, showed 85% sensitivity and 73.8% specificity, 44.7% while the remaining biomarkers were not different between PPV and 95.2% NPV. NT-proBNP>125 pg/ml was also an the two groups (Supplementary Figures 1, 2). The NT-proBNP independent predictor of AF (OR = 8.85; 95% CI, 2.17–35.95; distribution across the different subgroups is shown in Figure 2B. p = 0.017). In addition, AF burden was correlated with NT-proBNP (r = Sensitivity analysis was performed while excluding the patient 0.597, p = 0.024; Figure 2D). The discriminating ability (area who was anticoagulated for reasons other than AF, and the results under the ROC curve) of NT-proBNP was 0.900 (95% CI, 0.8274– were similar. 0.9726, p < 0.0001) to detect any AF (Figure 3A). The cut-off point of NT-proBNP of >95 pg/ml showed 95% sensitivity, 66.2% specificity, 41.3% positive predictive value (PPV), and 98.1% hAF vs. No AF negative predictive value (NPV) to detect any AF (Figure 3C). NT-proBNP levels were different between no AF and hAF [64.28 Anticoagulants (OR = 18.90; 95% CI, 2.10–169.83; p = 0.09) pg/ml (IQR 37.08–133.10) vs. 128.3 pg/ml (IQR 97.69–191.3), and NT-proBNP >95 pg/ml (OR = 22.42; 95% CI, 2.74–183.15; p p = 0.031; Figure 2C] while other biomarkers did not show = 0.04) were the only independent predictors of AF in the logistic differences (Supplementary Figures 1, 2). Patients without AF Frontiers in Neurology | www.frontiersin.org 4 November 2019 | Volume 10 | Article 1226 Palà et al. NT-proBNP in Paroxysmal Atrial Fibrillation FIGURE 2 | Plasma NT-proBNP level distribution. (A) Boxplot distribution between AF and no AF. (B) Boxplot distribution between different groups depending on past medical history (PMH) for AF, ECG findings, and Holter monitoring. (C) Boxplot distribution between Holter-detected AF (hAF) and no AF. (D) Correlation between NT-proBNP levels and AF burden. *P < 0.05. and with a Holter monitoring time below the median (457 h) higher than reported in previous screening studies (from 2.3 to were removed and all aforementioned comparisons were tested 12.3%) (10, 11). Most such studies have only used ECG or pulse again, with similar results (data not shown). The discriminating devices instead of long-term monitoring, so they have missed ability (area under the ROC curve) of NT-proBNP was 0.7464 cases of paroxysmal AF. Moreover, we used a targeted screening (95% CI 0.6087–0.8841, p = 0.031) to detect hAF (Figure 3B). approach, focussing on a high-risk population with advanced The cut-off point of NT-proBNP of >95 pg/ml showed 85.7 age and other conditions related to AF, such as hypertension % sensitivity, 66.2% specificity, 18.2% PPV, and 98.1% NPV to and diabetes, who should present a higher AF prevalence (23). detect hAF (Figure 3C). Longer monitoring in a high-risk population might therefore be The logistic regression analysis performed on this subgroup a good strategy to further explore in larger cohorts. Other studies confirmed that NT-proBNP ≥95 pg/ml was an independent using long-term monitoring with insertable cardiac devices or predictor of Holter-detected atrial fibrillation (OR, 11.778; pacemakers in patients without prior history of AF have also 95% CI, 1.35–102.85; p = 0.026). No clinical variable was an shown a higher prevalence of AF (24). In fact, in the recent independent predictor of AF detection. REVEAL AF trial, detection rates of AF rose with the monitoring Alternatively, the cut-off point of NT-proBNP of >125 time (25). Moreover, in patients with cryptogenic stroke, long- pg/ml showed 55.1% sensitivity, 73.8% specificity, 16% PPV, and term monitoring was found to be associated with higher rates of 95.2% NPV. NT-proBNP >125 pg/ml also was an independent AF detection, with a significant impact in secondary prevention predictor of AF (OR = 8.85; 95% CI 2.17–35.95; p = 0.017). NT- (26). However, although it could become the gold standard proBNP >125 pg/ml was not an independent predictor of hAF. for AF detection, long-term monitoring devices are expensive and unsuitable for application in primary care. In that sense, DISCUSSION biomarkers might be used as cost-effective alternatives for AF systematic screening. The prevalence of AF in our population was 20%, of which The present study suggests that NT-proBNP might be used 55% were newly detected AF individuals. This prevalence is as a screening biomarker to detect even paroxysmal AF in Frontiers in Neurology | www.frontiersin.org 5 November 2019 | Volume 10 | Article 1226 Palà et al. NT-proBNP in Paroxysmal Atrial Fibrillation FIGURE 3 | NT-proBNP discrimination power. (A) ROC curve: ability to discriminate between AF and no AF. (B) ROC curve, ability to discriminate between Holter-detected AF and no AF. The cut-off with the best specificity and sensitivity to detect hAF is marked with a circle in the two curves (95 pg/ml), as it is the cut-off value that was used to calculate sensitivity, specificity and predictive values. (C) Discrimination power of the NT-proBNP cut-off value of 95 pg/ml. All AF individuals had NT-proBNP plasma levels below the previous cut-off (except for a patient with paroxysmal AF), and the majority of no AF patients had values above the cut-off. asymptomatic, high-risk individuals. The focus of the present pacemakers is that their use is restricted to very specific study was to assess the predictive ability of NT-proBNP and populations with cardiac diseases that could affect some other biomarkers to diagnose paroxysmal atrial fibrillation and biomarker levels. be used as the first step of a screening workflow. In fact, the In the pilot study of Seegers et al. (14), in addition to a best screening biomarker in this context would be able to detect limited 7-day Holter ECG for the evaluation of paroxysmal AF, those cases impossible to notice with other available tools, such patients were preselected as those with lowest and highest NT- opportunistic ECG or hand-held ECG devices. It should be proBNP quartiles, overestimating the discriminatory value of mentioned that paroxysmal AF (pAF) patients also can have the biomarker. In our study, inclusion was performed before increased risk of stroke (27, 28). Moreover, pAF may progress to any biomarker determination, and 4-week Holter monitoring permanent AF within a year in almost 20% of cases in the aging was used as the gold standard to assess the usefulness of population (29, 30). blood-based biomarkers in AF screening. Although individuals Previous studies have suggested NT-proBNP as a with AF detected by other methods were not excluded from screening tool (14–16), but a limited length of monitoring the whole study, to increase the statistical power, specific might have misclassified some cases of paroxysmal comparisons were performed while excluding these patients and atrial fibrillation. Our study overcame this limitation by only taking into account AF cases detected with long-term recording patient’s heart rhythm continuously for 30 days monitoring, i.e., cases with neither previous AF history nor using a Holter monitor, allowing us to detect more AF detection during electrocardiographic assessment. Even though cases. This is to our knowledge the longest period over few patients were part of this group, specific comparisons which asymptomatic individuals have been monitored showed the usefulness of NT-proBNP to identify those patients to detect atrial fibrillation in a biomarker study, if we outside AF paroxysms. After validating those results with exclude studies using pacemakers (13). The limitation of a larger cohort, NT-proBNP could be implemented as part Frontiers in Neurology | www.frontiersin.org 6 November 2019 | Volume 10 | Article 1226 Palà et al. NT-proBNP in Paroxysmal Atrial Fibrillation of a screening programme to indicate long-term monitoring results from clinical trials of embolic strokes of undetermined in patients with a risk biological profile. Such a protocol source (ESUS) patients receiving non-vitamin K antagonists would identify patients at risk of AF who would benefit from (NOACs) (36, 37) indicate that this is a heterogeneous group anticoagulation, thereby improving primary stroke prevention of patients who need different management, only a minority of by preventing cardioembolic stroke, which is associated with whom have undetected paroxysmal AF (38). In fact, the ongoing the poorest disability and mortality rates among stroke ARCADIA trial (39) will test the use of anticoagulants in ESUS etiologies (31). patients with atrial cardiopathy using high NT-proBNP (>250 Additionally, we proposed for the first time a specific cut-off pg/ml) as one of the inclusion criteria. point for atrial fibrillation screening. Seegers et al. (14) suggested Our study has some limitations. First, the participant selection that NT-proBNP could be used for stroke prevention after finding was not randomized and might have suffered some selection increased values in AF individuals, without indicating any cut- bias: the high rates of AF might not be attributed only to off. Svenberg et al. (15) and Ghazal et al. (16) proposed a cut-off the selection of high-risk patients and long-term monitoring point of NT-proBNP >125 pg/ml, which is the recommended for but also to patients with more cardiac comorbidities being heart failure in the non-acute setting (32). Here, NT-proBNP >95 more prone to participate. The small sample size and the pg/ml reached high sensitivity (95%) even to detect paroxysmal small number of patients with AF suggest that these results AF individuals (85.7%). In fact, in our cohort, this cut-off point should be interpreted as hypothesis-generating. The protocol only misclassified one paroxysmal AF individual. Although the of the study did not include an echocardiography, missing the NT-proBNP biomarker is not specific for AF and might be opportunity to establish correlations between biomarkers and elevated in other cardiac diseases and renal dysfunction (33), left atrial diameter. Finally, the included population presented a high sensitivity is more useful than specificity for screening specific risk factors, and the results may not be directly applicable purposes. NT-proBNP might be rapid and easy to use to identify to populations with other combinations of risk factors. Further all patients who need follow-up Holter monitoring since it is validation studies in larger cohorts are needed. already widely employed as a point-of-care analysis tool for heart failure. CONCLUSIONS Although it might be suspected that elevated NT-proBNP levels were due to a higher burden of heart failure and other NT-proBNP was found to be elevated in AF individuals cardiovascular diseases in the atrial fibrillation subjects of our compared to controls. This biomarker also detected cases of AF cohort, in the hAF group, there were few individuals with not previously known or detected by ECG. NT-proBNP may known comorbidities that could affect NT-proBNP levels. In fact, be useful as a screening biomarker in asymptomatic high-risk in this group, there were no cases of heart failure, and there populations with a promising cut-off point of 95 pg/ml that were only differences in coronary heart disease when compared requires further validation. A screening strategy based on NT- to no AF individuals. In addition, the study was carried out proBNP, alone or in combination with other biomarkers, might at the outpatient clinic, and therefore, no patient was in a be of interest for the early initiation of anticoagulants, which situation of acute decompensated heart failure. This, combined could reduce cardioembolic stroke consequences. with multivariate analysis, confirmed the independent value of NT-proBNP in diagnosing paroxysmal atrial fibrillation. ApoC-III, vWF, ADAMTS13, uPAR, and uPA were not DATA AVAILABILITY STATEMENT elevated in AF individuals in the present study. Although The raw data supporting the conclusions of this manuscript will associated with cardioembolic stroke or AF recurrence, these be made available by the authors, without undue reservation, to biomarkers might not be useful in the context of AF any qualified researcher. screening. However, it would be interesting to perform further discovery studies to find new biomarkers that might be used in combination with NT-proBNP, increasing its specificity ETHICS STATEMENT and sensitivity. Interestingly, the present study showed that a higher level of The studies involving human participants were reviewed and NT-proBNP was correlated with higher AF burden, a conclusion approved by Ethics Committee of Research Institute IDIAP Jordi based on continuous Holter monitoring. This tendency was Gol (P15/047/2015) and Hospital Universitari Vall d’Hebron also reflected in the distribution of NT-proBNP levels between Clinical Research Ethics Committee [PR(AG)133-2015]. The the groups, with higher levels in individuals with a previous patients/participants provided their written informed consent to medical history of AF compared with individuals with hAF, participate in this study. as shown in Figure 1B. In fact, some studies have suggested that AF burden is related to higher stroke risk (34), and other AUTHOR CONTRIBUTIONS biomarker-based models have confirmed NT-proBNP as a stroke risk surrogate (35). AB, JC-E, FG-L, JB-O, and MM recruited patients. JA and Although the present study was performed in asymptomatic AP read Holter-ECG to verify AF episodes. EP, NG, and AC patients, NT-proBNP might also be useful for detecting performed biomarker measurements. AB, JC-E, JP, MM, and JM cryptogenic stroke patients with occult AF. Recently, the negative planned the whole project and drafted the study protocol. EP Frontiers in Neurology | www.frontiersin.org 7 November 2019 | Volume 10 | Article 1226 Palà et al. NT-proBNP in Paroxysmal Atrial Fibrillation performed statistical analysis and drafted the manuscript. All research contract (JR16/00008) from Instituto de Salud Carlos III. authors have critically reviewed the manuscript and approved the EP has received a predoctoral grant from Vall D’Hebron Institute final article version. of Research. FUNDING SUPPLEMENTARY MATERIAL The study received a research grant by Fundació Marató de TV3 The Supplementary Material for this article can be found in the research call La Marató 2014: malalties del cor. Grant online at: https://www.frontiersin.org/articles/10.3389/fneur. number: 201528-30-31-3. AB was supported by a Juan Rodés 2019.01226/full#supplementary-material REFERENCES 16. Ghazal F, Theobald H, Rosenqvist M, Al-Khalili F. Assessment of N- terminal pro-brain natriuretic peptide level in the screening for atrial 1. Stevens E, Emmett E, Wang Y, McKevitt C, DA Wolfe C. The Burden of Stroke fibrillation in the primary health care. PLoS ONE. (2019) 14:e0212974. in Europe Report. King’s College London for the Stroke Alliance for Europe doi: 10.1371/journal.pone.0212974 (SAFE), Brussels (2017). 17. Muñoz-Cortés G, Viveros-Sandoval ME, Areán-Martínez C, Vega-Gómez 2. Wolf P, Dawber T, Thomas HJ, Kannel W. Epidemiologic assessment of HE, López-Castañeda SE, et al. Von Willebrand factor plasma levels chronic atrial fibrillation and risk of stroke: the Framingham study. Neurology. variability in nonvalvular atrial fibrillation. J Atr Fibrillation. (2014) 7:1124. (1978) 28:973–7. doi: 10.1212/WNL.28.10.973 doi: 10.4022/jafib.1124 3. Roso-llorach A, Clua-espuny JL, Panisello-tafalla A, Lucas-noll J, Lo C, 18. Bustamante A, López-Cancio E, Pich S, Penalba A, Giralt D, García- Gallofre M, et al. Prevalence of undiagnosed atrial fibrillation and of that not Berrocoso T, et al. Blood biomarkers for the early diagnosis of stroke: the being treated with anticoagulant drugs : the AFABE study. Rev Esp Cardiol. stroke-chip study. Stroke. (2017) 48:2419–25. doi: 10.1161/STROKEAHA.117. (2013) 66:545–52. doi: 10.1016/j.rec.2013.03.003 017076 4. Hart RG, Pearce LA, Aguilar M. Meta-analysis: antithrombotic therapy to 19. Freynhofer MK, Bruno V, Jarai R, Gruber S, Höchtl T, Brozovic I, et al. prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Levels of von Willebrand factor and ADAMTS13 determine clinical outcome Med. (2007) 146:857–67. doi: 10.7326/0003-4819-146-12-200706190-00007 after cardioversion for atrial fibrillation. Thromb Haemost. (2011) 105:435–43. 5. Panisello-tafalla A, Clua-espuny JL, Gil-guillen VF, González-henares A, doi: 10.1160/TH10-09-0615 Queralt-tomas ML, López-pablo C, et al. Results from the registry of 20. Ichihara N, Miyamura M, Maeda D, Fujisaka T, Fujita SI, Morita H, atrial fibrillation (AFABE): gap between undiagnosed and registered atrial et al. Association between serum soluble urokinase-type plasminogen fibrillation in adults — ineffectiveness of oral anticoagulation treatment with activator receptor and atrial fibrillation. J Arrhythmia. (2017) 33:469–74. VKA. Biomed Res Int. (2015) 2015:134756. doi: 10.1155/2015/134756 doi: 10.1016/j.joa.2017.05.003 6. Morillas P, Pallarés V, Llisterri L, Sanchis C, Sánchez T. Prevalence 21. Pagola J, Juega J, Francisco-Pascual J, Moya A, Sanchis M, Bustamante A, et al. of atrial fibrillation and use of antithrombotics in hypertensive patients Yield of atrial fibrillation detection with Textile Wearable Holter from the aged ≥ 65 years. The FAPRES Trial. Rev Esp Cardiol. (2010) 63:943–50. acute phase of stroke: pilot study of Crypto-AF registry. Int J Cardiol. (2018) doi: 10.1016/S1885-5857(10)70188-2 251:45–50. doi: 10.1016/j.ijcard.2017.10.063 7. Kirchhof P, Benussi S, Kotecha D, Ahlsson A, Atar D, Casadei B, 22. January CT, Wann LS, Alpert JS, Calkins H, Cigarroa JE, Cleveland JC et al. 2016 ESC guidelines for the management of atrial fibrillation Jr, et al. 2014 AHA/ACC/HRS guideline for the management of patients developed in collaboration with EACTS. Rev Esp Cardiol. (2017) 70:50. with atrial fibrillation: executive summary: a report of the American doi: 10.1016/j.rec.2016.11.033 College of Cardiology/American Heart Association Task Force on practice 8. Freedman B. Screening for atrial fibrillation: a report of the AF- guidelines and the Heart Rhythm Society. Circulation. (2014) 130:2071–04. SCREEN international collaboration. Circulation. (2017) 135:1851–67. doi: 10.1161/CIR.0000000000000040 doi: 10.1161/CIRCULATIONAHA.116.026693 23. Lau DH, Nattel S, Kalman JM, Sanders P. Modifiable risk 9. Moran PS, Teljeur C, Ryan M, Smith SM. Systematic screening for factors and atrial fibrillation. Circulation. (2017) 136:583–96. the detection of atrial fibrillation. Cochrane Database Syst Rev. (2016) doi: 10.1161/CIRCULATIONAHA.116.023163 2016:CD009586. doi: 10.1002/14651858.CD009586.pub3 24. Van Gelder IC, Healey JS, Crijns HJGM, Wang J, Hohnloser SH, 10. Svennberg E, Engdahl J, Al-Khalili F, Friberg L, Frykman V, Gold MR, et al. Duration of device-detected subclinical atrial fibrillation Rosenqvist M. Mass screening for untreated atrial fibrillation and occurrence of stroke in ASSERT. Eur Heart J. (2017) 38:1339–44. the STROKESTOP study. Circulation. (2015) 131:2176–84. doi: 10.1093/eurheartj/ehx042 doi: 10.1161/CIRCULATIONAHA.114.014343 25. Reiffel JA, Verma A, Kowey PR, Halperin JL, Gersh BJ, Wachter R, et al. 11. Lowres N, Neubeck L, Redfern J, Ben Freedman S. Screening to identify Incidence of previously undiagnosed atrial fibrillation using insertable cardiac unknown atrial fibrillation: a systematic review. Thromb Haemost. (2013) monitors in a high-risk population: the REVEAL AF study. JAMA Cardiol. 110:213–22. doi: 10.1160/TH13-02-0165 (2017) 2:1120–7. doi: 10.1001/jamacardio.2017.3180 12. Llombart V, Antolin-Fontes A, Bustamante A, Giralt D, Rost NS, Furie K, et al. 26. Tsivgoulis G, Katsanos AH, Mac GB, Köhrmann M, Ricci BA, Tsioufis K, B-type natriuretic peptides help in cardioembolic stroke diagnosis. Stroke. et al. Prolonged cardiac rhythm monitoring and secondary stroke prevention (2015) 46:1187–95. doi: 10.1161/STROKEAHA.114.008311 in patients with cryptogenic cerebral ischemia. Stroke. (2019) 50:2175–80. 13. Wakula P, Neumann B, Kienemund J, Thon-Gutschi E, Stojakovic T, doi: 10.1161/STROKEAHA.119.025169 Manninger M, et al. CHA2DS2-VASc score and blood biomarkers to identify 27. Friberg L, Hammar N, Rosenqvist M. Stroke in paroxysmal atrial fibrillation: patients with atrial high-rate episodes and paroxysmal atrial fibrillation. report from the Stockholm Cohort of Atrial Fibrillation. Eur Heart J. (2010) Europace. (2017) 19:544–51. doi: 10.1093/europace/euw101 31:967–75. doi: 10.1093/eurheartj/ehn599 14. Seegers J, Zabel M, Grüter T, Ammermann A, Weber-Krüger M, Edelmann 28. Lilli A, Di Cori A, Zacà V. Thromboembolic risk and effect of oral F, et al. Natriuretic peptides for the detection of paroxysmal atrial fibrillation. anticoagulation according to atrial fibrillation patterns: a systematic Open Hear. (2015) 2:e000182. doi: 10.1136/openhrt-2014-000182 review and meta-analysis. Clin Cardiol. (2017) 40:641–7. doi: 10.1002/clc. 15. Svennberg E, Henriksson P, Engdahl J, Hijazi Z, Al-Khalili F, Friberg L, et al. 22701 N-terminal pro B-type natriuretic peptide in systematic screening for atrial 29. Proietti R, Hadjis A, Alturki A, Thanassoulis G, Roux JF, Verma A, et al. fibrillation. Heart. (2017) 103:1271–7. doi: 10.1136/heartjnl-2016-310236 A systematic review on the progression of paroxysmal to persistent atrial Frontiers in Neurology | www.frontiersin.org 8 November 2019 | Volume 10 | Article 1226 Palà et al. NT-proBNP in Paroxysmal Atrial Fibrillation fibrillation: Shedding new light on the effects of catheter ablation. JACC Clin 37. Hart RG, Catanese L, Perera KS, Ntaios G, Connolly SJ. Embolic Electrophysiol. (2015) 1:105–15. doi: 10.1016/j.jacep.2015.04.010 stroke of undetermined source: a systematic review and clinical 30. De Sisti A, Leclercq JF, Halimi F, Fiorello P, Bertrand C, Attuel P. Evaluation of update. Stroke. (2017) 48:867–72. doi: 10.1161/STROKEAHA.116. time course and predicting factors of progression of paroxysmal or persistent 016414 atrial fibrillation to permanent atrial fibrillation. Pacing Clin Electrophysiol. 38. Tsivgoulis G, Katsanos AH, Köhrmann M, Caso V, Lemmens R, (2014) 37:345–55. doi: 10.1111/pace.12264 Tsioufis K, et al. Embolic strokes of undetermined source: theoretical 31. Freeman WD, Aguilar MI. Prevention of cardioembolic stroke. construct or useful clinical tool? Ther Adv Neurol Disord. (2019) 12:1–12. Neurotherapeutics. (2011) 8:488–502. doi: 10.1007/s13311-011-0048-y doi: 10.1177/1756286419851381 32. Ponikowski P, Voors AA, Anker SD, Bueno H, Cleland JGF, Coats AJS, et al. 39. Kamel H, Longstreth WT, Tirschwell DL, Kronmal RA, Broderick 2016 ESC Guidelines for the diagnosis and treatment of acute and chronic JP, Palesch YY, et al. The AtRial Cardiopathy and Antithrombotic heart failure. Eur Heart J. (2016) 37:2129–200. doi: 10.1093/eurheartj/ehw128 Drugs In prevention After cryptogenic stroke randomized trial: rationale 33. Troughton RW, Lewis LK, Yandle TG, Pemberton CJ, Nicholls MG. B-type and methods. Int J Stroke. (2019) 14:207–14. doi: 10.1177/17474930187 natriuretic peptides: Looking to the future. Ann Med. (2011) 43:188–97. 99981 doi: 10.3109/07853890.2010.526630 34. Chen LY, Chung MK, Allen LA, Ezekowitz M, Furie KL, McCabe P, et al. Conflict of Interest: The authors declare that the research was conducted in the Atrial fibrillation burden: moving beyond atrial fibrillation as a binary entity: absence of any commercial or financial relationships that could be construed as a a scientific statement from the american heart association. Circulation. (2018) potential conflict of interest. 137:e623–44. doi: 10.1161/CIR.0000000000000568 35. Hijazi Z, Oldgren J, Lindbäck J, Alexander JH, Connolly SJ, Eikelboom Copyright © 2019 Palà, Bustamante, Clúa-Espuny, Acosta, Gonzalez-Loyola, JW, et al. The novel biomarker-based ABC (age, biomarkers, clinical Ballesta-Ors, Gill, Caballero, Pagola, Pedrote, Muñoz and Montaner. This is an history)-bleeding risk score for patients with atrial fibrillation: open-access article distributed under the terms of the Creative Commons Attribution a derivation and validation study. Lancet. (2016) 387:2302–11. License (CC BY). The use, distribution or reproduction in other forums is permitted, doi: 10.1016/S0140-6736(16)00741-8 provided the original author(s) and the copyright owner(s) are credited and that the 36. Hart RG, Sharma M, Mundl H, Kasner SE, Bangdiwala SI, Berkowitz SD, original publication in this journal is cited, in accordance with accepted academic et al. Rivaroxaban for stroke prevention after embolic stroke of undetermined practice. No use, distribution or reproduction is permitted which does not comply source. N Engl J Med. (2018) 378:2191–201. doi: 10.1056/NEJMoa1802686 with these terms. Frontiers in Neurology | www.frontiersin.org 9 November 2019 | Volume 10 | Article 1226