|
| HOME | SUBSCRIPTIONS | CURRENT ISSUE | PAST ISSUES | CARDIOSOURCE | SEARCH | HELP | FEEDBACK |
|
|
|
|||||||||
|
|||||||||||
|
J Am Coll Cardiol, 1998; 32:1987-1995 © 1998 by the American College of Cardiology Foundation |
* Department of Imaging (Division of Nuclear Medicine), Department of Medicine (Division of Cardiology), and CSMC Burns & Allen Research Institute, Cedars-Sinai Medical Center, Los Angeles, California, USA
b Department of Medicine, University of California Los Angeles, School of Medicine, Los Angeles, California, USA
Department of Medicine (Division of Cardiology), New York Hospital-Cornell Medical Center, New York, New York, USA
Department of Medicine, Emory University School of Medicine, Atlanta, Georgia, USA
Manuscript received March 17, 1998; revised manuscript received July 21, 1998, accepted August 6, 1998.
Address for correspondence: Daniel S. Berman, MD, Department of Imaging, Cedars-Sinai Medical Center, 8700 Beverly Boulevard, Room A042, Los Angeles, California 90048
bermand{at}cshs.org
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionReferences |
|---|
Background. Extent, severity and reversibility of exercise myocardial perfusion defects have been shown to correlate with prognosis. However, most studies examining the prognostic value of SPECT in chronic coronary artery disease (CAD) have been based on visual analysis by experts.
Methods. We studied 1,043 consecutive patients with known or suspected CAD who underwent rest Tl-201/exercise Tc-99m sestamibi dual-isotope myocardial perfusion SPECT and were followed up for at least 1 year (mean 20.0 ± 3.7 months). After censoring 59 patients with early coronary artery bypass grafting or percutaneous transluminal coronary angioplasty, <60 days after nuclear testing, the final population consisted of 984 patients (36% women, mean age 63 ± 12 years).
Results. During the follow-up period, 28 hard events (14 cardiac deaths, 14 nonfatal myocardial infarctions) occurred. Patients with higher defect extent (>10%), severity (>150) and reversibility (>5%) by quantitative SPECT defect analysis, as well as those with an abnormal scan (>2 abnormal segments, summed stress score >4 and summed difference score >2) by semiquantitative visual SPECT analysis, had a significantly higher hard event rate compared to patients with a normal scan (p < 0.001). With both visual and quantitative analyses, hard event rates of approximately 1% with normal scans and 5% with abnormal scans (p > 0.05) were observed over the 20-month follow-up period. A Cox proportional hazards regression model showed that chi-square increased similarly with the addition of quantitative defect extent and visual summed stress score variables after considering both clinical and exercise variables (improvement chi-square = 11 for both, p < 0.0007). There were no significant differences in the areas under receiver operating characteristic curves between quantitative and visual analysis (p > 0.70). Linear regression analysis also indicated that quantitative assessments correlated well with visual semiquantitative assessments.
Conclusions. The findings of this study indicate that automatic quantitative analysis of exercise stress myocardial perfusion SPECT is similar to semiquantitative expert visual analysis for prognostic stratification. These findings may be of particular clinical importance in laboratories with less experienced visual interpreters.
| ||||||||||||||||||||||||
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
Exercise myocardial perfusion protocol. All patients underwent rest Tl-201/exercise Tc-99m sestamibi separate acquisition dual-isotope myocardial perfusion SPECT, as previously described (9). Briefly, Tl-201 (3.0 to 4.0 mCi) was injected intravenously at rest, with dose variation based on patient weight. Rest Tl-201 SPECT was begun 10 min after radioisotope injection. Immediately after imaging, patients performed a symptom-limited exercise treadmill test (ETT) using the standard Bruce protocol with 12-lead electrocardiographic (ECG) recording each minute of exercise and continuous monitoring of leads aVF, V1 and V5. Blood pressure was measured and recorded at rest, at the end of each exercise stage and at peak exercise. Exercise end points included physical exhaustion, severe angina, sustained ventricular tachycardia, hemodynamically significant supraventricular dysrhythmias or significant exertional hypotension. The maximal degree of ST segment change at 80 ms after the J point of the ECG was measured and assessed as horizontal, upsloping or downsloping. Exercise ECG responses were considered uninterpretable if the patients were taking digoxin or had a paced ventricular rhythm or if the baseline ECG demonstrated left ventricular hypertrophy, nonspecific ST-T wave changes, left or right bundle branch block or Wolff–Parkinson–White syndrome. At near-maximal exercise, Tc-99m sestamibi (25 to 35 mCi) was injected intravenously, with dose variation based on patient weight. Exercise was continued at maximal workload for 1 min and at one stage lower for two additional minutes, when possible. Technetium-99m sestamibi SPECT imaging was begun 15 to 30 min after radiopharmaceutical injection.
SPECT acquisition protocol. All SPECT acquisitions were as previously described (9) employing a large field of view gamma camera and a low energy high resolution collimator to obtain 64 projections at 20 s/projection over a semicircular 180° arc extending from the 45° right anterior oblique to the 45° left anterior oblique projection. For Tl-201 SPECT, two energy windows were utilized, a 30% window centered on the 68- to 80-keV peak and a 20% window centered on the 167-keV peak. For Tc-99m sestamibi SPECT, a 15% window centered on the 140-keV peak was used. Images were acquired using a 64 x 64 image matrix. All images were subject to quality control measures, including cinematic display for assessment of patient motion, corrections for field nonuniformity and center of rotation. No attenuation or scatter correction was used. Patients with resting defects were often asked to return the next day for a 24-h delayed Tl-201 imaging to assess reversibility (10). Whenever available, these late redistribution thallium images were used in place of the resting ones for both visual and quantitative interpretation.
Automatic reconstruction and quantification of myocardial perfusion SPECT. Dual-isotope SPECT tomograms were reconstructed and reoriented using an automatic algorithm described by Germano et al. (11). When automatic reconstruction or reorientation failed, reconstruction limits and axes were assigned manually. The remainder of the automatic quantitative process was performed by modifying the CEqual algorithm originally described by Ezekiel et al. (12) and Garcia et al. (13) with the addition of an automatic segmentation algorithm described by Germano et al. (11). The left ventricle (LV) is described by an apex and base location, with a center coordinate and average radius. Myocardial sampling consists of generation of maximum count circumferential profiles using a spherical search for the apex and a cylindrical search for the remainder of the short-axis tomograms. The pixel locations corresponding to the maximal myocardial counts, as well as the count values, are determined. The LV radius is reset to the average radius in the circumferential profile, and the basal slice for assessment is determined by the slice with a maximum count of 50% or less than the hottest myocardial slice. The final sampling of the myocardium is made using these values. Modifications to the previously described CEqual algorithm were as follows. The identification of the LV is based on a clustering algorithm applied to short-axis reconstructed tomograms. By applying adaptive thresholding to iteratively define clusters corresponding to expected ventricular volumes, the algorithm separates myocardium from potential extracardiac structures reflecting hepatic, splenic or gut activity. If the results of the clustering approach fail to meet certain criteria, the algorithm applies a Hough transform and a scoring function weighted to favor a doughnut-like configuration (11). Prior to comparison to normal limits the stress and rest profiles are normalized to the most normal wall of the patient’s entire stress data set (4). Separate normal limits were generated and used for the myocardial perfusion evaluation of stress Tc-99m sestamibi and rest Tl-201 (14). A quality control report is generated to allow operator verification of the performance of the automatic algorithm and to manually correct any suboptimal values. The count values are plotted on a two-dimensional polar map representing the entire LV myocardium. The criteria for abnormality used for this evaluation were determined by receiver operating characteristic (ROC) curve analysis of the number of standard deviations from the mean which best separated visually normal from abnormal dual-isotope myocardial perfusion on a region by region basis (4). These optimal standard deviations were used as thresholds of abnormality for the total myocardium and three major myocardial regions (left anterior descending artery, left circumflex coronary artery and right coronary artery). Quantitative defect extent was defined by summation of the number of profile points falling below the dual-isotope normal limit expressed as percentage of the LV myocardium. Quantitative defect "severity" (actually a measure of defect extent and severity) was defined by the sum of the product of all profile points below the normal limits multiplied by their respective number of standard deviations below the normal mean count (15). Quantitative defect reversibility was calculated by scaling the rest and stress images to a common value of the most normal region of the stress distribution and subtracting the stress from the rest data (Fig. 1). Defect extent >10% of LV, defect severity index >150 and defect reversibility >5% of LV were defined as abnormal (16). When automatic LV identification or quantitation steps failed, the studies were subjected to observer-guided quantitative analysis using the standard commercially available algorithm (17).
|
Three global perfusion indices previously defined by our group were employed to combine assessments of defect extent and severity (3). A summed stress score (SSS) was obtained by adding the scores of the 20 segments of the stress sestamibi images. A summed rest score (SRS) was similarly obtained by adding the scores of the 20 segments of the rest Tl-201 images or 24-h delayed Tl-201 images. The sum of the differences between the stress and rest scores of each of the 20 segments was defined as the summed difference score (SDS) or reversibility score, an index of jeopardized myocardium. Summed difference score >2 was considered a reversible defect abnormality. A segment was considered abnormal if the segmental score was 
2. Segments with scores of 1 on both stress and rest images were considered to reflect attenuation and were considered as scores of 0-0 for purpose of analysis (1). Images were also visually assessed for the presence of transient ischemic dilation (TID) of the LV (19).
Likelihood of coronary artery disease. For the purpose of examining patients in different risk subsets, we used analyses of the pre- and post-ETT likelihood of CAD as aggregate clinical descriptors of proven prognostic importance. The pre-ETT likelihood was calculated using CADENZA (20) and was based on Bayesian analysis of age, gender, symptom classification, rest ECG, cardiac risk factors and the results of ECG stress testing. The post-ETT likelihood was based on pre-ETT likelihood of CAD, and the clinical and ECG responses to stress (3). Neither of these variables considered the information from the nuclear test.
Patient follow-up. Follow-up was performed by dedicated research personnel. The follow-up duration was at least 1 year, mean 20.0 ± 3.7 months. Hard events were defined as either cardiac death (confirmed by review of death certificate and hospital chart of physician’s records) or nonfatal myocardial infarction (documented by appropriate ECG and cardiac enzyme level changes). If a patient was found to have had both events after nuclear testing, only cardiac death was considered.
Statistical analysis. All continuous variables are expressed as means ± SD. The mean differences for continuous variables were compared using the Student t test (two tailed). All analyses comparing the frequency of cardiac events based on the perfusion defect size, severity and reversibility were performed using a chi-square statistic. Linear regression analysis was used to determine the correlation for perfusion defect between variables of visual and quantitative SPECT. A p value <0.05 was considered statistically significant in all analyses.
Receiver operating characteristic curve analysis represented sensitivities and false positive rates for predicting cardiac events for quantitative versus visual SPECT. The difference between the area under two ROC curves was calculated and compared (21). Receiver operating characteristic areas were expressed as the area ± SE.
The Cox proportional hazards regression model (22) was used to assess the incremental prognostic value of components of testing. A stepwise multivariate Cox regression analysis (SPSS version 7.5) was performed by using the following: 1) a composite clinical variable alone (pre-ETT likelihood of CAD/ischemia) (model 1); 2) a composite of clinical + exercise variables (post-ETT likelihood of CAD/ischemia) (model 2); 3) clinical + exercise + the best nuclear variable (quantitative defect extent or visual SSS). The increment in prognostic value by nuclear testing was determined after "forcing in" the most predictive prescan variables and then adding the most predictive nuclear variables (model 3). The chi-square value was calculated from the log of the ratio of maximal partial likelihood functions.
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
|
Comparison between quantitative and semiquantitative visual analyses of defect extent, severity and reversibility. Table 2 compares the various related quantitative and semiquantitative visual assessments in all patients by gender. Close correlations were noted for all assessments: quantitative defect extent versus number of abnormal segments (r = 0.68) and versus SSS (r = 0.70), quantitative defect severity versus SSS (r = –0.75) and quantitative defect reversibility versus SDS (r = 0.67). Comparison of quantitative and semiquantitative visual assessments by gender revealed significantly closer correlation in men than women (p < 0.001).
|
|
|
|
|
|
|
|
|
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
In the present study, the various objective quantitative indices performed equally well compared to analogous consensus expert visual analyses for assessment of prognosis. With both visual and quantitative analyses, hard event rates of approximately 1% with normal scans and 5% with abnormal scans (p > 0.05) were observed over the 20-month follow-up period. Multivariate analysis demonstrated the same incremental prognostic value for quantitative defect extent and visual SSS after clinical and exercise variables were considered. By ROC curve areas, in comparison to semiquantitative visual analysis, automatic quantitative analysis of corresponding scintigraphic variables performed equally well for prognostic assessment. Linear regression analysis also indicated close relationships between quantitative and semiquantitative visual assessments for defect extent, severity and reversibility variables. Since a single reader interprets images in most clinical settings, the fact that quantitative analysis was nearly equivalent to consensus visual analysis takes on greater clinical importance. It should be noted, however, that with the current quantitative method, a significantly greater proportion of patients were categorized as low risk by semiquantitative visual analysis than by quantitative analysis (Figs. 2 and 3, p < 0.001). Most likely this is related to the lack of incorporation of artificial intelligence rules for the recognition of artifacts with the current quantitative software. Evidence supporting this explanation is seen in the stronger correlations of quantitative analysis with expert visual analyses in men than women. Although we used gender-matched normal limits for quantitation, the program does not account for differences between women in breast size and breast attenuation, thus creating a problem in recognition of marked breast attenuation as an artifact. With expert visual analysis, on the other hand, this pattern is frequently identifiable. Even though automatic quantitation is intrinsically more reproducible, this particular quantitative implementation can still benefit from expert assessment of artifacts in the segment of the population most affected by them.
Quantitative Tl-201 or Tc-99m sestamibi myocardial perfusion SPECT has been shown to be accurate for detection and localization of CAD (26,27), and assessment of myocardial infarct size (28,29), and correlates highly with expert visual interpretation (5). Previous validations of CEqual were based on determination of the diagnostic accuracy of the program (5,17). These validations were conducted using both expert visual analysis and coronary angiography as the "gold standard." In preliminary reports, the results for CEqual with the dual-isotope imaging protocol have been shown to be similar for detection of the presence or absence of CAD (14). These results would be expected, since the same stress myocardial perfusion agent and dose were used for both protocols. Only limited data are available, however, regarding the prognostic value of quantitative myocardial perfusion SPECT. Miller et al. (30) and Mahmarian et al. (26) have documented the prognostic value of quantitative myocardial perfusion SPECT assessment in patients with acute myocardial infarction. Kamal et al. (31) and Pancholy et al. (32) have reported the prognostic value of adenosine quantitative thallium SPECT and exercise quantitative thallium SPECT in women. Both studies found that patients with large perfusion defect (15% of myocardium) had worse outcome than patients with no or small perfusion defect (<15% of myocardium). To our knowledge this is the first study to employ quantitative Tc-99m sestamibi SPECT analysis in assessment of prognosis in a large population of patients with chronic CAD and to systematically compare the results of semiquantitative visual and quantitative assessment for prognosis.
A large number of studies have documented the importance of stress myocardial perfusion SPECT with sestamibi in risk stratification (1,3,19,33,34) and in guiding patient management (3,35,36). All of these reports have relied on the expert semiquantitative visual analysis of myocardial perfusion SPECT studies. In previous studies from our laboratory, objective quantitative analysis was not employed for prognostic assessment, since it was not a routine part of our daily quality controlled data acquisition. The reliance on semiquantitative visual inspection by experts has limited the degree to which the results from our individual center can be generalized to other laboratories. The present study was made feasible by the advent of an automatic approach to identification of the LV for quantitative assessment of myocardial perfusion SPECT, since this new step allowed automatic batch processing of the scintigraphic data. The results of the present study, in which an objective quantitative analysis is found to contain prognostic information indistinguishable from that of the expert semiquantitative analysis, suggests that these prognostic results could be obtained in any laboratory providing high quality exercise imaging protocols, acquisition, processing and quantitation. This result has potential major implications for the field of nuclear cardiology and represents a clinically important advantage over stress echocardiography, which at the current time has no widely utilized objective quantitative analysis.
The main improvement noted between the previous CEqual program and this version was in the success rate observed for the automatic processing function. Improved methods for identifying the myocardium, and selection of apex, base, center and radius resulted in a processing success rate of 93% (911/984). The success rate for the previous CEqual version was clinically noted to be around 70% to 75%. This improvement in the automatic processing component of the algorithm resulted in an increase in the objectivity and likely the reproducibility of the CEqual quantitative analysis program.
In this study, a number of other variables that have been shown to be significant markers of severe and extensive CAD or harbingers of adverse outcomes, such as a multivessel disease pattern, were shown not to be significant predictors in a multivariable analysis after adjusting for the extent and severity of perfusion abnormalities. We have previously shown that the presence of TID, multivessel perfusion abnormalities and perfusion abnormalities in a left anterior descending artery distribution (3,37), after adjusting for the extent and severity of stress perfusion abnormalities in a multivariable model, are no longer predictive of adverse outcomes. For all of these variables, the greater associated risk is based in large part on the tendency of these patterns to be present in patients with severe and extensive perfusion abnormalities.
Study limitations. Although the present study was performed in a relatively large group of patients, the results are based on a population referred for nuclear testing, and thus potentially different from unbiased CAD populations. Not many hard events occurred in this study group, in which 61% (605 patients) had normal scans by visual analysis, limiting the ability to perform subanalyses such as the difference in prognostic importance of fixed or reversible defects or male versus female findings. Visually assessed TID was not of prognostic importance in this study. The potential incremental prognostic value of quantitative TID could not be assessed in this report, since the gated SPECT program required for the automatic analysis of TID was not being utilized at the time of the study.
Conclusions. The results of this study demonstrate that automatic quantitative analysis of exercise dual-isotope myocardial perfusion SPECT has comparable prognostic value to that of semiquantitative visual analysis. The findings suggest that the type of prognostic information reported using semiquantitative visual analysis by experts may be achieved by a more widely applicable, objective quantitative analysis, which may be of particular clinical importance in laboratories with less experienced visual interpreters.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. D. Metz, M. Beattie, R. Hom, R. F. Redberg, D. Grady, and K. E. Fleischmann The Prognostic Value of Normal Exercise Myocardial Perfusion Imaging and Exercise Echocardiography: A Meta-Analysis J. Am. Coll. Cardiol., January 16, 2007; 49(2): 227 - 237. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Yoshinaga, B. J.W. Chow, K. Williams, L. Chen, R. A. deKemp, L. Garrard, A. Lok-Tin Szeto, M. Aung, R. A. Davies, T. D. Ruddy, et al. What is the Prognostic Value of Myocardial Perfusion Imaging Using Rubidium-82 Positron Emission Tomography? J. Am. Coll. Cardiol., September 5, 2006; 48(5): 1029 - 1039. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Hachamovitch, S. W. Hayes, J. D. Friedman, I. Cohen, and D. S. Berman A prognostic score for prediction of cardiac mortality risk after adenosine stress myocardial perfusion scintigraphy J. Am. Coll. Cardiol., March 1, 2005; 45(5): 722 - 729. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Hachamovitch and M. F. Di Carli Are Human Readers Needed for Prognostication from Stress Myocardial Perfusion SPECT? Using Outcomes Research to Validate Medical Technology J. Nucl. Med., February 1, 2005; 46(2): 194 - 197. [Full Text] [PDF]
|
|
|
|
|
|
W. D. Leslie, S. A. Tully, M. S. Yogendran, L. M. Ward, K. A. Nour, and C. J. Metge Prognostic Value of Automated Quantification of 99mTc-Sestamibi Myocardial Perfusion Imaging J. Nucl. Med., February 1, 2005; 46(2): 204 - 211. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Aboul-Enein, S. Kar, S. W. Hayes, M. Sciammarella, A. Abidov, R. Makkar, J. D. Friedman, N. Eigler, and D. S. Berman Influence of Angiographic Collateral Circulation on Myocardial Perfusion in Patients with Chronic Total Occlusion of a Single Coronary Artery and No Prior Myocardial Infarction J. Nucl. Med., June 1, 2004; 45(6): 950 - 955. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Hachamovitch, S. W. Hayes, J. D. Friedman, I. Cohen, and D. S. Berman Stress myocardial perfusion single-photon emission computed tomography is clinically effective and cost effective in risk stratification of patients with a high likelihood of coronary artery disease (CAD) but no known CAD J. Am. Coll. Cardiol., January 21, 2004; 43(2): 200 - 208. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Hachamovitch, S. W. Hayes, J. D. Friedman, I. Cohen, X. Kang, G. Germano, and D. S. Berman Is there a referral bias againstcatheterization of patients withreduced left ventricular ejection fraction?: Influence of ejection fraction and inducible ischemia onpost-single-photon emission computed tomographymanagement of patients without a history of coronary artery disease J. Am. Coll. Cardiol., October 1, 2003; 42(7): 1286 - 1294. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Hachamovitch, S. W. Hayes, J. D. Friedman, I. Cohen, and D. S. Berman Comparison of the Short-Term Survival Benefit Associated With Revascularization Compared With Medical Therapy in Patients With No Prior Coronary Artery Disease Undergoing Stress Myocardial Perfusion Single Photon Emission Computed Tomography Circulation, June 17, 2003; 107(23): 2900 - 2907. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Hachamovitch, S. Hayes, J. D. Friedman, I. Cohen, L. J. Shaw, G. Germano, and D. S. Berman Determinants of risk and its temporal variation in patients with normal stress myocardial perfusion scans: What is the warranty period of a normal scan? J. Am. Coll. Cardiol., April 16, 2003; 41(8): 1329 - 1340. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. R. Corbett and E. P. Ficaro Gated SPECT and the Visual Gold Standard: Gold Standard or Not? J. Nucl. Med., November 1, 2001; 42(11): 1639 - 1642. [Full Text] [PDF]
|
|
|
|
|
|
M. J. Zellweger, H. C. Lewin, S. Lai, E. A. Dubois, J. D. Friedman, G. Germano, X. Kang, T. Sharir, and D. S. Berman When to stress patients after coronary artery bypass surgery?: Risk stratification in patients early and late post-CABG using stress myocardial perfusion SPECT: implications of apppropriate clinical strategies J. Am. Coll. Cardiol., January 1, 2001; 37(1): 144 - 152. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Vanzetto, O. Ormezzano, D. Fagret, M. Comet, B. Denis, and J. Machecourt Long-Term Additive Prognostic Value of Thallium-201 Myocardial Perfusion Imaging Over Clinical and Exercise Stress Test in Low to Intermediate Risk Patients : Study in 1137 Patients With 6-Year Follow-Up Circulation, October 5, 1999; 100(14): 1521 - 1527. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
