Author + information
- Received April 1, 2013
- Revision received July 11, 2013
- Accepted August 6, 2013
- Published online December 24, 2013.
- Joost Lumens, PhD∗,†∗ (, )
- Sylvain Ploux, MD∗,†,
- Marc Strik, MD†,
- John Gorcsan III, MD‡,
- Hubert Cochet, MD∗,
- Nicolas Derval, MD∗,
- Maria Strom, PhD§,
- Charu Ramanathan, PhD§,
- Philippe Ritter, MD∗,
- Michel Haïssaguerre, MD∗,
- Pierre Jaïs, MD∗,
- Theo Arts, PhD†,
- Tammo Delhaas, MD, PhD†,
- Frits W. Prinzen, PhD† and
- Pierre Bordachar, MD, PhD∗
- ∗Hôpital Cardiologique du Haut-Lévêque, CHU Bordeaux, L'Institut de rythmologie et modélisation cardiaque (LIRYC), Université Bordeaux, Bordeaux, France
- †Maastricht University Medical Center, Cardiovascular Research Institute Maastricht (CARIM), Maastricht, the Netherlands
- ‡University of Pittsburgh, Pittsburgh, Pennsylvania
- §CardioInsight Technologies, Cleveland, Ohio
- ↵∗Reprint requests and correspondence:
Dr. Joost Lumens, Maastricht University Medical Center, Cardiovascular Research Institute Maastricht (CARIM), Universiteitssingel 50, P.O. Box 616, 6200MD Maastricht, the Netherlands.
Objectives The purpose of this study was to enhance understanding of the working mechanism of cardiac resynchronization therapy by comparing animal experimental, clinical, and computational data on the hemodynamic and electromechanical consequences of left ventricular pacing (LVP) and biventricular pacing (BiVP).
Background It is unclear why LVP and BiVP have comparative positive effects on hemodynamic function of patients with dyssynchronous heart failure.
Methods Hemodynamic response to LVP and BiVP (% change in maximal rate of left ventricular pressure rise [LVdP/dtmax]) was measured in 6 dogs and 24 patients with heart failure and left bundle branch block followed by computer simulations of local myofiber mechanics during LVP and BiVP in the failing heart with left bundle branch block. Pacing-induced changes of electrical activation were measured in dogs using contact mapping and in patients using a noninvasive multielectrode electrocardiographic mapping technique.
Results LVP and BiVP similarly increased LVdP/dtmax in dogs and in patients, but only BiVP significantly decreased electrical dyssynchrony. In the simulations, LVP and BiVP increased total ventricular myofiber work to the same extent. While the LVP-induced increase was entirely due to enhanced right ventricular (RV) myofiber work, the BiVP-induced increase was due to enhanced myofiber work of both the left ventricle (LV) and RV. Overall, LVdP/dtmax correlated better with total ventricular myofiber work than with LV or RV myofiber work alone.
Conclusions Animal experimental, clinical, and computational data support the similarity of hemodynamic response to LVP and BiVP, despite differences in electrical dyssynchrony. The simulations provide the novel insight that, through ventricular interaction, the RV myocardium importantly contributes to the improvement in LV pump function induced by cardiac resynchronization therapy.
- ventricular interaction
- myocardial work
- cardiac resynchronization therapy
- electrophysiology mapping
Cardiac resynchronization therapy (CRT) is an effective treatment for patients with chronic heart failure (HF), decreased left ventricular (LV) ejection fraction (≤35%), and left bundle branch block (LBBB) (1,2). Its working action is generally believed to originate from resynchronization of the LV and right ventricular (RV) electrical activation, achieved by biventricular pacing (BiVP).
Paradoxically, single-site left ventricular pacing (LVP) has been shown to be as beneficial as BiVP for LV systolic pump function in acute hemodynamic studies (3–5), in long-term follow-up studies (6–8), and even in situations where LVP is unlikely to result in fusion of 2 activation wave fronts induced by LVP and intrinsic conduction (5,9). Therefore, the question arises whether electrical resynchronization is the primary working mechanism underlying the functional improvement induced by CRT. It is well known that ventricular pacing redistributes mechanical work in the LV wall so that the region of latest activation is associated with highest mechanical work (10). However, it is not known to what extent ventricular pacing affects mechanical work generated by the RV myocardium. Because direct mechanical coupling of the ventricles allows transmission of myocardial work between the ventricles, we hypothesize that a pacing-induced increase of RV myocardial work can benefit LV pump function.
To test this hypothesis, we measured local electrical and global hemodynamic function in an animal model of chronic HF with LBBB and in CRT candidates during baseline (LBBB), LVP, and BiVP. Furthermore, we used a computer model of the human heart and circulation (11–13) to investigate the consequences of LVP and BiVP for local LV and RV tissue mechanics. Together, these complementary data provide novel insights in the working mechanism of CRT, especially regarding the involvement of the RV myocardium in its hemodynamic effect.
Animal handling was performed according to the Dutch Law on Animal Experimentation and the European Directive for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes (86/609/EU). The protocol was approved by the Experimental Animal Committee of Maastricht University.
In 6 adult mongrel dogs (29 ± 3 kg), LBBB was induced by radiofrequency ablation and, subsequently, HF was induced by 4 weeks of tachypacing (14). Continuous, invasive hemodynamic and electrocardiographic measurements were performed during right atrial pacing at approximately 10 beats/min above intrinsic heart rate (baseline) and during atrial paced LVP and BiVP at the same heart rate and at short atrioventricular (AV) delay, ensuring full ventricular capture as noticed on the surface electrocardiogram. More details of the experimental protocol are provided in Online Appendix A.
Electrical activation maps were used to calculate 2 indexes of electrical dyssynchrony: total ventricular activation time (ATTOT) derived from all electrodes and LV activation time derived from the septal and LV free wall electrodes only (14).
The execution of the study conformed to the principles outlined in the Declaration of Helsinki on research in human subjects. The study protocol was approved by the Medical Ethics Committee of CHU Bordeaux. All patients granted their written approval to participate in the study.
The study included 24 consecutive patients who fulfilled the following criteria: 1) New York Heart Association functional class II, III, or IV, despite optimal medical therapy; 2) LV ejection fraction ≤35% during sinus rhythm; 3) QRS duration ≥120 ms; and 4) LBBB morphology on the surface electrocardiogram. Both QRS duration and LBBB morphology were defined according to the most recent American Heart Association/American College of Cardiology Foundation/Heart Rhythm Society recommendations (15). Etiology was considered ischemic in the presence of significant coronary artery disease (≥50% stenosis in 1 or more of the major epicardial coronary arteries), history of myocardial infarction, or prior revascularization.
Device Implantation, Pacing Protocol, and Assessment of Hemodynamic Function
All patients were implanted with a CRT device with leads in the RV apex and in a lateral or posterolateral branch of the coronary sinus. Within 72 h after device implantation, a high-fidelity pressure-recording micromanometer (Radi Medical Systems, St. Jude Medical, St. Paul, Minnesota) was introduced in the LV cavity. LV pressure data were acquired (16) during baseline (AAI mode; 10 beats/min above intrinsic heart rate) and during atrial paced LV and biventricular stimulation (DDD mode). The AV delay was set to the longest delay that did not lead to fusion between electrical activation waves originating from intrinsic RV conduction and from the LV pacing electrode during LVP. The same AV delay was used during BiVP with simultaneous LV-RV stimulation. Hemodynamic response was defined as % change in maximal rate of LV pressure rise (LVdP/dtmax) relative to baseline.
Noninvasive Electrocardiographic Mapping
In a subset of 10 patients, we used noninvasive, high-resolution electrocardiographic mapping (ECM) (CardioInsight Technologies Inc., Cleveland, Ohio) to acquire ventricular epicardial activation maps during baseline, LVP, and BiVP (17,18) and to quantify electrical dyssynchrony (ATTOT and LV activation time).
The CircAdapt model of heart and circulation (11,19) was used to quantify the acute effects of LVP and BiVP on ventricular mechanics and hemodynamics of the failing heart with LBBB. The model consists of modules representing cardiac walls, cardiac valves, large blood vessels, systemic and pulmonary peripheral vasculature, the pericardium, and local cardiac myofiber mechanics (Online Appendix B). It enables realistic beat-to-beat simulation of cardiovascular mechanics and hemodynamics under a wide variety of (patho-)physiological circumstances, including ventricular mechanical dyssynchrony (12,13).
First, mechanics and hemodynamics of the normal cardiovascular system with nonfailing myocardium and synchronous activation of the ventricular walls were simulated, as published previously (12,13). Second, a failing heart with LBBB was simulated (Online Appendix C). Third, LVP and BiVP were simulated so that they were in agreement with the electrocardiographic mapping data obtained in the patients and dogs, that is, LVP did not change ATTOT (135 ms), whereas BiVP was assumed to reduce ATTOT from 135 to 60 ms (Online Appendix C).
Local Ventricular Myofiber Mechanics
Simulated time courses of local Cauchy myofiber stress and natural strain were used to quantify regional differences in mechanical load and deformation of the myocardial tissue during LBBB, LVP, and BiVP. Peak systolic myofiber stress and external myofiber work were quantified as indexes of local myocardial tissue load. External myofiber work, expressed in joule per cardiac cycle (J/beat), was defined as the area enclosed by the stress-strain relation multiplied by tissue volume of the myocardial segment, which equaled 8.5 ml for each ventricular wall segment.
Values are presented as mean ± SD for continuous variables and as numbers and percentages for discrete variables. Statistical analysis was performed with the IBM SPSS Statistics 20 package for Windows (IBM Corp., Armonk, New York). Assumptions on homogeneity of variances and normality of residual distributions were checked using Mauchly's test of sphericity and Q-Q plots, respectively. One-way repeated measures analysis of variance (ANOVA) was used to test for significant effects of LVP and BiVP on baseline electrical and hemodynamic function parameters. If the sphericity assumption appeared to be violated, the Greenhouse-Geisser correction was used to adjust degrees of freedom for the averaged results of the ANOVA. If ANOVA showed significance, pairwise post-hoc analysis for differences between the 3 pacing conditions (no pacing/LVP/BiVP) was performed using the Fisher Least Significant Difference method. A p value <0.05 was considered statistically significant for all analyses.
Dogs and patients
Baseline conditions of dogs and patients are presented in Tables 1 and 2⇓⇓, respectively. Paced AV delay was relatively short compared with the PR interval in dogs (86 ± 26 ms vs. 141 ± 40 ms, respectively) as well as in patients (106 ± 19 ms vs. 213 ± 30 ms).
LVP and BiVP Similarly Improve Systolic LV Function
Both LVP and BiVP similarly increased LVdP/dtmax compared with baseline in dogs (LVP vs. BiVP; 21 ± 19% vs. 19 ± 17%; p = 0.33) (Table 1) and patients (16 ± 13% vs. 16 ± 11%; p = 0.95) (Table 3). Animal experimental data showed a trend toward increased LV stroke volume, pump stroke work, and systolic peak pressure during LVP and BiVP as compared with baseline, while LV end-diastolic volume and pressure remained unchanged (Table 1). In contrast, RV systolic peak pressure and maximal rate of right ventricular pressure rise (RVdP/dtmax) were decreased during LVP as compared with baseline and BiVP.
BiVP, But Not LVP, Reduces Electrical Dyssynchrony
Ventricular electrical activation maps of dogs and patients revealed the same characteristics (Fig. 1): during baseline, a classical LBBB pattern of electrical activation starting at the lateral RV free wall and gradually spreading towards the lateral LV free wall; during LVP, a mirrored LV-to-RV pattern of epicardial activation; and during BiVP, 2 fusing wave fronts of activation originating from the LV and RV pacing sites. In addition, the canine data showed that the septum is activated in an RV-to-LV transmural direction during baseline and BiVP and in an LV-to-RV direction during LVP. Compared with baseline, BiVP significantly reduced electrical dyssynchrony in dogs and in patients (Fig. 2), whereas LVP did not. In the dogs, activation times were significantly shorter during BiVP than during LVP (Table 1). In the patients, only ATTOT was significantly shorter during BiVP than during LVP (Table 3).
The model simulations also showed that LVP and BiVP similarly increased LVdP/dtmax by 15% (Table 4), despite the longer ventricular activation time during LVP. As in the dogs, both pacing strategies increased LV stroke volume, pump stroke work, and systolic peak pressure (Table 4), and LVP decreased RVdP/dtmax compared with baseline. In addition, simulations revealed that both LVP and BiVP increased RV pump stroke work by 16%.
LVP and BiVP Differently Affect Local Ventricular Myofiber Mechanics
Pronounced local differences are present in the pattern and amplitude of myofiber strain during baseline (LBBB), LVP, and BiVP (Fig. 3). Early-activated segments are characterized by rapid onset of systolic myofiber shortening followed by rebound stretch and, in some cases, a second phase of shortening at the end of systole. In late-activated regions, early-systolic stretch is followed by pronounced systolic myofiber shortening.
The regional differences in strain patterns translated into differences in local mechanical tissue load (Fig. 3: color maps). In the LBBB simulation, most mechanical myofiber work was generated by the LV free wall segments, whereas the RV free wall and septal segments generated little mechanical work or even dissipated mechanical work, as evidenced by the clockwise stress-strain relations (Fig. 3). Compared with LBBB, LVP reallocated mechanical work from the LV free wall to the septum, resulting in a spatially mirrored but equally dispersed distribution of mechanical work over the LV myocardium. BiVP was associated with less early-systolic myofiber stretch and shortening and a more homogeneous distribution of myofiber work than LVP (Fig. 3). In contrast, LV peak systolic myofiber stress was more homogeneously distributed during LVP, whereas the average values did not differ between LVP and BiVP (92 ± 7 kPa and 92 ± 13 kPa, respectively).
LBBB and LVP were associated with a comparable net amount of mechanical myofiber work generated by the LV myocardium (Fig. 4). The RV myocardium, however, generated more work during LVP than during LBBB. As a result, LVP acutely increased total ventricular myofiber work by 25%. BiVP resulted in a similar increase of total myofiber work (23%) as LVP, but now due to an increase of both LV and RV myofiber work.
Ventricular Interaction: Contribution of RV Myocardium to LV Pump Function
A more precise study on the role of left–right ventricular interaction on hemodynamic response to pacing therapy was performed by simulating LVP and BiVP with 5 different AV delays (60/80/100/120/140 ms) as well as 5 different velocities of activation, which resulted in a range of values for ATTOT (24/36/48/60/72 ms during BiVP and 54/81/108/135/162 ms during LVP). For the resulting 50 simulations, Figure 5 shows the relationship between ventricular myofiber work and LVdP/dtmax. The left panel indicates that total ventricular myofiber work increased linearly with LVdP/dtmax and that this linear relationship was virtually independent of the pacing mode. However, LVP and BiVP behaved differently when considering LV and RV myofiber work separately (Fig. 5, middle and right panel, respectively). While LVP and BiVP can lead to the same total ventricular myofiber work and LVdP/dtmax, their distribution of myofiber work over the LV and RV myocardium is rather different. During BiVP, the relative contribution of the RV myocardium to total ventricular myofiber work was rather constant and ranged from 22% to 24%. This contribution was considerably more variable during LVP, and increased from 28% in the simulation with highest conduction velocity to 38% in that with lowest. Overall, these simulation data highlight the important role of the RV myocardium as a contributor to LV pump function during LVP and, thus, the importance of ventricular interaction during CRT.
In the present study, we demonstrate that LVP and BiVP improve the systolic function of the dyssynchronous failing heart to a similar extent, both in experimental animals and in patients. With state-of-the-art techniques for electrical mapping, we showed in patients that pacing-induced hemodynamic improvement can occur without electrical resynchronization. These findings are corroborated by computer simulations, which showed that both pacing strategies increase total ventricular myofiber work to a similar extent, yet differently redistribute myofiber load over the LV and RV myocardium. Overall, LV systolic function correlates better with total ventricular myofiber work rather than with LV or RV myofiber work alone. These data provide the novel insight that left–right ventricular interaction is an important determinant of the hemodynamic effect of pacing therapy in dyssynchronous HF.
RV mechanical work: the missing link in the explanation for similarity of response to LVP and BiVP?
Our finding that LVP and BiVP improve LV systolic function to the same extent corroborates previously published data on acute hemodynamic response (3,4) and on long-term clinical response and reverse remodeling (6–8). In addition, the present study provides a potential mechanism underlying these observations.
It is known that contractile harmony is substantially disturbed in patients with LBBB or pacing-induced electrical dyssynchrony and that this contractile discordance compromises cardiac pump function. Regional differences in the systolic deformation pattern are related to local differences in sarcomere length and, consequently, myofiber contractile force (20) and work load (10). The simulations are in close agreement with experimental findings demonstrating that mechanical myofiber work is small or even negative in regions close to the pacing site and large in regions remote from the pacing site (10). So far, these insights remained limited to the LV wall. Our simulations show that the RV myocardium contributes significantly to the improvement of LV pump function in pacing therapies, especially LVP. While it may be intuitive that BiVP improves LV pump function by increasing LV myofiber work, it may be less intuitive that LVP similarly improves LV systolic pump function by exclusively increasing the amount of mechanical work generated by the RV myocardium. These findings emphasize the importance of ventricular interaction, that is, the property that the RV myocardium contributes to LV systolic pump function and vice versa.
Simulations of LVP and BiVP in hearts with different conduction velocities (Fig. 5) revealed that, during LVP, the relative contribution of RV myofiber work to total ventricular myofiber work increased with total ventricular activation time, whereas it stayed constant during BiVP. These simulation data suggest that LVP is less effective than BiVP in patients with slow intramyocardial conduction, with diminished RV contractile function, or in whom mechanical ventricular interaction is being impeded.
While indirect hemodynamic interaction results from the series coupling of the ventricles via the pulmonary and systemic circulations, direct mechanical interaction is due to the anatomical coupling via the interventricular septum and the surrounding pericardium (21). Because our animal experiments show no direct effect of LVP on indexes of LV filling, such as LV end-diastolic pressure and volume (Table 1), the positive effect of LVP on LV systolic pump function most probably results from direct mechanical interaction. Furthermore, the decreased values of RV systolic pressure and RVdP/dtmax with LVP suggest that the extra amount of mechanical work generated by the RV myocardium is largely converted into LV pump work through direct mechanical interaction.
Clinical implications and future perspectives
The demonstration that, during CRT, the RV myocardium can contribute to LV pump function and that this contribution differs between LVP and BiVP may explain why some patients respond better to LVP and others to BiVP, as demonstrated in the GREATER-EARTH (Evaluation of Resynchronization Therapy for Heart Failure in patients with a QRS duration >120 ms) study (6). We hypothesize that local differences in myocardial contractility (e.g., due to infarction, hibernation, and so on) determine a patient's response to LVP and BiVP in a way that hemodynamic improvement is compromised when the region with impaired contractile function coincides with the location of latest activation and, hence, highest mechanical load. Although experimental data point in this direction (22), it remains to be confirmed with prospective clinical studies.
Many studies demonstrated the acute deleterious effect of RV pacing on LV systolic function in terms of LVdP/dtmax (3,4). Similarly, our experimental and simulation data revealed that LVP acutely decreased RVdP/dtmax (Tables 1 and 4). Our simulations, however, additionally showed that RV pump stroke work was increased during both LVP and BiVP (Table 4). Therefore, it is questionable whether the pacing-induced decrease of RVdP/dtmax should be considered a sign of acute RV systolic impairment. Our simulations also showed that LVP increased mechanical myofiber work of the RV myocardial tissue by more than 100% (Fig. 4). Whether this acute LVP-induced increase of RV tissue load translates into compensatory RV remodeling and eventually RV decompensation and failure remains unknown and should be subject of future research.
In the present study, we evaluated the acute hemodynamic effect of CRT. Whether the observed acute hemodynamic improvements will evolve in chronic response to CRT, in terms of hard clinical endpoints or reverse remodeling, is unclear and should be the subject of future research.
In dogs and patients, LVP and BiVP were applied with atrial pacing at a short AV delay to ensure a constant heart rate and the absence of fusion between electrical activation waves originating from intrinsic conduction and from pacing electrode(s). These conditions have been chosen to clearly show the proof of principle that a pacing-induced hemodynamic benefit can be obtained in the absence of fusion in the case of LVP. Hence, our study is conceptually different from a previous study showing noninferiority of fusion-synchronized LVP compared with conventional simultaneous BiVP (23). We acknowledge that the AV delays used in our study may not have been the ones leading to optimal LV filling or systolic function. In a previous acute hemodynamic study (3), however, maximal aortic systolic or pulse pressure was observed at an AV delay of approximately 0.5 × (PR interval − 30 ms) for both LVP and BiVP. Applying this formula to our patient data, we obtained a predicted optimal AV delay of 92 ± 15 ms, which is close to the AV delay programmed in this study (106 ± 19 ms). Furthermore, the average paced AV delay in the patients was in good agreement with the value reported by Thibault et al. (6) in the GREATER-EARTH study (101 ± 16 ms), a study that also compared the effectiveness of LVP and BiVP in a conventional CRT population.
The multimodality of our study approach may have complicated interpretation of the results. At the same time, however, the consistency of the hemodynamic and electrocardiographic response to LVP and BiVP in animals, patients, and simulations provides firm evidence that electrical resynchronization is not always required for pacing therapy to improve systolic cardiac pump function. The invasive ECM data obtained in the dogs served as a control technique for our clinical ECM data, which was obtained by noninvasive indirect mapping of epicardial electrical activation. The animal experimental protocol also included measurement of RV pressure data. These data enabled evaluation of the effects of LVP and BiVP on RV systolic function. The simulation data for RV function showed good agreement with the experimental data, that is, LVP was associated with lower RVdP/dtmax than BiVP.
The computational model used in this study inherently provides a simplified representation of an average patient's failing heart with LBBB. Therefore, the conclusions drawn from these data should be interpreted with care. However, many model predictions agreed with measurements in patients and experimental animals. Moreover, the simplifications allowed a transparent view on complex fundamental mechanisms, which are hard to assess in experimental or clinical settings.
LVP and BiVP improve LV hemodynamic function to the same extent, despite substantial differences in electrical dyssynchrony. Both pacing strategies similarly increase total ventricular myofiber work, which is tightly linked with LV pump function. Our simulations show that CRT can improve LV systolic function by mechanical recruitment of the RV myocardium.
This research was performed within the framework of CTMM, the Center for Translational Molecular Medicine, project COHFAR (grant 01C-203) cofunded by the Dutch Heart Foundation. The study was supported by the French Government, Agence National de la Recherche au titre du programme Investissements d'Avenir (ANR-10-IAHU-04). Dr. Lumens received a grant within the framework of the Dr. E. Dekker program of the Dutch Heart Foundation (NHS-2012T010). Dr. Ploux was financially supported by “la Fédération Française de Cardiologie.” Dr. Gorcsan has received research grants from Biotronik, Medtronic, Toshiba, and St. Jude Medical; and is a consultant for CardioInsight Technologies Inc, GE, Toshiba, Biotronik, Medtronic, and St. Jude Medical. Drs. Strom and Ramanathan are paid employees and stock owners of CardioInsight Technologies Inc. Dr. Ritter has served as a consultant for Sorin CRM and Medtronic and has received consultant honoraria. Drs. Haïssaguerre and Jaïs are stock owners of CardioInsight Technologies Inc. Dr. Prinzen has received research grants from Medtronic, EBR Systems, and Merck Sharp & Dohme. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- analysis of variance
- total ventricular activation time
- biventricular pacing
- cardiac resynchronization therapy
- electrocardiographic mapping
- heart failure
- left bundle branch block
- left ventricle/ventricular
- % change in maximal rate of left ventricular pressure rise
- left ventricular pacing
- right ventricle/ventricular
- % change in maximal rate of right ventricular pressure rise
- Received April 1, 2013.
- Revision received July 11, 2013.
- Accepted August 6, 2013.
- 2013 American College of Cardiology Foundation
- Auricchio A.,
- Stellbrink C.,
- Block M.,
- et al.,
- The Pacing Therapies for Congestive Heart Failure Study Group. The Guidant Congestive Heart Failure Research Group
- Blanc J.J.,
- Etienne Y.,
- Gilard M.,
- et al.
- Thibault B.,
- Ducharme A.,
- Harel F.,
- et al.
- Gasparini M.,
- Bocchiardo M.,
- Lunati M.,
- et al.
- Boriani G.,
- Kranig W.,
- Donal E.,
- et al.
- Leclercq C.,
- Faris O.,
- Tunin R.,
- et al.
- Prinzen F.W.,
- Hunter W.C.,
- Wyman B.T.,
- McVeigh E.R.
- Leenders G.E.,
- Lumens J.,
- Cramer M.J.,
- et al.
- Lumens J.,
- Leenders G.E.,
- Cramer M.J.,
- et al.
- Strik M.,
- Rademakers L.M.,
- van Deursen C.J.,
- et al.
- Surawicz B.,
- Childers R.,
- Deal B.J.,
- et al.
- Derval N.,
- Steendijk P.,
- Gula L.J.,
- et al.
- Ploux S.,
- Lumens J.,
- Whinnett Z.,
- et al.
- Arts T.,
- Delhaas T.,
- Bovendeerd P.,
- Verbeek X.,
- Prinzen F.W.
- ter Keurs H.E.,
- Rijnsburger W.H.,
- van Heuningen R.,
- Nagelsmit M.J.
- Rademakers L.M.,
- van Kerckhoven R.,
- van Deursen C.J.,
- et al.