ACCURACY OF CT ANGIOGRAPHY AND SPECT MYOCARDIAL PERFUSION IMAGING FOR THE DIAGNOSIS OF CORONARY ARTERY DISEASE

Abstract

Background—

Establishing the diagnosis of coronary artery disease (CAD) in symptomatic patients allows appropriately allocating preventative measures. Nuclear myocardial perfusion imaging (SPECT-MPI) is frequently utilized for the evaluation of CAD but coronary CT angiography (CTA) has emerged as a valid alternative.

Methods and Results—

We compared the accuracy of SPECT-MPI and CTA for the diagnosis of CAD in 391 symptomatic patients who were prospectively enrolled in a multicenter study after clinical referral for cardiac catheterization. Area under the receiver-operating-characteristic curve (AUC) was used to evaluate the diagnostic accuracy of CTA and SPECT-MPI for identifying patients with CAD defined as presence of ≥1 coronary artery with ≥50% lumen stenosis by quantitative coronary angiography.

Sensitivity to identify patients with CAD was greater for CTA than SPECT-MPI (0.92 vs. 0.62, respectively, p<0.001), resulting in greater overall accuracy (AUC 0.91 [95% confidence interval 0.88–0.94]) vs. (0.69 [0.64–0.74], p< 0.001). Results were similar in patients without prior history of CAD (AUC 0.92 [0.89–0.96] vs. 0.67 [0.61–0.73], p< 0.001), and also for the secondary endpoints of ≥70% stenosis and multivessel disease, as well as subgroups, except for patients with calcium score ≥400 and those with ‘high risk’ anatomy in whom overall accuracy was similar because CTA’s superior sensitivity was offset by lower specificity in these settings. Radiation doses were 3.9 mSv for CTA and 9.8 for SPECT-MPI (p<0.001).

Conclusions—

CT angiography is more accurate than SPECT-MPI for the diagnosis of CAD as defined by conventional angiography and may be underutilized for this purpose in symptomatic patients.

Clinical Trial Registration—

URL: http://www.clinicaltrials.gov. Unique identifier: NCT00934037.

Each year 15 million Americans seek medical attention with symptoms concerning for coronary artery disease (CAD).¹ While there are considerations of requiring evidence of hemodynamic compromise associated with CAD to warrant revascularization in patients with stable symptoms², current practice guidelines rest the diagnosis of CAD and the associate appropriateness of preventative measures on the presence of anatomically defined coronary arterial stenoses on cardiac catheterization.^3–5 Stress testing – commonly combined with single photon emission computed tomography acquired myocardial perfusion imaging (SPECT-MPI) - functions as a gate keeper in patients with angina-like symptoms because of the risks and costs associated with cardiac catheterization.^4,5 Despite the use of conventional noninvasive testing prior to catheterization (including SPECT-MPI in 78%), an analysis of almost 400,000 coronary angiograms in patients without prior history of CAD revealed no obstructive disease in more than 60%, resulting in unnecessary risks and costs.^6,7 Conversely, failing to identify CAD may result in omitting preventative measures in patients who are at increased risk from adverse events. Therefore, accuracy of noninvasive testing for the diagnosis of CAD is an important public health concern. Several recent multicenter studies comparing SPECT-MPI and cardiac magnetic resonance imaging (MRI) for identifying patients with CAD on cardiac catheterization found only modest diagnostic accuracy (67–69%) for SPECT-MPI,^8–10 in contrast to results from specialized, single center reports.¹¹ On the other hand, CT coronary angiography (CTA) is an emerging noninvasive imaging modality that has yielded high diagnostic accuracy (93–96%) compared to cardiac catheterization in several multicenter and numerous single center studies^12–16 Several small investigations found greater diagnostic accuracy of CTA in direct comparison to SPECT-MPI, but a number of methodological concerns limit conclusions.^17–20 No data are available from a direct comparison of CTA and SPECT-MPI for detecting CAD in prospective cohorts using independent core laboratories.

The purpose of this study was to directly compare the diagnostic accuracy of CTA with SPECT-MPI for detecting CAD in symptomatic patients as defined by the gold standard of cardiac catheterization.

METHODS

Study Design

We studied patients who were prospectively enrolled in the Coronary Artery Evaluation Using 320-Row Detector Computed Tomography Angiography (CORE320) trial, which was performed at 16 tertiary medical centers in 8 countries to evaluate the diagnostic accuracy of cardiac CT for detecting coronary artery stenoses by conventional angiography with associated myocardial perfusion abnormalities by SPECT-MPI. The study is registered at www.clinicaltrials.gov, identifier: NCT00934037. Details of the CORE320 study design have been published.²¹ The primary sponsor of the CORE320 study, Toshiba Medical Systems, was not involved in any stage of the study design, data acquisition, data analysis, or manuscript preparation. All participating centers received study approval from their local institutional review boards in addition to approval from a central IRB at the coordinating site. All patients gave written informed consent prior to participation.

Patient Population

The patient population of the CORE320 study has been described in detail elsewhere.²² In brief, 444 study participants were selected for the study according to the following criteria: patients who were between 45–85 years of age, with symptoms suggestive of CAD and referral for conventional coronary angiography to define coronary anatomy. Patients were not eligible if they had history of cardiac surgery or cardiac device implantation, acute coronary syndrome, allergy to iodinated contrast or contrast induced nephropathy, multiple myeloma, organ transplantation, renal insufficiency, atrial fibrillation, New York Heart Association class III or IV heart failure, aortic stenosis, contraindications to vasodilator application, profound bradycardia, evidence of excessive radiation exposure from medical imaging in the recent past, percutaneous coronary intervention within the past 6 months, intolerance to beta-blockers, or a body-mass index greater than 40. For this analysis, we specified the important subgroup of patients without history of coronary artery disease, e.g., prior myocardial infarction or coronary revascularization, to test the utility of CT and SPECT-MPI in patients for whom a diagnosis of CAD had not been established. Pre-test probability of CAD was determined by the method of Morise et al.²³

CT Image Acquisition and Data Analysis

Methods for CT image acquisition in the CORE320 study have been described in detail elsewhere.²⁴ In brief, patients underwent coronary calcium scoring and CT coronary angiography using 320-detector row scanners with a slice thickness of 0.5 mm (Aquilion, Toshiba Medical Systems). The use of a wide detector system was mandated for the effective use of CT myocardial perfusion imaging for the primary study objective of CORE320. Total calcium score was determined by the Agatston method. For CTA, prospective ECG gating was used, with 120-kV tube voltage, 350–375 msec gantry rotation, and tube current adapted to patient body mass index (BMI) and gender. Radiation dose (mSv) for CT angiography (excluding calcium scanning) was estimated using the dose-length-product multiplied by a K factor of 0.014.²⁵ Radiation dose for contrast bolus tracking was not separately recorded but estimated using the mean total radiation doses for the entire CT examination subtracting the doses for calcium scanning, CT angiography, CT perfusion imaging, with the resultant value divided by 2 assuming equal doses for the two angiographic acquisitions.²⁶ Total dose for CT angiography was derived from adding doses from bolus tracking estimate and CT coronary angiography. The contrast medium iopamidol (Isovue 370, Bracco Diagnostics) was injected intravenously at rates between 4–5 ml/s. Beta blockers were given if the resting heart rate was above 60 beats per minute. Raw image data sets from all acquisitions were analyzed by an independent core laboratory. Two blinded observers visually graded each of 19 coronary artery segments (1.0 mm or more in diameter) according to an ordinal stenosis scale.²¹ Then, segments with at least one visible stenosis of 30% or more were manually quantified with the use of commercially available software (Vitrea2 version 5.0.0.0, Vital Images). Readers considered a ≥50% stenosis by quantitative assessment as diagnostic for obstructive CAD (= abnormal study). In heavily calcified segments, absence of a residual artery lumen at the lesion site (obscured by calcium) was recorded as obstructive CAD.²⁷ Stented segments were evaluated for in-stent obstruction following the same principles as in native segments. Segments with poor image quality – typically affecting smaller, distal segments – were recorded as negative for stenoses if there was no evidence for any atherosclerotic plaque, e.g., calcification, in the entire segment. The presence of atherosclerotic plaque in a segment with poor image quality prompted recording as positive for obstructive CAD. Disagreements on the presence and severity of percent diameter stenoses were resolved by consensus. No attempts were made to match secondary endpoints by the gold standard, e.g., multivessel disease or high grade stenoses, which derived from post hoc decisions.

Nuclear Myocardial Perfusion Imaging

Details on the SPECT-MPI image acquisition in the CORE320 study have been published.^22,28 Myocardial perfusion imaging was performed using SPECT. All imaging was performed within 60 days of cardiac catheterization either clinically driven (N=160) or as part of the research protocol (N=261). All SPECT-MPI cameras used in the study were required to undergo accreditation for quality assurance before commencement and throughout the enrollment period. The SPECT-MPI qualification involved evaluation of gamma camera performance and image quality. To account for variability in imaging equipment and image-acquisition techniques, the nuclear imaging core laboratory evaluated images for quality control following guidelines of the American Society of Nuclear Cardiology.²⁹ Attenuation correction was not available at most sites and thus was not utilized for this analysis. Whole body effective radiation dose was estimated using the administered radiopharmaceutical dose and the methodology of ICRP-80.³⁰ Myocardial territories were analyzed by SPECT-MPI for rest and stress myocardial perfusion abnormalities with a severity and reversibility-scored, 4-point scoring system using a 13-territory model.³¹ The summed stress score (SSS) was defined as the sum of abnormal myocardial segments at stress phase.²² Importantly, artifacts did not contribute to the summed stress score (SSS) and therefore a SSS ≥1 defined an abnormal SPECT-MPI study in accordance with methods used for large multicenter studies and independent core laboratories.³²

Diagnosis of Coronary Artery Disease by Invasive Coronary Angiography

Invasive coronary angiography was clinically driven and performed using standard angiographic techniques within the 60 days following CT image acquisition. Coronary angiographic images were saved in digital imaging and communication in medicine (DICOM) format, and forwarded to an independent angiographic core laboratory for analysis. The coronary tree segmentation for invasive coronary angiography was previously described.^21,33 Quantitative coronary angiography (QCA) was performed using edge-detection techniques (CAAS II QCA Research version 2.0.1 software, PIE Medical Imaging, Maastricht, the Netherlands). The most severe stenosis within each coronary segment was analyzed with quantitative assessment performed for all stenoses which were deemed ≥30% by visual assessment. Coronary artery disease was defined according to accepted standards, i.e., the presence of at least one ≥50% stenosis by QCA (equivalent to approximately ≥70% stenosis by regular visual estimate).^34–36 Secondary end points included stenoses by QCA of ≥70% (corresponding to approximately >80–90% by standard visual estimate)^35,36 and “high-risk” anatomy defined as DUKE CAD severity index ≥56 which includes three-vessel CAD, left main stenosis of ≥50%, or two vessel CAD with involvement of the proximal left anterior descending coronary artery.^16,37

Statistical Analysis

Statistical analyses were performed in an independent core laboratory at the Johns Hopkins University Bloomberg School of Public Health using SAS 9.3 (SAS Institute). Receiver-operating-characteristic (ROC) curves were used as the measure of diagnostic accuracy to identify patients with CAD, defined as at least one ≥50% stenosis by QCA. Consistent with the principles for a patient-based analysis, neither stenosis by CTA nor myocardial perfusion defect by SPECT-MPI had to match the vessel location/territory by conventional angiography for a true positive finding, i.e., patients were correctly identified as having CAD. ROC analysis was applied to compare the diagnostic performance of CTA (percent stenosis as continuous variable) and SPECT-MPI (SSS as continuous variable) for identifying patients with ≥50% stenosis by QCA by comparing the areas under the ROC curve (AUCs). Areas under ROC curves were compared using an algorithm proposed by DeLong et al.³⁸ Point statistics were determined using pre-defined thresholds (≥50% for CTA and SSS of ≥1 for SPECT-MPI) and compared using McNemar’s test for sensitivity/specificity and GEE Wald tests for predictive values. In addition, calibration curves were provided to identify point statistics with variable diagnostic thresholds for CTA and SPECT-MPI. Pre-specified subgroup analyses for this study included patients with obesity (BMI ≥30), severe calcification (calcium score ≥400), research vs. clinically triggered SPECT-MPI, and exercise vs. pharmacologic SPECT-MPI. All tests were two-tailed, the significance threshold was p<0.05, and confidence intervals were 95%.

RESULTS

Patient and Imaging Characteristics

Details on the enrollment characteristics of the CORE320 study have been published.^22,28 A flow chart for patient enrollment is shown in Figure 1. Of the 444 participants enrolled in the CORE320 study, 418 completed myocardial perfusion studies, 444 completed CTA, and 434 completed conventional coronary angiography. This investigation for the first time reports results on all 391 patients who were included in the final CORE320 analysis since 10 patients with incomplete imaging data for CT myocardial perfusion imaging – who were excluded in prior analyses with myocardial perfusion imaging being part of the endpoints^22,28 – were included in this study because complete CTA data were available in these patients. The demographic characteristics of the study population are presented in Table 1. The median age of the participants was 62 (interquartile range [IQR] 56–68) years and 66% were men. Risk factors for CAD such as arterial hypertension, diabetes mellitus, and hyperlipidemia were highly prevalent. The majority of patients (52%) were of intermediate pretest probability for CAD and 46% were of high risk (Table 1). Coronary artery disease prevalence by QCA was 60% in the entire cohort and 48% in patients without history of CAD, consistent with an intermediate-high risk profile of the study population. Median calcium score was 162 with 125 patients (32%) having severe coronary calcification (calcium score of ≥400). Of a total of 7,429 coronary artery segments, 213 (2.9%) segments could not be evaluated by CTA because of poor image quality. These uninterpretable segments affected test results (normal vs. abnormal) in 22/391 (5.6%) patients. Most of the unevaluable segments (79%) were in the distal coronary tree.

Figure 1.

Shown is the enrollment history for the patient cohort. Of 444 patients consented, 391 patients had complete data sets from CTA, SPECT-MPI, and cardiac catheterization for the final cohort.

Table 1.

Patient Baseline Characteristics

Characteristic	N(%) or Median(IQR)
Number of patients	391
Age (years)	62 (56–68)
Gender : Male	258 (66.0)
Current smoker	66 (17.6)
Diabetes	132 (33.8)
Dyslipidemia	261 (68.1)
Family history of coronary artery disease	167 (45.4)
Medication use (percent of patients)
Salicylates	252 (64.5)
Statins	214 (54.7)
Beta blocker	212 (54.2)
Nitrates	69 (17.6)
BMI (kg/m2)	26.6 (24.2–30.1)
Angina	295 (96.4)
Typical	143 (46.7)
Atypical	152 (49.7)
Stable	262 (67.0)
Unstable	9 (2.3)
History of MI	104 (26.6)
History of PCI	113 (28.9)
Calcium score	162 (9.0–548)
Pretest likelihood of CAD
Low	5 (2)
Intermediate	160 (52)
High	141 (46)

Abbreviations: N: Numbers; IQR: interquartile range; BMI: body mass index; CAD: coronary artery disease.

Diagnostic Accuracy of CTA and SPECT-MPI for the Diagnosis of CAD

Table 2 displays the diagnostic accuracy for identifying patients with CAD by CTA and SPECT-MPI along with point statistics derived by using pre-defined thresholds in the entire cohort and in patients without prior history of CAD. Diagnostic accuracy (AUC) was greater for CTA (0.91 [95% confidence interval 0.88–0.94]) than SPECT-MPI (0.69 [0.64–0.74], p< 0.001) in the entire cohort as well as in patients without prior history of CAD (0.92 [0.89–0.96] vs. 0.67 [0.61–0.73], p< 0.001) which was driven by higher sensitivity for CTA (Table 2). Figure 2 shows the ROC curves for CTA and SPECT-MPI with their respective calibration curves to identify the sensitivity and specificity for various reader-independent thresholds and broken down for the group of patients without CAD. The curves suggest most balanced performance for CTA at a stenosis threshold of 55% yielding 88% sensitivity and 83% specificity (compared to 92% and 75%, respectively, for the pre-specified threshold of 50% stenosis) while SPECT-MPI achieves the best balance at a SSS threshold of 3.5 corresponding to 58% sensitivity and 77% specificity (compared to 62% and 68%, respectively, for the pre-specified SSS threshold of 1). Table 3 shows results for subgroup analyses, which revealed consistently greater diagnostic accuracy for CTA than SPECT-MPI in obese and non-obese patients, exercise and pharmacologic stress testing, research and clinical driven SPECT-MPI, and patients with calcium score <400. Overall diagnostic accuracy was similar among the two modalities in patients with severe coronary calcification (calcium score ≥400) driven by lower specificity by CTA in the setting of high disease prevalence (Supplemental Tables). CTA was also more accurate (greater AUC) than SPECT-MPI for the secondary endpoints of multi-vessel CAD and high grade stenosis (≥70% stenosis by QCA) (Table 4). Point statistics for all subgroup analyses are presented in Supplemental Tables.

Table 2.

Overall Diagnostic Accuracy (AUC) and Point Statistics

	All (n=391)			No prior CAD (n=245)
	CTA	SPECT	p	CTA	SPECT	p
AUC	0.91 (0.88,0.94)	0.69 (0.64,0.74)	<0.001	0.92 (0.89,0.96)	0.67 (0.61,0.73)	<0.001
Sensitivity	216/234 0.92 (0.88,0.95)	145/234 0.62 (0.55,0.68)	<0.001	107/117 0.91 (0.85, 0.96)	64/117 0.55 (0.45,0.64)	<0.001
Specificity	117/157 0.75 (0.67,0.81)	107/157 0.68 (0.60,0.75)	0.23	103/128 0.80 (0.73,0.87)	90/128 0.70 (0.62,0.78)	0.08
PPV	216/256 0.84 (0.79,0.89)	145/195 0.74 (0.68,0.80)	0.001	107/132 0.81 (0.73,0.87)	64/102 0.63 (0.53,0.72)	<0.001
NPV	117/135 0.87 (0.80,0.92)	107/196 0.55 (0.47,0.62)	<0.0001	103/113 0.91 (0.84,0.96)	90/143 0.63 (0.54,0.71)	<0.001
Disease Prevalence	0.60			0.48

Abbreviations: AUC: area under the curve; SPECT: single photon emission computed tomography; CTA: CT angiography; PPV: positive predictive value; NPV: negative predictive value; CAD: coronary artery disease.

Figure 2.

Shown are the receiver-operator-characteristic (ROC) curves describing the diagnostic performance of quantitative CT angiography (Panel A) and SPECT-MPI (Panel B) in all patients (N=391) for identifying at least one ≥ 50% coronary arterial stenosis by quantitative coronary angiography (QCA). The solid lines are the results of AUC analysis and the dotted lines are calibration curves. The calibration curves identify the sensitivity and specificity for each corresponding threshold point (or vice versa) by CTA or SPECT-MPI. To arrive at the corresponding point statistic, extend a horizontal from the threshold on the right ordinate to the merging site with the calibration line. Then, connect this merging site with a vertical line to the position of the ROC curve to identify the corresponding sensitivity and specificity. For example, a threshold of 55% by quantitative CTA yields a sensitivity of 88% and a false positive rate (1 − specificity) of 17% = 83% specificity. Similarly, point statistics can be derived for various summed stress score (SSS) thresholds by SPECT-MPI. The area under the curve (AUC) was 0.91 for CTA and 0.69 for SPECT-MPI (p<0.001). Panels C and D show ROC curves for CTA and SPECT-MPI, respectively, in patients without prior history of coronary artery disease revealing similar differences in AUC: 0.92 for CTA and 0.67 for SPECT-MPI (p<0.001).

Table 3.

Diagnostic Accuracy (AUC) in Subgroups

	All Patients			Patients without prior CAD
Groups N (all), (no prior CAD)	CTA	SPECT	p Value	CTA	SPECT	p value
Obese patients (N = 100, 72)	0.95 (0.91, 0.99)	0.65 (0.55, 0.76)	<0.001	0.95 (0.91, 1.00)	0.62 (0.49, 0.75)	<0.001
Non obese patients (N = 291, 173)	0.89 (0.85, 0.93)	0.71 (0.65, 0.76)	<0.001	0.91 (0.87, 0.96)	0.69 (0.62, 0.76)	<0.001
Calcium score ≥ 400 (N = 125, 66)	0.76 (0.61, 0.91)	0.65 (0.51, 0.79)	0.33	0.70 (0.49, 0.92)	0.71 (0.58, 0.83)	0.96
Calcium score < 400 (N = 265, 178)	0.89 (0.85, 0.93)	0.67 (0.61, 0.73)	<0.001	0.91 (0.86, 0.96)	0.60 (0.52, 0.68)	<0.001
Exercise SPECT (N = 126, 82)	0.90 (0.85, 0.96)	0.60 (0.52, 0.69)	<0.001	0.91 (0.84, 0.98)	0.61 (0.50, 0.72)	<0.001
Pharmacologic SPECT (N = 264, 162)	0.91 (0.87, 0.94)	0.73 (0.67, 0.78)	<0.001	0.93 (0.89, 0.97)	0.69 (0.62, 0.77)	<0.001
Research SPECT (N = 231,137)	0.90 (0.86, 0.94)	0.72 (0.66, 0.78)	<0.001	0.92 (0.87, 0.98)	0.66 (0.60, 0.78)	<0.001
Clinical SPECT (N = 160, 108)	0.92 (0.88, 0.97)	0.65 (0.58, 0.73)	<0.001	0.92 (0.87, 0.98)	0.69 (0.60, 0.78)	<0.001

Abbreviations: SPECT: single photon emission computed tomography; CTA: CT angiography; QCA: quantitative coronary angiography; CAD: coronary artery disease;

Table 4.

Diagnostic Accuracy (AUC) for Secondary End Points

	All Patients			Patients without prior CAD
Groups	CTA	SPECT	p Value	CTA	SPECT	p value
Multivessel CAD1	0.85 (0.81, 0.89)	0.73 (0.68, 0.78)	<0.001	0.88 (0.83, 0.93)	0.77 (0.71, 0.84)	0.01
High Grade Stenosis2	0.88 (0.85, 0.92)	0.75 (0.71, 0.80)	<0.001	0.92 (0.89, 0.96)	0.77 (0.70, 0.84)	<0.001
High Risk CAD3	0.81 (0.76, 0.85)	0.77 (0.72, 0.82)	0.25	0.86 (0.81, 0.92)	0.80 (0.73, 0.87)	0.07

¹At least two vessels with 50% or greater stenosis by QCA.

²At least one coronary lesion of 70% or greater stenosis by QCA.

³DUKE CAD severity index of 56 or greater. Abbreviations: SPECT: single photon emission computed tomography; CTA: CT angiography; QCA: quantitative coronary angiography; CAD: coronary artery disease; AUC: area under curve.

Detection of Patients with High Risk Anatomy

Of 391 patients, 111 (28%) had high risk CAD as defined by QCA. In patients with high risk CAD, the sensitivity of CTA to detect CAD was 96% (107 /111) compared to 80% (89/111) by SPECT-MPI (p<0.001). A negative CTA test was 97% predictive of absence of high risk CAD in patients vs. 89% with SPECT-MPI (p=0.006). As anticipated, specificities (47% vs. 62% for CTA and SPECT-MPI, respectively) and positive predictive values (42% vs. 46%) were low for both modalities because of the discrepant disease definition compared to the reference standard (‘any stenosis ≥50% or perfusion defect’ by CTA and SPECT-MPI, respectively, vs. ‘high risk CAD DUKE severity index’), resulting in similar overall AUC (0.81 [0.76–0.85]) vs. (0.77 [0.72–0.82], p=0.25) (Table 4). Results for patients without history of CAD (N=245, disease prevalence 23% by QCA) were similar to the entire cohort with sensitivities of 95% vs. 77% for CTA and SPECT-MPI, respectively (p=0.004) and negative predictive values of 97% vs. 91% (p=0.02), respectively (Supplemental Tables).

Radiation Doses

Median radiation dose for CT angiography was 3.9 mSv vs. 9.8 mSv for SPECT-MPI (p<0.001). Details on radiation doses for various scan conditions were previously provided.²⁶

Adverse Events

One patient experienced a hypotensive response to beta blocker application in preparation for CT scanning as well as during pharmacologic SPECT-MPI testing. The patient was released after a brief observation period.

DISCUSSION

CT coronary angiography yielded considerably greater accuracy than SPECT-MPI for establishing or excluding the diagnosis of angiographic coronary artery disease in symptomatic patients defined by the gold standard of cardiac catheterization. The diagnostic accuracy by CTA was high for the entire cohort as well as for various subgroups except for patients with severe coronary calcification and those with ‘high risk anatomy’ in whom CTA had similar overall accuracy compared to SPECT-MPI due to lower specificity. Subgroup analyses, however, should be interpreted with caution as indicated by the wide confidence intervals.

The diagnostic accuracy of CTA for detecting CAD in this study is concordant with reported data for contemporary 64-slice technology and thus is representative of current practice.^12,16,39 It is important to note that the advantage of a 320-detector scanner – as used in this study – is confined to shorter scan acquisition times while the spatial resolution is the same as with a 64-slice system, leading to comparable diagnostic accuracy.^16,40 In fact, diagnostic accuracy was greater in the CorE-64 study than in our current analysis with very similar patient populations and almost identical reading teams.^12,16 On the other hand, diagnostic accuracy was substantially lower for SPECT-MPI compared to pooled analyses from single center studies.^39,41–43 Reported modest diagnostic performances by SPECT-MPI in multi-center study comparisons to MRI or PET were similar to results in this study.^8–10,44 Our data confirm fair diagnostic accuracy of SPECT-MPI for the diagnosis of CAD using a highly skilled nuclear imaging core laboratory with multi-center experience.³²

A ‘physiologic’ test, such as SPECT-MPI, may be at a disadvantage when comparing it to an anatomic gold standard for the diagnosis of CAD. Indeed, among angiographically obstructive stenoses on invasive coronary angiography or CTA, fewer than half cause myocardial ischemia.^2,45 While the appropriateness of defining CAD by this standard can be debated, it is certain that 1) stress testing is widely used for establishing/excluding CAD in symptomatic patients using the same diagnosis criteria as used in this study and 2) the presence of coronary arterial stenoses by angiography has been firmly anchored in our current management algorithms and thus determines the appropriateness of treatment.^4,5 Nonetheless, the incremental clinical benefit of detecting or excluding CAD by CTA remains to be determined. In fact, current guidelines for the management of patients with stable CAD recommend that stress testing for diagnosis and risk stratification is the appropriate initial study, and assign CTA a complementary role for patients with equivocal stress test results.^3,4 Given the higher sensitivity and negative predictive value of CTA, as shown in this study, CTA may be an effective initial test in symptomatic low-risk patients who are likely to have normal arteries.⁴⁶ Some studies suggest considering a hemodynamic assessment of coronary artery disease for guiding the decision to revascularize coronary arteries but this concept remains in evolution.^2,47–49

The purpose of our study was to establish the comparative diagnostic accuracy for identifying CAD among SPECT-MPI and CTA but prognostic considerations should be noted. Aggregate data for both CTA and SPECT-MPI suggest high annual risk (>2%) of myocardial infarction and mortality after abnormal study results.^50,51 Conversely, annual risk of myocardial infarction and death is exceedingly low (<0.1%) with a normal CT coronary angiogram while the risk after a normal SPECT-MPI ranges between (0.6–1.8%) depending on the population studied.^50,51 The clinical significance of non-obstructive CAD has gained recent attention and is associated with approximately 1% annual risk of myocardial infarction and death.^52–54 Most recently, the SCOT-HEART trial found lower rates of myocardial infarction in patients assigned to a CTA guided vs. traditional diagnostic strategy in patients with suspected CAD.⁵⁵ On the other hand, the PROMISE study reported equipoise between management strategies based on stress testing vs. CTA in low-risk patients with suspected CAD.⁵⁶

Radiation doses to patients were more than 50% lower with CTA compared to SPECT-MPI in this study. Newer approaches to SPECT imaging, including stress-only protocols as well as improved reconstruction algorithms and hardware, are associated with substantially lower radiation doses than those in this study.^57–59 However, these techniques are not widely applied at this time. Conversely, improvements in CTA technology, e.g., iterative image reconstruction etc., have yielded further reductions in radiation doses since commencement of this study.^60,61 Nonetheless, average radiation doses associated with CTA are considerably higher in clinical practice (≈ 11 mSv) than those achieved in the current study as we used commercial technology that may not be available in the community.^62–64 Cost considerations are also important. Preliminary data suggested higher rates of downstream testing and associated costs after CTA compared to SPECT-MPI - which are driven primarily by the increased frequency of revascularization after CTA compared to SPECT MPI.^65,66 However, cost analyses in the PROMISE study did not find differences between these two strategies.⁶⁷ Prospective investigations are needed how to utilize the information by both modalities for optimal patient management.

Study Limitations

All patients in this cohort were referred for cardiac catheterization with clinical suspicion of CAD and characterized as a population of intermediate-high risk profile. While intermediate risk patients benefit most from testing,⁶⁸ our results may not apply to low risk patients undergoing imaging. Referral bias is a concern in patients with prior test results (typically stress testing) scheduled for cardiac catheterization which may increase sensitivity but decrease specificity.⁶⁹ Nevertheless, we did not find differences in diagnostic accuracy for CTA and SPECT-MPI in patients with clinical and research imaging studies. Attenuation correction was not routinely used for SPECT-MPI acquisition which has shown to improve specificity.^70,71 However, its use is still limited - as shown in our study - and thus our results reflect current clinical practice. Furthermore, attenuation correction has not been shown to improve SPECT-MPI sensitivity for detecting CAD in patients, which largely accounted for the difference in diagnostic accuracy between the two modalities. Lastly, while we used the widely accepted standard for the definition of CAD, recent data raise the question if CAD defined by anatomic criteria is sufficient for patient management but alternative standards, e.g., fractional flow reserve, were not widely available for this study.²

Conclusions

CT coronary angiography yields greater accuracy than nuclear stress myocardial perfusion imaging for identifying patients with coronary artery disease as defined by invasive angiography, primarily driven by higher sensitivity of CT to identify anatomic stenoses. In appropriately selected patients with suspected coronary artery disease, CT angiography is an effective option among non-invasive imaging modalities to establish or exclude the diagnosis of coronary artery disease.