- Email: [email protected]
Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population
IJCA-25131; No of Pages 5 International Journal of Cardiology xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population☆ Paul D. Ziegler a,⁎, John D. Rogers b, Scott W. Ferreira c, Allan J. Nichols d, Mark Richards e, Jodi L. Koehler a, Shantanu Sarkar a a
Medtronic Cardiac Rhythm Heart Failure, Mounds View, MN, United States Scripps Clinic, La Jolla, CA, United States St. Louis University, St. Louis, MO, United States d OhioHealth Heart & Vascular Physicians, Columbus, OH, United States e ProMedica Physicians Cardiology, Toledo, OH, United States b c
a r t i c l e
i n f o
Article history: Received 24 April 2017 Received in revised form 5 June 2017 Accepted 9 June 2017 Available online xxxx Keywords: Atrial fibrillation Cryptogenic stroke Insertable cardiac monitor Stroke (ischemic)
a b s t r a c t Background: The long-term incidence of atrial fibrillation (AF) in cryptogenic stroke (CS) patients has been explored in carefully controlled clinical trials but real-world data are limited. We investigated the two-year incidence of AF in real-world clinical practice among a large cohort of patients with an insertable cardiac monitor (ICM) placed for AF detection following CS. Methods: Patients in the de-identified Medtronic Discovery™ Link database who received an ICM (Reveal LINQ™) for the purpose of AF detection following CS were included and monitored for up to 2 years. All detected AF episodes (≥2 min) were adjudicated. We quantified the AF detection rate using Kaplan-Meier survival estimates, analyzed the median time to initial detection of AF, and simulated the ability of various intermittent monitoring strategies to detect AF. Results: A total of 1247 patients (65.3 ± 13.0 years, 53% male) were included and followed for 579 ± 222 days. AF episodes (n = 4183) were detected in 238 patients, resulting in an AF detection rate of 21.5% at 2 years. The median time to AF detection was 112 [IQR 35–293] days. Intermittent monitoring for AF detection was inferior to continuous ICM monitoring with sensitivities ranging from 2.9% (annual 24-hour Holter) to 29.9% (quarterly 7-day Holters), p b 0.001. Conclusions: AF episodes were detected via continuous monitoring with ICMs in approximately 1 of every 5 CS patients within 2 years of follow-up. The vast majority of patients with AF would not have been detected with conventional external ambulatory monitors. ICMs should therefore be considered in the evaluation of CS patients. © 2017 Published by Elsevier Ireland Ltd.
1. Introduction Despite thorough evaluation following an ischemic stroke, a definitive cause is not identified in up to one-third of patients [1]. A significant proportion of these cryptogenic stroke survivors may have underlying atrial fibrillation (AF) which can be difficult to diagnose with conventional monitoring strategies since the arrhythmia is commonly asymptomatic [2] and often occurs sporadically [3,4]. Timely diagnosis of AF has important therapeutic consequences as this generally prompts a
☆ All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. ⁎ Corresponding author at: Medtronic Cardiac Rhythm Heart Failure, 8200 Coral Sea Street NE, Mailstop MVN41, Mounds View, MN 55112, United States. E-mail address: [email protected] (P.D. Ziegler).
change in pharmacologic treatment from antiplatelets to anticoagulants, given the superiority of the latter in preventing strokes among patients with AF [5]. The advent of small insertable cardiac monitors [6] (ICM) has improved our ability to diagnose infrequent arrhythmias such as AF in cryptogenic stroke patients as evidenced by numerous smaller singlecenter reports with relatively brief follow-up [7,8,9,10,11]. The CRYSTAL-AF study was a moderately sized trial which randomized patients with a recent cryptogenic stroke to standard of care monitoring or intensive monitoring with an ICM [12]. While the trial clearly demonstrated the superiority of monitoring with an ICM compared to standard of care monitoring [13], the study reflected the specific conditions and inclusion/exclusion criteria set forth by the trial design [12]. We previously reported the short-term (6-month) incidence of AF in a large, real-world cohort of patients with an ICM placed for AF
http://dx.doi.org/10.1016/j.ijcard.2017.06.039 0167-5273/© 2017 Published by Elsevier Ireland Ltd.
Please cite this article as: P.D. Ziegler, et al., Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.06.039
2
P.D. Ziegler et al. / International Journal of Cardiology xxx (2017) xxx–xxx
detection following cryptogenic stroke [14]. In the current report, we present the two-year incidence of AF in cryptogenic stroke patients as detected by ICMs in real-world clinical practice. 2. Methods We included all patients who received an ICM (Reveal LINQ™, Medtronic, MN) to monitor for the presence of AF after a cryptogenic stroke between February 24, 2014 and July 9, 2014. At the time of initial device configuration, implanting physicians designated that the reason for monitoring was a cryptogenic stroke. Device data was collected in the de-identified DiscoveryLink™ database along with patient's age and gender. Patients were monitored for up to 24 months after device insertion and all patients consented to the use of their data for research purposes. 2.1. Device characteristics and performance The Reveal LINQ™ ICM is a small (length 44.8 mm, width 7.2 mm, thickness 4.0 mm, volume 1.2 cm3, weight 2.5 g) cardiac monitor that is inserted into the subcutaneous tissue over the heart. Dedicated incision and insertion tools are used to form a small opening in the skin, create a uniformly sized pocket, and guide placement of the device under the skin. The device is typically inserted at 45 degrees relative to the sternum above the 4th intercostal space in the V2–V3 electrode orientation. ECG signals are measured between two electrodes located on each end of the device. Details of the AF detection algorithm operation [6] and performance [15,16] have been reported previously. Briefly, the AF detection algorithm analyzes changes in ventricular conduction over 2-minute intervals for patterns of irregularity and incoherence. Evidence of AF is reduced if p-waves are detected. At the end of each 2-minute interval, the device classifies the rhythm as sinus rhythm or AF. Performance of the AF detection algorithm has been evaluated in several studies by comparing it to simultaneously collected Holter data. In a population with known AF, the algorithm correctly identified the presence of AF in 96.1% of patients with AF (sensitivity) and correctly excluded the absence of AF in 97.4% (negative predictive value) [15]. Further improvements were made to reduce the duration of false positive detections by 55% and the number of false positive detections by 46% with no impact on detection sensitivity [16]. In addition to storing the cumulative time spent in AF for each discrete day of followup, the device also stores ECG waveforms for a portion of up to 14 detected AF episodes. Subsequently detected AF episodes then overwrite the earliest AF episode if the memory is not refreshed by in-office interrogation or remote transmission of the device data. All patients transmitted device data remotely via the Medtronic CareLink™ remote monitoring network. The device will automatically initiate wireless transmission of a portion of the ECG waveform every night for the longest AF episode recorded during the prior day. Patients may also initiate a full transmission of all AF episodes contained within the device's memory at any time. All recorded AF episodes were adjudicated by a single, blinded reviewer. The entire duration of the episode was considered to be either true or false based on the segment of the ECG waveform that was stored at the episode's onset. Only AF episodes determined to be true AF were included in the analysis. 2.2. Intermittent monitoring simulations The methodology for simulating the performance of various intermittent monitoring strategies from continuous monitoring data has been described elsewhere in detail [3,17]. In brief, AF detection data from a subset of days corresponding to the particular monitoring strategy were selected from the continuous daily AF detection memory of the ICM. The intermittent monitoring strategies that were evaluated included a single 24-hour Holter, 48-hour Holter, 7-day Holter, 21-day event recorder, 30-day event recorder, quarterly 24-hour Holters, quarterly 48-hour Holters, quarterly 7-day Holters, and monthly 24-hour Holters. For each intermittent monitoring strategy simulation, the initial day of monitoring was selected at random from a uniform distribution within the first 14 days following insertion of the ICM. For example, in the case of the 7-day Holter simulation, we examined if the ICM detected AF on any day among 7 consecutive days which began within the first 14 days (selected at random) after placement of the device. In the case of monitoring scenarios which were repeated at quarterly or monthly intervals (for the first year of follow-up), all subsequent monitoring periods occurred at a fixed interval from the randomly selected initial day of monitoring. For example, simulation of quarterly 24-hour Holter monitoring was performed by determining if the ICM detected AF on a day selected at random from days 1–14 following device insertion or on days 90, 180, and 270 from that initially selected day. The mean values of sensitivity and negative predictive value for each intermittent monitoring strategy are reported after repeating the simulations 10,000 times to ensure robustness. Sensitivity measures the proportion of patients with AF detected by the ICM who would also have been identified as having AF by intermittent monitoring. Negative predictive value measures the proportion of patients without AF detected by intermittent monitoring who were correctly identified as being free from AF based on the ICM data.
AF detection rates were also computed using AF episode duration thresholds of 6 min, 30 min, and 60 min. The median duration of the longest AF episode and the median time from device insertion to detection of the first AF episode were calculated. The t-test was used to compare the age of patients with AF episodes to those without AF episodes while an ANOVA model was used to compare the age of patients with different durations of their longest AF episode (b1 h, 1–4 h, 4–12 h, or N12 h). The log-rank test was used to compare the rate of AF detection between patients N65 years of age and patients ≤65 years of age. Comparisons of the proportion of patients with AF identified via intermittent monitoring strategies vs. continuous monitoring with the ICM were performed using the McNemar test. Continuous variables are reported as mean and standard deviation (SD) or median [Inter-Quartile Range (IQR)] while discrete variables are reported as counts and percentages, as appropriate. Statistical significance was assigned for p-values b0.05. All analyses were performed with SAS software version 9.4 (SAS Institute, Cary, NC, USA).
3. Results We studied 1247 patients (mean age 65.3 [SD 13.0] years, 53% male) over an average follow-up of 579 [SD 222] days. During this time, a total of 4183 episodes of AF were detected by the ICM in 238 patients. Multiple AF episodes were detected among 182 patients (76.5%) while a single AF episode was detected among the remaining 56 patients (23.5%). The median number of AF episodes detected among patients experiencing AF was 4 [IQR 2–14]. 3.1. AF detection rates and durations The rate of AF detection (≥2 min) increased from 4.6% at 30 days to 12.2% at 6 months, 16.3% at 12 months, and 21.5% at 24 months. Using more stringent AF episode duration thresholds resulted in AF detection rates of 19.4% (≥ 6 min), 17.9% (≥ 30 min), and 16.2% (≥ 60 min) at 24 months [Fig. 1]. The median duration of the longest AF episode experienced among patients with AF was 4.0 [IQR 0.8–12.7] hours. The distribution of the longest AF episode duration and the distribution of all adjudicated AF episode durations are presented in Fig. 2a and b, respectively. Among the 182 patients with multiple episodes of AF, 37 (20%) experienced their longest episode as their first AF episode while 10 patients (5%) had no change in AF episode duration and 135 (74%) had their longest episode at some point after the first AF episode. The median duration of the first detected episode among these 182 patients was 1.22 [0.20–4.50] hours while the duration of the longest subsequent AF episode was 5.25 [0.87–16.37] hours (p b 0.001), indicating that episode durations increased following the initial episode. 3.2. Role of age on AF detection Age data were available on 1219 of the 1247 patients studied (97.8%). Patients in whom AF was detected were significantly older than patients in whom AF was not detected (71.8 [SD 10.5] vs. 63.8 [SD 13.1] years, p b 0.001). Additionally, the AF detection rate at 24 months among patients N65 years old was significantly higher than among patients ≤ 65 years old (26.6% vs. 10.7%, p b 0.001). However,
2.3. Statistical analysis Kaplan-Meier survival estimates were used to compute AF detection rates at 30 days, 6 months, 12 months, and 24 months using a 2 minute duration criterion. At 24 months,
Fig. 1. Time to detection of AF using episode duration thresholds of 2, 6, 30, and 60 min.
Please cite this article as: P.D. Ziegler, et al., Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.06.039
P.D. Ziegler et al. / International Journal of Cardiology xxx (2017) xxx–xxx
3
Fig. 2. Distribution of maximum episode durations among 238 patients with AF detected (left panel) and all episode durations among 4183 episodes detected (right panel).
the median time to AF detection was similar between patients N65 years and ≤ 65 years (110.0 [IQR 32.0–294.0] days vs. 120.5 [IQR 72.0–278.0] days, p = 0.77). Moreover, there were no significant differences in age between patients whose longest AF episode was b1 h (70.3 [SD 11.8] years), 1–4 h (71.0 [SD 11.4] years), 4–12 h (73.1 [SD 9.5] years), or N 12 h (72.9 [SD 8.9] years), p = 0.40.
causes have AF detected by 24 months, the vast majority of these patients (approximately 4 out of 5) actually do not have AF detected within this timeframe despite continuous arrhythmia monitoring with an ICM. Therefore, unless the novel oral anticoagulants being tested also protect against non-AF related stroke, the risks of bleeding associated with this treatment strategy may outweigh the potential benefits in ischemic stroke reduction.
3.3. Impact of monitoring duration and simulations of intermittent monitoring 4.1. AF detection rates and monitoring durations The median time from device insertion to AF detection was 112 [IQR 35–293] days. If monitoring had stopped at 30 days after device insertion, 78.6% of patients who ultimately had AF detected would have gone undiagnosed. Between months 6–12 of follow-up, AF was detected for the first time in 45 additional patients (30% more patients identified at 12 months vs. 6 months on a relative basis). Furthermore, between months 12–24 of follow-up, AF was detected for the first time in yet an additional 45 patients (23% more patients identified at 24 months vs. 12 months on a relative basis). Simulations of various intermittent monitoring strategies revealed that the sensitivity, compared to the ICM, for detecting AF in this cryptogenic stroke population ranged from 2.9% (annual 24-hour Holter) to 29.9% (quarterly 7-day Holters) [Fig. 3]. Negative predictive values, compared to the ICM, ranged from 86.3% (annual 24-hour Holter) to 89.7% (quarterly 7-day Holters) [Fig. 3]. Both the sensitivity and negative predictive value for all intermittent monitoring strategies evaluated were significantly lower than continuous arrhythmia monitoring with the ICM (p b 0.001).
The AF detection rate in our study at 30 days of follow-up was 4.6% and this increased to 21.5% at 24 months of follow-up. Therefore, 78.6% of patients who were eventually found to have AF would not have been diagnosed if monitoring was only performed for up to 30 days as suggested by the current AHA/ASA guidelines [22]. The European Society of Cardiology AF guidelines [23], which were published since the release of the CRYSTAL-AF trial results, place a higher emphasis on long-term monitoring with a Class IIa, Level of Evidence B recommendation for additional ECG monitoring via ICMs to be considered in stroke patients.
4. Discussion This analysis represents the largest cohort of cryptogenic stroke patients who have been continuously monitored via ICMs for up to two years. The main finding of the study is that AF was diagnosed in approximately 1 of every 5 cryptogenic stroke patients (21.5%) by 24 months of follow-up. The AF detection rate in our study was very similar to the AF detection rate observed in the CRYSTAL-AF trial at 24 months (21.1%) [18] despite slight differences in the average age of the cohorts, the work-up required before a stroke was deemed cryptogenic, and how data were transmitted by the respective ICMs utilized in the studies. Recently, trials have been initiated [19,20] to evaluate whether prophylactic oral anticoagulation in patients with embolic strokes of undetermined source (ESUS) [21] is superior to antiplatelet therapy for recurrent stroke prevention. An underlying premise of these studies is that a significant proportion of ESUS patients have undiagnosed AF which would be better treated with oral anticoagulation compared to antiplatelet medication. Our study suggests that although a clinically meaningful percentage of patients with strokes due to undetermined
Fig. 3. Sensitivity and negative predictive values of various intermittent monitoring strategies for detection of AF compared to continuous monitoring with an ICM.
Please cite this article as: P.D. Ziegler, et al., Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.06.039
4
P.D. Ziegler et al. / International Journal of Cardiology xxx (2017) xxx–xxx
4.2. AF episode durations While the ICM used in this study is very sensitive to detecting AF episodes as brief as 2 min in duration [16], we found that most patients with device-detected episodes of AF experienced at least occasional episodes of much longer duration. As shown in Figs. 2, 74% of patients with AF experienced an episode longer than 1 h at some point during the 24-month follow-up. Furthermore, 43% experienced an episode longer than 6 h and 17% experienced episodes that persisted for longer than 24 h. These are likely underestimates of the true percentage of patients who met these episode duration thresholds since devices such as the ICMs used in this study may sometimes undersense the cardiac signal, which can result in partitioning a longer AF episode into multiple shorter AF episodes. Only 11% of patients with AF (or 2% of the overall population) had AF episodes which never exceeded 6 min in duration. We observed that among patients with multiple AF episodes, the longest AF episode generally occurred after the initial AF episode. This potentially supports the concept of atrial remodeling and additional study is required to determine if early active treatment of AF in this population could slow the progression of the disease. 4.3. Role of age on AF detection It is well established that the prevalence of AF increases sharply with advancing age in both the general population [24] and patients who have had a recent ischemic stroke [25]. Not surprisingly, we also detected AF more frequently among older (N65 years) compared to younger (≤ 65 years) cryptogenic stroke patients. However, there was no difference in the median time to initial detection of AF between older (110 [IQR 32–294] days) and younger (120.5 [72–278] days) patients, suggesting that short-term monitoring with external devices may not be a viable option regardless of age. In addition, a recent costeffectiveness analysis of ICMs for the detection of AF in cryptogenic stroke patients indicated that ICMs may actually be more costeffective in younger patients due to their longer life expectancy and a potential increase in quality-adjusted-life-years (QALYs) gained [26].
an ICM. Compared to the landmark CRYSTAL-AF study, this analysis represents N5 times as many patients with ICM data. Another strength is the extensive follow-up duration. A recent review article summarized many of the studies published in the last several years which have explored the incidence of AF via ICMs in cryptogenic stroke populations [27]. Most of these studies followed patients for one year or less. Our follow-up of up to 24 months, coupled with the large sample size, represents the most patient-years of continuously monitored cryptogenic stroke data ever assembled. This study also reflects actual clinical practice, rather than an artificial environment imposed by clinical trial inclusion criteria or protocols. As such, it is indicative of what might be expected when clinicians incorporate such a monitoring strategy into their practice. This study also has several limitations. Since the data were extracted from a de-identified device database, we lack many demographic characteristics, medication data or clinical actions taken, and clinical outcomes for these patients. The definition of cryptogenic stroke was also left to the discretion of the implanting physician and may include more heterogeneity than a rigorously controlled clinical trial. We may also have failed to detect extremely brief episodes of AF (b2 min) due to the technical limitations of the AF detection algorithm in the ICM. However, the clinical importance of such brief episodes is unknown. The RATE Registry recently demonstrated that extremely short episodes of AF were not associated with an increased risk of clinical events compared to patients without documented AF [28]. Furthermore, data from the CRYSTAL-AF trial established that cryptogenic stroke patients with initial AF episodes which are brief in duration often have longer duration AF episodes in the future [18]. This suggests that detection of AF in many patients with extremely brief episodes is likely to be only slightly delayed rather than missed altogether. Finally, our simulations of intermittent monitoring strategies assumed 100% patient compliance with the intended monitoring regimen. In reality, it has been shown that patient compliance with external monitoring devices decreases as the burden of monitoring (i.e., the monitoring frequency and duration) increases [29,30]. Therefore the sensitivities reported here are likely overestimates of what might be expected in clinical practice, particularly for the more aggressive monitoring schemes.
4.4. Value of long-term monitoring 5. Conclusion We previously reported AF detection rates in this cohort at 6 months of follow-up [14]. The current report significantly extends this follow-up and finds that the AF detection rate has increased from 12.2% at 6 months to 21.5% at 24 months, a relative increase of 76%. As AF is detected increasingly further from the cryptogenic stroke, it becomes more challenging to infer causality between AF and the index stroke event. However, the primary clinical goal remains prevention of stroke recurrence rather than definitive identification of the cause of the index stroke. In that context, detection of AF at any time post-stroke is crucial since oral anticoagulation is clearly indicated in AF patients with a prior stroke by current guidelines [22,23]. Where guidelines may be lacking is in the recommendation for arrhythmia monitoring following an ischemic stroke. The current AHA/ASA guidelines suggest that up to 30 days of arrhythmia monitoring is reasonable within the first 6 months following a stroke or transient ischemic attack (TIA) of unknown cause [22]. Our data demonstrates that the vast majority (78.6%) of patients with AF would be missed if monitoring had stopped at 30 days. Furthermore, our simulations of various intermittent monitoring strategies show that all evaluated strategies were inferior to continuous monitoring with ICMs. These results validate the findings of Choe et al. from similar simulations that were performed with data from the CRYSTAL-AF study [17].
AF episodes were detected via continuous monitoring with ICMs in approximately 1 of every 5 cryptogenic stroke patients within 2 years of follow-up. Consistent with the CRYSTAL-AF trial, these results demonstrate that 30 days of monitoring is insufficient to capture the vast majority of cryptogenic stroke patients who have AF and suggest that current monitoring guidelines may be inadequate. Conflict of interest PDZ = Employee and shareholder (Medtronic). JDR = Consultant (Medtronic, Biotronik); Speaker's bureau (Medtronic, Biotronik). SWF = Speaker's bureau (Medtronic). AJN = Speaker's bureau (Medtronic); Shareholder (Boston Scientific, Medtronic). MR = Speaker's bureau (Biotronik, Medtronic, Boston Scientific, Boehringer Ingelheim, Janssen, Pfizer); Consultant (Biotronik, Boston Scientific, Janssen). JLK = Employee and shareholder (Medtronic). SS = Employee and shareholder (Medtronic). References
4.5. Strengths and limitations One of the key strengths of this study is the large sample size of cryptogenic stroke patients who were continuously monitored for AF with
[1] P.L. Kolominsky-Rabas, M. Weber, O. Gefeller, B. Neundoerfer, P.U. Heuschmann, Epidemiology of ischemic stroke subtypes according to TOAST criteria: incidence, recurrence, and long-term survival in ischemic stroke subtypes: a populationbased study, Stroke 32 (2001) 2735–2740.
Please cite this article as: P.D. Ziegler, et al., Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.06.039
P.D. Ziegler et al. / International Journal of Cardiology xxx (2017) xxx–xxx [2] S.A. Strickberger, J. Ip, S. Saksena, K. Curry, T.D. Bahnson, P.D. Ziegler, Relationship between atrial tachyarrhythmias and symptoms, Heart Rhythm. 2 (2005) 125–131. [3] P.D. Ziegler, J.L. Koehler, R. Mehra, Comparison of continuous versus intermittent monitoring of atrial arrhythmias, Heart Rhythm. 3 (12) (2006 Dec) 1445–1452. [4] E.I. Charitos, U. Stierle, P.D. Ziegler, et al., A comprehensive evaluation of rhythm monitoring strategies for the detection of atrial fibrillation recurrence: insights from 647 continuously monitored patients and implications for monitoring after therapeutic interventions, Circulation 126 (2012) 806–814. [5] R.G. Hart, J.L. Halperin, L.A. Pearce, et al., Lessons from the stroke prevention in atrial fibrillation trials, Ann. Intern. Med. 138 (2003) 831–838. [6] S. Sarkar, D. Ritscher, R. Mehra, A detector for a chronic implantable atrial tachyarrhythmia monitor, IEEE Trans. Biomed. Eng. 55 (2008) 1219–1224. [7] M.A. Ritter, S. Kochhäuser, T. Duning, et al., Occult atrial fibrillation in cryptogenic stroke: detection by 7-day electrocardiogram versus implantable cardiac monitors, Stroke 44 (2013) 1449–1452. [8] P.E. Cotter, P.J. Martin, L. Ring, E.A. Warburton, M. Belham, P.J. Pugh, Incidence of atrial fibrillation detected by implantable loop recorders in unexplained stroke, Neurology 80 (2013) 1546–1550. [9] T. Etgen, M. Hochreiter, M. Mundel, T. Freudenberger, Insertable cardiac event recorder in detection of atrial fibrillation after cryptogenic stroke: an audit report, Stroke 44 (2013) 2007–2009. [10] E. Rojo-Martinez, M. Sandín-Fuentes, A.I. Calleja-Sanz, et al., High performance of an implantable Holter monitor in the detection of concealed paroxysmal atrial fibrillation in patients with cryptogenic stroke and a suspected embolic mechanism, Rev. Neurol. 57 (2013) 251–257. [11] L.M. Christensen, D.W. Krieger, S. Højberg, et al., Paroxysmal atrial fibrillation occurs often in cryptogenic ischaemic stroke. Final results from the SURPRISE study, Eur. J. Neurol. 21 (2014) 884–889. [12] A.M. Sinha, H.C. Diener, C.A. Morillo, et al., Cryptogenic Stroke and underlying Atrial Fibrillation (CRYSTAL AF): design and rationale, Am. Heart J. 160 (2010) 36–41. [13] T. Sanna, H.C. Diener, R.S. Passman, et al., Cryptogenic stroke and underlying atrial fibrillation, N. Engl. J. Med. 370 (2014) 2478–2486. [14] P.D. Ziegler, J.D. Rogers, S.W. Ferreira, et al., Real world experience with insertable cardiac monitors to find atrial fibrillation in cryptogenic stroke, Cerebrovasc. Dis. 40 (2015) 175–181. [15] G. Hindricks, E. Pokushalov, L. Urban, et al., Performance of a new leadless implantable cardiac monitor in detecting and quantifying atrial fibrillation: results of the XPECT trial, Circ Arrhythm Electrophysiol. 3 (2010) 141–147. [16] H. Pürerfellner, E. Pokushalov, S. Sarkar, et al., P-wave evidence as a method for improving algorithm to detect atrial fibrillation in insertable cardiac monitors, Heart Rhythm. 11 (2014) 1575–1583. [17] W. Choe, R.S. Passman, J. Brachmann, et al., A comparison of atrial fibrillation monitoring strategies after cryptogenic stroke: results from the CRYSTAL AF trial, Am. J. Cardiol. 116 (2015) 889–893.
5
[18] J. Brachmann, C.A. Morillo, T. Sanna, et al., Uncovering atrial fibrillation beyond short-term monitoring in cryptogenic stroke patients: three-year results from the cryptogenic stroke and underlying atrial fibrillation trial, Circ Arrhythm Electrophysiol. 9 (2016), e003333. [19] H.C. Diener, J.D. Easton, C.B. Granger, et al., Design of Randomized, double-blind, Evaluation in secondary Stroke Prevention comparing the EfficaCy and safety of the oral Thrombin inhibitor dabigatran etexilate vs. acetylsalicylic acid in patients with Embolic Stroke of Undetermined Source (RE-SPECT ESUS), Int. J. Stroke 10 (2015) 1309–1312. [20] R.G. Hart, M. Sharma, H. Mundl, et al., Rivaroxaban for secondary stroke prevention in patients with embolic strokes of undetermined source: design of the NAVIGATE ESUS randomized trial, Eur Stroke J. 1 (2016) 146–154. [21] R.G. Hart, H.C. Diener, S.B. Coutts, et al., Embolic strokes of undetermined source: the case for a new clinical construct, Lancet Neurol. 13 (2014) 429–438. [22] W.N. Kernan, B. Ovbiagele, H.R. Black, et al., Guidelines for the prevention of stroke in patients with stroke and transient ischemic attack: a guideline for healthcare professionals from the American Heart Association/American Stroke Association, Stroke 45 (2014) 2160–2236. [23] P. Kirchhof, S. Benussi, D. Kotecha, et al., 2016 ESC guidelines for the management of atrial fibrillation developed in collaboration with EACTS, Eur. Heart J. 37 (2016) 2893–2962. [24] A.S. Go, E.M. Hylek, K.A. Phillips, et al., Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial Fibrillation (ATRIA) Study, JAMA 285 (2001) 2370–2375. [25] R. Wachter, M. Weber-Krüger, J. Seegers, et al., Age-dependent yield of screening for undetected atrial fibrillation in stroke patients: the Find-AF study, J. Neurol. 260 (2013) 2042–2045. [26] A. Diamantopoulos, L.M. Sawyer, G.Y. Lip, et al., Cost-effectiveness of an insertable cardiac monitor to detect atrial fibrillation in patients with cryptogenic stroke, Int. J. Stroke 11 (2016) 302–312. [27] T.V. Glotzer, P.D. Ziegler, Cryptogenic stroke: is silent atrial fibrillation the culprit? Heart Rhythm. 12 (2015) 234–241. [28] S. Swiryn, M.V. Orlov, D.G. Benditt, et al., Clinical implications of brief devicedetected atrial tachyarrhythmias in a cardiac rhythm management device population: results from the registry of atrial tachycardia and atrial fibrillation episodes, Circulation 134 (2016) 1130–1140. [29] H. Kamel, B.B. Navi, L. Elijovich, et al., Pilot randomized trial of outpatient cardiac monitoring after cryptogenic stroke, Stroke 44 (2013) 528–530. [30] C.R. Vasamreddy, D. Dalal, J. Dong, et al., Symptomatic and asymptomatic atrial fibrillation in patients undergoing radiofrequency catheter ablation, J. Cardiovasc. Electrophysiol. 17 (2006) 134–139.
Please cite this article as: P.D. Ziegler, et al., Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.06.039
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population☆ Paul D. Ziegler a,⁎, John D. Rogers b, Scott W. Ferreira c, Allan J. Nichols d, Mark Richards e, Jodi L. Koehler a, Shantanu Sarkar a a
Medtronic Cardiac Rhythm Heart Failure, Mounds View, MN, United States Scripps Clinic, La Jolla, CA, United States St. Louis University, St. Louis, MO, United States d OhioHealth Heart & Vascular Physicians, Columbus, OH, United States e ProMedica Physicians Cardiology, Toledo, OH, United States b c
a r t i c l e
i n f o
Article history: Received 24 April 2017 Received in revised form 5 June 2017 Accepted 9 June 2017 Available online xxxx Keywords: Atrial fibrillation Cryptogenic stroke Insertable cardiac monitor Stroke (ischemic)
a b s t r a c t Background: The long-term incidence of atrial fibrillation (AF) in cryptogenic stroke (CS) patients has been explored in carefully controlled clinical trials but real-world data are limited. We investigated the two-year incidence of AF in real-world clinical practice among a large cohort of patients with an insertable cardiac monitor (ICM) placed for AF detection following CS. Methods: Patients in the de-identified Medtronic Discovery™ Link database who received an ICM (Reveal LINQ™) for the purpose of AF detection following CS were included and monitored for up to 2 years. All detected AF episodes (≥2 min) were adjudicated. We quantified the AF detection rate using Kaplan-Meier survival estimates, analyzed the median time to initial detection of AF, and simulated the ability of various intermittent monitoring strategies to detect AF. Results: A total of 1247 patients (65.3 ± 13.0 years, 53% male) were included and followed for 579 ± 222 days. AF episodes (n = 4183) were detected in 238 patients, resulting in an AF detection rate of 21.5% at 2 years. The median time to AF detection was 112 [IQR 35–293] days. Intermittent monitoring for AF detection was inferior to continuous ICM monitoring with sensitivities ranging from 2.9% (annual 24-hour Holter) to 29.9% (quarterly 7-day Holters), p b 0.001. Conclusions: AF episodes were detected via continuous monitoring with ICMs in approximately 1 of every 5 CS patients within 2 years of follow-up. The vast majority of patients with AF would not have been detected with conventional external ambulatory monitors. ICMs should therefore be considered in the evaluation of CS patients. © 2017 Published by Elsevier Ireland Ltd.
1. Introduction Despite thorough evaluation following an ischemic stroke, a definitive cause is not identified in up to one-third of patients [1]. A significant proportion of these cryptogenic stroke survivors may have underlying atrial fibrillation (AF) which can be difficult to diagnose with conventional monitoring strategies since the arrhythmia is commonly asymptomatic [2] and often occurs sporadically [3,4]. Timely diagnosis of AF has important therapeutic consequences as this generally prompts a
☆ All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. ⁎ Corresponding author at: Medtronic Cardiac Rhythm Heart Failure, 8200 Coral Sea Street NE, Mailstop MVN41, Mounds View, MN 55112, United States. E-mail address: [email protected] (P.D. Ziegler).
change in pharmacologic treatment from antiplatelets to anticoagulants, given the superiority of the latter in preventing strokes among patients with AF [5]. The advent of small insertable cardiac monitors [6] (ICM) has improved our ability to diagnose infrequent arrhythmias such as AF in cryptogenic stroke patients as evidenced by numerous smaller singlecenter reports with relatively brief follow-up [7,8,9,10,11]. The CRYSTAL-AF study was a moderately sized trial which randomized patients with a recent cryptogenic stroke to standard of care monitoring or intensive monitoring with an ICM [12]. While the trial clearly demonstrated the superiority of monitoring with an ICM compared to standard of care monitoring [13], the study reflected the specific conditions and inclusion/exclusion criteria set forth by the trial design [12]. We previously reported the short-term (6-month) incidence of AF in a large, real-world cohort of patients with an ICM placed for AF
http://dx.doi.org/10.1016/j.ijcard.2017.06.039 0167-5273/© 2017 Published by Elsevier Ireland Ltd.
Please cite this article as: P.D. Ziegler, et al., Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.06.039
2
P.D. Ziegler et al. / International Journal of Cardiology xxx (2017) xxx–xxx
detection following cryptogenic stroke [14]. In the current report, we present the two-year incidence of AF in cryptogenic stroke patients as detected by ICMs in real-world clinical practice. 2. Methods We included all patients who received an ICM (Reveal LINQ™, Medtronic, MN) to monitor for the presence of AF after a cryptogenic stroke between February 24, 2014 and July 9, 2014. At the time of initial device configuration, implanting physicians designated that the reason for monitoring was a cryptogenic stroke. Device data was collected in the de-identified DiscoveryLink™ database along with patient's age and gender. Patients were monitored for up to 24 months after device insertion and all patients consented to the use of their data for research purposes. 2.1. Device characteristics and performance The Reveal LINQ™ ICM is a small (length 44.8 mm, width 7.2 mm, thickness 4.0 mm, volume 1.2 cm3, weight 2.5 g) cardiac monitor that is inserted into the subcutaneous tissue over the heart. Dedicated incision and insertion tools are used to form a small opening in the skin, create a uniformly sized pocket, and guide placement of the device under the skin. The device is typically inserted at 45 degrees relative to the sternum above the 4th intercostal space in the V2–V3 electrode orientation. ECG signals are measured between two electrodes located on each end of the device. Details of the AF detection algorithm operation [6] and performance [15,16] have been reported previously. Briefly, the AF detection algorithm analyzes changes in ventricular conduction over 2-minute intervals for patterns of irregularity and incoherence. Evidence of AF is reduced if p-waves are detected. At the end of each 2-minute interval, the device classifies the rhythm as sinus rhythm or AF. Performance of the AF detection algorithm has been evaluated in several studies by comparing it to simultaneously collected Holter data. In a population with known AF, the algorithm correctly identified the presence of AF in 96.1% of patients with AF (sensitivity) and correctly excluded the absence of AF in 97.4% (negative predictive value) [15]. Further improvements were made to reduce the duration of false positive detections by 55% and the number of false positive detections by 46% with no impact on detection sensitivity [16]. In addition to storing the cumulative time spent in AF for each discrete day of followup, the device also stores ECG waveforms for a portion of up to 14 detected AF episodes. Subsequently detected AF episodes then overwrite the earliest AF episode if the memory is not refreshed by in-office interrogation or remote transmission of the device data. All patients transmitted device data remotely via the Medtronic CareLink™ remote monitoring network. The device will automatically initiate wireless transmission of a portion of the ECG waveform every night for the longest AF episode recorded during the prior day. Patients may also initiate a full transmission of all AF episodes contained within the device's memory at any time. All recorded AF episodes were adjudicated by a single, blinded reviewer. The entire duration of the episode was considered to be either true or false based on the segment of the ECG waveform that was stored at the episode's onset. Only AF episodes determined to be true AF were included in the analysis. 2.2. Intermittent monitoring simulations The methodology for simulating the performance of various intermittent monitoring strategies from continuous monitoring data has been described elsewhere in detail [3,17]. In brief, AF detection data from a subset of days corresponding to the particular monitoring strategy were selected from the continuous daily AF detection memory of the ICM. The intermittent monitoring strategies that were evaluated included a single 24-hour Holter, 48-hour Holter, 7-day Holter, 21-day event recorder, 30-day event recorder, quarterly 24-hour Holters, quarterly 48-hour Holters, quarterly 7-day Holters, and monthly 24-hour Holters. For each intermittent monitoring strategy simulation, the initial day of monitoring was selected at random from a uniform distribution within the first 14 days following insertion of the ICM. For example, in the case of the 7-day Holter simulation, we examined if the ICM detected AF on any day among 7 consecutive days which began within the first 14 days (selected at random) after placement of the device. In the case of monitoring scenarios which were repeated at quarterly or monthly intervals (for the first year of follow-up), all subsequent monitoring periods occurred at a fixed interval from the randomly selected initial day of monitoring. For example, simulation of quarterly 24-hour Holter monitoring was performed by determining if the ICM detected AF on a day selected at random from days 1–14 following device insertion or on days 90, 180, and 270 from that initially selected day. The mean values of sensitivity and negative predictive value for each intermittent monitoring strategy are reported after repeating the simulations 10,000 times to ensure robustness. Sensitivity measures the proportion of patients with AF detected by the ICM who would also have been identified as having AF by intermittent monitoring. Negative predictive value measures the proportion of patients without AF detected by intermittent monitoring who were correctly identified as being free from AF based on the ICM data.
AF detection rates were also computed using AF episode duration thresholds of 6 min, 30 min, and 60 min. The median duration of the longest AF episode and the median time from device insertion to detection of the first AF episode were calculated. The t-test was used to compare the age of patients with AF episodes to those without AF episodes while an ANOVA model was used to compare the age of patients with different durations of their longest AF episode (b1 h, 1–4 h, 4–12 h, or N12 h). The log-rank test was used to compare the rate of AF detection between patients N65 years of age and patients ≤65 years of age. Comparisons of the proportion of patients with AF identified via intermittent monitoring strategies vs. continuous monitoring with the ICM were performed using the McNemar test. Continuous variables are reported as mean and standard deviation (SD) or median [Inter-Quartile Range (IQR)] while discrete variables are reported as counts and percentages, as appropriate. Statistical significance was assigned for p-values b0.05. All analyses were performed with SAS software version 9.4 (SAS Institute, Cary, NC, USA).
3. Results We studied 1247 patients (mean age 65.3 [SD 13.0] years, 53% male) over an average follow-up of 579 [SD 222] days. During this time, a total of 4183 episodes of AF were detected by the ICM in 238 patients. Multiple AF episodes were detected among 182 patients (76.5%) while a single AF episode was detected among the remaining 56 patients (23.5%). The median number of AF episodes detected among patients experiencing AF was 4 [IQR 2–14]. 3.1. AF detection rates and durations The rate of AF detection (≥2 min) increased from 4.6% at 30 days to 12.2% at 6 months, 16.3% at 12 months, and 21.5% at 24 months. Using more stringent AF episode duration thresholds resulted in AF detection rates of 19.4% (≥ 6 min), 17.9% (≥ 30 min), and 16.2% (≥ 60 min) at 24 months [Fig. 1]. The median duration of the longest AF episode experienced among patients with AF was 4.0 [IQR 0.8–12.7] hours. The distribution of the longest AF episode duration and the distribution of all adjudicated AF episode durations are presented in Fig. 2a and b, respectively. Among the 182 patients with multiple episodes of AF, 37 (20%) experienced their longest episode as their first AF episode while 10 patients (5%) had no change in AF episode duration and 135 (74%) had their longest episode at some point after the first AF episode. The median duration of the first detected episode among these 182 patients was 1.22 [0.20–4.50] hours while the duration of the longest subsequent AF episode was 5.25 [0.87–16.37] hours (p b 0.001), indicating that episode durations increased following the initial episode. 3.2. Role of age on AF detection Age data were available on 1219 of the 1247 patients studied (97.8%). Patients in whom AF was detected were significantly older than patients in whom AF was not detected (71.8 [SD 10.5] vs. 63.8 [SD 13.1] years, p b 0.001). Additionally, the AF detection rate at 24 months among patients N65 years old was significantly higher than among patients ≤ 65 years old (26.6% vs. 10.7%, p b 0.001). However,
2.3. Statistical analysis Kaplan-Meier survival estimates were used to compute AF detection rates at 30 days, 6 months, 12 months, and 24 months using a 2 minute duration criterion. At 24 months,
Fig. 1. Time to detection of AF using episode duration thresholds of 2, 6, 30, and 60 min.
Please cite this article as: P.D. Ziegler, et al., Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.06.039
P.D. Ziegler et al. / International Journal of Cardiology xxx (2017) xxx–xxx
3
Fig. 2. Distribution of maximum episode durations among 238 patients with AF detected (left panel) and all episode durations among 4183 episodes detected (right panel).
the median time to AF detection was similar between patients N65 years and ≤ 65 years (110.0 [IQR 32.0–294.0] days vs. 120.5 [IQR 72.0–278.0] days, p = 0.77). Moreover, there were no significant differences in age between patients whose longest AF episode was b1 h (70.3 [SD 11.8] years), 1–4 h (71.0 [SD 11.4] years), 4–12 h (73.1 [SD 9.5] years), or N 12 h (72.9 [SD 8.9] years), p = 0.40.
causes have AF detected by 24 months, the vast majority of these patients (approximately 4 out of 5) actually do not have AF detected within this timeframe despite continuous arrhythmia monitoring with an ICM. Therefore, unless the novel oral anticoagulants being tested also protect against non-AF related stroke, the risks of bleeding associated with this treatment strategy may outweigh the potential benefits in ischemic stroke reduction.
3.3. Impact of monitoring duration and simulations of intermittent monitoring 4.1. AF detection rates and monitoring durations The median time from device insertion to AF detection was 112 [IQR 35–293] days. If monitoring had stopped at 30 days after device insertion, 78.6% of patients who ultimately had AF detected would have gone undiagnosed. Between months 6–12 of follow-up, AF was detected for the first time in 45 additional patients (30% more patients identified at 12 months vs. 6 months on a relative basis). Furthermore, between months 12–24 of follow-up, AF was detected for the first time in yet an additional 45 patients (23% more patients identified at 24 months vs. 12 months on a relative basis). Simulations of various intermittent monitoring strategies revealed that the sensitivity, compared to the ICM, for detecting AF in this cryptogenic stroke population ranged from 2.9% (annual 24-hour Holter) to 29.9% (quarterly 7-day Holters) [Fig. 3]. Negative predictive values, compared to the ICM, ranged from 86.3% (annual 24-hour Holter) to 89.7% (quarterly 7-day Holters) [Fig. 3]. Both the sensitivity and negative predictive value for all intermittent monitoring strategies evaluated were significantly lower than continuous arrhythmia monitoring with the ICM (p b 0.001).
The AF detection rate in our study at 30 days of follow-up was 4.6% and this increased to 21.5% at 24 months of follow-up. Therefore, 78.6% of patients who were eventually found to have AF would not have been diagnosed if monitoring was only performed for up to 30 days as suggested by the current AHA/ASA guidelines [22]. The European Society of Cardiology AF guidelines [23], which were published since the release of the CRYSTAL-AF trial results, place a higher emphasis on long-term monitoring with a Class IIa, Level of Evidence B recommendation for additional ECG monitoring via ICMs to be considered in stroke patients.
4. Discussion This analysis represents the largest cohort of cryptogenic stroke patients who have been continuously monitored via ICMs for up to two years. The main finding of the study is that AF was diagnosed in approximately 1 of every 5 cryptogenic stroke patients (21.5%) by 24 months of follow-up. The AF detection rate in our study was very similar to the AF detection rate observed in the CRYSTAL-AF trial at 24 months (21.1%) [18] despite slight differences in the average age of the cohorts, the work-up required before a stroke was deemed cryptogenic, and how data were transmitted by the respective ICMs utilized in the studies. Recently, trials have been initiated [19,20] to evaluate whether prophylactic oral anticoagulation in patients with embolic strokes of undetermined source (ESUS) [21] is superior to antiplatelet therapy for recurrent stroke prevention. An underlying premise of these studies is that a significant proportion of ESUS patients have undiagnosed AF which would be better treated with oral anticoagulation compared to antiplatelet medication. Our study suggests that although a clinically meaningful percentage of patients with strokes due to undetermined
Fig. 3. Sensitivity and negative predictive values of various intermittent monitoring strategies for detection of AF compared to continuous monitoring with an ICM.
Please cite this article as: P.D. Ziegler, et al., Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.06.039
4
P.D. Ziegler et al. / International Journal of Cardiology xxx (2017) xxx–xxx
4.2. AF episode durations While the ICM used in this study is very sensitive to detecting AF episodes as brief as 2 min in duration [16], we found that most patients with device-detected episodes of AF experienced at least occasional episodes of much longer duration. As shown in Figs. 2, 74% of patients with AF experienced an episode longer than 1 h at some point during the 24-month follow-up. Furthermore, 43% experienced an episode longer than 6 h and 17% experienced episodes that persisted for longer than 24 h. These are likely underestimates of the true percentage of patients who met these episode duration thresholds since devices such as the ICMs used in this study may sometimes undersense the cardiac signal, which can result in partitioning a longer AF episode into multiple shorter AF episodes. Only 11% of patients with AF (or 2% of the overall population) had AF episodes which never exceeded 6 min in duration. We observed that among patients with multiple AF episodes, the longest AF episode generally occurred after the initial AF episode. This potentially supports the concept of atrial remodeling and additional study is required to determine if early active treatment of AF in this population could slow the progression of the disease. 4.3. Role of age on AF detection It is well established that the prevalence of AF increases sharply with advancing age in both the general population [24] and patients who have had a recent ischemic stroke [25]. Not surprisingly, we also detected AF more frequently among older (N65 years) compared to younger (≤ 65 years) cryptogenic stroke patients. However, there was no difference in the median time to initial detection of AF between older (110 [IQR 32–294] days) and younger (120.5 [72–278] days) patients, suggesting that short-term monitoring with external devices may not be a viable option regardless of age. In addition, a recent costeffectiveness analysis of ICMs for the detection of AF in cryptogenic stroke patients indicated that ICMs may actually be more costeffective in younger patients due to their longer life expectancy and a potential increase in quality-adjusted-life-years (QALYs) gained [26].
an ICM. Compared to the landmark CRYSTAL-AF study, this analysis represents N5 times as many patients with ICM data. Another strength is the extensive follow-up duration. A recent review article summarized many of the studies published in the last several years which have explored the incidence of AF via ICMs in cryptogenic stroke populations [27]. Most of these studies followed patients for one year or less. Our follow-up of up to 24 months, coupled with the large sample size, represents the most patient-years of continuously monitored cryptogenic stroke data ever assembled. This study also reflects actual clinical practice, rather than an artificial environment imposed by clinical trial inclusion criteria or protocols. As such, it is indicative of what might be expected when clinicians incorporate such a monitoring strategy into their practice. This study also has several limitations. Since the data were extracted from a de-identified device database, we lack many demographic characteristics, medication data or clinical actions taken, and clinical outcomes for these patients. The definition of cryptogenic stroke was also left to the discretion of the implanting physician and may include more heterogeneity than a rigorously controlled clinical trial. We may also have failed to detect extremely brief episodes of AF (b2 min) due to the technical limitations of the AF detection algorithm in the ICM. However, the clinical importance of such brief episodes is unknown. The RATE Registry recently demonstrated that extremely short episodes of AF were not associated with an increased risk of clinical events compared to patients without documented AF [28]. Furthermore, data from the CRYSTAL-AF trial established that cryptogenic stroke patients with initial AF episodes which are brief in duration often have longer duration AF episodes in the future [18]. This suggests that detection of AF in many patients with extremely brief episodes is likely to be only slightly delayed rather than missed altogether. Finally, our simulations of intermittent monitoring strategies assumed 100% patient compliance with the intended monitoring regimen. In reality, it has been shown that patient compliance with external monitoring devices decreases as the burden of monitoring (i.e., the monitoring frequency and duration) increases [29,30]. Therefore the sensitivities reported here are likely overestimates of what might be expected in clinical practice, particularly for the more aggressive monitoring schemes.
4.4. Value of long-term monitoring 5. Conclusion We previously reported AF detection rates in this cohort at 6 months of follow-up [14]. The current report significantly extends this follow-up and finds that the AF detection rate has increased from 12.2% at 6 months to 21.5% at 24 months, a relative increase of 76%. As AF is detected increasingly further from the cryptogenic stroke, it becomes more challenging to infer causality between AF and the index stroke event. However, the primary clinical goal remains prevention of stroke recurrence rather than definitive identification of the cause of the index stroke. In that context, detection of AF at any time post-stroke is crucial since oral anticoagulation is clearly indicated in AF patients with a prior stroke by current guidelines [22,23]. Where guidelines may be lacking is in the recommendation for arrhythmia monitoring following an ischemic stroke. The current AHA/ASA guidelines suggest that up to 30 days of arrhythmia monitoring is reasonable within the first 6 months following a stroke or transient ischemic attack (TIA) of unknown cause [22]. Our data demonstrates that the vast majority (78.6%) of patients with AF would be missed if monitoring had stopped at 30 days. Furthermore, our simulations of various intermittent monitoring strategies show that all evaluated strategies were inferior to continuous monitoring with ICMs. These results validate the findings of Choe et al. from similar simulations that were performed with data from the CRYSTAL-AF study [17].
AF episodes were detected via continuous monitoring with ICMs in approximately 1 of every 5 cryptogenic stroke patients within 2 years of follow-up. Consistent with the CRYSTAL-AF trial, these results demonstrate that 30 days of monitoring is insufficient to capture the vast majority of cryptogenic stroke patients who have AF and suggest that current monitoring guidelines may be inadequate. Conflict of interest PDZ = Employee and shareholder (Medtronic). JDR = Consultant (Medtronic, Biotronik); Speaker's bureau (Medtronic, Biotronik). SWF = Speaker's bureau (Medtronic). AJN = Speaker's bureau (Medtronic); Shareholder (Boston Scientific, Medtronic). MR = Speaker's bureau (Biotronik, Medtronic, Boston Scientific, Boehringer Ingelheim, Janssen, Pfizer); Consultant (Biotronik, Boston Scientific, Janssen). JLK = Employee and shareholder (Medtronic). SS = Employee and shareholder (Medtronic). References
4.5. Strengths and limitations One of the key strengths of this study is the large sample size of cryptogenic stroke patients who were continuously monitored for AF with
[1] P.L. Kolominsky-Rabas, M. Weber, O. Gefeller, B. Neundoerfer, P.U. Heuschmann, Epidemiology of ischemic stroke subtypes according to TOAST criteria: incidence, recurrence, and long-term survival in ischemic stroke subtypes: a populationbased study, Stroke 32 (2001) 2735–2740.
Please cite this article as: P.D. Ziegler, et al., Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.06.039
P.D. Ziegler et al. / International Journal of Cardiology xxx (2017) xxx–xxx [2] S.A. Strickberger, J. Ip, S. Saksena, K. Curry, T.D. Bahnson, P.D. Ziegler, Relationship between atrial tachyarrhythmias and symptoms, Heart Rhythm. 2 (2005) 125–131. [3] P.D. Ziegler, J.L. Koehler, R. Mehra, Comparison of continuous versus intermittent monitoring of atrial arrhythmias, Heart Rhythm. 3 (12) (2006 Dec) 1445–1452. [4] E.I. Charitos, U. Stierle, P.D. Ziegler, et al., A comprehensive evaluation of rhythm monitoring strategies for the detection of atrial fibrillation recurrence: insights from 647 continuously monitored patients and implications for monitoring after therapeutic interventions, Circulation 126 (2012) 806–814. [5] R.G. Hart, J.L. Halperin, L.A. Pearce, et al., Lessons from the stroke prevention in atrial fibrillation trials, Ann. Intern. Med. 138 (2003) 831–838. [6] S. Sarkar, D. Ritscher, R. Mehra, A detector for a chronic implantable atrial tachyarrhythmia monitor, IEEE Trans. Biomed. Eng. 55 (2008) 1219–1224. [7] M.A. Ritter, S. Kochhäuser, T. Duning, et al., Occult atrial fibrillation in cryptogenic stroke: detection by 7-day electrocardiogram versus implantable cardiac monitors, Stroke 44 (2013) 1449–1452. [8] P.E. Cotter, P.J. Martin, L. Ring, E.A. Warburton, M. Belham, P.J. Pugh, Incidence of atrial fibrillation detected by implantable loop recorders in unexplained stroke, Neurology 80 (2013) 1546–1550. [9] T. Etgen, M. Hochreiter, M. Mundel, T. Freudenberger, Insertable cardiac event recorder in detection of atrial fibrillation after cryptogenic stroke: an audit report, Stroke 44 (2013) 2007–2009. [10] E. Rojo-Martinez, M. Sandín-Fuentes, A.I. Calleja-Sanz, et al., High performance of an implantable Holter monitor in the detection of concealed paroxysmal atrial fibrillation in patients with cryptogenic stroke and a suspected embolic mechanism, Rev. Neurol. 57 (2013) 251–257. [11] L.M. Christensen, D.W. Krieger, S. Højberg, et al., Paroxysmal atrial fibrillation occurs often in cryptogenic ischaemic stroke. Final results from the SURPRISE study, Eur. J. Neurol. 21 (2014) 884–889. [12] A.M. Sinha, H.C. Diener, C.A. Morillo, et al., Cryptogenic Stroke and underlying Atrial Fibrillation (CRYSTAL AF): design and rationale, Am. Heart J. 160 (2010) 36–41. [13] T. Sanna, H.C. Diener, R.S. Passman, et al., Cryptogenic stroke and underlying atrial fibrillation, N. Engl. J. Med. 370 (2014) 2478–2486. [14] P.D. Ziegler, J.D. Rogers, S.W. Ferreira, et al., Real world experience with insertable cardiac monitors to find atrial fibrillation in cryptogenic stroke, Cerebrovasc. Dis. 40 (2015) 175–181. [15] G. Hindricks, E. Pokushalov, L. Urban, et al., Performance of a new leadless implantable cardiac monitor in detecting and quantifying atrial fibrillation: results of the XPECT trial, Circ Arrhythm Electrophysiol. 3 (2010) 141–147. [16] H. Pürerfellner, E. Pokushalov, S. Sarkar, et al., P-wave evidence as a method for improving algorithm to detect atrial fibrillation in insertable cardiac monitors, Heart Rhythm. 11 (2014) 1575–1583. [17] W. Choe, R.S. Passman, J. Brachmann, et al., A comparison of atrial fibrillation monitoring strategies after cryptogenic stroke: results from the CRYSTAL AF trial, Am. J. Cardiol. 116 (2015) 889–893.
5
[18] J. Brachmann, C.A. Morillo, T. Sanna, et al., Uncovering atrial fibrillation beyond short-term monitoring in cryptogenic stroke patients: three-year results from the cryptogenic stroke and underlying atrial fibrillation trial, Circ Arrhythm Electrophysiol. 9 (2016), e003333. [19] H.C. Diener, J.D. Easton, C.B. Granger, et al., Design of Randomized, double-blind, Evaluation in secondary Stroke Prevention comparing the EfficaCy and safety of the oral Thrombin inhibitor dabigatran etexilate vs. acetylsalicylic acid in patients with Embolic Stroke of Undetermined Source (RE-SPECT ESUS), Int. J. Stroke 10 (2015) 1309–1312. [20] R.G. Hart, M. Sharma, H. Mundl, et al., Rivaroxaban for secondary stroke prevention in patients with embolic strokes of undetermined source: design of the NAVIGATE ESUS randomized trial, Eur Stroke J. 1 (2016) 146–154. [21] R.G. Hart, H.C. Diener, S.B. Coutts, et al., Embolic strokes of undetermined source: the case for a new clinical construct, Lancet Neurol. 13 (2014) 429–438. [22] W.N. Kernan, B. Ovbiagele, H.R. Black, et al., Guidelines for the prevention of stroke in patients with stroke and transient ischemic attack: a guideline for healthcare professionals from the American Heart Association/American Stroke Association, Stroke 45 (2014) 2160–2236. [23] P. Kirchhof, S. Benussi, D. Kotecha, et al., 2016 ESC guidelines for the management of atrial fibrillation developed in collaboration with EACTS, Eur. Heart J. 37 (2016) 2893–2962. [24] A.S. Go, E.M. Hylek, K.A. Phillips, et al., Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial Fibrillation (ATRIA) Study, JAMA 285 (2001) 2370–2375. [25] R. Wachter, M. Weber-Krüger, J. Seegers, et al., Age-dependent yield of screening for undetected atrial fibrillation in stroke patients: the Find-AF study, J. Neurol. 260 (2013) 2042–2045. [26] A. Diamantopoulos, L.M. Sawyer, G.Y. Lip, et al., Cost-effectiveness of an insertable cardiac monitor to detect atrial fibrillation in patients with cryptogenic stroke, Int. J. Stroke 11 (2016) 302–312. [27] T.V. Glotzer, P.D. Ziegler, Cryptogenic stroke: is silent atrial fibrillation the culprit? Heart Rhythm. 12 (2015) 234–241. [28] S. Swiryn, M.V. Orlov, D.G. Benditt, et al., Clinical implications of brief devicedetected atrial tachyarrhythmias in a cardiac rhythm management device population: results from the registry of atrial tachycardia and atrial fibrillation episodes, Circulation 134 (2016) 1130–1140. [29] H. Kamel, B.B. Navi, L. Elijovich, et al., Pilot randomized trial of outpatient cardiac monitoring after cryptogenic stroke, Stroke 44 (2013) 528–530. [30] C.R. Vasamreddy, D. Dalal, J. Dong, et al., Symptomatic and asymptomatic atrial fibrillation in patients undergoing radiofrequency catheter ablation, J. Cardiovasc. Electrophysiol. 17 (2006) 134–139.
Please cite this article as: P.D. Ziegler, et al., Long-term detection of atrial fibrillation with insertable cardiac monitors in a real-world cryptogenic stroke population, Int J Cardiol (2017), http://dx.doi.org/10.1016/j.ijcard.2017.06.039