INTRODUCTION

In the United States alone, more than 7.7 million patients are referred yearly for diagnostic cardiac stress testing with radionuclide imaging. Although patients may require pharmacologic stress testing because they are unable to perform adequate physical exercise, dynamic exercise on treadmill or bicycle remains the preferred stress modality. In our nuclear stress laboratory, the percentage of myocardial perfusion studies performed with exercise has remained approximately 50% since 2004. Exercise variables, including maximal workload, duration, hemodynamic response, exercise-induced symptoms, and electrocardiographic (ECG) changes provide invaluable additional information for assessing a patient’s prognosis that are not available with pharmacologic stress. The Duke exercise treadmill score incorporates some of these important exercise variables and allows for risk stratification.¹ However, because of the inherent diagnostic limitations of the exercise ECG and subjective symptoms, patients stratified by the Duke Treadmill Score can be further stratified on the basis of (semi)quantitative stress radionuclide myocardial perfusion imaging (MPI).²

PHYSIOLOGY OF EXERCISE STRESS TESTING

Dynamic exercise increases the metabolic demand of the exercising skeletal muscle which can be met only by increasing the blood flow to the exercising muscle. Under conditions of maximal exertion, the cardiac output may increase more than fourfold to meet this need.³ As shown in Figure 14-1, at low levels of exercise, the increase in cardiac output is due to an increase in both stroke volume and heart rate. However, as the intensity of exercise increases, the contribution of the stroke volume to cardiac output reaches a plateau, and the augmentation of cardiac output becomes primarily dependent on the ability to increase the heart rate further. Of note, the inability to appropriately increase the heart rate in response to exercise predicts severity of coronary artery disease (CAD) and mortality.^4,5 The increase in cardiac output results in a greater myocardial metabolic demand, inducing coronary vasodilation. A recent positron emission tomography (PET) study in human subjects quantifying the change in myocardial perfusion with exercise demonstrated that myocardial blood flow can increase more than fourfold to meet the metabolic demands of the heart during exercise.⁶

Figure 14-1 Hemodynamic and metabolic responses to exercise. A, During low levels of exercise, the increase in cardiac output is due to increases in both heart rate and stroke volume. However, as the exercise increases, stroke volume cannot increase further, and the augmentation of cardiac output is due to continued increases in heart rate. These changes are related to increased sympathetic drive as well as withdrawal of parasympathetic stimulation to the heart. B, Under resting, fasted conditions, fatty acid oxidation is the predominant fuel source for the body. However, with increasing levels of exercise, glucose oxidation plays an increasingly important role in meeting the energy demands of the exercising skeletal muscle. At the point of exhaustion, anaerobic glycolysis and glucose oxidation are required to meet the energy demands.

The above-mentioned hemodynamic changes occur through both neurohormonal and metabolic mechanisms. With the onset of exercise, heart rate and stroke volume both increase through sympathetic activation and parasympathetic withdrawal. During exercise there are several peripheral effects that augment the effects provided by the central sympathetic activation. These include α-adrenergic stimulation of the venous capacitance vessels, resulting in venoconstriction and enhanced venous return, which will increase preload and cardiac inotropy. In addition, skeletal muscle production of lactic acid and adenosine during exercise causes dilation of the arteriolar resistance vessels, reducing the systemic vascular resistance and thereby increasing cardiac output and blood flow to the exercising skeletal muscle.

Metabolic changes occur in association with these hemodynamic adaptations. Specifically, in exercising skeletal muscles, there is a decreasing dependence on fatty acid oxidation and an increasing dependence on glucose oxidation to meet the energy needs until maximum oxygen consumption is achieved. At that point, glucose is the exclusive oxidative substrate. Further increases in the level of exercise above the anaerobic threshold cause the release of lactate from exercising skeletal muscle. Interestingly and in contrast, with increasing exercise, the myocardium continues to oxidize both fatty acids and glucose and also increases the utilization of lactate produced by the exercising skeletal muscle for energy (see Fig. 14-1).

The intensity of exercise can continue to increase until the point at which the heart cannot meet the metabolic demands of the exercising muscles (i.e., blood flow to muscles can no longer provide adequate amounts of substrate or remove metabolic byproducts). In the case of a healthy individual, this is generally at the level of the exercising skeletal muscle. However, in the case of an individual with critical coronary artery stenoses, the inability to exercise can be due to the limited ability to provide adequate blood flow to the heart muscle. It is in this latter scenario that myocardial ischemia is manifested and provides the rationale for exercise stress testing.

EXERCISE PROTOCOLS

In the United States, most exercise tests are performed using a motorized treadmill and graded Bruce, modified Bruce, or Naughton protocols. In many other countries, the upright bicycle is the preferred exercise equipment. An important aspect of diagnostic exercise protocols is the gradual, linear increase in workload from a very low level to the maximally tolerated workload. This gradual increase in workload is of importance, since the purpose of physical exercise testing is to reproduce symptoms, and one can generally not predict at which workload symptoms will occur. It is therefore important to also evaluate the patient carefully prior to starting a stress test to determine if the patient is appropriate for exercise stress testing (Table 14-1). Exercise protocols are generally terminated because of the onset of symptoms or fatigue, but they may also be terminated because of the development of significant ECG changes or because the patient becomes hypotensive or develops pulmonary edema. Termination of exercise solely because the predicted target heart rate (85% of age-predicted max) was achieved is a commonly made mistake. Many elderly patients, for example, are capable of achieving a considerably higher heart rate and workload at which symptoms may be unmasked.

Table 14-1 Contraindications to Exercise Stress Testing

PREPARATION FOR EXERCISE

As mentioned, the purpose of exercise testing is to provoke and to reproduce ischemic symptoms during maximal physical effort. The patient should be well informed about the details of the procedure and be motivated to give his/her best effort. Cardioactive medications, such as β-blockers, calcium blockers, or nitrates, may diminish the sensitivity of testing. Therefore, if the exercise test is performed specifically to diagnose CAD in a patient with no prior cardiac history, such medications should be stopped for at least 24 hours and in the case of long-acting medications, 48 hours before the day of testing. On the other hand, in patients with known CAD, a physician may be more interested in the results of testing with the patient on his/her routine medication, which will provide insight into the effectiveness of treatment.

Patients should be instructed to wear comfortable clothing and shoes and be in a fasted state on the morning of the exercise test. Moreover, it is prudent to instruct all patients scheduled for stress testing not to consume any caffeine-containing beverages or food on the morning of testing. The reason for this is that it is not uncommon that a patient is not able to perform adequate exercise. If the patient has not consumed caffeine-containing items, it is easy to switch to pharmacologic stress and ensure a diagnostic test result without having to reschedule the patient. On the other hand, a recent study suggested that a small amount of caffeine does not significantly affect the presence and magnitude of perfusion abnormalities. Although it appears prudent to continue to recommend that patients abstain from caffeine usage prior to the test, consumption of a single cup of coffee may not be a valid reason for cancellation of a vasodilator stress test.⁷

BASIS OF STRESS IMAGING

The conceptual underpinning of radionuclide imaging is the visualization of heterogeneity of regional myocardial blood flow secondary to impaired regional coronary flow reserve downstream of coronary arteries with significant obstructive disease (Fig. 14-2).

Figure 14-2 Schematic representation of the principle of rest/stress myocardial perfusion imaging. Top: Two branches of a coronary artery are schematically shown; the left branch is normal, and the right branch has a significant stenosis. Middle: Rest and stress myocardial perfusion images of the territories supplied by the two branches. Bottom: Schematic representation of coronary blood flow in the coronary branches at rest and during stress. At rest, myocardial blood flow is similar in both coronary artery branches. Coronary blood flow in the abnormal vascular bed was maintained through autoregulatory mechanisms that lower vascular resistance (r) distal of a significant stenosis. When a myocardial perfusion imaging agent is injected at rest, myocardial uptake therefore will be homogenous (normal image on the left). During stress, peripheral vascular resistance (R) decreases in the normal bed (r), resulting in a 2 to 2.5-fold increase of blood flow over baseline. In the abnormal bed, distal from the stenosis, peripheral vascular resistance (r) cannot decrease much further. This creates heterogeneity of regional myocardial blood flow that can be visualized with thallium-201- or technetium-99m-labeled agents as an area with relatively decreased radiotracer uptake (darker left area on stress image).

(Modified from Wackers FJ: Exercise myocardial perfusion imaging, J Nucl Med 35:726–729, 1994.)

In order to visualize such heterogeneity of blood flow reliably, a linear relationship must exist between regional myocardial blood flow and regional myocardial uptake of a radiotracer.⁸ Although in general this is true for relatively low flow ranges (up to 1 mL/min/g), as is present under resting conditions and resting myocardial ischemia, at the higher flow ranges (>2 mL/min/g) achieved during stress, especially during vasodilator stress, regional myocardial radiotracer uptake may no longer be linear. At higher rates of blood flow, the uptake of many radiotracers presently used in clinical imaging demonstrates a plateau (Fig. 14-3) known as the roll-off phenomenon.

Figure 14-3 Relationship between coronary blood flow and radiotracer uptake. An ideal flow imaging agent would be plotted on the line of identity. All commonly used radiotracers deviate below the line of identity and show a plateau (“roll-off”) at higher blood flow levels. Typical ranges of coronary blood flow during physical exercise (EX) and pharmacologic vasodilation (Pharm) are shown. As shown on the graph, the roll-off is the least for thallium-201 and more for technetium-99m sestamibi and ^99mTc tetrofosmin.

The suboptimal linearity at higher flow ranges is caused by the limited first-pass extraction fraction of many single-photon radiopharmaceuticals used for MPI. Table 14-2 shows the myocardial extraction fractions of commonly used radiotracers. In addition to myocardial extraction fraction, mechanisms involved in the transmembrane transport and myocardial washout and wash-in affect net myocardial uptake of radiopharmaceuticals. Fortunately this roll-off phenomenon does not significantly affect the ability to identify patients with high-degree (>70%) coronary stenoses. However, in patients with relatively mild CAD, the roll-off phenomenon may potentially result in falsely normal images (Fig. 14-4).

Table 14-2 First-Pass Myocardial Extraction Fraction of Various Radiopharmaceuticals

Figure 14-4 Effect of radiotracer “roll-off” on myocardial perfusion images. In the graph, regional myocardial uptake of a radiotracer uptake is plotted against regional coronary blood flow. The radiotracer shows significant roll-off at flows exceeding 2.5 mL/min/g. In the relative low flow range, because of good linearity of uptake, a myocardial region with 0.75 mL/min/g has significantly less radiotracer uptake than a region with1.5 mL/min/g, resulting in an abnormal image with defect (left image). However, at higher flow ranges, a similar 0.75 mL/min/g flow difference may result in no significant difference in myocardial radiotracer uptake and in a normal image without perfusion defect (right image).

Thallium (²⁰¹Tl), a potassium analog, has a relatively high initial extraction fraction and good linearity to flow. After intravenous administration, continuous washout and wash-in occurs in the myocardium, resulting in Tl-201 myocardial redistribution. Technetium (^99mTc)-labeled agents, such as sestamibi and tetrofosmin, are lipophilic compounds that have lower first-pass extraction fractions than ²⁰¹Tl and bind to mitochondrial membranes in a stable manner based on the mitochondrial membrane potential, and demonstrate minimal washout. Consequently the ultimate net retention of ²⁰¹Tl-labeled and ^99mTc-labeled agents in the myocardium is similar.⁹

INJECTION OF RADIOTRACER

The timing of injection of radiotracer relative to peak exercise and termination of exercise is an important and often not well-appreciated aspect of stress testing. Because the first-pass extraction of the radiotracers is not 100%, myocardial uptake of a radiotracer is not instantaneous at the moment of injection. After injection during stress, the injected bolus reaches the central circulation in about 10 to 15 seconds, and first-pass myocardial extraction takes place. After the initial first transit, recirculation occurs with a rapidly decreasing blood concentration of radiotracer. On average, it takes about 1.5 to 2 minutes for blood radioactivity to decrease to 50% of maximum (Fig. 14-5). This may differ among patients, depending on the level of exercise effort. Furthermore, blood clearance curves are different for various radiotracers (e.g., slightly faster for ^99mTc-labeled agents than for ²⁰¹Tl). The timing of radioisotope injection is based on the fact that it is important to have most of the dose accumulate in the myocardium under conditions in which the heterogeneity of blood flow is maximal, that is, at peak exercise. Termination of exercise at about 1 minute after injection, as is frequently done in clinical practice, may be too early. A substantial amount of radiotracer may still be recirculating at that time. Therefore, it is advisable to encourage patients to continue exercising for at least 2 minutes after injection of radiotracer.

Figure 14-5 Blood clearance of radiotracer. Arterial blood disappearance curve of technetium-99m-sestamibi after injection at peak exercise or pharmacologic vasodilation with dipyridamole. Peak blood pool activity occurs at 0.5 minute for both exercise and pharmacologic testing. At 1.5 minutes after injection, the blood pool activity is 50% of maximum for exercise stress testing. The decrease in blood pool activity is relatively quicker for pharmacologic stress testing.

(From Sand NP, Juelsgaard P, Rasmussen K, et al.: Arterial concentration of ^99mTc-sestamibi at rest, during peak exercise and after dipyridamole infusion. Clin Physiol Funct Imaging 24:394–397, 2004.)

IMAGING PROTOCOLS

Reliable interpretation of stress myocardial perfusion images depends for a large part on optimal quality images. Consistently good quality can be ensured by following image acquisition guidelines published by the American Society of Nuclear Cardiology.¹⁰

Two important variables affect quality of images: the amount of radiotracer injected and imaging time. The injected dose can be adjusted to a patient’s weight, keeping in mind the balance between the image quality and radiation exposure. Imaging time, or time per stop per projection, can also be adjusted to the dose and the patient’s weight. One way to anticipate suboptimal image quality is to record the count rate emanating from a patient’s chest before setting image acquisition parameters. If the count rate is less than that usually recorded for good-quality studies, the time per stop should be lengthened.

At present, four image-acquisition protocols are most commonly used: the 1-day or 2-day rest-exercise ^99mTc-agent imaging protocol, the exercise-redistribution Tl-201 imaging protocol, and the rest/exercise dual-isotope imaging protocol. Numerous studies have shown that each of these protocols yields equivalent clinical results for detection of CAD and for risk stratification. If many patients referred to an imaging facility are overweight (body mass index > 30 kg/m²), ^99mTc-labeled agents and 2-day imaging protocols may be preferred. The most frequently used imaging protocols for ²⁰¹Tl and ^99mTc-labeled agents are schematically shown in Figure 14-6A-D. Technical details of imaging are tabulated in Table 14-3.

Figure 14-6 Commonly used rest/exercise (EX) SPECT imaging protocols. A, Two-day imaging protocol used for ^99mTc-based radiotracer stress/rest imaging, typically used in obese patients. B, One-day imaging used for Tc-99m-based radiotracer stress/rest imaging, typically used in nonobese patients. C, One-day, stress-redistribution imaging protocol used for thallium-201 imaging. D, One-day, dual-isotope imaging protocol. Time intervals and total protocol times are shown.

Increasing attention is being paid to the radiation exposure associated with imaging procedures, including myocardial perfusion studies. The exposure and the biological effect of radiation, generally assessed in terms of the effective dose, is dependent not only on the absolute amount of radioactivity that is injected but also the quality of the radiation that is emitted, which is the same for the gamma-photon emitting radionuclides used in single-photon emission computed tomography (SPECT) MPI, the half-life of the agent, and the sensitivities of specific organs to the radiation. Based on these factors, the lowest effective dose is achieved with ^99mTc-labeled agents used in a 1-day stress/rest protocol, whereas the highest effective dose is achieved with a dual-isotope protocol (Table 14-4).¹¹ In patients with low to intermediate pretest likelihood of CAD, normal exercise parameters, and normal MPI, it may be possible to perform only stress imaging, thereby reducing the radiation exposure.

Table 14-4 Estimated Effective Doses for Various Myocardial Perfusion Imaging Protocols

Based on data from Einstein AJ, Moser KW, Thompson RC, et al: Radiation dose to patients from cardiac diagnostic imaging. Circulation 116:1290–1305, 2007.

DIAGNOSTIC VALUE OF EXERCISE SPECT IMAGING

Since the introduction of stress radionuclide MPI in the mid-1970s, numerous clinical studies have demonstrated the clinical usefulness for detection of CAD. The main objective of exercise SPECT imaging in present-day clinical practice is still predominantly to elucidate whether symptoms suggestive of CAD, such as chest discomfort or dyspnea on exertion, are caused by CAD or whether other etiologies should be pursued. Management decisions may then be guided by extensive published evidence that certain image patterns have important prognostic significance.

Because of the proven prognostic value of stress MPI, it is unlikely that angiographic correlation and sensitivity and specificity will continue to be assessed in large numbers of patients using state-of-the-art technology, other than for local quality assurance purposes or to evaluate new technical advances such as attenuation-correction devices. The sensitivity, specificity, and normalcy values for exercise SPECT MPI have been evaluated in multiple studies and confirm the diagnostic power of this modality.^12–24 Based on a meta-analysis of 79 studies that included a total of 8964 patients, the sensitivity, specificity, and normalcy rate of SPECT MPI are 86%, 74%, and 89%, respectively.^25,26 Figure 14-7 shows representative data for the diagnostic yield for identifying disease in individual coronary arteries.²⁷ Sensitivity for recognizing disease in the left anterior descending coronary artery is higher than that for disease in the right coronary artery or left circumflex coronary artery, with no significant difference in accuracy. The relatively low specificity of SPECT imaging suggests that artifacts due to attenuation and motion are not always recognized. Referral bias has been proposed as another potential explanation for the limited specificity of SPECT imaging.

Figure 14-7 Detection of coronary artery disease in individual coronary arteries. The sensitivity (Sens), specificity (Spec), and accuracy (Acc) for detecting ≥ 50% stenosis in the left anterior descending coronary artery (LAD), right coronary artery (RCA), and left circumflex coronary artery (LCx) are shown.

(Reproduced with permission from Elhendy A, Sozzi FB, van Domburg RT, et al: Accuracy of exercise stress technetium-99m sestamibi SPECT imaging in the evaluation of the extent and location of coronary artery disease in patients with an earlier myocardial infarction, J Nucl Cardiol 7:432–438, 2000.)

Through the years, it has also become clear that complete agreement with coronary angiography is an elusive goal, because stress perfusion imaging and coronary angiography evaluate two different aspects of CAD. Whereas contrast coronary angiography visualizes coronary anatomy and the presence or absence of regional coronary luminal narrowings, radionuclide stress MPI provides noninvasive information about the pathophysiologic consequences of coronary atherosclerosis. Specifically, patients with apparently significant angiographic CAD but normal exercise myocardial perfusion images nevertheless have a favorable prognosis.²⁸ In contrast, patients with relatively mild angiographic CAD but markedly abnormal SPECT images have a less favorable outcome. One study²⁹ demonstrated that in patients with apparently “false-positive” ²⁰¹Tl images and “normal” epicardial coronary arteries, coronary blood flow response to acetylcholine infusion was blunted, indicating endothelial dysfunction (Fig. 14-8). This suggests that “false-positive” images probably were true positives. A better measure of the true specificity of stress MPI is the assessment of normalcy rate in subjects with low (<3%) likelihood of CAD. In our own laboratory, the normalcy rate of ^99mTc sestamibi SPECT was 99% using quantitative analysis of exercise myocardial perfusion SPECT images acquired in 40 normal subjects.

Figure 14-8 Impaired endothelium-dependent vasodilation in abnormal exercise SPECT. Coronary blood flow response to increasing doses of acetylcholine (AC) and to papaverine (PAPA) in patients with normal or minimally diseased (<30% luminal narrowing) epicardial coronary arteries and abnormal (n = 13) or normal (n = 14) exercise SPECT thallium-201 (²⁰¹Tl) imaging. Patients with abnormal ²⁰¹Tl SPECT showed blunted response to endothelial-dependent vasodilation with AC compared to those with normal ²⁰¹Tl SPECT.

(Modified from Zeiher AM, Krause T, Schachinger V, et al: Impaired endothelium-dependent vasodilation of coronary resistance vessels is associated with exercise-induced myocardial ischemia. Circulation 91:2345–2352, 1995.)

ECG-GATED SPECT

Evaluating the left ventricular ejection fraction (LVEF) is important in any cardiac patient. Currently, more than 80% of all SPECT studies are acquired with ECG gating. Gated SPECT provides a means of assessing postexercise resting LVEF, as well as regional wall motion and wall thickening. The routine assessment of LVEF by ECG-gated SPECT has added an important dimension to the overall assessment of patients referred for exercise testing and MPI.³⁰ Although the appearance of motion and the change in count density on cine display of ECG-gated SPECT myocardial perfusion studies is in fact artifactual and caused by improved count recovery during the cardiac cycle due to diminishing partial volume effect, a linear relationship exists with myocardial thickening.³¹

Assessment of LVEF, either normal or abnormal, in a patient with exercise-induced ischemia has important clinical and management consequences.^32,33 Furthermore, it is not rare that a patient is diagnosed as having a previously unknown cardiomyopathy on the basis of the information from a gated SPECT. In some patients, postexercise LVEF may be significantly lower than LVEF on rest SPECT imaging.³⁴ This may indicate postexercise ischemic stunning, which is a marker signifying severe CAD, although the specificity of this finding reflecting stunning may be limited in the setting of a large reversible defect.³⁵

Visual assessment of regional myocardial thickening on gated SPECT has also improved the diagnostic yield of exercise MPI by improving identification of attenuation artifacts and thereby increasing the certainty of the image interpretation. Specifically, the inspection of gated SPECT images has been shown to decrease the number of “borderline” interpretations by 68%, increasing the number of normal studies in patients with a low likelihood of CAD and increasing the detection of abnormal perfusion in patients with a known history of CAD.³⁶

When nongated exercise thallium SPECT was compared to ECG-gated exercise sestamibi SPECT images in women, specificity improved from 67% to 92%, mainly because of better identification of breast attenuation.³⁷ A fixed defect that shows normal wall thickening is most likely due to attenuation. Unfortunately, attenuation artifacts, either caused by breast or diaphragm, may be at times more severe on stress images than on rest images, mimicking myocardial ischemia. In this scenario, normal wall thickening is unhelpful in differentiating between the two possibilities. It is then necessary to consider other available clinical and exercise information to solve this conundrum. However, when a rest defect is more severe than a stress defect and regional wall thickening is normal, the defect in question is very likely artifactual.

It has been shown that the quality and accuracy of LVEF derived from ECG-gated SPECT studies are dependent on the amount of injected dose, stable heart rhythm, size of left ventricle, and the amount of gastrointestinal activity adjacent to the heart.³⁸ Sixteen frames per R-R interval acquisition provide more accurate (~5% higher) LVEF than the traditionally used 8-frame acquisition.³⁹ Although in many laboratories, both the rest and exercise studies are acquired in ECG-gated mode, gated SPECT acquisition is preferably applied to the higher dose study. Provided that a SPECT study has adequate count density, no gating problems due to arrhythmias, no intense adjacent gastrointestinal activity, and normal left ventricular size, LVEF quantification by gated SPECT is highly reproducible and accurate in most patients.⁴⁰ Most software packages also provide end-diastolic and end-systolic volumes.⁴¹ These volume measurements generally demonstrate greater variation than that of LVEF. Normal values for gated SPECT LVEF are significantly lower in men than in women (median LVEF: 52% versus 56%), whereas normal end-diastolic volumes are larger in men than in women (median volume: 109 mL versus 79 mL). In addition to the assessment of global left ventricular function, which is an important diagnostic and prognostic variable in cardiac patients, evaluation of regional wall motion has become an extremely useful aid for identifying attenuation artifacts and thus has helped to improve specificity.³⁷

MOTION CORRECTION

Technologists use a number of ways to ensure that patients remain immobile during image acquisition. Velcro straps are commonly used to immobilize a patient’s arms and body. Furthermore, technologists emphasize to patients the importance of lying still during image acquisition. Nevertheless, a substantial number of patients show motion during the acquisition. All vendors provide motion correction software. However, it is not always possible to correct appropriately for motion in all directions. Prevention is better than correction.⁴²

STATE-OF-THE-ART MYOCARDIAL PERFUSION SPECT WITH ATTENUATION CORRECTION

One of the main obstacles encountered in the interpretation of SPECT myocardial perfusion images is the influence of soft-tissue attenuation on the generation of a perceived perfusion defect that is actually an artifact. The most common of these artifacts arise from attenuation caused by breast tissue, diaphragm, or obesity. Although the artifacts can be evaluated to a certain extent through the use of prone imaging to identify diaphragmatic attenuation or gating of the SPECT images to evaluate regional wall motion, attenuation correction is the optimum method of dealing with these artifacts.

Many gamma cameras are presently equipped with non-uniform attenuation correction devices as an option. Each vendor has championed different approaches consisting of scanning external line sources in varying configurations or x-ray CT for generating transmission attenuation maps. No one approach has yet been identified as clearly superior over the others, but the increasing interest in hybrid imaging will likely result in a greater reliance on x-ray CT for attenuation correction.

Recent clinical studies using non-uniform attenuation-correction devices (Table 14-5) show improved specificity and normalcy rate with preserved sensitivity compared to uncorrected studies.^43–47 In one study, it was suggested that attenuation correction not only improves specificity but also enhances the identification of multivessel disease.⁴⁸ In clinical practice, experienced interpreters are well capable of recognizing attenuation artifacts from indirect evidence and thus avoiding false-positive interpretations. However, it is of interest to note that the application of attenuation correction has been shown to improve the overall diagnostic performance of individual readers with different interpretive attitudes.^47,49

In clinical practice, the performance of attenuation correction devices has not been consistently successful, in part due to the accentuation of scatter from adjacent bowel activity or improper registration of CT and SPECT images.⁵⁰ Although significant progress has been made, further development of scatter correction algorithms is necessary. It is important that rigorous and practical criteria for quality assurance of attenuation maps and their registration with the SPECT images be developed to improve the reliability of attenuation correction in every patient.

DISPLAY OF EXERCISE SPECT IMAGES

Tomographic reconstruction of the heart generates a multitude of images. To simplify image interpretation, the display of SPECT images has been standardized (Fig. 14-9).⁵¹ This is also important for communication and exchange of images between imaging facilities. Three sets of reconstructed tomographic slices are usually displayed for interpretation: short-axis slices, horizontal long-axis slices, and vertical long-axis slices (Fig. 14-10). Stress and rest (or delayed) images are displayed in two rows (stress on top and rest below) to facilitate comparison. The short-axis slices are displayed from apex (upper left) to base (lower right), the vertical long-axis slices are displayed from septum (left) to lateral wall (right), and the horizontal long-axis slices are displayed from inferior wall (left) to anterior wall (right).

Figure 14-9 Standardized display of stress/rest SPECT images. Representative normal exercise stress and rest SPECT images. The reconstructed short-axis slices (top four rows), vertical long-axis slices (rows 5 and 6), horizontal long-axis slices (rows 7 and 8) shown are displayed according to the format explained in Figure 14-7.

Figure 14-10 Standardized display of reconstructed SPECT slices. Stress and rest slices are displayed in alternating rows with the stress slices on top. Short-axis slices are displayed with the right ventricle at the left on the image. The slices are cut from apex to base and displayed from left to right. Vertical long-axis slices are displayed with the apex pointing towards, the right. The slices are cut from septum to lateral wall and displayed from left to right. Horizontal long-axis slices are displayed with the right ventricle on the left with the apex pointing up. The slices are cut from inferior wall to anterior wall and displayed from left to right.

(Reproduced with permission from Standardization of cardiac tomographic imaging. From the Committee on Advanced Cardiac Imaging and Technology, Council on Clinical Cardiology, American Heart Association; Cardiovascular Imaging Committee, American College of Cardiology; and Board of Directors, Cardiovascular Council, Society of Nuclear Medicine, Circulation 86:338–339, 1992.)

Images are preferably displayed on a monitor using a linear grayscale, a monochromatic color scale, or a multicolor scale. It is important for the interpreter to routinely use one standard display mode and not change frequently to different color scales. Certain color scales have a tendency to exaggerate subtle differences in myocardial radiotracer uptake, whereas other color scales may have the opposite effect.

To simplify and standardize interpretation further, myocardial slices are divided into segments.⁵¹ The American Heart Association, the American College of Cardiology, and the American Society of Nuclear Cardiology adopted a 17-segment model for interpretation of nuclear, echocardiographic, and magnetic resonance images (Fig. 14-11).^51,52 Each segment can be assigned to one of three coronary artery perfusion territories, as shown in the diagram in Figure 14-11. A detailed three-dimensional atlas of coronary anatomy territories for MPI has been developed based on extensive angiographic correlation (Fig. 14-12).⁵³ Last, the nomenclature for each segment has been standardized as well (Fig. 14-13). The most important difference with the older nomenclature is that the term posterior is no longer used.

Figure 14-11 Coronary artery territories on SPECT reconstructed slices. The standard 17-segment model is shown with assignment of segments to the territories of the left anterior descending coronary artery (LAD), right coronary artery (RCA), and left circumflex coronary artery (LCX).

(Reproduced with permission from Cerqueira MD, Weissman NJ, Dilsizian V, et al: Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart: A statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association, J Nucl Cardiol 9:240–245, 2002.)

Figure 14-12 SPECT myocardial perfusion zones and atlas of coronary arteries. AV, atrioventricular; D1, first diagonal branch; D2, second diagonal branch, RI, ramus intermedius.

(Reproduced with permission from Nakagawa Y, Nakagawa K, Sdringola S, et al: A precise, three-dimensional atlas of myocardial perfusion correlated with coronary arteriographic anatomy, J Nucl Cardiol 8:580–590, 2001.)

Figure 14-13 Standardized nomenclature of 17 segments. The display is the bull’s eye plot. The segmentation and numbers are the same as in Figure 14-8.

(Reproduced with permission from Cerqueira MD, Weissman NJ, Dilsizian V, et al: Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart: A statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association, J Nucl Cardiol 9:240–245, 2002.)

SEMIQUANTITATIVE VISUAL ANALYSIS AND QUANTITATIVE ANALYSIS OF MYOCARDIAL PERFUSION IMAGES

Myocardial perfusion images should not be interpreted simply in a binary fashion as either “normal” or “abnormal.” Myocardial perfusion abnormalities should be characterized according to the degree of decreased radiotracer uptake. A commonly employed scoring system uses a 4-point scale: 0 = normal; 1 = mildly reduced; 2 = moderately reduced; 3 = severely reduced; 4 = absent uptake.⁵⁴ By applying this scoring system to each segment of the 17-segment model, to both the rest and stress images, a summed stress score, a summed rest score, and a summed difference score can be derived. These semiquantitative visual scores have been shown to provide important prognostic information.² A normal image thus has a score of 0, whereas the maximal abnormal score is 68 (no heart visualized). A summed score of less than 8 is considered a small perfusion abnormality, 9 to 13 a moderate abnormality, and over 13 a large perfusion abnormality.

A purely visual assessment of SPECT images is subjective and may result in suboptimal reproducibility of the results. Therefore, an important complement to visual inspection of images is the quantitative evaluation of perceived perfusion defects, which is based on the inherently quantifiable nature of nuclear images. A number of validated software packages based on the concept of count profiles extracted from the emission tomographic images have been developed, and most are commercially available for quantification of myocardial perfusion and function (QPS-QGS; Emory Toolbox; 4D-MSPECT, and WLCQ). While evaluation of SPECT images using a quantitative program provides greater reproducibility than visual assessment,⁵⁵ a recent comparative study showed that the performance of currently available software packages for quantification of myocardial perfusion differs significantly in their diagnostic performance and therefore cannot be used interchangeably.⁵⁶

The basic quantitative approaches are similar for each of the programs: regional radiotracer uptake is quantitatively compared to normal databases, although each of the programs is based on slightly different models that are used to generate the quantitative profiles. Specific information about the various quantitative programs has been published in a recent review that will provide the reader with a greater understanding of the assumptions and modeling upon which the programs are based.^57–61 The difference is largely in the display of data. Four of the programs display relative radiotracer uptake on SPECT images as polar plots or bull’s eyes. The fourth package, WLCQ, used routinely in our laboratory, involves the generation of circumferential count distribution profiles.⁶² Circumferential count profiles are generated for each of the short-axis slices and displayed with a curve representing the lower limit of normal myocardial count distribution. This provides a readily appreciable quantitative measure of the degree of abnormality of the patient’s image compared to normal image files. Table 14-6 compares various ways to categorize the extent of myocardial perfusion abnormalities.⁶³

We believe that reliable quantification of myocardial perfusion images is extremely useful and should be used more frequently for the following reasons:

However, quantitative analysis should always be viewed as complementary to visual analysis. Image interpretation should always start with visual inspection of images. Quantitative display then serves to confirm the visual impression.

INTERPRETATION OF EXERCISE SPECT STUDIES

It is recommended that clinical interpretation of exercise SPECT images should follow a systematic approach to extract the maximal amount of information. The following routine is used in our laboratory:

Figure 14-14 Evidence of poor gating of SPECT images. In the volume curve for the optimally gated study (left panel), the end-diastolic volumes in the first and last frame bin are the same. In contrast, in the study with poor gating due to the presence of frequent premature ventricular contractions (right panel), the end-diastolic volume in the last frame bin is 50% greater than the end-diastolic volume in the first frame bin.

IMAGE INTERPRETATION

It is helpful to divide the numerous reconstructed short-axis slices into three groups: apical slices, midventricular slices, and basal slices. After interpretation of the myocardial free walls on short-axis slices, the apex and base of the left ventricle are reviewed on the long-axis slices. The following define the various image patterns and abnormalities that can be seen in myocardial perfusion images:

Normal. The uptake of radiopharmaceutical is homogeneous throughout the left ventricular myocardium. However, in studies that are not attenuation corrected, normal regional variation in uptake can occur in specific areas. Specifically, the lateral wall is usually hotter than other walls because of its closer proximity to the gamma camera head. Quantitatively normal images show regional radiotracer uptake that is above lower limits of normal distribution.

Care should always be taken in interpreting SPECT images as showing no perfusion abnormalities. SPECT imaging only reflects relative flow heterogeneity but offers no information about absolute flow. As a result, multivessel disease causing balanced ischemia may not be appreciated. It is therefore critical to evaluate all of the imaging and exercise data to identify any findings that might suggest the presence of diffuse ischemia.⁶⁵ For example, the evaluation of function by gated SPECT can increase the identification of abnormal segments in patients with three-vessel disease by almost 50% compared to perfusion imaging alone.⁶⁶ Furthermore, identification of transient ischemic dilation will increase the identification of high-risk patients with left main disease from 56% to 83%.⁶⁷

Defect. A defect represents a localized myocardial area with a relative decrease in radiotracer uptake exceeding normal variation. Quantitatively abnormal regional radiotracer uptake is below lower limits of normal. Defects may vary in intensity from slightly reduced activity to almost total absence of activity. Using the earlier-mentioned semiquantitative scoring system, mildly reduced uptake = 1; moderately reduced uptake = 2; severely reduced uptake = 3; and absent uptake = 4.⁵⁴ The degree of abnormality can be quantified in terms of the extent and severity of abnormal uptake. A defect can be further quantified as “percent of total left ventricle.” Severe breast or diaphragmatic attenuation may occasionally create a quantitative defect, but it is vital that the reader exclude the possibility that a true perfusion abnormality is contributing to the defect before interpreting the defect as being artifact.

Reversible Defect. A reversible defect is defined as a defect that is present on the initial stress images and no longer present, or present to a lesser degree, on resting or delayed images. This pattern usually indicates myocardial ischemia. Visually discerniable reversibility corresponds quantitatively to at least a 25% improvement of defect.

Fixed Defect. A fixed defect is defined as a defect that is unchanged and present on both exercise and rest (delayed) images. This pattern generally indicates infarction and the presence of scar tissue. However, in some patients with fixed ²⁰¹Tl defects on redistribution imaging, improved uptake can be noted after a new resting injection of ²⁰¹Tl or on 24-hour redistribution imaging. Similarly resting reinjection of ^99mTc-labeled agents after administration of oral nitrates may reveal defect reversibility. These additional imaging maneuvers can help to identify viable myocardium with severely decreased resting blood flow.

Reverse Defect. The initial images are either normal or show a defect, whereas the delayed or rest images show a more severe defect. This pattern is frequently observed in patients who had thrombolytic therapy or percutaneous coronary intervention for acute coronary syndrome. This pattern may persist for years, and the phenomenon is thought to be caused by initial excess radiotracer uptake in a reperfused myocardial area consisting of a mixture of scar tissue and viable myocytes. Initial excess accumulation is subsequently followed by rapid clearance from scar tissue. Although the clinical significance of this finding is controversial, it does not represent evidence of exercise-induced ischemia.⁶⁸

Lung Uptake. Normally no, or very little, radiotracer is noted in the lung fields on postexercise images. Increased lung uptake on planar images can be quantified as the lung-to-heart ratio (normal < 0.5 for ²⁰¹Tl and < 0.4 for ^99mTc-labeled agents).^69–71 Increased lung uptake of radiotracer represents an important abnormal image pattern, which is associated with an elevated left ventricular end-diastolic pressure,⁷² and indicates exercise-induced ischemic left ventricular dysfunction and severe multivessel CAD. Occasionally, increased lung uptake is also observed after pharmacologic stress and has a similar unfavorable significance. Not surprisingly, increased lung uptake occurs also in patients with severely decreased resting LVEF, with or without demonstrable exercise-induced ischemia.

Transient Left Ventricular Dilation. Occasionally, the left ventricle is noted to be larger following exercise than on the rest or delayed image.⁷³ This pattern is more likely caused by apparent thinning of the myocardium by circumferential endocardial ischemia rather than true and persistent dilation of the left ventricular cavity.⁷⁴ At times this image pattern may occur without apparent regional perfusion abnormalities.⁷⁵

Transient Right Ventricular Visualization. The right ventricle is more clearly visualized on postexercise images than on rest images. This pattern indicates ischemic left ventricular dysfunction during exercise.⁷⁶ The mechanism responsible for this finding remains unknown but may involve either increased right ventricular strain or a relative decrease in count intensity in the left ventricle due to diffuse hypoperfusion.

LOW-RISK AND HIGH-RISK IMAGE PATTERNS

Interpretation of myocardial perfusion images must take into account all of the above factors, because certain image patterns have important diagnostic and prognostic significance. It is important to consider both ends of the diagnostic spectrum of SPECT imaging. Categorization of outcome based on these various imaging parameters has been stratified into low, intermediate, and high risk, which are associated with a < 1%, 1% to 5%, and > 5% annual risk of cardiac death or nonfatal myocardial infarction, respectively.

Low-Risk Patterns

Patients with entirely normal or near-normal exercise myocardial perfusion images, even in the presence of known angiographic CAD, have a very low (<1%) yearly cardiac event rate.^{28,54,77–84} This was already demonstrated in the early 1980s for planar exercise ²⁰¹Tl and sestamibi images, and was subsequently confirmed in the 1990s in large numbers of patients with SPECT imaging. In recent years, it has been recognized that although this notion is generally correct, there are small but important subgroups of patients with a less favorable outcome (Fig. 14-15).⁸⁴ These subgroups are elderly patients, patients (especially women) with diabetes,⁸⁵ African-Americans,⁸⁶ and patients with known CAD. In addition, patients who require pharmacologic testing with either adenosine or dobutamine⁸⁷ have a relatively poorer outcome despite normal myocardial perfusion images. Obviously, patients who require pharmacologic stress testing often have additional significant comorbidity. Furthermore, patients who exercise less than 7 METs, patients who have ischemic ECG changes during adenosine infusion,⁸⁸ or patients who have transient ischemic dilation of the left ventricle⁷⁵ may have a higher cardiac event rate even though myocardial perfusion images appear normal. It was assumed previously that the “warranty period” of normal stress myocardial perfusion images was about 2 years. In view of the data, it seems reasonable that patients in the higher-risk categories should undergo repeat stress imaging after 1 year.⁸⁴

Figure 14-15 Annual rates of cardiac death or myocardial infarction based on the results of myocardial perfusion imaging. Overall, normal myocardial perfusion images are associated with low annual cardiac event rates, although the risk is affected by important patient factors and factors related to the patient response to the stressor. With the exception of the pronounced risk of cardiac events in patients with ischemic ECG changes during adenosine infusion with normal perfusion images, there is a graded increase in risk as the myocardial perfusion images become more abnormal.

High-Risk Patterns

When assessing the overall results of an exercise SPECT study, exercise variables should be taken into consideration as well. The Duke Treadmill Score provides a simple first estimate of risk that can be refined subsequently by considering additional information derived from SPECT imaging.¹ The Duke Treadmill Score is calculated as follows: exercise time (minutes) minus 5 times maximal exercise ST segment deviation (mm) minus 4 times angina index (0 = no angina, 1 = angina occurs; 2 = exercise-limiting angina) or can be assessed from a simple nomogram.¹ In addition to low exercise tolerance (<7 METs), exercise-limiting angina, and marked ST depression, exercise-induced hypotension, or exercise-induced ventricular arrhythmias indicate also poor outcome.

High-risk SPECT myocardial perfusion images (Fig. 14-16) are characterized by any one or more of the following image patterns:

• A large exercise-induced myocardial perfusion defect (see Table 14-2)

• Multiple myocardial perfusion defects in more than one coronary artery territory (see Table 14-6)

• Extensive reversibility of an exercise-induced defect

• Abnormal postexercise LVEF, normal rest LVEF

• Increased postexercise lung uptake

• Transient (postexercise) ischemic left ventricular dilation

• Transient (postexercise) right ventricular visualization

• Abnormal rest LVEF

Figure 14-16 A-D, Abnormal and high-risk ^99mTc sestamibi exercise/rest SPECT images of a 63-yr-old male with diabetes and atypical chest discomfort. A and B, The reconstructed exercise SPECT slices show an enlarged left ventricle with a large anteroseptal and apical myocardial perfusion defect that shows partial reversibility. Transient ischemic dilation and transiently enhanced visualization of the right ventricle can be noted. These three image features indicate a high-risk study. C, Three-dimensional representation of the slices shown in A and B. On the left, surface rendering of four myocardial walls: septal, anterior, lateral, and inferior. Relative radiotracer uptake is shown normalized to the area with maximal uptake (i.e., lateral wall). The large perfusion defect involving all walls except the lateral wall can be appreciated. Although partial defect reversibility is present in the anteroapical wall, a large rest defect in the same area is present. On the right, comparison to normal data files. Abnormal areas are indicated in color and normal areas in white. D, Quantification of relative radiotracer uptake by circumferential profile analysis in comparison to a normal database. Blue, stress count distribution profiles; red, rest count distribution profiles; green, lower limit of normal count distribution. Representative apical, midventricular, and basal and apex profiles are shown. The blue exercise curves are below the lower limit of normal in the anteroseptal region from apex to base. The red rest curves show improvementöthat is, closer to the lower limit of normalöfrom apex to base. The overall exercise defect is large (38% of the left ventricle), the rest defect is also large (16% of the left ventricle), and the ischemic burden is large as well, involving 22% of the left ventricle.

When one or more of these exercise/image patterns are present, the patient should be considered to be at high risk for future cardiac events.

REPORTING SPECT IMAGING RESULTS

The results of exercise SPECT may have an important impact on decision making and management of cardiac patients. If communication with referring physicians is ineffective, patients may not fully benefit from nuclear cardiology procedures. In recent years it has become clear that in many imaging facilities, the quality of reporting was inadequate.⁸⁹ Reports should be concise and clear. In most circumstances, it is possible to decide if a study is normal or abnormal. If a study is abnormal, the degree of abnormality should be defined (i.e., small, moderate, or large) as well as the type of abnormality, fixed or reversible. The anatomic location of a perfusion defect, the possible coronary territory involved, and presence of high-risk features should be stated. Interpretation should also be formulated with consideration of exercise parameters: one should never interpret images alone. A relatively small exercise-induced myocardial perfusion abnormality has markedly different clinical significance in a patient who achieved only a low workload as compared to a patient who exercised for 15 minutes. Reasons for suboptimal quality of images that affect the confidence of interpretation (attenuation artifacts, intense gastrointestinal uptake, motion) should be mentioned as well. Examples of optimized reports have been published.^89,90

Exercise radionuclide MPI is a mature, powerful, and well-validated clinical tool. Optimal results are obtained when careful attention is paid to many details of the entire process, from performing the exercise stress, to optimized image acquisition, thoughtful image interpretation, and formulating a meaningful report.