Author + information
- Received February 25, 2002
- Revision received July 30, 2002
- Accepted September 20, 2002
- Published online February 5, 2003.
- Daniel B Fram, MD, FACC*,†,* (, )
- Rabih R Azar, MD*,
- Alan W Ahlberg, MA*,
- Linda D Gillam, MD, FACC*,
- Joseph F Mitchel, DO, FACC*,†,
- Francis J Kiernan, MD, FACC*,†,
- Jeffrey A Hirst, MD, FACC*,†,
- Jeffrey F Mather, MS*,
- Edward Ficaro, PhD*,
- Gizelle Cyr, RN*,
- David Waters, MD, FACC*,† and
- Gary V Heller, MD, PhD, FACC*
- ↵*Reprint requests and correspondence:
Dr. Daniel B. Fram, Cardiac Lab, Hartford Hospital, 80 Seymour Street, P.O. Box 5037, Hartford, Connecticut 06102-5037 USA.
Objectives This study was designed to determine how long nuclear myocardial perfusion imaging (MPI) remains abnormal following transient myocardial ischemia.
Background Acute rest MPI identifies myocardial ischemia with a high sensitivity when the radionuclide is injected during chest pain. However, the sensitivity of this technique is uncertain when the radionuclide is injected following the resolution of symptoms.
Methods Forty patients undergoing successful coronary angioplasty were randomized into four equal groups. Tc-99m sestamibi was injected intravenously during the last balloon inflation (acute MPI) in 30 patients and then reinjected 1, 2, or 3 h later (delayed MPI). In a fourth group, the radiopharmaceutical was injected at 15 min following balloon deflation (delayed MPI). A final injection was performed at 24 to 48 h (late MPI) in 37 patients (93%).
Results A perfusion defect was detected in all 30 acute MPI studies; in 7/10 patients (70%) injected at 15 min; in 11/30 patients (37%) injected at 1, 2, or 3 h; and in 7/37 patients (19%) injected at 24 to 48 h. Perfusion scores were 13.0 ± 9.2 on acute MPI, 5.1 ± 2.8 at 15 min (p < 0.001 vs. acute MPI); 2.6 ± 3.0 at 1, 2, and 3 h (p < 0.001 vs. acute MPI); and 1.3 ± 2.4 at 24 to 48 h (p < 0.001 vs. acute MPI; p < 0.03 vs. delayed MPI).
Conclusions Myocardial perfusion imaging may remain abnormal for several hours following transient myocardial ischemia even when normal flow is restored in the epicardial coronary artery.
Acute rest myocardial perfusion imaging (ARMPI) with Tc-99m sestamibi is emerging as a useful technique in the assessment of patients with spontaneous chest pain and a nondiagnostic electrocardiogram (ECG). Recent studies have shown that when the radiopharmaceutical is injected while the patient is symptomatic, ARMPI is highly accurate in the diagnosis of acute myocardial ischemia as the cause of chest pain (1,2). However, in patients whose symptoms have resolved, the utility of ARMPI is less clear. Whereas some studies have shown that ARMPI will reveal a defect in most patients when injected several hours after the resolution of ischemic symptoms (3), other studies have demonstrated that ARMPI normalizes in the majority of such patients (4).
Several factors may account for the disparity in these studies. First, it is not always possible to establish the exact time of onset and resolution of ischemia. Second, the extent and severity of coronary stenoses and ischemia vary widely among patients presenting with acute chest pain and may vary over time in any one patient. Third, some patients have had a myocardial infarction (MI) and, therefore, the defects will persist.
In contrast to the acute clinical setting, during percutaneous coronary intervention the onset, duration, and extent of ischemia are more precisely known. As an experimental model, a radiopharmaceutical agent with insignificant redistribution, such as sestamibi (5), can be injected during balloon occlusion, and myocardial perfusion imaging (MPI) can be performed later. Once MPI is completed, a second injection can be done to assess the fate of the defect after restoration of normal coronary flow. We used this model to determine how long MPI remains abnormal following transient myocardial ischemia.
Patients scheduled to undergo elective, single native vessel balloon angioplasty or stenting were screened for study participation. Patients were excluded if they had evidence of a previous MI, if there was a >30% residual luminal narrowing following angioplasty, or if normal coronary flow by angiography with complete resolution of ischemic symptoms and ECG changes did not occur after the final balloon inflation. Patients with evidence of MI (creatine kinase [CK] >2× the upper limit of normal) within 24 h of angioplasty were also excluded. The hospital institutional review board approved the study and written informed consent was obtained from each patient.
Coronary balloon angioplasty and/or intracoronary stenting was performed using the femoral approach according to standard techniques. Five minutes after the last balloon inflation performed for clinical purposes, an additional “study inflation” was performed at the same site and was maintained for 3 min unless the patient developed manifestations of severe ischemia, in which case the inflation was terminated sooner. During the final minute of the study inflation, a seven-lead ECG was performed and the patient’s symptoms were assessed. Immediately following balloon deflation, angiography was performed in multiple projections. Angiograms were analyzed for residual percent stenosis and Thrombolysis In Myocardial Infarction (TIMI) flow by an angiographer who was blind to the results of MPI. At 18 to 24 h following the angioplasty, cardiac enzymes were drawn and a 12-lead ECG was performed.
Radionuclide injection protocol
The radionuclide injection protocol is outlined in Figure 1. Forty patients were sequentially assigned to one of four time interval groups: 15 min, 1 h, 2 h, and 3 h. Patients assigned to the 1-h, 2-h, and 3-h groups (30 patients) were intravenously injected with 10 to 15 mCi of Tc-99m sestamibi during the study inflation. This was administered 30 s after initiation of the 3-min inflation, allowing at least 1 min for radionuclide circulation during balloon occlusion. This was termed acute MPI. After image acquisition these patients then received a second sestamibi injection of 30 to 35 mCi at 1, 2, or 3 h following termination of the study inflation. This was referred to as delayed MPI.
Patients in the 15-min group (10 patients) were injected with 30 to 35 mCi of Tc-99m sestamibi at precisely 15 min following termination of the study inflation. This was also referred to as delayed MPI. These patients were not injected with tracer during the study inflation. Patients in all four groups were again injected with 30 to 35 mCi of Tc-99m sestamibi at 24 to 48 h after the study inflation. This was termed late MPI. Therefore, patients in the 1-, 2-, and 3-h groups underwent three MPI studies (acute, delayed, and late) and patients in the 15-min group underwent two studies (delayed and late).
Single photon emission computed tomography (SPECT) imaging acquisition began within 60 min after each sestamibi injection using an ADAC/Vertex dual-headed SPECT camera equipped with low-energy, high-resolution collimators. Sixty-four projections were obtained over a 180-degree semicircular arc. Processing was performed using a low-pass Butterworth filter with a frequency cutoff of 0.60 cycles/pixel and an order of 5.0 for reconstruction of the transaxial slices to a thickness of 6.6 mm. No preprocessing or attenuation correction was used.
Image interpretation and scoring
Images were interpreted visually by a consensus of three experienced nuclear cardiologists. Each study was interpreted separately in a blinded fashion. A standard 17-segment model (6)was used to assess perfusion using a 0 to 4 scale (0 = normal perfusion, 4 = absent perfusion). A visual summed defect score was calculated for each image. Myocardial perfusion images were considered abnormal if the visual summed defect score was ≥4, or if the score was 3 with any single segment having a score ≥2.
Quantitative methods were used to assess both global and regional sestamibi uptake. Results for global quantitation are presented as the percent of the entire left ventricle that had a perfusion defect and the results for regional quantitation as the percent of the vascular territory that was assigned to the target vessel that had a perfusion defect. Quantitation was performed by an experienced technologist using a commercially available software program (7). In this analysis, gender-specific databases were used in the comparison between the patient and normal distributions. Left ventricular regions in which perfusion intensity was >2.5 SD below the mean were classified as a defect.
In order to assess ventricular wall motion, transthoracic echocardiography was performed in all 10 patients in the 15-min group. Imaging was begun immediately following termination of the study inflation and was completed within 15 min of balloon deflation.
Data are presented as the mean ± SD or as proportions. Nominal variables were compared using a two-tailed Fisher exact test if the expected frequency in any cell was <5 and a chi-square test if the expected frequency was ≥5. Summed defect scores and quantitative defect extent were compared between >2 groups using the Kruskall-Wallis test followed by group-to-group comparisons using the Mann-Whitney Utest. Within group comparisons were made using the Friedman test for >2 within-group measurements, followed by the Wilcoxon signed-rank test. In all cases, where post-hoc testing was required, the Bonferroni correction was used to correct for multiple comparisons. A p value <0.05 was considered statistically significant.
We had estimated the proportion of patients displaying abnormal images to be 99%, 75%, 26% (4), and 20% (3)in the acute, 15-min, 1 to 3 h, and 24 to 48 h groups, respectively. At an alpha level of 0.05, a sample size of 10 in each group would provide a >90% power to detect a difference between these groups.
Clinical and procedural characteristics
The clinical and procedural characteristics of the study patients are summarized in Table 1. The dose of sestamibi was 11.7 ± 1.1 mCi for the acute MPI studies, 31.8 ± 2.5 mCi for the delayed MPI studies, and 30.8 ± 2.9 mCi for the late MPI studies.
During the study inflation, 37/40 patients (93%) had typical angina and/or an ST segment shift ≥1 mm that resolved within minutes of balloon deflation. After the procedure, all patients had a ≤20% residual stenosis with normal contrast dye flow (TIMI grade 3). There were no significant dissections and all intervention sites were well opacified. In all patients in the 15-min group, echocardiographic left ventricular wall motion was normal. At 18 to 24 h, eight patients had CK elevations that were <2× the upper limit of normal with an abnormal CK-MB fraction (mean 19.7, range: 9.7 to 46 IU; normal 0 to 7.5 IU).
Visual assessment of nuclear myocardial perfusion
The results of SPECT imaging are presented in Table 2and in Figure 2. All 30 patients injected with sestamibi at the time of coronary occlusion (acute MPI) had a perfusion defect in a territory supplied by the target vessel. The mean defect score for the acute MPI studies was 13.0 ± 9.2 (range 4 to 38). Defect scores for the 30 patients who underwent acute imaging improved from 13.0 ± 9.2 on acute MPI to 2.6 ± 3.0 on delayed MPI (p < 0.001 compared with acute MPI) and to 1.3 ± 2.4 on late MPI (p < 0.001 compared with acute MPI, p < 0.03 compared with delayed MPI). The mean defect score for the 15-min group on delayed MPI was 5.1 ± 2.8 and decreased to 1.8 ± 2.2 on late MPI (p < 0.02). This defect score for delayed MPI in the 15-min group was intermediate between that of the acute MPI group (13.0 ± 9.2) and the combined 1-, 2-, and 3-h delayed MPI group (2.6 ± 3.0) (acute vs. 15-min, p < 0.001; 15-min vs. 1-, 2-, 3-h delay, p = 0.03).
On delayed MPI, a defect was present in 18 of the 40 patients (45%). The mean defect score for these 18 patients was 6.3 ± 1.4 (range 4 to 9). A defect was present in 7/10 patients in the 15-min group, 3/10 patients in the 1-h group, 5/10 patients in the 2-h group, and 3/10 patients in the 3-h group. In the 15-min group, the mean perfusion score for the abnormal delayed images (n = 7) was 6.5 ± 1.6 (range 4 to 9), in the 1-h group (n = 3) it was 6.6 ± 1.5 (range 5 to 8), in the 2-h group (n = 5) it was 6.6 ± 1.1 (range 5 to 8), and in the 3-h group (n = 3) it was 5.3 ± 1.5 (range 4 to 7).
Thirty-seven patients underwent late imaging at 24 to 48 h. Myocardial perfusion imaging was abnormal in seven patients (19%). None of the seven patients with abnormal late images had clinical, electrocardiographic, or enzymatic evidence for MI. Among the 18 patients who had an abnormal delayed MPI, 12 patients had normalized on late imaging (Figs. 3 and 4) ⇓and 6 patients remained abnormal. In the 12 patients ⇓who normalized, the mean defect score was 6.1 ± 1.5 on delayed imaging and decreased to 0.8 ± 0.9 on late imaging. In the six patients who remained abnormal on late imaging, the mean defect score was 6.8 ± 0.9 on delayed imaging and was similar, 6.0 ± 1.2, on late imaging. Of the 22 patients whose MPI was normal on delayed imaging, in 21 patients it remained normal on late imaging and in one patient it had become abnormal.
Quantitative assessment of nuclear myocardial perfusion
Results from the quantitative assessment of myocardial perfusion correlated with those found on visual assessment and are presented in Table 3. In the 30 patients injected with sestamibi during coronary occlusion, the extent of both regional and global defects on acute MPI were similar between the 1-, 2-, and 3-h groups. In comparison with acute MPI, regional and global defect extents on delayed and on late MPI were significantly lower within each group, although there were no significant differences between the groups. For the combined 1-, 2-, and 3-h groups, mean regional defect extent improved from 49.1 ± 31.3% on acute MPI to 9.6 ± 15.2% on delayed MPI (p < 0.001 compared with acute MPI) and to 1.4 ± 4.8% on late MPI (p < 0.001 compared with acute MPI; p < 0.001 compared with delayed MPI). When global assessment was performed, mean defect extent improved from 20.2 ± 15.2% on acute MPI to 3.8 ± 5.0% on delayed MPI (p < 0.001 compared with acute MPI) and to 1.5 ± 3.5% on late MPI (p < 0.001 compared with acute MPI; p = 0.08 compared with delayed MPI).
Determinants of defect persistence
To identify factors that may have been associated with an abnormal delayed MPI, we compared patients with and without a defect on delayed imaging (Table 4). The mean acute MPI defect score was significantly greater in those patients with an abnormal delayed MPI compared with those patients with a normal delayed study on both visual and quantitative analysis. However, there was considerable overlap of acute imaging scores between these two groups. No other factor had a significant impact on the results.
This study demonstrates that sestamibi SPECT imaging may remain abnormal for up to 3 h following the resolution of transient myocardial ischemia and, in a small minority of patients, for as long as 24 to 48 h. Although abnormal, the defects on delayed imaging were considerably less intense than those found when MPI was performed during balloon inflations. Further, perfusion imaging tended to normalize as time from the ischemic episode increased. Fifteen min after the restoration of normal coronary flow, perfusion imaging was abnormal in 70% of patients. Between 1 and 3 h imaging was abnormal in 37%, and at 24 to 48 h, it was abnormal in 19% of patients. The probability of finding a defect on delayed MPI correlated with the intensity of the acute MPI defect. However, some patients with relatively low acute MPI scores also had persistent abnormalities on delayed imaging. No other variable predicted the presence or absence of a delayed defect. The validity of the findings by visual interpretation was supported by the quantitative assessment, which revealed similar results.
Eighteen of the 40 patients had an abnormal MPI on delayed study (sensitivity for detection of recent ischemia −45%). In 6 of these 18 patients imaging remained abnormal at 24 to 48 h. These abnormalities on late imaging could not be explained on the basis of MI. It is possible that these abnormal images were due to attenuation artifact. However, in 12 of the 18 patients with abnormal delayed images, the MPI had normalized at 24 to 48 h, implying that these abnormal delayed studies were not due to attenuation artifact and thus represented a postischemic abnormality in myocyte sestamibi uptake.
Previous studies in patients with recent episodes of chest discomfort have also demonstrated abnormal nuclear perfusion studies following resolution of all clinically detectable signs of ischemia (1,3,4,8). This phenomenon was first noted by Wackers et al. (8), who found that 50% of patients presenting with unstable angina had an abnormal nuclear scan when injected with thallium within 6 h of pain resolution. Bilodeau et al. (1)found that 65% of patients injected >4 h following an episode of ischemic chest discomfort had a nuclear perfusion defect. Varetto et al. (3)demonstrated that 11 patients with documented coronary artery disease had abnormal nuclear scans when injected with sestamibi 2 to 8 h following the resolution of ischemic chest discomfort, and in 9 of these patients, the perfusion scans had normalized by 24 to 48 h. In a model similar to ours, Gallik et al. (4)injected 35 angioplasty patients with sestamibi during balloon occlusion and then again 2 to 3 h postangioplasty. Similar to our findings, in that study 26% of patients had an abnormal scan on delayed imaging. Our study extends these observations by performing MPI at multiple, prespecified time points and by rigorously excluding patients with prior or periprocedural infarctions.
Ambrosio et al. (9)studied this phenomenon in patients with exercise-induced demand ischemia who developed postischemic contractile dysfunction. Patients were injected with sestamibi 30 min postexercise and, contrary to our findings, all scintigrams were normal. This model differed from our model in that the mechanism of ischemia was different (demand vs. supply) and there was no evidence for postischemic myocardial stunning in our patients. Together, these studies suggest that the occurrence of postischemic sestamibi defects may be influenced by the mechanism of ischemia.
Reason for postreperfusion defects
In the earlier studies it was assumed that the defects on delayed imaging were in many cases due to either significant residual stenoses with silent myocardial ischemia or to recent or remote infarction (1,3,4,10). In the present study, patients with MI were excluded, and therefore this could not explain our findings. Ongoing ischemia was also unlikely, as no patient had a significant residual stenosis, all had normal epicardial coronary artery flow, and all parameters of ischemia were normal at the time of sestamibi injection. In the 15-min group, echocardiographic left ventricular wall motion was normal in all 10 patients, indicating that the defects were not related to the phenomenon of postischemic myocardial stunning. These findings are consistent with animal models of transient ischemia where, following similar brief periods of ischemia, myocardial function returned to normal within several minutes (11).
One possible explanation for the delayed defects is that there were subtle microcirculatory flow abnormalities that were not significant enough to generate clinically apparent ischemia or be detected angiographically, but were significant enough to cause a sestamibi perfusion abnormality. Relevant to this hypothesis is the study of Sinusas et al. (12). Using a transient ischemia model they demonstrated a decrease in subendocardial flow as measured by microspheres, even when flow in the epicardial coronary artery had returned to normal. Thus, a persistent but relatively subtle reduction in subendocardial flow following ischemia may explain our findings of persistent sestamibi defects.
Another possible explanation for the persistent defects involves the impact of transient ischemia on the cellular mechanism of sestamibi uptake. Sestamibi enters the myocyte by passive diffusion in response to an electrical gradient across the cell and mitochondrial membranes (13,14). When this electrical potential is reduced, there is decreased movement of sestamibi into the cytosol and mitochondria (14). It has been demonstrated in animal models that transient ischemia induces biochemical and ultrastructural alterations within the myocyte that may last up to several days (15)and that may negatively impact myocardial energetics (15,16). In a clinical study, Camici et al. (17)has shown that in the postischemic period there are changes in the rates of myocyte glucose metabolism, implying altered glycogen metabolism and depletion of cellular energy stores. As a corollary to this, Kloner et al. (15)found microscopic evidence of depleted glycogen stores in dog hearts exposed to transient ischemia. Also in a dog model, DeBoer et al. (16)demonstrated that transient coronary occlusion significantly decreased subendocardial adenosine triphosphate levels, which remained depressed for up to 72 h following reperfusion. In models of pharmacologically simulated ischemic injury, Piwnica-Worms et al. (18)have demonstrated that there are metabolic derangements that cause a decrease in the membrane potentials with a subsequent decrease in sestamibi uptake.
Such metabolic abnormalities may be the basis for our findings. It is possible that the defects in delayed MPI found in the present study could be explained by a transient, postischemic decrease in the membrane potential attributable to altered myocyte energetics, ultrastructure, or metabolism as described in the earlier studies. If this were the case, it would have to be assumed that such a metabolic derangement would not be profound enough to cause a detectable decrease in contractility or change in the surface ECG, but would be significant enough to interfere with the electrical gradient across the cell membrane. Such a scenario, a “stunned membrane,” would be analogous to the transient postischemic contractile dysfunction that has been well described (19). If such a mechanism were confirmed in future studies, it would imply that sestamibi is not a pure bloodflow tracer and that its uptake may be influenced by postischemic alterations in the transmembrane electrochemical gradient.
Although this study has demonstrated that perfusion defects may persist following brief episodes of coronary occlusion, the clinical utility of this finding may be limited, as the sensitivity of delayed MPI was relatively low, with only 37% of patients having defects at 1 to 3 h following reperfusion. Furthermore, a perfusion abnormality on SPECT imaging after this interval likely underestimates the extent of jeopardized myocardium. There are, however, several important differences between the design of this study and the clinical situation.
In this model, myocardial ischemia was induced by brief coronary occlusion followed by complete, or almost complete, elimination of the culprit narrowing with restoration of normal epicardial flow. In contrast, in patients presenting with angina at rest, the culprit lesion is usually a nonocclusive, complex, severe narrowing that remains severely stenosed even after symptoms have resolved (20–22). Myocardial ischemia results from cyclic worsening of the stenosis attributable to changes in coronary tone or thrombus size (23), and bloodflow across the lesion may not have returned to normal following resolution of clinically detectable ischemia. Therefore, it might be expected that in the clinical setting, perfusion defects following the resolution of chest pain would occur more often and persist for longer periods of time than would be predicted from the results of our study. Thus, the findings from this study may underestimate the clinical utility of delayed MPI in patients with recently resolved chest pain.
These findings may also have prognostic implications. Tatum et al. (24)found that normal sestamibi imaging in patients presenting within several hours of the onset of chest discomfort was associated with a very low incidence of adverse cardiac events over the ensuing 12 months. Our finding that normal delayed imaging correlated with less severe defects on acute imaging suggests that a smaller area of myocardium was at ischemic risk in these patients compared with patients having abnormal delayed studies. Such a smaller area of ischemic risk may provide the basis for the findings in the Tatum et al. (24)study.
This study has several important limitations. First, the design of the study necessitated a low-dose/high-dose, “stress/rest” protocol. Therefore, it is possible that the delayed imaging defects were to some extent artifactual because of inadequate “fill-in” of the acute imaging defects. Both Heo et al. (25)and Taillefer et al. (26)have addressed this issue, and although they found less reversibility with a stress/rest sequence than with a rest/stress sequence, the absolute magnitude of the difference was small, with approximately 90% concordance between studies. Further, it is notable that in our study, patients in the 15-min group did not undergo acute imaging yet had abnormal delayed imaging. This argues strongly that inadequate fill-in cannot entirely explain the findings.
Second, the study inflation was maintained for 30 to 150 s following the injection of sestamibi. It is known that myocardial uptake of sestamibi occurs for up to 300 s following intravenous injection and, therefore, it is possible that the defect scores were underestimated.
Finally, attenuation correction was not available at the time of this study. Therefore, attenuation artifact as a contributor to the persistent defects cannot be excluded.
This study demonstrates that myocardial perfusion imaging with Tc-99m sestamibi may remain abnormal for at least several hours following the resolution of transient coronary ischemia despite the absence of coronary obstruction, the normalization of epicardial flow, and the resolution of all clinical manifestations of ischemia. The mechanistic basis for this finding is unclear; however, previous reports raise the possibility that transient, ischemia-induced alterations in myocyte membrane function may be a factor. These results are relevant to the use of myocardial perfusion imaging in patients who have recently presented with chest pain at rest.
- acute rest myocardial perfusion imaging
- creatine kinase
- myocardial infarction
- myocardial perfusion imaging
- single photon emission computed tomography
- Thrombolysis In Myocardial Infarction
- Received February 25, 2002.
- Revision received July 30, 2002.
- Accepted September 20, 2002.
- American College of Cardiology Foundation
- Bilodeau L.,
- Théroux P.,
- Gregoire J.,
- Gagnon D.,
- Arsenault A.
- Hilton T.C.,
- Thompson R.C.,
- Williams H.J.,
- Saylors R.,
- Fulmer H.,
- Stowers S.A.
- Varetto T.,
- Cantalupi D.,
- Altieri A.,
- Orlandi C.
- Gallik D.M.,
- Obermueller S.D.,
- Swarna U.S.,
- Guidry G.W.,
- Mahmarian J.J.,
- Verani M.S.
- Okada R.D.,
- Glover D.,
- Gaffney T.,
- Williams S.
- Cerqueira M.D.,
- Weissman N.J.,
- Dilsizian V.,
- et al.
- Kritzman J.N.,
- Ficaro E.P.,
- Liu Y.H.,
- Wackers F.J.T.,
- Corbett J.R.
- Wackers F.J.T.,
- Lie K.I.,
- Liem K.L.,
- et al.
- Ambrosio G.,
- Betocchi S.,
- Pace L.,
- et al.
- Becker L.C.
- Egeblad H.,
- Haunso S.,
- Amtorp O.
- Sinusas A.J.,
- Trautman K.A.,
- Bergin J.D.,
- et al.
- Chiu M.L.,
- Kronauge J.F.,
- Piwnica-Worms D.
- Piwnica-Worms D.,
- Kronauge J.F.,
- Chiu M.L.
- DeBoer L.W.V.,
- Ingwall J.S.,
- Kloner R.A.,
- Braunwald E.
- Camici P.,
- Araujo L.I.,
- Spinks T.,
- et al.
- Piwnica-Worms D.,
- Chiu M.L.,
- Kronauge J.F.
- Kloner R.A.,
- Bolli R.,
- Marban E.,
- Reinlib L.,
- Braunwald E.
- Ambrose J.A.,
- Winters S.L.,
- Stern A.,
- et al.
- Dangas G.,
- Mehran R.,
- Wallenstein S.,
- et al.
- Théroux P.,
- Fuster V.
- Heo J.,
- Kegel J.,
- Iskandrian A.S.,
- Cave V.,
- Iskandrian B.B.