Author + information
- Received July 21, 2016
- Revision received October 11, 2016
- Accepted October 21, 2016
- Published online March 1, 2017.
- E. Kevin Heist, MD, PhDa,∗ (, )
- Andres Belalcazar, PhDb,
- Wyatt Stahl, BSc,
- Tom F. Brouwer, MD, MScd and
- Reinoud E. Knops, MDd
- aCardiac Arrhythmia Service, Massachusetts General Hospital, Boston, Massachusetts
- bIndependent Researcher, Minneapolis, Minnesota
- cBoston Scientific, St. Paul, Minnesota
- dDepartment of Clinical and Experimental Cardiology, Heart Center, Amsterdam Medical Center, University of Amsterdam, Amsterdam, the Netherlands
- ↵∗Reprint requests and correspondence:
Dr. E. Kevin Heist, Cardiac Arrhythmia Service, Massachusetts General Hospital, 55 Fruit Street, GRB 109, Boston, Massachusetts 02114.
Objectives This study determined the impact of subcutaneous implantable cardioverter-defibrillator (S-ICD) coil and generator position on defibrillation threshold (DFT).
Background S-ICD implantation can occasionally result in unacceptably high DFT. Implant position characteristics associated with high DFTs in S-ICD patients have not been fully elucidated.
Methods A 3.8-million-element computer model built from magnetic resonance images was used to simulate the electric fields that occur during defibrillation. Generator positions were tested from posterior to anterior in 4-cm increments. The left parasternal coil was tested with 0, 5, and 10 mm of underlying subcutaneous fat and the generator with 20 mm of underlying fat. The estimated DFT for the S-ICD was defined as the energy delivered when producing an electric field of 4 volts/cm in at least 95% of the ventricular myocardium.
Results Estimated DFTs were 22, 29, 64, and 135 joules for posterior, standard (lateral), mid-anterior, and anterior generator locations, respectively. Defibrillation thresholds were 29, 58, and 95 joules with 0, 5, and 10 mm subcoil fat, respectively, and 45 joules with 20 mm subgenerator fat. Combining anterior generator position with subcoil fat resulted in a very high DFT (379 joules). Shock impedance increased with both subcoil and subgenerator fat but was minimally affected by anterior/posterior generator position.
Conclusions The model suggests that an S-ICD implantation strategy involving posterior generator location and coil and generator directly over the fascia without underlying fat is likely to markedly lower DFTs with the S-ICD and assist in troubleshooting of patients with unacceptably high DFTs.
- defibrillation threshold
- subcutaneous implantable cardioverter-defibrillator
- transvenous implantable cardioverter-defibrillator
The subcutaneous implantable cardioverter-defibrillator (S-ICD) is now widely used for the treatment of life-threatening ventricular arrhythmias as an alternative to the transvenous ICD (T-ICD) (1,2). Studies demonstrate that most S-ICD shocks are successful in terminating ventricular tachycardia (VT) and ventricular fibrillation (VF) (1). The rate of successful S-ICD defibrillation is comparable to that of T-ICD, with the caveat that the S-ICD requires higher energy shocks than the T-ICD (3). In the IDE (Investigational Device Exemption) study of the S-ICD, at the time of implantation, 82.8% of patients had successful consecutive conversion of the initial 2 induced VF episodes after final device positioning, and all patients ultimately had successful defibrillation according to the testing protocol (requiring 2 successive successful conversions at 65 J) (4). An analysis of 111 clinical VT/VF events treated by the S-ICD demonstrated that 90.1% were successfully converted by the first shock (1), comparable to a T-ICD rate of 89.6% with dual coil leads in the MADIT-CRT (Multicenter Automatic Defibrillator Implantation Trial with Cardiac Resynchronization Therapy) study (5). Furthermore, 98.2% of the 111 events were successfully converted by the 5 available S-ICD shocks. Factors associated with the defibrillation threshold (DFT) of the T-ICD, including lead and generator positions, have been described previously (6). Implant factors associated with DFT of the S-ICD have not yet been fully elucidated, however.
In this study, we used a computer model to simulate defibrillation with the S-ICD to explore the impact of position of both the S-ICD coil and generator implants on DFT.
Computer modeling of defibrillation is a well-established technique representing the thorax anatomy by a 3-dimensional (3D) matrix of small circuit elements with appropriate tissue resistivities. Electrodes are applied that introduce currents. A numerical solution is obtained, resulting in voltages and gradients that can be measured anywhere in the thorax. Similar computer models have been described (7–14), including some focused on S-ICD simulation (8,9). The particular model we used has been described previously (7,15).
The computer model and magnetic resonance imaging (MRI) data set were licensed from an academic research laboratory (Bakken MIND Laboratory, University of Minnesota, Minneapolis, Minnesota). MRI data were de-identified. The subject was a 63-year-old male, 100 kg, 180 cm tall, with 80% stenosis in the left anterior descending coronary artery and normal ventricular size and function. A 1.5-T scanner (Sonata, Siemens, Malvern, Pennsylvania) was used. Images were obtained from abdomen to neck, and those at end-diastole represented the arrested heart. The images were manually segmented at the axial resolution of the MRI (1.5 mm), yielding a model with 3.8 million elements and 1.5 × 1.5 × 5 mm thoracic resolution (Figure 1). Each tissue was assigned an electrical resistivity value according to published studies (16–19) as follows: muscle was 225 ohms-cm, lung was 1,400 ohms-cm, blood was 150 ohms-cm, myocardium was 250 ohms-cm, liver and kidneys were 600 ohms-cm, and bone and fat were 2,000 ohms-cm. Resistivities were assumed to be isotropic, including myocardial fibers (20–22). Homogenous resistivity was used for each tissue type, as in similar studies (7–14). However, the model has sufficient resolution to allow a highly heterogeneous thorax, with distinction among tissues including lateral thoracic arteries, azygous vein, esophagus, parasternal cartilage, ribs, epicardial fat, mediastinal and retrosternal fat, and pulmonary vessels, among others (Figure 1). Other assumptions included a passive monodomain representation of the heart and thorax, with negligible capacitance. The latter assumption was based on the widely cited reference of Gabriel et al. (16). They surveyed and measured tissue resistance and capacitance properties (i.e., the permittivity which determines polarization and capacitance) across a large frequency range. After tabulating resistive and reactive parts of the complex impedance of tissues in our model at defibrillation modeling frequencies, we found that the electric field magnitude errors incurred by neglecting permittivity were <5%. In this manner, the resistive-only representation of the thorax was justified.
Validation of the model centered on thoracic electric field accuracy and not on determining a DFT, which when measured clinically, is a probabilistic variable and not a fixed number. As with other investigators using similar models, our bridge from an electric field to a DFT estimation was the critical mass concept, experimentally developed by Zipes et al. (23), Zhou et al. (24), and Witkowski et al. (25), discussed below. To confirm electric field accuracy, a comparison to actual measurements in the source subject with skin and esophageal electrodes was conducted. The subject swallowed 2 pairs of tethered bipolar electrodes (Arzco, Chicago, Illinois) to position behind the left atrium, using atrial electrogram guidance. The pair was then separated by pulling one tether 4 cm. A second pair of electrodes was placed on the skin at various positions on the thorax, and currents were injected into it. This allowed verification that near-cardiac electric fields caused by extrathoracic currents were simulated with less than 10% error. Additional tests in saline-filled elliptical tanks with actual S-ICD generators and coils (Boston Scientific, St. Paul, Minnesota) further ensured the accuracy of electric field simulations, errors were less than 2%. In addition, the model’s conventional structure and its history (7,15), along with verification that results were within clinical ranges, provided confidence in its accuracy.
In our computer simulations, DFTs were approximations. They were estimated to occur at the voltage at which 95% of the ventricular mass has greater than 4 volts/cm of electric field strength when simulating biphasic defibrillation, the critical mass concept cited previously. Estimated DFT results are given as energy delivered based on a fixed 50% tilt, 98-μF capacitor system. These parameters determine the S-ICD’s biphasic truncated exponential waveform depending on the presenting impedance. The device automatically adjusts the pulse duration of each phase of the waveform until the trailing edge voltage amplitude has decayed to 50% of the leading edge one. We report DFTs not in duration-corrected current terms but in energy terms according to widespread practice. The estimated DFT is calculated by adding the energy in both phases of the waveform, using standard electrical engineering equations for a capacitor discharge when the initial discharge voltage, impedance, capacitance, and tilt are known. The computer model determines the first 2 of these terms from the critical mass voltage and thoracic resistivities and the rest are known from the device specification (see discussion in Geddes et al. ).
An S-ICD generator (Boston Scientific) 82 × 64 × 12.7 mm was simulated. The coil was 80 mm long and 3 mm in diameter. For the standard placement, the S-ICD generator and coil were located according to their instructions for use. Namely, the generator was placed at the midaxillary line, centered at the height of the sixth rib, and the coil was placed at the left parasternal border, beginning 3 cm above the xiphoid process.
To test effects of generator position changes, it was moved from the standard lateral location by 4-cm increments as follows: posterior (−4 cm), midanterior (+4 cm), and anterior (+8 cm) and was also moved 4 cm cranially and 4 cm caudally. To test the effect of dilated cardiomyopathy in a separate model, the heart was manually enlarged following typical dilated cardiomyopathy radiological imaging; the left ventricular end-diastolic diameter (LVEDD) increased from 48 to 70 mm. To test hypertrophic cardiomyopathy effects, symmetric hypertrophy was simulated by increasing the LV wall thickness from 13 to 23 mm, and thereby decreasing the LVEDD from 48 to 29 mm. To test the effect of increasing fat thickness under the coil, the coil was moved from its standard position laying against the parasternal fascia, with 5 and 10 mm of intervening fat (the latter separation resulted in the coil being just under the skin). A right parasternal position of the coil was also modeled. The generator was separated from its standard position directly over the fascia of the serratus anterior muscles by a 20-mm layer of intervening fat (13 mm of fat remained between the generator and the skin). Estimated DFT and shock impedance were then modeled with each of these changes (Figure 2).
Model comparison with clinical measurements
For validation, the model was also analyzed by simulating a conventional T-ICD (using a single coil in the right ventricle and a left pre-pectoral generator), to verify that DFTs and shock impedances were within the range observed clinically. The T-ICD model had 7.7-joule DFT and 59-ohm shock impedance. In comparison, clinical values with a T-ICD are reported to be 11.1 ± 8.5 joules (3) and 62 ± 9 ohms (27), demonstrating that the model is in range of clinical experience with the T-ICD. For the S-ICD, the clinical DFT was 36.6 ± 19.8 joules (3). In the IDE study (4), the impedance data from 841 delivered shocks for 316 patients undergoing intraoperative testing demonstrated a mean shocking impedance of 78.8 ± 21.8 ohms with a range of 33 to 161 ohms (U.S. Food and Drug Administration submission documents). The computer simulation of the S-ICD resulted in an estimated DFT of 29 joules and an impedance of 83 ohms, which are well within the clinical values observed.
Impact of generator and coil placement on DFT
Compared to the estimated DFT for the standard coil and generator positions (29 J), displacement of the generator 4 cm posteriorly decreased DFT slightly, whereas displacement by 4 or 8 cm anteriorly resulted in marked elevations in DFT (Figure 3), as a result of shunting of the current anteriorly in the thorax (Figure 2). The dilated heart model demonstrated a similar impact of posterior/anterior generator positioning, with slightly reduced DFT values for any given generator position, while a hypertrophied heart produced estimated DFT values which were similar to the normal heart (Figure 3). Moving the generator 4 cm cranial or 4 cm caudal from the standard position resulted in small reductions in DFT compared to that in the standard position (Figure 4). In comparison to the standard left parasternal coil (20 mm left of the midline; DFT of 29 J), a right parasternal coil (20 mm right of midline, directly on the fascia) increased DFT slightly (37 J), whereas a more extreme displacement of the right parasternal coil (40 mm right of midline, also directly on the fascia) increased DFT more substantially (64 J).
Impact of subcoil and subgenerator fat on DFT
In the model, standard positioning involved the coil and generator directly against the underlying fascia, without intervening fat. We then modeled fat under the coil, demonstrating progressive increases in estimated DFT with 5 and 10 mm of subcoil fat (Figure 5). Twenty millimeters of subgenerator fat also increased DFT, although the impact of subgenerator fat was smaller than subcoil fat, due to the smaller electrode surface area of the coil compared to the generator. The DFT increased when subcoil and subgenerator fat were modeled together (Figure 5). To further study the impact of combined factors on DFT, we compared the impact of 8 cm anterior generator displacement and 10 mm subcoil fat, resulting in a very high estimated DFT (379 J), representing the least favorable implant position of the S-ICD system (Figure 6).
Impact of generator/coil placement and underlying fat on impedance
Our simulation modeled shock impedance in addition to estimated DFT for the generator and coil perturbations described above. Subcoil, and to a lesser degree, subgenerator fat resulted in substantial increases of shock impedance (Figure 7). Combining subcoil with subgenerator fat resulted in a further increase of impedance compared to either alone. In contrast, posterior (−4 cm) and anterior (+4 cm and +8 cm) displacement of the generator had negligible impact on shock impedance (Figure 7).
Our analysis resulted in several important findings. First, computer modeling of the S-ICD resulted in estimated DFT and shock impedance within the range of published clinical results with the device. Second, compared to a standard lateral generator position, a posterior one (by 4 cm) decreased the estimated DFT modestly, whereas anterior generator position (by 4 or 8 cm) substantially and progressively increased the DFT, with similar results in standard, dilated, and hypertrophic hearts. Third, moving the generator cranially or caudally by 4 cm or moving the coil to a right parasternal position only modestly impacted DFT. Fourth, subcoil fat and to a smaller degree subgenerator fat substantially increased the estimated DFT. Fifth, combining unfavorable factors (i.e., subcoil plus subgenerator fat or subcoil fat plus anterior generator position) markedly increased the estimated DFT. Sixth, shock impedance increased with subcoil and/or subgenerator fat but was not substantially affected by anterior/posterior generator position.
The results highlight the fact that there are 2 main conditions required to maintain effective defibrillation with the S-ICD, as is also true with external defibrillators and T-ICDs: first, a good current pathway that avoids shunting and traverses the heart, and, second, a low shock impedance to allow the necessary current through the thorax. The first condition is intuitive. One must place the 2 electrodes across the heart to maximize current through it and avoid shunting current away from this pathway. In our model, anterior generator positioning caused the bulk of the current to shunt through the anterior thorax, with reduction in the current required for defibrillation in the heart (particularly posteriorly). This anterior/posterior generator position had a large impact on estimated DFT but negligible impact on shock impedance.
The finding of slightly decreased DFT with the S-ICD for a dilated compared to a normal heart may seem surprising and contrary to typical findings with a T-ICD. It should be noted, however, that with a T-ICD, the electrical field generated by the shock “radiates” outward from the right ventricle shocking coil, and increased distance of the myocardium from the coil (as would be found in a dilated heart) results in lower field strengths encountered by the myocardium. In contrast, the S-ICD produces an electrical field through the thorax, including the heart. Cardiac dilation (which tends to result in replacement of higher impedance lung with lower impedance myocardium and intracardiac blood in the shocking path) decreases shock impedance and thereby increases electrical current through the myocardium. In regard to hypertrophic cardiomyopathy, the resistivity of myocardium and blood are similar (compared with fat or lung), so the replacement of intracardiac blood by hypertrophied myocardium has little impact on current flow through the myocardium and therefore little impact on estimated DFT.
The second condition for successful defibrillation (requiring low impedance to create effective current for defibrillation) relates to Ohm’s law (current = voltage/resistance), where the current will be reduced if the resistance is increased for a given device voltage output. Fat underlying the coil or generator increases the impedance in our model and largely through that mechanism increases the DFT, despite a vector between the coil and generator which traverses the heart. In reality, defibrillator outputs are not simple nor is defibrillation of the heart (6). Implantable cardioverter-defibrillators (both S-ICDs and T-ICDs) have compensation mechanisms that prolong the defibrillation pulse width if the impedance encountered is larger. Although lengthening the pulse duration via the tilt mechanism may ensure constant energy delivery with higher impedances, it does not imply a constant defibrillation efficacy (28). There is a known relationship between high shock impedance and reduced defibrillation efficacy (29). It should be noted that the coil in T-ICDs is in the blood pool of the right ventricle in a relatively similar impedance environment among patients, whereas in the S-ICD, the coil may be in contact with fascia/muscle (with favorable low impedance) or with subcutaneous fat (with unfavorable high impedance).
This study allows for relatively straightforward recommendations for device placement and for troubleshooting high DFTs. For patients with unusual cardiac or thoracic anatomy, device positioning which interposes the heart between the coil and generator will produce a shock vector traversing the heart which is likely to maximize the probability of successful defibrillation. Our model also suggests a modest elevation in the DFT for a right (as opposed to left) parasternal coil, although either left or right parasternal coil positioning is likely to be acceptable if the device is otherwise well positioned. If high DFTs are encountered, a relatively high lead impedance may suggest fat under the generator or coil, while anterior generator position can be identified only by anatomic landmarks and not by shock impedance. In addition, multiple factors may converge to result in unacceptably high DFT in a given patient; a slightly anterior generator position or small amount of subcoil fat may produce an acceptable DFT in isolation but may produce an unacceptably high DFT if combined in a given patient.
Although there are no large clinical studies evaluating predictors of successful defibrillation with the S-ICD, a published case describes failed defibrillation (impedance of 114 ohms) in a patient with both an anterior generator and subcoil fat, with successful defibrillation when the generator was repositioned posteriorly and the coil was moved against the fascia (impedance reduced to 80 ohms) (Figure 8), consistent with our study findings (30).
We believe that the general principles of successful S-ICD defibrillation presented here (a shock vector traversing the heart and minimization of subcoil and sub-can fat to reduce shock impedance) are broadly applicable to S-ICD implants. Therefore, our findings will improve the practitioner’s understanding of S-ICD defibrillation, particularly for those who are new to the procedure, and help in managing difficult DFT cases in clinical practice. Failure to defibrillate with the S-ICD (either in testing at implantation or during clinical VT/VF) is fortunately an issue of outliers, with most S-ICD patients never experiencing an unsuccessful shock, either during testing at implantation or during clinical ventricular arrhythmias (1,4,31).
A limitation of this study is that it represents an anatomy derived from a single patient. Despite this, it is likely that the major findings of our model are valid for most patients, based on basic principles of electrical circuits: anterior generator position will tend to short-circuit the current, and fat under the coil and/or generator will increase shock impedance. Our findings cannot be extrapolated to patients with extreme anatomic variation from the norm (i.e., dextrocardia or severe chest wall deformity).
Bidomain computer models can be used to predict DFTs with greater accuracy. They include important defibrillation variables that our simpler monodomain heart model does not, such as intracellular and extracellular currents, conductivity anisotropy, cellular membrane potentials and activation, virtual electrode polarization (affects arrhythmia re-initiation), and wave propagation in myocardium (32–34).
Our conventional model represents enough important factors to allow approximation of DFT versus device placement. Notable among them is the electric field strength to which the heart is subjected, a critical input for any model. For the dramatic perturbations that we are studying (device position and coil impedance mediated by fat), our model includes sufficient elements to accurately calculate the various experimental cardiac field strengths: true geometries derived from MRI, detailed organ and tissue resolution, true tissue resistivity values obtained from recognized experimental sources, and a validated electric field simulator. Future studies are needed to replicate our findings using bidomain models and ideally clinical data.
Our computer modeling suggests that lateral-to-posterior generator placement (in patients with typical cardiac and thoracic anatomy), with minimal fat underlying the coil and generator, reduces estimated DFT with the S-ICD. Subcoil and subgenerator fat but not anterior/posterior generator position results in increased shock impedance and decreases the effective current traversing the heart.
COMPETENCY IN MEDICAL KNOWLEDGE: The subcutaneous implantable cardioverter-defibrillator (S-ICD) is now in widespread clinical use for the treatment of life-threatening ventricular arrhythmia. During treatment of clinical ventricular arrhythmias with the S-ICD, the first shock success rate is approximately 90% (and much higher with multiple shocks), but implantation predictors of defibrillation efficacy with this device have not been well defined. Our computer modeling study has identified posterior generator position, and lack of intervening fat between the coil and/or the generator and the underlying fascia, as predictors of successful defibrillation. Shock impedance is elevated by fat between the coil and/or generator and the fascia, but not by anterior generator position.
TRANSLATIONAL OUTLOOK: Results from this computer modeling study should be validated by clinical data with the S-ICD, ideally in prospective studies or possibly by retrospective analysis of clinical cases assessing defibrillation efficacy.
This study was funded by Boston Scientific Corp. Dr. Heist is a consultant for and has received research grants from Biotronik, Boston Scientific, and Medtronic and has received research grants from LivaNova and St. Jude Medical. Dr. Belalcazar is a consultant for Boston Scientific. Mr. Stahl is an employee of Boston Scientific. Dr. Knops is a consultant and speaker for Boston Scientific, Medtronic, and St. Jude Medical. Dr. Brouwer has reported that he has no relationships relevant to the contents of this paper to disclose.
Drs. Heist and Belalcazar contributed equally to this work.
- Abbreviations and Acronyms
- defibrillation threshold
- left ventricular
- magnetic resonance imaging
- subcutaneous implantable cardioverter-defibrillator
- transvenous implantable cardioverter-defibrillator
- ventricular fibrillation
- ventricular tachycardia
- Received July 21, 2016.
- Revision received October 11, 2016.
- Accepted October 21, 2016.
- The Authors
- Burke M.C.,
- Gold M.R.,
- Knight B.P.,
- et al.
- Lambiase P.D.,
- Barr C.,
- Theuns D.A.,
- et al.
- Weiss R.,
- Knight B.P.,
- Gold M.R.,
- et al.
- Noro M.,
- Zhu X.,
- Enomoto Y.,
- et al.
- de Jongh A.L.,
- Entcheva E.G.,
- Replogle J.A.,
- et al.
- Mocanu D.,
- Kettenbach J.,
- Sweeney M.O.,
- et al.
- ↵de Jongh AL, Entcheva E, Replogle J, Claydon FJ. Effects of cardiac anisotropy on modeling transvenous defibrillation in the human thorax. Engineering in Medicine and Biology Society, 1997. Proceedings of the 19th Annual International Conference of the IEEE, 1997;1.
- Zhou X.,
- Daubert J.P.,
- Wolf P.D.,
- Smith W.M.,
- Ideker R.E.
- Witkowski F.X.,
- Penkoske P.A.,
- Plonsey R.
- Geddes L.A.,
- Baker L.E.
- Gold M.R.,
- Olsovsky M.R.,
- Pelini M.A.,
- Peters R.W.,
- Shorofsky S.R.
- Brouwer T.,
- Driessen A.,
- Olde Nordkamp L.,
- et al.
- Frommeyer G.,
- Zumhagen S.,
- Dechering D.G.,
- et al.
- Rantner L.J.,
- Vadakkumpadan F.,
- Spevak P.J.,
- Crosson J.E.,
- Trayanova N.A.
- Trayanova N.,
- Rantner L.J.