Exercise Myocardial Perfusion SPECT in Patients Without Known Coronary Artery Disease
Incremental Prognostic Value and Use in Risk Stratification
Background We evaluated the incremental prognostic value, the role in risk stratification, and the impact on patient management of myocardial perfusion single-photon emission computed tomography (SPECT) in a population of patients without prior myocardial infarction, catheterization, or revascularization.
Methods and Results We examined 2200 consecutive patients who at the time of their dual-isotope SPECT had not undergone catheterization, coronary artery bypass surgery, or percutaneous transluminal coronary angioplasty and had no known history of previous myocardial infarction. Follow-up was performed at a mean of 566±142 days (97% complete) for hard events (cardiac death and myocardial infarction) and for referral to cardiac catheterization or revascularization within 60 days after nuclear testing. Examination of clinical, exercise, and nuclear models by use of pre–exercise tolerance test (ETT), post-ETT, and nuclear information using a stepwise Cox proportional hazards model and receiver-operating characteristic curve analysis revealed that nuclear testing added incremental prognostic value after inclusion of the most predictive clinical and exercise variables (global χ2=12 for clinical variables; 31 for clinical+exercise variables; 169 for nuclear variables; gain in χ2, P<.0001 for all; receiver-operating characteristic areas: 0.66±0.04 for clinical, 0.73±0.04 for clinical+ exercise variables, 0.87±0.03 for nuclear variables, P=.03 for gain in area with exercise variables; P<.001 for increase with nuclear variables). Multiple logistic regression analysis revealed that scan information contributed 95% of the information regarding referral to catheterization with further additional information provided by presenting symptoms and exercise-induced ischemia. Referral rates to early catheterization and revascularization paralleled the hard event rates in all scan categories−very low referral rates in patients with normal scans and significant increases in referral rates as a function of worsening scan results. Even after stratification by clinical and exercise variables such as the Duke treadmill score, pre- and post-ETT likelihood of coronary artery disease, presenting symptoms, sex, and age, the nuclear scan results further risk-stratified the patient subgroups, thus demonstrating clinical incremental value.
Conclusions In a patient population with no evidence of previous coronary artery disease at overall low risk (1.8% hard event rate), myocardial perfusion SPECT adds incremental prognostic information and risk-stratifies patients even after clinical and exercise information is known. It appears that referring physicians use this test in an appropriate manner in selecting patients to be referred to catheterization or revascularization.
Coronary artery disease is the leading cause of morbidity and mortality in industrialized societies.1 Because effective treatments can reduce the likelihood of subsequent adverse events, it is important to identify individuals at high risk of future cardiac events in whom medical therapy or revascularization may enhance survival.1
Although the incremental prognostic value of nuclear testing has been demonstrated in a number of studies, these study cohorts have consisted of patients both with and without prior myocardial infarction, revascularization, or catheterization.2 3 4 5 6 7 8 9 In those few studies evaluating the incremental prognostic value of nuclear testing in patients with “suspected coronary disease,” the study cohorts included some patients with prior catheterization documenting coronary disease2 or have been limited to those patients who have undergone both nuclear testing and subsequent catheterization.3 4 5 Although some previous studies have looked at purely diagnostic patient populations, they did not assess the incremental value of nuclear testing over prior information.10 11 To the best of our knowledge, the incremental prognostic value of myocardial perfusion scintigraphy in a homogeneous patient cohort without previously documented CAD has not been examined.
The goals of this study were to define the statistical incremental prognostic value of exercise stress myocardial perfusion SPECT in a patient population without previously defined CAD and to define the clinical role of the test in risk stratification of this population. Since effective risk stratification per se does not necessarily imply a subsequent change in patient management, the utility of nuclear testing, as measured by its impact on subsequent patient referral, takes on an important role. Thus, this study also sought to determine the impact of this test on patient management as measured by the post–nuclear testing referral of patients to catheterization and revascularization.
We identified 2268 consecutive patients who underwent exercise dual-isotope SPECT between January 1, 1991, and December 1, 1993, who at the time of their nuclear tests had not undergone cardiac catheterization, coronary artery bypass surgery, or percutaneous transluminal coronary angioplasty and had no known history of previous myocardial infarction. Patients who underwent pharmacological dual-isotope SPECT or who were known to have valvular heart disease or primary cardiomyopathy were not included in this study. Of the initial population, 68 patients were lost to follow-up, and the remaining 2200 patients with successful follow-up (97%) were included in this study. At the time of nuclear study, all patients completed a questionnaire detailing the characteristics of their presenting symptomology. All patients were then interviewed by a physician who reviewed their responses prior to the initiation of the treadmill portion of their tests.
We have shown previously that referral to revascularization in the first 60 days after nuclear testing tends to be based on the results of the scan, and referral to revascularization more than 60 days after nuclear testing tends to be based on worsening of the patient’s clinical status.10 12 For this reason, the 87 patients in our cohort who were revascularized in the first 60 days after nuclear testing were censored from the prognostic portion of the analyses. Thus, the prognostic data presented here are based on a subset of 2113 patients.
Exercise Myocardial Perfusion Protocol
All patients underwent exercise dual-isotope myocardial perfusion imaging as previously described.13 Whenever possible, β-blockers and calcium channel antagonists were discontinued 48 hours before testing, and nitrate compounds were discontinued at least 6 hours before testing; 105 (4.8%) patients were under the influence of β-blockers and 180 (8.2%) were under the influence of calcium channel blockers at the time of their tests. Thallium-201 (2.5 to 3.5 mCi) was injected intravenously at rest, with dose variation based on patient weight. Rest thallium-201 SPECT imaging was initiated 10 minutes after injection of the isotope. Immediately after imaging, all patients performed a symptom-limited treadmill exercise test using standard protocols with 12-lead 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. 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.
The ECG response to exercise was categorized as either nonischemic (no significant ECG changes), ischemic (significant ST-segment elevation or depression), equivocal (borderline ECG changes), or nondiagnostic (ECG uninterpretable because of digoxin use, paced rhythm, bundle branch block, and so forth). The clinical response to exercise was also assessed as either nonischemic, ischemic (typical angina pectoris or anginal equivalent during exercise), equivocal, or abnormal (exertional hypotension or inappropriate shortness of breath).
At near-maximal exercise, a 20- to 30-mCi dose of technetium-99m sestamibi was injected (actual dose varied with patient weight), and exercise continued for 1 additional minute after injection. Technetium-99m sestamibi SPECT imaging was begun 30 minutes after isotope injection.
SPECT Acquisition Protocol
SPECT imaging was performed as previously described.13 The SPECT studies were performed using a circular 180° acquisition for 64 projections at 20 seconds per projection. For thallium-201 imaging, two energy windows were used, including a 30% window centered on the 68- to 80-keV peak and a 10% window centered on the 167 keV peak. For technetium-99m sestamibi SPECT, a 15% window centered on the 140-keV peak was used. Images were acquired using a 64×64 image matrix and were subject to quality control measures as previously described.13
A semiquantitative visual interpretation was performed using short-axis and vertical long-axis myocardial tomograms that were divided into 20 segments for each study (Fig 1⇓). These segments were assigned on six evenly spaced regions in the apical, midventricular, and basal slices of the short-axis views and two apical segments on the midventricular long-axis slice.14 Each segment was scored by consensus of two experienced observers using a 5-point scoring system (0=normal, 1=equivocal, 2=moderate, 3=severe reduction of radioisotope uptake, and 4=absence of detectable tracer uptake in a segment).
We defined three nuclear variables using the above-described 20-segment, 5-point scoring system. A summed stress score (SSS) was obtained by adding the scores of the 20 segments of the stress sestamibi images. Summed stress scores less than 4 were considered normal, between 4 and 8 mildly abnormal, and greater than 8 severely abnormal. A summed rest score (SRS) was obtained by similarly adding the scores of the 20 segments of the rest thallium images (Fig 1⇑). The sum of the differences between each of the 20 segments on the stress and rest images was defined as the summed difference score (SDS). Each of these variables incorporate both the extent and severity of perfusion defects, both of which independently add prognostic information.12 Scans with defects in more than one region associated with a particular coronary artery were considered to demonstrate multivessel disease.15 Scans that appeared to have left ventricular dilation on stress images but not on rest images were considered to have transient ischemic dilation.
Patient follow-up was performed by scripted telephone interview by individuals blinded to the patient’s test results. Events were defined as either cardiac death (confirmed by review of death certificate and hospital chart or physician’s records) or nonfatal myocardial infarction (documented by appropriate cardiac enzyme and ECG changes). When interventions (cardiac catheterization, coronary artery bypass surgery, or percutaneous transluminal coronary angioplasty) were identified, these outcomes were confirmed by hospital records or the physician’s office records. All patients included in this report were followed for at least 1 year. The mean follow-up interval was 566±142 days.
Likelihood of Coronary Artery Disease
For purposes of analyzing patients in different risk subsets, we used analysis of the pre- and post-ETT likelihood of CAD as aggregate descriptors of proven prognostic importance based on bayesian analysis of age and calculated with the use of CADENZA.16 The variables included in the calculations of pre- and post-ETT likelihoods of CAD are listed in Table 1⇓.
Comparisons between patient groups were performed using a one-way ANOVA (with a Bonferroni correction as appropriate) for continuous variables and a χ2 test for categorical variables. All continuous variables are described as mean±SD. A value of P<.05 was considered statistically significant.
To determine the incremental prognostic value of a test, it is necessary to include all other information known regarding the patient prior to that time. With this in mind, the Cox proportional hazards model (BMDP version 7, program 2L)17 18 was used in a stepwise fashion to determine four distinct statistical models: (1) a clinical model predictive of events, (2) a clinical and exercise model predictive of events, (3) a nuclear model predictive of events, and (4) a model to determine the increase in prognostic information after adding the most predictive nuclear variable(s) (model 3) to a model that “forces in” the best clinical and exercise variables (model 2). The dependent variable in the Cox proportional hazards analysis is the time to an event rather than the occurrence of the event within a determined time period. The threshold for entry of variables into all models was P<.05. A statistically significant increase in the global χ2 of the model (as determined by Cox proportional hazards testing) after the addition of the nuclear variables defined incremental prognostic value. The value of χ2 obtained is proportional to the information content of the model examined.
In light of the limitation of 39 outcome events of interest, we avoided underpowered Cox proportional hazards models19 by using validated aggregate variables rather than deriving optimal models based on our raw data. The variables entered into the models consisted of the pre-ETT likelihood of CAD in the clinical model and the post-ETT likelihood of CAD and the Duke treadmill score in the exercise model. In the absence of generally accepted nuclear aggregate variables, we entered the raw data of the derived nuclear variables. These included the SSS, the SDS, and the presence of multivessel abnormalities on the scan. For each model, the global χ2 of the model and the proportion of the information content of the model contained in the leading variable were both expressed.
Incremental value was also determined by using the same variables tested in the Cox model to construct logistic models for clinical, clinical+exercise, and clinical+exercise+nuclear variables. These logistic regression models were used to calculate the probability of an event for each patient using each model, and ROC curves were constructed for each of these models. These curves were analyzed by comparing the area underneath them, a measure that reflects the discriminatory power of the test in question independent of factors such as diagnostic threshold, the baseline event rate in the study sample, or selection bias.20 21 22 23 ROC curves have a potential area that has values ranging from 0 to 1; an area of 0.5 corresponds to no discriminatory power, while an area of 1 defines perfect discrimination. ROC areas were expressed as the area±SEM.
Predictors of Post–Nuclear Referral to Catheterization or Revascularization
To determine the most powerful predictors of referral to catheterization within 60 days after nuclear testing (early catheterization) and referral to revascularization within this same time span (early revascularization), we performed multiple logistic regression (BMDP version 7, program LR)18 using the uncensored patient cohort (n=2200). The temporal restriction of 60 days was placed to limit the interventions studied (end points of interest) to those most likely related to the results of the index noninvasive study. The number of variables entered into the regression model were limited to 1 per 10 events of interest to prevent overfitting of the model.19 Global χ2 for each final model and the percent contribution of each covariate to the global χ2 were determined.
Initial Patient Population
The 2113 patients included in this study are characterized in Table 2⇓. Patients who had events on follow-up had greater pre-ETT likelihood of CAD, more frequent uninterpretable ECG, and more commonly had cardiac risk factors when compared with patients who had no events on follow-up. The presenting symptoms and pre-ETT likelihood of CAD in our population are shown in Table 3⇓. Of those patients with atypical and typical angina at the time of presentation, virtually all had intermediate or high pre-ETT likelihood of CAD. Almost half the patients with nonanginal chest discomfort and the majority of patients with no symptoms at the time of presentation had a low pre-ETT likelihood of CAD. Of these 930 patients with low pre-ETT likelihood of CAD, 320 had more than one cardiac risk factor, 171 had abnormal rest ECGs, 378 were greater than 60 years of age, and 549 had a positive ETT, as demonstrated by an increase in likelihood of CAD after consideration of exercise variables, that is, post-ETT greater than pre-ETT likelihood of CAD.
Among the 2113 patients included in this study, 39 hard events occurred. These included 13 cardiac deaths and 26 nonfatal myocardial infarctions (1.8% hard event rate). In the overall uncensored population of 2200 patients, 185 catheterizations and 87 revascularizations (44 percutaneous transluminal coronary angioplasties, 43 coronary artery bypass grafts) occurred in the first 60 days after nuclear testing. This represents an early catheterization rate of 8.4% and an early revascularization rate of 4.0%.
Descriptive patient characteristics, exercise, and nuclear variables in patients with and without events on follow-up are presented in Tables 2, 4, and 5. In general, patients who had events had lowered exercise tolerance, more frequent abnormal ECG responses during exercise, greater post-ETT likelihood of CAD, and lower Duke treadmill scores. With respect to nuclear variables, patients who had events had more severe, extensive, and frequent scan abnormalities.
The results of the Cox analysis are shown in Table 6⇓. Significant increases in global χ2 occurred both with the addition of the exercise variables, yielding the clinical+ exercise model, as well as with the addition of nuclear variables after forcing in the clinical+exercise model. There was almost a twofold increase in information content with the addition of exercise variables and a fivefold gain in incremental information with the addition of nuclear information (Table 6⇓).
Incremental Value: ROC Curve Analysis
The area under the ROC curve for clinical variables alone was 0.66±0.04. This area increased significantly with the addition of exercise information (area for clinical+ exercise variables=0.73±0.04; P=.03). Forcing in the clinical and exercise information and adding the nuclear variable (clinical+exercise+nuclear model) resulted in a further significant gain in area and, thus, information (area for final model=0.87±0.03). The log odds of the areas for these three models were 0.6, 1.0, and 1.9, suggesting a doubling of prognostic information with the addition of information at each step.
Logistic Analysis of Predictors of Intervention
The logistic model evaluating referral to catheterization was powerful (overall χ2=511), with 95% of the information provided by the SDS and further small gains added by presenting symptoms. We also developed similar logistic models that separately evaluated predictors of early catheterization in patients with normal scans and in patients with abnormal scans. The subgroup analysis focusing on referral to early catheterization after a normal scan revealed post-ETT likelihood of CAD to be the best predictor of referral. This suggests that in the setting of a low-risk scan, clinical and exercise variables as well as uncertainty regarding the scan results contributed to the referral to catheterization. In patients with abnormal scans, the subgroup analysis revealed the extent and severity of reversible defects present on the scan (as measured by the SDS) to be the best predictor of referral to early catheterization, with further information added by the presence of anginal symptoms. Thus, the predominant information for the referral to catheterization in the overall cohort, as well as higher-risk patients (abnormal scans), was provided by the results of nuclear testing. Only in those patients with low-risk scans did clinical information play the major role in the referral. See Table 7⇓.
Nuclear Threshold for Abnormality Versus Proportion of Hard Events Detected
Fig 3⇓ demonstrates the proportion of hard events detected and the proportion of the total patient population examined with abnormal scans as a function of the SSS threshold defining a normal scan. The number of hard events missed (hard events occurring in patients with normal scans) would be similar at all SSS thresholds less than 5, suggesting that no gain in patient benefit (increased detection of high-risk patients) would be achieved by altering this threshold. Increases in the value of SSS used as a threshold would, however, result in decreases in the number of patients with abnormal scans and fewer subsequent referrals to catheterization, at the expense of increasing numbers of hard events not detected (decreased patient benefit).
Events and Subsequent Management Versus Extent and Severity of Hypoperfusion
The frequency of hard events, early catheterization, and revascularization as a function of the scan result is shown in Table 8⇓. Patients with normal scans had an exceedingly low hard event rate over the follow-up period (0.3%). The hard event rate increased significantly with worsening scan findings. Referral rates to early catheterization and revascularization paralleled the hard event rates in all scan categories– very low referral rates in patients with normal scans and significant increases in referral rates as a function of worsening scan results.
The rate of referral to early catheterization was low in normal, probably normal, and equivocal scans but increased dramatically for abnormal and probably abnormal scans. On the other hand, the rate of referral to late catheterization also increased significantly but far less dramatically than for early catheterization. Thus, even though we only assessed early catheterization rates in our population, the vast majority of catheterizations are included in this analysis (>80% of all catheterization).
Events and Subsequent Management Versus Clinical and ETT Factors and Scan Results
We compared the rate of hard events as a function of scan result and the Duke treadmill score,24 a widely used aggregate prognostic index of exercise variables (Fig 2A⇓). As expected, a significant increase in hard event rate was noted with increasing Duke treadmill score. Importantly, in the low and intermediate Duke score groups, the patients were further stratified by the scan result into very low risk groups for normal scans and an increasing event rate in mild and severely abnormal scans. This finding reveals the incremental clinical information yielded by nuclear testing. Of interest, significant stratification was achieved in the low-risk Duke group despite its low (0.9%) overall event rate. From a clinical standpoint, however, the most pronounced stratification occurred in the intermediate Duke score group that comprised 55% of our population (1187 patients). The relatively high event rate in patients with normal scans in the high-risk Duke score may be attributable to the small patient size for that group (one event in 28 patients).
Fig 2B⇑ demonstrates the catheterization rate as a function of the same classification. The rates of referral to catheterization and revascularization paralleled the hard event rates as a function of the Duke score category and scan result. As noted with hard events, patients with normal scans were very infrequently referred to catheterization in all Duke treadmill score categories. This referral rate increased significantly as a function of scan result. Similar findings were noted in our cohort when pre- or post-ETT likelihood of coronary disease was used in place of the Duke treadmill score in this stratification analysis (Tables 9⇓ and 10⇓). Patients with normal scans had low event rates and low referral rates to catheterization regardless of pre- and post-ETT likelihood of CAD category. In patients with mildly and severely abnormal scans, there were significant increases in hard event rates as a function of worsening scan results but not as a function of pre- or post-ETT likelihood of CAD. There were, however, significant increases in referral rates to catheterization and revascularization as a function of both scan result and pre- and post-ETT likelihood of CAD.
Although the results of the nuclear scan risk-stratified patients with both anginal and nonanginal presentations (Table 9⇑), there was a trend toward higher event rates in the patients with anginal symptoms (P=NS). Interestingly, patients with anginal presentations were referred to catheterization twice as often as those patients without symptoms in all scan categories (Table 9⇑). This difference was also noted in the rates of revascularization (Table 10⇑), as patients with mildly and severely abnormal scans were referred for revascularization more frequently when the patient presented with anginal symptoms. Thus, while anginal symptoms did not confer added risk, they did play a secondary role in influencing patient management.
Events and Subsequent Management Versus Sex and Age
In both men and women the hard event rate increased significantly as a function of SSS (Table 9⇑). Importantly, this increase was more dramatic in the women compared with the men, resulting in significantly greater hard event rates in the women with severely abnormal scans when compared with the men (P<.01). Interestingly, the rate of referral to catheterization and revascularization (Table 10⇑) was virtually identical in men and women–very low referral rates in normal scans, significantly greater rates increasing as a function of SSS. Thus, while no sex-related referral bias was present, relative undercatheterization was present in women relative to their risk in the setting of a severely abnormal scan.
With respect to patient age, there were similar increases in event rates as a function of the results of nuclear scan, with no differences between three age groups within scan categories (Table 9⇑). Similarly, the rate of referral to catheterization and revascularization paralleled the scan findings, with no differences between age groups (Table 9⇑). Thus, the age-related effect in this study was that of increasing prevalence of abnormal scans rather than a change in risk associated with a particular scan result.
The intent of the current study was twofold: to determine the incremental prognostic value of exercise sestamibi myocardial perfusion SPECT in a population of patients without known CAD and to evaluate the impact on patient management of this test as measured by the rates of referral to catheterization and revascularization early (within 60 days) after nuclear testing. Regarding prognosis, we found that even after the most predictive clinical and exercise variables were forced into the Cox model, the addition of nuclear information provided statistical incremental prognostic value−more than a fivefold increase in prognostic information. The ROC curve also demonstrated a similar incremental value for nuclear variables over clinical and exercise variables. Further, the results of the nuclear scan significantly stratified the patient population examined; those patients with normal scans had a very low event rate (<1%), and those with abnormal scans had increasing event rates with worsening scan result. Even after stratification by clinical and exercise characteristics, such as Duke treadmill score, pre- and post-ETT likelihood of CAD, presenting symptoms, sex, and age, the nuclear scan results further stratified the patient subgroups.
With respect to patient management, referral rates to catheterization and revascularization paralleled patient risk as a function of nuclear variables both in the overall cohort as well as after patient subgrouping by clinical and exercise variables as described above. These results indicate that physicians appropriately used the nuclear scan results−the best predictors of outcome−in referring patients to subsequent intervention. Clinical characteristics, such as anginal symptoms and pre- and post-ETT likelihood of CAD, also appropriately modified referral patterns. Thus, it would appear that in patients without previously known CAD, sestamibi myocardial perfusion SPECT both adds incremental prognostic value to clinical and exercise variables and is used in an appropriate manner by referring physicians.
Comparison to Previous Studies
A number of previous studies have evaluated the prognostic implications of exercise thallium myocardial perfusion scintigraphy in patients with suspected CAD.2 3 4 5 10 11 The work of Staniloff et al10 in a purely diagnostic population, one of the earliest studies on prognostic value of thallium imaging, revealed the potential prognostic power of thallium imaging but did not assess it in an incremental fashion. The findings of the current study agree with those of the previous work by Ladenheim et al2 from our laboratory, which was the first to assess the incremental prognostic value of nuclear testing. A greater increment in information from the nuclear test was shown in the present study compared with the Ladenheim study. This may be due to differences in the patient population used (the Ladenheim study included patients with previous catheterization), SPECT versus planar imaging, or sestamibi versus thallium as the isotope utilized. Three important studies were reported from Kaul and colleagues3 4 5 at the University of Virginia in patients with “suspected CAD”; however, all three of these studies included patients with known prior myocardial infarction (43% of the combined populations3 4 ). The low event rates in the normal scan group confirms the observations from previous studies.7 8 25 26
Nuclear Scan Low-Risk Threshold
As shown in Fig 3⇓, the threshold we have previously used to define normal scans is appropriate for prognostic purposes. The proportion of hard events missed using this threshold is not affected by shifting to a lower SSS value, and more hard events would have been missed using a higher SSS threshold.
Risk Stratification by Nuclear Testing
Our study goals were to evaluate nuclear testing in patients in whom the coronary anatomy was not known and, importantly, to determine the prognostic value and stratification utility of nuclear testing prior to referral to catheterization. The results of our study could be used to guide the decision process after nuclear testing with respect to whether to refer to intervention. Our results indicate that the majority of diagnostic patients referred to our laboratory, the 1624 patients with normal scans (76% of cohort, 0.4% event rate), can be safely managed without need for intervention. It can be further argued that many of these patients did not need nuclear testing and could have been risk-stratified by clinical and exercise variables alone. Thus, the 1068 patients with low post-ETT likelihood of CAD or the 926 patients with low-risk Duke treadmill score could have been managed without nuclear testing (Fig 2A⇑ and Table 8⇑). The exclusion of these patients, however, does not alter the successful stratification of the remaining patients by nuclear scan results.
The ability of the nuclear scan to further risk-stratify patients after initial clinical stratification is key to understanding its potential clinical role. Although the scan results stratified patients in the low-risk Duke treadmill score group, the cost-effectiveness of this stratification is doubtful, since few high-risk patients would be identified by the scan. On the other hand, nuclear testing in intermediate-risk Duke groups would result in considerable potential savings, since the overall group risk is intermediate, yet more than two thirds of these patients had normal scans, thus not requiring any further evaluation (such as catheterization). Importantly, this group comprised 55% of our study population. This noninvasive approach may also work well even in patients with high-risk Duke scores, but this approach will require further study in larger populations.
Impact of Nuclear Testing on Patient Management
An important result of the current study is the striking parallel between event rates and referral to catheterization after nuclear testing. Patients with normal and equivocal scans had uniformly low referral rates to catheterization. The results of this study suggest that referral to catheterization was based on appropriate criteria and occurred at an acceptable rate relative to risk or symptoms in patients with both normal and abnormal scans. These results are similar to those found in a general population referred to nuclear testing.27
The scintigraphic studies used in the current work were assessed by experienced observers using a standardized, semiquantitative approach to visual interpretation that we have developed14 and have documented to be highly reproducible.13 Consequently, the reliance on the expertise of the observer limits the extrapolation of our results to those of other centers. Objective quantitative methods for analysis of technetium-99m myocardial perfusion SPECT studies have been developed28 that correlate highly with both visual scan assessment and coronary angiography.29 At the time of collection of the SPECT studies in this patient population, we did not have a quantitative analysis technique in operation on all of our camera/computer systems. The visual interpretation methods that were used form the basis for the quantitative analysis programs developed by Cedars-Sinai Medical Center and Emory University.28 Therefore, the results of semiquantitative analysis in this study should correlate strongly with those of quantitative analysis. Further prognostic studies using quantitative analysis would be of interest. Finally, since our patient population had infrequent fixed defects, the preponderance of the information was derived from the stress scan performed with sestamibi. We believe that the results of this study are applicable to any of the high-dose stress sestamibi study protocols.30
Statistical and Clinical
The patients in this study are those referred to a university-affiliated community hospital in a major urban area, and the results of this study should be applicable to this setting. With respect to the statistical analysis, the use of multivariate models is limited by the number of events accumulated during the follow-up period. We limited the number of variables entered into these models to 1 per 10 outcome events to prevent overfitting of the model and thus enhance its accuracy.19 The low loss to follow-up rate, the large patient group used, and the adequate number of events favor the likelihood that our multivariate results are accurate.19 Since we were limited to using four variables in the Cox proportional hazards analysis, we chose to use previously derived and validated aggregate variables (likelihood of CAD and Duke treadmill score) rather than to derive models consisting of variables specific to our data set. The disadvantage to this approach was that the nonnuclear models were not optimized for our population while the nuclear model was (since no accepted nuclear aggregate variable is accepted). We also performed an analysis using models specific to our population and found similar results; thus, the use of optimized nuclear variables and generalized nonnuclear variables did not significantly alter our results.
The results of this study reveal that exercise sestamibi myocardial perfusion SPECT adds incremental prognostic information when used in patients who have not undergone previous catheterization or revascularization and have not had previous myocardial infarction and who are at overall low-intermediate risk (1.8% hard event rate, 1.2% per year of follow-up). Further, physicians referred patients to catheterization and revascularization in proportion to the extent and severity of their scan results and, thus, to their risk of cardiac events. In light of this, the effect of testing on patient management appears to be both powerful and appropriate.
Selected Abbreviations and Acronyms
|CAD||=||coronary artery disease|
|ETT||=||exercise tolerance test|
|SPECT||=||single-photon emission computed tomography|
This work was supported in part by a grant from Dupont-Pharma.
Presented in part at the 41st annual meeting of the Society of Nuclear Medicine, Orlando, Fla, June 1994.
- Received August 21, 1995.
- Revision received September 28, 1995.
- Accepted October 4, 1995.
- Copyright © 1996 by American Heart Association
Rutherford JD, Braunwald E. Chronic ischemic heart disease. In: Braunwald E, ed. Heart Disease. 1992:1293-1364.
Kaul S, Lilly DR, Gascho JA, Watson DD, Gibson RS, Oliner CA, Ryan JM, Beller GA. Prognostic utility of the exercise thallium-201 test in ambulatory patients with chest pain: comparison with cardiac catheterization. Circulation. 1988;77:745-758.
Pollock SG, Abbott RD, Boucher CA, Beller GA, Kaul S. Independent and incremental prognostic value of tests performed in hierarchical order to evaluate patients with suspected coronary artery disease. Circulation. 1992;85:237-248.
Strattman HG, Williams GA, Wittry MD, Chaitman BR, Miller DD. Exercise technetium 99-m sestamibi tomography for cardiac risk stratification of patients with stable chest pain. Circulation. 1994;89:615-622.
Iskandrian AS, Johnson J, Le TT, Wasserleben V, Cave V, Heo J. Comparison of the treadmill score and single-photon emission computed tomographic thallium imaging in risk assessment. J Nucl Cardiol. 1995;1:144-149.
Staniloff HM, Forrester JS, Berman DS, Swan HJC. Prediction of death, myocardial infarction, and worsening chest pain using thallium scintigraphy and exercise electrocardiography. J Nucl Med. 1986;27:1842-1848.
Berman DS, Kiat H, Friedman JD, Wang FP, Van Train K, Matzer L, Maddahi J, Germano G. Separate acquisition rest thallium-201/stress technetium 99m sestamibi dual-isotope myocardial perfusion single-photon emission computed tomography: a clinical validation study. J Am Coll Cardiol. 1993;22:1455-1464.
Berman DS, Kiat H, Van Train K, Garcia E, Friedman J, Maddahi J. Technetium-99m sestamibi in the assessment of chronic coronary artery disease. Semin Nucl Med.. 1991;3:190-212.
Palmas W, Friedman JD, Diamond GA, Silber H, Kiat H, Berman DS. Incremental value of simultaneous assessment of myocardial function and perfusion with Tc-99m sestamibi for the prediction of coronary artery disease extent. J Am Coll Cardiol. 1995;1024-1031.
Cox DR. Regression models and life-tables. J R Stat Soc (B). 1972;34:187-202.
Dixon WJ, ed. BMDP Statistical Software Manual. Berkeley, Calif: University of California Press; 1992.
Diamond GA. Future imperfect: the limitations of clinical prediction models and the limits of clinical prediction. J Am Coll Cardiol. 1989;14:12A-22A.
Diamond GA. ROC steady: a receiver operator characteristic curve that is invariate relative to selection bias. Med Decis Making. 1987;7:238-243.
McNeil BJ, Hanley JA. Statistical approaches to the analysis of receiver operator characteristic (ROC) curves. Med Decis Making. 1984;4:137-150.
Diamond GA. Affirmative actions: can the discriminant accuracy of a test be determined in the face of selection bias? Med Decis Making. 1991;11:48-56.
Brown KA. Prognostic value of thallium-201 myocardial perfusion imaging: a diagnostic tool comes of age. Circulation. 1991;83:363-381.
Van Train K, Garcia EV, Maddahi J, Areeda J, Cooke CD, Kiat H, Silagan G, Folks R, Friedman J, Matzer L, Germano G, Bateman T, Ziffer J, DePuey EG, Fink-Bennett D, Cloninger K, Berman DS. Multicenter trial validation for quantitative analysis of same-day rest-stress technetium 99m-sestamibi myocardial tomograms. J Nucl Med. 1994;35:609-618.
Berman DS, Kiat HS, Van Train KF, Germano G, Maddahi J, Friedman JD. Myocardial perfusion imaging with technetium-99-m-sestamibi: comparative analysis of available imaging protocols. J Nucl Med. 1994;35:681-688.