We studied 294 consecutive patients without known CAD—defined by history of myocardial infarction, coronary revascularization, or presence of angiographically documented coronary stenoses of ≥50% luminal diameter by invasive coronary angiography—participating in the Cedars-Sinai Cardiac Imaging Database Registry and undergoing both CCTA and SPECT-MPI for suspected ischemic heart disease at Cedars-Sinai Medical Center within 6 months. These patients did not experience change in symptoms or coronary revascularization between the examinations. Two patients with excessive motion artifact on CCTA were excluded. Pre-test likelihood of CAD before the first test performed was determined according to the Diamond and Forrester method using <15%, >85%, and 15-85% for low, high, and intermediate pre-test likelihood, respectively.13,14 The study protocol was approved by the Cedars-Sinai Institutional Review Board, and all patients included in the study provided informed consent for the use of their clinical and imaging data.

CCTA was performed on a SOMATOM Definition dual-source computed tomography scanner (Siemens Medical Systems, Forchheim, Germany) which employs a flying focus along the z-axis with two different focal spots, acquiring 64 overlapping .6-mm image slices with a temporal resolution of 83 ms.15 The CCTA protocol has been previously described in detail.16 Unless contraindicated, a sublingual spray of .4 mg nitroglycerin (Sciele Pharma, Alpharetta, Georgia) was administered and 80 mL of intravenous contrast (Omnipaque or Visipaque, GE Healthcare, Princeton, New Jersey) was injected into the antecubital vein at 5 mL · s⁻¹, followed by 80 mL of saline at 5 mL · s⁻¹. ECG-gated scanning from 1 cm below the tracheal bifurcation to the level of the diaphragm was performed in a single breath-hold. Scanning parameters included heart rate-dependent pitch (.2 to .45), 330 ms gantry rotation time, 100 or 120 kVp tube voltage, and 600 mAs tube current. Dose modulation protocols were implemented according to our previously described methods.17 The effective radiation dose in the patients, using the various acquisition protocols, was 11.3 ± 6.4 mSv.

Retrospectively gated reconstruction of raw CCTA data was performed at 40% (when available), 65%, 70%, 75%, and 80% of the R-R interval using the following parameters: .6-mm slice thickness (.75 mm if BMI >35 kg · m⁻²), .3-mm slice increment, 250-mm field-of view, 512 × 512 matrix, and B26f “medium smooth” kernel or a B46f “sharp” kernel in patients with dense coronary calcium (calcium score >100).

Reconstructed data were transferred to a Siemens workstation (Leonardo; Siemens Medical solutions) or a Hewlett-Packard workstation (Palo Alto, California) running the Vitrea 2 clinical software (Vital Images, Minnetonka, Minnesota). CCTA examinations were read for this research by two experienced CCTA readers (B. T. and R. N.) blinded to patient clinical status, SPECT-MPI, and clinically reported CCTA findings. Observers visually assessed each coronary segment using standard axial images, oblique multiplanar reformations, and oblique maximum intensity projections.18 Twenty-five coronary segments including the left main (LM) were analyzed and for the purpose of per-vessel analysis, the maximal stenosis from each of the three major coronary arteries were used.19 With respect to branching vessels, the diagonal and ramus intermediate branch were combined with the left anterior descending artery (LAD), the obtuse marginal (OM) branch combined with the left circumflex (LCX) and the posterolateral and posterior descending artery combined with the right coronary artery (RCA) or the LCX depending on whether they originated from the RCA or LCX. Maximal diameter stenosis severity was visually graded by consensus between the two observers on a scale of 0 to 6 (0 = 0%, 1 = 1% to 24%, 2 = 25% to 49%, 3 = 50% to 69%, 4 = 70% to 89%, 5 = 90% to 99%, and 6 = 100%).20 We had demonstrated previously that visual stenosis assessment by this grading system and quantitative CT-based stenosis assessment were equivalent when compared to stenosis quantification by invasive angiography.20 Plaque was further characterized as calcified or non-calcified plaque as described previously.21 When plaque contained both calcified and non-calcified components with neither of them constituting >75% of the plaque volume it was described as a partially calcified or mixed plaque. Data were analyzed after excluding all patients with densely calcified plaque for per-patient analysis and after excluding all vessels with dense calcification for per-vessel analysis. A secondary analysis was performed by including all patients and vessels with dense calcification. For this analysis the artery with dense calcification obscuring the coronary lumen was considered to have a maximal luminal stenosis ≥90%. For assignment of plaque distribution within individual vessels, the ramus intermediate branch was considered an extension of the proximal LAD, the diagonal branch an extension of the mid-LAD, the OM branches an extension of the proximal circumflex and the PDA and posterolateral branches an extension of the distal RCA or distal LCX depending on coronary dominance. For quantification of coronary artery calcium, all CT images were reviewed by an expert reader, using semi-automatic commercially available software (ScImage, Los Altos, CA). Total Agatston CCS was calculated as the sum of calcified plaque scores of all coronary arteries.22

Patients were instructed to discontinue beta-blockers and calcium antagonists 48 hours and nitrates 24 hours before testing. Rest perfusion images were acquired 10 minutes after infusion of 7-9 mCi of ^99mTc-sestamibi or 3-4.5 mCi of ²⁰¹Tl (based on body weight). Stress testing was performed with a symptom-limited Bruce treadmill exercise protocol or vasodilator challenge, as described previously.23 At near-maximal exercise, ^99mTc-sestamibi (32-40 mCi based on patient weight) was injected intravenously, after which treadmill exercise was continued at maximal workload for 1 minute and at one stage lower for two additional minutes whenever possible. ^99mTc-sestamibi MPI acquisition was started 15-30 minutes after radiopharmaceutical injection. For vasodilator stress, adenosine was infused at 140 μg · kg⁻¹ · min⁻¹ for 5 minutes and in ambulatory patients, a low-level treadmill exercise was performed during adenosine infusion.24 At the end of the second minute, ^99mTc-sestamibi (32-40 mCi) was injected and myocardial imaging was started approximately 60 minutes later.6 Twelve lead electrocardiography (ECG) was monitored continuously during stress testing. Horizontal or downsloping ST-segment depression ≥1 mm or upsloping ≥1.5 mm were considered positive for ischemia.

SPECT images were acquired with a 2-detector gamma camera (Philips Adac Forte or Vertex, Philips Medical Systems, Cleveland, Ohio or E-Cam, Siemens Medical Solutions). High-resolution collimators were used, and acquisition consisted of 64 projections over a 180° orbit, with 64 projections at 25 s/projection for supine ^99mTc acquisition followed immediately by 15 s/projection for prone ^99mTc acquisition.25 Rest ²⁰¹Tl acquisition was performed at 35 s/projection in supine position only. At each of the 64 projection angles, the image data were recorded into 16 equal ECG-gated time bins. No attenuation or scatter correction was applied. After iterative reconstruction (12 iterations) with Butterworth prefiltering (cutoff, .66 cycle/pixel for supine ^99mTc, .55 cycle/pixel for prone ^99mTc; order 5), short-axis images were automatically generated.26

All SPECT MPI studies were reanalyzed quantitatively for purposes of this research study, using an objective, automatic computer analysis.27 Automatically generated myocardial contours were evaluated by a core laboratory technologist without knowledge of any clinical data, and when necessary, contours were adjusted to correspond to the myocardium (8% of cases). A qualitative assessment of image quality using a visual 5-point scale (1 = uninterpretable, 2 = poor, 3 = fair, 4 = good, and 5 = excellent) was performed. The quantitative perfusion variable employed was total perfusion deficit (TPD), which is a validated, computer-derived analog of the percent myocardium abnormal by visual analysis, representing both extent and severity of perfusion defect.28 TPD was calculated as the percentage of the total surface area of the left ventricle below the predefined uniform average deviation threshold using QPS software.27 TPD was measured at stress and rest, and patients with a stress TPD of ≥5% were considered abnormal and were considered to represent myocardial ischemia in these patients with no known CAD.28,29 A computer-derived maximum severity score was also quantified using QPS software and was expressed on a scale of 0-4 (0 = normal and 4 = absence of tracer uptake) for each patient and for each coronary territory.27 Defects in the anterior and septal wall were allocated to the left anterior descending coronary artery (LAD); in the lateral wall, the left circumflex coronary artery (LCX); in the inferior wall, right coronary artery (RCA). When the assignment of a perfusion defect to a vascular territory was not clear (watershed regions), we used automated fusion software described previously and assigned the vessel according to the large vessel supplying the region in question.30

Continuous variables were expressed with median and range. Categorical variables were expressed as percentages. The severity of luminal stenosis, presence perfusion defects by SPECT-MPI, and frequency of perfusion defects were considered categorical variables. The Wilcoxon sum test, and a Kruskall Wallis test were used to compare continuous variables and Pearson Chi-square tests were used to compare ordinal variables. Sensitivity, specificity, and other diagnostic test measures were used to compare the different cut points of degree of stenosis to predict ischemia. Receiver operating characteristic (ROC) curves were constructed to assess the discriminatory value of various CCTA stenosis cutoffs in detecting the presence and absence of inducible ischemia by SPECT-MPI. The measured areas under the ROC curves were compared as described by Hanley and McNeil.31 CCTA-based imaging features of coronary artery stenosis including luminal stenosis severity, plaque composition, number of coronary arteries with stenosis ≥50%, and the presence of serial stenoses ≥50% were included in a stepwise fashion in a multivariable logistic regression analysis and adjusted for baseline clinical risk factors to determine the strongest significant predictors of ischemia. A P-value of <.05 was considered significant. Data were analyzed using STATA software version 9 (www.stata.com).

Exercise stress testing was used in 201 (68%) of patients, of whom 97% achieved >85% of maximal predicated heart rate. Adenosine stress was employed in the remaining 91 (32%) patients. Stress perfusion abnormalities were seen in 46 patients (15.4%). In patients with abnormal perfusion, stress TPD ranged from 5% to 48% with a median defect of 14% and ischemic TPD (Stress TPD-Rest TPD) ranged from 4% to 16% with a median defect of 6%. The proportion of patients with stress perfusion abnormalities was similar between pharmacologic and exercise stress. Perfusion defects were localized to the LAD in 29 patients, LCX in 13 patients and RCA in 17 patients. There were 12 patients with stress-induced perfusion defects in more than 1 vascular territory.

CCTA study quality was excellent or good in 283 studies (97%) and fair in 6 studies. Calcium scores ranged from 0 to 4305 with a median score of 35. Nine patients had dense calcification in one or more coronary arteries were excluded from the primary analysis. A secondary analysis was performed by considering these patients to have a maximal stenosis of ≥90%. Ninety-four patients (32%) had no detectable coronary artery plaque, 104 (35%) had plaque causing a maximal luminal stenosis of >0% and <50% of the vessel diameter, 58 (20%) had a maximal stenosis of ≥50% and <90%, and 27 (9%) had a maximal stenosis of ≥90%. In 27 patients, two or more coronary arteries exhibited luminal stenosis of ≥70%. Maximal stenosis was found in a proximal segment in 54% of patients, in a mid segment in 16%, in a distal segment in 6%, and in multiple locations in the same vessel (serial stenoses) in 24%. In patients with any detectable plaque, calcified plaque caused the maximal stenosis in 42%, partially calcified plaque in 41%, and non-calcified plaque in 17%.

On a per-vessel basis, 473 arteries were found to have no stenosis, 271 had a maximal stenosis of >0% and <50%, 80 had a maximal stenosis of ≥50% and <90%, and 38 had a maximal stenosis of ≥90%. There were 13 coronary arteries with dense calcium in which luminal stenosis could not be reliably assessed by both expert readers.

With increasing grades of stenosis severity from 0% to 100%, there was an increase in the proportion of individuals with abnormal myocardial perfusion by SPECT-MPI, P = .001 (Figure 1). Only a small proportion of patients with <50% stenosis on CCTA had stress perfusion defects, including three of 94 patients (3%) without coronary artery plaque (stress TPD 5-6%) (Figure 1) and three of 59 patients (5%) with atherosclerotic plaque and a maximal luminal narrowing of 1-24% (stress TPD 5-8%). A progressive increase in the frequency of ischemia was observed in the >50% stenosis categories. Among patients with a luminal stenosis severity of 50-70%, 17% were found to have perfusion abnormalities by SPECT-MP increasing to 42% and 74% when maximal stenosis was 70-89% and 90-100%, respectively (Figure 1).

Figure 1 Prevalence of ischemia on SPECT-MPI in patient groups defined by maximal stenosis detected during CCTA. Bars represent percent of individuals in each stenosis category with ischemia, defined by TPD ≥ 5%, on SPECT-MPI. The numbers in parentheses represent number of individuals in each stenosis category. In a per-patient analysis, prevalence of ischemia increased in proportion to stenosis severity with P = .001 across all categories of stenosis severity

---

Among all patients with luminal stenosis of 0-24%, 25-49%, and 50-69%, mean TPD was low, .05 ± .31, .12 ± .48, and .72 ± 1.14, respectively, P = ns (Figure 2). There was a significant increase in the mean stress TPD to 3.2 ± 3.5 and 12.9 ± 12.84 when stenosis severity increased to 50-89% and 90-100% (Figure 2). Similarly, when the mean severity of stress perfusion defects alone was considered, there was no significant difference in severity scores in patients with maximal luminal narrowing of 0-24%, 25-49%, and 50-69% (.19 ± .31, .001 ± .0, and .29 ± .48, respectively, P = ns (Figure 3). When maximal luminal stenosis increased to the 70-89% and 90-100% categories, there was a proportional increase in the severity of stress perfusion abnormality to .72 ± .14 and 2.58 ± 1.34, respectively (P < .001 and P < .0001, respectively) (Figure 3). While the stenosis category 50-69% was usually not associated with ischemia, 17% of these patients manifested stress-induced ischemia on SPECT-MPI as noted above.

Figure 2 Comparison of maximal stenosis detected by CCTA and TPD quantified by SPECT-MPI and expressed as percent perfusion defect on a continuous scale. Mean and standard deviation of the TPD for patients in each stenosis category are shown. On a per-patient basis, TPD increases with increase in maximal stenosis severity with P < .0001 across all categories of stenosis severity and P < .001 when patients with stenosis severity of all categories except 90-100% are compared

---

Figure 3 Comparison of maximal stenosis detected by CCTA and the severity of ischemia quantified by SPECT-MPI and expressed on a scale of 0-4. Mean and standard deviation of the severity of ischemia for patients in each stenosis category are shown. On a per-patient basis, the severity of ischemia increases with increase in maximal stenosis severity with P < .0001 across all categories of stenosis severity and P < .001 when patients with stenosis severity of all categories except 90-100% are compared

---

On a per-patient basis, there was a strong association between the coronary artery with the maximal luminal stenosis and the region of myocardium affected by ischemia: 40 (87%) patients exhibited stress-induced ischemia in territories perfused by arteries with luminal stenosis of maximal severity that ranged from 50% to 100%. An example of a patient with a 70-89% stenosis in the proximal RCA with a stress perfusion defect in the corresponding vascular territory is shown in Figure 4. There were 12 patients with stress-induced perfusion defects in more than one vascular territory. In three of these patients, the ischemic territory was in a watershed region that could have been ascribed to the LCX or the RCA. In these patients the assignment of the perfusion territory was based on fusion imaging and coronary dominance.

Figure 4 An example of stress-induced ischemia in a 58-year-old male with a 70-89% stenosis in the proximal RCA with a non-calcified plaque. (A) A 2.5 mm thick maximum intensity projection (MIP) of a CCTA showing the RCA. The arrow points to the partially calcified plaque causing 70-89% stenosis at the ostium of the RCA. (B) Stress and rest SPECT-MPI showing a reversible perfusion defect in this same patient in the mid and distal inferior wall. (C, D) Polar maps of stress and rest SPECT-MPI showing the reversible perfusion defect in the RCA territory

---

In six patients, luminal stenosis was absent or minimal (0-24%) in the vessels that corresponded to the ischemic territories. The quality of SPECT-MPI images in five of these patients was suboptimal from motion or breast attenuation artifact and they were unable to undergo prone imaging. A sixth patient was found to have severe left ventricular hypertrophy by CCTA without significant coronary artery stenosis.

Univariable analysis was used to identify variables that were associated with ischemia among patients with coronary artery stenosis. In addition to traditional risk factors such as increasing age, male gender, hypertension, hyperlipidemia, and diabetes, the severity of luminal stenosis, presence of multiple vessels containing ≥50% stenoses and serial ≥50% stenoses in one coronary artery were strongly associated with ischemia.

Previous studies have shown an association between non-calcified and partially calcified plaque with myocardial ischemia.32-35 In our analysis, there was a significant association of non-calcified plaque with ischemia in a univariable model; however, this association was not preserved when other confounding factors such as stenosis severity were considered in the multivariable model. Our analysis also revealed a strong relationship between multivessel ≥50% stenotic disease and ischemia that faded when accounting for the presence of high grade stenosis.

When we evaluated the relationship between serial stenoses and ischemia, we found that presence of multiple significant stenoses in the same vessel was strongly associated with ischemia on SPECT-MPI. The strong relationship between serial stenoses and ischemia remained even after adjustments for severity of individual luminal stenosis were made. This observation is consistent with the physiological effect of serial stenoses on resistance to coronary blood flow that has previously been described in both experimental animals and in patients with epicardial coronary artery disease.36,37 In the presence of serial stenoses, the net resistance to coronary blood flow is determined by the sum of the individual resistances and can result in greater stress-induced impairment of myocardial perfusion compared to a single stenosis with the same reduction in luminal diameter.36,37 The incremental value of detecting serial ≥50% stenosis when assessing coronary arteries by CCTA could potentially improve our ability to identify individuals likely to manifest inducible ischemia.