Author + information
- Received May 12, 2014
- Revision received June 21, 2014
- Accepted July 3, 2014
- Published online February 1, 2015.
- Maksymilian P. Opolski, MD∗,†∗ (, )
- Stephan Achenbach, MD‡,
- Annika Schuhbäck, MD‡,
- Andreas Rolf, MD∗,
- Helge Möllmann, MD∗,
- Holger Nef, MD§,
- Johannes Rixe, MD§,
- Matthias Renker, MD§,
- Adam Witkowski, MD†,
- Cezary Kepka, MD‖,
- Claudia Walther, MD∗,
- Christian Schlundt, MD‡,
- Artur Debski, MD†,
- Michal Jakubczyk, MSc¶ and
- Christian W. Hamm, MD∗,§
- ∗Department of Cardiology, Kerckhoff Heart Center, Bad Nauheim, Germany
- †Department of Interventional Cardiology and Angiology, Institute of Cardiology, Warsaw, Poland
- ‡Department of Internal Medicine 2 (Cardiology), University of Erlangen, Erlangen, Germany
- §Department of Cardiology and Angiology, Justus-Liebig University of Giessen, Giessen, Germany
- ‖Department of Coronary and Structural Heart Diseases, Institute of Cardiology, Warsaw, Poland
- ¶Institute of Econometrics, Warsaw School of Economics, Warsaw, Poland
- ↵∗Reprint requests and correspondence:
Dr. Maksymilian P. Opolski, Department of Cardiology, Kerckhoff Heart Center, Benekestrasse 2, 61231 Bad Nauheim, Germany.
Objectives This study sought to establish a coronary computed tomography angiography prediction rule for grading chronic total occlusion (CTO) difficulty for percutaneous coronary intervention (PCI).
Background The uncertainty of procedural outcome remains the strongest barrier to PCI in CTO.
Methods Data from 4 centers involving 240 consecutive CTO lesions with pre-procedural coronary computed tomography angiography were analyzed. Successful guidewire (GW) crossing ≤30 min was set as an endpoint to eliminate operator bias. The CT-RECTOR (Computed Tomography Registry of Chronic Total Occlusion Revascularization) score was developed by assigning 1 point for each independent predictor, and then summing all points accrued. Continuous distribution of scores was used to stratify CTO into 4 difficulty groups: easy (score 0); intermediate (score 1); difficult (score 2); and very difficult (score ≥3). Discriminatory performance was tested by 10-fold cross-validation and compared with the angiographic J-CTO (Multicenter CTO Registry of Japan) score.
Results Study endpoint was achieved in 55% of cases. Multivariable analysis yielded multiple occlusions, blunt stump, severe calcification, bending, duration of CTO ≥12 months, and previously failed PCI as independent predictors for GW crossing. The probability of successful GW crossing ≤30 min for each group (from easy to very difficult) was 95%, 88%, 57%, and 22%, respectively. Areas under receiver-operator characteristic curves for the CT-RECTOR and J-CTO scores were 0.83 and 0.71, respectively (p < 0.001). Both the original model fit and 10-fold cross-validation correctly classified 77.3% of lesions.
Conclusions The CT-RECTOR score represents a simple and accurate noninvasive tool for predicting time-efficient GW crossing that may aid in grading CTO difficulty before PCI. (Computed Tomography Angiography Prediction Score for Percutaneous Revascularization for Chronic Total Occlusions [CT-RECTOR]; NCT02022878)
- chronic total occlusion
- clinical prediction rule
- coronary computed tomography angiography
- percutaneous coronary intervention
Marked advances in endovascular techniques and device technology have resulted in substantial improvements of procedural success rates of percutaneous coronary intervention (PCI) in chronic total occlusion (CTO) (1,2). Despite this, the majority of patients with CTO are still being managed medically or referred for bypass surgery (3). The most common reasons for deferring PCI in CTO appear to be the uncertainty of predicting the procedural outcome and the difficulty of gauging the time and use of resources during PCI (4).
The availability of a simple and reliable prediction model to evaluate the probability of successful guidewire (GW) crossing through CTO would aid in identifying patients in whom an interventional approach to revascularization is likely to succeed. The recent introduction and validation of the angiographic J-CTO (Multicenter Chronic Total Occlusion Registry of Japan) score was intended to assess CTO complexity and therefore to determine the time required for PCI (4). In the search for innovative imaging modalities with improved visualization of CTO when compared with conventional coronary angiography (CCA), coronary computed tomography angiography (CCTA) has emerged as an accurate noninvasive tool to assess the trajectory and morphology of occluded coronaries (5). Therefore our goals were as follows: 1) to determine the CCTA characteristics of CTO that influence the procedural outcome of PCI in a contemporary clinical setting; 2) to develop a CCTA-based prediction model for grading CTO suitability for PCI; and 3) to investigate the diagnostic accuracy of this model for prediction of PCI duration compared with the J-CTO score.
Study design and population
The CT-RECTOR (Computed Tomography Registry of Chronic Total Occlusion Revascularization) is a multicenter, retrospective, observational study of consecutive patients undergoing CCTA before attempted PCI of CTO between March 2007 and November 2013 in 4 European medical centers (Kerckhoff Heart Center, Bad Nauheim, Germany; University Hospital Giessen, Giessen, Germany; University Hospital Erlangen, Erlangen, Germany; and Institute of Cardiology, Warsaw, Poland). All included patients had typical effort angina or positive functional stress tests related to CTO. Institutional Review Board approval was obtained at each center, and the study complied with the declaration of Helsinki. The study protocol was registered at ClinicalTrials.gov (NCT02022878).
CTO was defined as a TIMI (Thrombolysis In Myocardial Infarction) flow grade 0 within the occluded segment with no antegrade filling of the distal vessel other than via collaterals (6). All patients had a native vessel or a bypass graft occlusion of ≥2.5 mm estimated to be of ≥3 months’ duration based on either previous angiography or a history of sudden chest pain or myocardial infarction consistent with the location of CTO. The type of CTO was either de novo or due to in-stent restenosis. All PCI were performed by highly experienced operators performing a minimum of 50 CTO cases per year as advocated by the EuroCTO Club (7). Dedicated wires were used in a step-up approach starting from soft polymeric wires at the beginning with a subsequent switch to stiff flat or tapered wires. The strategies used included the parallel wire technique, subintimal tracking and re-entry techniques, as well as retrograde wire approach. The interventional strategy was left to the discretion of the operator.
As in the J-CTO registry (4), the primary endpoint was defined as successful GW crossing through the CTO within 30 min of procedure time. This was viewed as the most objective parameter reflecting the level of difficulty intrinsic to the CTO and minimizing the operator-related bias associated with the final outcome of PCI. Furthermore, according to Morino et al. (1) the GW manipulation time is inversely related to GW crossing success rate corresponding to the final restoration of flow through CTO. The secondary endpoints were defined as: 1) successful GW crossing through CTO at any time; and 2) successful GW crossing through CTO with restoration of flow (<50% residual stenosis and TIMI flow grade 2 to 3).
CCTA protocol and image reconstruction
All CCTA scans were performed during the 4 weeks prior to PCI. During the study period, 2 dual-source scanners were used (Somatom Definition CT and Somatom Definition Flash CT, Siemens, Erlangen, Germany). Unless contraindicated, an intravenous or oral dose of metoprolol was given to target a heart rate of <65 beats/min, and sublingual nitroglycerin was administered before CCTA. The contrast transit time was estimated by injection of a test bolus or using a real-time bolus tracking technique. For acquisition of the volume dataset, 80 to 120 ml iodinated contrast material containing between 350 and 400 mg/ml iodine was injected followed by a mixture of 20% contrast agent and 80% saline. The scan parameters were as follows and varied according to the scanner type: beam collimation 64 × 0.6 mm; tube voltage 100 or 120 mV; gantry rotation time 330 or 280 ms; tube current 330 to 438 mAs/rotation or 320 mAs/rotation. Dose reduction strategies including electrocardiogram-gated tube current modulation and prospective axial triggering were used whenever feasible. The estimated radiation dose for CCTA was determined based on the recommended conversion factors (8) and ranged from 1.1 to 28 mSv. Scan data were reconstructed in mid-to-end systole and diastole (30% to 45% and 55% to 75% of the R-R interval). The slice thickness was 0.6 or 0.75 mm with an increment of 0.4 mm.
All CT data were analyzed by a highly experienced reader (M.P.O.) on a dedicated workstation (Leonardo workstation, Siemens). The assessment was performed blinded to all clinical and angiographic data after confirmation of the CTO vessel in which PCI was attempted. The CCTA datasets were evaluated using multiplanar reconstructed images and thin-slab maximum intensity projections. Additionally, 3-dimensional reconstruction of the central vessel line running through the occlusion path and entry/exit sites was created using a dedicated software tool (Circulation, Siemens) to calculate the least foreshortened length of the CTO and evaluate orthogonal cross sections of the vessel area (9). Initially, a detailed qualitative analysis was made to assess the potential variables associated with GW crossing. All CTO characteristics were separately surveyed at the entry site, occlusion route, exit site, and the overall segment (entry, occlusion route, and exit combined). The presence of multiple occlusion sites, defined as ≥2 complete interruptions of the contrast opacification separated by contrast-enhanced segment of ≥5 mm, was determined (10). Stump morphology was classified as tapered or blunt. Additionally, a tapered stump was classified as either central or eccentric based on its location along the central vessel line. Side branch presence at the entry or exit site was defined as the visualization of any side branch within 3 mm proximal to the entry or exit of the occlusion. The presence of any calcium was noted. Severe calcium was defined as the presence of high-density plaque involving ≥50% of the vessel cross-sectional area on visual assessment (9,11). Bending was defined as the presence of at least 1 bend of >45° throughout the occlusion route (11,12). Importantly, any vessel tortuosity outside the occlusion segment was excluded from the assessment of bending (4). The presence of vessel intramyocardial course within the occlusion route was assessed. Quantitative analysis included the measurement of occlusion length, and proximal and distal vessel diameters. Occlusion length was classified as either <20 or ≥20 mm according to the consensus of the EuroCTO Club (7). Interobserver variability was assessed by a second experienced reader (M.R.) in 25% of all CTO lesions.
Coronary angiographic analyses were performed by a single experienced reader (M.P.O.) blinded to the results from CCTA and clinical characteristics, using a commercial software (CASS QCA, version 5.10, Pie Medical Imaging, Maastricht, the Netherlands). The angiographic variables of the J-CTO score, including blunt entry site, calcification within CTO, bending >45°, occlusion length ≥20 mm, and previously failed lesion, were determined as previously described (4). The J-CTO score for an individual CTO was then obtained by assigning 1 point for each variable present and then summing all points accrued, with a minimal and maximal total score of 0 and 5, respectively. The presence of heavy calcification was assessed as previously described (13). Other angiographic indexes included ostial CTO, multiple occlusions, the presence of side branches within 3 mm proximal to the entry site, the presence of bridging collaterals, the degree of retrograde collaterals according to the Rentrop classification (14), and poor distal vessel opacification (7).
Data are presented as mean ± SD for continuous variables and frequencies for categorical variables. Continuous variables were compared by using Student t test or the Mann-Whitney U test. Differences in categorical data were analyzed by the Fisher exact test. The prediction model for successful GW crossing ≤30 min was developed by using a multivariable logistic regression analysis. As a first step, only candidate categorical variables that reached statistical significance in the univariable model were entered into the multivariable analysis using the standard method of entry. The results of the multivariable model were then used to assign an integer score to each independent variable based on the model’s beta coefficients. Furthermore, all individual integer score values were summed to calculate a total score for each CTO (CT-RECTOR score). The discriminatory performance of the model was validated using 10-fold cross-validation in which the original set was randomly partitioned in 10 equal size subsamples. Of the 10 subsamples, a single subsample is retained as the validation set for testing the model, and the remaining 9 subsamples are used as the training set. The cross-validation process is repeated 10 times with each of the subsamples used exactly once as the validation set (15). To exclude the potentially confounding influence of retrograde wiring on the developed model, the multivariable regression analysis was repeated in the subgroup of CTO treated by antegrade approach only. Agreement between the CT-RECTOR and J-CTO scores was assessed using the Spearman correlation coefficient. The diagnostic accuracy of the CT-RECTOR score was compared with that of the J-CTO score using paired comparisons between areas under receiver-operating characteristic (ROC) curves given by C statistics according to the method of DeLong et al. (16). The interobserver agreement was analyzed using the kappa statistics (a method of agreement for categorical variables) for each independent predictor of the multivariable model. Statistical significance was defined as a p value of <0.05. Statistical analyses were performed with SPSS (version 20.0, SPSS Inc., Chicago, Illinois) and MedCalc (version 12.3, MedCalc Software, Mariakerke, Belgium).
A total of 229 patients (mean age 63 ± 10 years, 79% male) with 240 CTO lesions were enrolled. Successful GW crossing within 30 min was achieved in 55% of lesions. Clinical characteristics between successful and failed GW crossing were similar except for more frequent previous coronary artery bypass grafting and slightly lower left ventricular ejection fraction in the GW-failure group than the GW-success group (Table 1).
Most CTO were attempted for the first time (85%) and by antegrade approach (89%), and the overall procedural success defined as recanalization of CTO with restoration of flow (<50% residual stenosis and TIMI flow grade 2 to 3) was 62%. Table 2 summarizes angiographic and procedural CTO characteristics in relation to successful and failed GW crossing ≤30 min. Regarding procedural characteristics, the GW-failure group showed higher prevalence of reattempted PCI, retrograde injection, and retrograde wiring, as well as a higher number of overall introduced wires than the GW-success group did.
CCTA characteristics and selection of variables
The GW-failure group showed higher prevalence of lesions ≥20 mm and multiple occlusions than the GW-success group (Table 3). Regional qualitative CCTA characteristics surveyed at different sites of the CTO such as blunt stump at the entry site, severe calcification at the entry site and occlusion route, and bending at the entry and exit sites and occlusion route were more prevalent in the GW-failure group (Figures 1 and 2⇓⇓). For the purpose of simplicity, all significant regional variables were additionally assessed throughout the overall CTO (entry, occlusion route, and exit combined) demonstrating relevant relationships for the blunt stump, bending, and the presence of calcification. Because severe calcification was a stronger correlate of GW-crossing ≤30 min compared with the presence of any calcium, only the former was selected for further statistical assessment. Thus, the final qualitative CCTA variables entered into the model included occlusion length ≥20 mm, multiple occlusions, blunt stump, bending, and severe calcification assessed throughout the overall CTO segment.
Univariate and multivariate regression analyses
The univariate clinical, angiographic, and CCTA predictors for successful GW crossing ≤30 min included the following: previous coronary artery bypass grafting; lesion in the right coronary artery; lesion in the left anterior descending artery; poor distal CTO visualization by CCA; reattempted PCI at CTO; occlusion duration ≥12 months or unknown; occlusion length ≥20 mm by CCTA; multiple occlusions by CCTA; blunt stump by CCTA; bending by CCTA; and severe calcification by CCTA. In the multivariable regression analysis, 2 clinical and 4 CCTA variables (reattempted PCI at CTO, occlusion duration ≥12 months or unknown, multiple occlusions, blunt stump, bending, and severe calcification) were independent predictors for GW crossing ≤30 min (Table 4). The subanalysis in CTO treated by antegrade approach only yielded similar independent predictors for GW crossing ≤30 min. Of the independent CCTA variables, both multiple occlusions (22% vs. 7%, p < 0.001) and severe calcification (34% vs. 21%, p < 0.001) were more often detected by CCTA, with similar frequencies of the blunt stump (47% vs. 46%, p = 0.927) and bending (23% vs. 25%, p = 0.749), than were detected with CCA. Interobserver agreement for independent CCTA variables was 0.81 for multiple occlusions, 0.80 for bending, 0.71 for severe calcification, and 0.57 for blunt stump.
Development of the CT-RECTOR score
Based on the regression model’s beta coefficients ranging from 1.04 to 1.28, each independent variable was assigned an integer score of 1 point. For each CTO lesion, a total difficulty score for PCI (CT-RECTOR score) was calculated by summing all points accrued (Figure 3). Figure 4 shows the relationship between the CT-RECTOR score and successful GW crossing ≤30 min. As shown in Figure 5A, the resulting continuous distribution of total difficulty scores across all lesions was then categorized into the following 4 groups of scores according to the varying probabilities of successful GW crossing ≤30 min: 1) easy (score of 0); 2) intermediate (score of 1); 3) difficult (score of 2); and 4) very difficult (score ≥3). Figure 5B depicts the declining probabilities of total GW success proportionally to the difficulty level of CTO.
The original model fit correctly classified 76.3% of lesions, whereas the 10-fold cross-validation showed similar discriminatory performance without misclassification of any case. The CT-RECTOR score correlated positively with the J-CTO score (r = 0.561, p < 0.001). The area under the ROC curve for the CT-RECTOR score to predict successful GW crossing ≤30 min was significantly higher compared with the area under the ROC curve for the J-CTO score (0.83 vs. 0.71, p < 0.001) (Figure 6). Table 5 shows the diagnostic performance for classification of CTO difficulty level by CT-RECTOR and J-CTO scores for prediction of successful GW crossing ≤30 min.
To our knowledge, the present study is the first large-scale, multicenter evaluation of a CCTA-based diagnostic rule for predicting the time-efficient PCI in CTO. Specifically, we have developed the CT-RECTOR score as a simple and easy-to-use noninvasive prediction tool for grading CTO suitability for successful GW crossing within 30 min. Of note, the proposed CCTA prediction rule has the potential to exceed the angiographic J-CTO score in terms of the diagnostic performance. We believe that clinicians may find the CT-RECTOR score particularly useful to better estimate the time required for PCI, specifically in patients with poor CTO visualization by CCA.
Despite marked advances in endovascular techniques and CTO-dedicated devices, PCI of CTO remains a significant technical challenge with unpredictable procedural success and often long duration (17). During the last decade, CCTA has emerged as a valuable noninvasive tool to provide guidance during PCI in CTO with improved visualization of calcification and vessel trajectory as compared with CCA (5). Previous reports on CCTA in CTO identified the presence of severe calcification, lesion length, and bending as independent predictors of GW passage (9,11,12). However, such previous regression models were limited by including a relatively small number of lesions; additionally, the contribution of each independent CCTA variable should be re-evaluated in light of the improvements in PCI of CTO.
We performed a comprehensive qualitative analysis to determine the most robust CCTA predictors for successful GW crossing through CTO. To simplify the application of the final regression model, the scrutinized parameters were categorical and represented “presence or absence” of CTO characteristics. Moreover, after demonstrating no relevant differences between the regional variables, only significant CCTA predictors surveyed at the overall CTO segment were selected for further assessment. Ultimately, the multivariable regression analysis yielded the presence of 4 CCTA-based (multiple occlusions, blunt stump, bending, and severe calcium) and 2 clinical (reattempted PCI at CTO, occlusion duration ≥12 months or unknown) predictors for successful GW crossing ≤30 min. Interestingly, our model complemented previous CCTA observations by introducing 2 novel predictors (multiple occlusions and blunt stump) for PCI in CTO. Although a blunt stump had already been reported as a significant correlate for PCI outcome in a previous angiographic study (11), the presence of multiple occlusions has to our knowledge never been studied in this context. Of note, our results indicate that the presence of tandem occlusion segments with double entry and exit sites constitutes a significant technical hurdle to successful GW passage. Although the lesion length has been frequently reported as a relevant correlate for successful PCI in CTO (1,11), it was not an independent predictor for GW passage in our study, possibly indicating a stronger influence of other CCTA variables from the regression model. In accordance with previous reports (4,18), we found previously failed PCI at CTO and long or unknown duration of occlusion to be the only independent clinical predictors for GW crossing.
Clinical application of the CT-RECTOR score
The equal contribution of all independent predictors to the final model fit resulted in a simple and easy-to-remember total difficulty score for CTO (CT-RECTOR score). We advocate clinical application of the CT-RECTOR score in patients with pre-procedural availability of CCTA, primarily for the assessment of CTO complexity and prediction of GW manipulation time. Our CCTA prediction rule was found to be highly sensitive for predicting GW success ≤30 min in lesions with a CT-RECTOR score ≤2. Thus, the proposed CCTA score may be particularly applied to estimate the need for additional CTO devices or dedicated revascularization techniques in lesions with respectively higher difficulty levels. Furthermore, the excellent specificity of our model in lesions with a low CT-RECTOR score (≤1) ensures high GW success in the hands of experienced operators and could therefore be used for training purposes among less-skilled interventionalists. Finally, the CT-RECTOR score is related to the overall GW manipulation time and might be indirectly used for estimating the final PCI success.
Comparison of the CT-RECTOR and J-CTO scores
Recently, the J-CTO score has been introduced as an accurate angiographic rule for prediction of successful GW crossing ≤30 min (4). In contrast to the CT-RECTOR score, the J-CTO scoring system consists of 5 angiographic variables including the lesion length and excepting the multiple occlusions and CTO duration. Further dissimilarities include the amount of surveyed calcification and the location of the blunt stump. Importantly, we have demonstrated a significantly higher diagnostic accuracy of the CT-RECTOR score as compared with the J-CTO score for prediction of successful GW crossing. Lower accuracy of the J-CTO score in our study is likely due to weaker discriminatory performance of angiographic variables such as calcification and blunt stump that failed to show significant differences in GW group comparison. Additionally, CCA underestimated the presence of multiple occlusions and severe calcification when compared with CCTA. This is in line with previous observations demonstrating that severe calcification visualized by CCTA but not by CCA is independently related to successful PCI in CTO (11,19). Furthermore, the enhanced diagnostic performance of CCTA over CCA might be a result of the 3-dimensional depiction of the CTO bending and accurate detection of multiple occlusions that are often not readily seen on invasive angiograms (20,21). Indeed, this has been recently suggested by Rolf et al. (21) who demonstrated a significantly higher CTO revascularization rate in the CCTA-guided PCI after translation of 3-dimensional CCTA images to the catheterization laboratory. Finally, the lower number of variables in the J-CTO score compared with the CT-RECTOR score (5 vs. 6 variables) might particularly predispose to improved discriminatory performance of the latter in lesions with low angiographic scores.
First, it was a retrospective and observational experience with a potential for case selection bias resulting from different reasons for referral to CCTA. Nevertheless, we included all consecutive CTO patients with pre-procedural CCTA data, and 2 of our 4 centers established a comprehensive CTO program requiring CCTA prior to PCI in all patients without contraindications for additional coronary imaging. Second, although we included the largest CTO population with pre-procedural CCTA reported thus far, the number of CTO was still inadequate for application of the holdout validation with splitting the original data into independent training and testing sets. Thus, further analyses are needed to evaluate the CT-RECTOR applicability in other patient populations. Third, the operator, site, and resources selection biases are inevitable and might influence the predictive performance of our model. Finally, and similar to the J-CTO registry, we decided to include CTO with a retrograde wiring approach so as not to discard more complex lesions. However, as CTO with a retrograde approach represented the minority (11%) of our lesions, the applicability of the CT-RECTOR score among those CTO should be re-evaluated in future studies.
We introduce the CT-RECTOR score as a simple and accurate computed tomographic prediction rule for grading CTO difficulty by predicting successful GW crossing within 30 min. The proposed CCTA-based scoring system does not suffer from the limitations of CCA and thus has the potential to exceed the discriminatory performance of the angiographic J-CTO score. Clinicians may find the CT-RECTOR score particularly useful to better estimate the time and resources required for the interventional treatment of CTO.
The authors thank Elizabeth Martinson, PhD, for her help in editing this manuscript.
This study was supported by a grant from the German Cardiac Society, Düsseldorf, Germany. Dr. Opolski was supported by a research grant from the German Cardiac Society; and received a scholarship from the Foundation for Polish Science. Dr. Achenbach has received grant support from Siemens Healthcare and Abbott; and has served on the Speakers Bureau of Siemens Healthcare. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- conventional coronary angiography
- coronary computed tomography angiography
- chronic total occlusion
- percutaneous coronary intervention
- receiver-operating characteristic
- Thrombolysis In Myocardial Infarction
- Received May 12, 2014.
- Revision received June 21, 2014.
- Accepted July 3, 2014.
- American College of Cardiology Foundation
- Morino Y.,
- Kimura T.,
- Hayashi Y.,
- et al.,
- for the J-CTO Registry Investigators
- Fefer P.,
- Knudtson M.L.,
- Cheema A.N.,
- et al.
- Morino Y.,
- Abe M.,
- Morimoto T.,
- et al.
- ↵Bongartz G, Golding SJ, Jurik AG, et al. European guidelines on quality criteria for computed tomography, EUR 16262EN, 2000. Available at: http://www.drs.dk/guidelines/ct/quality/htmlindex.htm. Accessed February 20, 2013.
- Opolski M.P.,
- Kepka C.,
- Achenbach S.,
- et al.
- Rentrop K.P.,
- Cohen M.,
- Blanke H.,
- Phillips R.A.
- Grantham J.A.,
- Marso S.P.,
- Spertus J.,
- House J.,
- Holmes D.R.,
- Rutherford B.D.