Quantitative coronary angiography was performed in case a coronary lesion was defined by visual analysis. Analysis was performed using a computer-assisted, automated, edge detection system (Philips Easy Vision). The person analyzing the angiograms was blinded to clinical and MPS data. The guiding catheter was used as a scaling device. The minimal lumen diameter and the proximal and distal reference vessel diameters were determined to allow calculation of the percent diameter stenosis. Significant coronary artery stenosis was excluded if (1) by visual analysis no coronary lesion could be identified or (2) quantitative coronary angiography on coronary lesions identified by visual analysis had less than 50% diameter stenosis.

Cineventriculography performed in 30° RAO projection with injection of 30 cc of contrast agent and acquisition of 30 frames/sec was used to evaluate left ventricular ejection fraction. Analysis of ejection fraction was performed using computer-assisted, automated, edge detection system (Philips Easy Vision).

Stress and rest studies were done in a one-day (1 × 300 MBq followed by 1 × 750 MBq) or a two-day (2 × 450 MBq) protocol using ^99mTc-Tetrofosmin or MIBI. A total of 67 patients were stressed by bicycle exercise stress and 51 patients with pharmacological stress using dipyridamole. In case of bicycle exercise stress, ^99mTc-Tetrofosmin or MIBI were injected during maximal or symptom-limited exercise with the workload increased by 25 W every 2 minutes and at rest. In case of dipyridamole stress, ^99mTc-Tetrofosmin or MIBI were injected 3 minutes after completion of a 4-minute dipyridamole infusion (0.56 mg dipyridamole/kg body weight) and at rest. The SPECT protocol usually started with the stress part, followed by SPECT at rest either the same day (one-day protocol; N = 54) or a few days later (two-day protocol; N = 64). Antianginal drugs were stopped on the day of the stress test while β-blockers were discontinued 2 days before the test. In case of dipyridamole stress, patients were instructed not to consume any food or beverages containing caffeine before the stress test. Acquisition was done in a 64 × 64 matrix on a Siemens MULTISPECT 3 (triple-head gamma camera; Siemens Gammasonics Inc., Hoffman Estates, IL) 60 minutes after tracer injection, with 60 views using a zoom factor of 1.23. All data sets were reconstructed from 360° data using filtered back projection (third-order Butterworth filter, critical frequency 0.5 Nyquist). The data sets were transferred to a Siemens e.soft system, where they were reoriented according to the cardiac long axes, yielding oblique data sets.

Using an automatic quantitative analysis package of perfusion SPECT images (Cedars Sinai QPS 2008, Cedars-Sinai Medical Center, Los Angeles, CA), the stress and rest data sets of the 118 patients were analyzed in fully automatic operation mode. The summed stress score (SSS), summed rest score (SRS), and summed difference score (SDS) values were calculated based on comparison with the corresponding gender-specific institutional normal database. The left ventricular myocardium was divided into 20 segments2,7 and scored for extent and severity of perfusion abnormalities in each segment during stress and rest according to a 5-point scale: 0: normal, 1: slight reduction, 2: moderate reduction, 3: severe reduction of radiotracer uptake, and 4: no radiotracer uptake. The SSS and SRS are the sum of scores in these 20 segments.2,7,8 In case the segmental rest score had a higher value than the stress score it was assigned the stress score value. The sum of the differences between each of the 20 segments on the stress and rest images was defined as SDS, also called the reversibility score. It is an index of jeopardized myocardium. SSS and SRS were classified using the following classification system: <4: normal, 4-8: mildly abnormal, 9-13: moderately abnormal, and ≥14: severely abnormal.2,9,10

Follow-up information on patient survival was obtained in all patients by dedicated research personnel. Data were obtained by telephone contact with the patient, with one of his or her immediate relatives and complemented by information obtained by the patient’s general physician or patient charts from recurrent hospital admissions. In case neither patient nor relatives could be contacted and patient’s general physician did not know about the patient’s outcome, the local population registries were contacted to obtain information about the patient’s possible death or current location. Patients were followed for at least 4.5 years (mean of 6.3 ± 1.2 years) for hard events defined as death, nonfatal myocardial infarction, or myocardial revascularization procedure. Myocardial infarction was verified by hospital document and the diagnosis was based on the accepted criteria of characteristic chest pain, electrocardiographic, and enzyme changes. In addition, a composite of major adverse events (MAE: death, nonfatal myocardial infarction, bypass surgery, or percutaneous coronary intervention) was evaluated. Furthermore, using a standard medical questionnaire, the level of current limitations due to angina pectoris considering the CCS (Canadian Cardiology Society) class and dyspnoe considering the NYHA (New York Heart Association) class were determined at the end of the follow-up period. Patients with symptoms determined as CCS class II-IV and/or NYHA class II-IV within the last 2 months prior to follow-up questionnaire were considered as symptomatic.

Values are expressed as mean ± SD or as percentages. For the comparisons of clinical and angiographic characteristics and the extent of SPECT perfusion defects, the Student’s t-test for unpaired samples and analysis of variance to evaluate differences between different groups was used as indicated. Proportions were compared by Chi-square test or with Fischer test when appropriate. Receiver operating characteristics (ROC) curves were constructed to define the optimal cut-off values of SRS and SSS for prediction of death or MAE during follow-up. These allowed definition of diagnostic accuracy of SRS and SSS for clinical events. The areas under the curve of the ROC curves were expressed as the area ± SD. Patient survival without adverse events was analyzed and Kaplan-Meier curves were constructed. Differences between survival curves were evaluated using the log-rank test. Cox proportional hazard regression analysis was used to determine univariate and multivariate predictors of death as well as MAE at follow-up. Age, gender, diabetes, stress level, SRS, SSS, and SDS were included in the analysis. A P-value < 0.05 was considered statistically significant.

There were 16 deaths, 9 myocardial infarctions, and 11 coronary revascularization procedures during follow-up. Thus, annual mortality rate during follow-up was 2.1%. The total number of patients with major adverse events was 29. A total of 64 patients had symptoms of dyspnea (NYHA II-IV) and/or chest pain (CCS II-IV) at follow-up. A total of 76 patients had either MAE or symptoms during follow-up.

By quantitative analysis of SPECT studies mean SRS was 3.1 ± 5.4 and mean SSS was 5.9 ± 5.9. There were 81 patients with a SRS of <4, 19 patients with a SRS of 4-8, and 18 patients with a SRS of >8. The SSS was <4 in 48 patients, 4-8 in 35 patients, 9-13 in 16 patients, and ≥14 in 19 patients. The SDS was <4 in 77 patients, 4-8 in 35 patients, and ≥8 in 6 patients.

Considering quantitative analysis of MPS, there was a significant difference between patients with and without death during follow-up in the SRS (6.4 ± 6.7 vs. 2.6 ± 5.0, P = 0.048) but no significant difference between patients with and without MAE in the SRS (5.4 ± 7.7 vs. 2.4 ± 6.27, P = 0.073). There were significant differences in SSS between patients who died and those alive at follow-up (9.5 ± 6.9 vs. 5.4 ± 5.6, P = 0.012). Similarly, there were significant differences in SSS between patients with and without MAE at follow-up 8.7 ± 7.2 vs. 5.2 ± 5.0, P = 0.010, respectively. Patients with any MAE or significant clinical symptoms during follow-up had also higher SSS than those without the respective event, 7.2 ± 7.1 vs. 4.6 ± 6.2, P = 0.042. The SDS was not significantly higher for patients who died (3.1 ± 2.4 vs. 2.6 ± 2.4), had MAE (2.9 ± 2.4 vs. 2.7 ± 2.4), or MAE or significant clinical symptoms (2.8 ± 2.5 vs. 2.7 ± 2.5) at follow-up compared to those without the respective event.

ROC analysis demonstrated for SRS an area under the ROC curve of 0.64 (95% CI: 0.53-0.72, cut-off SRS >6) for death during follow-up. For any MAE, ROC analysis demonstrated for SRS an area under the curve of 0.56 (95% CI: 0.47-0.66, cut-off SRS >6). ROC analysis demonstrated for SSS an area under the curve of 0.70 (95% CI: 0.60-0.79, cut-off SSS >13) for death during follow-up. For any MAE, ROC analysis demonstrated for SSS an area under the curve of 0.61 (95% CI: 0.51-0.69, cut-off SSS >13) (Figure 1). Annual mortality rates were 1.3% in patients with a SSS ≤13 and 4.5% in patients with a SSS >13. Figure 2 demonstrates the impact of SSS level on rate of survival (left panel) and rate of MAE-free survival (right panel) during the follow-up period. The prognosis of patients with a SSS level >13 was found to be significantly worse than in those with a lower SSS.

Figure 1 Receiver operator characteristics curves for death at follow-up considering SSS (left panel) and for MAE at follow-up considering SSS (right panel)

---

Figure 2 Kaplan-Meier curves for survival (left panel) and MAE-free survival (right panel) for patients with a SSS ≤13 (continuous line) and for patients with a SSS >13 (interrupted line)

---

A number of previous studies have evaluated the prognostic implications of stress thallium MPS in patients with suspected CAD. Several of these studies on patients with “suspected CAD” included patients with known prior myocardial infarction. Based only on their clinical history these patients represent a high-risk group. The total width of patient risks just considering the clinical factors was comparatively high in these studies. Consequently, annual event rates ranged by a factor of up to 10 between low-risk patients and high-risk patients of these studies.2,6,10-13 In particular, negative perfusion scintigraphy studies have been proven to be associated with very low event rates.14

In contrast, the patient population included in this study was very selective. All patients have had coronary angiography. Thus, clinical history implied a high likelihood of significant coronary artery disease. However, the insignificant coronary artery disease determined by angiography indicates that very high-risk patients were excluded from the analysis. Thus, patients included in this study had a smaller width of risks as determined by clinical and angiographic findings than in several previous studies.2,12,13 The relatively narrow range of patient risks creates a difficult situation for separation of different patient risks. Perfusion defects tended to be rather small with a low prevalence of large myocardial perfusion defects. Thus, accurate quantitative analysis appears to be required to detect even minor differences, while a visual analysis may be too crude in this preselected patient cohort.

ROC analysis demonstrated a SSS of 13 to be the cut-off value for a high risk of subsequent clinical events. This cut-off value relates to reports by Georgoulias et al15 on 246 consecutive asymptomatic patients evaluated after coronary artery stenting. In this study, the cumulative hard event-free rate at 10 years was 30.7% for patients with a SSS >13, whereas it was 97.3% for patients with a SSS <13. The annual mortality rates were similar to a report on MPS in patients with normal baseline electrocardiograms in which mortality rates of 1.1% in patients with normal perfusion and 3.4% in patients with reversible defects were reported.16

The study allows important insights into the clinical impact of minor severity coronary artery disease found to be nonsignificant by coronary angiography. Thus, it extends the prognostic evaluation of coronary artery disease to a level below those defined by significant epicardial lesions as determined by coronary angiography. Myocardial perfusion defects of increasing severity were associated with an impaired prognosis. This may be due to perfusion defects caused by atherosclerotic disease on the microvascular level which remains undetected by coronary angiography.

The findings of this study confirm results of a recent report by Alqaisi et al17 which demonstrated significant differences in patient prognosis between MPS positive and negative patients having angiographically insignificant coronary artery disease. However, this study demonstrates additionally that the extent of test positivity has important incremental information to test positivity alone.

The number of patients included in this analysis is limited. However, we applied strict inclusion criteria to evaluate the predefined study targets. Patients with valvular heart disease, history of myocardial infarction, or revascularization procedure or cardiomyopathy were excluded. Furthermore, patients with no significant coronary artery disease and no revascularization being performed based on the angiographic results and an available perfusion scintigraphy represent only a small fraction of the patients presenting for MPS assessment. However, this is a patient subgroup which allows important insights into the impact of coronary artery disease of minor severity as determined by angiography. The relatively small number of patients was in part substituted by a relatively long follow-up period of more than 6 years. To choose medical therapy without revascularization in patients with signs of coronary artery disease by angiography and finding of perfusion defect by stress MPI may imply that hemodynamically significant coronary stenosis were not treated adequately. Some of the patients included in the analysis may have had hemodynamically significant stenosis in spite of an angiographically determined diameter stenosis of <50%. However, treatment strategy in each patient was decided upon by two interventional cardiologists based on the full knowledge of all findings. Still, medical therapy was chosen primarily based on the angiographic assessment of a nonsignificant disease not requiring revascularization therapy. Coronary angiography was performed after MPS in all patients. Fifty patients included in the study had a SSS <4. It may appear inadequate to perform coronary angiography on a patient with a negative MPS. However, coronary angiography was performed based on clinical symptoms and the potential for balanced ischemia. In this study, two different radionuclide tracers were used. However, this should not have had an impact on the study results as both tracers have high sensitivity to detect perfusion defects.