- Email: [email protected]
Markers for silent atrial fibrillation in esophageal long-term electrocardiography
Available online at www.sciencedirect.com
ScienceDirect Journal of Electrocardiology 49 (2016) 496 – 503 www.jecgonline.com
Markers for silent atrial fibrillation in esophageal long-term electrocardiography Andreas Haeberlin, MD, PhD, a, b,⁎ Lucca Lacheta, MMed, a Thomas Niederhauser, PhD, b, c Thanks Marisa, MSc, b, c Reto A. Wildhaber, MD, MSc, c Josef Goette, PhD, c Marcel Jacomet, PhD, c Jens Seiler, MD, a Juerg Fuhrer, MD, a Laurent Roten, MD, a Hildegard Tanner, MD, a Rolf Vogel, MD, PhD, MSEE d a
Department of Cardiology, Bern University Hospital and University of Bern, Bern, Switzerland ARTORG Center for Biomedical Engineering Research, University of Bern, Bern, Switzerland Institute for Human Centered Engineering, Bern University of Applied Sciences, Biel, Switzerland d Department of Cardiology, Bürgerspital Solothurn, Solothurn, Switzerland b
c
Abstract
Purpose: Paroxysmal atrial fibrillation (PAF) often remains undiagnosed. Long-term surface ECG is used for screening, but has limitations. Esophageal ECG (eECG) allows recording high quality atrial signals, which were used to identify markers for PAF. Methods: In 50 patients (25 patients with PAF; 25 controls) an eECG and surface ECG was recorded simultaneously. Partially A-V blocked atrial runs (PBARs) were quantified, atrial signal duration in eECG was measured. Results: eECG revealed 1.8‰ of atrial premature beats in patients with known PAF to be PBARs with a median duration of 853 ms (interquartile range (IQR) 813-1836 ms) and a median atrial cycle length of 366 ms (IQR 282-432 ms). Even during a short recording duration of 2.1 h (IQR 1.2– 17.2 h), PBARs occurred in 20% of PAF patients but not in controls (p = 0.05). Left atrial signal duration was predictive for PAF (72% sensitivity, 80% specificity). Conclusions: eECG reveals partially blocked atrial runs and prolonged left atrial signal duration – two novel surrogate markers for PAF. © 2016 Elsevier Inc. All rights reserved.
Keywords:
atrial fibrillation; esophageal ECG; long-term ECG; surrogate markers
Background Paroxysmal atrial fibrillation (PAF) is common and may have severe consequences such as ischemic strokes [1]. PAF may remain clinically silent and undiagnosed if patients are asymptomatic [1]. Nevertheless, even silent PAF increases the risk for strokes [2,3]. Thus, a reliable screening tool for silent PAF is highly desirable in order to prevent adverse outcomes. Commonly, long-term ECG recordings are performed to screen for PAF [4]. Continuous long-term ECG recordings for several days outperforms 24-h ECG monitoring [5] or a discontinuous event triggered ECG monitoring [6]. However, continuous long-term recordings
⁎ Corresponding author at: Department of Cardiology, Inselspital, Bern University Hospital, 3010, Bern, Switzerland. E-mail address: [email protected] http://dx.doi.org/10.1016/j.jelectrocard.2016.03.006 0022-0736/© 2016 Elsevier Inc. All rights reserved.
increase the data volume to be reviewed, thus, the time needed for analysis may become high [7]. Furthermore, surface long-term ECG recordings very often cause patient discomfort due to skin irritation [6]. Complementary, ECG surrogate markers like frequent atrial premature beats (APBs) allow identifying patients with a high pretest probability for PAF [8], even if no PAF episode has been recorded in a long-term ECG yet. Patients with high PAF probability may then undergo repeated long-term ECG recordings to confirm presence of PAF. However, the performance of this approach is clearly limited by the fact that the atrial electrogram cannot be analyzed directly in conventional surface long-term ECG recording systems. Instead, R-R intervals are used as surrogate markers of the atrial activity [7]. As an alternative to conventional surface ECG monitoring, esophageal electrocardiography (eECG) may be used. eECG provides excellent atrial electrograms [9], which allow a precise and direct analysis of the atrial electric activity. The purpose of the present study
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
was to identify and validate eECG surrogate markers for PAF by comparing PAF patients with matched controls. Methods Study design In this investigator-initiated case–control study, we analyzed esophageal long-term ECG recordings for the presence of surrogate markers to detect PAF. For this purpose, we compared 25 patients with known PAF (but presenting in sinus rhythm) to 25 control patients 1:1 matched for age (± one year) and gender. The local ethics committee approved the study (according to the declaration of Helsinki 1964). Informed consent was obtained from all individual participants included in the study.
497
(BlueSensor VL®, Ambu, Denmark). The first sECG channel was derived between an electrode at the cranial sternal border to an electrode at the lower left border of the ribcage on the anterior axillary line. The latter in conjunction with an electrode at the lower right border of the ribcage on the mid-clavicular line defined the second channel. Synchronization of the eECG and sECG recordings was obtained using a two-step approach. Prior to the registration, the internal clocks of both devices were synchronized. After the registration, fine adjustment was done manually by synchronizing both ECG traces using clearly identifiable events on both traces (e. g. ventricular premature beats). Patients were asked to grade their subjective overall discomfort during lead insertion and one hour later. Possible answers were “no discomfort”, “low discomfort”, “moderate discomfort”, “strong discomfort” and “discomfort not tolerable”.
Study population The eECG recordings were acquired from patients aged ≥ 18 years presenting to the Department of Cardiology, Bern University Hospital. All “case” recordings were acquired from patients with known PAF referred to our clinic for any reason. The case patients were enrolled into the study while being in sinus rhythm on the standard 12-channel ECG. All “control” recordings were acquired from patients without known PAF referred to our clinic for any reason. PAF was ruled out by at least one negative continuous 7-day ECG within one year prior to inclusion. In addition, patients had no history suggestive of PAF. Exclusion criteria for cases and controls were: - Pacemaker/ICD with an atrial electrode - History of prior ablation of atrial fibrillation or other supraventricular tachyarrhythmias - Heart transplantation - Valve replacement less than 4 weeks ago - Carcinoma of the esophagus or nasopharynx - Acute myocardial infarction or hemodynamically unstable patients Pregnancy ECG registration All patients underwent eECG acquisition for several hours using an esophageal lead (Esosoft/Esoflex 6S®, FIAB, Italy), connected to a dedicated high-resolution eECG recorder [10]. The eECG recorder continuously acquires a bipolar eECG signal with a sampling frequency of 512 Hz and 24 bit resolution [10]. Patients carried the device similar to a hearing aid behind the ear in a small cushioned splash waterproof bag. The lead was inserted nasally, the lead insertion depth was chosen according to a previously described approach [9]. Simultaneously and during the same time period, a conventional surface ECG (sECG) was registered using a long-term ECG recorder (Lifecard CF®, Spacelabs Healthcare, USA). This device continuously records two bipolar sECG channels with 1024 Hz sampling frequency and 12 bit resolution using three dedicated long-term sECG electrodes
ECG processing and analysis sECG recordings were analyzed using the Pathfinder SL® software, version 1.7.1 (Spacelabs Healthcare, USA) without changing any default parameters. The presence and number of APBs and atrial runs (three or more beats with supraventricular origin, total episode duration less than 30 s [11]) were analyzed for cases and controls. Subsequently, the eECG recordings were analyzed (direct visual analysis) by the same person, an experienced eECG interpreter (physician), for the presence of potential surrogate markers for PAF. These a priori defined features comprise partially A-V blocked atrial runs (PBARs) and right and left atrial signal duration of the atrial eECG signal. The ECG interpreter was blinded with respect to the patient group (case/control) or other patient data (all ECG recordings were given numbers not corresponding to the patient database). The signal was filtered by a 0.5–200 Hz bandpass filter using custom-made eECG visualization software (Fig. 1). The presence of APBs and atrial runs – in particular PBARs – was analyzed. Furthermore, the duration of PBARs (from the first to the last atrial electrogram of each PBAR) and the cycle lengths (of the atrial electrograms of each PBAR) were measured. After long-term eECG analysis, the original eECG data were processed using custom-made software based on MATLAB® (MathWorks, USA) to calculate signalaveraged eECGs. No filters were applied. Detection of normal A-V-sequential (presumably sinus) beats was performed using a template matching algorithm (triggered by the atrial electrogram in the eECG). Subsequently, all these beats were averaged, resulting in a “typical mean waveform morphology” of the atrial signal for each patient (Fig. 2 shows an example). The averaged atrial eECG signal was analyzed for the presence of the potential surrogate markers specified above. “Right atrial signal duration” – appearing as right atrial “far-field signal” in the eECG – was defined to start at the beginning of the atrial eECG signal (point 1 in Fig. 2) and to last until the beginning of the main deflection of the atrial eECG signal (point 2 in Fig. 2). The interval from point 2 (where the slope of the
498
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
Fig. 1. Analysis window of the eECG visualization software showing an eECG sequence. At the top, the registration date and the acquisition time are displayed. The eECG is shown below (one continuous channel). In the first fully visible beat, the atrial (A) and ventricular (V) electrogram are labeled.
main deflection exceeded ± 45°) to the end of the atrial eECG signal (point 3 in Fig. 2) was defined to reflect the “left atrial signal duration”. Statistical analysis Statistical analysis was made using R version 3.0.1 for Windows. Continuous variables are expressed as median and interquartile range (IQR). Medians between the two paired patient groups were compared by a Wilcoxon signed-rank test. Non-parametric 95%-confidence intervals (CI's) are reported. Categorical variables are expressed as numbers and percentages, proportions were compared by Fisher's exact test.
Receiver operating characteristics (ROC) analysis was performed to assess sensitivity and specificity of ECG surrogate markers. Areas under the ROC curve and corresponding 95% bootstrap confidence intervals (CI's) for sensitivity and specificity were calculated. A P-value ≤ 0.05 was considered to be significant. Results Baseline patient characteristics We observed no differences between the case and control group regarding age or gender (matching variables), cardiovascular risk factors, disease etiology, echocardio-
Fig. 2. Signal-averaged atrial electrogram as derived from the eECG. The beginning of the right (1.) and the end of the left atrial signal (3.) are highlighted. 2.) represents the end of the right atrial and the beginning of the left atrial signal. The main deflection of the signal does not reflect the QRS-complex but the main activation of the left atrium.
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
graphic parameters or betablockers (Table 1). The rate of oral anticoagulation or class III antiarrhythmic agents (amiodarone and dronedarone) was higher in patients with known PAF. N = 5 (20%) of control patients received oral anticoagulation due to recurrent pulmonary embolism (n = 3) or low LV ejection fraction (n = 2). The PAF burden in the study population was low, just a single PAF episode was detected in one patient.
499
16 ms longer (95%-CI: 11–21 ms, p = 0.003). Furthermore, there was no difference in left atrial signal duration of cases with class III antiarrhythmic agents vs. cases without class III agents (p = 0.31). The predictive value of the total atrial signal duration was optimal at a cutoff of 114.6 ms, offering 76% sensitivity and 76% specificity for PAF (AUC 0.83, Fig. 5). Patient tolerance of long-term esophageal ECG
Surrogate markers for PAF in the long-term ECG During the recording period, patients with known PAF showed more APBs per hour in comparison to controls in sECG and eECG (4.9 (0.9–18.9) vs. 1.5 (0–5.1), p = 0.02), whereas the number of atrial runs was not different. Median ECG recording duration was not different between cases and controls (1.9 (IQR 1.1–15.6) vs. 3.1 (IQR 1.8–17.6) hours, p = 0.22) and identical for sECG/eECG recordings (simultaneous registration). The longest eECG recording duration was 49 h. In addition to the sECG analysis, the eECG revealed 68 of 38,757 (1.8‰) of the recorded APBs to be PBARs (Fig. 3). PBARs were found in five PAF patients (20%) but not in controls (p = 0.05). Median duration of PBARs was 853 ms (813–1836 ms) and median atrial cycle length of PBARs was 366 ms (282–432 ms). Detailed numerical data about incidence of atrial premature beats, atrial runs and PBARs can be found in Table 2. Surrogate markers for PAF in the signal-averaged esophageal ECG Median atrial signal duration was 16 ms (95%-CI: 9– 25 ms) longer in patients with known PAF vs. control patients (p b 0.0001). We attributed this difference to a longer median left atrial signal duration (16 ms, 95%-CI: 9– 22 ms, p b 0.001, Fig. 4). In contrast, median right atrial signal duration was not different (Fig. 4). The difference in left atrial signal duration was not caused by a higher rate of class III antiarrhythmic agents in patients with known PAF. Median left atrial signal duration of case patients without class III antiarrhythmic agents vs. control patients was also
During insertion, patients only experienced low (66%) or moderate discomfort (34%). One hour later, patient discomfort had decreased significantly (p = 0.01); most patients reported only low (92%) or even no (4%) discomfort. Only 4% of patients still suffered from moderate discomfort. However, no eECG registration had to be aborted prematurely. Discussion This is the first study evaluating the performance of long-term eECG surrogate markers for PAF. eECG offers high quality atrial signals, which allow direct analysis of the atrial electric activity, unlike sECG. PBARs and the duration of the left atrial activation are indicative for PAF. These surrogate markers may be used to identify patients with a high likelihood of PAF. The role of partially blocked atrial runs and atrial premature beats In our study, we observed more APBs in patients with known PAF compared to controls. This is not a novel finding and has been reported previously by others [8,12,13]. However, interestingly, the eECG revealed some APBs to be in fact short atrial runs, which were partially blocked (PBARs, Fig. 3). We hypothesize that, from an electrophysiologic viewpoint, such “unmasked” atrial runs with short cycle lengths (therefore partially blocked) may be linked to AF even closer than single APBs or relatively slow conducted atrial runs (visible in sECG) due to their higher arrhythmogenicity. Remarkably, the patient showing the
Table 1 Baseline patient characteristics.
Age [years] Female gender (%) Arterial hypertension (%) Diabetes mellitus (%) CHADS2-Score (median) Ischemic heart disease (%) Hypertensive heart disease (%) Valvular heart disease (%) Left ventricular ejection fraction [%] Left atrial diameter, parasternal [mm] Oral anticoagulation (%) Acetylsalicylic acid (%) Class III antiarrhythmic drugs (%) Betablockers (%) IQR: interquartile range.
Cases (n = 25)
Controls (n = 25)
P-value
69 (IQR 61–76) 18 (72) 17 (68) 1 (4) 1 (1–2) 11 (44) 3 (12) 3 (12) 60 (60–65) 44 (39–49) 13 (52) 11 (44) 10 (40) 13 (52)
68 (IQR 61–73) 18 (72) 15 (60) 2 (8) 1 (1–2) 15 (60) 6 (24) 3 (12) 60 (53–60) 42 (39–46) 5 (20) 8 (32) 0 (0) 12 (48)
0.18 1 0.77 1 0.95 0.4 0.46 1 0.27 0.23 0.04 0.56 b0.001 1
500
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
highest number of PBARs (3.5 per hour) had the only episode of PAF we registered during long-term e/sECG monitoring. Furthermore, the fact that APBs with short coupling intervals appear to be more arrhythmogenic than APBs with longer coupling intervals [14] may support the hypothesis of the higher PBAR arrhythmogenicity. Similarly, pulmonary veins are considered arrhythmogenic as they exhibit very fast electric activity. Atrial captures of such trigger activity can easily be revealed by long-term eECG as PBARs, whereas conventional sECG may just reveal one or several atrial premature beats. Thus, PBARs may be very specific surrogate markers for PAF. Right and left atrial signal duration in the signal-averaged esophageal ECG Patients with PAF showed longer left atrial signal duration than controls in the signal-averaged eECG (Fig. 4). The distinction between the “right” and the subsequent “left” atrial signal is academic, since right and left atrial activation are not strictly sequential. However, the main deflection of the atrial eECG signal (interval from (2.) to (3.) in Fig. 2) predominantly reflects left posterior atrial activity, as described by Binkley et al. [15]. The preceding small deflection (interval from (1.) to (2.) in Fig. 2) is mainly governed by right atrial activation and activation of the paraseptal region. The right atrial conduction time also corresponds to findings reported by Binkley et al. [15] as measured by intracardiac leads. The left atrial signal duration of PAF patients was prolonged by 16 ms compared to controls. This prolongation
Table 2 Differences between groups in the long-term ECG. Cases (n = 25) Atrial premature beats 33′764 (84%) (% of patients) Atrial runs (% of patients) 45 (32%) PBARS # of PBARs (% of patients) 68 (20%) duration of PBARs [ms] 853 (IQR 813–1836) atrial cycle length [ms] 366 (IQR 282–432)
Controls (n = 25)
P-value
4993 (64%) 0.02 31 (28%)
n. s.
0 (0%) -
0.05 -
IQR: interquartile range. PBARs: partially A-V blocked atrial runs.
was not attributed to a higher rate of class III antiarrhythmic agents in the PAF patients. The difference remained significant when comparing only the PAF patients without class III antiarrhythmic medication to controls (who did all not take class III antiarrhythmic agents). This prolongation of left atrial signal duration is supported by invasive three-dimensional electroanatomic mapping studies, reporting a prolongation of 14.4 ms in PAF patients compared to controls [16]. The prolonged left atrial signal duration may be explained by the structural and functional remodeling of the atrial substrate in patients with PAF. Atrial fibrillation may promote [17] or follow atrial fibrosis, leading to slow and anisotropic conduction which itself – a vicious circle – begets PAF [16]. Unlike left atrial signal duration, we did not find a prolonged right atrial signal duration in PAF patients, even
Fig. 3. Examples of partially blocked premature atrial runs (PBARs) from two different patients (top and bottom panel). Each panel shows the eECG (red) and a conventional surface ECG lead derived from an electrode at the cranial sternal border to an electrode at the lower left border of the ribcage on the anterior axillary line (black). Atrial (A) and ventricular (V) signals are labeled. Atrial premature beats as recognized automatically by the software are highlighted in blue. To the human interpreter, two APBs are visible in each panel in the sECG lead.
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
501
Fig. 4. Boxplot showing the right and left atrial signal duration measured in the signal-averaged eECG for cases and controls.
though the right atrium might also constitute an arrhythmogenic substrate in PAF patients. However, there is evidence that arrhythmogenic atrial remodeling is less pronounced in
the right as compared to the left atrium [18]. This discordance of the biatrial substrate may explain why we did not find a difference in the right atrial signal duration.
Fig. 5. ROC curve showing the predictive performance of the total atrial signal duration measured in the signal-averaged eECG. A 95% bootstrap confidence interval for the optimal cutoff point (left atrial signal duration of 66.4 ms) is shown in red. The area under the curve (AUC) is given, including a confidence interval.
502
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
Methodological considerations on the esophageal long-term ECG Due to the vicinity of the esophagus and the left atrium, the eECG offers excellent left atrial signals. The esophageal electrode catheter is tolerated well by patients even during prolonged eECG recordings [9]. Thus, eECG allows continuous and direct surveillance of the atrial rhythm unlike conventional surface long-term ECG systems [7]. “APBs” recognized in the sECG can sometimes be PBARs which may only be detected by eECG [19]. Moreover, atrial waveform morphology analysis is possible with the eECG [10,20]. Atrial waveform analysis may in future even allow drawing conclusions on the anatomical origin of APBs, PBARs or conducted atrial runs. In particular, this may become important to monitor the atrial arrhythmogenic activity after catheter ablation of PAF. The esophageal vs. surface signal-averaged ECG Recent studies on P-wave duration in patients with/ without PAF as derived from the signal-averaged sECG have shown conflicting results, mainly attributed to different technical methodologies used [21]. The P-wave duration seems to depend heavily on filter settings and is affected severely by noise [22]. This may also be true for the eECG. However, we believe, that our results are reliable since the eECG features higher atrial signal amplitudes and a better signal-to-noise ratio [9].
interpreted cautiously for such patients (e. g. patients after pulmonary vein isolation have been reported to experience procedure-related modification of the atrial signal duration). Conclusion This is the first study evaluating the performance of long-term eECG for the detection of PAF. eECG offers excellent atrial signals, revealing surrogate markers which may be of clinical importance to identify patients likely suffering from PAF. Notably, partially blocked atrial runs and prolonged left atrial signal duration are indicative for PAF. Funding and conflicts of interest Funding: This work was supported by the Commission for Technology and Innovation, Switzerland; the Swiss Heart Foundation, Switzerland; the Department of Cardiology, Bern University Hospital and University of Bern, Switzerland and the Department of Cardiology, Bürgerspital Solothurn. Conflicts of interest None.
Clinical implications
Ethical approval
Based on our results, a prolonged left atrial signal duration measured in the signal-averaged eECG may be a simple noninvasive marker (independent of age and gender) of atrial remodeling in the diseased atria of PAF patients. Moreover, PBARs seem to be a new and specific surrogate marker for PAF. The predictive value of left atrial signal duration as surrogate markers for PAF was moderate in our patient population (Fig. 5). It has to be mentioned that in a real-life patient cohort with a lower PAF pre-test probability than in our study (50%), the positive predictive value of left atrial signal duration will be even lower and requires further investigation. We hypothesize that patients with a high number of APBs, blocked APBs in the 12-channel ECG, long left atrial signal duration or PBARs in eECG may benefit from extensive and repeated long-term rhythm monitoring in order to detect PAF. EECG registration requires a dedicated electrode inserted in the esophagus. Esophageal leads are tolerated well by patients as we already have shown earlier [9], thus, may be used more often in future.
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Limitations To the author's knowledge, this is the first characterization of long-term eECG surrogate markers for PAF. However, since our results based on a small number of patients, further evaluation of the method is required. We excluded patients after atrial catheter ablation or with an atrial pacemaker lead. Thus, our results have to be
Acknowledgements None. References [1] European Heart Rhythm Association, European Association for Cardio-Thoracic Surgery, Camm AJ, Kirchhof P, Lip GYH, Schotten U, et al. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC). Eur Heart J 2010;31:2369–429. [2] Healey JS, Connolly SJ, Gold MR, Israel CW, Van Gelder IC, Capucci A, et al. Subclinical atrial fibrillation and the risk of stroke. N Engl J Med 2012;366(2):120–9. [3] Camm AJ, Lip GY, De Caterina R, Savelieva I, Atar D, Hohnloser SH, et al. 2012 focused update of the ESC guidelines for the management of atrial fibrillation: an update of the 2010 ESC guidelines for the management of atrial fibrillation–developed with the special contribution of the European Heart Rhythm Association. Europace 2012; 14(10):1385–413. [4] Dobreanu D, Svendsen JH, Lewalter T, Hernandez-Madrid A, Lip GY, Blomstrom-Lundqvist C. Current practice for diagnosis and management of silent atrial fibrillation: results of the European Heart Rhythm Association survey. Europace 2013;15(8):1223–5. [5] Arya A, Piorkowski C, Sommer P, Kottkamp H, Hindricks G. Clinical implications of various follow up strategies after catheter ablation of atrial fibrillation. Pacing Clin Electrophysiol 2007;30(4):458–62.
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503 [6] Roten L, Schilling M, Haberlin A, Seiler J, Schwick NG, Fuhrer J, et al. Is 7-day event triggered ECG recording equivalent to 7-day holter ECG recording for atrial fibrillation screening? Heart 2012;98(8):645–9. [7] Haeberlin A, Roten L, Schilling M, Scarcia F, Niederhauser T, Vogel R, et al. Software-based detection of atrial fibrillation in long-term ECG's. Heart Rhythm 2014;11(6):933–8. [8] Wallmann D, Tuller D, Wustmann K, Meier P, Isenegger J, Arnold M, et al. Frequent atrial premature beats predict paroxysmal atrial fibrillation in stroke patients: an opportunity for a new diagnostic strategy. Stroke 2007;38(8):2292–4. [9] Haeberlin A, Niederhauser T, Marisa T, Goette J, Jacoment M, Mattle D, et al. The optimal lead insertion depth for esophageal ECG recordings with respect to atrial signal quality. J Electrocardiol 2013;46(2):158–65. [10] Niederhauser T, Marisa T, Haeberlin A, Goette J, Jacoment M, Vogel R. High-resolution esophageal long-term ECG allows detailed atrial wave morphology analysis in case of atrial ectopic beats. Med Biol Eng Comput 2012;50(7):769–72. [11] Tanner H, Hindricks G, Kobza R, Dorszewski A, Schirdewahn P, Piorkowski C. Trigger activity more than three years after left atrial linear ablation without pulmonary vein isolation in patients with atrial fibrillation. J Am Coll Cardiol 2005;46(2):338–43. [12] Binici Z, Intzilakis T, Nielsen OW, Kober L, Sajadieh A. Excessive supraventricular ectopic activity and increased risk of atrial fibrillation and stroke. Circulation 2010;121(17):1904–11. [13] Kochhauser S, Dechering DG, Dittrich R, Reinke F, Ritter MA, Ramtin S. Supraventricular premature beats and short atrial runs predict atrial fibrillation in continuously monitored patients with cryptogenic stroke. Stroke 2014;45(3):884–6. [14] Yamane T, Date T, Kanzaki Y, Inada K, Matsuo S, Shibayama K. Behavior of atrial ectopic beats before and after pulmonary vein
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
503
isolation in patients with atrial fibrillation: a reduction in the number and arrhythmogenicity of ectopic firings. Heart Rhythm 2006;3(12): 1421–7. Binkley PF, Bush CA, Fleishman BL, Leier CV. In vivo validation of the origin of the esophageal electrocardiogram. J Am Coll Cardiol 1986;7:813–8. Lin Y, Yang B, Garcia FC, Ju W, Zhang F, Chen H. Comparison of left atrial electrophysiologic abnormalities during sinus rhythm in patients with different type of atrial fibrillation. J Interv Card Electrophysiol 2014;39(1):57–67. He X, Gao X, Peng L, Wang S, Zhu Y, Ma H. Atrial fibrillation induces myocardial fibrosis through angiotensin II type 1 receptorspecific arkadia-mediated downregulation of Smad7. Circ Res 2011; 108(2):164–75. Chang SL, Tai CT, Lin YJ, Wongcharoen W, Lo LW, Tuan TC, et al. Biatrial substrate properties in patients with atrial fibrillation. J Cardiovasc Electrophysiol 2007;18(11):1134–9. Haeberlin A, Niederhauser T, Marisa T, Mattle D, Jacomet M, Goette J, et al. Esophageal long-term ECG reveals paroxysmal atrial fibrillation. Circulation 2012;125(18):2281–2. Haeberlin A, Niederhauser T, Tanner H, Vogel R. Atrial waveform analysis using esophageal long-term electrocardiography reveals atrial ectopic activity. Clin Res Cardiol 2012;101(11):941–2. Blanche C, Tran N, Carballo D, Rigamonti F, Burri H, Zimmermann M. Usefulness of P-wave signal averaging to predict atrial fibrillation recurrences after electrical cardioversion. Ann Noninvasive Electrocardiol 2014;19(3):266–72. Holmqvist F, Platonov PG, Havmoller R, Carlson J. Signal-averaged P wave analysis for delineation of interatrial conduction - further validation of the method. BMC Cardiovasc Disord 2007;29:1–8.
ScienceDirect Journal of Electrocardiology 49 (2016) 496 – 503 www.jecgonline.com
Markers for silent atrial fibrillation in esophageal long-term electrocardiography Andreas Haeberlin, MD, PhD, a, b,⁎ Lucca Lacheta, MMed, a Thomas Niederhauser, PhD, b, c Thanks Marisa, MSc, b, c Reto A. Wildhaber, MD, MSc, c Josef Goette, PhD, c Marcel Jacomet, PhD, c Jens Seiler, MD, a Juerg Fuhrer, MD, a Laurent Roten, MD, a Hildegard Tanner, MD, a Rolf Vogel, MD, PhD, MSEE d a
Department of Cardiology, Bern University Hospital and University of Bern, Bern, Switzerland ARTORG Center for Biomedical Engineering Research, University of Bern, Bern, Switzerland Institute for Human Centered Engineering, Bern University of Applied Sciences, Biel, Switzerland d Department of Cardiology, Bürgerspital Solothurn, Solothurn, Switzerland b
c
Abstract
Purpose: Paroxysmal atrial fibrillation (PAF) often remains undiagnosed. Long-term surface ECG is used for screening, but has limitations. Esophageal ECG (eECG) allows recording high quality atrial signals, which were used to identify markers for PAF. Methods: In 50 patients (25 patients with PAF; 25 controls) an eECG and surface ECG was recorded simultaneously. Partially A-V blocked atrial runs (PBARs) were quantified, atrial signal duration in eECG was measured. Results: eECG revealed 1.8‰ of atrial premature beats in patients with known PAF to be PBARs with a median duration of 853 ms (interquartile range (IQR) 813-1836 ms) and a median atrial cycle length of 366 ms (IQR 282-432 ms). Even during a short recording duration of 2.1 h (IQR 1.2– 17.2 h), PBARs occurred in 20% of PAF patients but not in controls (p = 0.05). Left atrial signal duration was predictive for PAF (72% sensitivity, 80% specificity). Conclusions: eECG reveals partially blocked atrial runs and prolonged left atrial signal duration – two novel surrogate markers for PAF. © 2016 Elsevier Inc. All rights reserved.
Keywords:
atrial fibrillation; esophageal ECG; long-term ECG; surrogate markers
Background Paroxysmal atrial fibrillation (PAF) is common and may have severe consequences such as ischemic strokes [1]. PAF may remain clinically silent and undiagnosed if patients are asymptomatic [1]. Nevertheless, even silent PAF increases the risk for strokes [2,3]. Thus, a reliable screening tool for silent PAF is highly desirable in order to prevent adverse outcomes. Commonly, long-term ECG recordings are performed to screen for PAF [4]. Continuous long-term ECG recordings for several days outperforms 24-h ECG monitoring [5] or a discontinuous event triggered ECG monitoring [6]. However, continuous long-term recordings
⁎ Corresponding author at: Department of Cardiology, Inselspital, Bern University Hospital, 3010, Bern, Switzerland. E-mail address: [email protected] http://dx.doi.org/10.1016/j.jelectrocard.2016.03.006 0022-0736/© 2016 Elsevier Inc. All rights reserved.
increase the data volume to be reviewed, thus, the time needed for analysis may become high [7]. Furthermore, surface long-term ECG recordings very often cause patient discomfort due to skin irritation [6]. Complementary, ECG surrogate markers like frequent atrial premature beats (APBs) allow identifying patients with a high pretest probability for PAF [8], even if no PAF episode has been recorded in a long-term ECG yet. Patients with high PAF probability may then undergo repeated long-term ECG recordings to confirm presence of PAF. However, the performance of this approach is clearly limited by the fact that the atrial electrogram cannot be analyzed directly in conventional surface long-term ECG recording systems. Instead, R-R intervals are used as surrogate markers of the atrial activity [7]. As an alternative to conventional surface ECG monitoring, esophageal electrocardiography (eECG) may be used. eECG provides excellent atrial electrograms [9], which allow a precise and direct analysis of the atrial electric activity. The purpose of the present study
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
was to identify and validate eECG surrogate markers for PAF by comparing PAF patients with matched controls. Methods Study design In this investigator-initiated case–control study, we analyzed esophageal long-term ECG recordings for the presence of surrogate markers to detect PAF. For this purpose, we compared 25 patients with known PAF (but presenting in sinus rhythm) to 25 control patients 1:1 matched for age (± one year) and gender. The local ethics committee approved the study (according to the declaration of Helsinki 1964). Informed consent was obtained from all individual participants included in the study.
497
(BlueSensor VL®, Ambu, Denmark). The first sECG channel was derived between an electrode at the cranial sternal border to an electrode at the lower left border of the ribcage on the anterior axillary line. The latter in conjunction with an electrode at the lower right border of the ribcage on the mid-clavicular line defined the second channel. Synchronization of the eECG and sECG recordings was obtained using a two-step approach. Prior to the registration, the internal clocks of both devices were synchronized. After the registration, fine adjustment was done manually by synchronizing both ECG traces using clearly identifiable events on both traces (e. g. ventricular premature beats). Patients were asked to grade their subjective overall discomfort during lead insertion and one hour later. Possible answers were “no discomfort”, “low discomfort”, “moderate discomfort”, “strong discomfort” and “discomfort not tolerable”.
Study population The eECG recordings were acquired from patients aged ≥ 18 years presenting to the Department of Cardiology, Bern University Hospital. All “case” recordings were acquired from patients with known PAF referred to our clinic for any reason. The case patients were enrolled into the study while being in sinus rhythm on the standard 12-channel ECG. All “control” recordings were acquired from patients without known PAF referred to our clinic for any reason. PAF was ruled out by at least one negative continuous 7-day ECG within one year prior to inclusion. In addition, patients had no history suggestive of PAF. Exclusion criteria for cases and controls were: - Pacemaker/ICD with an atrial electrode - History of prior ablation of atrial fibrillation or other supraventricular tachyarrhythmias - Heart transplantation - Valve replacement less than 4 weeks ago - Carcinoma of the esophagus or nasopharynx - Acute myocardial infarction or hemodynamically unstable patients Pregnancy ECG registration All patients underwent eECG acquisition for several hours using an esophageal lead (Esosoft/Esoflex 6S®, FIAB, Italy), connected to a dedicated high-resolution eECG recorder [10]. The eECG recorder continuously acquires a bipolar eECG signal with a sampling frequency of 512 Hz and 24 bit resolution [10]. Patients carried the device similar to a hearing aid behind the ear in a small cushioned splash waterproof bag. The lead was inserted nasally, the lead insertion depth was chosen according to a previously described approach [9]. Simultaneously and during the same time period, a conventional surface ECG (sECG) was registered using a long-term ECG recorder (Lifecard CF®, Spacelabs Healthcare, USA). This device continuously records two bipolar sECG channels with 1024 Hz sampling frequency and 12 bit resolution using three dedicated long-term sECG electrodes
ECG processing and analysis sECG recordings were analyzed using the Pathfinder SL® software, version 1.7.1 (Spacelabs Healthcare, USA) without changing any default parameters. The presence and number of APBs and atrial runs (three or more beats with supraventricular origin, total episode duration less than 30 s [11]) were analyzed for cases and controls. Subsequently, the eECG recordings were analyzed (direct visual analysis) by the same person, an experienced eECG interpreter (physician), for the presence of potential surrogate markers for PAF. These a priori defined features comprise partially A-V blocked atrial runs (PBARs) and right and left atrial signal duration of the atrial eECG signal. The ECG interpreter was blinded with respect to the patient group (case/control) or other patient data (all ECG recordings were given numbers not corresponding to the patient database). The signal was filtered by a 0.5–200 Hz bandpass filter using custom-made eECG visualization software (Fig. 1). The presence of APBs and atrial runs – in particular PBARs – was analyzed. Furthermore, the duration of PBARs (from the first to the last atrial electrogram of each PBAR) and the cycle lengths (of the atrial electrograms of each PBAR) were measured. After long-term eECG analysis, the original eECG data were processed using custom-made software based on MATLAB® (MathWorks, USA) to calculate signalaveraged eECGs. No filters were applied. Detection of normal A-V-sequential (presumably sinus) beats was performed using a template matching algorithm (triggered by the atrial electrogram in the eECG). Subsequently, all these beats were averaged, resulting in a “typical mean waveform morphology” of the atrial signal for each patient (Fig. 2 shows an example). The averaged atrial eECG signal was analyzed for the presence of the potential surrogate markers specified above. “Right atrial signal duration” – appearing as right atrial “far-field signal” in the eECG – was defined to start at the beginning of the atrial eECG signal (point 1 in Fig. 2) and to last until the beginning of the main deflection of the atrial eECG signal (point 2 in Fig. 2). The interval from point 2 (where the slope of the
498
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
Fig. 1. Analysis window of the eECG visualization software showing an eECG sequence. At the top, the registration date and the acquisition time are displayed. The eECG is shown below (one continuous channel). In the first fully visible beat, the atrial (A) and ventricular (V) electrogram are labeled.
main deflection exceeded ± 45°) to the end of the atrial eECG signal (point 3 in Fig. 2) was defined to reflect the “left atrial signal duration”. Statistical analysis Statistical analysis was made using R version 3.0.1 for Windows. Continuous variables are expressed as median and interquartile range (IQR). Medians between the two paired patient groups were compared by a Wilcoxon signed-rank test. Non-parametric 95%-confidence intervals (CI's) are reported. Categorical variables are expressed as numbers and percentages, proportions were compared by Fisher's exact test.
Receiver operating characteristics (ROC) analysis was performed to assess sensitivity and specificity of ECG surrogate markers. Areas under the ROC curve and corresponding 95% bootstrap confidence intervals (CI's) for sensitivity and specificity were calculated. A P-value ≤ 0.05 was considered to be significant. Results Baseline patient characteristics We observed no differences between the case and control group regarding age or gender (matching variables), cardiovascular risk factors, disease etiology, echocardio-
Fig. 2. Signal-averaged atrial electrogram as derived from the eECG. The beginning of the right (1.) and the end of the left atrial signal (3.) are highlighted. 2.) represents the end of the right atrial and the beginning of the left atrial signal. The main deflection of the signal does not reflect the QRS-complex but the main activation of the left atrium.
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
graphic parameters or betablockers (Table 1). The rate of oral anticoagulation or class III antiarrhythmic agents (amiodarone and dronedarone) was higher in patients with known PAF. N = 5 (20%) of control patients received oral anticoagulation due to recurrent pulmonary embolism (n = 3) or low LV ejection fraction (n = 2). The PAF burden in the study population was low, just a single PAF episode was detected in one patient.
499
16 ms longer (95%-CI: 11–21 ms, p = 0.003). Furthermore, there was no difference in left atrial signal duration of cases with class III antiarrhythmic agents vs. cases without class III agents (p = 0.31). The predictive value of the total atrial signal duration was optimal at a cutoff of 114.6 ms, offering 76% sensitivity and 76% specificity for PAF (AUC 0.83, Fig. 5). Patient tolerance of long-term esophageal ECG
Surrogate markers for PAF in the long-term ECG During the recording period, patients with known PAF showed more APBs per hour in comparison to controls in sECG and eECG (4.9 (0.9–18.9) vs. 1.5 (0–5.1), p = 0.02), whereas the number of atrial runs was not different. Median ECG recording duration was not different between cases and controls (1.9 (IQR 1.1–15.6) vs. 3.1 (IQR 1.8–17.6) hours, p = 0.22) and identical for sECG/eECG recordings (simultaneous registration). The longest eECG recording duration was 49 h. In addition to the sECG analysis, the eECG revealed 68 of 38,757 (1.8‰) of the recorded APBs to be PBARs (Fig. 3). PBARs were found in five PAF patients (20%) but not in controls (p = 0.05). Median duration of PBARs was 853 ms (813–1836 ms) and median atrial cycle length of PBARs was 366 ms (282–432 ms). Detailed numerical data about incidence of atrial premature beats, atrial runs and PBARs can be found in Table 2. Surrogate markers for PAF in the signal-averaged esophageal ECG Median atrial signal duration was 16 ms (95%-CI: 9– 25 ms) longer in patients with known PAF vs. control patients (p b 0.0001). We attributed this difference to a longer median left atrial signal duration (16 ms, 95%-CI: 9– 22 ms, p b 0.001, Fig. 4). In contrast, median right atrial signal duration was not different (Fig. 4). The difference in left atrial signal duration was not caused by a higher rate of class III antiarrhythmic agents in patients with known PAF. Median left atrial signal duration of case patients without class III antiarrhythmic agents vs. control patients was also
During insertion, patients only experienced low (66%) or moderate discomfort (34%). One hour later, patient discomfort had decreased significantly (p = 0.01); most patients reported only low (92%) or even no (4%) discomfort. Only 4% of patients still suffered from moderate discomfort. However, no eECG registration had to be aborted prematurely. Discussion This is the first study evaluating the performance of long-term eECG surrogate markers for PAF. eECG offers high quality atrial signals, which allow direct analysis of the atrial electric activity, unlike sECG. PBARs and the duration of the left atrial activation are indicative for PAF. These surrogate markers may be used to identify patients with a high likelihood of PAF. The role of partially blocked atrial runs and atrial premature beats In our study, we observed more APBs in patients with known PAF compared to controls. This is not a novel finding and has been reported previously by others [8,12,13]. However, interestingly, the eECG revealed some APBs to be in fact short atrial runs, which were partially blocked (PBARs, Fig. 3). We hypothesize that, from an electrophysiologic viewpoint, such “unmasked” atrial runs with short cycle lengths (therefore partially blocked) may be linked to AF even closer than single APBs or relatively slow conducted atrial runs (visible in sECG) due to their higher arrhythmogenicity. Remarkably, the patient showing the
Table 1 Baseline patient characteristics.
Age [years] Female gender (%) Arterial hypertension (%) Diabetes mellitus (%) CHADS2-Score (median) Ischemic heart disease (%) Hypertensive heart disease (%) Valvular heart disease (%) Left ventricular ejection fraction [%] Left atrial diameter, parasternal [mm] Oral anticoagulation (%) Acetylsalicylic acid (%) Class III antiarrhythmic drugs (%) Betablockers (%) IQR: interquartile range.
Cases (n = 25)
Controls (n = 25)
P-value
69 (IQR 61–76) 18 (72) 17 (68) 1 (4) 1 (1–2) 11 (44) 3 (12) 3 (12) 60 (60–65) 44 (39–49) 13 (52) 11 (44) 10 (40) 13 (52)
68 (IQR 61–73) 18 (72) 15 (60) 2 (8) 1 (1–2) 15 (60) 6 (24) 3 (12) 60 (53–60) 42 (39–46) 5 (20) 8 (32) 0 (0) 12 (48)
0.18 1 0.77 1 0.95 0.4 0.46 1 0.27 0.23 0.04 0.56 b0.001 1
500
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
highest number of PBARs (3.5 per hour) had the only episode of PAF we registered during long-term e/sECG monitoring. Furthermore, the fact that APBs with short coupling intervals appear to be more arrhythmogenic than APBs with longer coupling intervals [14] may support the hypothesis of the higher PBAR arrhythmogenicity. Similarly, pulmonary veins are considered arrhythmogenic as they exhibit very fast electric activity. Atrial captures of such trigger activity can easily be revealed by long-term eECG as PBARs, whereas conventional sECG may just reveal one or several atrial premature beats. Thus, PBARs may be very specific surrogate markers for PAF. Right and left atrial signal duration in the signal-averaged esophageal ECG Patients with PAF showed longer left atrial signal duration than controls in the signal-averaged eECG (Fig. 4). The distinction between the “right” and the subsequent “left” atrial signal is academic, since right and left atrial activation are not strictly sequential. However, the main deflection of the atrial eECG signal (interval from (2.) to (3.) in Fig. 2) predominantly reflects left posterior atrial activity, as described by Binkley et al. [15]. The preceding small deflection (interval from (1.) to (2.) in Fig. 2) is mainly governed by right atrial activation and activation of the paraseptal region. The right atrial conduction time also corresponds to findings reported by Binkley et al. [15] as measured by intracardiac leads. The left atrial signal duration of PAF patients was prolonged by 16 ms compared to controls. This prolongation
Table 2 Differences between groups in the long-term ECG. Cases (n = 25) Atrial premature beats 33′764 (84%) (% of patients) Atrial runs (% of patients) 45 (32%) PBARS # of PBARs (% of patients) 68 (20%) duration of PBARs [ms] 853 (IQR 813–1836) atrial cycle length [ms] 366 (IQR 282–432)
Controls (n = 25)
P-value
4993 (64%) 0.02 31 (28%)
n. s.
0 (0%) -
0.05 -
IQR: interquartile range. PBARs: partially A-V blocked atrial runs.
was not attributed to a higher rate of class III antiarrhythmic agents in the PAF patients. The difference remained significant when comparing only the PAF patients without class III antiarrhythmic medication to controls (who did all not take class III antiarrhythmic agents). This prolongation of left atrial signal duration is supported by invasive three-dimensional electroanatomic mapping studies, reporting a prolongation of 14.4 ms in PAF patients compared to controls [16]. The prolonged left atrial signal duration may be explained by the structural and functional remodeling of the atrial substrate in patients with PAF. Atrial fibrillation may promote [17] or follow atrial fibrosis, leading to slow and anisotropic conduction which itself – a vicious circle – begets PAF [16]. Unlike left atrial signal duration, we did not find a prolonged right atrial signal duration in PAF patients, even
Fig. 3. Examples of partially blocked premature atrial runs (PBARs) from two different patients (top and bottom panel). Each panel shows the eECG (red) and a conventional surface ECG lead derived from an electrode at the cranial sternal border to an electrode at the lower left border of the ribcage on the anterior axillary line (black). Atrial (A) and ventricular (V) signals are labeled. Atrial premature beats as recognized automatically by the software are highlighted in blue. To the human interpreter, two APBs are visible in each panel in the sECG lead.
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
501
Fig. 4. Boxplot showing the right and left atrial signal duration measured in the signal-averaged eECG for cases and controls.
though the right atrium might also constitute an arrhythmogenic substrate in PAF patients. However, there is evidence that arrhythmogenic atrial remodeling is less pronounced in
the right as compared to the left atrium [18]. This discordance of the biatrial substrate may explain why we did not find a difference in the right atrial signal duration.
Fig. 5. ROC curve showing the predictive performance of the total atrial signal duration measured in the signal-averaged eECG. A 95% bootstrap confidence interval for the optimal cutoff point (left atrial signal duration of 66.4 ms) is shown in red. The area under the curve (AUC) is given, including a confidence interval.
502
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503
Methodological considerations on the esophageal long-term ECG Due to the vicinity of the esophagus and the left atrium, the eECG offers excellent left atrial signals. The esophageal electrode catheter is tolerated well by patients even during prolonged eECG recordings [9]. Thus, eECG allows continuous and direct surveillance of the atrial rhythm unlike conventional surface long-term ECG systems [7]. “APBs” recognized in the sECG can sometimes be PBARs which may only be detected by eECG [19]. Moreover, atrial waveform morphology analysis is possible with the eECG [10,20]. Atrial waveform analysis may in future even allow drawing conclusions on the anatomical origin of APBs, PBARs or conducted atrial runs. In particular, this may become important to monitor the atrial arrhythmogenic activity after catheter ablation of PAF. The esophageal vs. surface signal-averaged ECG Recent studies on P-wave duration in patients with/ without PAF as derived from the signal-averaged sECG have shown conflicting results, mainly attributed to different technical methodologies used [21]. The P-wave duration seems to depend heavily on filter settings and is affected severely by noise [22]. This may also be true for the eECG. However, we believe, that our results are reliable since the eECG features higher atrial signal amplitudes and a better signal-to-noise ratio [9].
interpreted cautiously for such patients (e. g. patients after pulmonary vein isolation have been reported to experience procedure-related modification of the atrial signal duration). Conclusion This is the first study evaluating the performance of long-term eECG for the detection of PAF. eECG offers excellent atrial signals, revealing surrogate markers which may be of clinical importance to identify patients likely suffering from PAF. Notably, partially blocked atrial runs and prolonged left atrial signal duration are indicative for PAF. Funding and conflicts of interest Funding: This work was supported by the Commission for Technology and Innovation, Switzerland; the Swiss Heart Foundation, Switzerland; the Department of Cardiology, Bern University Hospital and University of Bern, Switzerland and the Department of Cardiology, Bürgerspital Solothurn. Conflicts of interest None.
Clinical implications
Ethical approval
Based on our results, a prolonged left atrial signal duration measured in the signal-averaged eECG may be a simple noninvasive marker (independent of age and gender) of atrial remodeling in the diseased atria of PAF patients. Moreover, PBARs seem to be a new and specific surrogate marker for PAF. The predictive value of left atrial signal duration as surrogate markers for PAF was moderate in our patient population (Fig. 5). It has to be mentioned that in a real-life patient cohort with a lower PAF pre-test probability than in our study (50%), the positive predictive value of left atrial signal duration will be even lower and requires further investigation. We hypothesize that patients with a high number of APBs, blocked APBs in the 12-channel ECG, long left atrial signal duration or PBARs in eECG may benefit from extensive and repeated long-term rhythm monitoring in order to detect PAF. EECG registration requires a dedicated electrode inserted in the esophagus. Esophageal leads are tolerated well by patients as we already have shown earlier [9], thus, may be used more often in future.
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Limitations To the author's knowledge, this is the first characterization of long-term eECG surrogate markers for PAF. However, since our results based on a small number of patients, further evaluation of the method is required. We excluded patients after atrial catheter ablation or with an atrial pacemaker lead. Thus, our results have to be
Acknowledgements None. References [1] European Heart Rhythm Association, European Association for Cardio-Thoracic Surgery, Camm AJ, Kirchhof P, Lip GYH, Schotten U, et al. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC). Eur Heart J 2010;31:2369–429. [2] Healey JS, Connolly SJ, Gold MR, Israel CW, Van Gelder IC, Capucci A, et al. Subclinical atrial fibrillation and the risk of stroke. N Engl J Med 2012;366(2):120–9. [3] Camm AJ, Lip GY, De Caterina R, Savelieva I, Atar D, Hohnloser SH, et al. 2012 focused update of the ESC guidelines for the management of atrial fibrillation: an update of the 2010 ESC guidelines for the management of atrial fibrillation–developed with the special contribution of the European Heart Rhythm Association. Europace 2012; 14(10):1385–413. [4] Dobreanu D, Svendsen JH, Lewalter T, Hernandez-Madrid A, Lip GY, Blomstrom-Lundqvist C. Current practice for diagnosis and management of silent atrial fibrillation: results of the European Heart Rhythm Association survey. Europace 2013;15(8):1223–5. [5] Arya A, Piorkowski C, Sommer P, Kottkamp H, Hindricks G. Clinical implications of various follow up strategies after catheter ablation of atrial fibrillation. Pacing Clin Electrophysiol 2007;30(4):458–62.
A. Haeberlin et al. / Journal of Electrocardiology 49 (2016) 496–503 [6] Roten L, Schilling M, Haberlin A, Seiler J, Schwick NG, Fuhrer J, et al. Is 7-day event triggered ECG recording equivalent to 7-day holter ECG recording for atrial fibrillation screening? Heart 2012;98(8):645–9. [7] Haeberlin A, Roten L, Schilling M, Scarcia F, Niederhauser T, Vogel R, et al. Software-based detection of atrial fibrillation in long-term ECG's. Heart Rhythm 2014;11(6):933–8. [8] Wallmann D, Tuller D, Wustmann K, Meier P, Isenegger J, Arnold M, et al. Frequent atrial premature beats predict paroxysmal atrial fibrillation in stroke patients: an opportunity for a new diagnostic strategy. Stroke 2007;38(8):2292–4. [9] Haeberlin A, Niederhauser T, Marisa T, Goette J, Jacoment M, Mattle D, et al. The optimal lead insertion depth for esophageal ECG recordings with respect to atrial signal quality. J Electrocardiol 2013;46(2):158–65. [10] Niederhauser T, Marisa T, Haeberlin A, Goette J, Jacoment M, Vogel R. High-resolution esophageal long-term ECG allows detailed atrial wave morphology analysis in case of atrial ectopic beats. Med Biol Eng Comput 2012;50(7):769–72. [11] Tanner H, Hindricks G, Kobza R, Dorszewski A, Schirdewahn P, Piorkowski C. Trigger activity more than three years after left atrial linear ablation without pulmonary vein isolation in patients with atrial fibrillation. J Am Coll Cardiol 2005;46(2):338–43. [12] Binici Z, Intzilakis T, Nielsen OW, Kober L, Sajadieh A. Excessive supraventricular ectopic activity and increased risk of atrial fibrillation and stroke. Circulation 2010;121(17):1904–11. [13] Kochhauser S, Dechering DG, Dittrich R, Reinke F, Ritter MA, Ramtin S. Supraventricular premature beats and short atrial runs predict atrial fibrillation in continuously monitored patients with cryptogenic stroke. Stroke 2014;45(3):884–6. [14] Yamane T, Date T, Kanzaki Y, Inada K, Matsuo S, Shibayama K. Behavior of atrial ectopic beats before and after pulmonary vein
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
503
isolation in patients with atrial fibrillation: a reduction in the number and arrhythmogenicity of ectopic firings. Heart Rhythm 2006;3(12): 1421–7. Binkley PF, Bush CA, Fleishman BL, Leier CV. In vivo validation of the origin of the esophageal electrocardiogram. J Am Coll Cardiol 1986;7:813–8. Lin Y, Yang B, Garcia FC, Ju W, Zhang F, Chen H. Comparison of left atrial electrophysiologic abnormalities during sinus rhythm in patients with different type of atrial fibrillation. J Interv Card Electrophysiol 2014;39(1):57–67. He X, Gao X, Peng L, Wang S, Zhu Y, Ma H. Atrial fibrillation induces myocardial fibrosis through angiotensin II type 1 receptorspecific arkadia-mediated downregulation of Smad7. Circ Res 2011; 108(2):164–75. Chang SL, Tai CT, Lin YJ, Wongcharoen W, Lo LW, Tuan TC, et al. Biatrial substrate properties in patients with atrial fibrillation. J Cardiovasc Electrophysiol 2007;18(11):1134–9. Haeberlin A, Niederhauser T, Marisa T, Mattle D, Jacomet M, Goette J, et al. Esophageal long-term ECG reveals paroxysmal atrial fibrillation. Circulation 2012;125(18):2281–2. Haeberlin A, Niederhauser T, Tanner H, Vogel R. Atrial waveform analysis using esophageal long-term electrocardiography reveals atrial ectopic activity. Clin Res Cardiol 2012;101(11):941–2. Blanche C, Tran N, Carballo D, Rigamonti F, Burri H, Zimmermann M. Usefulness of P-wave signal averaging to predict atrial fibrillation recurrences after electrical cardioversion. Ann Noninvasive Electrocardiol 2014;19(3):266–72. Holmqvist F, Platonov PG, Havmoller R, Carlson J. Signal-averaged P wave analysis for delineation of interatrial conduction - further validation of the method. BMC Cardiovasc Disord 2007;29:1–8.