Author + information
- Received December 1, 2014
- Revision received January 26, 2015
- Accepted January 30, 2015
- Published online June 1, 2015.
- J. Tobias Kühl, MD, PhD∗∗ (, )
- Jesper J. Linde, MD∗,
- Lars Køber, MD, DMSc∗,
- Henning Kelbæk, MD, DMSc∗ and
- Klaus F. Kofoed, MD, DMSc∗,†
- ∗Department of Cardiology, The Heart Centre, Rigshospitalet, University of Copenhagen, Denmark
- †Department of Radiology, Diagnostic Centre, Rigshospitalet, University of Copenhagen, Denmark
- ↵∗Reprint requests and correspondence:
Dr. J. Tobias Kühl, Department of Cardiology, 9441 The Heart Centre, Rigshospitalet, University of Copenhagen, Blegdamsvej 94, 2100-Cph Copenhagen, Denmark.
Objectives The objective of this study was to test the hypothesis that the extent and severity of left ventricular myocardial hypoperfusion at rest, in addition to signs of left ventricular myocardial scar, are related to adverse long-term outcome in patients with non–ST-segment elevation myocardial infarction (NSTEMI).
Background Multidetector computed tomography (MDCT) is a noninvasive test with a spatial resolution that allows for the assessment of transmural myocardial perfusion. In patients with suspected NSTEMI, the assessment of myocardial hypoperfusion could be clinically useful.
Methods MDCT was performed at rest before invasive treatment in 396 patients with NSTEMI. The transmural involvement of left ventricular hypoperfusion, the presence of intramyocardial fat or calcification, a summed defect score adding the extent of left ventricular myocardial hypoperfusion (0 to 64 point scale), and the transmural attenuation ratio between the subendocardial and the subepicardial myocardium were assessed. The study endpoint was a combination of death and hospitalization due to heart failure.
Results The median follow-up time of the study was 50 months, and the study endpoint was reached in 56 (15%) of the patients. In a Cox proportional hazards survival model with adjustments for known risk factors, both the summed defect score and transmural attenuation ratio were independently associated with adverse outcome (hazard ratio [HR]: 1.07; 95% confidence interval [CI]: 1.02 to 1.11; p = 0.004 and HR: 0.61; 95% CI: 0.44 to 0.85; p = 0.003, respectively). The presence of intramyocardial fat or calcification was also associated with adverse outcome (HR: 3.5; 95% CI: 1.2 to 10.7; p = 0.03) when compared with patients without any perfusion defect.
Conclusions The extent and severity of left ventricular myocardial hypoperfusion at rest and signs of left ventricular myocardial scar assessed with MDCT before invasive treatment is strongly linked to adverse long-term outcome in patients with NSTEMI.
The correlation between the magnitude of myocardial ischemia and adverse outcome has been demonstrated in patients with ST-segment elevation myocardial infarction (1). In contrast, the relationship between the severity and extent of transmural myocardial hypoperfusion on one hand and clinical outcome after treatment on the other is less well elucidated in patients with non–ST-segment elevation myocardial infarction (NSTEMI).
Multidetector computed tomography (MDCT) coronary angiography is indicated in patients with chest pain and intermediate likelihood of coronary artery disease, but recently it also has been proposed as a valuable diagnostic tool in patients with suspected acute coronary syndrome, thus including patients with NSTEMI (2,3). Interestingly, assessment of regional myocardial perfusion (including areas of hypoperfusion) in patients with myocardial infarction may be performed with use of coronary computed tomography angiography (CTA) (4–8). MDCT has the advantage of submillimeter spatial resolution, which permits the assessment of transmural myocardial perfusion gradients in addition to signs of myocardial scar such as the presence of myocardial calcification or myocardial lipomatous metaplasia (9,10). Myocardial calcification and lipomatous metaplasia are regarded as signs of myocardial scar formation (11).
We tested the hypothesis that the extent and severity of left ventricular myocardial hypoperfusion at rest in addition to signs of myocardial scar as assessed by MDCT are related to adverse long-term outcome in patients with NSTEMI.
Consecutive patients with NSTEMI referred to Rigshospitalet, Copenhagen, Denmark, were scanned with cardiac MDCT prior to invasive investigation as part of a research protocol. Included patients were classified into 3 groups according to regional myocardial perfusion findings at rest by qualitative MDCT analysis: 1) patients with normal myocardial perfusion; 2) patients with myocardial hypoperfusion who did not have a scar; and 3) patients with myocardial hypoperfusion who had a scar. In patients with myocardial hypoperfusion, the regional transmural extent of hypoperfusion was graded as mild, moderate, or severe. Semiquantitative parameters of myocardial perfusion were related to long-term outcome.
All patients underwent medical and invasive treatment according to contemporary clinical practice. The clinical outcome of all patients was recorded for a minimum of 2 years. The primary endpoint was a composite of death from any cause and hospitalization for heart failure. Information on mortality was obtained from validated databases in Denmark (Green System, CSC Scandihealth, Aarhus, Denmark) and the Faroe Islands (COSMIC registry, Cambio Healthcare Systems, Aarhus, Denmark). Information regarding hospitalization due to heart failure was obtained from a systematic review of all hospital admissions after the index myocardial infarction and was defined as admission because of dyspnea, with objective signs of pulmonary congestion and treatment with intravenous diuretics. Verification of hospitalization due to heart failure was performed by an independent reviewer blinded to clinical data.
NSTEMI was defined according to European guidelines as symptoms with acute chest pain and/or electrocardiography changes without persistent ST-segment elevation and a characteristic rise and fall in plasma troponin T (12).
Patients with a contraindication to coronary CTA, including a history of chronic renal disease or serum creatinine >125 μmol/l, cardiac arrhythmia, and known allergy to iodine contrast medium, along with patients with hemodynamic instability and those who declined to participate, were not enrolled in the study. The protocol was approved by the Committees on Biomedical Research for the Capital Region of Denmark (protocol #KF01318727), and written informed consent was obtained from all patients.
Invasive coronary angiography was performed within 2 weeks of diagnosis according to Danish practice during the period of study inclusion (13). The median time between coronary CTA and invasive coronary angiography was 4 h (interquartile range: 2 to 18 h) with no delays due to a coronary CTA. Treatment was determined by an interventional cardiologist who was blinded to coronary CTA findings. The number of diseased vessels was determined on the basis of invasive coronary angiography as >50% diameter stenosis in the 3 major branches and the left main stem. Clinical signs of heart failure within 5 days prior to coronary CTA were recorded according to Killip class.
Scanning was performed with a 64-slice MDCT scanner (Toshiba Aquillion, Tochigi, Japan) using a retrospective scanning technique. The scanner settings were as follows: tube voltage: 120 to 135 kV; detector collimation: 64 × 0.5 mm; and rotation time: 350 to 500 ms. Depending on the expected scan time, 70 to 100 ml of contrast agent (Visipaque 320, GE Healthcare, Buckinghamshire, United Kingdom) was infused with a rate of 5 ml/s, and image acquisition was initiated by automatic bolus triggering. An automatic raw data motion analysis tool (PhaseXact, Toshiba Medical Systems Corp., Otawara, Japan) was used to select the optimal motion-free mid-diastolic phase used for myocardial analysis. Images were reconstructed with 0.5-mm slice thickness. Image data were transferred to an external workstation (Vitrea 6.2, Vital Images, Minnetonka, Minnesota, myocardial perfusion application) for further analysis blinded to clinical information and findings of the coronary angiography.
Myocardial perfusion image analysis was assessed using a window width and level of 300/150 Hounsfield units (HU) and a slice thickness of 3.5 mm (average projection) with slight adjustments for each individual patient. All studies were examined systematically for artifacts as a consequence of motion, misalignment, and beam hardening as previously described (14).
Qualitative evaluation of myocardial perfusion
The left ventricular myocardium was divided into 16 segments based on the standard segment model, with the apex region excluded from the analysis. Each myocardial segment were assigned a visual severity score as follows: 0 = normal myocardial perfusion, defined as myocardium with a homogenous contrast distribution relative to the remainder of the left ventricular myocardium; 1 = mild hypoperfusion, defined as an area with hypoperfusion (= myocardial perfusion defect) involving less than one-third of the transmural extent of the myocardium (Figure 1A); 2 = moderate hypoperfusion greater than one-third but less than one-half transmural (Figure 1B); 3 = severe hypoperfusion involving greater than one-half transmural (Figure 1C) or hypoperfusions with myocardial thinning (Figure 1D, see definition in the following text); and 4 = MDCT sign of myocardial scar (Figures 1E and 1F, see definition in the following text). Each myocardial segment was assigned a full severity score if >50% of the circumferential extent of the segment as presented on the short axis view was involved and half severity score if 25% to 50% of the myocardial segment was involved. If <25% of the segment was involved, the severity score was assigned a value of 0.
Patients were classified according to their worst visual severity score into 5 groups: normal perfusion, mild hypoperfusion, moderate hypoperfusion, severe hypoperfusion, or MDCT signs of myocardial scar.
Myocardial hypoperfusion accompanied by the presence of myocardial thinning was assigned 3 points in this severity score scale (3). Myocardial thinning was defined as 1 or more myocardial segments that were thinned >50% relative to the adjacent myocardium.
“MDCT signs of myocardial scar” was defined by the presence of myocardial calcification or myocardial lipomatous metaplasia. Myocardial lipomatous metaplasia was defined as myocardial perfusion defects in which a 5-mm2 region of interest in the defect was >3 SD below the baseline attenuation of normal myocardium on noncontrast images (25 HU). Using noncontrast images of the first 100 included subjects without a history of previous myocardial infarction, this threshold was defined as the mean attenuation density of a 10-mm2 region of interest placed in the lateral wall, in the apex, and in the septum on a mid-ventricular short axis plane, and the result was in accordance with previous studies assessing lipomatous metaplasia (9,10).
Semiquantitative assessment of myocardial perfusion
On the basis of qualitative severity scores, a semiquantitative summed defect score (SDS) was calculated for each patient as the sum of the severity scores of all segments. Because the presence of extensive perfusion defects may give the same risk of adverse outcome compared with MDCT signs of myocardial scar, we performed a sensitivity analysis in which segments with MDCT signs of myocardial scar were given a visual severity score of 3, 2, and 1 in the SDS calculation.
Semiquantitative evaluation of the transmural myocardial perfusion gradient between the subendocardial myocardium and the subepicardial myocardium was achieved by dividing the myocardium into 3 equally sized layers with the use of semiautomated software and subsequently calculating the transmural attenuation ratio (TAR) as the ratio between the average attenuation of subendocardial and subepicardial voxels as previously described (15,16).
Interobserver variability of SDS and presence of myocardial scar was assessed for all patients by 2 independent observers. Interobserver variability of TAR was assessed in 200 randomly selected patients. Left ventricular end-diastolic volume indexed to body surface area and left ventricular ejection fraction (LVEF) were measured with use of MDCT images.
Continuous variables that were normally distributed were presented as mean ± SD, and continuous data that were not normally distributed were presented as median and interquartile range. Categorical variables were presented as frequencies and percentages.
For statistical comparisons, a 2-tailed Student t test for independent samples was used for continuous values, and the chi-square test was used for categorical variables. Continuous variables that were not normally distributed were compared with the Kruskal-Wallis test.
Interobserver variability of SDS and TAR was assessed using the Bland-Altman method, and the presence of any perfusion defect, lipomatous metaplasia, and calcification was assessed with Kappa statistics.
The relationship between outcome and the presence of myocardial perfusion defects or scar and tertiles of SDS and TAR was plotted according to the Kaplan-Meier method. A univariable Cox proportional hazards survival analysis was performed for potential predictors of clinical events. A multivariable regression analysis with forced entry of SDS and TAR was performed separately with adjustment for established clinical predictors including age, diabetes, previous myocardial infarction, number of diseased vessels, left ventricular end-diastolic volume indexed to body surface area, LVEF, and Killip class. The overall Wald chi-square of a model with the aforementioned established variables was compared with the same models in which either SDS or TAR was added to assess the incremental information of the semiquantitative assessment of the myocardium.
Patients were grouped according to their worst defect score—mild, moderate, severe, or myocardial scar—and tested with use of the Cox proportional hazards survival model (patients with no perfusion defect were used as reference). The effect was tested in a univariable model and a model including the aforementioned known predictors of outcome.
The proportional hazard assumption was checked through the method of cumulative residuals. The presence of myocardial scar, SDS, and TAR was tested for interaction with age. Statistical analysis was performed using SAS version 9.13 (SAS Institute, Cary, North Carolina).
Of 1,409 patients screened for participation between December 1, 2006 and January 31, 2009, 396 patients underwent scanning in the study (Figure 2). A total of 699 patients could not enter the study because of logistical problems, mainly because patients were moved from the referring hospital to Rigshospitalet the evening before the invasive procedure and were scheduled for invasive angiography early the next morning before MDCT could be performed. The image quality in 20 patients did not permit interpretation of myocardial perfusion, leaving a study population of 376. The mean heart rate during MDCT in the excluded patients was 75 ± 16 beats/min compared with 60 ± 13 beats/min in the remaining group (p < 0.001). Patient demographics and treatment are presented in Table 1.
Qualitative and semiquantitative myocardial perfusion
Using qualitative assessment, we found 144 patients (38%) with normal myocardial perfusion. Among patients with no myocardial perfusion defect, 91 (63%) required revascularization according to the invasive angiography. In patients who had at least 1 area with hypoperfusion, the worst segment was categorized as mild in 47 patients (13%), moderate in 98 (26%), and severe in 46 (12%), and 41 (11%) had a myocardial scar. All patients with MDCT signs of myocardial scar also had at least 1 perfusion defect. In patients with myocardial thinning (n = 12), the thinnest part of the myocardium was 3.8 ± 0.7 mm, or 38 ± 7% of the thickness of the surrounding myocardium.
Myocardial calcification was found in 5 patients (1%), and myocardial lipomatous metaplasia was found in 40 (11%). Mean attenuation density in the myocardium with lipomatous metaplasia was –15 ± 31 HU.
The median SDS of all patients was 4 (range 0 to 33). TAR measurements could be performed for all patients except for 6 as a result of pacemaker artifacts (n = 2), parts of the left ventricle outside the scan field (n = 3), or software segmentation failure (n = 1). The mean TAR (N = 370) was 1.1 ± 0.5.
The interobserver variability for detecting the presence of any perfusion defect was κ = 0.76, and the interobserver variability for detecting myocardial scar was κ = 0.93. A bias of 0.4 SDS points existed between the observers, with 95% confidence intervals (CIs) of –4 to 5. No bias existed in TAR measurements, with a 95% CI of 0.05 to –0.05.
Clinical outcome and relation to MDCT findings
The median follow-up time was 50 months (range 27 to 65 months). During follow-up, 56 of the patients (15%) reached 1 of the combined endpoints: 44 (12%) died as their first event, and 12 (3%) were admitted to the hospital because of heart failure. Using a univariable Cox regression analysis, we found that patients reaching the primary endpoint were characterized by having higher age, higher prevalence of previous myocardial infarction and diabetes, higher prevalence of 3-vessel or left main disease, lower LVEF, and higher Killip class (Table 2). Clinical outcome according to the presence of myocardial hypoperfusion or myocardial scar and according to tertiles of the semiquantitative perfusion variables is provided in Figure 3.
In a multivariable Cox regression model with adjustment for known predictors of adverse outcome, both SDS and TAR remained significantly associated with outcome (Table 3). Changing the visual severity score for segments with MDCT signs of myocardial infarcts to 3, 2, or 1 did not change the results significantly. Upon using the same model in the subgroups of patients with no history of previous myocardial infarction and patients without myocardial scar by MDCT, both SDS and TAR remained independently correlated to adverse outcome. Models containing SDS or TAR had significantly higher Wald chi-square values than did models including previously known predictors of outcome alone (Figure 4). A model containing SDS and risk factors had significantly higher Wald chi-square value than did models including risk factors and the presence of myocardial scar (70.3 vs. 64.9; p < 0.01).
The extent of the worst transmural myocardial involvement was found by unadjusted evaluation to correlate with an increasing risk of events (Figure 5, top panel). In a model adjusting for known predictors of adverse outcome, a similar relation was noted, but only the presence of myocardial lipomatous metaplasia or calcification remained significantly associated with the endpoint (Figure 5, bottom panel).
We have demonstrated that the extent and severity of myocardial hypoperfusion in addition to signs of myocardial scar assessed with MDCT prior to invasive coronary evaluation and revascularization is significantly and independently related to long-term outcome in patients with NSTEMI. Whereas previous studies have shown an excellent agreement between MDCT and both single-photon emission computed tomography (5–7) and cardiac magnetic resonance (4,8) for determining the presence and size of myocardial perfusion defects in various patient groups, the current study is the first to show that these MDCT findings translate into adverse clinical outcomes.
Our data also demonstrate a significant effect on outcome after adjustment for LVEF, which suggests that the relation between CT-derived myocardial perfusion and global left ventricular contractile function is more complex than a simple linear correlation.
To test the proposed method, we used a high-risk group of patients with confirmed NSTEMI. Currently the indication for coronary CTA is reserved for patients with chest pain at lower pre-test likelihood. The clinical value of the proposed method is found in studies that point toward an expansion of the use of coronary CTA to include patients suspected of having acute coronary syndrome in the emergency department even before the results of troponin tests are available. Some of these patients will have NSTEMI (2,3). Further studies should explore whether coronary CTA-assessed myocardial findings translate into an adverse outcome in patients with lower probability chest pain syndromes. The present study does not propose and/or recommend coronary CTA in an NSTEMI population but explores one of the potentially diagnostic benefits that could be achieved by using such a strategy.
We found that the absence of perfusion abnormalities and MDCT signs of myocardial scar correlated with a low risk of an adverse outcome. Conversely, compared with patients with normal myocardial perfusion, we found that patients with moderate and severe hypoperfusion had an unadjusted increase in hazard ratio of 2.4 and 5.2, respectively. However, only the sum of the total hypoperfusion defect load assessed in SDS was associated with outcome independently of previously known risk factors. This result was also found when excluding patients with previous myocardial infarction and myocardial fat and calcification; this finding suggests that the extent of ischemia in patients with NSTEMI has important prognostic value.
When patients with no perfusion defects are used as reference group, the presence of myocardial scar remained independently significant when adjusting for known risk factors. In 40 of 41 patients with a myocardial scar, the presence of lipomatous metaplasia was found. The histopathologic evolution of a myocardial infarction has been described thoroughly, including myocyte death and subsequent pressure load increase, triggering a cascade of reparative changes that include myocardial thinning, left ventricular dilation, hypertrophy, and the formation of collagen scar (17). Interestingly, the presence of adipose tissue was not described in the classical literature on left ventricular remodeling after myocardial infarction until 1997, when Baroldi et al. (11) published the finding of lipomatous metaplasia in 26 of 38 explanted hearts from patients with chronic ischemic heart disease.
Lipomatous metaplasia appears in myocardial scars more than 10 months after infarction (10). Its presence could be speculated to be linked to the pathogenesis of post-myocardial infarction arrhythmias. However, the link between the presence of MDCT sign of scar and adverse outcome could also well be explained by a larger extent of dysfunctional left ventricular myocardium in patients with a previous scar rather than properties (e.g., arrhythmogenesis) of the scar. Our data suggest that in the calculation of SDS, segments with MDCT sign of myocardial scar could count the same as having a perfusion defect in terms of calculating risk. The clinical relevance of adipose tissue should be the focus of further studies. The fact that myocardial calcification was found in only 5 patients indicates its limited relevance for assessment in patients with NSTEMI.
Under resting conditions, myocardial perfusion is higher in the subendocardium than in the subepicardium because the coronary tree adapts to compensate for the flow-limiting effect of myocardial compressive forces that are greatest at the endocardial surface (18). The MDCT spatial resolution allows for an evaluation of the subendocardial/subepicardial ratio in humans, and mean values between 1.12 and 1.16 have been found in nonischemic regions during resting conditions (15,16). The subendocardium is the most vulnerable part of the myocardium (18,19), and in the presence of myocardial ischemia, the subendocardial/subepicardial ratio drops below 1 (16). This study is the first to evaluate the subendocardial/subepicardial ratio in patients with myocardial infarction, and we found that a reduced ratio (0.92 to 1.08 in the lowest tertile) overall in the left ventricle is correlated to adverse outcome. Measurements of TAR, however, require additional semiautomated software, which may limit the relevance of this technique in an emergency situation.
Of the patients with no perfusion defect, 63% required revascularization, and thus CT myocardial perfusion at rest was not able to rule out significant stenosis. However, it is not surprising that perfusion at rest cannot predict the need for intervention, because resting flow requires a stenosis of 90% to 100% before flow limitation is hemodynamically detectable (20). The hemodynamic significance and the threshold for intervention of a coronary stenosis are best evaluated during rest-stress testing (21). This notion has driven the recent development of stress CT myocardial perfusion imaging (22). Further studies should focus on the prognostic value of CT myocardial perfusion imaging, preferably also in lower risk patient groups for whom coronary CTA is clinically indicated.
Contrast-induced nephropathy was found in 5% of patients, and all increased levels of serum creatinine returned spontaneously to normal without any clinical manifestation. There were no reports of serious adverse reactions to contrast media (unpublished observations). Patients received a total of 19 ± 4 mSv, which is comparable to the dose range reported from contemporary MDCT studies at the time of inclusion. Nevertheless, the image analysis in this study was constrained to late diastolic images that are available in prospectively gated scans, and thus the method described here can be performed with much lower levels of radiation.
No additional beta-blocker was administered prior to MDCT examination to lower heart rate, and in 20 patients the images were nonevaluable, predominantly because of motion artifacts. Further studies should focus on the importance of lowering the heart rate.
Although only mid-diastolic images were used for analysis, it is a further limitation of the study that it was not performed with contemporary scanning techniques with prospective gating and lower tube voltage (100 kV), which may change the appearance of contrast defects.
Myocardial thinning is described as part of early remodeling of the left ventricle after a myocardial infarction (17) or may occur as part of a chronic myocardial infarction. The underlying mechanism for myocardial thinning is difficult to distinguish using MDCT. The presence of calcified or lipomatous contents in a scar suggests a chronic component. In contrast, the presence of a mere perfusion defect in a patient with NSTEMI will most often be a sign of acute infarction but may in some cases may be caused by a chronic (fibrotic) scar. In these cases a clear distinction between chronic and acute scar is not possible with MDCT.
Patients at the highest risk were not enrolled, and thus the study describes a low-risk group without cardiac arrhythmias or renal dysfunction suitable for MDCT scanning. Accordingly, the present results may not be generalized to the average NSTEMI population (23).
Although we find a strong statistical relation to the study endpoint, the overall number of endpoints is low, and the results should be interpreted with appropriate caution.
The extent and severity of left ventricular myocardial hypoperfusion and signs of myocardial scar by MDCT assessed before invasive treatment correlates to long-term outcome in our group of patients with NSTEMI. An optimized risk assessment could possibly be achieved by adding MDCT myocardial perfusion findings to conventional risk assessment in patients with acute coronary syndrome. Further studies are required to confirm our results, preferably also in patient groups at lower risk and using contemporary scanning techniques and a multicenter setup.
COMPETENCY IN MEDICAL KNOWLEDGE: The extent and severity of left ventricular myocardial hypoperfusion and sign of myocardial scar can be assessed with MDCT and has prognostic implications incremental to that of traditional risk factors in patients with NSTEMI.
TRANSLATIONAL OUTLOOK: Although reporting signs of hypoperfusion and myocardial scar is possible on the basis of MDCT images, an assessment using MDCT in more patient target groups—including patients with chest pain who have intermediate risk of coronary artery disease—will lead to a better understanding of the clinical value of the method.
This research was supported by the Danish Agency for Science, Technology and Innovation, The Danish Council for Strategic Research (Eastern Denmark Initiative To imprOve Revascularization Strategies, grant 09-066994), The Research Fund of Rigshospitalet, and the A.P. Møller and Chastine McKinney Møller Foundation, Copenhagen, Denmark. Drs. Linde and Kofoed have received lecturing fees from Toshiba Medical Systems. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- computed tomography angiography
- left ventricular ejection fraction
- multidetector computed tomography
- non–ST-segment myocardial infarction
- summed defect score
- transmural attenuation ratio
- Received December 1, 2014.
- Revision received January 26, 2015.
- Accepted January 30, 2015.
- American College of Cardiology Foundation
- Miller T.D.,
- Christian T.F.,
- Hopfenspirger M.R.,
- Hodge D.O.,
- Gersh B.J.,
- Gibbons R.J.
- Feuchtner G.M.,
- Plank F.,
- Pena C.,
- et al.
- Ichikawa Y.,
- Kitagawa K.,
- Chino S.,
- et al.
- Bassand J.P.,
- Hamm C.W.,
- Ardissino D.,
- et al.
- Wallentin L.,
- Lagerqvist B.,
- Husted S.,
- Kontny F.,
- Stahle E.,
- Swahn E.
- George R.T.,
- Arbab-Zadeh A.,
- Miller J.M.,
- et al.
- Sutton M.G.,
- Sharpe N.
- Duncker D.J.,
- Bache R.J.
- Rochitte C.E.,
- George R.T.,
- Chen M.Y.,
- et al.
- Kuhl J.T.,
- Moller J.E.,
- Kristensen T.S.,
- Kelbaek H.,
- Kofoed K.F.