Clinical evaluation of a small implantable cardiac monitor with a long sensing vector

Abstract Introduction We conducted this study to show the safety and efficacy of a new implantable cardiac monitor (ICM), the BioMonitor 2 (Biotronik SE & Co. KG; Berlin, Germany), and to describe the arrhythmia detection performance. Methods The BioMonitor 2 has an extended sensing vector and is implanted close to the heart. It can transmit up to six subcutaneous electrocardiogram strips by Home Monitoring each day. We enrolled 92 patients with a standard device indication for an ICM in a single‐arm, multicenter prospective trial. Patients were followed for 3 months, and 48‐h Holter recordings were used to evaluate the arrhythmia detection performance. Results One patient withdrew consent and in one patient, the implantation failed. Two study device‐related serious adverse events were reported, satisfying the primary safety hypothesis. Implantations took 7.4 ± 4.4 min from skin cut to suture. At 1 week, the R‐wave amplitude was 0.75 ± 0.53 mV. In the 82 patients with completed Holter recordings, all patients with arrhythmias were correctly identified. False positive detections of arrhythmia were mostly irregular rhythms wrongly detected as atrial fibrillation (episode‐based positive predictive value 72.5%). Daily Home Monitoring transmission was 94.9% successful. Conclusion Safety and efficacy of the new device has been demonstrated. The detected R‐wave amplitudes are large, leading to a low level of inappropriate detections due to over‐ or undersensing.

help risk stratification in structural heart disease. [1][2][3][4][5] Remote monitoring has overcome the problem of limited ICM storage capacity (∼1 h of electrocardiogram [ECG]) and allows earlier detection of and response to clinically relevant arrhythmia. 2,3 Further device miniaturization, extended automation and features, and more accurate detection algorithms could expand the field of application for ICMs. 3 The novel BioMonitor 2 ICM (Biotronik SE & Co. KG, Berlin, Germany) is a successor device to the BioMonitor, 4,5 with a USB-stick-like shape, 60% reduced weight and volume, and twice longer sensing vector to increase R-wave amplitudes and improve diagnostic accuracy.
The objective of the prospective, multicenter, single-arm, nonrandomized BIO|MASTER.BioMonitor 2 study was to evaluate safety and efficacy of the BioMonitor 2 and its dedicated insertion tools.

Patients
Study patients had to be at least 18 years old; be able and willing to comply with study procedures, including remote monitoring surveillance; and fulfill any of the following: (1) having a standard indication for ICM such as unexplained syncope or other symptoms possibly caused by heart rhythm disturbances or being (2) currently planned for ICM-guided therapy management of paroxysmal AF, (3) indicated for catheter ablation of persistent AF, or (4) ablated for persistent AF within 4 weeks before enrolment.
Patients were excluded if they had any cardiac rhythm management device implanted (e.g., pacemaker), had life expectancy <6 months, were pregnant or breast-feeding or considering becoming pregnant during the study, or if they participated in another interventional clinical investigation.
All patients provided written informed consent. The study was done in compliance with good clinical practice guidelines and the Declaration of Helsinki, including approval of the study protocol by appropriate national and local ethics committees, and study registration with ClinicalTrials.gov, number NCT02565238.

Device studied
The device has a volume of 5 cc and a weight of 10 g ( Figure 1A).
Attached to the 55-mm × 15-mm × 6-mm rigid part is a flexible "antenna" of 33 mm, which can adapt to the shape of the body while extending the sensing vector to increase the signal amplitude. A dedicated fast insertion tool (FIT) set is provided to place the BioMonitor 2 into a subcutaneous pocket ( Figure 1B). Typical ICM positions, diagonal and vertical, are illustrated in Figure 1C.

Study protocol
During ICM implantation, the implanters evaluated handling characteristics of the FIT set, separately for the pocket tool and the lead support tool.
Most devices were programmed to standard settings: asystole duration ≥3 s; bradycardia rate ≤40 beats/min for at least 10 s; high ventricular rate ≥180 beats/min for at least 16 beats; for AF, R-R variability ≥12.5%, with confirmation time of 6 min; and sudden rate drop OFF.
Patients were followed for 3 months after ICM insertion. At the 1-week and 3-month follow-ups, R-wave amplitudes were measured, noise burden and other ICM data were retrieved, and adverse events and device deficiencies were assessed.
A continuous 48-h Holter-ECG obtained between the 1-week and 3-month follow-ups was used to validate the corresponding ICM detections. Adjudication was done by one of the authors who did not enroll study patients (DS). Characteristics of binary classification, such as positive predictive value (PPV), negative predictive value, sensitivity, and specificity, were calculated for different arrhythmia types using episode-based and patient-based approaches, and for AF, also a duration-based approach (see Appendix A).

Study endpoints
The primary endpoint was freedom from serious adverse device effects (SADEs) related to the BioMonitor 2 or FIT. The primary hypothesis was that the proportion of patients without SADE would be >90% from the beginning of implantation to study termination. The secondary hypothesis was that the mean R-wave amplitude at the 1-week followup would be greater than 0.3 mV, the historical result for the predecessor device. 4 Additional data of interest were handling characteristics of the FIT set, overall sensing behavior including noise burden, accuracy of arrhythmia detection by the ICM verified by Holter-ECG findings, Home Monitoring transmission performance, and adverse events.

Statistical methods
The sample size was calculated using the POWER procedure of SAS

RESULTS
Between September 2015 and July 2016, 92 patients were enrolled at 13 investigational sites in Germany (10 sites), Australia (one), Austria (one), and Czech Republic (one) (see Appendix B). Table 1 shows patient characteristics. One patient withdrew informed consent before implantation. The insertion of the BioMonitor 2 was successful in 90 of 91 patients (98.9%). Insertion failure in one patient was caused by bent pocket tool.
Of the 90 patients with ICM, 84 terminated the study regularly (93.3%) and six prematurely (6.7%). The reason for premature termination were pacemaker implantation (N = 3; asystole or intermittent complete atrioventricular block detected by ICM), loss to follow-up (N = 2), or withdrawal of consent (N = 1).

Primary hypothesis (safety)
The primary endpoint was evaluated in 91 patients undergoing ICM Data are mean ± standard deviation or number (percent). NYHA = New York Heart Association.

Secondary hypothesis (R-wave amplitude)
The R-wave measurements at 1 week were available in 79 patients. The mean R-wave amplitude of 0.75 ± 0.53 mV (median, 0.63; IQR, 0.41-0.97) was significantly higher than the historical result of 0.30 mV for the BioMonitor 4 (P < .001).

Handling characteristics of the insertion tool
In the 90 patients with successfully inserted ICM, the time from first skin cut to last suture was 7.4 ± 4.4 min (median, 6.1; IQR, 4.4-9.3).
One-third of this time was spent on ICM positioning (mean, 2.8 ±

Overall sensing behavior
There was no significant difference in the mean R-wave amplitude at

Accuracy of arrhythmia detection
Of

Home monitoring transmission performance
The ratio of the number of days with a transmitted message and the number of days in the transmission period (day of first transmitted message to day of last transmitted message of each patient) was 94.9% in pooled data for all patients.

Adverse events
Ten device-related adverse events were reported in eight patients.
Of the two serious events that were evaluated as primary endpoints, one was connected to an infection. Eight adverse device effects were nonserious, including acute (N = 2) and late (N = 2) pain or nausea, hematoma (N = 2), device migration (N = 1), and infection (N = 1).

DISCUSSION
The new ICM (BioMonitor 2) was developed to facilitate minimally invasive surgical procedure by reduced device size and to improve R-wave amplitudes and arrhythmia detection by a longer sensing vector. The present study demonstrated the safety and efficacy of the BioMonitor 2 and its FIT set.

Asystole
The safety was shown by an SADE-free rate of 97.8%, which is significantly higher than the hypothesized 90%. The efficacy was indicated by the increased R-wave amplitude, reliable arrhythmia detection, and good handling characteristics of the FIT set.

Sensing performance
The mean R-wave amplitude of 0.75 mV for the BioMonitor 2 is markedly larger than 0.3 mV for the predecessor device, 4 owing to the longer sensing vector and the implantation closer to the heart. It compares favorably with the other devices on the market (eg, 25% larger than Medtronic LINQ TM ) 8,9 and leads to an improved noise burden, which has been reduced considerably, from median 4.0%   9 We identified three of 82 (3.7%) patients with such episodes in our short-term observation.

Arrhythmia detection accuracy
False-positive detections, which are also reduced but not eradicated with the BioMonitor 2, will rarely lead to unnecessary treatments because ICM misclassifications can be largely overcome by manual analysis of the corresponding sECG snapshots by the physician. 5 Even if the snapshots are overwritten due to ICM memory limitation, they may be available in the remote monitoring system archive. 5  the workload connected to it. It should be kept in mind that the devices in this study were mostly set to standard programming, which can be optimized depending on the patient's indication.
The majority of inappropriate detections were caused by irregular rhythms detected as AF. One possibility to improve the performance of AF detection is to extend the minimum duration for an AF episode to be detected. 10 Unfortunately, the fact that our device programming was not standardized and our limited sample size limits our ability to report on differences in performance depending on detection criteria. reported in a similar approach. 12 The successor device to Reveal XT, Reveal LINQ, using both R-R interval variation and a new P-wave recognition algorithm for AF detection, demonstrated an improved sensitivity of 97.4% and specificity of 97.0% for patient-based analysis of AF. 13 In the duration-based approach, Reveal LINQ showed 98.4% sensitivity and 99.5% specificity, 13 compared with 93.6% and 99.2% for the BioMonitor 2, respectively (Table 4). However, realworld performance of devices is often worse than in controlled trials and depends on the AF incidence in the monitored population, the programmed sensitivity of AF algorithm, and the duration of detected AF episodes. 10 For other rhythm disturbances (asystole and bradycardia), longterm data from the BioMonitor 2 are needed to decide if there are major differences in detection accuracy between Reveal LINQ 9,14 and

Further results
The rating of the FIT set by BioMonitor 2 implanters was good or acceptable in the vast majority of patients, with the implantation procedure lasting for a median of 6.1 min (first cut to last suture).
The ICM miniaturization trend continues and allows safe, minimally invasive surgical procedures. 2,16,17 Although the Reveal LINQ (weight

Study limitations
The study had several limitations. First, it did not assess the long-term sensing performance, but it may be assumed that sensing remains sta-

CONCLUSION
The In the present study, this matrix of performance characteristics was calculated for the episode-based and the patient-based arrhythmia detection accuracy, as explained below. For AF, also durationbased analysis was made. All results were derived from the adjudicated 48-h Holter-ECG and BioMonitor 2 ICM findings.

Episode-based approach
In the episode-based approach, all episodes are counted equally, irrespective of the patient who had the episode. The start time and the end time of an arrhythmia episode never match exactly in the BioMon- In contrast to TP, FP, and FN, the TN episodes cannot be defined and calculated meaningfully in the episode-based approach. Therefore, only PPV and sensitivity can be determined for this approach, while NPV, specificity, and accuracy cannot be determined.

Patient-based approach
The patient-based approach refers to an analysis that classifies patients as with or without a certain arrhythmia type occurring during the 48-h Holter-ECG recording, and with or without BioMonitor 2detected episodes of this type during this period. A patient with some FN and at least one TP detection of AF counts as TP patient, because he or she has been correctly identified as having AF.
All characteristics of binary classification, including TN, can be calculated, where "detection" refers to the BioMonitor 2 and "episodes"

Duration-based analysis in AF
The clinical investigation protocol required to synchronize the clocks of the BioMonitor 2 and the Holter device at the time of Holter start.
The periods of AF as seen by the BioMonitor 2 and by Holter monitoring were aligned time wise; the timeline started and ended according to Holter. The AF episodes ongoing at Holter start were included. The following time periods were determined: True positive (TP): Both BioMonitor 2 and Holter see AF.
False positive (FP): BioMonitor 2 sees AF but Holter does not see AF.
True negative (TN): Both BioMonitor 2 and Holter do not see AF.
False negative (FN): BioMonitor 2 does not see AF while Holter sees AF.
The AF timelines according to adjudicated Holter findings in 82 patients were determined as follows: • In 67 patients without AF, the whole Holter timeline was "no AF." • In six patients with incessant AF, the whole Holter timeline was "AF." • Of nine patients with intermittent AF, raw Holter recordings were available at the time of this analysis in four patients.

○
In four patients with intermittent AF and raw Holter data, the Holter program was used to identify the onset and the end of AF episodes automatically. Episodes shorter than 6 min were discarded because they could not be detected by the ICM due to programmed AF confirmation time of 6 min.

○
In five patients with intermittent AF and no raw Holter data, the printout of the original analysis contained a graphical presentation of the periods in which the automatic analysis found AF. The start and the end times of these episodes were identified manually in the printouts.

■
In one patient, the Holter program did not recognize AF that was evident from the presence or absence of P-waves and from the changing cycle lengths in the ECG.
The decision of the adjudicator to classify this episode as AF overruled the automatic AF recognition of the Holter system.
The ICM-based AF timelines were derived from the adjudicated list of episodes stored in the ICM.