Noninvasive Imaging in Coronary Artery Disease

Abstract

Noninvasive cardiac imaging is widely used to evaluate the presence of coronary artery disease. Recently, with improvements in imaging technology, noninvasive imaging has also been used for evaluation of the presence, severity, and prognosis of coronary artery disease. Coronary CT angiography and MRI of coronary arteries provide an anatomical assessment of coronary stenosis, whereas the hemodynamic significance of a coronary artery stenosis can be assessed by stress myocardial perfusion imaging, such as SPECT/PET and stress MRI. For appropriate use of multiple imaging modalities, the strengths and limitations of each modality are discussed in this review.

Introduction

Coronary artery disease (CAD) is one of the leading causes of death, and the prevalence has increased worldwide.^1-3 To diagnose and define the severity of CAD, various imaging modalities have been introduced. For choosing the appropriate test, we review the strengths and limitations of each modality. Available modalities include the following: coronary CT angiography (CCTA), CT perfusion (CTP), computed fractional flow reserve (FFR) derived from CT (FFR_CT), nuclear myocardial perfusion imaging (MPI) (SPECT/PET), and cardiac MRI (CMR).

Coronary CT Angiography

As coronary revascularization is the main approach to treat CAD, CAD imaging has often been driven by the evaluation of anatomical stenosis severity. For this anatomical approach toward CAD, CCTA has gained its value with a rich evidence base that supports its clinical utility. In 3 prospective multicenter studies, the diagnostic performance of a 64-slice CCTA was reported in various populations.^4-6 The assessment by coronary computed tomographic angiography of individuals undergoing invasive coronary angiography (ACCURACY) trial was a 16–U.S. center study in patients without known CAD. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of CCTA to detect ≥70% stenosis were 94%, 83%, 48%, and 99%, respectively.⁴ The coronary artery evaluation using a 64-row multidetector CT angiography (CORE64) study followed the ACCURACY trial as another multicenter study that was conducted in 291 patients both with and without known CAD who had baseline coronary artery calcium score less than 600 Agatston units. In this study, the per-patient sensitivity, specificity, PPV, and NPV for detecting ≥50% stenosis were 85%, 90%, 91%, and 83%, respectively. Furthermore, evaluation of stenosis using CCTA demonstrated that an area under the curve (AUC) was 0.91 for high-grade anatomical stenosis for invasive coronary angiography (ICA)–confirmed CAD, and it was similar to that demonstrated by ICA for the prediction of subsequent coronary artery revascularization.⁵ In a third multicenter trial with 360 patients without known CAD presenting with both acute and stable chest pain, the CAD prevalence (≥50% stenosis) was high (68%), and diagnostic performance of CCTA was maintained, demonstrating a per-patient sensitivity, specificity, PPV, and NPV of 99%, 64%, 86%, and 97%, respectively.⁶ The NPV for CCTA has generally been high (eg, 95%-100% in the 2 multicenter studies that constrained enrollment to those individuals without known CAD).^4,6,7 Based on these results, CCTA is now generally considered as a valuable tool for the exclusion of significant CAD.^8,9

In addition to coronary stenosis severity, CCTA enables the study of other important atherosclerotic plaque characteristics with generally high accuracy. As an example, CCTA offers the diagnostic ability to assess measures of arterial remodeling, plaque composition, presence of intraplaque calcification, and aggregate plaque volume. Of these, positive arterial remodeling, low attenuation plaque, and spotty intraplaque calcification have shown a relation to both presently occurring and future acute coronary syndrome.^10-14 Plaque volume on CCTA determined by manual methods demonstrates a high correlation and a modest agreement with that determined by intravascular ultrasound.^12,15,16

CCTA findings of CAD also inform measures of risk stratification in both asymptomatic and symptomatic individuals. In a meta-analysis evaluating 9592 patients, CCTA demonstrated a graded increase in adverse events following CCTA performance for no CAD, nonobstructive CAD, and obstructive CAD (any vessel with greater than 50% luminal stenosis).¹⁷ Particularly for individuals without known CAD, the pooled annualized event rate was low compared with that of obstructive CAD (0.17% vs 8.8%).¹⁷ CCTA also demonstrated a good prognostic value in various groups, such as young patients with a family history of CAD¹⁸ and ethnic groups.¹⁹ Combining CAD data with measures of left ventricular (LV) function provides incremental prognostic utility, with LV ejection fractions less than 50%, conferring a worsened prognosis. Perhaps most importantly, a normal finding on CCTA was associated with a very low annual mortality rate of 0.13%.²⁰

Current-generation CCTA is nevertheless not without limitations. Although anatomical assessment by CCTA is diagnostically sound with a robust ability to predict prognosis,^11,14,20,21 CCTA measurements of stenoses do not accurately predict their functional significance.²² Furthermore, usage of iodinated contrast material and exposure to radiation are also important considerations. Pertaining to the latter, numerous radiation dose–reduction methods have been introduced and now enable CCTA performance at levels similar to that of background radiation exposure from radon.

The near-term future of CCTA holds promise for better detection of coronary artery lesions. One such effort to overcome certain limitations of current-generation CCTA is in dual-energy CT, which exhibits improved image quality by a reduction in noise without additional radiation dose than that with single-energy CT.²³ When compared with ICA images, dual-energy CT showed 90% sensitivity, 94% specificity, and 93% accuracy for the detection of greater than 50% luminal stenosis.²⁴ Certain CT acquisition systems also demonstrate increased temporal resolution than single-energy CT does,²⁵ an attribute that obviates the need for heart rate control in all individuals and may improve the ability of CT to evaluate the severity of lesion stenosis, particularly in patients who do not achieve adequate heart rate control during the examination.

Cardiac CTP

Recent advances in technology allow for assessment of myocardial perfusion by CT. Multidetector CT systems can image in a dynamic mode, in which sequential images are obtained over a period to record the kinetics of iodinated contrast in the arterial blood pool and the myocardium.²⁶ George et al,²⁶ using a 64-detector CT in a canine ischemia model, performed CTP during adenosine infusion. They found strong correlations between the ratio of myocardial to LV upslope and microsphere-derived myocardial blood flow (MBF). The authors replicated the study in humans with adenosine stress 64- and 256-row detector CCTA and CTP (Fig. 1). In the human study, they calculated transmural perfusion ratio (subendocardial attenuation-subepicardial attenuation), which had a significant inverse linear correlation (r = 0.63, P = 0.001) with percentage diameter stenosis on quantitative ICA. Furthermore, the combination of CTP and CCTA was 86% sensitive and 92% specific for identifying patients with atherosclerosis causing perfusion abnormalities when compared with that of the combination of ICA and SPECT-MPI as the gold standard.²⁷ In a prospective multi-center international trial, the combined coronary atherosclerosis and myocardial perfusion evaluation using 320-detector row CT (CORE320) trial evaluated CT for the identification of CAD (≥50% luminal stenosis) and corresponding myocardial perfusion defect in patients with suspected CAD. This method demonstrated a utility of CTP in patients with chest pain, as CCTA with concomitant CTP was as effective as sequential SPECT-MPI and invasive angiography for identifying flow-limiting atherosclerotic lesions.²⁸ In 381 patients from 16 centers, the patient-based diagnostic accuracy defined by AUC of integrated CTA-CTP for detecting or excluding flow-limiting CAD was 0.87 (95% CI: 0.84-0.91). It showed better accuracy in patients without prior myocardial infarction (MI) (AUC = 0.90) and without prior CAD (AUC = 0.93).²⁹

Figure 1.

(1) Images from a 64-row detector CTP. (A) A partially reversible perfusion deficit in the territory of the LAD and a primarily fixed perfusion deficit in the inferior wall on radionuclide myocardial perfusion imaging with increased subdiaphragmatic tracer uptake (stress, upper panels; rest, lower panels). (B and C) Adenosine stress CTP shows a dense perfusion deficit in the LAD territory and a subendocardial perfusion deficit in the inferior and lateral walls. (D) An invasive coronary angiogram demonstrates a left-dominant system with a totally occluded LAD (black arrow) as well as intermediate- and high-grade stenoses in a large ramus intermedius, the body of the left circumflex, and the ostium of the obtuse marginal artery (white arrows). (2) Images from a 256-row detector CTP. (A) A partially reversible perfusion deficit in the inferior and inferolateral wall on radionuclide MPI in this patient with exertional angina (stress, upper panels; rest, lower panels). Rest (B) and stress (C) CTP imaging shows a reversible subendocardial perfusion deficit in the inferior and inferolateral walls. Noninvasive angiography confirms a significant stenosis (white arrows) in the proximal right coronary artery (D and E) and the proximal left circumflex artery (F).²⁷ (Color version of figure is available online.)

Given the nascent stage for CTP, data regarding the prognostic utility of CTP deficits remain unknown and require evaluation in prospective multicenter efforts. Furthermore, the cost and the clinical effectiveness need to be examined, with a particular focus on safety because of additional ionizing radiation exposure. Continuing development of CT technology may ameliorate some of these concerns, with new acquisition protocols such as using prospective electrocardiogram-gated scans .³⁰

Fractional Flow Reserve Derived From CT (FFR_CT)

At present, the “gold”-standard assessment of the hemodynamic significance of coronary stenosis is FFR.³¹ Recent advances in computational fluid dynamics now enable the calculation of coronary flow and pressure fields from anatomical image data (Fig. 2).³² Applied to CT, these technologies enable calculation of FFR, which is defined as the ratio of maximal MBF through a diseased artery to the blood flow in the hypothetical case in which the artery was normal. Several studies have demonstrated an incremental value of FFR_CT for the diagnosis of hemodynamically significant CAD. In the diagnosis of ischemia-causing stenoses obtained via noninvasive fractional flow reserve (DISCOVER-FLOW) trial, when compared with invasive FFR, noninvasive FFR_CT demonstrated a per-vessel accuracy, sensitivity, specificity, PPV, and NPV for lesions causing ischemia of 84.3%, 87.9%, 82.2%, 73.9%, and 92.2%, respectively.³³ The performance of FFR_CT was superior to CT stenosis for diagnosing ischemic lesions; the latter of which demonstrated an accuracy, sensitivity, specificity, PPV, and NPV of 58.5%, 91.4%, 39.6%, 46.5%, and 88.9%, respectively. The determination of fractional flow reserve by anatomic computed tomographic angiography (DeFACTO) trial—a multicenter international study evaluating FFR_CT against CT for diagnostic accuracy of ischemia³⁴— consisted of 252 patients in whom 407 vessels were directly interrogated by invasive FFR. On a per-patient basis, FFR_CT was superior to CT stenosis for diagnosis of ischemic lesions for accuracy (73% vs 64%), sensitivity (90% vs 84%), specificity (54% vs 42%), PPV (67% vs 61%), and NPV (84% vs 72%). The AUC of FFR_CT also showed improved discrimination over CT stenosis alone (0.68 [95% CI: 0.62-0.74] vs 0.81 [0.75-0.86]; P < 0.001). In patients with intermediate stenosis (30%-69%), there was more than a 2-fold increase in sensitivity, from 37%-82%, with no loss of specificity.³⁵ Evaluation of stenosis by FFR_CT also demonstrated superior discrimination when compared with that by CCTA based on per-patient and per-vessel analysis in patients with intermediate CAD. In a subsequent study of 96 patients, the use of FFR_CT to select patients for ICA and percutaneous coronary intervention (PCI) resulted in 30% lower costs and 12% fewer events at 1 year when compared with that in the most commonly used ICA or visual strategy.³⁶ Most recently, the HeartFlowNXT study showed a more accurate determination of which arterial blockages are associated with coronary ischemia than that shown by standard CCTA or by ICA.³⁷ In this prospective, international study that enrolled 254 patients scheduled to undergo nonemergent clinically indicated ICA owing to suspected CAD, FFR_CT data matched closely with invasively measured FFR. The AUC for FFR_CT (≤0.80) was 0.82 vs 0.63 for CCTA (≤0 stenosis) (P < 0.0001) with invasive FFR as the reference standard. The per-patient sensitivity and specificity were 86% and 79% for FFR_CT vs 94% and 34% for CCTA, and 91% and 51% for ICA (>50% luminal stenosis), respectively.

Figure 2.

A mathematical model for pulsatile coronary flow. (A) Lumped parameter models are coupled to the aortic inlet and noncoronary vasculature and coronary microcirculation. P is the pressure, R the resistance, C the capacitance, L the inductance, and E(t) the elastance. Subscript LA is for the left atrium, AV for atrioventricular, V-Art for ventricle-arterial, p for proximal, d for distal, a for arterial, im for intermyocardial, and V for venous. (B) Pulsatile pressure and flow rate waveforms demonstrate reduced coronary flow in early systole because of contraction of the ventricle followed by increased flow in diastole, resulting from relaxation of the ventricle. (C) Velocity fields and 3-dimensional pressure are computed throughout the cardiac cycle. The pressure gradients and high-velocity jets across stenoses in the left anterior descending artery (LAD), left circumflex artery (LCX), and right coronary artery (RCA) can be noted.³² (Color version of figure is available online.)

Recently, FFR_CT demonstrated its value on not only diagnostic area but also therapeutic prediction modeling. In a study conducted in 44 patients, FFR_CT was shown to be feasible for predicting the functional outcome of PCI by virtual stenting models, that is, prediction of ischemia reduction. Diagnostic accuracy of FFR_CT to predict ischemia (FFR ≥ 0.8) after stenting was 96% (sensitivity = 100%, specificity = 96%, PPV = 50%, and NPV = 100%). This study showed that virtual coronary stenting of CT-derived computational models is feasible and that this novel noninvasive technology may be useful in predicting functional outcomes after coronary stenting.³⁸ Further studies are needed to elucidate the role of FFR_CT figures in the evaluation of in-stent restenosis, coronary artery bypass grafts, and prognosis.

FFR_CT offers several practical advantages in that it does not require modification of CT angiography protocols, does not require administration of additional medications beyond what is typically administered for CCTA, and does not result in any additional radiation exposure. However, certain artifacts may affect CT interpretability, including coronary motion. Additional limitations of FFR_CT relate to assumptions in the physiological models that include population as well as patient-specific data. Relationships relating myocardial mass to total coronary flow, the relative coronary microvascular resistance based on vessel size, or reductions in resistance in response to adenosine-mediated hyperemia vary amongst patients. In patients with microvascular disease, models of adenosine-mediated hyperemia may overestimate the degree of vasodilation, resulting in FFR_CT values that are less than those of measured FFR.

MPI by SPECT and PET

SPECT is a robust imaging modality that has been widely used in clinical practice and its diagnostic accuracy, value in risk assessment, and prognosis are well established. A meta-analysis of 79 SPECT studies, including 8964 patients, demonstrated average sensitivities and specificities of 86% and 74%, respectively.³⁹ A more recent meta-analysis of 117 SPECT studies revealed a sensitivity of 88.3% and specificity of 75.8%.⁴⁰ Patients with abnormal myocardial perfusion SPECT (MPS) have on average an annual event rate (cardiac death or MI) of 6.7%.³⁹ By contrast, a normal MPS confers generally good clinical outcomes. A total of 16 studies performed between 1994 and 2001, which reported 20,983 patients with normal MPS and a mean follow-up of 28 months, observed a rate of death due to cardiac disease or MI of 0.7% per year, a rate similar to that of an asymptomatic population.⁴¹ Similar findings have recently been reported in a multi-center registry of 4728 patients. The observed annualized survival rate for those patients with a normal finding on ^99mTc-tetrofosmin evaluation was 0.6%. Similarly, excellent survival rates were noted for the male and female subsets of this population as well as for patients who could exercise and for those undergoing pharmacologic stress testing.⁴² Accuracy of SPECT may improve in the setting of newer imaging equipment, radiotracers, image acquisition, and reconstruction techniques. The recently introduced camera systems with optimized acquisition geometry, collimator design, and reconstruction software have the potential to improve image quality with significantly shorter image acquisition times and reduced isotope doses, but whether these technological advances translate to improved clinical performance remains to be definitively addressed.⁴³

The incremental value of hybrid cardiac imaging with SPECT and CT has also been of interest. In the first report of 38 patients with perfusion defects on SPECT, the number of lesions with equivocal hemodynamic relevance was significantly reduced using SPECT/CT fusion when compared with that using side-by-side analysis.⁴⁴ This added clinical value was observed in 29% of patients and was particularly common in patients with multivessel disease and intermediate severity stenosis or in patients with diseases side branches. A study implementing motion-frozen SPECT data and CT-guided SPECT contour and territory adjustments found that the improved diagnostic value of hybrid imaging was mainly driven by higher diagnostic indices in left circumflex and right coronary artery territories.⁴⁵ In a study including more than 500 patients with a normal SPECT study who also underwent concomitant CCTA, patients an abnormal CCTA was associated with a higher rate the combined end point.⁴⁶

The following 2 new high-efficiency cardiac cameras have been introduced: Discovery NM 530c (GE Healthcare, Israel) and D-SPECT (Spectrum Dynamics, Israel), and they are based on an array of cadmium-zinc-telluride (CZT)–pixilated detectors. The use of CZT detectors has advantages of superior energy resolution and spatial resolution while acquiring 19 cardiac views.⁴⁷ CZT SPECT imaging demonstrated 95% sensitivity and 69% accuracy for detecting obstructive CAD by ICA defined as ≥70% narrowing.⁴⁸ In a study of consecutive patients without known CAD, patients undergoing CZT SPECT required a lower mean total administered isotope dose per patient and shorter imaging time than the ones undergoing the hybrid SPECT/64-slice CT.^49,50

PET has also demonstrated high diagnostic performance for the diagnosis of CAD. A study comparing the diagnostic accuracy between PET and SPECT observed the image quality to be excellent in 78% of PET studies when compared with that in 62% of SPECT studies, a finding that was associated with higher accuracy with PET (89% vs 79%). The superiority of PET was maintained in patients who were obese, a problematic population for SPECT.⁵¹ The superior quality and accuracy of PET have been attributed to its better spatial resolution and attenuation correction, an advantage that is particularly relevant in patients prone to soft tissue attenuation artifacts. A meta-analysis⁵² of 114 SPECT studies (13,741 patients) and 15 PET studies (1319 patients) demonstrated average sensitivities and specificities of 88% and 61% for SPECT and 84% and 81% for PET, respectively. Another meta-analysis demonstrated that⁵³ PET MPI using rubidium-82 (⁸²Rb) possesses superior sensitivity and specificity than that of SPECT for the detection of CAD. This study showed sensitivities of 85% (CI: 0.82-0.87) and 90% (CI: 0.88-0.92) and specificities of 85% (CI: 0.82-0.87) and 88% (CI: 0.85-0.91) for SPECT and PET, respectively. Summary of receiver operating characteristic curves was computed: AUC was 0.90 and 0.95 for SPECT and PET, respectively (P < 0.0001).

PET MPI is also employed for evaluating intramyocardial microvascular function through the measurement of absolute MBF at rest and at peak stress (Fig. 3). With dynamic perfusion images, image-derived time-activity curves from the arterial blood and myocardial tissue regions are used as inputs to a tracer kinetic model. The rate of uptake of the tracer into the myocardial tissue provides an estimate of MBF on an absolute scale of mL/min/g. This technique also allows the calculation of the myocardial flow reserve (MFR), which is the ratio of MBF at peak hyperemia to resting MBF.^54,55 MBF and MFR demonstrated prognostic value as an additional marker for adverse cardiac events.^56-59 Herzog et al⁵⁶ demonstrated that in the setting of a normal MPI result, a preserved CFR-MFR of more than 2.0 is a strong predictor of a good event-free survival through 3 years with an adverse prognosis seen in the setting of a reduced MFR. In addition, MFR results have been observed to effectively stratify patients based on their risk of adverse cardiovascular events in those with normal and those with abnormal stress perfusion.

Figure 3.

Determination of regional myocardial blood flow from the serially acquired PET images. On the reoriented short-axis image of the left ventricular myocardium (upper left panel), a region of interest is assigned to the left ventricular blood pool. The polar map displays the relative distribution of the flow tracer throughout the left ventricular myocardium. Based on a comparison with a normal database, no regional reductions in radiotracer activity are identified. Regions of interest are assigned to the territories of the left anterior descending (LAD), the left circumflex (LCX), and the right coronary arteries (RCAs). The corresponding time-activity curves for arterial blood and the myocardial regions are shown in the left lower panel. The final readout is shown in the lower left panel; extent and severity of regional flow defects are indicated but listed as zero because of the normal homogenous radiotracer distribution. Values of flow in mL/min/g are indicated for each of the vascular territories as well as the entire left ventricular myocardium. ROI, region of interest; S.F., spillover fraction, for cross contamination of activity from blood into myocardium.⁹⁹ (Color version of figure is available online.)

Recently, integrated PET/CT has been demonstrated to offer accurate spatial coregistration of myocardial perfusion defects and subtending coronary arteries (Fig. 4). In a recent study, Javadi et al⁶¹ compared standardized myocardial vascular territories with individual coregistered myocardial territories using ⁸²Rb PET/CT and found disagreement in 9% of the segments. Notably, in 72% of patients, a disagreement in at least one myocardial segment was found. Most frequently, standard right coronary segments were reassigned to the left circumflex artery territory (39% of reassigned segments), and standard circumflex segments were reassigned to the left anterior descending territory (30%). The PET/CT images allowed for identi-fication of flow-limiting coronary lesions that required a revascularization procedure (as defined by ICA and PET) with a sensitivity, specificity, PPV, and NPV of 90%, 98%, 82%, and 99%, respectively. These results were confirmed by a similar study with SPECT/CT that demonstrated the hybrid approach to result in a significant improvement in specificity (from 63%-95%) and PPV (from 31%-77%) when compared with that by CCTA alone for detecting flow-limiting coronary stenosis.⁶² A similar diagnostic performance was reported by Groves et al. using a ⁸²Rb PET/CT hybrid system.⁶³ Notably, most nonevaluable severely calcified vessels in the left anterior descending artery had positive findings on stress nuclear MPI, whereas most nonevaluable vessels with motion artifacts in the right coronary artery had negative findings.

Figure 4.

(A) Fused 3D reconstructions of a coronary arteriogram and stress myocardial perfusion obtained in the same setting, assessed through integrated PET/CT. A full-motion cine can be viewed in the online-only Data Supplement Movie V. LCX indicates left circumflex artery; LAD, left anterior descending coronary artery.⁶⁰ (B) An integrated PET/CT study. The CTA images demonstrate a noncalcified plaque (arrowhead) in the proximal left anterior descending coronary artery (LAD) with 50%-70% stenosis; however, the rest and peak dobutamine stress myocardial perfusion PET study (lower left panel) demonstrates only minimal inferoapical ischemia. In addition, LVEF was normal at rest and demonstrated a normal rise during peak dobutamine stress. Full-motion cines can be viewed in the online-only Data Supplement (Movies VI and VII). Ant, anterior; Endo Epi, endocardial plus epicardial; Inf, inferior; LCX, left circumflex; LM, left main; RCA, right coronary artery.⁶⁰

Some studies have described the development of “normal databases” characterizing relative perfusion data for patients with a low likelihood of CAD.^64,65 These databases objectively define the normal range of relative perfusion values to evaluate patients with suspected CAD and determine whether the perfusion distribution is normal or abnormal. These databases can be used as an adjunct to the subjective visual scan interpretation by the clinician. Most of the previous studies that have developed low likelihood normal databases have done so only in the context of SPECT or 2D PET MPI. Recently, 3-dimensional (3D) ⁸²Rb PET normal database containing the relative perfusion scores of patients without obstructive CAD demonstrated a good diagnostic value. Its normalcy rate was 95%. Sensitivity was 100% for detecting patients with either 50% or 70% stenosis. Optimal specificity was 87% for either 50% or 70% stenosis.⁶⁵

PET also enables assessment of myocardial viability and hibernation. In a recent meta-analysis (24 studies, 756 patients), the sensitivity and specificity values of PET were 92% and 63%, respectively.⁶⁶ Regarding recovery of LV function after revascularization, PET demonstrated a higher sensitivity but comparable specificity to all imaging techniques.⁶⁶ In another meta-analysis that included more recent studies from 1998-2006, patients with ischemic LV dysfunction (LV ejection fraction ≤45%) were more sensitively examined by PET, albeit at a cost of specificity when compared with nonnuclear techniques to predict recovery of global LV dysfunction.⁶⁷ However, the risk of death after revascularization was similar unless the extent of viable tissue was more than 20% of LV.⁶⁸

Although PET has shown its usefulness in clinical assessment of myocardial perfusion and viability, there are several limitations, such as limited availability, methodological complexity, and high costs. Nonnegligible radiation doses are associated with PET, but current PET perfusion stress-rest imaging (including absolute quantification of myocardial perfusion) can be performed with much lower doses (<4 mSv) when compared with that for SPECT, particularly with the use of new 3D PET/CT scanners.^69,70 Furthermore, absolute quantification of myocardial perfusion has several limitations that need to be recognized. The optimal cutoff values for absolute perfusion need to be defined in large populations and for various tracers, and potential sources of errors in mean flow rate measurements from PET-based perfusion imaging need to be remembered, including the effect of high coronary driving pressures and resting flow rates. In addition, a further study is required as to how MFR may effectively differentiate between patients who have epicardial stenosis and patients who have abnormalities with subendocardial microvascular perfusion. Furthermore, additional information is needed to address important subpopulations such as patients who are diabetic and obese, patients with heart failure, and patients who have undergone revascularization.

Cardiac MRI

Perfusion CMR studies employ the use of “first pass” of an intravenously injected gadolinium contrast agent during vasodilator (ie, adenosine or dipyridamole) challenge to identify hemodynamically significant coronary artery stenosis. At present, this technique has been well validated, showing similar or better diagnostic accuracies than routinely used techniques, such as SPECT^71-73 or stress echocardiography, did.⁷⁴ The MRI for myocardial perfusion assessment in CAD trial⁷³ was a multicenter, multivendor prospective trial, which compared perfusion CMR with SPECT for the detection of ischemia, with quantitative coronary angiography as the gold standard. It was reported that perfusion CMR in the entire study population was superior to SPECT. Nagel et al⁷⁴ compared dobutamine stress CMR (DSMR) with dobutamine stress echocardiography (DSE) in the detection of CAD (≥50% stenosis) using ICA as the gold standard. They reported that DSMR has a higher diagnostic accuracy than that of DSE in patients with suspected CAD. The sensitivity increased from 74%-86% and specificity increased from 69.8%-85.7% (both P < 0.05) for DSE and DSMR, respectively. A meta-analysis of DSMR using 14 data sets demonstrated a sensitivity of 83% and a specificity of 86% on a per-patient level.⁷⁵ These studies included patients who were selected to undergo catheterization and had a relatively high disease prevalence of 71%. Perfusion of the myocardium can be assessed qualitatively or quantitatively by calculating the myocardial perfusion reserve index (MPRI).⁷⁶ To calculate MPRI, the relative upslope of the signal intensity of a given segment during stress is divided by the relative upslope of the same segment at rest. The MPRI has been validated during ICA, and an MPRI of 1.5⁷⁷ or 2.0⁷⁸ correlated with FFR measurements of ≤0.75 or with ≥50% luminal stenosis. In a prospective study, perfusion CMR was compared with PET and quantitative coronary angiography in detection and sizing of compromised myocardium in 48 patients and 18 healthy subjects.⁷⁹ This study observed that a perfusion CMR approach reliably identifies patients with coronary artery stenosis and provides information on the amount of compromised myocardium. Furthermore, an array of studies showed that ischemia by stress perfusion CMR and DSMR is an independent predictor of cardiac events.⁸⁰ Normal stress perfusion CMR or DSMR predicted a 3-year event-free survival of 99.2%. This concept has been further validated for stress perfusion CMR, where strong prognostic values were demonstrated for both male and female patients,⁸¹ and also for DSMR, in which test results were highly predictive of cardiac events in 1369 patients who were followed up for 4 years.⁸² The advantage of stress perfusion CMR over modalities such as SPECT is the ability to distinguish between subendocardial and transmural perfusion defects.

A potential limitation of CMR stress imaging is that current pulse sequences often demonstrate a “dark-rim” artifact that can be mistaken for a true perfusion abnormality.⁸³ This artifact is explained by a multitude of factors, such as a motion during data acquisition, Gibbs ringing caused by resolution limitations, or susceptibility artifacts from the passage of the contrast agents.^84-86 Recent technology such as higher spatial resolution CMR might reduce this artifact. However, more data are needed to define the optimal application of this technique on CMR perfusion.⁸⁷

Delayed-Enhancement MRI

Patients with ischemic cardiomyopathy may be more likely to benefit from revascularization if viable tissue is present. Direct imaging of myocardial fibrosis is possible with the use of an inversion recovery–prepared T1-weighted gradient echo sequence after the intravenous administration of gadolinium. This CMR technique has been named delayed-enhancement MRI (DE-MRI) and demonstrates nonviable tissue as hyper-enhanced or bright. The clinical utility of DE-MRI in the delineation of nonviable myocardium has been confirmed by direct comparison with several clinically established markers of myocardial viability, including contractile reserve, perfusion, metabolism, and, most recently, electromechanical mapping.⁸⁸ The performance of DE-MRI for the detection of MI was tested in an international, multicenter, randomized trial.⁸⁹ In total, 282 patients with acute and 284 with chronic first-time MI were scanned in 26 centers. The study showed that the sensitivity of DE-MRI increased with increasing gadolinium dose, reaching 99% and 94% in acute and chronic MI, respectively, with the 0.3-mmol/kg dose. Furthermore, with doses 0.2 mmol/kg or higher, when MI was identified, it was in the correct location in more than 97% of patients.⁸⁹ However, optimal dose of gadolinium contrast might be challenging, as it needs to be balanced between higher dose for better image quality⁸⁹ and limitations such as the cost, brighter blood pools that may obscure subendocardial infarcts,⁹⁰ and an increased concern for nephrogenic systemic fibrosis in patients with severe renal impairment.⁹¹ Furthermore, to date, although DE-MRI has been demonstrated to be a robust prognostic tool, its use to guide revascularization strategies has not been observed.

MR Coronary Angiography

Tremendous advances in MRI technique during the past decade have demonstrated potential for noninvasive diagnosis of CAD. The image resolution of current coronary artery MR techniques is approximately 1 mm, which is sufficient for the detection of stenosis in large coronary arteries but likely inadequate for accurate detection of disease in smaller branches of the coronary vasculature. A multicenter study from Japan was reported for 138 patients using a whole-heart MR coronary imaging technique. The acquisition of images was successfully completed in 92% of patients. The sensitivity of MR coronary angiography was 88% and the specificity was 72% for CAD detection (≥50% luminal stenosis by ICA).⁹² However, a significant limitation of MR coronary angiography is that test performance with contemporary technology (1.5 T, 3D navigator echo) can vary significantly with operator experience. The potential magnitude of this effect was illustrated in a study in which a 1.5-T MRI scanner was used in 21 patients; all images from 2 patients and 4 coronary segments from the remaining 19 patients were excluded from analysis because of poor image quality.⁹³ The remaining 186 proximal, middle, or distal coronary segments were interpreted by 2 observers. For the prediction of ≥50% stenosis by the 2 observers, sensitivity was 44% and 56% and the specificity was 95% and 84%. A meta-analysis of 51 studies compared the diagnostic performance of CCTA and MR coronary angiography for the detection of ≥50% diameter stenosis.⁹⁴ CCTA had significantly higher sensitivity (85% vs 72% with CMR) and specificity (95% vs 87%). A distinct meta-analysis by Schuetz et al⁹⁵ yielded a mean sensitivity and specificity of 97.2% and 87.4% for CCTA and 87.1% and 70.3% for MR coronary angiography, respectively. For ruling out CAD, CCTA is more accurate than MR coronary angiography, when scanners with more than 16 rows improve sensitivity, as do slowed heart rates.

Plaque Assessment Using CMR

CMR has shown potential for atherosclerotic plaque assessment.⁹⁶ Hyperintensity plaque in T1-weighted imaging was related with positive remodeling, ultrasound attenuation, lower Hounsfield units, and transient slow flow after PCI, which are related to vulnerable plaques.^10,96,97 Hyperintensity plaque also related with presence of intracoronary thrombus detected by optical coherence tomography in patients with angina.⁹⁸ Accordingly, the future role of coronary MRI may be as an important adjunct to the comprehensive CMR examination, which combines anatomical, functional, and plaque assessment. Direct correlation of MR angiographic abnormalities with perfusion and wall motion abnormalities may also prove to be a powerful combination in the evaluation of CAD.

Conclusion

An array of high-quality imaging technologies exist to study individuals with suspected CAD, with benefits and limitations to each of them. Selection of the best test depends on the patient risk factors and the local availability and expertise of each modality. This integrated approach combining anatomical and functional imaging is critical for comprehensive evaluation of the clinically relevant aspects of CAD.