Author + information
- Satoshi Aita, MDa,∗,
- Kuniomi Ogata, MEngc,∗,
- Kentaro Yoshida, MDa,b,∗ (, )
- Takeshi Inaba, RDCSa,
- Hisanori Kosuge, MDa,d,
- Takeshi Machino, MDa,
- Yasuaki Tsumagari, MDa,
- Ai Hattori, MDa,
- Yoko Ito, MDa,
- Yuki Komatsu, MDa,
- Kensuke Sekihara, PhDe,
- Hitoshi Horigome, MDf,
- Kazutaka Aonuma, MDa,
- Akihiko Nogami, MDa,
- Akihiko Kandori, PhDc,† and
- Masaki Ieda, MDa,†
- aDepartment of Cardiology, University of Tsukuba, Tsukuba, Japan
- bDepartment of Cardiology, Ibaraki Prefectural Central Hospital, Kasama, Japan
- cResearch and Development Group, Hitachi Ltd., Kokubunji, Japan
- dTsukuba Advanced Imaging Center, Tsukuba, Japan
- eSignal Analysis Inc., Hachioji, Japan
- fDepartment of Pediatrics, University of Tsukuba, Tsukuba, Japan
- ↵∗Address for correspondence:
Dr. Kentaro Yoshida, Department of Cardiology, Faculty of Medicine, University of Tsukuba, 1-1-1 Tennodai, Tsukuba 305-8575, Japan.
Objectives This study aimed to develop a novel premature ventricular contraction (PVC) mapping method to predict PVC origins in whole ventricles by merging a magnetocardiography (MCG) image with a cardiac computed tomography (CT) image.
Background MCG can noninvasively discriminate PVCs originating from the aortic sinus cusp from those originating from the right ventricular outflow tract.
Methods This study was composed of 22 candidates referred for catheter ablation of idiopathic PVCs. MCG and CT were performed the same day before ablation. Estimated origins by MCG-CT imaging using the recursive null steering spatial filter algorithm were compared with origins determined by electroanatomic mapping (CARTO, Biosense Webster, Inc., Diamond Bar, California) during the ablation procedure. Radiopaque acrylic markers for the CT scan and coil markers generating a weak magnetic field during MCG measurements were used as reference markers to merge the 2 images 3-dimensionally.
Results PVC origins were determined by endocardial and epicardial mapping and ablation results in 18 (86%) patients (right ventricular outflow tract in 10 patients, aortic sinus cusp in 2 patients, interventricular septum in 1 patient, near His bundle in 1 patient, right ventricular free wall in 1 patient, and left ventricular free wall in 3 patients). Estimated origins by MCG-CT imaging matched the origins determined during the procedure in 94% (17 of 18) of patients, whereas the electrocardiography algorithms were accurate in only 56% (10 of 18). Discrimination of an epicardium versus an endocardium or right- versus left-sided septum was successful in 3 of 4 patients (75%).
Conclusions The diagnostic accuracy of noninvasive MCG-CT mapping was high enough to allow clinical use to predict the site of PVC origins in the whole ventricles.
- catheter ablation
- computed tomography
- noninvasive mapping
- premature ventricular contraction
Radiofrequency catheter ablation (RFCA) for premature ventricular contractions (PVCs) is an established treatment option to eliminate patient symptoms and prevent PVC-induced cardiomyopathy and heart failure (1). Although technologies such as 3-dimensional (3-D) mapping, irrigated-tip ablation catheters, and contact force monitoring help operators to map and ablate PVCs, procedural complexity, efficacy, and safety still depend on the location of the arrhythmia origin (2). It is especially difficult to eliminate PVCs with an origin near the His-bundle region or in the left ventricular (LV) summit, the intramural layer, or the epicardial aspect. Electrocardiography (ECG) algorithms can identify the area of origin of ventricular arrhythmias, but their spatial resolution is fundamentally low and their diagnostic accuracy can be impaired by the patient’s physical characteristics (e.g., obesity and cardiac rotation and offsets of ECG lead placement) (3). Although the utility of magnetocardiography (MCG) to discriminate PVCs originating from the right ventricular outflow tract (RVOT) from those originating from the aortic sinus cusps was previously reported from our center (4,5), this system did not allow mapping of the whole right and left ventricles and identification by the absolute location (coordinates) of the origins. To overcome these limitations and increase the clinical utility of the MCG system for arrhythmia treatments, we developed a custom-made technology to merge MCG images with 3-D computed tomography (CT) images. The aim of this prospective study was to validate this method by comparing the estimated PVC origin on the MCG-CT merged images with that on the map depicted on a 3-D mapping system during catheter ablation.
Twenty-two consecutive patients with PVCs refractory to antiarrhythmic drugs referred for catheter ablation in our hospital between March 2017 and February 2018 were prospectively included in this study. All except 2 patients with coronary artery disease were free of any structural heart disease. The study protocol was approved by the local Institutional Review Board, and all patients provided their informed written consent.
MCG and CT
MCG can noninvasively measure the weak magnetic field generated by the human heart. The electric current distributions in the heart are estimated from the magnetic field. A superconducting quantum interference device magnetometer is used as the magnetic sensor in MCG. Because the permeability of the human body is constant, detection of the heart’s magnetic field is barely affected by other organs around the heart such as fat, bone, and lungs (6–8).
The entire study protocol is outlined in Figure 1 and the Central Illustration. CT examinations were performed 1 week before catheter ablation with a 320-row CT scanner (Aquilion ONE GENESIS Edition, Canon Medical Systems, Otawara, Japan). Just before the CT scan, 5 custom-made radiopaque spherical markers fabricated with acryl and fixed on an L-shaped board were placed on the patients’ chests to determine their 3-D location relative to the heart (Figure 1A). The parameters for CT imaging were prospective ECG-gated acquisitions, tube potential of 120 kVp, and detector configuration of 320 × 0.5 mm. Scanning was initiated after a threshold of 150 Hounsfield units was reached in the ascending aorta. The CT images were reconstructed at a 0.5-mm slice thickness in the diastolic phase (basically 75% to 77%) of the R-R interval. The CT Digital Imaging and Communications in Medicine images were viewed and edited with Expert INTAGE (CYBERNET SYSTEMS CO., Tokyo, Japan).
Just after the CT scan, the patient was moved to the magnetically shielded room to undergo MCG recording. MCG methodology has been described in detail previously (4,5,9,10). An MCG system (MC-6400, Hitachi High-Technologies Corporation, Tokyo, Japan) with 64 magnetic sensors was used to measure the MCG signal and the magnetic field signal generated from the marker coils. The magnetic sensors were in an 8 × 8 matrix with a pitch of 25 mm and a measurement area of 175 × 175 mm. The MCG signals from each patient were recorded in the resting state in the anteroposterior direction for 2 min at a sampling rate of 1 kHz. The MCG signals were passed through a 0.1- to 50-Hz band-pass filter and a 50-Hz power line noise filter. As during CT scanning, the patients’ arms were toward their head during MCG recording. Before MCG recording for PVCs, triaxial (x-y-z axis) circular marker coils, which were fixed on an L-shaped board as the acrylic markers for the CT scan, were placed at the same position as that during the CT scan (Figure 1B). A coil driver applied alternating electric current (40 Hz) to each marker coil. The positions of the marker coils on the patient were estimated from the measured magnetic field signals by using an optimization method (Online Appendix). Then, the coordinates of the coil markers for MCG were merged with those of the acrylic markers for CT (Figure 1C). Finally, the MCG signals from the PVCs were recorded for 2 min, averaged, and then used to reconstruct the 3-D current distribution by applying the recursive null steering (RENS) spatial filter algorithm (Figure 1D) (Online Appendix) (11). We developed analysis software in MATLAB (MathWorks, Inc., Natick, Massachusetts) for estimating the position of the marker coils and 3-D current distribution from the MCG measurement.
Electrophysiological study and catheter ablation
After informed consent was obtained, electrophysiological study and RFCA were performed. Antiarrhythmic drugs were discontinued for at least 5 half-lives before the procedure. Electrode catheters were placed near the His bundle, coronary sinus, and right ventricular apex. When a target PVC was suspected to originate from the left outflow tract, a 2-F mapping catheter (Japan Lifeline Co., Ltd., Tokyo, Japan) was advanced into the great cardiac vein through the inner lumen of the catheter placed at the coronary sinus (4-mm interelectrode spacing; Inquiry Luma-Cath Fixed Diagnostic Catheter, Abbott, Saint Paul, Minnesota) inserted from the subclavian vein (2). A 7-F quadripolar 3.5-mm-tip open-irrigated ablation catheter (Navistar ThermoCool STSF, Biosense Webster, Inc.) was used for mapping and ablation. A 3-D electroanatomic mapping system (CARTO System Version 4.3.5, Biosense Webster, Inc.) was used to identify the earliest activation site of the target PVC. The radiofrequency energy was delivered at a maximum power of 50 W and a maximum temperature of 42°C. Pace mapping was also performed to search for optimal RFCA sites. Successful RFCA was defined as complete elimination of the clinical PVC and noninducibility by programmed ventricular stimulation or by a drug provocation test (isoproterenol up to 0.06 μg/kg/min, 20 mg adenosine triphosphate, and 10 mg edrophonium). We defined the site of successful RFCA as the site of PVC origin.
Comparison between MCG-CT map and electroanatomic map
The right and left ventricles were divided into 9 and 11 segments, respectively, that were derived from schemas depicted in a previous study (Figure 2) (12). The segment in which a procedure-derived PVC origin was identified was determined based on the electroanatomic map merged with CT imaging by the CARTO system (red closed dots in Figure 2). The operator judged the segment of the procedure-derived origin, and then 2 electrophysiology fellows blinded to the RFCA results independently judged the segment of the MCG-CT–derived origin. These fellows also judged whether the MCG-CT–derived origin was considered endocardial versus epicardial or a left- versus right-sided septum. Discrepancies were resolved by consensus. Although we prospectively included patients in this study, an estimated PVC origin could not be determined before catheter ablation because data analysis took 3 to 4 weeks.
ECG monitoring was continued in all patients from the end of RFCA and throughout hospital stay. After discharge, the patients were followed up at the outpatient clinic, with visits at 1, 3, and 6 months after RFCA, and elimination of the PVCs was confirmed by 12-lead and Holter ECGs recorded at each follow-up visit.
Eighteen patients were included for further analysis after 4 patients with failed ablation were excluded from the study. Patient details and PVC characteristics are presented in Table 1. The mean patient age was 54 ± 17 years, and 12 (67%) women were included. The mean LV ejection fraction was 65 ± 6%. The QRS morphology of the PVC was a right bundle branch block pattern in 3 (17%) patients and a left bundle branch block pattern in 15 (83%) patients.
PVC origins and MCG-CT mapping accuracy
The distribution of procedure-derived PVC origins in all 18 patients is shown in Table 1 and Figure 2. In 15 cases (68%), spontaneous PVC constantly occurred during the procedure, and we could map the PVC origin by activation mapping. Pace mapping was mainly used in the remaining 7 patients. A correct diagnosis by MCG-CT mapping was obtained in 17 of the 18 patients (94%). Patients with an epicardial origin in the LV anterior wall (Patient #18) (Figure 3), an endocardial origin in the LV lateral wall (Patient #7), and right-sided ventricular septum origin (Patient #3) were correctly diagnosed by MCG-CT mapping. In the 1 patient (Patient #16) with failed diagnosis, the correct origin was the epicardial aspect of the LV posterior wall, whereas the MCG diagnosis was the endocardial aspect of the lateral wall (Online Figure 1). Overall, diagnosis of an epicardium versus an endocardium and a right-sided versus left-sided septum was correctly discriminated by MCG-CT mapping in 3 of 4 patients (75%).
Accuracy of ECG algorithms
This 57-year-old man had frequent PVCs originating from the LV outflow tract. A 12-lead ECG of the PVCs is shown in Figure 3A. Figure 3B shows the MCG current maps merged with CT imaging in the horizontal, coronal, and sagittal planes where the maximum current during the PVC is shown in red. Figure 3C indicates the coordinates of PVC origin by the crossed lines, and Figure 3D indicates them 3-dimensionally on the CT map. Of note, the left coronary artery on the epicardial aspect (red dashed arrows) was adjacent to the PVC origin (red closed dot). The results from CARTO and fluoroscopic imaging during the ablation procedure are shown in Figure 3E. Based on the recording of the local electrograms (Figure 3F), the great cardiac vein was activated earlier than the LV endocardium and aortic sinus cusps. First, we applied RFCA to the left coronary cusp followed by the LV endocardium (left and bottom panels of fluoroscopic images in Figure 3E). Transient suppression of the PVCs during the applications was observed, but the PVCs immediately recurred. Next, we advanced the ablation catheter into the great cardiac vein (right panel of fluoroscopic imaging in Figure 3E). Its local activation time preceded QRS onset by 30 ms, and ablation in the vein eliminated the PVCs completely. It was notable that the successful ablation site matched the estimated origin by MCG-CT mapping.
The estimated origin by MCG-CT mapping (red closed dot) was the endocardial lateral wall in the left ventricle (Figures 4A to 4C, Table 1, Online Figure 2). Ablation at the same site on the CARTO map (Figure 4D) successfully eliminated the PVCs. Notably, MCG-CT mapping predicted an endocardial origin but not an epicardial origin in this case.
Patient #2, 10, and 11
Three patients with PVCs originating from the RVOT are shown in Figures 5A to 5C (Table 1, Online Figure 3). The red closed dots in Figures 5A to 5C indicate the estimated origins by MCG-CT mapping, and the ablation sites are shown by red tags on the respective CARTO maps. The posterior site of the RVOT (Figure 5A), anterior site of the RVOT (Figure 5B), and free wall site of the RVOT (Figure 5C) were correctly diagnosed by MCG-CT mapping. The PVCs in these patients were successfully eliminated by ablation.
The mismatch of results is described in detail in Online Figure 1.
The details of failed ablation are described in Table 1. In Patient #19, MCG-CT mapping estimated the origin in the epicardial aspect of the anterior wall of the left ventricle. Although we tried to ablate it from the great cardiac vein, high impedance did not allow delivery of radiofrequency applications. In Patient #20, MCG-CT mapping estimated the origin in the septal aspect of the RVOT, but the aortic sinus cusp was targeted for ablation. Because the PVC was infrequently observed during the procedure, it was difficult to identify the correct origin. In Patient #21, MCG-CT mapping estimated the origin in the LV septum, but only the right ventricular septum was actually targeted for ablation and the LV septum was not mapped. In Patient #22, MCG-CT mapping estimated the origin in the right ventricular septum. Ablation was applied to the lateral wall of the RV, but it did not eliminate the PVCs.
The key observations of the present study are as follows: 1) a novel method to merge the MCG map with the CT image using custom-made markers worked successfully and identified the sites of origin in patients with PVCs originating from both ventricles; 2) accuracy was 94% in patients who had PVCs successfully eliminated by catheter ablation, whereas that of the ECG algorithm was lower as previously reported (16); and 3) from the theoretical and mathematical viewpoints, this technology may have the potential to identify transmural distribution of sites of origin (epicardial vs. endocardial origin and right-sided vs. left-sided septal origin). Although this study provides only preliminary results, the findings from a few cases shown in the figures have encouraged us to undertake additional study of more subjects and update the merge function.
Fundamentally, MCG has higher spatial resolution than ECG because the magnetic field is not distorted by flow through tissues such as lungs, bones, and muscles. The RENS spatial filter algorithm can reconstruct a source distribution from bioelectromagnetic data with a spatial resolution considerably higher than that of conventional methods. The robustness of this method to the source correlation was previously validated in our computer simulation and our experiments using auditory-evoked magnetoencephalographic data (11). The present method to 3-dimensionally merge MCG and CT images further enhanced these advantages and the clinical utility of the MCG system in arrhythmia treatments.
Merging the MCG map with the CT image
Although our mathematical evaluation of the MCG-CT merged method calculated its error to be within 3 mm, its accuracy may be impaired by several factors in humans such as effects caused by the respiratory phase or cardiac cycle (systole or diastole), volume status, PVC coupling intervals, number of PVCs analyzed and averaged, and the distance between the MCG sensors and PVC origin. However, most of these negative factors can be overcome. The respiratory effect may be minimized by performing CT scanning and CARTO mapping during the expiratory phase. The end-diastolic phase was selected throughout the CT scan, MCG recording, and CARTO mapping. The volume status was consistent throughout the measurements because CT and MCG mapping were consecutively performed within a few hours of each other on the same day, and all measurements and treatment were performed in the fasting state. Although the PVC coupling intervals varied (Table 1), accuracy did not appear to be affected in this patient series. One possible negative contributor is the small number of PVCs analyzed or averaged. In the single patient (Patient #16) with failed localization by MCG-CT mapping and with PVCs originating from the LV posterior wall, only 1 PVC was recorded during the MCG measurement. This factor may be associated with a decreased signal-to-noise ratio. Another factor is the longer distance between the MCG sensors and the origin of the PVCs. Importantly, the LV posterior wall is further from the sensors than the RVOT area. This could be 1 reason for failed localization because magnetic signals attenuate in proportion to the distance.
Taken together, overall accuracy was ethically acceptable to conduct a prospective validation cohort in a manner that informs the operator of the estimated PVC origin before catheter ablation. Also, the overall accuracy of the ECG algorithms was lower than that of the MCG method, and, particularly, their accuracy in the subdivision of the PVC region was quite low, supporting the clinical value of MCG over ECG (16).
Preventing coronary artery injury
It is well-known that coronary artery injury can occur as a direct thermal effect of RFCA. Although the right coronary artery and left circumflex artery are likely to suffer from such a complication in the ablation of Wolff-Parkinson-White syndrome and mitral isthmus ablation, the left anterior descending artery on the epicardial aspect is also at risk for thermal injury (17). Although PVCs can potentially be treated by ablation from the coronary sinus and its branches (2), the risk of coronary artery injury should be assessed, ideally by a noninvasive method, before the ablation procedure is performed. In this study, we noticed that the locations of the coronary arteries were helpful as landmarks to recognize the detailed and exact location of sites of origin. Because the coronary arteries were precisely depicted in our CT images without processing such as smoothing, attenuation, and volume reconstruction, they provided important safety information to physicians before the procedure and may be 1 of the advantages of our mapping technology.
Noninvasive body surface mapping systems for the detection of PVCs, so-called electrocardiographic imaging (ECGi) (CardioInsight Technologies, Inc., Cleveland, Ohio) (16,18) and a noninvasive epicardial and endocardial electrophysiology system (EP Solutions SA, Yverdon-les-Bains, Switzerland) (19–22), were previously introduced. Jamil-Copley et al. (16) investigated 24 patients with outflow tract PVCs using ECGi (RVOT in 18 patients and LVOT in 6 patients). The accuracy for the discrimination between RVOT versus LVOT was 96%. Wissner et al. (21) reported that in 18 (86%) of 21 patients with PVC/VT, the correct ventricular segment was diagnosed as a result of nonquantitative assessment. Of note, these systems depend not on the original but the mathematically reconstructed unipolar electrograms and apply a method to project the location data on the simplified heart torso created from CT images. This may result in decreased accuracy or an error in discriminating endocardial, epicardial, and intramural origins. Our novel method can potentially overcome these limitations by not projecting but simply merging the coordinates of origins determined by MCG mapping on an original 3-D CT image. In fact, our method could potentially allow the accurate diagnosis of epicardial origins and the side (right vs. left) of septal origin. Of note, a very recent study reported negative results of ventricular mapping by the ECGi system, which questions many of the conclusions previously drawn from ECGi mapping. This study directly compared the ECGi map and invasive epicardial contact map, and primary epicardial breakthroughs imaged using both methods resulted in very large mean absolute errors of over 70 mm (23).
First, the study volume is small. Only 2 epicardial origins and 2 aortic sinus cusp origins were included. However, the wide distribution of origins other than those in the common outflow tracts was helpful in evaluating the accuracy of this system in the whole ventricles.
Although we prospectively included patients in this study, the estimated PVC origin could not be determined before catheter ablation because data analysis took 3 to 4 weeks. Currently, we have established a method to analyze data and reconstruct and merge the 3-D map within a few days and conducted a validation study in which the estimated result is informed to the operators before the ablation procedure.
Although we attempted to minimize negative factors impairing the accuracy of this mapping system, respiratory and cardiac motion should particularly be taken into consideration to maximize the performance of this mapping technology in clinical practice.
This study was not a direct comparison of MCG and technologies such as ECGi; it only investigated the system’s ability to detect arrhythmia origin (electrical onset). The ability of this technology in other types of arrhythmias remains to be evaluated in future studies.
A quantitative analysis that can provide the spatial resolution of this mapping system could not be conducted because MCG recording requires a magnetically shielded room, making it difficult to perform during pace mapping. Because our technology was completely custom-made, we could not load MCG-CT data into the commercially developed CARTO mapping system. Errors in merging the electroanatomic map depicted by the ablation catheter and CT imaging (within the CARTOMERGE system, Biosense Webster, Inc.) cannot be ignored in the assessment of this study. We are now conducting a new validation study to allow more accurate merging of the CT, MCG, and CARTO maps. This upgrade will allow complete quantitative analysis of the accuracy of this MCG-CT mapping method in the near future.
In this preliminary study, MCG-CT mapping could identify sites of PVC origins in both ventricles noninvasively and semiquantitatively. A validation cohort including more patients and further updates to the system are needed to justify the use of contrast medium and radiation during CT scanning and to clarify whether this technology improves procedural outcomes, procedural efficiency, and the safety of catheter ablation for ventricular arrhythmias.
COMPETENCY IN MEDICAL KNOWLEDGER: adiofrequency catheter ablation for PVCs has become an established treatment option. However, procedural complexity, efficacy, and safety still depend on the location of the PVC origin. Although electrocardiographic algorithms can identify the area of origin of ventricular arrhythmias, their accuracy is fundamentally limited due to the low spatial resolution. A custom-made technology to merge magnetocardiography images with 3-D computed tomography images could accurately identify sites of PVC origin distributed in the whole ventricles (17 of 18, 94%).
TRANSLATIONAL OUTLOOK: This technology may have the potential to improve procedural outcomes, procedural efficiency, and the safety of catheter ablation for ventricular arrhythmias.
↵∗ Drs. Aita and Ogata contributed equally to this work and are joint first authors.
↵† Drs. Kandori and Ieda contributed equally to this work and are joint senior authors.
This work was supported by JPS KAKENHI grant number JP17K09484. Dr. Aonuma has received a research grant from Hitachi, Ltd. Dr. Nogami has received honoraria from Abbott and an endowment from Medtronic and Johnson and Johnson. Mr. Ogata holds the patent JP6393173B, position estimation method, and position estimation system. Dr. Kandori holds the patent JP6393173B, positional estimation method and positional estimation system. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- computed tomography
- electrocardiographic imaging
- left ventricular
- premature ventricular contraction
- recursive null steering
- radiofrequency catheter ablation
- right ventricular outflow tract
- Received January 31, 2019.
- Revision received May 14, 2019.
- Accepted June 19, 2019.
- 2019 The Authors
- Komatsu Y.,
- Nogami A.,
- Shinoda Y.,
- et al.
- Inaba T.,
- Nakazawa Y.,
- Yoshida K.,
- et al.
- Ito Y.,
- Shiga K.,
- Yoshida K.,
- et al.
- Yamada S.,
- Kandori A.
- Yoshida K.,
- Ogata K.,
- Inaba T.,
- et al.
- Yoshida K.,
- Liu T.Y.,
- Scott C.,
- et al.
- Yoshida N.,
- Yamada T.,
- McElderry H.T.,
- et al.
- Kimata A.,
- Igarashi M.,
- Yoshida K.,
- Takeyasu N.,
- Nogami A.,
- Aonuma K.
- Cakulev I.,
- Sahadevan J.,
- Arruda M.,
- et al.
- Wissner E.,
- Saguner A.M.,
- Metzner A.,
- et al.
- Wissner E.,
- Revishvili A.,
- Metzner A.,
- et al.
- Wang Y.,
- Cuculich P.S.,
- Zhang J.,
- et al.
- Duchateau J.,
- Sacher F.,
- Pambrun T.,
- et al.