Author + information
- Received February 6, 2018
- Revision received June 11, 2018
- Accepted June 19, 2018
- Published online October 15, 2018.
- Charlotte Brouwer, MD,
- Gijsbert F.L. Kapel, MD,
- Monique R.M. Jongbloed, MD, PhD,
- Martin J. Schalij, MD, PhD,
- Marta de Riva Silva, MD and
- Katja Zeppenfeld, MD, PhD∗ ()
- ↵∗Address for correspondence:
Dr. Katja Zeppenfeld, Leiden University Medical Centre, Department of Cardiology, C5-P, P.O. Box 9600, 2300 RC Leiden, the Netherlands.
Objectives This study sought to evaluate the relation between 12-lead ventricular tachycardia (VT) electrocardiography (ECG) and VT-related anatomical isthmuses (AIs) in repaired tetralogy of Fallot (rTOF).
Background Slow-conducting AIs are the dominant VT substrate in rTOF. Whether an AI is considered critical relies on pace mapping (PM) guided by the VT ECG.
Methods VT ECGs, electroanatomical mapping data and PM results were analyzed in 25 rTOF patients (group 1) (age 57 ± 13 years). Selection of PM and ablation sites was guided by VT ECG. In 7 patients (group 2) (age 33 ± 14 years), PM was systematically performed within all AIs, irrespective of the VT ECG.
Results In group 1, all 35 induced VTs (median VT cycle length 270 [interquartile range: 240 to 310] ms) were AI related. All 11 right bundle branch block (RBBB) VTs were related to AI3 (right ventricular septum if positive concordant [7 of 7]), coronary cusp if V2 transition break [3 of 4]). Left bundle branch block (LBBB) VTs with transition <V5 were mapped to AI3 (8 of 10) or AI2 (2 of 10) and LBBB VTs with transition ≥V5 to AI1 (8 of 14), AI3 (5 of 14), and AI4 (1 of 14). In group 2, all 8 induced VTs (median VT cycle length 240 [interquartile range: 230 to 268] ms) were AI related. All RBBB VTs were related to AI3 (right ventricular septum). For LBBB VTs, paced matches were obtained in AI3 and AI1. Activation mapping and/or ablation success confirmed AI3 to be critical for all 8 VTs.
Conclusions In rTOF with only AI1 and AI3, RBBB VTs are due to clockwise and LBBB VTs to counterclockwise activation of AI3. Involvement of both AIs in the VT circuit limits the role of the 12-lead VT ECG and PM. AI3 can always be targeted irrespective of the 12-lead VT ECG.
Although survival into adulthood has improved, patients with repaired tetralogy of Fallot (rTOF) remain at risk for late ventricular arrhythmias (1–3). The majority of ventricular arrhythmias are monomorphic re-entrant ventricular tachycardia (MVT) (4). The dominant substrate for MVT are slow-conducting anatomical isthmuses (AIs) located in the right ventricular outflow tract (RVOT) (Figure 1) (3,5,6). The most important pathological feature of TOF is the anterior deviation of the infundibular septum (i.e., outlet septum), located between the pulmonary outflow tract and the aortic valve. This deviation results in malalignment of the outlet septum and consequently a subaortic ventricular septal defect (VSD), subpulmonary stenosis, and dextroposition of the aorta overriding the ventricular septum (7). In the present surgical era of transatrial-transpulmonary repair, the majority of rTOF patients (94%) have only AI1 (located between a transannular [TA] patch or RVOT incision and the tricuspid annulus) and AI3 (infundibular septum, bordered inferiorly by the VSD patch and superiorly by the pulmonary and aortic valves) (6,8). Isthmus 2 (between an RV incision and the pulmonary valve [PV]) is only present in case of a transventricular repair and AI4 only in patients with a subaortic VSD with muscular posteroinferior rim, which occurs in the minority of TOF patients (5,9).
Transection of a VT-related AI by ablation is highly effective to prevent VT (3,5,10–12). Whether an AI is related to clinical VTs and selected for ablation is determined mainly by pace mapping because of the poorly tolerated VTs in rTOF (13). Preselection of pace-mapping sites is guided by the 12-lead VT electrocardiography (ECG). The VT QRS complex begins when the excitation wavefront activates viable myocardium (14,15). In rTOF, location and extent of surgical scars may significantly affect the RV activation sequence and thereby the resultant VT ECG. As a consequence, the VT ECG may misdirect pace mapping to an AI that might be less approachable by ablation, disregarding that other AIs may be involved, but also may misclassify a VT as not being AI related. The objective of this study was to systematically evaluate: 1) the relation between the 12-lead VT ECG and VT-related AI in rTOF; and 2) whether pace mapping facilitates selection of the AI most suitable for ablation.
Consecutive rTOF patients who underwent detailed electroanatomical mapping (EAM) and VT ablation between 2007 and 2016 were included. All patients were treated according to our standard clinical protocol, which has been adapted over time (6). Accordingly, patients were divided into group 1 (2006 to 2013): selection of the AI for ablation was based on limited pace mapping guided by the 12-lead VT ECG; and group 2 (2014 to 2017): systematic pace mapping within all AIs by pacing along and at both sides of the shortest straight line between 2 unexcitable boundaries was performed irrespective of the 12-lead VT ECG, followed by identification of the AI related to VT. If more than 1 AI was considered VT related, the best approachable AI was selected for ablation.
Medical records were reviewed for details on prior surgeries and documented VTs. Biventricular function was obtained from transthoracic echocardiograms and cardiac magnetic resonance imaging studies.
EAM: AI characteristics and geometry
Programmed electrical stimulation was performed from the RV apex and RVOT (≥3 extrastimuli at ≥3 cycle lengths, including administration of isoproterenol). Twelve-lead ECGs of all induced VTs were stored. A 3-dimensional reconstruction of all AIs was obtained as previously described (6). The geometry and electrophysiological characteristics of AIs were assessed by defining isthmus length, isthmus width, and conduction velocity during sinus rhythm, as described previously (6).
Briefly, AIs were considered abnormal in the absence of continuous normal bipolar electrograms (>1.5 mV) throughout the isthmus. If conduction velocity across the AI (AI length divided by difference between the local activation time at isthmus entrance and exit) was <0.5 m/s, the AI was considered slow conducting (6).
The length of the RV incision or TA patch was measured as distance between PV and the lower edge site at the RVOT or RV free wall that did not capture during high output pacing. The PV level was determined by the first ventricular electrogram recorded when withdrawing the catheter from the pulmonary artery at any site of its circumference (Figure 2).
Selection of ablation target sites
In group 1, the 12-lead ECG of each induced VT was analyzed and pace mapping was initiated within the AI likely related to VT, based on the experience of the operator. An AI was considered VT related if the paced QRS ECG matched the VT ECG (pace match ≥11 or 12 leads) regardless of the stimulus-to-QRS interval. Once a VT-related AI was identified, the adjoining anatomical boundaries were connected by a linear radiofrequency lesion. If VT was hemodynamically tolerated, participation of a given AI in the VT circuit was confirmed by recording of diastolic activity, entrainment mapping, or VT termination during ablation (3,5). Ablation was considered successful if conduction through the corresponding AI was blocked and VT was no longer inducible. In all studies, AI block was assumed if no capture along the ablation line could be obtained with high-output pacing (10 mA, 2 ms) in combination with the presence of double potentials or a change in activation sequence during sinus rhythm or RV pacing (3). Differential pacing to confirm bidirectional conduction block was performed in all procedures since 2008. All paced 12-lead ECG and corresponding pacing sites on 3-dimensional maps were stored.
In group 2, 12-lead ECG of induced VTs were also analyzed but a systematic pacing protocol was applied within all AIs irrespective of the 12-lead VT ECG. If a good pace-match was obtained within more than 1 AI, the narrowest and easiest approachable AI was selected for ablation.
Analysis of 12-lead ECG morphologies and their relation to AIs
Twelve-lead ECGs of spontaneous, induced VTs and paced morphologies were analyzed for: 1) frontal plane QRS axis; 2) bundle branch block–like morphology (right bundle branch block [RBBB] morphology defined as dominant R-wave or qR complex in V1, left bundle branch block [LBBB] morphology defined as dominant S-wave in V1); and 3) precordial QRS transition (first lead with a predominant R-wave or S-wave for LBBB and RBBB, respectively). QRS transition ≥V5 was defined as late precordial transition. A transition break in V2 was defined as a RBBB morphology with dominant S in V2 and dominant R in V3 to V6. Based on the location on 3-dimensional maps, pacing sites were assigned to an AI if a ≥11 of 12 pace match was obtained within the AI. The stimulus-to-QRS interval at each pacing site was measured.
Continuous data are presented as mean ± SD or median (interquartile range [IQR]) according to distribution. Categorical data are reported as percentage and frequency. Continuous variables were compared using the Student's t test or the Mann-Whitney U test where appropriate. Categorical variables were compared using the chi-square test. Multiple morphologically distinct VTs in 1 patient were considered as independent for analysis. All analyses were performed with SPSS 22.0 (IBM Corporation, Armonk, New York). A p value <0.05 was considered significant.
Group 1 consisted of 25 patients (age 57 ± 13 years, 64% men), median age at repair 12 (IQR: 5 to 19) years. Sixteen (64%) patients had a RV incision extending from the PV into the RV, with annulus augmentation by a TA patch in 11 (44%) patients. In 9 (36%) patients, the RV incision spared the PV. Twelve (48%) patients had spontaneous MVT, recorded on 12-lead ECG (6 of 12) or implantable cardioverter-defibrillator recordings (6 of 12).
Group 2 consisted of 7 patients (33 ± 14 years, all men), median age at repair 2 (IQR: 1 to 5) years. Four patients (57%) had a transatrial-transpulmonary total repair approach with an incision extending into the RVOT in 2 (29%) and annulus augmentation by a TA patch in 2. One patient was repaired with an RV incision sparing the PV. Three (43%) patients had documented spontaneous MVT, all recorded on 12-lead ECG.
Table 1 summarizes patient characteristics according to groups.
EAM and geometry of AIs
In group 1, 24 (96%) patients had both AIs 1 and 3. The mean RV incision length was 41 ± 13 mm. Isthmus 1 was wider compared with AI3 (36 ± 14 mm vs. 20 ± 6 mm; p < 0.0001). Isthmuses 1 and 3 were electroanatomically abnormal in 9 (36%) and 21 (84%) patients, respectively. All abnormal AIs 1 and 3 were also slow conducting. Isthmus 2 was observed in 9 (36%) patients, were electroanatomically abnormal in 5 (20%) patients, and were slow conducting in 3 (12%) patients. Three patients had AI4, which was electroanatomically abnormal and slow conducting in 1 patient.
In group 2, AIs 1 and 3 were observed in all patients. The mean RV incision length was 36 ± 10 mm. AI width was 47 ± 11 mm for AI1 and 28 ± 9 mm for AI3 (p = 0.053).
Electroanatomically abnormal AI1 and AI3 were observed in 1 (14%) and 5 (71%) patients, respectively. All electroanatomically abnormal AIs 1 and 3 were also slow conducting. One patient had an electroanatomically normal AI2.
In group 2, AI 1 and AI 3 were on average 11 mm and 8 mm wider compared with group 1 (p = 0.076 and p = 0.016, respectively). The length of the RV incision was comparable between groups. Table 2 summarizes EAM data for both groups.
Characteristics of induced VTs
In group 1, all patients, including all 12 with documented VT, were inducible for a total of 35 VTs (median VT cycle length 270 [IQR: 240 to 310] ms). Twenty-six (74%) VTs were hemodynamically not tolerated. Eleven (31%) VTs had a RBBB morphology, all with an inferior axis. The QRS complex was positive concordant in 7 and a V2 transition break was observed in 4. Twenty-four (69%) VTs had an LBBB morphology, with an inferior axis in 17 and a superior axis in 7. An early QRS transition was observed in 10 and a late QRS transition was observed in 14.
Thirty-one of the 35 (89%) VTs were mapped to a slow-conducting AI (electroanatomically abnormal in 30), which was determined by pace mapping in all and confirmed by activation mapping and/or VT termination during ablation in 8 of 33 (23%). Four VTs were mapped to an AI with normal conduction (Table 3).
All 11 VTs with RBBB morphology were related to slow-conducting and electroanatomically abnormal AI3. Seven RBBB VTs with positive concordant morphology were mapped to the RV septal site of AI3. The 4 remaining RBBB VTs with V2 transition break were mapped to the left ventricular side of AI3, either to the opposite coronary cusp (3 of 4) or the subaortic left ventricular outflow tract (1 of 3).
Of 24 VTs with LBBB morphology, 20 (83%) were mapped to a slow-conducting AI (electroanatomically abnormal in 19). Thirteen VTs were mapped to AI3, 8 to AI1, 2 to AI2, and 1 to AI4.
In all 9 patients with AI2, VTs with early QRS transition (all inferior axis) were mapped to AI3 (presumed exit free wall) (6 of 8) or AI2 (2 of 8). All 3 VTs with late QRS transition (1 superior axis) were mapped to AI1.
In patients without AI2, both VTs with early QRS transition (both inferior axis) were mapped to AI3 (exit free wall). VTs with late QRS transition were mapped to AI3 (exit free wall) (5 of 11; 3 superior axis) and AI1 (5 of 11; 2 superior axis), and 1 VT was mapped to AI4. Figure 3 demonstrates the location of the presumed critical isthmus site for each induced VT in group 1 according to VT ECG.
In group 2, all patients, including those 3 with documented VT, were inducible for 8 VTs (median VT cycle length 240 [IQR: 230 to 268] ms). Six VTs (75%) were hemodynamically not tolerated. Three (38%) VTs had RBBB morphology (all inferior axis, positive concordant). Five (63%) VTs had LBBB morphology (2 superior axis, all QRS transition ≥V5).
Seven (88%) VTs were related to a slow-conducting and electroanatomically abnormal AI, determined by pace mapping in all.
All RBBB VTs were related to AI3 (exit RV septum) (3 of 3). Remarkably, for all LBBB VTs, a good paced match was obtained within AI3 but also within AI1 (Figure 4). Because AI3 was narrower and easier approachable by catheter ablation, AI3 was targeted for ablation in all.
A total of 39 of 43 induced VTs (91%) were mapped to a slow-conducting AI (electroanatomically abnormal in 38). However, a prolonged S-QRS interval (>40 ms) at AI sites with a good pace match was only observed for 15 (39%) VTs: 14 VTs in group 1 (10 in AI3, 2 in AI1, and 1 in AI2) and 1 VT in group 2 (AI3). A progressive shortening of the S-QRS interval between the VT entrance to VT exit site was only observed in 3 VTs.
Geometry of AIs in relation to LBBB VT morphology
In total, 18 patients had 19 LBBB VTs with late QRS-transition. Seven patients had 10 LBBB VTs with early QRS transition.
In patients with LBBB VTs and late QRS transition, the RV incision (as 1 unexcitable boundary of AI1) was significantly longer compared with patients with LBBB VTs and early QRS transition (37.1 ± 8.5 mm vs. 26.3 ± 7.3 mm; p = 0.005) (see Figure 5).
VT-related AIs and ablation results
In group 1, a total of 31 VTs were targeted. Of 22 VTs initially mapped to AI3, 15 were successfully transected by ablation (68%). Seven VTs mapped to AI3 could not be abolished after targeting AI3; in 5 VTs, the AI was protected by a pulmonary homograft. For 2 VTs, an ablation line from the coronary cusp extending toward the aortic root would have been required to transect AI3, which was discarded considering the risk of coronary artery and aortic valve damage. As this procedure was performed in the early years, targeting another AI was not considered.
Of 7 VTs mapped to AI1, only 1 (14%) could be successfully abolished by transecting AI1. For the remaining 6 VTs in 5 patients, a second AI was targeted. Four VTs were abolished by transecting AI3 in 3 patients and AI2 in 1 patient. In 1 patient, both AI1 and AI3 were protected by a pulmonary homograft and could not successfully be transected.
Two VTs mapped to AI2 were successfully abolished by transecting AI2.
In group 2, a total of 8 VTs were targeted. AI3 was successfully transected by ablation abolishing all 8 VTs, including the 5 LBBB VTs with a good pace match in both AI1 and AI3. Figure 6 shows ablation results for the targeted VTs.
This is the first study to systematically evaluate the 12-lead VT ECG in rTOF and its relation to VT-related AIs. We observed a remarkable variation of VT ECG patterns, although the critical VT substrate was always confined to the RVOT (Figure 7).
RBBB VTs had a more uniform morphology (inferior axis with a positive concordant or V2 transition break precordial pattern) and AI3, activated during VT in a clockwise direction (from an anterior view), was always part of the re-entry circuit. The morphology of LBBB VTs was more variable and dependent on the presence of AI2 and the geometry of AI1.
In the absence of AI2, LBBB VTs propagated through AI3 in a counterclockwise direction. Variations in the VT morphology were due to the geometry of AI1; a long RV incision or TA patch resulted in LBBB VTs with late transition. In these cases, because both AI1 and AI3 are involved in the re-entry circuit, AI3 (which is usually easier approachable by ablation) can be targeted without the need of pace mapping.
In the presence of AI2, variations in the precordial transition of LBBB VTs can be influenced by the activation of AI2. LBBB VTs with an early transition can be due to clockwise activation of AI2 or can exit from the free wall of AI3 if the RV incision is short. LBBB VTs with a late transition can result from counterclockwise activation of AI2, exiting at the free-wall site of AI1 if the RV incision is long. Of importance, AI2 can be a bystander, a critical part of the circuit, or part of a figure-of-8 mechanism involving AI1, AI2, and AI3 (Figure 8). In these cases, mapping is still required to identify the critical VT isthmus. Alternatively, ablation without mapping could target AI3 and AI2 if easily approachable.
Substrate for ventricular arrhythmias in rTOF
The majority of ventricular arrhythmias in rTOF are fast MVTs, arising from slow-conducting AIs bordered by unexcitable tissue. Slow-conducting AIs responsible for VT can be identified by EAM during sinus rhythm, depending mainly on pace mapping (91% of VTs in this cohort) (3,5). Pace mapping can provide a measure of slow conduction as indicated by a prolonged stimulus-to-QRS interval (16). Remarkably, in our cohort the stimulus-to-QRS interval during pace mapping was not useful to identify slow-conducting VT–related AIs, even when pacing was performed at rates close to VT cycle length. This can be explained by the short AI length, resulting in limited conduction delay over a short distance during pacing and during VT. The latter is in line with the typically short VT cycle length in rTOF.
Transection of the VT-related AI by ablation can be highly effective in preventing VT recurrence in rTOF. Demonstration of conduction block after isthmus transection is an accepted and clearly defined procedural endpoint (3,5). However, isthmus geometry may determine ablation success. Isthmuses 2 and 3 have been described to be narrow and thin walled, whereas AI1 was longer and thicker, and includes the region of the atrioventricular groove with the right coronary artery, which makes it less amenable for ablation (6,9). In line with these findings, in our cohort, AI1 was broader than AI3 and AI2, and attempts to transect AI1 by radiofrequency failed in 6 of 7 patients, followed by successful transection of AI3 or AI2.
Of note, the type of prior surgery affects isthmus geometry. The modern transatrial-transpulmonary approach avoids long RV incisions and large TA patches, preventing AI2 and resulting in an even broader AI1 (8,17,18).
From an ablation point of view, the narrowest and easiest accessible AI with the lowest risk for collateral damage should be targeted.
12-lead VT ECG in rTOF and its relation to AI
Preselection of pace-mapping sites is usually based on the 12-lead VT ECG. Limited information is available concerning the 12-lead VT ECG in rTOF and its relation to AIs. In a series of 11 rTOF patients with 15 AI-related VT, 2 RBBB, and 13 LBBB VTs with a RV critical isthmus site were induced. The QRS precordial transition tended to be at V3 or later in AI1-related VTs and earlier in AI3- or AI4-related VTs (5).
Notably, a RBBB VT morphology was observed in 25% of VTs in our series, which is usually considered to indicate a left ventricular site of origin. Despite the RBBB morphology, 71% of VTs were approachable from the RV. A positive concordant 12-lead VT ECG was typical for AI3 dependent VT with clockwise activation of AI3. Of interest, RBBB VTs with V2 transition break was consistent with a left-sided exit within the coronary cusp or subaortic left ventricular outflow tract.
The unexpected high prevalence of RBBB VTs arising from the RVOT may be explained by the anterocephalic deviation of the infundibular and often muscular outlet septum in TOF, which results in early activation of the left ventricle if AI3 is clockwisely activated during VT (19,20).
Identification of VT-related AI for LBBB morphology VT in rTOF patients is less straightforward.
In idiopathic LBBB VTs, the QRS transition becomes progressively later as the VT site of origin moves from high septal toward the RV free wall (14,15).
However, in rTOF the presence and extension of surgical scars from prior repair further influences the RV activation pattern and hence the LBBB VT morphology. The VT QRS transition progresses if the RV incision or TA patch extends further into the RV. The resulting late transition may direct pace mapping and ablation to AI1 (which may be less suitable for ablation), or may even suggest a remote substrate not related to the RVOT. In the absence of AI2, AI1 and AI3 were both part of the re-entry circuit. Accordingly, in group 2, we could demonstrate that a good pace match could be obtained from both AIs.
Implications for VT ablation in contemporary patients with rTOF
Our findings suggest that pace mapping to identify ablation target sites might no longer be necessary in the majority of rTOF patients with re-entry VTs. The presence of AIs can often be obtained from prior operation reports. If AI2 is not present, AI3 can always be targeted irrespective of the 12-lead VT morphology. If AI2 is present, EAM can be used to evaluate which AI is most easily approachable by ablation. In rare cases with presence of AI4, mapping is required to identify the VT-related isthmus.
The ability to target the VT substrate without the need for pace or activation mapping is appealing because it eliminates the need of inducibility of the clinical VT and is likely to shorten procedure time and increase patient comfort.
In our study, we did not completely define re-entry circuits in the majority of induced VTs because of hemodynamic intolerance. Our study cohort is relatively small and derived from a tertiary referral center with expertise in TOF patients, which might have influenced our results. Applying our systematic pace mapping protocol irrespective of VT morphology in a large, mixed prospective cohort is desirable to support our findings.
In rTOF patients with AI-related VTs, AI3 is always a part of the re-entry circuit in the absence of AI2 and AI4. VTs with RBBB morphology result from clockwise activation and VTs with LBBB morphology from counterclockwise activation of AI3. Variations in precordial transition of LBBB VTs depend on the geometry of AI1. Involvement of both AI1 and AI3 limits the role of the 12-lead VT ECG and pace mapping in determining the critical VT isthmus. AI3, which is usually narrower and easier to approach, can always be targeted. In presence of AI2, ablation of AI3 and AI2 should be considered.
COMPETENCY IN MEDICAL KNOWLEDGE: In patients with rTOF and AI-related VT, AI3 is always part of the VT re-entry circuit in absence of AI2 and AI4. AI3 is narrower and easier accessible by ablation compared with AI1. Involvement of both AI1 and AI3 in the re-entry circuit makes substrate identification by pace mapping obsolete. Ablation of AI 3 can be performed without the need for pace mapping first. In the presence of AI2, ablation of both AI 2 and AI3 should be considered.
TRANSLATIONAL OUTLOOK: This study is performed in a selected group of rTOF patients in a tertiary referral center with expertise in patients with congenital heart disease. Our study cohort is relatively small. Larger prospective studies applying our systematic pace-mapping protocol irrespective of VT morphology in rTOF patients would be desirable.
The Department of Cardiology Leiden has received unrestricted research and fellowship grants from Abbott, Boston Scientific, Medtronic, and Biotronik. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
All 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
- anatomical isthmus
- electroanatomical mapping
- left bundle branch block
- pulmonary valve
- right bundle branch block
- repaired tetralogy of Fallot
- right ventricular
- right ventricular outflow tract
- ventricular septal defect
- ventricular tachycardia
- Received February 6, 2018.
- Revision received June 11, 2018.
- Accepted June 19, 2018.
- 2018 American College of Cardiology Foundation
- Marelli A.J.,
- Ionescu-Ittu R.,
- Mackie A.S.,
- Guo L.,
- Dendukuri N.,
- Kaouache M.
- Marelli A.,
- Mackie A.,
- Ionescu-Ittu R.,
- Rahme E.,
- Pilote L.
- Kapel G.F.,
- Reichlin T.,
- Wijnmaalen A.P.,
- et al.
- Khairy P.,
- Aboulhosn J.,
- Gurvitz M.Z.,
- et al.
- Zeppenfeld K.,
- Schalij M.J.,
- Bartelings M.M.,
- et al.
- Kapel G.F.
- Kapel G.,
- Laranjo S.,
- Blom N.A.,
- et al.
- Moore J.P.,
- Seki A.,
- Shannon K.M.,
- Mandapati R.,
- Tung R.,
- Fishbein M.C.
- Kriebel T.,
- Saul J.P.,
- Schneider H.,
- Sigler M.,
- Paul T.
- Furushima H.,
- Chinushi M.,
- Sugiura H.,
- et al.
- Khairy P.,
- Harris L.,
- Landzberg M.J.,
- et al.
- de Riva M.,
- Watanabe M.,
- Zeppenfeld K.
- Stevenson W.G.,
- Sager P.T.,
- Natterson P.D.,
- Saxon L.A.,
- Middlekauff H.R.,
- Wiener I.
- Padalino M.A.,
- Cavalli G.,
- Albanese S.B.,
- et al.
- Kapel G.F.,
- Reichlin T.,
- Wijnmaalen A.P.,
- et al.
- Anderson R.H.,
- Jacobs M.L.