The Diagnostic Accuracy of 64-Slice Computed TomographyCoronary Angiography Compared With Stress Nuclear Imaging
in Emergency Department Low-Risk Chest Pain Patients
Michael J. Gallagher, MDMichael A. Ross, MDGilbert L. Raff, MDJames A. Goldstein, MDWilliam W. O’Neill, MDBrian O’Neil, MD
From the Departments of Cardiology (Gallagher, Raff, Goldstein, O’Neill), and Emergency Medicine,(Ross, O’Neil), William Beaumont Hospital, Royal Oak, MI; and the Department of EmergencyMedicine, Wayne State University School of Medicine, Detroit, MI (Ross, O’Neil).
Study objective: We compared the accuracy of multidetector computed tomography (CT) coronaryangiography with stress nuclear imaging for the detection of an acute coronary syndrome or 30-daymajor adverse cardiac events in low-risk chest pain patients.
Methods: This was a prospective study of the diagnostic accuracy of myocardial perfusion imagingand multidetector CT in low-risk chest pain patients. The target condition was an acute coronarysyndrome (confirmed �70% coronary stenosis on coronary artery catheterization) or major adversecardiac events within 30 days. Patients were low risk by Reilly/Goldman criteria and had negativeserial ECGs and cardiac markers. All had both rest/stress sestamibi nuclear imaging andmultidetector CT. Patients with abnormal stress nuclear imaging results (reversible perfusion defects)or multidetector CT results (stenosis �50% or calcium score �400) were considered for cardiaccatheterization, and those with discordant results had a greater than 30-day reevaluation (includingECG) by a cardiologist. All were followed up for evidence of major adverse cardiac events within 30days by review of hospital records and structured telephone interview. Primary outcomes were theaccuracy of multidetector CT and myocardial perfusion imaging for the detection of an acute coronarysyndrome and 30-day major adverse cardiac events.
Results: Of the 92 patients, 7 (8%) were excluded because of uninterpretable multidetector CTscans. Of the remaining 85 study patients (49�11 years, 53% men), 7 (8%) were found to have thetarget condition, with all having significant coronary stenosis (88%�9%) and none having myocardialinfarction or major adverse cardiac events during 30 days. Stress nuclear imaging results werenegative in 72 (85%) patients, and multidetector CT results were negative in 73 (86%) patients. Thesensitivity of stress nuclear imaging was 71% (95% confidence interval [CI] 36% to 92%), andmultidetector CT was 86% (95% CI 49% to 97%), and the specificity was 90% (95% CI 81% to 95%)and 92% (95% CI 84% to 96%), respectively. The negative predictive value of stress nuclear imagingand multidetector CT was 97% (95% CI 90% to 99%) and 99% (95% CI 93% to 100%), respectively,and the positive predictive value was 38% (95% CI 18% to 64%) and 50% (95% CI 25% to 75%),respectively.
Conclusion: The accuracy of multidetector CT is at least as good as that of stress nuclear imagingfor the detection and exclusion of an acute coronary syndrome in low-risk chest pain patients. [AnnEmerg Med. 2007;49:125-136.]
Volume , . : February Annals of Emergency Medicine 125
Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography Gallagher et al
INTRODUCTIONBackground
Each year, roughly 6 million patients are evaluated for chestpain in emergency departments (EDs).1 Accurate and efficientscreening for patients with an acute coronary syndrome isessential. Historically, 2% to 10% of patients with an acutecoronary syndrome are inappropriately sent home from theED.2 Missed diagnosis of acute myocardial infarction isassociated with significant morbidity, and it is the leadingcontributor to malpractice claims paid by emergencyphysicians.3 Because of this, many patients are admitted forfurther testing, with the majority found to not have an acutecoronary syndrome. It is estimated that the cost of thesenegative evaluations is $10 to $13 billion per year. To addressthese issues, EDs have developed chest pain units and diagnosticprotocols. These protocols commonly include serial cardiacserum marker evaluations and ECGs, followed by stress testing,with or without radionuclide imaging.4,5
ImportanceRapid advances in multidetector computed tomographic
(CT) technology have allowed noninvasive coronary arteryimaging. Studies comparing 64-slice multidetector CT with
Editor’s Capsule Summary
What is already known on this topicMost patients admitted to chest pain observation unitsreceive stress testing before discharge from the unit.Computed tomography (CT) coronary angiography hasbeen shown to have excellent correlation with cardiaccatheterization, so it might provide an alternative methodof evaluation.
What question this study addressedThis study compared the use of multidetector CTcoronary angiography with stress nuclear imaging for thedetection of coronary artery disease or 30-day majoradverse cardiac events in low-risk chest pain patients.
What this study adds to our knowledgeThis study demonstrated that multidetector CT was asgood as stress nuclear imaging for detection of the 7patients with coronary artery disease or 30-day adversecardiovascular events in this low-risk cohort.
How this might change clinical practiceAlthough a small nonrandomized study, this study doessuggest that multidetector CT is an alternative to stresstesting in patients who have completed an uneventfulstay in a chest pain observation unit. If confirmed byothers, multidetector CT may replace stress testing forsome patients admitted to the chest pain unit.
invasive coronary angiography have shown that multidetector
126 Annals of Emergency Medicine
CT performs well in the detection of significant coronarystenosis, with sensitivities ranging from 82% to 95% andspecificities of 82% to 98%.6-9 The presence of coronarycalcification in patients with acute chest pain has also beenshown to be predictive of future cardiac events.10 However,there are no studies that compare the ability of multidetectorCT to detect an acute coronary syndrome with traditional stressnuclear imaging in low-risk chest pain patients.
Goals of This InvestigationOur study objective was to compare the diagnostic accuracy
of multidetector CT with traditional stress nuclear imaging forthe detection of an acute coronary syndrome in ED low-riskchest pain patients.
MATERIALS AND METHODSThis was a prospective institutional review board–approved
study comparing the diagnostic accuracy of multidetector CTand radionuclide rest/stress imaging in a convenience sample oflow-risk chest pain patients. Study patients received amultidetector CT and a radionuclide stress test, thus allowingeach patient to serve as his or her own control.
Setting and Selection of ParticipantsThe setting of this study was a suburban teaching hospital
with an annual ED census of 115,894 visits per year. Thedepartment included a 21-bed ED observation unit that hasbeen operational for more than 10 years and is fully staffed byattending emergency physicians.11 A “low-risk” chest paindiagnostic protocol is used in the ED observation unit. All studypatients had negative serial ECG and cardiac marker results inthe ED and ED observation unit, as described below.
Patients were identified as study candidates by the initialemergency physician and then screened by trained research staffat their admission to the ED observation unit. Patient screeningtook place from 7:00 PM Sunday through 5:00 PM Friday(according to research staff and CT availability). Patients wereexcluded from the study if they had positive initial cardiacmarker results or new ischemic ECG changes. Patients were alsoexcluded from the study if they were younger than 18 years,pregnant, had known coronary artery disease (�30% coronarystenosis), existing cardiomyopathy, congestive heart failure withan ejection fraction less than or equal to 45%, acontraindication to iodinated contrast or �-blocking drugs,atrial fibrillation or markedly irregular rhythm, renalinsufficiency (creatinine �1.5 mg/dL), or had received CTimaging or contrast within the past 48 hours. Screening anddata collection by trained research assistants began at studyenrollment and occurred prospectively.
Patients were first treated in the ED and found to havesymptoms suggestive of an acute coronary syndrome.“Suggestive” symptoms were included because up to one thirdof patients with an acute coronary syndrome do not present
with chest pain.12 Suggestive symptoms included discomfort,
Volume , . : February
Gallagher et al Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography
burning, and dyspnea. The term “chest pain” is used as asimplified descriptor of these patients. With criteria previouslyvalidated by Reilly et al,13 patients were identified as “low risk”according to their symptoms, examination results, and initialECG findings. By these criteria, low-risk patients had no ECGevidence of acute infarction or ischemia (including new leftbundle branch block), no pain that was worse than usual anginaor like a previous myocardial infarction, no recentrevascularization, no rales above both bases, and a systolic bloodpressure that was greater than 110 mm Hg. In the ED, patientsalso had an initial chest radiograph and an initial set of cardiacmarkers (CK-MB and troponin I) to screen for non–ST-segment-elevation myocardial infarction or other majornonacute coronary syndrome conditions such as pneumothoraxor pneumonia.
Study patients were admitted to the ED observation unit forthe chest pain diagnostic protocol. The protocol consisted ofcardiac monitoring and serial ECG and cardiac marker tests(CK-MB fraction, troponin I, and myoglobin) 4 hours after EDarrival. Positive test results were defined as dynamic ischemicECG changes, positive CK-MB mass and index, or positivetroponin I level. Patients whose myoglobin result was positive orwhose CK-MB level was normal but doubled had a repeatedECG, CK-MB, and troponin I test 8 hours after ED arrival. Ifthese results were negative, the patient then received anappropriate stress test, with sestamibi stress nuclear imaging, aswell as a multidetector CT coronary angiography. To reduce thelikelihood of �-blocker-induced chronotropic incompetenceduring stress testing, most patients (87%) underwent stressnuclear imaging before multidetector CT. The treatingemergency physician, in concurrence with a consultingcardiologist, was provided with stress nuclear imaging andmultidetector CT clinical readings. They selected patientmanagement according to these combined results. It was notthought to be appropriate to withhold multidetector CT resultsfrom the treating physicians for the purposes of this study.
Patients were followed up for evidence of an acutecoronary syndrome or major adverse cardiac events (seebelow) within 30 days of their index visit by structured chartreview, telephone follow-up, and scheduled return visits forselected patients. All patients with discordant imaging results(abnormal stress nuclear imaging result but normalmultidetector CT result or normal stress nuclear imagingresult but abnormal multidetector CT result) who did notreceive cardiac catheterization were called back to be seen bya cardiologist at a cardiology clinic for a repeated ECG,physician interview, and physical examination at least 30days after the index visit specifically to determine whether anacute coronary syndrome or major adverse cardiac events hadoccurred after discharge. Readmission to either the studyhospital or a proximate affiliated hospital was also followedup by electronic record review for evidence of 30-day major
adverse cardiac events.
Volume , . : February
Data Collection and ProcessingThe collection of data for all patient characteristics and study
outcomes occurred prospectively and began at study enrollmentwith a standardized case report form. Data were collected bytrained research assistants or study physicians. Data validity wascross-checked with double data entry methods.
The target conditions being tested for in this study were anacute coronary syndrome on the index visit and major adversecardiac events within 30 days of the index visit. Acute coronarysyndromes encompass any of 3 conditions: an ST-segment-elevation myocardial infarction, a non–ST-segment-elevationmyocardial infarction, and unstable angina. Because serialtesting essentially excludes the first 2 conditions, the primarycondition being tested for was unstable angina. Definingunstable angina according to symptoms alone is problematicand would lead to several false-positive cases.14 Alternatively,stress nuclear imaging results could not be used as a referencestandard because stress nuclear imaging itself is an imperfecttest,15 and in this study it was being compared withmultidetector CT for their respective abilities to detect an acutecoronary syndrome. In this study, patients with positive testingsuggestive of unstable angina underwent cardiac catheterizationfor confirmation of the diagnosis. As such, unstable angina wasdefined by cardiac catheterization showing at least 70%coronary artery stenosis. Coronary stenosis was confirmed byquantitative coronary angiography (QuantCor QuantitativeCoronary Angiography, Pie Medical Systems, Masstricht,Netherlands). Catheterization of all low-risk patients was notthought to be appropriate or necessary, so we chose to alsofollow a clinical outcome as a reference standard for disease or asurrogate marker of disease. The clinical outcome was majoradverse cardiac events, for which all patients were followed upfor at least 30 days from the index visit. Major adverse cardiacevents was defined as an acute coronary syndrome, as definedabove, the development of new Q waves on subsequent ECGs,new congestive heart failure or cardiogenic shock, majordysrhythmias (high-grade atrioventricular block, ventriculartachycardia, ventricular fibrillation), cardiac arrest, or deathfrom an acute coronary syndrome.
Patients received a rest/stress nuclear imaging protocol. Forrest imaging, a weight-adjusted dose of 99mTc-sestamibi (8 to 10mCi) was injected, followed by image acquisition 10 minutesafter injection. Two or more hours after rest injection, patientsunderwent symptom-limited standard exercise treadmill testingor a pharmacologic stress (dipyridamole intravenous injection ata dose of 0.57 mg/kg during 4 minutes) protocol. At peakexercise, a weight-adjusted dose of 99mTc-sestamibi (25 to 40mCi) was injected, followed by image acquisition 30 minutesafter injection. Patients with equivocal or probably abnormalsupine stress nuclear imaging study results underwent proneimaging to minimize attenuation artifacts. The ECG responseto exercise was categorized as either nonischemic or ischemic(�1 mm flat or down-sloping ST-segment depression or
sustained ventricular tachycardia). The clinical response to
Annals of Emergency Medicine 127
Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography Gallagher et al
exercise was categorized as either nonischemic or ischemic(typical angina pectoris during exercise). Hemodynamiccharacteristics, including age-adjusted heart rate response, andDuke treadmill score were also recorded.
Qualitative and semiquantitative visual analysis was made bya board-certified nuclear physician using a 17-segment model,according to previously published methods.16,17 Each of the 17segments was classified into a qualitative severity scale. All studyimages were read by a nuclear physician involved in the patients’care and later by a second blinded nuclear physician. Both wereblinded to multidetector CT results. Discordant readings wereadjudicated by consensus with a third physician. Images wereclassified as abnormal or probably abnormal if there weresignificant reversible perfusion defect(s) and normal or probablynormal if reversible defects were not present.
All patients were scanned on a 64-slice multidetector CTscanner (Sensation 64 Cardiac; Siemens Medical Systems,Forchheim, Germany), as previously described.6 Patients weregiven 50 to 100 mg of atenolol 1 hour before multidetector CTimaging or intravenous metoprolol (5 to 30 mg) as needed toachieve a target pulse rate less than or equal to 65 beats/min.Nitroglycerin 0.4 mg sublingual was given 1 minute beforeimage acquisition. Pulse rate, ECG, and blood pressure weremonitored before, during, and after imaging. An initialunenhanced scan was performed for calcium scoring. Acontrast-enhanced scan (Visipaque; GE Healthcare, Waukesha,WI) was obtained using 100 mL of contrast injected through anantecubital vein at 5 mL/s, followed by 20 mL of contrastinjected at 3 mL/s, followed by a 40-mL saline solution chaser.Timing was performed with a program sensing a threshold of160 Hounsfield units in a region of interest in the aorta, using asingle breath-hold cranial-to-caudal acquisition (aortic arch toinferior heart border). The scan parameters were 32 � 0.6 mmcollimation with dual focal spots per detector row, tube rotationtime 330 ms, table feed/rotation 3.8 mm, tube voltage 120 mV,effective mAs 750 to 850, pitch 0.2, volumetric CT dose index59 mGy. Tube current modulation was used to decrease tubeoutput during ECG systole in the majority of patients. ECG-gated data sets were reconstructed automatically at 65% and35% of the R-R cycle length, and additional reconstructionwindows were created after examination of data sets if motionartifacts were present. In addition, 2-mm- and 3-mm-thickreconstruction data sets were created with skin-to-skin field-of-view coverage for evaluation of the lung parenchyma andmediastinum.
Calcium scores were analyzed with SYNGO software(Siemens Medical Systems). Multidetector CT angiograms wereanalyzed on a 3-dimensional workstation (Aquarius; TeraRecon,San Mateo, CA). Scans were analyzed by a board-certifiedcardiologist with level 3 clinical competence in cardiovascularCT imaging18 who was unaware of the clinical data and blindedto the results of the stress nuclear imaging. A 15-segment modelof the coronary tree was used.19 Each lesion was identified by
using maximum intensity and multiplanar reconstruction
128 Annals of Emergency Medicine
techniques transversely and along multiple longitudinal axes.Lesions were classified by a qualitative severity scale: 0�nostenosis, 1�1% to 25% stenosis, 2�26% to 50% stenosis,3�51% to 70% stenosis, 4�71% to 99% stenosis, 5�totalocclusion. Multidetector CT stenosis of greater than 50% wasclassified as positive. Multidetector CT was also classified aspositive if there was coronary calcium encroachment on thelumen in transaxial multiplanar reconstructions greater than50% or a coronary artery calcium score greater than 400.Patients with uninterpretable images were excluded from thisstudy.
Sensitivity, specificity, positive predictive value, and negativepredictive value were determined for CT and stress nuclearimaging separately, including 95% confidence intervals (CIs).Patients with the reference outcome are shown separately frompatients without for demographics, ECG data, stress nuclearimaging, and CT data. With few reference outcome patients,statistical comparisons were not completed between these 2groups. Statistical analyses were completed using SAS version9.1 (SAS Institute, Inc., Cary, NC).
RESULTSPatient enrollment occurred during 7 months, from
September 2004 to March 2005. Throughout the study period,there were 68,367 ED visits, of which 2,182 patients (age58�16 years; 46% men) were sent to the ED observation unitfor the chest pain diagnostic protocol; 13% were subsequentlyadmitted.
Characteristics of Study SubjectsNinety-six patients were enrolled with consent during the
study period. Four patients did not complete the protocol andwere excluded. Reasons for exclusion included computertechnical failure and no data acquisition (1), cancellation of CT(1), treadmill testing without nuclear imaging (1), and stressecho testing (1). Additionally, 7 patients’ CT angiographyimages were uninterpretable and were also excluded. Reasons foruninterpretable images were motion artifact (3), poor contrastto noise (2), machine artifacts (1), and ECG gating deficiency(1). No patients were excluded because of positive serial ECG orcardiac marker results preceding imaging. Follow-upinformation was obtained by structured review of hospitalrecords for 30-day outcomes for all patients. Structuredtelephone interview was completed for 83 (98%) patients at amedian interval of 76 days (range 54 to 91 days); the remaining2 patients were queried in the hospital database and governmentdeath registry, with no reference outcomes identified. Both ofthese patients had normal multidetector CT and stress nuclearimaging results.
Among the 85 study patients with interpretable images, theaverage age was 49�11 years, with 53% being men (Table 1).Chest pain was the presenting chief complaint in 94% ofpatients. Admission ECG results were normal in 60% of
patients, with T-wave flattening in 34% and Q waves present in
Volume , . : February
Gallagher et al Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography
5.9% of results. Patients had an average probability of acutecardiac ischemia of 23%�14% according to Acute CardiacIschemia Time Insensitive Predictive Instrument (ACI-TIPI)scores, and patients had an average Thrombolysis in MyocardialInfarction (TIMI) risk score of 0.8 (�0.8).20-22
Main ResultsOverall, 7 patients (8%) met criteria for a final reference
outcome, all with an acute coronary syndrome defined bysignificant coronary artery stenosis. No patients had myocardialinfarction and no patients had subsequent 30-day major adversecardiac events. All 7 underwent cardiac catheterization, 6 duringthe index visit and 1 on a subsequent scheduled return visit.Their average percentage of stenosis was 88%�9%, with arange of 70% to 99%. Of these, 6 patients had single-vesselcoronary stenosis treated with percutaneous coronaryintervention, and 1 patient was treated with medical therapy.Cardiac catheterization was also performed on 5 additionalpatients, with 3 showing normal coronary arteries and 2 withmild to moderate disease.
Of the 85 study patients, both stress nuclear imaging andmultidetector CT results were negative in 66 patients (78%),with none having evidence of an acute coronary syndrome. Bothstress nuclear imaging and multidetector CT results werepositive in 6 patients (7%), with 4 having significant coronaryartery disease. Thirteen patients (15%) had discordantmultidetector CT and stress nuclear imaging results, with 5receiving cardiac catheterization and 3 of these patients havingsignificant coronary artery stenosis (Figure 1 and Table 2). Ofthe 8 discordant patients who did not undergo catheterization,
2 had 50% and 1 had 50% to 70% multidetector CT stenosis,
Volume , . : February
all with a normal stress nuclear imaging result. Additionally, 4patients had a possible reversible perfusion defect (suspicious forischemia), with less than 50% stenosis on multidetector CT,and 1 with an abnormal stress ECG result, with a normal stressnuclear imaging result and 0% stenosis on multidetector CT.Of the discordant patients who did not receive a cardiaccatheterization, none had evidence of a reference outcome onsubsequent repeated medical history, examination, ECG, or30-day medical record review (Table 2).
For the detection of the stated clinical outcomes, an acutecoronary syndrome, or significant coronary artery stenosis, thesensitivity of stress nuclear imaging was 71% (95% CI 36% to92%), and multidetector CT was 86% (95% CI 49% to 97%),whereas the specificity of stress nuclear imaging andmultidetector CT was 90% (95% CI 81% to 95%) and 92%(95% CI 84% to 96%), respectively. The negative predictivevalue for stress nuclear imaging and multidetector CT was 97%(95% CI 90% to 99%) and 99% (95% CI 93% to 100%),respectively, whereas the positive predictive values were 38%(95% CI 18% to 64%) and 50% (95% CI 25% to 75%),respectively (Table 3).
Ninety-one percent of patients received an exercise sestamibistress test, 8% received a disopyrimadole stress test, and 1%received a dobutamine sestamibi stress test. Treadmill patientsexercised for an average of 9.0�2 minutes. A maximal pulse rateresponse was not achieved in 13% of patients. Seven percent ofpatients showed ischemic changes on their stress ECG, and 26%had anginal-like symptoms during their stress test. By Dukescore, 83% of patients were low risk and 17% were intermediaterisk. Sestamibi perfusion imaging result was abnormal in 11%of patients, probably abnormal in 5%, probably normal in 34%,and normal in 51%. Of patients with a positive referenceoutcome, imaging result was abnormal or probably abnormal in5 of 7 patients (71%). Eight patients had normal perfusionimaging results but achieved a submaximal pulse rate (average74% predicted maximal pulse rate), with 1 of these havingsignificant coronary artery stenosis (Table 4).
Eighty-seven percent of patients underwent stress nuclearimaging before multidetector CT. The average pulse rate beforepreparation for multidetector CT was 74�12 beats/min. Inpreparation for multidetector CT imaging, oral atenolol(88�22 mg) was initiated in the ED in 13 patients (12%), andintravenous metoprolol (27�17 mg) was administered in themultidetector CT holding area in 75 patients (88%) before scanacquisition. There were no adverse events related to � blockadeor iodinated contrast administration. Coronary calcium scorewas greater than 100 in 9 (11%) patients, with 2 of these havingsignificant coronary artery stenosis. Of the seventy-six patients(89%) with calcium scores less than 100, 5 had significantcoronary artery stenosis. Multidetector CT found greater than50% stenosis in 11 patients (13%), with 5 of these havingsignificant coronary artery disease (Figures 2 and 3). Of the 7patients with 25% to 50% stenosis, 1 had significant coronary
artery disease, with 99% stenosis of the mid right coronary
Annals of Emergency Medicine 129
Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography Gallagher et al
artery found on cardiac catheterization. The multidetector CTimage quality in this patient was limited by poor contrastopacification and significant background noise (Figure 4). Ofthe 65 patients (76%) with less than 25% stenosis onmultidetector CT, none were found to have significant coronaryartery disease (Figure 2). In comparing multidetector CT withcardiac catheterization, of 12 patients who had cardiaccatheterization, 7 had coronary stenosis greater than 70%, withCTA being positive (�50% stenosis) in 6 of these 7 cases andfalsely negative in 1 (discussed above). Of the 5 catheterizationcases that did not have greater than 70% coronary stenosis (or30-day clinical outcomes), CTA result was positive in 3 andnegative in 2.
LIMITATIONSThis study has several important limitations to consider.
Using the results of the multidetector CT and stress nuclearimaging to determine who would undergo angiography orwho would have a follow-up appointment may haveintroduced an incorporation bias. Because we did not feel itethical to withhold test results from those treating patients,knowledge of multidetector CT or stress nuclear imagingresults may have also introduced a referral bias. Becausecardiac catheterization did not occur in all low-risk patients,there is an inherent referral bias based on positivenoninvasive diagnostic test results. This bias has been well
reported for both stress ECG and stress nuclear testing.23,24
130 Annals of Emergency Medicine
If the accuracy of multidetector CT is comparable tocoronary catheterization, as has been suggested inpreliminary studies, then performing both stress nuclearimaging and multidetector CT on all patients may havedecreased referral bias.6 Correcting for this bias woulddecrease the sensitivity of each test. Attempts to minimizethis were made using 30-day clinical outcomes. The use of asurrogate clinical marker, instead of an invasive criterionstandard, is another limitation. However, this approach isreasonable in studies of diagnostic tests in which theprevalence of disease is low.25 Limiting the study to patientswho agree to participate and to hours in which researchassistants could enroll patients made the study population aconvenience sample, and this may have introduced samplingbias. However, in comparing all chest pain observation unitpatients with study patients, the 2 groups were comparablein terms of their admission rate (13% versus 14%) andpercentage of male patients (46% versus 53%), with overlapin age distributions (58�16 years versus 49�11 years).Exclusion of patients with uninterpretable images may haveoverestimated the diagnostic accuracy for multidetector CT.Our rate of uninterpretable studies (8%), however, isconsistent with the average rates cited in the literature.6-9
The relatively small sample size of this study limits the abilityto fully address safety. Feasibility was not addressed by thisstudy; a cohort study would be needed for that. Finally,
ents
ents
7 excluded
PositiveMDCT
12 (14.1%)
NegativeSNI
6 (7.1%)
PositiveSNI
6 (7.1%)
ACSACS
2 4
+ -
4 2
+ -
lear imaging in the detection of an acute coronaryany segment or CAC �400. Positive stress nuclearmultidetector computed tomography; SNI, stress nuclear
Pati
Pati
nuc% inDCT,
issues of cost, length of stay, clinical impact, or long-term
Volume , . : February
Gallagher et al Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography
prognosis was not addressed in this study. These issues andthe relative clinical value of multidetector CT, relative totraditional stress nuclear imaging, would be better answeredby a randomized controlled study. These are fertile areas for
Table 2. Summary of patients with discordant test results.
Age,Sex
TIMIRisk
ACI-TIPI
StressECG
Stress NuclearResult
PMDuk
MDCT negative, SNI positive (no cardiac catheterization)44 F 1 3 Neg Possible
MDCT positive, SNI negative (no cardiac catheterization)47 M 1 N/A§ Neg Normal 103, Lo
63 F 0 36 Neg Normal 107, Lo
63 F 1 36 Neg Normal 92, Low
MDCT positive, SNI negative (cardiac catheterization)60 M 2 43 Neg Normal 73, Low
48 M 0 7 Pos ProbableNormal
88, Low
40 M 1 7 Neg Normal 99, Low
LAD, Left anterior descending; RCA, right coronary artery; PDA, posterior descend*Reevaluation: return visit for medical history or examination findings suggestivetesting not indicated in any cases.†Underwent cardiac catheterization 1 year after study enrollment because of recu‡Catheterization results.§ACI-TIPI score not applicable, because of baseline ECG repolarization abnormalit
Table 3. Myocardial perfusion imaging and multidetector CTaccuracy (n�85).
ImagingMethod
Sensitivity(n)
Specificity(n) PPV (n) NPV (n)
Stressnuclearimaging(95% CI)
71% (5/7)(36%–92%)
90% (70/78)(81%–95%)
38% (5/13)(18%–64%)
97% (70/72)(90%–99%)
MultidetectorCT (95%CI)
86% (6/7)(49%-97%)
92% (72/78)(84%-96%)
50% (6/12)(25%-75%)
99% (72/73)(93%-100%)
future studies.
Volume , . : February
DISCUSSIONThis study shows that multidetector CT has accuracy that is
comparable to that of stress nuclear imaging for the detection ofan acute coronary syndrome in ED low-risk chest pain patients.As such, it appears to be a reasonable alternative to stress nuclearimaging in a “low-risk” ED chest pain population after negativeserial ECG and cardiac marker results.
During the last 2 decades, there has been a progressiveevolution in the management of ED chest pain patients at riskfor an acute coronary syndrome. Traditionally, these patientswere admitted to an inpatient hospital bed for 2 to 3 days tocomplete a diagnostic evaluation that usually consisted of serialenzyme levels, ECG results, and stress testing. Subsequently,chest pain observation units were developed to provideaccelerated diagnostic protocols for chest pain patients.4,26,27
Compared with traditional admission, these protocols have beenshown to decrease the length of stay to roughly half a day,
decrease cost, and lead to fewer missed myocardial infarctions,
Annals of Emergency Medicine 131
Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography Gallagher et al
with comparable clinical outcomes.5,28-32 With this approach,roughly 80% of these patients are discharged home aftercompletion of their protocol. The advent of multidetector CTcoronary angiography offers to take this evolution 1 step fartherwith a technology that appears to be as accurate as stress nuclearimaging.
Recently, there has also been an evolution in cardiac CTimaging. Early studies suggested that normal to minimalcoronary artery calcification alone was sufficient to accuratelyrisk-stratify ED patients with acute chest pain, with reported
-10
0
10
20
30
40
50
60
MD CT Lum inal Stenosis (% )
Positive Cath/ACS
Positive M DCT
Negative M DCT
55
10 9
2 1
13 2 1
0% 1-25% 25-50% 50-75% 75-100% CAC>400
No
.of
Stu
dy
Pat
ien
t sN
o.o
fp
tsw
ith
Po
siti
ve
Cat
h/A
CS
8
Figure 2. Dual-axis distribution of multidetector CTstenosis percentage and patients with a positivecatheterization (number at end of bar).
*Includes 1 patient with normal perfusion and positive clinical and electrocardiog†Includes 8 patients with normal perfusion and submaximal pulse-rate response.
sensitivities of 96% to 100% for the detection of significant
132 Annals of Emergency Medicine
cardiac events.10 Five of the 7 patients in the current study withcatheterization-proven severe coronary artery disease hadcoronary artery calcification scores less than 100, and 2 had nodetectable coronary calcium (ie, CAC�0). Not surprisingly,these 5 patients were young (average age 46.8 years) but withmultiple coronary risk factors. These findings suggest that theaddition of multidetector CT coronary angiography to CACscore alone is an important and necessary adjunct for the triageof this low-risk, often younger, patient population. Early studiesof multidetector CT were done using 16-slice CT machines andhave been followed by studies using 64-slice CT machines. Werecently reported the accuracy of 64-slice multidetector CTcompared with invasive coronary angiography in the detectionof greater than 50% stenosis and found multidetector CT to be95% sensitive and 90% specific.6 In that study population,significant stenosis was present in 57% of patients. Bayes’theorem has been used to describe how a test may performdifferently, depending on the prevalence of disease in thepopulation to which it is applied. According to this principle,the current study adds to the evaluation of multidetector CTimaging by evaluating its diagnostic performance in a “low-prevalence” population. In the current study, the prevalence ofcoronary stenosis was 8%, and multidetector CT was able toidentify 86% of patients (6/7) with disease, with a specificity of92%, which suggests that multidetector CT performs well inboth high- and low-prevalence populations.
The accuracy of stress nuclear imaging in our study(sensitivity 71%, specificity 92%) was slightly lower than thatreported in the literature. Previous reports have demonstratedthat the selective addition of stress nuclear imaging to a
ositive, N � 7 (8%) Reference Negative, N � 78 (92%)
comprehensive ED chest pain evaluation protocol excludes
Volume , . : February
and
Gallagher et al Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography
acute coronary syndrome with a sensitivity of 99% andspecificity of 87%.5 Our diagnostic accuracy, however, iscomparable to a recent stress nuclear imaging meta-analysis of19 studies that reported a sensitivity and specificity of 87% and73%, respectively, for detection of greater than 50%catheterization-proven stenosis.15 Our lower sensitivity is likelyattributed to the overall low prevalence of disease (7 ACS-positive cases), as well as our decision to define patients withnormal stress perfusion (without ECG or clinical evidence ofischemia) and submaximal pulse rate response as “probablynormal” studies. One of the 2 false-negative stress nuclearimaging results occurred in a patient who achieved 74% peakpredicted pulse rate (while receiving long-term �-blockertherapy), whereas the other occurred in a patient with stenosis(70%) localized to a large diagonal branch vessel.
Multidetector CT and stress nuclear imaging technologiesdiffer in that stress nuclear imaging provides “physiologic”perfusion information intended to identify anatomic coronarystenosis, whereas multidetector CT shows “anatomic” coronarystenosis without perfusion data. Recent developments suggestthat multidetector CT may ultimately be capable of providingboth perfusion and anatomic information.33 However, we didnot find that the lack of perfusion information limited theability of multidetector CT to identify patients with an acutecoronary syndrome. Additionally, multidetector CT is capableof identifying aortic dissection or pulmonary embolus bymodifying the scan protocol and contrast dosing.34 In fact, this“triple rule-out” protocol was used on a discretionary basis in 35of the 85 study patients; none were found to have major acutepathology. For suboptimal image quality, stress nuclear imagingoffers an advantage because patients may be reimaged while theisotope remains active, with different positioning andacquisition techniques to overcome poor-quality images.Unfortunately, reimaging multidetector CT patients involves
Figure 3. True positive multidetector CT example. Severe didiagonal coronary artery (arrow) noted on CT angiogram (A)
repeated doses of contrast and radiation, making reimaging less
Volume , . : February
feasible. It is hoped that, as multidetector CT technologymatures, this will be less of a problem.
Although others have reported good diagnostic outcomes inchest pain protocols in which a stress ECG is done withoutimaging, we chose to compare multidetector CT against both stressECG and stress nuclear information.35 In our study, performing astress ECG without imaging would have missed 5 of the 7 patientsfound to have an acute coronary syndrome. Because this is a newtechnology, we also thought that it was safest to perform serialcardiac marker and ECG tests in combination with imaging, ratherthan imaging alone. Further studies may disprove the need for thiscombination. We concur with the American Heart Associationadvisory committee about the need for testing beyond serialmarkers in a low-risk chest pain population.36 We had previouslyfound that serial cardiac marker levels identified only 19% of acutecoronary syndrome patients in this “low-risk” population, whereasstress imaging identified 60% of these acute coronary syndromepatients.37 Recent studies suggest that cardiac marker protocols of 2hours or 90 minutes are effective in ruling out myocardialnecrosis.5,38 It is possible that combining these rapid markerprotocols with multidetector CT will allow the emergencyphysician to provide a completed evaluation in 2 to 5 hours whilethe patient remains within the ED. This combined approach meritsfurther study.
Although our reference outcome was defined as any patient withan acute coronary syndrome and 30-day major adverse cardiacevents, all positive reference outcomes were cardiac catheterizationstenosis greater than 70%. Because no patients had 50% to 70%stenosis, using a catheter stenosis greater than 50% would not havechanged our results. We assume that this stenosis caused thesymptoms that prompted the patient to seek treatment andaccounted for the patient’s unstable angina. In this study, therewere also 8 patients with discordant multidetector CT and stressnuclear imaging results who did not undergo cardiac
e involving the left anterior descending (arrowheads) andcardiac catheterization (B).
seas
catheterization according to their overall clinical characteristics.
Annals of Emergency Medicine 133
Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography Gallagher et al
None of these patients had highly abnormal multidetector CT orstress nuclear imaging results, and none were found to haveevidence of a missed diagnosis on follow-up.
From the results of this study and previous studiescomparing multidetector CT with cardiac catheterization, itseems prudent to consider further testing, such as additionalstress nuclear imaging, for any patient with greater than 25%stenosis on multidetector CT or any patient with a coronaryartery calcium score greater than 100.6 It also seems reasonableto consider cardiac catheterization of any patient with greaterthan 70% stenosis on multidetector CT.
In defining an acute coronary syndrome, we chose to useobjective findings, such as coronary stenosis, rather than abroader clinical diagnosis of unstable angina. This approach hasbeen encouraged in a consensus document on standardized
Figure 4. False-negative multidetector CT. A, Artifact noted osevere distal right coronary artery stenosis. C, reevaluationnoncalcified and calcified stenosis.
reporting criteria for studies of an acute coronary syndrome.14
134 Annals of Emergency Medicine
This approach may have resulted in fewer patients with an acutecoronary syndrome because some patients diagnosed with“unstable angina” are diagnosed according to their medicalhistory alone. However, we believed that the combination ofmultidetector CT and stress nuclear imaging results withrigorous 30-day outcomes provided a more reliable studydiagnosis.
In conclusion, in this prospective study of ED low-risk chestpain observation unit patients, multidetector CT performed atleast as well as stress nuclear imaging in the detection andexclusion of an acute coronary syndrome after myocardialinfarction had been ruled out.
Supervising editors: Judd E. Hollander, MD; Michael L.
Gallagher et al Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography
Author contributions: MJG, MAR, GLR, JAG, and BO conceivedand designed the study. MJG, GLR, JAG, and WWO obtainedinternal funding to support the study. All were engaged inpatient enrollment. MJG, MAR, GLR, and BO were involved indata analysis and article preparation. All authors wereinvolved in final article revisions. MR takes responsibility forthe paper as a whole.
Funding and support: Sponsored in part by a grant from theMinestrelli Advanced Cardiac Research Imaging Center, RoyalOak, MI.
Publication dates: Received for publication March 7, 2006.Revision received June 16, 2006. Accepted for publicationJune 20, 2006. Available online September 15, 2006.
Presented in part at the American Heart Association, ScientificSessions, November 2005, Dallas, TX.
Reprints not available from the authors.
Address for correspondence: Michael A. Ross, MD,Department of Emergency Medicine, William BeaumontHospital, 3601 West 13 Mile Road, Royal Oak, MI 48073;248-898-3080, fax 248-898-2015; [email protected].
REFERENCES1. McCaig LF, Burt CW. National Hospital Ambulatory Medical Care
Survey 2003: Emergency Department Summary: Advance Datafrom Vital and Health Statistics, No. 358. Hyattsville, MD:National Center for Health Statistics; 2005.
2. Pope JH, Aufderheide TP, Ruthazer R, et al. Missed diagnoses ofacute cardiac ischemia in the emergency department. N EnglJ Med. 2000;342:1163-1170.
3. Karcz A, Korn R, Burke MC, et al. Malpractice claims againstemergency physicians in Massachusetts: 1975-1993. Am JEmerg Med. 1996;14;341-345.
4. Zalenski RJ, Rydman RJ, Ting S, et al. A national survey ofemergency department chest pain centers in the United States.Am J Cardiol. 1998;81:1305-1309.
5. Fesmire F, Hughes A, Fody E, et al. The Erlanger chest painevaluation protocol: a one-year experience with serial 12-lead ECGmonitoring, two-hour delta serum marker measurements, andselective nuclear stress testing to identify and exclude acutecoronary syndromes. Ann Emerg Med. 2002;40:584-594.
6. Raff GL, Gallagher MJ, O’Neill WW, et al. Quantitative andqualitative accuracy of non-invasive coronary angiography using64-slice helical computed tomography. J Am Coll Cardiol. 2005;46:552-557.
7. Mollet NR, Cademartiri F, van Mieghem CAG, et al. High-resolutionspiral computed tomography coronary angiography in patientsreferred for diagnostic conventional coronary angiography.Circulation. 2005;112:2318-2323.
8. Leschka S, Alkadhi H, Plass A, et al. Accuracy of MSCT coronaryangiography with 64-slice technology: first experience. Eur Heart J.2005;26:1482-1487.
9. Leber AW, Knez A, von Ziegler F, et al. Quantification ofobstructive and nonobstructive coronary lesions by 64-slicecomputed tomography: a comparative study with quantitativecoronary angiography and intravascular ultrasound. J Am CollCardiol. 2005;46:147-154.
10. Georgiou D, Budoff MJ, Kaufer E, et al. Screening patients with
chest pain in the emergency department using electron beam
Volume , . : February
tomography: a follow-up study. J Am Coll Cardiol.2001;38:105-110.
11. Ross MA, Compton S, Richardson D. The use and effectivenessof an emergency department observation unit for elderly patients.Ann Emerg Med. 2003;41:668-677.
12. Canto JG, Shlipak MG, Rogers WJ. Prevalence, clinicalcharacteristics, and mortality among patients with myocardialinfarction presenting without chest pain. JAMA. 2000;283:3223-3229.
13. Reilly BM, Evans AT, Schaider JJ, et al. Impact of a clinicaldecision rule on hospital triage of patients with suspected acutecardiac ischemia in the emergency department. JAMA. 2002;288:342-350.
14. Hollander JE, Blomkalns AL, Brogan GX, et al. Standardizedreporting guidelines for studies evaluating risk stratification ofemergency department patients with potential acute coronarysyndromes: Multidisciplinary Standardized Reporting Criteria TaskForce. Ann Emerg Med. 2004;44:589-598.
15. Shaw LJ, Berman DS. Sequential single-photon emissioncomputed tomography myocardial perfusion imaging. Am JCardiol. 2005;96(suppl):28J-39J.
16. Cerqueira MD, Weissman NJ, Dilsizian V, et al. AHA scientificstatement: standardized myocardial segmentation andnomenclature for tomographic imaging of the heart: a statementfor healthcare professionals from the Cardiac Imaging Committeeof the Council on Clinical Cardiology of the American HeartAssociation. Circulation. 2002;105:539-542.
17. Berman DS, Hachamovitch R, Kiat H, et al. Incremental value ofprognostic testing in patients with known or suspected ischemicheart disease: a basis for optimal utilization of exercisetechnetium-99m sestamibi myocardial perfusion single-photonemission computed tomography. J Am Coll Cardiol. 1995;26:639-647.
19. Austen WG, Edwards JE, Frye RL, et al. A reporting system onpatients evaluated for coronary artery disease: report of the AdHoc Committee for Grading of Coronary Artery Disease, Councilon Cardiovascular Surgery. Circulation. 1975;51:5-40.
20. Selker HP, Zalenski RJ, Antman EM, et al. An evaluation oftechnologies for identifying acute cardiac ischemia in theemergency department: executive summary of a national heartattack alert program working group report. Ann Emerg Med.1997;29:1-88.
21. Selker HP, Griffith JL, D’Agotion RB. A tool for judging coronarycare unit admission appropriateness, valid for both real-time andretrospective use: a time-insensitive predictive instrument (TIPI)for acute cardiac ischemia: a multicenter study. Med Care. 1991;29:610-627.
22. Antman E, Cohen M, Bernink P, et al. The TIMI risk score forunstable angina/non-ST elevation MI: a method forprognostication and therapeutic decision making. JAMA. 2000:284:835-842.
23. Gibbons R, Balady G, Bricker JT, et al. ACC/AHA 2002 guidelineupdate for exercise testing: summary article: a report of theAmerican College of Cardiology/American Heart Association TaskForce on Practice Guidelines. Circulation. 2002;106:1883-1892.
24. Klocke FJ, Baird MG, Bateman TM, et al. ACC/AHA/ASNCguidelines for the clinical use of cardiac radionuclide imaging:executive summary: a report of the American College ofCardiology/American Heart Association Task Force on PracticeGuidelines (ACC/AHA/ASNC Committee to Revise the 1995Guidelines for the Clinical Use of Cardiac Radionuclide
Imaging).Circulation. 2003;108:1404-1418.
Annals of Emergency Medicine 135
Diagnostic Accuracy of 64-Slice Computed Tomography Coronary Angiography Gallagher et al
26. Lee T, Juarez G, Cook E, et al. Ruling out acute myocardialinfarction: a prospective multicenter validation of a 12-hourstrategy for patients at low risk. N Engl J Med. 1991;324:1239-1246.
27. Gibler W, Runyon J, Levy R, et al. A rapid diagnostic andtreatment center for patients with chest pain in the emergencydepartment. Ann Emerg Med. 1995;25:1-8.
28. Gomez MA, Anderson JL, Karagouris LA, et al. An emergencydepartment-based protocol for rapidly ruling out myocardialischemia reduces hospital time and expense: results of arandomized study (ROMIO). J Am Coll Cardiol. 1996;28:25-33.
29. Roberts RR, Zalenski RJ, Mensah EK, et al. Costs of anemergency department-based accelerated diagnostic protocolversus hospitalization in patients with chest pain: a randomizedcontrolled trial. JAMA. 1997;278:1670-1676.
30. Zalenski RJ, McCarren M, Roberts R, et al. An evaluation of achest pain diagnostic protocol to exclude acute cardiac ischemiain the emergency department. Arch Intern Med. 1997;157:1085-1091.
31. Farkouh ME, Smars P, Reeder GS, et al. A clinical trial of a chestpain observation unit for patients with unstable angina. N Engl
J Med. 1998;339:1882-1888.
136 Annals of Emergency Medicine
32. Graff L, Delara J, Ross M, et al. Chest Pain Evaluation Registry(CHEPER) Study: impact on the care of the emergencydepartment chest pain patient. Am J Cardiol. 1997;80:563-568.
33. George RT, Cordeiro MA, Silva C, et al. Multi-detector computedtomography quantifies myocardial perfusion during adenosinestress. Circulation. 2005;112(suppl II):465-466.
34. Vrachliotis TG, Bis KG, Haidary A, et al. Enhancement ofcoronary, aortic and pulmonary vasculature using a biphasicsingle injection 64-slice CT-angiography protocol in emergencydepartment patients with atypical chest pain. Radiology. In press.
35. Polanczyk CA, Johnson PA, Hartley LH, et al. Clinical correlatesand prognostic significance of early negative exercise tolerancetest in patients with acute chest pain seen in the hospitalemergency department. Am J Cardiol. 1998;81:288-292.
36. Stein RA, Chaitman BR, Balady GJ, et al. Safety and utility ofexercise testing in emergency room chest pain centers: anadvisory from the Committee on Exercise, Rehabilitation, andPrevention, Council on Clinical Cardiology, American HeartAssociation. Circulation. 2000;102:1463-1467.
37. Trappe K, Jackson RE, Ross M. A significant proportion ofpatients admitted from a chest pain observation unit have cardiacpathology [abstract]. Ann Emerg Med. 2003;42:S8-S9.
38. McCord J, Nowak RM, McCullough PA, et al. Ninety-minuteexclusion of acute myocardial infarction by use of quantitativepoint of care testing of myoglobin and troponin I. Circulation.
2001;104:1483-1488.
Did you know?
You can personalize the new Annals of Emergency Medicine Web site to meet your individual needs.
Visit www.annemergmed.com today to see what else is new online!
