Yield and Diagnostic Value of Stress Myocardial Perfusion Imaging in Patients Without Known Coronary Artery Disease Presenting With SyncopeClinical Perspective
Background—American College of Cardiology/American Heart Association appropriate use criteria recommend performing stress myocardial perfusion imaging (MPI) for intermediate- to high-risk patients presenting with syncope but not for low-risk patients. However, there are limited data to support these recommendations. We investigated the yield of stress MPI for the evaluation of syncope in patients at risk but without known coronary artery disease.
Methods and Results—Using the Cleveland Clinic Institutional Review Board–approved MPI database, we identified consecutive patients without known coronary artery disease who underwent stress MPI between 2006 and 2012 for diagnostic workup of syncope. Patients were stratified into low-, intermediate-, and high-risk groups using the Framingham risk score. For patients with abnormal MPI, left heart catheterization were reviewed if performed. There were 700 patients (mean age, 62±15 years; 55% female) who had undergone stress MPI for syncope; 659 patients (94%) had normal perfusion. Of the 41 patients with abnormal MPI, 18 had left heart catheterization (9 were false-positive); there were 23 remaining patients with abnormal MPI (16 having moderate to severe perfusion defect size) but who did not have a left-side angiogram and could have undiagnosed significant coronary artery disease. The diagnostic yield of stress MPI was similarly low among all cardiovascular risk categories.
Conclusions—Stress MPI for evaluation of syncope in patients without known coronary artery disease has a low-diagnostic yield among all risk categories; thus, reaffirmation and revision of the appropriateness criteria should be considered.
American College of Cardiology/American Heart Association (ACC/AHA) appropriate use criteria recommend performing stress myocardial perfusion imaging (MPI) for intermediate- to high-risk patients presenting with syncope (Level A7) but not for low-risk patients (Level I3).1 The American Society of Nuclear Cardiology (ASNC) recommends performing stress MPI for unexplained syncope.2 Despite the limited data to support these recommendations, stress MPI is frequently the initial diagnostic test obtained. One of the rationales is that syncope may be a manifestation of significant coronary artery disease (CAD) and ischemia.3 However, on the basis of our clinical experience, the utility of such testing, particularly for those without previous diagnosis of CAD, is limited. Hence, we investigated the diagnostic utility of stress MPI for patients presenting with syncope.
Editorial see p 358
Clinical Perspective on p 391
From the stress MPI database, we identified consecutive patients who underwent stress MPI between October 2006 and August 2012 for diagnostic workup of syncope. Inclusion criteria included age >18 years and syncope as the clinical indication that was entered for the stress test. Patients with known history of CAD, previous myocardial infarction, or revascularization were excluded. Patient demographics, comorbidities, and medications entered at the time of stress testing were extracted for analysis.
CAD was defined on coronary angiography as >50% stenosis of a major epicardial coronary artery or q waves on an ECG with clinical diagnosis of myocardial infarction. Hypertension was defined as systolic blood pressure (SBP) ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg at the time of the stress test or clinic visits, self-reported history, or the use of antihypertensive medications. Hyperlipidemia was defined as an abnormal fasting lipid panel according to the Adult Treatment Panel III guidelines, self-reported history, or the use of statins. Diabetes mellitus was defined as fasting glucose ≥126 mg/dL, self-reported history, or the use of hypoglycemic medications. We also retrieved from the electronic medical record (looking 6 months before and up to 6 months after the stress test) whether there was history or documentation of aortic stenosis, hypertrophic cardiomyopathy, ventricular arrhythmia, or high-degree atrioventricular block, which are common causes of cardiac syncope. The study was approved by the Cleveland Clinic Institutional Review Board with waiver of consent and complied with the Declaration of Helsinki.
Cardiovascular Risk Stratification
The 10-year risk of cardiovascular disease was estimated from the Framingham risk score. The latter was calculated using age, sex, blood pressure, history of smoking, and fasting lipid panel obtained at the time of the echocardiogram, and a score was generated from standardized tables.4 Patients were then categorized as low risk (10-year risk <10%), intermediate (10% to 20%), and high risk (>20%).
Exercise or Pharmacological Stress Testing
The methods for exercise treadmill testing in our laboratory have been described in detail previously.5 Briefly, standard protocols (Bruce, modified Bruce, and Cornell) were chosen with goal test duration between 8 and 12 minutes. Patients were asked to hold β- blockers (if any) at least 12 to 24 hours before the stress testing. All patients exercised to exhaustion, regardless of the achieved heart rate, and were asked not to hold on to the handrails. However, the test was terminated prematurely if patients developed severe chest pain, symptomatic hypotension, SBP >250 mm Hg, significant arrhythmia, or severe ST- segment changes or as per patient request. Patients were monitored for heart rate, blood pressure, rhythm, symptoms, ECG changes, and rate of perceived exertion (on a 1-to-10 scale, where 10 is maximum exertion) at rest and at every stage of the exercise protocol. Exercise capacity in metabolic equivalent (1 metabolic equivalent = 3.5 mL//kg per minute of oxygen consumption)6 was estimated on the basis of the protocol, speed, and grade achieved.6,7
The pharmacological tests were performed with dipyridamole infusion of 0.56 mg/kg over 4 minutes, adenosine infusion of 140 μg/kg per minute for 4 to 6 minutes, or regadenoson a 0.4-mg intravenous bolus.
Single Photon Emission Tomography MPI
The full protocol was described in detail previously.8,9 Briefly, gated single photon emission tomography MPI (SPECT MPI) was obtained at rest (low dose ≈9–15 mCi) and stress (exercise [Bruce, modified Bruce, or Cornell] or pharmacological [dipyridamole, adenosine, or regadenoson]; high dose, ≈30–45 mCi) using Tc-99m tetrofosmin according to ASNC guidelines. Imaging was started 30 to 60 minutes after resting or pharmacological testing and 10 to 20 minutes after exercise testing using a dual-headed detector gamma camera with high-resolution, low-energy collimators. Images were acquired using a 64×64 matrix with a step- and- shoot protocol, 180° elliptical orbit, 64 total projections, and 16 frames per RR cycle. They were then reconstructed by filtered back projection using Ramp and Butterworth filters or iterative reconstruction (starting September 2008) without attenuation correction. All images were reviewed for quality assurance by a senior nuclear technologist and a board-certified nuclear cardiologist.
The left ventricular volumes and ejection fraction (EF) were calculated from the stress-gated images (higher counts). The presence and extent of perfusion defect (summed stress score [SSS], summed difference score, and summed rest score) were quantified using semiautomated polar maps as described previously.10 An abnormal MPI was defined as SSS ≥3 (approximate perfusion defect size [PDS], 5%) and categorized as mild (SSS 3–7 or PDS 5%–10%), moderate (SSS 7–14 or PDS 10%–20%), and severe (SSS ≥14 or PDS >20%). Patients with a perfusion defect that was suspected to be artifact or attenuation related (ie, diaphragm attenuation of the inferior wall in male patients or breast attenuation of the anterior wall) but unconfirmed and reported as equivocal were grouped with those with abnormal MPI tests.
Stress Positron Emission Tomography MPI
Gated positron emission tomography (PET) 82-Rb images were acquired on 2 scanners as described in detail previously.11 The stress test was performed pharmacologically using dipyridamole infusion or regadenoson (starting December 2010); 82-Rb infusion during stress and rest (60 mCi with the dedicated PET scanner, 40 mCi with the hybrid PET/computed tomography scanner) started at 8 minutes. Data acquisition, attenuation correction, gating, reconstruction, nominal transaxial/axial spatial resolution, estimated radiation dose delivered to patients, and postprocessing were as described previously.11 PET images were reoriented along the standard cardiac axes and displayed using Corridor4DM software (Invia, Ann Arbor, MI). Left ventricular volumes and EF were computed by the software using stress-gated images (higher counts). The perfusion defects (SSS, summed difference score, and summed rest score) were quantified using semiautomated polar maps.12 All quantitative measures were visually reviewed and adjusted when appropriate by a nuclear cardiologist.
We retrieved the pathogenesis of syncope (when available or identified) from chart review. We evaluated the proportion of patients with abnormal stress test results. For those with abnormal MPI (ie, SSS ≥3), we reviewed patients’ records to identify whether left heart catheterization (LHC) was performed at the Cleveland Clinic, its satellite facilities, or another facility with documentation of the angiography results up to 6 months from the stress test date (mean interval time, 4.1±3.5 days). We then stratified the results by the Framingham risk score (low, intermediate, or high). All-cause death was reported using the Social Security Death Index master file with previously reported high specificity.13 Censoring date was November 1, 2012. Mortality rates were annualized by dividing by the mean follow-up time.
A descriptive analysis was performed examining selected variables for each group. Continuous variables were expressed as mean (SD) or median (quartile 1–3) and compared using 1-way ANOVA or Kruskal-Wallis test as appropriate. Categorical variables were expressed as number (percentage) and compared using Pearson χ2 test or Fisher exact test as appropriate. Kaplan–Meier survival curves stratified by the Framingham risk scores and MPI were generated and compared using the log-rank test. Survival analysis treated the time of stress MPI as time 0. A value of P <0.05 was set a priori and considered statistically significant. All statistical analyses were performed using the Statistical Package for Social Sciences, version 19, for Windows (SPSS, Chicago, IL).
The study consisted of 700 patients without known diagnosis of CAD (mean age, 62±15 years; 55% female) who had syncope and underwent stress MPI according to the standard clinical practice at the Cleveland Clinic. Patients had significant comorbidities, including hypertension (62%), hyperlipidemia (54%), diabetes mellitus (20%), and smoking (44%; Table 1). The prevalences of history of ventricular arrhythmia, aortic stenosis, and hypertrophic cardiomyopathy were 5%, 1.4%, and 1.6%, respectively, and no patients had known high-degree atrioventricular block. The baseline demographics, comorbidities, and medications are listed in Table 1 and were stratified by the Framingham cardiovascular risk score. Patients in the high-risk group had more cardiovascular risk factors and were taking more cardiac medications (Table 1).
There were 165 patients (24%) who performed treadmill stress MPI, exercised for a mean of 7.6±2.0 minutes, and achieved 92±8% maximal predicted heart rate and 8.7±3.0 metabolic equivalent. The resting blood pressures were 129±18 mm Hg (SBP) and 82±10 mm Hg (diastolic); the peak stress blood pressures were 174±26 mm Hg (SBP) and 84±13 mm Hg (diastolic). Only 2 patients experienced a drop in peak SBP >10 mm Hg (10 and 32 mm Hg, respectively) but did not have any associated symptoms. None of the patients developed hypotension or SBP <100 mm Hg during exercise. Similarly, none developed bradycardia or heart block. Patients in the high-risk category achieved lower metabolic equivalents and exercise time (P=0.002 and 0.021 for the trend across risk groups, respectively) but no significant difference in hypotension.
Among patients who underwent pharmacological vasodilator stress testing (n=535), 26 patients (4.9%) had hypotension (SBP <100 mm Hg) and 24 patients (4.5%) had >30-mm Hg drop in SBP. None of the patients had syncope, bradycardia, or heart block. There was no significant difference in the incidence of hypotension when stratified by the Framingham risk score.
Almost three quarters of the stress tests (n=535) were pharmacological (510 SPECT and 25 were PET). The mean left ventricular EF and end-diastolic volume index were 68±13% and 46±23 mL/m2, respectively (Table 2). There were 53 patients (7.6%) with EF <50% and 41 patients (5.9%) with abnormal or equivocal MPI, with a significantly higher percentage among those in the high-risk category (P=0.004 and 0.036, respectively; Table 2 and Figure 1). The histogram plots of EF and SSS/summed rest score/summed difference score are illustrated in Figure 2.
Yield and Diagnostic Utility
There were 41 studies (5.9%) with abnormal MPI and 53 studies (7.6%) with EF <50% (Table 2). Of the 41 studies with abnormal MPI (5.9%), 24 studies had reversible perfusion defect (alone or mixed), and 17 studies had only fixed perfusion defect. Subsequent to the stress tests, 18 LHCs were performed, 9 of which showed no obstructive CAD and were therefore deemed false-positive. The overall yield of documented hemodynamically and angiographically significant CAD was 9 of 700 (1.3%; 95% confidence interval, 0.68–2.4) with 23 remaining patients with abnormal MPI (16 patients having moderate to severe PDS) but who did not have an LHC and could have undiagnosed significant CAD (Figure 1).
Yield of MPI Stratified by Framingham Risk Score
There were significantly higher numbers of abnormal MPI and EF <50% among patients in the intermediate- and high-risk groups (Table 2 and Figure 1). Among those in the low-risk group (n=339), there were 13 abnormal MPIs and 10 LHCs (7 false-positive) with a yield of 0.88% (95% confidence interval, 0.3–2.6), whereas the risk scores in the intermediate- and high-risk groups were 1.3% (95% confidence interval, 0.3–4.0) and 2.4% (95% confidence interval, 0.61–7.3), respectively. Assuming that the remaining patients with at least moderate PDS but without LHC (1 in the low-risk group, 11 in the intermediate-risk group; and 4 in the high-risk group) had significant CAD, the yield would be 1.2% (95% confidence interval, 0.38–3.2), 6.0% (95% confidence interval, 3.4–10), and 5.5% (95% confidence interval, 2.4–11), respectively (Figure 1).
Yield of MPI Stratified by Type of Stress Test
There were 165 exercises SPECTs, 535 pharmacological SPECTs, and 25 PETs. The presence of abnormal MPI, LHC results, and overall diagnostic yield of stress MPI workup of syncope were stratified by the type of stress test and summarized in Table 3.
Aortic Stenosis, Hypertrophic Cardiomyopathy, and Ventricular Arrhythmia
There were 10, 11, and 36 patients with a history of aortic stenosis, hypertrophic cardiomyopathy, and ventricular fibrillation, respectively, in the entire cohort, with 80% to 100% having normal MPI and ≈82% with EF ≥50%. There was no statistically significant difference of these conditions between patients with normal and those with abnormal MPI (P=0.06, 0.4, and 0.08 for aortic stenosis, hypertrophic cardiomyopathy, and ventricular fibrillation, respectively) or among different Framingham risk scores (Table 1).
Pathogenesis of Syncope
The pathogenesis of syncope is summarized in Table 4. Almost a third were idiopathic without a clear diagnosis, followed by vasovagal and neurogenic, whereas cardiac pathogenesis, particularly ischemia and CAD, were uncommon.
Over a mean follow-up time of 2.6 years (median, 2.6 years; quartile 1–3, 1.4–3.7 years), 45 (6/4%) patients died. There were significantly increased event rates for the high-risk group (1%/y [low risk] versus 1.5%/y [intermediate risk] versus 8.7%/y [high risk]; log-rank P<0.0001; Figure 3). Similar trends were observed when results were stratified by the type of stress test.
Among patients with abnormal MPI (n=41), there was no significant difference in outcome between those who underwent LHC and those who did not (2 of 18 versus 3 of 23; log-rank P=0.9). Also, there was no significant difference in annual mortality between those with normal and those with abnormal MPI (2.3% versus 3.3%, respectively; log-rank P=0.3; Figure 4). Patients who had exercise stress tests had lower mortality than those undergoing pharmacological stress tests (0.86%/y versus 3.0%/y; P<0.001), and those with both exercise testing and normal MPI had the lowest mortality (0.68%/y). There was a slight trend toward worse survival for those with any abnormality on the stress test (ie, EF <50% or abnormal MPI) compared with those with normal EF and MPI (log-rank P=0.09).
Syncope is the abrupt or transient loss of consciousness followed by complete and spontaneous recovery. Although often a benign finding, it may represent severe underlying disease. Vasovagal and cardiac disease are the most common causes, but unexplained syncope is present in 20% to 40% of cases.14,15 Although history, physical examination, and ECG are the initial steps in the evaluation of syncope and most often identify the pathogenesis, advanced testing is often needed.16 However, even after extensive cardiovascular workup, the pathogenesis of syncope might remain unexplained.17 The most recent ACC/AHA appropriate use criteria recommend performing stress MPI for intermediate- to high-risk patients presenting with syncope (Level A7) but not for low-risk patients (Level I3).1 Also, the ASNC recommends performing stress MPI for unexplained syncope.2 The data supporting these recommendations, however, are limited. Still, stress MPI is frequently ordered and sometimes one of the first diagnostic tests to be performed. At our institution, patients presenting to the syncope clinic, for example, undergo stress testing per protocol (approximately two thirds undergo MPI, one third undergo stress echo cardiography). There have been few published cases showing causality between myocardial ischemia and clinical vasodepressor syncope.3 However, it has been our experience that the yield for stress MPI for patients presenting for syncope in the absence of known history of CAD is very low. For this reason, we sought to assess the clinical yield of stress MPI in our tertiary referral center.
Our study showed that 5.9% of 700 patients had abnormal MPI and only 9 patients had significant CAD. There were 23 patients with abnormal or equivocal stress MPI who did not undergo an LHC (at least not within our healthcare system or as shown in their records), with 16 patients having at least moderate PDS who could have undiagnosed CAD, raising the overall yield to potentially 3.6% (4.6% if all the abnormal MPI were true-positive). This is significantly lower than the percentage of abnormal stress MPI of all patients who are referred to our institution for various diagnoses during the same time period (≈32%).
To better stratify the diagnostic yield of stress MPI on the basis of patient risk, we used the Framingham risk score to stratify patients as low, intermediate, and high risk. Patients in the low-risk group had a low yield of CAD (3 of 339), which is consistent with the guideline recommendation for stress test as inappropriate for this cohort. Those in the intermediate- and high-risk groups also had low yield of CAD (6 of 361 with confirmed CAD and 15 potential additional patients with at least moderate PDS but without LHC). Hence, in intermediate- and even high-risk patients without known CAD presenting with syncope, an appropriateness level of A7 for such groups may need to be revised, particularly given the limited identified cardiac causes of syncope by MPI (Table 4). Also, in this cohort, an abnormal stress test did not confer a higher mortality outcome compared with a normal test, likely attributable to a high false-positive rate.
Although the ACC/AHA recommend exercise testing for patients at risk for CAD,1 the European Society of Cardiology guidelines recommend exercise testing for patients presenting with syncope during or shortly after exertion.18 The purpose of exercise testing is to assess for hemodynamic changes, autonomic dysfunction, or arrhythmia that may reproduce the same symptoms. Yet, the majority of stress MPI performed in real life is pharmacological (76% in this cohort). However, when we stratified by the type of stress test, the overall yield between exercise and pharmacological SPECT was not significantly different (0.6% versus 0.7%; P=0.99; Table 3). Patients who underwent cardiac PET had a higher yield, likely driven by selection bias. Also, among those who exercised (n=165), none had hypotension, significant drop in blood pressure, bradycardia, or heart block that might explain the pathogenesis of the syncope.
Strengths and Limitations
This is the first study to the best of our knowledge that evaluated the yield and diagnostic utility of stress MPI for detection of CAD in the assessment of syncope for patients without known CAD. We have confirmed the low yield in patients with low risk (indeed inappropriate test) and showed that even among those at intermediate and high risk, the yield is still low, raising the question of whether the current guidelines reflect the low diagnostic yield and prognostic value. The study addresses an important question in an era of cost containment. We acknowledge several limitations. This is a retrospective study from a single tertiary referral center with probable referral and selection bias. Patients with normal MPI had higher annual mortality than expected, particularly those undergoing pharmacological stress testing who are more likely to be inpatients with additional comorbidities; those with exercise stress testing, however, had <1%/y mortality. We did not evaluate whether patients with normal MPI had LHC and CAD (ie, false-negative test). However, patients with normal MPI often did not undergo LHC unless the pretest probability for CAD was very high, and in such cases, stress MPI would not have been the appropriate test. Also, even if the diagnosis of CAD was missed among patients with normal MPI (ie, false-negative test), this strengthens our findings because it further reduces the clinical utility of stress MPI.
In addition, more than half of the patients with positive MPI did not have an LHC for various reasons: LHC performed elsewhere with no documentation; myocardial perfusion defect that could have been small, suggestive of an artifact, or equivocal test that persuaded the physician not to order an LHC; lost to follow-up; or unknown reasons. Regardless, there were no significant differences in outcomes between those who had an LHC and those who did not and similarly between those with normal and those with abnormal MPI (Figure 4). One explanation is that many of the abnormal MPI tests were false-positives, were clinically insignificant, or did not explain the syncope (attributable to other pathogeneses). Finally, although we chose patients who had syncope entered as the indication for stress testing, it is likely that some patients might have had other concomitant symptoms such as chest pain or shortness of breath. This reinforces our results because excluding those with angina or other symptoms would likely further reduce the diagnostic yield of stress MPI for isolated syncope.
Stress MPI for evaluation of syncope in patients without known CAD has a low diagnostic and minimal prognostic yield among low-risk patients in whom such testing is indeed inappropriate. For those in the intermediate- and high-risk groups, the yield is still relatively low; therefore, reaffirmation and revision of the appropriateness criteria should be considered.
- Received November 30, 2012.
- Accepted February 6, 2013.
- © 2013 American Heart Association, Inc.
- Hendel RC,
- Berman DS,
- Di Carli MF,
- Heidenreich PA,
- Henkin RE,
- Pellikka PA,
- Pohost GM,
- Williams KA
- Ward RP,
- Al-Mallah MH,
- Grossman GB,
- Hansen CL,
- Hendel RC,
- Kerwin TC,
- McCallister BD Jr.,
- Mehta R,
- Polk DM,
- Tilkemeier PL,
- Vashist A,
- Williams KA,
- Wolinsky DG,
- Ficaro EP
- Germano G,
- Kiat H,
- Kavanagh PB,
- Moriel M,
- Mazzanti M,
- Su HT,
- Van Train KF,
- Berman DS
- Strickberger SA,
- Benson DW,
- Biaggioni I,
- Callans DJ,
- Cohen MI,
- Ellenbogen KA,
- Epstein AE,
- Friedman P,
- Goldberger J,
- Heidenreich PA,
- Klein GJ,
- Knight BP,
- Morillo CA,
- Myerburg RJ,
- Sila CA
- Moya A,
- Sutton R,
- Ammirati F,
- Blanc JJ,
- Brignole M,
- Dahm JB,
- Deharo JC,
- Gajek J,
- Gjesdal K,
- Krahn A,
- Massin M,
- Pepi M,
- Pezawas T,
- Ruiz Granell R,
- Sarasin F,
- Ungar A,
- van Dijk JG,
- Walma EP,
- Wieling W
The American College of Cardiology/American Heart Association appropriate use criteria recommend performing stress myocardial perfusion imaging (MPI) for intermediate- to high-risk patients presenting with syncope (Level A7) but not for low-risk patients (Level I3). Despite the limited data to support these recommendations, stress MPI is frequently the initial diagnostic test obtained. On the basis of our clinical experience, the utility of such testing, particularly for those without previous diagnosis of coronary artery disease, is limited. Hence, we investigated the diagnostic utility of stress MPI in 700 patients without known coronary artery disease (mean age, 62±15 years; 55% female) presenting with syncope. Patients were stratified into low-, intermediate-, and high-risk categories using the Framingham risk score. For patients with abnormal MPI, left heart catheterization were reviewed if performed. Of the 700 patients, 659 patients (94%) had normal perfusion and 41 patients had abnormal MPI with 18 subsequent left heart catheterization performed (9 of which were false-positive). There were 23 remaining patients with abnormal MPI (16 patients having moderate to severe perfusion defect size), but who did not have a left-side angiograms and could have undiagnosed significant coronary artery disease. The diagnostic yield of stress MPI was similarly low among all cardiovascular risk categories. Therefore, reaffirmation and revision of the appropriateness criteria should be considered.