Author + information
- Received August 24, 2001
- Revision received December 18, 2001
- Accepted December 21, 2001
- Published online March 20, 2002.
- Louise Emmett, MB, ChB*,
- Robert M Iwanochko, MD*,
- Michael R Freeman, MD, FACC†,
- Alan Barolet, MD‡,
- Douglas S Lee, MD* and
- Mansoor Husain, MD*,* ()
- ↵*Reprint requests and correspondence:
Dr. Mansoor Husain, Robert J. Burns Nuclear Cardiology Laboratory, Toronto Western Hospital, EN12-221, 200 Elizabeth Street, Toronto, Ontario, Canada M5T-2S8.
Objectives We sought to determine the level of angiographic stenosis at which reversible regional wall motion abnormalities (RWMA) are present on exercise stress technetium-99m (Tc-99m)– gated single photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI), and whether assessments of stress and rest RWMA add incremental diagnostic information.
Background Stress and rest gated SPECT MPI enables the detection of post-exercise stunning. Although some studies have correlated RWMA to the severity of MPI defects, only one previous study correlated RWMA on gated MPI to angiographic findings. However, this correlation excluded patients with rest perfusion defects and did not involve gating of rest images.
Methods One hundred patients undergoing angiography within six months of exercise stress Tc-99m (sestamibi)–gated SPECT MPI (in the absence of interim cardiac events or revascularization) were recruited. Images were acquired 15 to 30 min after stress and interpreted without knowledge of the Duke treadmill score, left ventricular ejection fraction and angiographic data.
Results The sensitivity of reversibleRWMA for angiographic stenoses >70% was 53%, with a specificity of 100%. The presence of reversible RWMA was able to stratify patients with angiographic stenoses of 50% to 79% and 80% to 99% with a high positive predictive value. A good correlation was noted between the presence of reversible RWMA and the coronary artery jeopardy score (R = 0.49, p < 0.0001). Multivariate analysis showed that the post-stress RWMA, Duke treadmill and reversible RWMA scores were significant predictors of angiographic severity.
Conclusions Post-stress and reversible RWMA, as shown by exercise stress Tc-99m–gated SPECT MPI, are significant predictors of angiographic disease and add incremental value to MPI for the assessment of angiographic severity.
Myocardial stunning can occur due to an acute ischemic injury or after significant stress-induced myocardial ischemia (1–3). The ability of stress and rest electrocardiography-gated technetium-99m (Tc-99m) single photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI) to provide myocardial perfusion data, as well as to determine left ventricular (LV) function, affords the opportunity to further evaluate the frequency and clinical relevance of stress-induced myocardial stunning.
The value of electrocardiography-gated SPECT MPI of higher dose post-stress images has been demonstrated for LV function (4,5), myocardial viability (6)and the assessment of perfusion artifacts due to tissue attenuation (7,8). With newer cameras and software, gating of both rest and stress Tc-99m sestamibi SPECT perfusion data in single-day protocols is possible, and a good correlation has been demonstrated between the severity of myocardial perfusion defects and a drop in post-stress LV ejection fraction (LVEF) (9,10). An association has also been demonstrated between regional wall motion abnormalities (RWMA) and severe perfusion defects (11). However, to date, there is little information available on the significance of exercise stress-induced RWMA, with respect to the angiographic findings (12). Indeed, this relationship has not previously been studied in patients with resting perfusion defects, nor has it involved gating of rest images.
The purpose of this study was to determine the level of angiographic stenosis at which reversibleRWMA are present on exercise stress Tc-99m–gated SPECT MPI, and whether assessments of stress and rest RWMA add incremental information to the treadmill exercise electrocardiographic (ECG) and MPI data.
The study included consecutive patients referred for routine single-day Tc-99m sestamibi rest and exercise electrocardiography-gated SPECT MPI between January 1999 and June 2000 (n = 2,330), who also underwent coronary angiography within six months of their perfusion imaging (n = 132) and had no cardiac events (myocardial infarct) or a revascularization procedure in the interval between the MPI and angiogram (n = 112). Twelve more patients were excluded because of incomplete exercise, MPI or coronary angiographic data, leaving a total of 100 patients to form the study group.
Exercise stress was performed in all patients, using a Bruce protocol treadmill test. Nitrates and beta-blockers are not routinely stopped before MPI in our laboratory. In our study group, 13% were taking long-acting nitrates, and 40% were taking beta-blockers at the time of exercise. Accepted end points for exercise stress included achievement of 85% of the target heart rate, >2 mm ST segment depression on the exercise ECG or typical ischemic chest pain. A total of 70 patients reached 85% of their target heart rate. Of the 30 who did not achieve their target heart rate, 21 had a positive exercise ECG, and 6 had typical exercise-induced angina; only 3 did not achieve a target end point.
A same-day rest/exercise Tc-99m sestamibi gated SPECT myocardial perfusion protocol (n = 67) or exercise/rest-gated SPECT MPI (n = 33) was performed. The initial study was performed with 300 MBq of Tc-99m sestamibi, followed by the second portion of the examination, using 900 MBq at least 4 h after the first injection. Rest images were obtained 1 h after the injection of Tc-99m sestamibi. In all patients, post-exercise image acquisition was commenced within 15 to 30 min after the end of exercise stress.
The SPECT imaging protocol
Images were obtained over a 180° orbit from the right anterior oblique 45° view to the left posterior oblique 45° view, using a dual-head, variable-angle gamma camera equipped with ultra-high-resolution collimators. For image acquisition, a 20% acceptance window around the 140-KeV photopeak was used. Sixty-four projections were acquired at 25 s/projection for 300-MBq studies, whereas 21 s/projection was used for 900-MBq studies. A 64 × 64 × 16 matrix was utilized for all studies. Both rest and stress acquisitions were gated at 8 frames/cycle, with 100% beat acceptance. The projection data sets were pre-filtered using a Butterworth filter (order 5, cut-off 0.66 for high-dose acquisitions; order 10, cut-off 0.50 for low-dose acquisitions) and reconstructed using filtered backprojection. When necessary, the images were motion-corrected manually.
All images were visually interpreted using a semi-quantitative method by the consensus agreement of at least two experienced observers who had no knowledge of the clinical or angiographic findings of the patient. Scoring of perfusion images was performed on non-gated images. However, gated images were made available at this session, for the exclusion of attenuation defects. Actual scoringof wall motion segments was performed with gated images only and was carried out several weeks after the session at which the perfusion scores were assigned. Neither the non-gated perfusion images nor the segmental perfusion scores were available at the time of wall motion scoring. The rest and post-stress images were interpreted for the presence, extent, severity and reversibility of perfusion and wall motion defects. A 20-segment model of the left ventricle was used for scoring perfusion defects, with a 5-point scoring system for defect severity (0 = normal perfusion; 1 = equivocal or mildly reduced; 2 = moderate reduction; 3 = severe reduction; 4 = absent perfusion) (13). The summed stress score (SSS) was the sum of all scores on the stress scan; the summed rest score (SRS) was the sum of all scores on the rest scan; and the summed difference (i.e., reversibility) score (SDS) was calculated as the difference between the summed stress and rest scores.
For interpretation of the wall motion component of the study, both rest and stress-gated images were assessed simultaneously by two experienced observers who had no knowledge of the results of the myocardial perfusion, clinical or angiographic findings of the patients. Gated images were displayed as three short-axis slices with wall contours removed. Rest and stress-gated images were displayed adjacently to allow for assessment of a change in regional wall motion. The left ventricle was again divided into the same 20-segment model for scoring. A composite severity score was utilized, which accounted for both wall motion and wall thickening within each segment. A severity scoring system of 0 to 4 was used for each segment: 0 = normal wall motion and thickening; 1 = mild hypokinesia/reduced thickening; 2 = moderate hypokinesia/reduced thickening; 3 = severe hypokinesia/reduced thickening; and 4 = akinesia/absent thickening. A summed stress score for wall motion (SSSWM) was determined from the sum of the scores on the stress-gated image, which reflects the degree and extent of post-stressRWMA. The summed rest score for wall motion (SRSWM) was calculated as the sum of the scores on the rest images, thus reflecting the degree and extent of fixed RWMA. The summed difference (i.e., reversibility) score for wall motion (SDSWM) was determined by the difference between SSSWM and SRSWM, thus reflecting the degree and extent of reversibleRWMA.
To allow for statistical correlation with angiographic findings, the 20 perfusion segments for both perfusion and wall motion were divided into anterior, inferiorand lateralvascular territories, as previously described (14,15). Briefly, segments 1–3, 7–9, 13, 14 and 19 were assigned to the anterior territory; segments 4, 10, 15, 16 and 20 to the inferior territory; and segments 5, 6, 11, 12, 17 and 18 to the lateral territory.
The end-systolic volume and LVEF were also recorded for both the rest and stress images, using a previously validated automated algorithm, and this estimation was confirmed visually (16). Johnson et al. (11)determined the serial reproducibility of LVEF estimation to be ±5% (2 SD from the mean value). We considered a drop in LVEF ≥5% to be significant for the purposes of statistical analysis. The presence or absence of transient ischemic dilation (TID) was determined by the visual consensus of the two experienced observers. No attempt to record the severity of TID was made in this study.
All coronary angiograms were read by two experienced observers, with consensus; they had no knowledge of the clinical and perfusion findings of the patients. The site (distal or proximal) of angiographic stenoses was identified visually; however, the severity of angiographic stenoses was assessed by a validated quantitative coronary analysis (17). For patients with a previous coronary artery bypass graft surgery (n = 7), both the site and severity of disease in the native vessels and bypass grafts were documented. Dominance of the right versus circumflex coronary artery was determined.
A coronary artery jeopardy score was calculated according to the method described by Califf et al. (18). Briefly, this method assigns a score of 2 for coronary angiographic stenosis >75% in the left anterior descending coronary artery, major septal perforator, left circumflex coronary artery, obtuse marginal branch and posterior descending coronary artery. In patients with a left-dominant system, the right coronary artery was assigned no points. Each vessel distal to ≥75% stenosis was given a score of 2 points (to a maximum of 12 points). The presence or absence of a collateral channel was not considered in determining either the degree of angiographic stenosis or the jeopardy score.
For the purposes of statistical analysis and comparison with perfusion imaging, angiographic findings were also divided into anterior, lateral and inferior vascular territories. The anterior territory included stenoses within the left main, left anterior descending, diagonal and septal perforator coronary arteries. The lateral territory included the circumflex and obtuse marginal coronary arteries, and the inferior territory included the right coronary and posterior descending coronary arteries. This assignment was validated for each patient.
Continuous data are expressed as the mean value ± SD. Comparisons were made using the Student ttest for normally distributed variables and Wilcoxon rank-sum test for nonparametric variables. Categorical data were assessed using the chi-square or Fisher exact test, as appropriate. For multivariate analysis, a backward stepwise elimination method was employed in a least squares linear regression model. Variables with a level of significance (p) <0.10 on univariate analysis were entered into the multivariate regression model by employing a multivariate significance level of p < 0.05 (two-sided). The model was also re-evaluated with a stepwise multivariate regression approach, by adding univariate predictors into the model and assessing for the incremental change in the residual variance of the model. We were cautious to avoid highly collinear co-variates and variables resulting in high variance inflation in the modeling strategy. Residual plots were constructed to test the validity of the statistical assumptions of the linear regression model. A Pearson correlation coefficient was used to evaluate associations between the predefined perfusion variables and RWMA scores, if the variables assessed satisfied the assumptions of normality. A comparison of sample correlation coefficients was performed by using Fisher ztransformation of sample R estimates (two-sided). Sensitivity, specificity and positive and negative predictive values were determined for SSS and SDS, as well as for SSSWM and SDSWM, for predicting angiographic stenoses of both >50% and >70%. Statistical analyses were performed using SYSTAT version 9.0 (SPSS Science, Chicago, Illinois) and SAS version 8.0 (SAS Institute, Cary, North Carolina).
The study group consisted of 23 women and 77 men (mean age 60 ± 9 years). The most common reason for referral was the diagnosis of chest pain (55%), which was almost equal for patients with typical (27%) and atypical (28%) angina. Another 29% of patients had a documented history of myocardial infarction, and these patients were referred for risk stratification. The patients’ clinical characteristics, stratified by the presence or absence of reversible RWMA, are shown in Table 1. Patients with reversible RWMA were older (64 ± 8 years) and more likely to be male (34 [92%] of 37) than patients who did not show reversible RWMA (58 ± 10 years, p = 0.003; 43 [68%] of 63, p = 0.007). Significantly more patients with reversible RWMA had a post-stress (36 [97%] of 37) or rest (27 [73%] of 37) perfusion defect than those who did not have reversible RWMA (37 [59%] of 63, p = 0.001; and 29 [46%] of 63, p = 0.009, respectively).
All patients in our study had undergone coronary angiography within six months (58 ± 50 days) of their MPI study, and only those with no interval cardiac event or revascularization procedure were included. Eighty-eight percent of patients underwent coronary angiography after MPI. In our group, 24% of patients demonstrated no hemodynamically significant lesions (<50% stenosis) on the coronary angiogram; 28% had single-vessel disease; 20% had double-vessel disease; and 28% had triple-vessel disease.
Perfusion imaging revealed 9% of patients to have only fixedperfusion defects (SRS ≥4, SDS ≤2), 47% to have fixed and reversibleperfusion defects (SRS ≥4 and SDS >2) and 24% to have only reversibleperfusion defects (SRS <4 and SDS >2). A standard cut-off value of SSS <4 was used to define normalperfusion (19,20), and 20% of the study group was categorized as such. The sensitivity and specificity for the perfusion component of the study are shown in Figure 1. In our study group, a post-stress perfusion defect (SSS ≥4) alone achieved a sensitivity of 89% and specificity of 63% for the diagnosis of angiographic stenosis >70%.
Regional wall motion
Thirty-seven percent (n = 37) of patients in the study demonstrated a reversible regional wall motion abnormality (SDSWM >2). The presence or absence of a reversible (SDSWM >2) and post-stress (SSSWM ≥4) RWMA was compared to the grade of angiographic stenoses within the anterior, lateral and inferior coronary vascular territories. The specificity of reversible RWMA for angiographic stenosis >70% was 100% overall and 96%, 97% and 94% for the anterior, lateral and inferior vascular territories, respectively (Fig. 1). This compared well to the specificity of 87% overall and 76%, 82% and 87% for the anterior, lateral and inferior walls, respectively, for post-stress RWMA. However, the overall sensitivity of reversible RWMA for detecting angiographic stenosis >70% (53%) was lower than that of post-stress RWMA (71%) (Fig. 1).
When complete occlusions were excluded and patients were stratified by the severity of their angiographic stenoses (50% to 79% and 80% to 99%), the presence of reversible RWMA distinguished a higher angiographic severity, with a positive predictive value of 77%, 88% and 86% for the anterior, lateral and inferior vascular territories, respectively. Figure 2compares the prevalence of reversible RWMA and reversible perfusion defects in patients with angiographic stenoses of 50% to 79% (mean and median values ∼70%) and 80% to 99% (mean and median values ∼90%) angiographic stenoses.
Although the numbers of patients are not large, a subgroup analysis was undertaken to explore potential differences between the rest/stress (n = 67) and stress/rest (n = 33) imaging protocols. The sensitivity of a post-stress perfusion defect for the detection of angiographic stenoses >70% was lower in stress/rest versus rest/stress protocols (77% [20/26] vs. 95% [42/44]); however, the specificity of this finding was somewhat higher (71% [5/7] vs. 61% [14/23]). The sensitivity and specificity of post-stress RWMA followed a similar pattern in the stress/rest versus rest/stress protocols (sensitivity 65% [17/26] vs. 75% [33/44]; specificity 86% [6/7] vs. 69% [16/23]). Unexpectedly, the sensitivity of reversible RWMA for >70% angiographic stenoses was greater in the stress/rest group (61% [16/26] vs. 47% [21/44]). However, the specificity of a reversible RWMA remained perfect in both imaging protocols (100% [7/7] vs. 100% [23/23]).
Compared with men, women in this study had a higher incidence of normal coronary angiograms (<50% stenoses in all epicardial arteries; 48% [11/23] vs. 17% [13/77], p < 0.001) and a lower incidence of multivessel disease (22% [5/23] vs. 64% [43/67], p < 0.01). The sensitivities of a post-stress perfusion defect for detecting angiographic stenoses >70% were 90% (9/10) in women and 88% (53/60) in men, and the corresponding specificities were 77% (10/13) and 53% (9/17). Although the sensitivity of reversible RWMA was only 30% (3/10) in women and 57% (34/60) in men, the specificity of reversible RWMA was 100% in both genders (13/13 women vs. 17/17 men).
Correlation between perfusion and wall motion
Figure 3reveals the correlation between the severity of reversible RWMA (SDSWM) and reversible perfusion defects (SDS). Although a significant correlation between the severity of reversible perfusion defects and reversible RWMA was observed in all patients, this relationship tended to be stronger when patients with significant, fixed perfusion defects (SRS ≥4, n = 55) were excluded from the analysis (R = 0.61, p < 0.0001, n = 100 vs. R = 0.76, p < 0.0001, n = 45) (Fig. 3).
Of all clinical, ECG, exercise, perfusion, wall motion and LV function variables recorded (see Methods), only the 1) SSSWM; 2) SDSWM; 3) SSS; 4) Duke treadmill score; 5) presence of TID; and 6) >5% reduction in LVEF from rest to post-stress correlated with the angiographic jeopardy score on univariate analysis (p < 0.05). Of these, only the SSSWM, SDSWM and Duke treadmill score emerged as independent incremental predictors of the angiographic jeopardy score on stepwise multivariate regression (p < 0.0001, p < 0.02 and p < 0.0001, respectively). Thus, after adjusting for reversible and post-stress RWMA and Duke treadmill scores, the effects of SSS, TID or >5% drop in LVEF were no longer significant predictors of the angiographic jeopardy score.
The major finding of this study is that stress-induced (i.e., reversible) RWMA on a single-day exercise Tc-99m sestamibi-gated cardiac SPECT is highly specific for severe angiographic stenosis. Moreover, the presence or absence of reversible RWMA correlated well with the coronary anatomy, as measured by the number of diseased vessels or coronary artery jeopardy score. Therefore, the presence of reversible RWMA on a single-day MPI protocol may indicate a patient at high risk of future cardiac events (18). Although RWMA present on both the rest and stress images were more sensitive for the presence of angiographic stenoses >70%, they were less specific than reversible RWMA. Both reversible and post-stress RWMA, as well as high Duke treadmill scores, were significant independent predictors of the severity of the angiographic jeopardy score.
Only one previous study compared RWMA on MPI to the severity and extent of angiographic stenoses (12). Although this study showed that post-exercise RWMA predicted coronary artery disease, it did not evaluate the incremental value of comparing post-stress to rest LV function. Moreover, this study excluded patients with fixed perfusion defects, thus limiting its application from a large number of patients in whom gated cardiac SPECT is frequently performed. In contrast, the current study included patients with a broad range of perfusion defects and assessed the value of gating both rest and stress MPI. We have examined both the level of angiographic stenosis required to induce reversible RWMA and the independent incremental information that an assessment of RWMA adds to the perfusion and ECG data.
By definition, myocardial stunning is present if LV dysfunction is reversible and persists when myocardial perfusion has returned to normal (21). Ambrosia et al. (3)studied 30 patients with known coronary artery disease to determine the period of time that stunning of the myocardial wall persisted after cessation of exercise. Consistent with our findings, they showed that less severe angiographic lesions were associated with more prompt resolution of regional contractile abnormalities. We demonstrate the absence of reversible RWMA with angiographic stenoses <70%. In fact, the presence of reversible RWMA was able to differentiate well between angiographic stenoses of 50% to 79% and 80% to 99%.
Johnson et al. (11)used a method of chordal shortening to quantify RWMA on post-stress images in patients with reversible perfusion defects. They showed a significant reduction in post-stress regional wall motion in ischemic versus non-ischemic segments. We also detected a strong correlation between ischemia, as demonstrated by reversible perfusion defects, and the presence of reversible RWMA. This association was particularly strong when patients with purely reversible perfusion defects were examined, suggesting that the presence of a myocardial infarct and fixed wall motion abnormality may make it more difficult to identify regions of reversible RWMA in adjacent ischemic regions.
Sensitivity of RWMA
The presence of reversible RWMA was relatively insensitive for the detection of high-grade angiographic stenoses. This may be for a number of reasons. In our study, hibernating myocardium with persistent RWMA due to severe ischemia would be classified as a fixed RWMA rather than as a reversible one. It is also possible that the visual scoring technique and inherent resolution of SPECT is insensitive for the detection of mild stunning.
Our study group included a broad range of patients, with up to 55% demonstrating partially or completely fixed perfusion defects. Regions of infarction are likely to have fixed rather than reversible RWMA. Reflecting this, the sensitivity of post-stress RWMA for angiographic lesions >70% was higher than that of reversible RWMA, as both fixed and reversible RWMA are encompassed in the former.
Given that the sensitivity of reversible RWMA was somewhat reduced in women, compared with men, the fewer RWMA observed in women may partly reflect a gender-dependent difference in diagnostic accuracy. However, given that the women in our study had an approximate threefold higher incidence of normal coronary angiograms and a significantly lower burden of multivessel disease, their lower incidence of RWMA is likely to reflect a true paucity of stunning, based on a bias in the referral pattern for angiography.
The angiographic jeopardy score was developed and validated by Califf et al. (18)as a method of determining the prognosis on the basis of angiographic findings. The five-year survival rate was 97% in patients with a score <2 and only 56% in patients with a maximal score of 12. We used the jeopardy score to indirectly explore the possible prognostic implications of perfusion defects, RWMA, TID, >5% reduction in LVEF from rest to post-stress and the Duke treadmill score. Multivariate regression analysis demonstrated that both reversible and post-stress RWMA and Duke treadmill scores added incremental information to the perfusion data in predicting the severity of angiographic disease. Further prognostic studies are indicated to determine whether the finding of reversible RWMA on MPI increases the risk of cardiac events and warrants early intervention.
Both rest/stress and stress/rest single-day MPI protocols were employed in this study, in accordance with our laboratory’s routine clinical practice. Although the lower-dose, stress-gated portion of the stress/rest protocol appeared to reduce the sensitivity of post-stress perfusion defects and RWMA, this was not the case for reversible RWMA. Moreover, the major finding of this study—that reversible RWMA was highly specific for high-grade angiographic stenosis—was demonstrated in both protocols. Although the validity of gating low-dose Tc-99m–based perfusion protocols has been confirmed by a number of other studies demonstrating a high correlation between rest and stress LVEF (9,10,22), the evaluation of reversible RWMA may be improved by high-dose, stress andrest gated imaging. Thus, a similar study using a two-day protocol may yield more accurate results.
Another potential limitation of our study is that in patients with severe perfusion defects and few counts, wall motion may be erroneously underestimated. This issue was not specifically addressed in this study, in which a semiquantitative assessment of wall motion was made.
An adequate collateral circulation in territories with high-grade angiographic stenoses may have influenced both the perfusion and RWMA results. However, the angiographic information required to address this question was not available in the current study.
All patients in this study underwent angiography within six months of the myocardial perfusion study. As the majority of the patients were referred for angiography as a result of an abnormal myocardial perfusion scan, an inherent selection bias may affect our results. However, the sensitivity and specificity of our perfusion data are comparable to those in the published data, and given that the study group was composed of patients with and without significant coronary artery disease, we believe that meaningful extrapolation of our results may be possible.
Reversible RWMA on Tc-99m–gated cardiac SPECT is highly specific for high-grade angiographic stenosis. Incremental information from assessment of RWMA, including post-exercise and reversible RWMA, added to the clinical and perfusion data, thus improving the evaluation of coronary artery disease. More widespread implementation of gated rest and stress MPI may be warranted. The prognostic significance of these findings requires prospective evaluation.
☆ Dr. Lee is a Research Fellow of the Heart and Stroke Foundation of Canada, Canadian Institutes of Health Research. Dr. Husain is recipient of a Clinician Scientist Award of the Canadian Institutes of Health Research.
- electrocardiogram or electrocardiographic
- left ventricular
- left ventricular ejection fraction
- myocardial perfusion imaging
- regional wall motion abnormality
- summed difference (perfusion) score (for wall motion)
- single photon emission computed tomography
- summed rest (perfusion) score (for wall motion)
- summed stress (perfusion) score (for wall motion)
- transient ischemic dilation
- Received August 24, 2001.
- Revision received December 18, 2001.
- Accepted December 21, 2001.
- American College of Cardiology Foundation
- Homans D.C.,
- Laxson D.D.,
- Sublett E.,
- Pavek T.,
- Crampton M.
- Ambrosio G.,
- Betocchi S.,
- Pace L.,
- et al.
- Williams K.A.,
- Taillon L.A.
- Chua T.,
- Kiat H.,
- Germano G.,
- et al.
- DePuey E.G.,
- Rozanski A.
- Smanio P.E.,
- Watson D.D.,
- Segalla D.L.,
- Vinson E.L.,
- Smith W.H.,
- Beller G.A.
- Johnson L.L.,
- Verdesca S.A.,
- Aude W.Y.,
- et al.
- Berman D.S.,
- Hachamovitch R.,
- Kiat H.,
- et al.
- Matzer L.,
- Kiat H.,
- Friedman J.D.,
- van Train K.,
- Maddahi J.,
- Berman D.S.
- Berman D.S.,
- Kiat H.,
- Friedman J.D.,
- et al.
- Germano G.,
- Kiat H.,
- Kavanagh P.B.,
- et al.
- Califf R.M.,
- Phillips H.R. 3rd.,
- Hindman M.C.,
- et al.
- Hachamovitch R.,
- Berman D.S.,
- Kiat H.,
- et al.
- Hachamovitch R.,
- Berman D.S.,
- Shaw L.J.,
- et al.
- Braunwald E.,
- Kloner R.A.