Author + information
- Received March 13, 2017
- Revision received August 14, 2017
- Accepted August 17, 2017
- Published online October 16, 2017.
- Jacob S. Koruth, MDa,∗ (, )
- Jin Iwasawa, MDa,
- Yoshinari Enomoto, MDa,
- Meir Bar-Tal, MScb,
- Yigal Ultchin, PhDb,
- Alona Sigal, BScb,
- Liron Mizrahi, MScb,
- Abraham Berger, MScc,
- Ori Hazan, MScc,
- Srinivas R. Dukkipati, MDa and
- Vivek Y. Reddy, MDa
- aHelmsley Electrophysiology Center, Mount Sinai Medical Center, New York, New York
- bBiosense Webster, Inc., Tirat-Hacarmel, Israel
- cBerger Thermal Research, Tel Aviv, Israel
- ↵∗Address for correspondence:
Dr. Jacob S. Koruth, Helmsley Electrophysiology Center, Mount Sinai Hospital and School of Medicine, One Gustave L. Levy Place, P.O. Box 1030, New York, New York 10029.
Objectives This study sought to compare a novel lesion dimension estimation approach to actual measurements of lesion dimensions on necropsy in porcine atria and ventricles.
Background An irrigated-tip, force-sensing radiofrequency catheter with 6 temperature (tip-tissue interface) sensors allows for assessment of lesion dimensions based on estimated tissue temperature. Lesion dimension assessment has not been attempted previously in atrial tissue.
Methods Ablations were performed using this catheter in all chambers. Irrigated radiofrequency was delivered using 20 to 50 W for durations that ranged from 15 to 90 s with contact force ranging from 5 to 45 g to replicate a wide spectrum of clinical conditions. All swine were then sacrificed and lesions were identified and photographed. Three independent observers made offline measurements, which were then averaged to obtain lesion width and depth for comparison with estimated dimensions based on interface tissue temperature.
Results In 9 swine, 54 atrial and 61 ventricular lesions were assessed. In the atria, the mean difference between the measured and estimated depth and width was 0.9 ± 0.74 mm and 1.2 ± 0.9 mm, respectively. Eighty percent of all lesions had a difference of ≤1.7 mm for depth and ≤1.74 mm for width. In the ventricle, the mean difference between the measured and estimated depth and width was 0.75 ± 0.6 mm and 1.66 ± 1.1 mm, respectively. Eighty percent of all lesions had a difference of ≤1.1 mm ventricular depth and ≤2.6 mm for width.
Conclusions Estimation of lesion dimensions can be achieved with clinically relevant accuracy using unique temperature signatures. These data have important implications for understanding the adequacy of lesion overlap and assessment of transmurality.
The ability to predict the dimensions of a lesion during radiofrequency (RF) applications is an unmet need in catheter ablation. Such a capability could reduce the occurrence of nontransmural or noncontiguous lesions and reduce collateral damage by avoiding delivery of RF current beyond what is necessary. Although several approaches to lesion assessment have been proposed, measurement of tissue temperature during RF ablation is uniquely attractive, given that thermal injury is the fundamental determinant of lesion formation with RF current (1,2). Thus, it is reasonable to postulate that tissue temperature estimation during irrigated RF ablation would be an attractive parameter for lesion assessment. However, it is well described that the hot spot created during irrigated RF is located subendocardial to the point of contact of the catheter tip and tissue. This prevents thermocouples placed within catheter tips from recording the tissue temperature accurately (3,4).
One approach to the estimation of tissue temperature is to use a tip design wherein thermally isolated thermocouples are placed along the ablative tip surface. This design allows for interface temperature assessment and can provide a global perspective of the thermal environment at the interface despite the presence of external irrigation. This is particularly important with newer generations of irrigated catheter tips, for which the degree of tip cooling can be significant (5). Additionally, assessment of the interface in this manner might allow insight into the thermal dynamics of the RF lesion. Such a capability could improve safety by minimizing the occurrence of char and steam pops that are related to high temperatures. In this preclinical porcine study, we sought to investigate the performance of a novel lesion estimation approach that incorporates interface temperature measurements during irrigated RF ablation in atria and ventricles.
The preclinical experiments were approved by the institutional animal care and use committees. After an overnight fast, 9 swine (weight >70 kg) were pre-medicated, intubated, and mechanically ventilated with isoflurane (1% to 3%). Percutaneous femoral venous access was obtained, and transseptal puncture was performed after systemic heparin administration to gain access to the left atrium (LA) and left ventricle (LV). A deflectable sheath was used for all ablations. Detailed electroanatomic maps (Carto 3 V4, Oracle System, Biosense Webster, Irvine, California) were created using a multielectrode mapping catheter (PentaRay, Biosense Webster) for all chambers to define anatomy. Discrete RF applications were then placed with additional guidance from intracardiac echocardiography to facilitate lesion identification on autopsy. Trabeculated portions of the atria were avoided. For each lesion, RF energy was delivered using an interrupted point-by-point ablation technique during power control mode after stable catheter position had been confirmed. All animals were pre-medicated with intravenous beta-blockers and lidocaine to reduce the risk of ventricular arrhythmias during ventricular ablation.
All swine were humanely sacrifice at the end of each experiment. The heart was explanted and then perfusion stained with tetrazolium chloride. Individual lesions were identified on the basis of their anatomic position documented during ablation and also by comparison to locations on the electroanatomic maps. Each identified lesion was then sectioned (tissue was not fixed with formalin) centrally along its long axis and photographed with a digital microscope. Three separate observers independently measured the width and depth of the lesions using Image J (National Institutes of Health) software. Observers were blinded to RF and temperature data. The mean of all 3 measurements was then taken to represent the lesion dimensions.
The catheter (Oracle, Biosense Webster) is a modification of the force-sensing, deflectable, 56-hole irrigated catheter (ThermoCool SmartTouch Surround Flow, Biosense Webster). Its 3.5-mm ablative tip has embedded along its surface 6 microelectrodes that are both electrically and thermally isolated from the surface of the catheter tip. Their construction includes a layer of polyimide insulation that surrounds the microelectrode, except for the exposed outermost portion that contacts tissue. Each microelectrode is connected to a pair of copper/constantan wires. Thus during ablation, the temperature at the microelectrode can be obtained by reading the voltage between the copper and the constantan wires and serve as thermocouples (Figure 1). Ablation was performed using an RF generator (EP Shuttle, Stockert, Freiburg, Germany) and a dispersive pad. A proprietary module provided real-time display of the tip-tissue interface temperature, numerical display of estimated depth and width of the lesion during RF and estimated in-tissue hot-spot temperatures, and display of signals from each of the microelectrodes.
Irrigation was provided by a custom pump (CoolFlow, Biosense Webster). The settings used were as follows: 1) for powers 20 to 30 W, temperature cutoff was set to 75°C with irrigation rate of 8 ml/min; and 2) for powers >30 W, the irrigation rate was set to 15 ml/min. RF parameters were selected to reflect the wide variation of settings used in clinical practice. Power settings were 20, 25, 30, 35, 40, and 50 W, and durations ranged between 15 and 90 s.
The approach to lesion estimation was based on simulated modeling of the catheter tip during RF ablation. This model is a multiphysics coupled model that incorporates the following: 1) the catheter tip and surrounding tissue structures, with an assumption that the tip remains stationary during ablation; 2) the electrical field generated from the current source, assumed to be constant; 3) the electrical and thermal properties of the tissue and their relation to temperature; 4) the fluid dynamics of the irrigation fluid and its effect on the thermal field; and 5) the lethal temperature isotherm of 57°C, which was selected to define the lesion size for this study.
To simplify the model, a single parameter of catheter penetration depth was used to represent the possible configurations of the contact interface. This in turn permitted the use of a simplified geometry for the catheter-tissue interface. The catheter itself was based on the actual structure of the study catheter, with respect to its dimensions and materials, with some simplifications made to its internal structure. Similarly, the tissue was simplified into layers of constant thickness, with an overall flat cylindrical shape.
The main result of the simulation was the time-dependent development of the temperature field, which was used to produce the database used by the lesion assessment algorithm. A set of simulations were executed for various scenarios that spanned all combinations of ablation currents, degrees of catheter-tip penetration, and durations to generate the database. The ablations were assumed to continue up to 125 s, with the current in the range 0.45 to 0.71 A, which is equivalent to 20 to 50 W, thus covering the typical ranges used clinically. The penetration depths at which the simulations were run included 0.01 to 1.2 mm for thin atria tissue, 0.01 to 1.5 mm for thin ventricular tissue and thick atrial tissue, and 0.01 to 3.0 mm for thick ventricular tissue, the penetration ranges being a function of both tissue thickness and mechanics (e.g., stiffness). The penetration depth is a parameter that represents the degree of electric and thermal contact between the catheter and the tissue, which parameterizes all unknown factors such as angle and variability in material properties. Each specific simulation run generated a temperature field around the catheter as a function of time (Figure 2). Lesion dimensions and maximum temperature (hot spot) achieved in the tissue were then defined for each simulation.
Several different physical models related to the wall thicknesses were also taken into consideration (Figure 3). For lesion assessment, estimation was divided into different models: thick (∼10 mm), intermediate (∼5 mm), and thin (∼2.5 mm).
For the thin-walled atria, a thickness of 2.5 mm was selected empirically for the muscle tissue as a compromise to reflect thin walls. For tissues of intermediate thickness, the model was identical to the thin tissue model except that the thickness of tissue was increased to 5 mm (Figure 3). The selection of the appropriate model was left to the discretion of the user, with the general guideline of selection based on known tissue thickness. For purposes of this experiment, for all LV lesions and right ventricular (RV) septal lesions, we used the thick model given that the relative thickness of tissue in this area was more than ∼10 mm, and for the RV free wall and atrial lesions, we used the thin model based on the wall being thinner and closer to 5 mm in thickness. During actual ablation, the measured generator current and maximum temperature readings were used to access this database for the relevant scenario. This selected a specific model, and through comparison of the measured temperature with the model, the lesion parameters (depth, width, and maximum temperature) were extracted and used to define lesion size. The system takes advantage of the relation between the measured temperature and the tissue-catheter interface (or penetration) for a known (measured) ablation current. For example, at a specific generator current and RF duration, the higher the measured temperature, the stronger the coupling between the tissue and catheter. Finally, for comparison, RF parameters were also used to calculate the ablation index, as described previously (6,7).
Continuous variables are expressed as mean ± SD. Differences between measured and estimated dimensions were compared by presenting means and SDs. The absolute error in measurements was presented as histograms and cumulative frequencies (Microsoft Excel 2010). Bland-Altman and scatterplots were created with MATLAB version 7.0 (MathWorks, Natick, Massachusetts). An analysis of variance was used to generate the variance components (i.e., rater mean square, patient and error mean squares) necessary to calculate the appropriate interexaminer reliability coefficient with SAS system software version 9.4 (SAS Institute Inc., Cary, North Carolina).
A total of 143 lesions were created in all chambers in 9 swine. A total of 28 ablations (13 atrial and 15 ventricular lesions) were assessed to be of suboptimal quality for accurate measurement of dimensions (poor-quality photographs, indistinct staining of borders, or eccentric sectioning of the lesion) and were therefore excluded. The remaining 115 ablation lesions were selected for lesion assessment, including 54 atrial lesions and 61 ventricular lesions.
The interexaminer reliability coefficient for the mean rater scores was 0.78 for width and 0.82 for depth. Lesions placed within the RV had a mean width and depth of 7.70 ± 0.24 mm and 3.48 ± 0.17 mm, respectively, and in the LV, the width was 9.36 ± 0.30 mm and depth was 4.59 ± 0.18 mm. In the atria, 27 of 54 lesions were nontransmural, and only these were selected for atrial depth calculation. The mean width of lesions in the atria was 6.24 ± 0.19 mm, and the mean depth was 2.69 ± 0.15 mm. The mean ratio of the width to depth for lesions in the atria, RV, and LV was 2.75 ± 0.90 (1.81 to 4.73), 2.36 ± 0.60 (1.49 to 3.87), and 2.08 ± 0.60 (1.2 to 3.1), respectively. Figure 4 displays the maximum and minimum temperatures recorded from the embedded thermocouples at the catheter tip during atrial and ventricular ablations. There were no instances of steam pop formation.
The mean duration of lesions was 25 ± 11 s (12 to 60 s), with a mean contact force of 14 ± 5 g (5 to 25 g) and power of 37.0 ± 8.6 W (20 to 50 W). One-half of the lesions in the atria (27 of 54) were nontransmural, and depth calculations were restricted to these lesions only. The mean absolute difference between the measured and calculated depth of lesions was 0.9 ± 0.74 mm (95% confidence interval [CI]: 0.61 to 1.19). The mean difference between the measured and calculated width of lesions was 1.2 ± 0.9 mm (95% CI: 0.96 to 1,44). Eighty percent of all lesions (cumulative percentage) had a difference of ≤1.7 mm for depth measurements (root mean square [RMS] = 1.18 mm) and ≤1.74 mm for width measurements (RMS = 1.44 mm) (Figures 5 and 6). Bland-Altman plots are provided to show the spread of these differences (Online Figure 1). This compared favorably with the lesion dimensions predicted by ablation index, where 80% of all lesions had a difference of up to 3 mm for both depth and width (RMS = 2.3 for depth and 2.95 for width).
The mean duration of lesions was 43.5 ± 21 s (15 to 90 s), with a mean contact force of 17.5 ± 5.0 g (7 to 43 g) and power of 38 ± 7 W (20 to 50 W). The mean difference between the measured and calculated depth of lesions was 0.75 ± 0.60 mm (95% CI: 0.6 to 0.9). The mean difference between the measured and calculated width of lesions was 1.66 ± 1.10 mm (95% CI: −1.39 to 1.93). Eighty percent of all lesions (cumulative percentage) had a difference of ≤1.1 mm (RMS = 0.94) for ventricular depth and ≤2.6 mm (RMS = 1.67) for ventricular width (Figures 5 and 6). Bland-Altman plots are provided to show the directional spread of these differences (Online Figure 1). This compared favorably with the dimensions predicted by ablation index, where 80% of all lesions had a difference of up to 2.5 mm for depth and 3.5 mm for width (RMS = 2.01 for depth and 2.59 for width).
Irrigated RF energy, by allowing for cooling of the catheter tip-tissue interface, enables increased delivery of RF power while reducing the risk of char formation from elevated interface temperatures; however, irrigation simultaneously reduces the ability to measure the interface temperature, which renders traditional tip temperatures unfit for lesion estimation. Investigators have therefore focused on several other parameters that reflect catheter-tissue coupling during ablation, such as contact force, impedance drops, and electrogram characteristics, and have used them to assess lesion formation and progression (8,9). These approaches, when used individually, have been criticized as unable to consistently predict lesion dimension, which led to the creation of indices such as force time integral, lesion size index, and more recently, the ablation index (6,10). These formulaic approaches incorporate multiple parameters and aim to provide a more consummate assessment of lesion formation. Such approaches are important because they are geared toward improving lesion transmurality and overlap.
We investigated a novel catheter that by design allows for the reintroduction of catheter-tip temperatures despite the use of enhanced saline tip irrigation. The interface temperature profile is used to provide lesion dimension assessment by using a catheter-specific algorithm that is based on the biophysics of RF ablation and tailored to different tissue environments and thicknesses. Other authors have reported on other indices (using force, power, and time) that predict depth, but these did not have the capability to detect surface temperature. Second, they were typically assessed only in ventricular myocardium in preclinical beating heart models (7). Third, their experimental method targeted predetermined depths predicted by the algorithm as opposed to the approach used in this study, in which pre-determined depths were not targeted.
Unique to our in vivo report is that we examined a wide spectrum of RF power, duration, and forces that simulated real-life clinical settings, and we assessed the performance of this approach in atrial tissue. Although several studies have reported on atrial lesion characteristics or have suggested values for parameters such as the force time integral and the ablation index to guide ablation, this is the first study that evaluated an approach that estimates both depth and width, addressing both transmurality and contiguity (11,12). We specifically placed lesions in areas of thick atrial tissue such as the septum to allow us an opportunity to assess nontransmural atrial lesions (50% of all atrial lesions in the study).
For the ventricle, the depth of lesions ranged from 1.97 to 6.97 mm, and their width ranged from 5.22 to 12.15 mm, with 80% of errors within 1.1 mm for depth and 2.6 mm for width. Ventricular depth and width prediction might be useful clinically; for example, one might decide to terminate RF delivery once a lesion reaches a certain depth (RV outflow tract sites). Indeed, with the increasing role of pre-procedural image integration, one can envision using such an approach to help guide ablation depth and, at the very least, increase operator confidence in the durability of lesion sets. Rozen et al. (13) recently described the ability of a very similar catheter-tip design to predict lesion depth in the ventricular myocardium by ablating for a fixed time interval of 60 s. Although their lesion depth prediction model takes into account the temperatures from the thermocouples, their approach uses RF parameters such as time and impedance and differs from the estimation technique used in this report. They reported a mean difference between predicted and measured ablation lesion depth of 0.72 ± 0.56 mm, which is comparable to the depth estimation difference observed in this study (0.75 ± 0.60 mm).
A greater need for lesion dimension prediction exists clinically for atrial ablation, in which long contiguous and transmural lesion sets are required to achieve durable conduction block (e.g., wide-area circumferential pulmonary vein ablation). The interface temperature-based approach could provide a means to potentially improve ablation contiguity and transmurality while minimizing the risk of collateral damage, char, and pop formation. However, formatting lesion tags based on dimensions on the electroanatomic map is useful only if there is an understanding of local tissue thickness. This would allow the operator to select the appropriate tissue thickness model for lesion prediction (e.g., atrial thin vs. thick tissue) and indicate when the RF application can be expected to be transmural. Such approaches might prevent application of RF beyond what is needed, thereby potentially reducing collateral injury, RF, and even procedure times. Atrial lesions were restricted to clinically relevant contact forces (<25 g). The depth of all lesions created in the atrium varied between 0.72 and 4.07 mm, whereas the width measured between 2.23 and 9.60 mm. The majority (80%) of errors for both dimensions were limited to <1.7 mm.
We chose to describe the difference between estimated and measured lesions as histograms and cumulative frequencies (Figure 5) because this displays the variation in errors across all lesions. In addition, we have provided data on RMS as an estimate of the accuracy of the model’s predictive capability (Figure 6), as well as Bland-Altman plots (Online Figure 1) so that the reader can readily appreciate agreement. Although not investigated in this report, our qualitative observation is that interface temperature appears to be quite sensitive to contact and stability; for example, a lesion that is accompanied by minimal temperature elevation will likely be small (difficult to find on necropsy). The data in this report provide us with some useful insights into ablation dynamics, especially within the atria. Atrial lesion dimension predictions were less accurate compared with the ventricle. This suggests that site-specific and other unaccounted for factors likely play a role in lesion formation. It is possible that certain RF settings, such as high power or lower rates of irrigation, could create lesions with less variation from the prediction model, and further investigation of this is needed. Although not the focus of this study, the mean temperature across all thermocouples for all atrial ablations was 41.4°C, with Figure 4 demonstrating the variation in maximum and minimum temperatures recorded at the catheter tip across both ventricular and atrial ablation lesions. We observed that lesions had qualitatively different appearances when the mean temperatures were on either extreme of this spectrum. The lesions that were characterized by temperatures <37°C often had some degree of endocardial sparing (Figures 7A and 7B) and had less distinct borders than lesions associated with the higher end of the temperature spectrum.
The model for selected RF parameters shown in Figure 3 demonstrates certain important concepts of RF thermal dynamics in tissues that have various thicknesses. The model indicates that in thin tissues such as the atrium (Figure 3A), the thermal contours during ablation increase along the lesion width because they are constrained by heat sinks on either side; for example, at the free wall of the atrium, there is the blood pool endocardially and the lung epicardially. This results in a toroidal-shaped lesion with the hottest regions on either side of the catheter (Figure 7C). This unique lesion morphology can have the appearance of 2 separate lesions fused in the center when sectioned along the diameter, but it is in fact a single lesion. This appearance was seen in 3 atrial and 2 RV lesions. As the tissue thickness progressively increases (Figure 3B [RV free wall] to Figure 3C [LV wall]), epicardial cooling has decreasing influence on the lesion such that it tends to take on a more spherical contour. This is highlighted by the decreasing ratio of the width to depth for lesions in the atria, RV, and LV, the mean ratio being 2.75 ± 0.90 (range 1.81 to 4.73), 2.36 ± 0.60 (range 1.49 to 3.87), and 2.08 ± 0.60 (range 1.2 to 3.1), respectively.
Catheter tip design
In addition to allowing for the preceding lesion estimation algorithm, the catheter design used in this study has other attributes that, although not examined in this study, can be advantageous both for monitoring ablation and from a safety perspective. Specifically, the availability of microelectrode recordings from the surface of the catheter tip (Figure 1B) can allow for real-time and immediate detection of ablation lesion formation at that site by demonstrating rapid reduction in local electrogram amplitude. Also, the availability of local interface temperature assessment by the 6 thermocouples allows for a more global assessment of temperatures at the catheter tip, and this can have value in detecting and therefore avoiding significant elevations in temperatures that can result in charring or steam pop formation.
On the basis of this study, in the atrium, 80% of dimensions were within 1.75 mm (diameter) of the observed lesion. Practically, an operator can use this information by implementing an overlap strategy of 1 mm (radius of lesion tag = half of predicted diameter) between adjacent ablation tags on the electroanatomic map. The operator would also take into consideration the thickness at the site of ablation to ensure a transmural lesion. This approach, however, has to be validated prospectively in further preclinical experiments, as well as clinical studies.
The study was performed in healthy swine endocardially; therefore, the findings of this report might not apply to clinical situations where the presence of fibrosis can alter RF ablation dynamics or to epicardial RF ablation. The lack of use of tissue fixation before lesion measurement in this study is a significant limitation. It is also possible that errors were introduced by artifacts accompanying postmortem measurement of tissue, including varying amounts of shrinkage of tissue (desiccation of tissue from high temperatures) or ablation-related edema errors in thinner tissues. The exclusion of a significant portion of 28 of 143 (16%) of the ablation lesions, the overall limited number of lesions per chamber (especially for assessment of atrial depth), and the lack of a comparator arm using conventional catheter-tip temperature recordings are major limitations and need to be considered when interpreting the results of this study. The approach used to estimate lesion size in this model was based on several assumptions as detailed in the Methods section, all of which have their own respective limitations. Only the hottest or peak thermocouple temperature was used for lesion estimation, and therefore, this approach does not take into account disparities in temperature readings across all 6 thermocouples that can occur with different catheter positions on tissue.
Estimation of RF lesion dimension can be achieved with clinically relevant accuracy using unique temperature signatures acquired from the surface of the catheter along with other RF variables. These data have important implications for assessing the adequacy of lesion overlap and assessment of transmurality. Whether such a strategy can potentially lead to reductions in RF delivery or procedure times without predisposing to gap formation remains to be evaluated.
COMPETENCY IN MEDICAL KNOWLEDGE: Lesion dimension estimation can be enhanced by incorporating temperature signatures acquired from the surface of an irrigated RF catheter tip along with other RF variables. This approach can be used by operators to determine the extent of overlap between adjacent ablation lesions and to target different lesion depths that can be tailored to thickness at that anatomic location. Tailored RF delivery in this fashion could help improve ablation efficacy and prevent excessive RF application.
TRANSLATIONAL OUTLOOK: This approach must be validated prospectively in further preclinical and clinical studies. Whether such a strategy can potentially lead to improved ablation outcomes and reduce unnecessary RF delivery remains to be evaluated clinically.
The authors would like to acknowledge the contribution of Alan D. Weinberg, MS, Department of Biostatistics, Mount Sinai School of Medicine, New York, New York, for his assistance in statistical analysis.
This study was funded by Biosense Webster, Inc. Drs. Koruth and Reddy have served as consultants to and received grant support from Biosense-Webster Inc. Meir Bar-Tal, Dr. Ultchin, Alona Sigal, and Liron Mizrachi are employees of Biosense-Webster. Abraham Berger and Ori Hazan are employees of Berger Thermal Research; and have served as consultants to Biosense Webster. All other 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
- confidence interval
- left atrium
- left ventricle
- root mean square
- right ventricle
- Received March 13, 2017.
- Revision received August 14, 2017.
- Accepted August 17, 2017.
- Koruth J.S.,
- Dukkipati S.,
- Gangireddy S.,
- et al.
- Bruce G.K.,
- Bunch T.J.,
- Milton M.A.,
- Sarabanda A.,
- Johnson S.B.,
- Packer D.L.
- Nakagawa H.,
- Yamanashi W.S.,
- Pitha J.V.,
- et al.
- Theis C.,
- Rostock T.,
- Mollnau H.,
- et al.
- Yokoyama K.,
- Nakagawa H.,
- Shah D.C.,
- et al.
- Neuzil P.,
- Reddy V.Y.,
- Kautzner J.,
- et al.
- Das M.,
- Loveday J.J.,
- Wynn G.J.,
- et al.
- Nakagawa H.,
- Ikeda A.,
- Govari A.,
- et al.
- Granier M.,
- Winum P.F.,
- Granier M.,
- et al.
- Rozen G.,
- Ptaszek L.,
- Zilberman I.,
- et al.