CHAPTER 19: POSITRON EMISSION TOMOGRAPHY IN HEART DISEASE

Overview: Myocardial Perfusion and Metabolism—Now and at the Beginning

Low coronary blood flow may cause acute coronary syndromes, angina pectoris, heart failure, arrhythmias, and even death, as reviewed in detail in Chapter 34. The capacity for increased coronary blood flow is associated with cardiovascular health, and its variations may reflect our emotional states and lifestyles, including the foods we eat—even our most recent meal. Intense risk factor treatment prevents or stabilizes atherosclerosis, thereby preserving coronary blood flow, which is the most basic vital sign of life on which blood pressure, heart rate, and respiration depend. Assessing, maintaining, or restoring coronary blood flow are the objectives of all coronary procedures and most cardiac diagnostic testing. Next to “heart,” the commonest critical word in cardiovascular practice is “ischemia,” implicitly referring to low myocardial perfusion.

Coronary blood flow is a personalized, vital sign of individual status. When severely depressed, similar to blood pressure, it heralds imminent adverse events in an individual, not just a statistical probability of adverse outcome as do risk factors or the diagnosis of coronary artery disease (CAD). Advanced imaging technology and an extensive research literature document its measurement and relevance. Yet quantitative assessment of myocardial perfusion or coronary blood flow remains largely unused and unmeasured in clinical practice; its use is restricted to a research tool in a small number of academic centers where it usually has limited impact on clinical decisions. However, in view of revascularization trials failing to improve event-free survival in patients with CAD or reduced left ventricular function, quantitative perfusion and metabolic imaging should play a larger role in selecting patients optimally benefiting from revascularization.

How does the “now” of coronary blood flow in acute coronary syndromes relate to “the beginning” of the four-chambered mammalian heart that had already evolved 200 million years ago? A stem mammal (eg, Morganucodon watsoni) was a 1-inch long weasel-like animal during dinosaurs' domination for 250 million years.¹ The Chicxulub asteroid impact² and associated Deccan volcanism blocked the sun, causing the end-Cretaceous mass extinction of most life.^3,4 However, small mammals survived,⁵ radiating into multiple open niches and large sizes, reflecting adaptability to extreme demands and environments that required an extraordinary heart and cardiovascular system before large brains and facile limbs evolved. As documented in humans,⁶ canines,⁷ and swine,⁸ the branching coronary artery tree of mammals is structured in mathematically exact arterial size and length for myocardial mass.^6,7

This arterial size-length-mass relationship and the branching structure of the coronary arterial tree are precisely related to coronary blood flow by a power equation for minimum energy loss. This relationship is mediated by normalized wall shear force to distribute efficiently the oxygenated coronary blood to myocardium⁶ while preserving capacity for high demands of coronary blood flow or coronary flow reserve (CFR).^9,10 The same equation characterizes the observed mother-daughter relationship of coronary arterial branch diameters.⁶ This power equation may even explain some of the different manifestation of CAD in women compared to men, as addressed in Chapter 34. The heart’s phasic contraction and relaxation is the oldest, most developed mechanism for pumping its own fuel supply and systemic needs. Evolutionary functional demands drove coronary physiology, thence anatomy, to its current essential role for understanding and managing CAD addressed in this chapter on clinical quantitative myocardial perfusion by positron emission tomography (PET).

Cellular metabolism and contractile proteins evolved for survival even earlier than physiology or coronary blood flow.^11,12 One of many such mechanisms is myocardial metabolism that switches from aerobic metabolism of fatty acids when there is adequate supply of oxygenated coronary blood to anaerobic metabolism of glucose when coronary blood flow is impaired or hypoxic.¹³ Under such adverse circumstances, the flow-deprived myocardium “hibernates” or stops contracting but remains “alive or viable” for potential return of normal contractile function when coronary blood flow is restored. Transient severely impaired coronary blood flow may temporarily stop or “stun” myocardial contraction after the low-flow episode is over, with slow contractile recovery over hours to days after normal blood flow is restored. PET metabolic or viability imaging drives optimal personalized management of patients with advanced coronary artery disease and impaired left ventricular function as addressed in this chapter.

Although current literature provides many excellent supporting papers and reviews,^14–42 quantitative myocardial perfusion and metabolism remain at the edges of mainstream cardiology despite their value in relation to myocardial ischemia (see Chap. 34). Consequently, this chapter on PET takes a different approach. Rather than providing a detailed summary of the literature on the current “average” use of cardiac PET perfusion imaging, this chapter emphasizes two major themes developed in the two leading centers prominent in developing clinical cardiac PET—the David Geffen School of Medicine at the University of California at Los Angeles and the McGovern Medical School, University of Texas–Houston.

First, we use quantitative PET to elucidate complex, but clinically relevant, coronary pathophysiology and integrate its complexity into comprehensive decisional “smart” analytical displays to guide binary clinical decisions personalized for individuals. Second, we demonstrate, by case examples, how quantifying myocardial perfusion and metabolism drives personalized assessment of CAD, its severity, and its management, including invasive procedures and revascularization for individuals or for optimally determining average group treatment outcomes. For conceptual simplicity, the term coronary blood flow is used interchangeably with myocardial perfusion because of their physiologic interchangeability of mechanisms and consequences while having slightly different units of cc/min versus cc/min/g due to different measurement methodology.

Each case illustrates a major class of clinical conditions uniquely addressed by PET, providing personalized insight so that clinical readers and patients can see its value in the delivery of individually tailored cardiovascular practice, and not just as a remote research tool capable of generating bulk data without personalized relevance. Although somewhat unconventional, our approach provides the vision for readers, cardiologists, and patients to see what can and should be the optimal, objective, and personalized mechanism for establishing and managing CAD. The quantitative, personalized data from PET allows for individualized decisions about the intensity of therapy recommended—lifestyle changes alone or combined with medical therapy with or without revascularization. In essence, PET provides quantitative data highly specific for each individual’s personalized coronary care, with a scientific and technically valid basis, that physicians and patients themselves visually readily understand, with a corresponding powerful influence on their management.

Why Cardiac Positron Emission Tomography for Myocardial Perfusion?

Diagnostic Sensitivity and Specificity

The diagnostic sensitivity and specificity of PET for the diagnosis of CAD are 90% to 95% and 88% to 95%, respectively, and are higher than those for single-photon emission computed tomography (SPECT) at 85% and 85%, respectively.^24,25,43,44 The area under the receiver-operating characteristic for PET was 0.95 compared to 0.90 for SPECT, a significant difference (P < .0001).²⁴ This greater diagnostic accuracy of PET versus SPECT results from better resolution and attenuation correction of PET.

However significant, these comparisons fail to demonstrate the enormous differences between these two technologies for several reasons. In these studies, the reference gold standard for defining CAD to which PET and SPECT were compared was visually interpreted angiographic percent diameter stenosis. The variability of visually estimated percent diameter stenosis; its poor correlation with coronary blood flow or myocardial perfusion; and failure to account for diffuse narrowing, complex anatomy, or arterial remodeling contravene the validity of the angiogram as a reference for identifying or assessing either anatomic or physiologic severity of stenosis or diffuse disease.^{16,31,37,38,39,40,41} When compared to objective automated quantitative coronary analysis (QCA) of angiographic severity of discrete coronary artery stenosis, diagnostic sensitivity and specificity of PET perfusion images approaches 95%.³⁴

However, even QCA remains limited as a gold standard of severity of CAD,^{16,31,37,38,39,40,41} for several reasons. The coronary angiogram fails to quantify the essential and more basic physiologic function of the coronary arteries now shown to define severity for guiding management.^16,31,39 Diffuse epicardial atherosclerosis cannot be measured. Epicardial endothelial function or small-vessel disease may profoundly affect coronary blood flow but is not accounted for on an angiogram. Moreover, coronary blood flow relates to the arterial diameter raised to the fourth power so that small changes in arterial size that cannot be measured accurately on an arteriogram markedly alters coronary blood flow. Finally, the varying diameters and irregularities over different lengths, angles, and shapes affect coronary blood flow but are not reliably accounted for on anatomic images. The computed tomography (CT) angiogram, with best resolution of 0.5 mm, has a large uncertainty at each arterial border that precludes projecting or calculating blood flow capacity in 3- to 5-mm diameter coronary arteries because the flow-diameter to the fourth power relation magnifies the uncertainty of arterial dimensions.³⁹ Coronary anatomic-functional relationships are detailed further in Chapter 34 on coronary blood flow and myocardial ischemia.

Fractional flow reserve (FFR), measured by trans-stenotic invasive pressure wire during cardiac catheterization, was originally validated and compared to CFR by quantitative PET perfusion imaging.⁴⁵ Although FFR has advanced cardiovascular science from anatomic characterization to one based on the physiologic severity of stenosis to guide percutaneous coronary intervention (PCI),^46,47 FFR is also an inadequate reference standard because of the substantial discordance between FFR and CFR by invasive technologies and also by quantitative and relative myocardial perfusion by PET.^48,49 Moreover, FFR was low or discordant with the absence of visually significant stenosis when all arteries were interrogated as opposed to measuring FFR in only arteries with stenosis visually severe enough to be considered for PCI.⁵⁰

While offering a major advantage over SPECT imaging, relative PET perfusion images fail to utilize PET capacity for absolute quantification of perfusion in cc/min/g and absolute CFR. Such quantitative measurements move PET into the highest level of assessing CAD in order to personalize the physiologic severity of CAD that is unmatched by any other technology. To date, all trials of elective revascularization fail to reduce myocardial infarction (MI) or cardiovascular death, despite relief of angina.^51,52,53,54 Consequently, simply diagnosing CAD should no longer be considered the primary, or even an adequate goal of diagnostic testing, because establishing evidence-driven physiologic severity is now the optimal scientific basis for revascularization versus medical management alone.

The evidence-driven goal of diagnostic testing now requires objectively identifying those patients with quantitative physiologically severe CAD of sufficiently high risk in whom revascularization would reduce the risk for MI or mortality, while also relieving angina.^{35,36,37,38,39} Quantitative assessment of perfusion provides comprehensive personalized quantification of physiologic severity of CAD, integrating information from each individual coronary artery and its secondary and tertiary branches to delivery an integrated gold standard. Within the confines of biologic and methodological variability, quantitative perfusion defines the flow through a coronary artery as a vital sign essential for survival, just like blood pressure, where a defined low range may bode illness.

As addressed in Chapter 34, randomized trials are needed to determine the threshold at which coronary perfusion or pressure is so low that the risk of adverse events is high enough to justify revascularization as a means to improve outcomes. However, a deep truth resides in individual cases, representative of the variations of CAD, where quantitative PET provides unique insights and management guidance as the basis for its current clinical use. The cases also illustrate the need for more advanced interventional trials; those completed to date lack adequate low flow thresholds high enough risk for optimal benefit from revascularization.

The Coronary Flow Capacity Map: Complexity Made Simple for Personalized Patient Decisions

Three measures of coronary blood flow, or myocardial perfusion, quantify its profound effects in all manifestations of CAD. Rest and stress myocardial perfusion in cc/min/g and CFR as the ratio of stress perfusion to rest perfusion on a regional pixel basis are the final common pathways of coronary anatomy and physiology on patient outcomes. Myocardial perfusion is commonly heterogeneous regionally and temporally associated with multiple, interacting mechanisms that may contravene some providers' preference for a simple, binary, positive or negative, one-result-fits-all test to guide management. Rather than a diagnostic limitation of methodology or biological variability, these three perfusion measurements provide rich insights into the reality of heterogeneous coronary pathophysiology essential for individualized management when integrated to provide a valid binary directive—“complexity made knowledgeably simple.” These insights for individual personalized management provide an equally rational basis for randomized trials of “group” management or revascularization based on quantitative physiologic severity of CAD.

Accordingly, at the outset, we account for and incorporate these three measures into a coronary flow capacity map color-coded for severity from healthy young volunteers to patients having angina and significant electrocardiogram (ECG) changes during dipyridamole stress PET imaging.^55,56,57 After we demonstrate its clinical utility, we then address the technical details of quantifying myocardial perfusion. Figure 19–1 orients topographic PET images derived from tomographic views, as previously described.^{16,31,37,40,48,55,56,57}

Stress flow and CFR require integrated interpretation caused by frequent discordance of rest and stress flow, as seen in Fig. 19–2. High rest flow (A, white) caused by a high pressure-rate product, anxiety, medications, or female sex,⁴¹ either separately or cumulatively (proportionately higher than stress flow) may cause low CFR (A, blue) that alone erroneously implies flow-limiting stenosis. Alternatively, CFR may be high (B, white or red) with no ischemia despite low flow at rest and stress (B, blue) because of beta-blockers, physical conditioning, increased vagal tone, or CAD with low stress flow erroneously suggesting ischemia if it is considered alone without CFR. By plotting values for each pixel of stress flow in cc/min/g on the horizontal axis and CFR on the vertical axis (C), all three perfusion measures for each pixel are accounted for in the coronary flow capacity map panel (D, red) because CFR is the ratio of stress to rest flow. Caffeine, beta-blockers, inadequate vasodilator stress, and diffuse or small-vessel disease may lower stress perfusion in cc/min/g to apparently low ischemic levels; therefore, measuring rest flow and CFR commonly provides essential diagnostic information for correct clinical interpretation from the coronary flow capacity map.

The coronary flow capacity map plot is color-coded red for stress flow and CFR in 241 young healthy volunteers with no risk factors, no medical conditions, no measurable blood caffeine, and no measurable urinary cotinine as an objective test for smoking, as previously reported.^40,45,56,57 The lower limit of red is the mean minus one standard deviation (SD) of CFR and stress flow for this group. Blue indicates the optimal stress flow and CFR thresholds for patients with angina, ECG changes, and/or a regional stress defect during dipyridamole stress PET imaging. Green indicates patients with either angina or ECG, changes but not both, during stress imaging with a regional stress defect. Orange indicates patients with CFR and stress flow at the lower one SD of the healthy volunteers. Yellow indicates the upper one SD above the patients with angina or ST-segment changes during stress imaging. The orange-to-yellow transition is halfway between these SD limits.

Each color-coded pixel is then mapped to its original position and displayed as the coronary flow capacity map in four projections (lateral, inferior, septal, and anterior views) showing coronary flow capacity comparable to that in young healthy volunteers within the distribution of the coronary arteries (D). The specific ranges of stress perfusion in cc/min/g and CFR for each color used for all PET perfusion images in this chapter are detailed in Fig. 19–3.

Complete relevant perfusion information in a single clinical case involve, five sets of different data in four views involving 6720 pixels spatially distributed over a total of 20 different image views. Each of these views has 5 continuous variables (rest-stress uptake, rest-stress flow and CFR) color coded into 7 ranges from healthy young volunteers to patients with overt ischemia during dipyridamole stress as in Figs. 19-3 and 19-4. The rest and stress relative images (A) are abnormal with subtle differences between the distal anterior region (green) and the distal septum (yellow) that are difficult to interpret, inconsistent with a single left anterior descending coronary artery (LAD) territory and providing little insight into the detailed regional severity and size of abnormal stress perfusion.⁵⁷ Rest perfusion and stress perfusion images (B) are proportionately larger and more severe than the relative perfusion defects, but are complicated by spatial heterogeneity of both rest and stress perfusion. The CFR display (B) reflects the rest-stress perfusion heterogeneity into better defined normal or abnormal regions but still fails to define the subtle differences between the distal septum and the distal anterior regions on the relative images. Visually interpreting or simply quantifying the 6720 pixels spatially distributed in the 20 different views is difficult or limited for even an experienced reader.

However, based on the color-coded pixel graph (C), the coronary flow capacity map (D) integrates all these data into a single four-view color-coded image for the entire range of flows from active ischemia (blue) to flow in healthy young volunteers (red), accounting for all the flow data in a particular patient.⁵⁷ The map demonstrates preserved flow capacity (red) in left circumflex coronary artery (LCX) and right coronary artery (RCA) distributions, minimally reduced flow capacity on the proximal LAD distribution tapering to mild (yellow) diffuse narrowing in the distal LAD and thence to the moderate (green) border zones around a small severe stress defect (blue) in the distribution of a small distal diagonal branch off the patent but diffusely narrowed distal LAD. The small apical defect shows myocardial steal, indicating an occluded collateralized distal diagonal branch. The PET findings are confirmed by the angiogram inset done as part of a research protocol because the defect is too small and distal to warrant PCI.

Clinical PET Cases: Personalized to Quantify Severity for Guiding Management

The following fourteen cases visually demonstrate the principles and personalized management for the wide range of manifestations of CAD for which PET uniquely drives patient management as examples seen daily at the Weatherhead PET Imaging Center for Preventing and Reversing Atherosclerosis. The clinical power of cardiac PET illustrated in these cases requires knowledge of coronary physiology and myocardial ischemia addressed in the chapter on coronary blood flow and myocardial ischemia (Chap. 34). Here, to begin, quantitative PET is integrated with coronary physiology as comprehensive analytical displays of complex perfusion data made knowledgeably “simple” for clinical use in both physician and patient understanding. Later sections address technical aspects of measuring myocardial perfusion by PET.

Case 1: Coronary Flow Capacity in Multivessel Complex CAD

A 63-year-old man with 6 months of mild stable angina (Fig. 19–5)⁴⁷ illustrates the complex data inherent in all quantitative perfusion imaging that is integrated into a “simple,” easily understood image for clinical decisions by validated directive software based on a large patient database detailed in a later section. Figure 19–5 displays relative myocardial perfusion images of rubidium (Rb)-82 at rest (A) and during dipyridamole stress (B). With the arterial input function as detailed later, stress perfusion in cc/min/g (C) and CFR (D) for every pixel in every topographic image in each of four topographic views are plotted (E), color-coded and back-projected to the coronary flow capacity map (F).

For the patient whose images are depicted in Fig. 19–5, the coronary flow capacity map (F) shows occlusion or subtotal stenosis of the mid LAD, of the RCA, and of the distal LCX. Coronary flow capacity is severely reduced in 47% of the left ventricle (LV), of which 28% of the LV shows myocardial steal in the central area of the stress defect, a typical pattern indicating collaterals beyond occlusions or subtotal stenosis. There is a more proximal moderate or “intermediate” LAD stenosis involving the first septal perforator (yellow, 22% of the LV) with further reduced border zones (green, 10% of the LV). The diagonals and ramus intermedius distributions have only minimally reduced (orange, 17% of the LV) or good (red, 4% of the LV) coronary flow capacity. Because of the definitive images of complex multivessel disease, the patient was scheduled for coronary artery bypass graft (CABG) surgery with preoperative coronary angiography confirming the PET findings. He has done well over the following 3 years with a good flow capacity on follow-up PET.

Case 2: Saved from Unnecessary CABG Surgery

A 76-year-old asymptomatic man with preserved left ventricular ejection fraction (LVEF) had PET for a second opinion after CABG surgery was recommended following abnormal SPECT imaging and angiography reported to show the following stenoses: 70% left main; 80% LAD; 80% first diagonal; 70% first obtuse marginal; and 70% and 80% serial stenoses in the RCA. In Fig. 19–6, his rest-stress relative uptake images show no stress defect but only a small inferoapical motion artifact from the inferior apex recoiling briefly into the most inferior image plane during systole. The inferior apex is outside this lowest imaging plane during diastole, lasting two-thirds of the contraction cycle, thereby acquiring less activity. It normalized on ECG-gated systolic images. As the coronary flow capacity map shows, 69% of the LV has flow capacity comparable to that of healthy young volunteers, with 21% of the LV having minimal reduction (orange) and 10% with mildly reduced flow capacity consistent with age and mild, diffuse, nonobstructive, calcific CAD with coronary flow capacity well above the low ischemic flows that would be coded green or blue if present. The patient was advised against CABG surgery and did not have it performed. After the PET, the prior angiogram on which the recommendation for CABG surgery was based was obtained for review by the author K. Lance Gould (KLG). It showed luminal irregularities, mild nonobstructive CAD, and arterial lip dividers at branch points with no significant stenosis that were overinterpreted by the prior cardiologist. The SPECT images were false positives.

Case 3: Missed Severe CAD with Normal Treadmill Test

A 62-year-old asymptomatic man with risk factors but no known CAD recently had a normal ECG exercise stress test, and he underwent PET in a research protocol. He had a history of back pain related to activity thought to be the result of degenerative disc disease that he experienced during his normal treadmill stress test. His relative myocardial uptake images in Fig. 19–7 show a large, severe, anterior, septal and apical stress-induced defect involving 55% of the LV in the distribution of the LAD proximal to the first septal perforator and wrapping around the apex and up the distal inferior wall.

Absolute myocardial perfusion capacity was severely reduced, with myocardial steal in 37% of the LV indicating occlusion just distal to the first septal perforator. The proximal LAD had additional proximal moderately severe stenosis involving the first septal perforator (green, 10% of LV) with small-vessel or diffuse disease of distal the RCA supplying posterior perforators (yellow, 23%). The LCX had preserved coronary flow capacity (red, 14% of LV) with mild diffuse disease tapering to yellow distally (orange, 17% of LV). The CT scan done for attenuation of PET data shows moderate coronary calcification in the LAD and RCA. Gated PET perfusion images showed normal left ventricular contraction with an LVEF of 73% at stress. Based on the PET scan, medical therapy and coronary angiography with revascularization were recommended due to the size and severity of the rest-stress PET abnormalities. An angiogram confirmed the PET findings, for which his cardiologist had CABG carried out.

Case 4: Complex Post PCI Perfusion—Septal Perforator Caged by LAD Stent

A 58-year-old man with atypical chest discomfort, separate episodes of dyspnea—at night or inconsistently while walking—had PET as part of a research protocol. His baseline coronary flow capacity map in Fig. 19–8A shows a severe stress defect typical of a severe stenosis of the LAD proximal to the first septal perforator and wrapping around the apex. An angiogram confirmed these findings, and PCI was done with a proximal stent. The post-PCI PET showed a residual small, severe, basal septal, stress defect comprising 4% of the LV in the distribution of the first septal perforator caged by the LAD stent. In our experience, most coronary branches caged by stents have reduced coronary flow capacity even when open on post-PCI angiogram.

Case 5: Normal Relative Myocardial Uptake Images Miss Severe Diffuse CAD

A 56-year-old man with mild stable angina had a PET scan as part of a research protocol. During dipyridamole stress he had angina with 1 mm ST-segment depression with normal relative uptake images in Fig. 19–9. Despite completely normal relative stress images (A), stress perfusion in cc/min/g (B) and CFR (C) were severely reduced to low flow ischemic levels caused by severe diffuse CAD. The histogram color bars to right of each image show color-coded severity and percent of the LV in each range of severity. He developed angina and greater than 1 mm ST-segment depression at 5 minutes (early stage 2) of the Bruce protocol. An angiogram confirmed severe diffuse disease with superimposed visually estimated 70% stenosis in all three epicardial arteries. He underwent CABG surgery, after which his symptoms resolved and treadmill duration increased to 7 minutes without angina but with still with greater than 1 mm ST-segment depression. At follow up PET, global stress flow increased from 1.25 to 2.0 cc/min/g reflecting improved coronary flow capacity but still limited by residual diffuse CAD at the borderline threshold for ischemia.

Case 6: Complex Post-PCI Angiogram with Question of Left Main

A 52-year-old physically active asymptomatic man had PET for a second opinion for CABG surgery. He had a prior PCI with two stents to the LAD, with a moderate estimated 50% to 55% diameter left main stenosis that was not stented. Treadmill exercise stress testing and myocardial perfusion imaging were reported as normal; however, the patient developed nonsustained supraventricular tachycardia in recovery from exercise, leading his cardiologist to recommend CABG surgery (given the moderate left main disease on angiography). In Fig. 19–10, rest relative uptake is normal (A). Stress relative uptake (B) and the coronary flow capacity map (C) show the following:

1. A severe stress defect in the distribution of LCX proximal to its the first obtuse marginal branch (blue on the flow capacity map)

2. A severe stress defect in the distribution of the first diagonal branch off the LAD (blue on the flow capacity map)

3. Mildly reduced but adequate coronary flow capacity of the proximal LAD and first septal perforator above low flow ischemic levels (yellow on the flow capacity map) tapering to more severely reduced flow at the apex (green on the flow capacity map) consistent with severe diffuse narrowing

4. Mildly reduced but adequate flow capacity above ischemic levels in the ramus intermedius distribution anterior to the lateral stress defect caused by the first obtuse marginal branch stenosis

5. Excellent high flow capacity in the RCA distribution

Objective computer analysis and review of the angiogram by KLG measured the left main as a 52% diameter stenosis with diffuse LAD narrowing with moderate stenosis of the first obtuse marginal branch of the LCX and first diagonal branch. The left main stenosis alone did not severely limit flow capacity. However, the cumulative effects of the left main in sequence with diffuse disease and downstream moderate anatomic stenosis severely reduced flow capacity heterogeneously. The PET results suggested that CABG surgery was appropriate and also helped elucidate appropriate bypass targets (LAD with its diffuse narrowing, the large first diagonal branch, and first obtuse marginal branch with branches on either side of the ramus). In the absence of symptoms, the patient was reluctant to have CABG surgery but after reviewing the PET with him and his cardiologist with the high risk of cumulatively severe CAD, CABG surgery was performed with the patient doing well at follow-up 9 years later.

Case 7: Left Main Stenosis with Patent RCA

A 49-year-old man had PET as part of a research protocol. Nine years previously, he had a prior ascending aorta graft for type A aortic dissection complicated by severe aortic regurgitation. The left main coronary artery had been reattached to the graft. He followed an intense exercise regimen consisting of 1 hour on the elliptical and 30 minutes on the bicycle five times weekly without symptoms or limitation of exercise, except over the prior week when he felt that “something was not quite right.” After rest and dipyridamole stress images were completed (Fig. 19–11), he developed hypotension, deep ST-segment depression, and cardiac arrest. He was supported by advanced cardiac life support to the catheterization laboratory, with subsequent extracorporeal membrane oxygenation. An angiogram showed a severe stenosis at the anastomosis of the left main to the aortic graft that was stented open. He recovered fully with normal brain function and normal LVEF.

Case 8: Left Main Stenosis with Occluded RCA

A 69-year-old man with poorly controlled risk factors, right bundle branch block, and past angiogram showing an occluded RCA had mild exertional angina for 3 years. Although his angina was variable, he thought that perhaps it had slightly worsened over the prior 3 weeks; however, he was reluctant to have an angiogram. Dipyridamole stress led to severe angina, 3-mm ST-segment depressions, and blood pressure falling from 126/78 to 90/56 mm Hg, requiring early imaging and termination of stress with intravenous aminophylline, metoprolol, and nitroglycerin resolving these stress effects. The relative images in Fig. 19–12 show mild perfusion heterogeneity or small mild separate stress defect with no severe large regional perfusion defect. However, the stress flow, CFR, and coronary flow capacity map show severely reduced stress perfusion to low ischemic flow levels. Angiogram confirmed a severe stenosis of the left main coronary artery and an occluded RCA; he underwent successful CABG surgery.

Case 9: Abnormal Relative Stress Image But Adequate Absolute Perfusion

A 60-year-old man had known angiographic occlusion of the LCX with extensive collaterals and variable angina over the prior 10 years, depending on adherence to exercise and healthy living. After a half year of unhealthy high-stress living, his exertional angina worsened, leading to a follow-up PET scan (Fig. 19–13). While there is stress perfusion abnormality on the relative perfusion images, the flow capacity map shows only mildly to moderately reduced flow capacity in a small lateral region with good flow capacity in the remaining myocardium. The collaterals supplying collaterals to the LCX distribution were sufficiently well developed to prevent myocardial steal. Collateral perfusion improves with regular exercise, as well as low-fat and low-carbohydrate foods by complex mechanisms, including lowering the postprandial lipid surge that inhibits endothelial function.^{58,59,60,61,62,63} Accordingly, an angiogram was avoided, and return to healthy habits resolved his angina.

Case 10: Progression of Diffuse CAD, Compromising Collateral Supply

A 69-year-old asymptomatic man with total occlusion of the RCA by angiogram and poor risk-factor control had a baseline PET as part of a research protocol (Fig. 19–14A). The baseline coronary capacity map showed a severe perfusion defect comprising 13% of the LV (blue) with the central 7% revealing myocardial steal (dark blue) associated with collateral perfusion. Although he was still asymptomatic 6 years later, protocol follow-up PET in Fig. 19–14B showed a larger, severe defect comprising 23% of the LV (blue) with an additional 9% border zones (green) and 13% central steal (dark blue). In the rest of the myocardium outside the severe stress defect, the best coronary flow capacity (red) had declined from 56% to 16% of the LV, with a corresponding increase in the areas of mildly reduced flow capacity (yellow) from 10% to 26% of the LV. Therefore, this enlarging stress defect in the RCA distribution was likely the result of progressive diffuse disease in the other coronary arteries supplying the collaterals to the totally occluded RCA. Accordingly, coronary angiography confirmed the PET findings but could not quantify the diffuse CAD. Attempted PCI of the occluded RCA was not successful. Because the patient was asymptomatic with good LV function, CABG surgery was not indicated after the failed PCI of the chronically occluded RCA.

Case 11: Severe Angina at High Coronary Flow—Abnormal Adenosine A1 Receptor?

A 66-year-old man had continued intermittent atypical chest pain before and after PCI, with two stents to the LAD. Rather than perform another angiogram, his cardiologist requested PET for assessing progression of his CAD (Fig. 19–15). He developed moderately severe rest chest pain during initial rest perfusion imaging; rest imaging was normal and symptoms resolved spontaneously. Consequently, repeat rest imaging was followed by regadenoson stress imaging. All rest and stress images were normal. Stress perfusion and CFR showed extraordinarily high coronary flow capacity. However, at these very high flows, at the end of image acquisition, the patient developed extreme chest pain (the worst he had ever had), requiring intravenous aminophylline and metoprolol as well as two sublingual nitroglycerin for relief. There were no ECG changes. LVEF on ECG-gated PET perfusion images increased from 75% to 78% with normal wall motion. The high flows argued against either coronary spasm or microvascular disease. Regadenoson is considered to be a specific adenosine A2 agonist mediating coronary arteriolar vasodilation with weak affinity for the adenosine A1 receptor mediating cardiac pain. In this case, regadenoson triggered severe cardiac pain without ischemia, suggesting aberrant A1 receptor activation by possibly binding regadenoson, or a high density of partially activated A1 receptors, thereby causing severe angina despite high coronary flow mediated by A2 receptor agonism.

Case 12: Abnormal Relative Stress Images That Miss Diffuse CAD

A 72-year-old asymptomatic man had routine follow-up PET scan with abnormal stress test 15 years previously leading to angiogram then demonstrating a chronic total occlusion of the LAD. He had no PCIs at that time, and he remained asymptomatic. Fifteen years later, PET relative images (Fig. 19–16), showed a mid-anterior stress defect distal to the first septal perforator and first diagonal branches. The coronary flow capacity map showed myocardial steal, indicating collateral flow with diffuse underlying disease not apparent on relative images. With stability of a known occlusion for 15 years, stability of diffuse disease, absence of symptoms, and preserved left ventricular function, the patient elected for continued medical treatment.

Case 13: Severe Microvascular Dysfunction Without Obstructive CAD

A 78-year-old woman presented with angina manifest as jaw and arm pain for 11 years at rest and exertion that had progressed to limit her walking to 100 feet. She has hypertension and hyperlipidemia. Five coronary angiograms over the years showed no obstructive CAD. She was referred for dipyridamole PET-CT scan with images shown in Fig. 19–17. Relative stress perfusion images were normal (A). Global stress flow of 1.66 cc/min/g and CFR of 1.99 are moderately impaired without angina or ECG changes during dipyridamole stress. CT for attenuation correction showed no coronary calcium, and ELVF was 70% on gated perfusion images. These findings indicated microvascular dysfunction without calcific coronary atherosclerosis.

Case 14: Saved from Unnecessary Coronary Angiogram

A 43-year-old asymptomatic woman requested a second opinion and PET scan after a cardiologist recommended coronary angiography because of a positive ECG stress test performed at routine physical examination followed by a positive SPECT myocardial perfusion imaging reported to show mild anterior ischemia. Her PET images (Fig. 19–18) were normal with CFR of 3.9 and a coronary flow capacity map showing uniformly high-flow capacity comparable to that in young healthy volunteers.

Clinical Patient Groups and the Color-Coded Ranges for the Coronary Flow Capacity Map

The case examples demonstrate the power of quantitative perfusion, CFR, and coronary flow capacity for virtually every manifestation of CAD, where the coronary flow capacity map correlates with high statistical certainty with large classes of healthy young subjects and patients with clinical and subclinical CAD (Fig. 19–19 and Table 19–1). This extensive validation of the coronary flow capacity map serves to personalize management of CAD or forms the basis for randomized trials of intervention versus medical management by identifying patients with sufficiently large, physiologically severe CAD to optimally benefit from revascularization.^31,37,40

Heterogeneous Myocardial Perfusion: Simplifying Nature’s Complexity

Resting and stress perfusion are variable in different regions of the heart because of differences in regional endothelial function, wall stress and thickness, regional workload, spatially variable systolic compression and relaxation, differing regional metabolic demands, and multiple control mechanisms as reviewed in the chapter on coronary blood flow (Chap. 34). This regional perfusion variability, or perfusion heterogeneity, is profoundly influenced by risk factors, early subclinical and clinically manifest coronary atherosclerosis, medications, emotions and emotional stress, physical activity or training, and food (even a single recent meal). Perfusion heterogeneity is commonly more prominent in resting perfusion, as in Fig. 19–20 (A with only lateral and inferior views) because vasodilatory stress substantially overrides autoregulation and many resting control mechanisms (B). Consequently, CFR (C) may show apparent severe regional perfusion defects (blue and green) caused by localized high rest perfusion or may show regional high CFR values (red) caused by low resting flow regions.

The coronary flow capacity map integrates the regional variations in color-coded ranges of perfusion for the entire range of people from healthy young volunteers to patients having angina and significant ECG changes during dipyridamole stress. The percentage of the LV with low stress perfusion in cc/min/g (> 5% or approximately 10% in B) or with low CFR (18% in C) as stand-alone metrics overestimates severity compared to the coronary flow capacity map, accounting for both stress flow and CFR together that shows only mildly reduced flow capacity diffusely (yellow and orange) caused by mild, nonobstructive, calcific CAD. Perfusion heterogeneity provides important clinical insights into endothelial dysfunction and heterogeneous distribution of coronary atherosclerosis rather than being a limitation of quantitative perfusion of the technology. Overprocessing or oversmoothing images to eliminate perfusion heterogeneity destroys essential data that reveal important insights into coronary pathophysiology.

Types of Stress for Quantitative Perfusion Imaging

Quantitative perfusion imaging requires acquisition of the first-pass blood activity of the radionuclide perfusion tracer. Treadmill exercise precludes imaging this first-pass arterial input because of patient motion and the vertical position of PET scanners with horizontal gantries for supine patient positioning. Although supine exercise is reported for quantitative PET imaging,^26,38,42 pharmacologic stress is the standard using either arteriolar vasodilators or dobutamine, as listed in Table 19–2, with their differing biological effects. Dipyridamole and adenosine are the oldest, best established for achieving near-maximum hyperemia and a CFR of 4.0 or higher in healthy volunteers required for the optimal diagnosis or assessment of the physiologic severity of CAD.

Residual blood caffeine inhibits the hyperemic effects of these vasodilators, thereby invalidating diagnostic accuracy, as shown in Fig. 19–21. The baseline coronary flow capacity map shows no clinically significant abnormality after withholding caffeine for 24 hours but with residual blood caffeine levels of 1.0 μg/mL. After 48 hours of abstinence from caffeine and no measurable blood caffeine level, repeat rest-dipyridamole PET caused a large severe perfusion defect involving 25% to 30% of the LV in an RCA or LCX distribution with a minimum CFR of 0.69 indicating myocardial steal with stress perfusion reduced to 69% of rest perfusion. Myocardial steal is associated with collaterals beyond an occluded artery corresponding in this case to a chronic asymptomatic occlusion of a dominant RCA on angiogram.

The adenosine A2 agonist regadenoson is marketed for vasodilator stress myocardial perfusion imaging. When used according to the protocol recommended by the manufacturer, the maximum stress flow and CFR with regadenoson is only 80% of stress perfusion with dipyridamole in patients undergoing sequential stress imaging with both agents.⁶⁴ Inadequate or submaximal vasodilator stress reduces the relative size and severity of stress perfusion defects compared to their size and severity with maximum hyperemia induced by dipyridamole or adenosine. The relative PET images for Case Y in Fig. 19–22 show a large dipyridamole-induced stress defect caused by a chronic asymptomatic LAD occlusion. The regadenoson stress defect in the same patient is substantially smaller and less severe, thereby failing to quantify size and severity.

Submaximal stress has the opposite misleading effect on absolute perfusion by eliciting lower stress flow in cc/min/g compared to maximum stress, thereby erroneously appearing to reflect diffuse disease. In Case Z of Fig. 19–22, relative stress images (A, red) are both normal. However, stress perfusion in cc/min/g (B, blue) is moderately reduced to approximately 1.1 cc/min/g (by the right stress flow color bar scale) also indicated by the coronary flow capacity map (C, yellow by the color bar scale of combined stress flow of 1.1 cc/min/g and CFR of approximately 2.0). With dipyridamole, stress perfusion is 2.0 cc/min/g and the flow capacity is high (C, red). In a systematic study on the time course of regadenoson stress perfusion by PET using Rb-82, maximum perfusion occurred 55 seconds after the 10-second regadenoson infusion, achieving 90% of dipyridamole stress perfusion.⁶⁴ Therefore, injecting radionuclide at 55 seconds after the 10-second regadenoson infusion produces greater hyperemia and was associated with better quantification of size and severity of stress-induced perfusion defects but not the same as dipyridamole. Finally, regadenoson has more adverse events than adenosine.⁶⁵

Technical Components of Quantifying Myocardial Perfusion

Measuring myocardial perfusion in cc/min/g using radionuclides requires three components: the time-integrated arterial blood concentration of activity, the myocardial tissue activity, and the equation or model for calculating perfusion from these measured components (Fig. 19–23). Radiolabeled microspheres used in experimental animals are 100% trapped in the myocardium so that the model calculation of perfusion is a simple ratio with resulting units of cc/min/g. Radionuclides used clinically are variably trapped in myocardium with decreasing myocardial trapping or “extraction” declining as perfusion increases.

The model for calculating flow from the measured arterial input function and myocardial uptake for the common clinical radionuclides employ a correction term for this flow-dependent extraction, E, in Fig. 19–23. The extraction fraction is expressed as a fraction of 1.0 for 100% extraction-like microspheres. At resting perfusion levels, the extraction for nitrogen (N)-13 ammonia is about 0.80 or 80%, falling to about 55% at maximum stress perfusion; for Rb-82, the resting extraction is about 65%, falling to 35% at maximum stress perfusion (Table 19–3).

The extraction correction for each radionuclide can be determined in two ways. The extraction-perfusion relation can be experimentally determined in animals and inserted into the equation with the remaining two unknowns, arterial input and myocardial uptake activities, measured by the scanner for each pixel or specified region with perfusion in cc/min/g displayed regionally. The other method uses multicompartmental curve fitting of the arterial input and myocardial activities over time to account for extraction as part of the flow model equations to calculate perfusion in cc/min/g. The latter approach requires acquiring serial 10- to 15-second images for the first usually 2 minutes, followed by serial images every 30 to 60 seconds for at least 5 minutes in order to construct time activity curves of arterial input and myocardial activity with time. Each of these common radionuclides has a specific acquisition protocol and flow model.

Radionuclides for Quantifying Myocardial Perfusion

PET radionuclides for myocardial perfusion imaging (see Table 19–3) have differing biological behavior, availability, and clinical advantages and disadvantages. The partially extracted radionuclides, cyclotron-produced N-13 ammonia and generator-produced Rb-82, are the most widely used with the largest literature as recently reviewed.³¹ The positron ranges and extraction or retention fraction vary with their definition and how they are measured. Each has a specific flow model incorporating the arterial input function, myocardial uptake, and flow-dependent extraction or retention, as described below. Both provide accurate measures of myocardial perfusion but with differing operational advantages and disadvantages.

N-13 ammonia has a smaller positron range, associated sharper images, and a longer half-life that is less demanding of scanner performance suitable for three-dimensional (3D) cardiac imaging; the first-pass bolus for arterial input has much lower activity than Rb-82. On the other hand, it requires an on-site or nearby cyclotron, close coordination of patient imaging with the cyclotron run, and slower patient throughput because of the longer half-life of N-13 (10 minutes) requiring decay time between rest and stress images.

Rb-82 has the most user-friendly generator source of the PET radionuclides, a short half-life permitting rapid serial rest-stress imaging or serial stress-stress images for research protocols, and a well-validated “simple” two-image acquisition protocol. However, the images are slightly less sharp than N-13 ammonia, although the small difference has no clinical consequences. The full dose of 40 to 50 mCi of Rb-82 for the first-pass arterial input requires high scanner performance optimal with two-dimensional (2D) imaging but not feasible with 3D scanners that, therefore, require half-dose Rb-82 to avoid saturating the scanner, as discussed in a later section. The radiation dose for both radionuclides are similar at 1 to 2 mSV per image, compared to three times that dose from the CT attenuation scan of a PET-CT scanner.

O-15 produced by an on-site cyclotron requires the arterial input function and myocardial uptake but is 100% extracted by myocardium, an important theoretical advantage over the partially extracted radionuclides. However, myocardial uptake imaging is challenging because of its high concentration in the blood pool that requires subtraction of the blood-pool counts from the original image to visualize the myocardium. This subtraction requires acquiring a second set of images after a single inhalation of 40 to 50 mCi of O-15 carbon monoxide. Therefore, the imaging and processing protocol for the O-15 flow model is substantially more complex than the partially extracted tracers such as N-13 ammonia and Rb-82.⁵⁰ O-15 has a short half-life of (2 minutes); thus, an on-site cyclotron is necessary. Because of its very short half-life and more complex processing, the images are somewhat noisy with greater variability, which is suboptimal for routine clinical applications and regional perfusion needed for clinical decisions. However, in experienced sites, it has provided important research data, particularly for validating in humans other radionuclides more suitable for clinical use, such as Rb-82.⁶⁶

F-18 flurpiridaz is a new perfusion tracer not yet approved by the US Food and Drug Administration.⁶⁷ F-18 has a longer half life (110 minutes) than N-13 or Rb-82. Its first-pass extraction is reportedly 90%, and it requires its own specific flow model. However, the frequent claim that any one of the radionuclides is better than another because of higher extraction, as for O-15, fails to acknowledge that all of these perfusion tracers work well for measuring perfusion; the extraction characteristics are inherently incorporated into the specific flow model for each radionuclide. The disadvantage of F-18 flurpiridaz is its long half-life, which complicates quantitative timely sequential rest-stress quantitative imaging as well as requiring a nearby cyclotron. As addressed below, the primary source of variability in cc/min/g among the different perfusion tracers resides in the arterial input acquisition, not the radionuclide or details of flow models.

Flow Models for Calculating Myocardial Perfusion in cc/min/g

There are two basic flow models for calculating myocardial perfusion in cc/min/g (Fig. 19–24). After intravenous injection of radionuclide, the arterial activity rises rapidly, peaks, and then falls as the systemic circulation dilutes its concentration after intravenous injection. The spreading out of the arterial input activity is caused by cardiac output or systemic circulation diluting the activity concentration and dispersing it over time by multilength pathways or transit times through the lungs. Significant myocardial uptake begins as the broadened arterial pulse of activity passes into the coronary arteries off the aortic root.

The primary data needed for perfusion in cc/min/g are the area under the time-integrated curve of the arterial input activity and the “instantaneous” myocardial uptake at completion of the arterial input function. For Rb-82, the myocardial uptake is flat after the first 2 minutes, so that a single high-quality 5-minute image is obtained, representing the “instantaneous” myocardial uptake. As indicated above and in Fig. 19–24, the multicompartmental model acquires serial short images to construct arterial input and myocardial time activity curves that account for flow-dependent extraction from the time activity curves best fit to the compartmental model equation for calculating perfusion. A variation of this approach uses the experimental extraction-flow relation entered into the equation to simplify the best fit of the time activity curves for solving the equation for perfusion in cc/min/g.

The “simple” or “retention” flow model for Rb-82 acquires a single, early phase, 2-minute first-pass arterial input image followed by a 5-minute myocardial uptake image (or a 10-minute myocardial image for N-13 ammonia), as in Fig. 19–24 and Fig. 19–25. Flow in cc/min/g is calculated using an experimentally determined extraction correction (the exponential term of Fig. 19–24).⁶⁸ All flow models use iterative solutions for calculating perfusion in cc/min/g.

An extensive literature addresses the several variations in radionuclides, flow models, and acquisition protocols for quantitative myocardial perfusion, supporting whichever one available is the “best” for that facility. Although “best” at any facility implies that its methodology is objectively selected and optimized, needing no change or critical internal comparison to alternatives, the method used at most sites is usually driven by convenience, prior experience, preexisting facilities, prior expertise in a given method, and particularly funding support. Established research PET facilities with on-site cyclotron generated radionuclides typically have physicists developing the more complex multicompartmental flow models and serial image acquisition protocols. Other PET sites that are primarily clinical may use the “simple” or simpler “retention” flow models for Rb-82 that are more widely available with simpler facility requirements.

Direct comparison of the various models experimentally,⁶⁸ theoretically,⁶⁹ or clinically^32,33,70 show that the “simple model” provides perfusion in cc/min/g comparable to multicompartmental models (Fig. 19–26) with less variability than multicompartmental models.^32,69 In experimental animals, the simple model using a single arterial input image (A) correlated tightly with the compartmental model using serial images to construct time activity curves and with perfusion by radiolabeled microspheres (B).⁶⁸ In patients, the simple model for Rb-82 (C) and 13-N ammonia (D) provided “higher sensitivity for detection and localization of abnormal flow and myocardial perfusion reserve…without the computational complexity and sensitivity to noise—of the multicompartmental model.”³²

The variability of stress perfusion of flow models parallels their complexity because the assumptions necessary to calculate flow introduce noise that their greater mathematical or theoretical precision fails to overcome. For example, the most complex distributive model developed to date had greater variability than standard versions for this reason and proved less clinically useful than simpler versions.⁷¹ However, most of the variability among different flow models does not result from the equations for calculating perfusion. Rather, the greatest variability in absolute perfusion is caused by variable arterial input imposed by short, noisy, serial images using back-projected regions of interest (ROIs),⁷² as discussed below.

As a result of differences in PET scanners, radionuclides, acquisition protocols, and flow models, rest and stress perfusion in cc/min/g may be different among different PET centers. However, the ranges of CFR (myocardial perfusion reserve) are more comparable among different PET facilities because the ratio of stress to rest flow for CFR cancels out the acquisition and modeling factors; this leads to variations in the absolute perfusion in the numerator and denominator of this ratio.³³ Complete integrated quantification of stenosis or diffuse CAD requires both CFR and stress flow in cc/min/g; therefore, reducing variability of absolute perfusion in cc/min/g is essential, as reviewed below.

The Arterial Input Function: The Achilles Heel of Quantitative Perfusion Imaging

An extensive literature addresses all the components of quantitative perfusion imaging—scanners, radionuclides, imaging protocols, myocardial resolution, partial volume correction, flow models, and software—except one, the arterial input function. A single, excellent, unique recent editorial⁷³ on the topic related to a detailed study of the issue,⁷² emphasizing the paucity of literature on this critical but neglected component of quantitative perfusion.

For multicompartmental flow models, the short, serial images of the first-pass arterial activity are so noisy that an ROI cannot be reliably located directly on the aorta, left atrium (LA), or LV cavity for the arterial input function. For the same reason, the arterial ROI cannot be placed to avoid spillover activity from the immediately adjacent superior vena cava, right atrium, and pulmonary artery or its major branches having the highest activity after intravenous injection. Multicompartmental models based on these noisy, short, serial images therefore require locating the arterial ROI on late myocardial images just superior to the estimated atrioventricular ring at the base of the LV and back projecting it onto the early-phase noisy serial images. The position of this back-projected ROI is assumed to be the LA from which the arterial input curve is obtained.

However, cardiac position, translation, and motion vary greatly during the cardiac and respiratory cycle at rest; movement of still several centimeters more occurs during the tachycardia and tachypnea that accompany the administration of vasodilator stress.^74,75,76,77 In a systematic study, an ROI located on late myocardial images back-projected onto first-pass images were outside the LA or aortic root for significant portions of the cardiac cycle in a substantial number of patients,⁷² thereby giving erroneous perfusion values. Moreover, this study demonstrated that each individual has a personally optimal ROI in central either LA or aorta that avoids spillover from high-activity adjacent structures containing venous blood. No single ROI site was optimal for all patients or even for rest and stress in the same patient.

An arbitrary fixed ROI in the LA or aortic root, particularly back-projected from late myocardial images, yielded erroneous arterial inputs (Fig. 19–27) and perfusion values in 50% or more of all patient studies (Fig. 19–28). Erroneous location of arterial ROI may cause profound life-threatening errors of quantitative perfusion imaging illustrated in Fig. 19–29. Incorrect ROI location that fails to capture the true arterial activity may cause the arterial input function to be too low, with resulting erroneously high perfusion measurements (A) compared to the true arterial input (B) proven by the angiogram.

The commonly unrecognized great strength of the single 2-minute arterial input image of the simple model resides in its high-quality arterial images that allow optimal selection of the most reliable arterial input and therefore perfusion measurements. Arterial ROI selection is so critical that the details of venous or pulmonary artery anatomy adjacent to arterial input anatomy for perfusion imaging need review because they are not widely understood. Fig. 19–30 views the supine patient through the left side where the scanner tomographic planes are rotated counterclockwise and viewed from the feet looking toward the head in traditional radiologic views. Tomographic planes in this schematic labeled A, B, and C correspond to the CT and PET image planes of Fig. 19–31.

In Fig. 19–31, the high-activity venous blood in the superior vena cava, right atrium, right ventricle, and pulmonary arteries is immediately adjacent to the LA and aortic root. In addition, the heart recoils downward and medially during systole and also moves inferiorly with inspiration, all motions that are greatly increased with the tachycardia and tachypnea during vasodilator stress over 2- to 3-centimeter ranges.^74,75,76,77 A fixed back-projected ROI from late myocardial images is therefore commonly not over the optimal central LA or aorta site but may be over high-activity venous structures or low-activity lung.⁷²

With semiautomated software, the high-quality 2-minute arterial images allow rapidly placing five to eight ROIs over pulmonary artery, aorta, right ventricle, and LV sites, as well as several LA sites. The high activity in the pulmonary artery and right ventricle bracket the upper range of activity for selecting ROIs with no spillover from high-activity venous blood structures. Of all possible ROI sites for measuring arterial activity, the LA or aorta site with the highest activity without spillover from adjacent venous blood activity is the best arterial input activity for that image personalized for that patient. The optimal ROI may be different at rest conditions compared to stress is the same patient. These conclusions are based on repeated serial quantitative PET perfusion imaging for test-retest precision, day-to-day variability, and multiple ROIs in the same subject in more than 200 volunteers or patients and more than 6000 patients having clinical, protocol, and follow-up quantitative PET at the Weatherhead PET Center for Preventing and Reversing Atherosclerosis of the University of Texas at Houston.^{16,31,37,39,40,48,55,57,64,72,74,75,76,77} In this PET center, the technicians are highly skilled with every ROI checked for any potential errors, complete with perfusion, CFR, and coronary flow capacity maps within 3 to 5 minutes after image reconstruction by KLG.

Dedicated mathematical models focusing on the details and mathematical precision of complex multicompartment models largely overlook variability of the arterial input function resulting from heart translation and motion during the cardiac cycle.^74,75,76,77 These arterial input errors may cause errors in perfusion measurements that are greater than what could be ascribed to the flow models.^74,75,76,77 The simple retention models appear to have the greatest ease of application for routine clinical use with the least variability.^{32,33,68,69,70,71,72,73,74,75,76,77} However, there is not, and will likely will not be, a standardized protocol given the wide range of radionuclides, protocols, and flow models already in place. Nevertheless, as discussed below, for all the different approaches there is a relatively simple standard of performance for accuracy of whichever tracer, protocol, or flow model used that makes quantitative perfusion a universal measure, a universal vital sign, such as blood pressure, regardless of the specific technology used to measure it.

The Universal Standard of Performance for Quantitative Myocardial Perfusion

The standards to which we adhere are based on well-established physiologic facts of maximal and minimal myocardial perfusion proven in experimental models and in humans long before cardiac PET or magnetic resonance imaging (MRI) even existed. In experimental models and healthy young humans, that CFR is 4.0 or higher was established in humans as early as 1973,⁷⁸ followed by confirmatory invasive measurements^{79,80,81,82,83} and reconfirmed by later PET literature recently summarized for 14,962 individuals previously reported.³¹ For healthy young volunteers without risk factors and no detectable blood caffeine, CFR by PET averaged 4.2 ± 0.8; absolute stress perfusion in cc/min/g, 2.9 ± 0.5 cc/min/g⁴⁵; and transmural myocardial scar perfusion, 0.2 cc/min/g.^31,40,57

Accordingly, for the many variations in PET protocols for quantifying myocardial perfusion to define physiologic severity, the simple standard performance test combining measurement accuracy and clinical coronary pathophysiology to ensure correct clinical decisions is the capacity to measure (1) rest perfusion of 0.2 cc/min/g in transmural scar in at least five patients to test low perfusion accuracy and (2) regional and global CFR of 4.0 or higher and stress perfusion of 2.9 cc/min/g or higher on two sequential rest-stress PET perfusion studies in the same subject with ± 15% variability for at least 15 young healthy volunteers with no risk factors and no measurable blood caffeine levels.^40,55

The PET Scanner

Most cardiac PET literature is based on using 2D scanners, initially with rotating rod attenuation correction and more recently PET-CT scanners with CT attenuation correction. At present, only 3D PET-CT scanners are manufactured largely for cancer applications. The CT attenuation correction of the PET-CT scanner may provide slightly better reconstructed resolution with sharper images than rotating rod attenuation correction. However, this small difference has little clinical impact. PET-CT also detects coronary calcification that is useful clinically for identifying clinical or subclinical coronary atherosclerosis, but also incurs two to three times the radiation dose as the rotating rod.

2D scanners have lead septa and in-plane reconstruction algorithms that reduce random coincidences and allow first-pass acquisition of full dose 50 mCi of Rb-82 with high-quality, high-count arterial and myocardial images without scanner saturation. 3D scanners lack septa and therefore have greater sensitivity acquiring more counts that, however, include high random coincidences. Therefore, the first-pass arterial activity of full dose Rb-82 (40-50 mCi) saturates the scanner or shuts down acquisition of 3D scanners and some older 2D scanners (Fig. 19–32), thereby precluding quantitative perfusion measurements.

Therefore, for 3D scanners, the dose of Rb-82 has to be reduced to 20 to 25 mCi infused slowly over 30 to 60 seconds. Higher doses lower noise equivalent counts rate by 50% (Fig. 19–33).^74,84 Although studies from a site experienced in 3D PET for cardiac PET with Rb-82 report adequate quality images, primarily global perfusion measurements are reported or used clinically as guides to the angiogram on which clinical decisions rely without demonstrating regional perfusion for clinical guidance shown in the examples above. For N-13 ammonia or F-18 radionuclides that are given in lower doses of 10 to 20 mCi, scanner saturation is not an issue with either 2D or 3D PET scanners.

Most PET scanners have sufficiently comparable reconstructed resolution as to constitute little differentiation among them for cardiac PET. However, the order of magnitude better resolution of PET over SPECT at heart depth in the chest is a powerful advantage for PET, as illustrated in Fig. 19–34. For SPECT imaging of an unmoving static Jaszczak phantom under ideal imaging conditions, any negative defect having no activity that is smaller than 3 centimeters in diameter fails to be detected. In addition on the SPECT images, attenuation also reduces activity recovery in the center of the 20-cm phantom by 50%, thereby creating attenuation artifacts. In contrast, PET imaging sees all the abnormal cold spots with no central attenuation less.

The final critical component of the scanner is software to display and shift attenuation or emission data in order to achieve correct coregistration (Fig. 19–35). Despite alignment by external body markers in this case, both rest and stress helical CT images do not coregister with the emission data, thereby causing severe abnormalities that are misregistration artifacts. However, the cine CT shifted to achieve correct coregistration provides accurate size and severity of a mild-to-moderate medium-sized stress defect in the distribution of a chronically occluded collateralized ramus intermedius coronary artery on angiogram.

With both rotating rod and CT attenuation data, misregistration is common in 20% to 40% of all cases, because of cardiac position changing within the chest attenuating structures differentially at rest and stress even with the patient body position unchanged compared to external markers.^74,75,76,77 This misregistration may cause gross artifactual abnormalities or reduce accuracy of quantitative perfusion measurements.^74,75,76,77

Over the breathing and heart cycles, cine CT for attenuation correction averages the Hounsfield density units of chest attenuating structures, diaphragm, and heart motion over several breathing and heart cycles.^58,59,60,61 The resulting cumulative or averaged attenuation data are more comparable to the 5-minute emission image acquisition. In contrast, a short helical CT attenuation scan at some point in the cardiac or respiratory cycle or breath-hold may not coregister with the emission data acquired over 5 minutes with multiple cardiac and breathing cycles. In Weatherhead PET Center at the University of Texas at Houston, every case is checked for misregistration and corrected in order to ensure optimal perfusion quantification even in the absence of relative perfusion defects.

Standard CT attenuation correction incurs a higher radiation dose than rotating rod attenuation correction. Accordingly, modifying the CT dose for lower resolution suitable for attenuation correction and using a single poststress CT scan reduces total dose for a rest-stress study to approximately 7 to 8 mSv.⁷⁷

Complexity Made Simple: Quantitative Myocardial Perfusion for Personalized Coronary Care

Coronary blood flow is a determinant of survival and quality of life. As a basic vital sign, its accurate measurement should guide cardiovascular management (Fig. 19–36³¹) based on physiologic knowledge and careful technology serving to eliminate inappropriate procedures and maximize patient outcomes. Of the single views of 12 different cases in Fig. 19–36, the coronary flow capacity maps definitively identified and quantified patients with excellent flow capacity, those with risk factors only, those with mild to moderate focal and/or diffuse CAD optimally managed medically, and those requiring angiography or potentially appropriate revascularization.

Chronic Low Coronary Blood Flow Syndromes and Myocardial Viability

Just as acutely reduced coronary blood flow causes acute coronary syndromes, chronically reduced coronary blood flow causes “chronic coronary syndromes”—hibernating myocardium, stunned myocardium, or ischemic cardiomyopathy. For these syndromes, the physiologic severity of coronary artery stenosis by perfusion imaging is not the question addressed in the prior section on physiologic severity of coronary stenosis. The distinction characterizing these chronic coronary syndromes is usually chronic total coronary occlusion with collateral perfusion or chronic subtotal occlusion with low forward coronary flow.

Myocardial perfusion for these conditions remains essential for interpreting the metabolic images. Equally important for clinical decisions, quantitative perfusion also provides valuable information on the flow capacity of surrounding donor coronary arteries supplying the collaterals at risk from stenosis or diffuse disease of the supply arteries feeding the collaterals. Collateral steal during vasodilator stress is integral to quantifying the interaction of supply arteries, the region supplied by collaterals, and the “viability” of this region.

Clinical Myocardial Metabolism

With adequate oxygenated coronary blood flow, the myocardium preferentially metabolizes fatty acids aerobically as its primary source of energy for contraction. Following a high-sugar or a high-carbohydrate meal, the myocardium adapts by switching a substantial portion of its energy requirements for contraction to aerobic glucose metabolism. Ischemia resulting from low blood flow or hypoxia interrupts aerobic metabolism of either fatty acids or glucose. The myocardium then switches to anaerobic metabolism of glucose with production of lactic acid because complete metabolism to carbon dioxide requires an adequate supply of oxygenated coronary blood flow.

Under conditions of reduced coronary flow or hypoxia, the myocardial shift to anaerobic metabolism of glucose can provide only limited energy for contraction as a result of the accumulation of lactic acid, impeding further anaerobic metabolism of glucose and impeding left ventricular contraction. Although this low-level fuel source from anaerobic metabolism of glucose at low coronary perfusion may inhibit regional contraction, it provides sufficient metabolism to maintain noncontracting myocardial cell integrity for prolonged periods of hours to even years, at least in some patients. The dysfunctional myocardium recovers aerobic metabolism and contractile function on restoration of adequate oxygenated blood flow, hence the term hibernating myocardium.

Sustained metabolic activity for production of high-energy phosphate is fundamental for survival of functionally compromised myocardium. Maintenance of cell membrane integrity and transmembrane ion concentration gradients depends on sustained ATP production and availability. As demonstrated in both human and animal investigations, chronic segmental myocardial dysfunction is associated with profound alterations of fuel substrate metabolism.^85,86,87 Selection of fuel substrates shifts from free fatty acid to the more energy-efficient glucose with an increase in regional anaerobic glycolysis. This transcriptionally regulated shift in regional substrate usage and metabolism has been referred to as “recapitulation of the fetal gene program.”⁸⁸ Regionally increased glucose uptake is therefore considered a hallmark of reversibly dysfunctional myocardium that is readily identifiable with ¹⁸F-2-fluoro-2-deoxy-D-glucose (¹⁸F-FDG) PET imaging. Such imaging is clinically useful for risk assessment and for predicting postrevascularization regional and global left ventricular function or long-term survival.

The noninvasive study of the myocardium substrate’s metabolism with PET extends beyond coronary perfusion and delivery of products necessary for metabolism. Energy-generating substrate oxidation relates to external cardiac work, myocardial efficiency, myocardial changes in energy needs, and substrate availability by shifting its metabolism among substrates in normal healthy myocardium. A wide spectrum of radionuclide probes is available for imaging these metabolic shifts, ranging from radiolabeled myocardial fatty acid, glucose, lactate, and amino acid metabolism to neuronal control and activity.

The radioligand ¹⁸F-FDG is the most important and widely used for clinical applications. As a glucose analog, it tracks the initial transport of glucose from blood into cells and its phosphorylation to glucose-6-phosphate as the initial metabolic step of transformation of exogenously derived glucose. Because phosphorylated ¹⁸F-FDG cannot be metabolized further, it is metabolically trapped in the cell and accumulates in tissue in proportion to rates of exogenous glucose utilization. However, the radiotracer ¹⁸F-FDG is not cell specific for myocardium. It also traces glucose utilization in different organs, such as skeletal muscle and brain, as well as in tumors and inflammatory processes. Therefore, ¹⁸F-FDG is clinically useful for imaging regional myocardial metabolism and for identifying inflammatory disease of the cardiovascular system.

Definitions of Myocardial Viability, Hibernating, and Stunned Myocardium

The term viable is rather broad because, by definition, it also applies to normally contracting myocardium with adequate perfusion as well as ischemic myocardium with normal resting contraction without scar. The term potentially reversible contractile dysfunction is a more accurate term but is seldom used in clinical cardiology. By common use, myocardial viability imaging implies poorly contracting myocardium caused by chronic or transient low myocardial perfusion that recovers contractile function with restoration of normal perfusion. Although improved segmental contractile function following revascularization serves as a measure of myocardial viability, more definitive clinical outcome measures include postrevascularization improvements in global left ventricular function, in heart failure symptoms, and in long-term survival.

Mechanisms of reversible contractile dysfunction include myocardial hibernation and myocardial stunning. Myocardial hibernation is a response to chronically reduced resting myocardial blood flow, where the myocardium downregulates its energy expenditures for contractile work so that the diminished energy demand matches the diminished energy supply. Associated with this downregulation, the myocardium switches substrate selection from fatty acid to glucose as a more oxygen-efficient substrate to accommodate the reduced supply of coronary perfusion that is available.

Alternatively in animal experiments, hibernation may result from severe reductions in CFR without occlusion associated with increased glucose extraction, leading to a secondary decrease in resting blood flow due to the efficiency of aerobic glucose metabolism.^89,90 Findings in patients with ischemic cardiomyopathy support the latter mechanism where the extraction of ¹⁸F-FDG and, thus, of glucose, increases with declining CFR. Rates of glucose extraction were found to be highest at flow reserves of about 1.2.⁹¹ In hibernating myocardium, importantly, energy-producing cellular processes are maintained, although at a lower level, with high-energy phosphates expended mostly for the maintenance of transmembrane ion concentration gradients and for facilitation of basic cell maintenance rather than active contraction.⁹²

Myocardial stunning is the myocardial response to a transient ischemic episode caused by a severe, but transient, decline in coronary flow or an increase in cardiac work without adequate flow increase (demand-induced ischemia). Following the ischemic episode, blood flow promptly recovers, but contractile dysfunction of the postischemic myocardium initially persists (ie, “stunning”) and recovers only gradually over subsequent hours, weeks, or months.^93,94 In this scenario, repeated ischemic episodes interrupt the recovery of contractile function, leading to chronic impairment of contractile function despite normal resting coronary blood flow. A distinct feature of “repetitive” or “chronic stunning” is, therefore, the relatively well-preserved coronary flow at rest, which differs from the diminished resting coronary flow in “hibernating myocardium.” Apart from hibernation, regional flow reductions may also reflect scar tissue and infarcted myocardium with irreversible loss or impairment of contractile function. As summarized in Table 19–4, the diagnostic challenge is, therefore, to identify the potential reversibility of functionally impaired myocardium with diminished blood flow.

In patients, hibernation likely coexists with stunning in dysfunctional but viable myocardium. The delicate balance between supply and demand, although at a lower level, combined with the severe impairment in flow reserve renders “hibernating” myocardium susceptible to temporary increases in demand as, for example, related to physical and mental stress during daily life so that ischemia and stunning are superimposed on hibernation.^91,93,95 This possibly accounts for the “incomplete adaptation to ischemia” and for progressive loss of viable tissue.⁹⁶

Imaging Myocardial Viability

Imaging Perfusion and ¹⁸F-FDG Uptake

Regional increases in myocardial glucose utilization are most reliably identified by comparing the myocardial ¹⁸F-FDG uptake with myocardial perfusion. Resting myocardial perfusion is typically acquired first followed by assessment of ¹⁸F-FDG uptake. ¹⁸F-FDG is administered intravenously with image acquisition beginning 1 hour later. The F-18 glucose analog, ¹⁸F-FDG, is taken up by viable myocardium in the LV regions of low resting perfusion, thereby imaged as a perfusion-FDG mismatch.⁹⁷ This perfusion-FDG “abnormal mismatch” pattern of high ¹⁸F-FDG uptake in regions of low resting relative perfusion is the hallmark of hibernating myocardium.

Stress myocardial perfusion imaging after the rest perfusion image before ¹⁸F-FDG imaging may add clinically useful diagnostic information. Although not essential for the assessment of viability, rest-stress perfusion imaging can delineate the extent and severity of epicardial coronary artery disease and provide explanations for increases in myocardial ¹⁸F-FDG uptake in normally perfused but dysfunctional regions of myocardium. A stress-induced perfusion defect reflects a regional impairment in myocardial flow reserve and identifies the dysfunctional myocardium as chronically stunned. Furthermore, stress perfusion images aid in identifying possible reasons for impairments in left ventricular function. Different from ischemic cardiomyopathies with regional stress-induced perfusion defects, myocardial perfusion in nonischemic idiopathic dilated cardiomyopathy is characteristically homogeneous both at rest and during vasodilator stress.

Patterns of Perfusion: ¹⁸F-FDG Images

Figure 19–37 illustrates the four basic clinical perfusion-FDG image patterns in topographic simple 3D format as in the prior section of this chapter:

• Normal perfusion and normal FDG or “normal match” of healthy subjects or patients with low LVEF caused by nonischemic cardiomyopathy (A)

• Abnormal perfusion and abnormal FDG or “abnormal match” of myocardial scar (B)

• Abnormal perfusion and normal ¹⁸F-FDG or “abnormal mismatch” of hibernating myocardium in the distal half of the LV (C)

• Normal perfusion and low FDG uptake or “reverse mismatch” of adequately perfused myocardium metabolizing fatty acids rather than glucose (D)

In C, the base of the LV also shows a “reverse mismatch” with normal perfusion and preferential fatty acid metabolism. There is no ¹⁸F-FDG uptake proximal to the chronic LAD occlusion with distal hibernating myocardium demonstrating low perfusion and high ¹⁸F-FDG uptake.

Despite carbohydrate loading, some patients may have normally perfused LV regions that preferentially metabolize fatty acids with no ¹⁸F-FDG uptake thereby causing a “normal reverse mismatch” pattern (see Fig. 19–37D). Finally, nonischemic cardiomyopathy with poor contractile function has normal perfusion and ¹⁸F-FDG images with a “normal match” ¹⁸F-FDG–perfusion pattern as in a normal heart with a reduced LVEF. Commonly, patients may have combinations of all these patterns in different regions of the LV, particularly diabetic patients with myocardial transmural scar, substantial border zones of nontransmural scar, hibernating myocardium, and/or ischemia with a component of diabetic cardiomyopathy as well.

Tomographic Display and Analysis of Perfusion and Metabolism Images

Transaxially acquired perfusion and metabolism images are reoriented into short-axis and vertical and horizontal long-axis slices, using the same reorientation parameters for both (Figs. 19–38, 19–39, 19–40, 19–41). Visual assessment of myocardial radiotracer concentrations employs the standard 17-segment model and grades segmental radiotracer activity concentrations separately for the perfusion and ¹⁸F-FDG images on a four-point scale, where 1 is normal; 2, mildly reduced; 3, moderately reduced; and 4, severely reduced or absent radiotracer activity.⁹⁸ Differences in segmental radiotracer uptake between perfusion and metabolism images of at least one grade are considered “segmental metabolism perfusion-mismatches.” The sum of segments with mismatch scores represents the extent of potentially reversible dysfunctional myocardium. The sum of all segmental mismatch scores serves as a measure of the combined extent and severity of the “mismatch myocardial region” or the amount of viable myocardium.

Semiquantitative information on regionally increased ¹⁸F-FDG uptake relative to blood flow is available through polar or topographic map analysis of the PET ¹⁸F-FDG and perfusion images as in Fig. 19–42. The analysis approach defines myocardial regions with the highest relative flow-tracer uptake as “normal” as the reference for normalizing the ¹⁸F-FDG polar maps. “Mismatches” between metabolism and perfusion are identified by subtracting the perfusion from the normalized metabolism polar maps (regional differences) (see Fig. 19–42). Regional count differences are compared to a database of normal values and predetermined thresholds defined as “mismatches” of ¹⁸F-FDG uptake relative to perfusion. Accordingly, the quantitative image analysis approach generates a map of the geographic distribution and the total amount of viable (perfusion defect metabolism mismatch), nonviable (perfusion defect metabolism match), and normal myocardium (no perfusion defect normal metabolism match). Based on postrevascularization outcomes in regional wall motion, differences of more than two SDs from the normal indicate potential reversibility of wall motion impairment.⁹⁹ Importantly, the approach identifies those myocardial regions with increased extraction of ¹⁸F-FDG and thus of glucose relative to blood flow as an indication of persisting energy-producing metabolic processes that are essential for cell survival and identification of myocardial viability likely to benefit from improved coronary blood flow.

Evaluation of Glucose Metabolism, Only Without Perfusion Imaging

Some laboratories employ ¹⁸F-FDG imaging only for the assessment of reversible contractile dysfunction. Myocardial segments with normal or only mildly reduced (ie, < 50%) ¹⁸F-FDG uptake are considered reversibly dysfunctional, whereas segments with more severe reductions in ¹⁸F-FDG uptake (usually > 50%) are considered to be irreversibly dysfunctional.^{100,101,102,103} Acquisition of gated ¹⁸F-FDG images affords assessments of regional myocardial systolic thickening as an additional parameter of viability.¹⁰³

Although relatively high predictive accuracies for identifying reversibly and irreversibly dysfunctional myocardial segments have been reported,^101,104 imaging ¹⁸F-FDG only may be diagnostically limited. Dysfunctional myocardial regions with only mild matched perfusion and metabolism defects that do not recover function following revascularization are identified as viable by ¹⁸F-FDG imaging only.¹⁰⁵ Conversely, in regions with more severe reductions in ¹⁸F-FDG uptake, defined by the ¹⁸F-FDG–only approach as “nonviable,” the ¹⁸F-FDG uptake may substantially exceed regional flow and, thus, reflect the presence of viable myocardium. Moreover, the variability of ¹⁸F-FDG uptake in normal myocardium (ie, low or no uptake because of fasting, fatty food, or diabetes), frequently limits accurate identification of “viable myocardium” without characterization of perfusion. Importantly, it is the regionally increased extraction fraction of glucose as the hallmark of “myocardial viability,” which can be determined only by relating ¹⁸F-FDG uptake to myocardial perfusion.⁹¹

Absolute rates of glucose utilization have little value for predicting postrevascularization outcomes¹⁰⁰ because of substantial heterogeneity of functional cellular compromise and injury in hypoperfused myocardial regions, lack of adequate control normal regions or patients, assumptions in the metabolic models, and lack of standardized protocols for quantifying myocardial glucose consumption.

Pre-PET Patient Preparation and Standardization of Study Conditions

In view of these complex adaptive metabolic pathways, the preceding preparation in the 24 hours before and at the time of the ¹⁸F-FDG PET scan is essential for clinically reliable results. This pre-PET preparation is largely due to the myocardium’s ability to shift its energy needs among several fuel substrates including free fatty acid, glucose, and lactate. Selection of fuel substrates depends on their concentrations in blood, adequate oxygenated coronary blood flow, prior carbohydrate or fatty food, and hormonal influences. Complicating factors include diabetes mellitus, insulin resistance, and elevated blood catecholamine concentrations, frequently present in heart failure patients. Standardization of study conditions for perfusion metabolism imaging seeks to maximize myocardial glucose utilization and, thus, ¹⁸F-FDG uptake for diagnostic assessment of viability. High carbohydrate meals in the 24 hours before the ¹⁸F-FDG PET are necessary to precondition myocardium toward glucose metabolism and inhibit fatty acid metabolism. For resting perfusion-FDG imaging without stress, high-carbohydrate food can be eaten up to 2 hours before the scan. For combined vasodilator stress imaging with ¹⁸F-FDG after the stress perfusion image, patients should fast for four hours before the study. At the PET scan, several standardization protocols are available:

1. Oral glucose loading. After an approximate 5-hour fasting period, oral glucose (50-100 g) is administered orally about 1 hour prior to the ¹⁸F-FDG injection. Besides suppressing circulating free fatty acid levels, glucose loading raises plasma glucose concentrations and stimulates insulin secretion.

2. Administration of the antilipolytic agent acipimox. Acipimox diminishes lipolysis and, therefore also circulating free fatty acid levels. It also increases glucose levels caused by prior carbohydrate ingestion.

3. Euglycemic-hyperinsulinemic clamping.⁹⁰ This entails continuous infusion of insulin with coinfusion of glucose at rates to maintain the plasma glucose concentrations within the range of normal. Among the standardization approaches, glucose clamping most consistently produces high-quality diagnostic ¹⁸F-FDG images. However, it is so labor intensive that its use has remained confined to investigational settings.

4. Hybrid approach. After high-carbohydrate foods during the preceding 24 hours, a modified relatively simple insulin semiclamp can be easily applied clinically according to the protocol in Fig. 19–43 with excellent results in 98% of cases (protocol B¹⁰⁶). Just prior to the PET ¹⁸F-FDG study, measure fingerstick blood sugar to determine the need for intravenous glucose and insulin to enhance myocardial uptake of ¹⁸F-FDG for optimally differentiating viable myocardium from scar. The loading dose of intravenous glucose and addition of regular insulin intravenously depends on the baseline blood glucose as in Fig. 19–43. The aim is a controlled intravenous glucose load followed by a decreasing blood glucose indicating adequate response to native or exogenous injected insulin, thereby ensuring myocardial uptake of FDG.

Complex Cases Management Driven by PET Perfusion and Viability Imaging

The following cases show progressively complex cases in whom PET perfusion–¹⁸F-FDG imaging guided management, particularly revascularization procedures versus medical management.

Case 15: Primarily Hibernating Myocardium with Minimal Nontransmural Scar

A 75-year-old diabetic woman with daily moderate to severe resting and exertional angina was referred for rest/viability PET to determine treatment of an occluded LCX, occluded RCA artery, and severe LAD-diagonal disease. Resting perfusion images (Fig. 19–44) showed large, severe, contiguous, anterior, apical, lateral, and inferior perfusion defects involving 60% of the LV in the distribution of the mid LAD, mid LCX, and RCA coronary arteries. ¹⁸F-FDG metabolic images showed myocardial uptake of ¹⁸F-FDG, indicating viability of essentially the entire heart with only a small distal inferior nontransmural scar comprising 3% of the LV taking up less ¹⁸F-FDG. ECG-gated PET perfusion images showed LVEF reduced to 36%. The PET viability scan suggested that surgical revascularization was appropriate because of the large area of underperfused, viable, hibernating myocardium comprising 56% of the LV.

Case 16: Transmural Scar, Hibernating Myocardium, and Low LVEF in Multivessel Distribution

The 65-year-old man with images in Fig. 19–45 was referred for PET because of recent abnormal ECG and 6-month history of exertional, epigastric discomfort, chest pain, and throat tightness, all provoked by dyspnea. He had two prior PCIs with drug-eluting stents to the LAD in 2005 and the third obtuse marginal in 2007; the most recent pharmacologic SPECT nuclear stress test demonstrated normal perfusion images.

The relative rest and stress PET images showed a large, severe, lateral, apical, mid-anterior, and distal inferior, resting scar or hibernating myocardium involving 60% to 70% of the LV in the distribution of the LCX, mid-posterior descending artery, and mid-LAD. After dipyridamole, the defect was minimally worse, indicating fixed scar or viable, low-flow, hibernating myocardium.

The viability ¹⁸F-FDG images showed substantial mismatch with active FDG uptake in anterior, septal, and apical regions, indicating hibernating, viable, hypoperfused myocardium in mid to distal LAD and RCA distributions comprising approximately 30% of the LV. The lateral LV had a predominantly transmural scar comprising 25% of the LV with an additional 10% of the anterior, anterolateral wall, apex, and basal inferior wall being viable and hibernating in the distribution of a large first obtuse marginal branch. Therefore, cumulatively, 40% of the LV was metabolically viable, underperfused, hibernating, and akinetic, but it was potentially salvageable by revascularization of the LAD and obtuse marginal coronary arteries. Approximately 25% of the LV was fixed scar and 35% was normally perfused.

Absolute myocardial perfusion combining CFR and maximal cc/min/g showed severely reduced perfusion in the large defect and normal perfusion in the basal regions. In the transmural scar, resting and stress perfusion were 0.2 cc/min/g and minimum CFR was 0.8, indicating myocardial steal with border zones of mixed hibernating myocardium, scar, and myocardial steal associated with collaterals in the OM distribution. Gated PET perfusion images showed abnormal left ventricular contraction with severe, anterior, apical and lateral hypokinesis and an LVEF of 40%. Based on the PET scan, coronary arteriography with revascularization procedures are indicated because of the size and severity of the rest-stress-FDG PET findings. After CABG surgery, the LVEF increased to 53% with residual apical hypokinesis and resolution of symptoms.

Case 17: Complex Mix of Transmural Scar, Border Zone Nontransmural Scar, and Hibernating and Ischemic Myocardium in Multivessel Distribution

The 82-year-old man whose images are shown in Fig. 19–46 was referred for follow-up PET after a prior abnormal rest stress PET study 22 years previously demonstrated an occluded collateralized LAD. He was asymptomatic, and no intervention was performed at that time. He remained asymptomatic until 6 months prior to the current follow-up PET, when he developed leg weakness, fatigue, and dyspnea during walking without chest pain.

Relative PET images showed a moderate-sized, severe, anterior, resting perfusion defect involving 15% of the LV in the distribution of a distal diagonal branch. In addition, there was a large, severe, lateral resting perfusion defect involving 20% of the LV in the distribution of the mid-LCX distal to the first obtuse marginal branch. After adenosine, the anterior defect was slightly larger in the proximal border zones, indicating predominantly nonviable myocardium with a small proximal border zone of viable myocardium with reduced CFR. The lateral resting defect was also larger, indicating viable ischemic border zones with reduced coronary flow capacity. ¹⁸F-FDG images showed separate small transmural scars each comprising 10% of LV (anterior) and 8% of LV (basal lateral), cumulatively comprising 18% of the LV as scar. Approximately 27% of the LV was viable and ischemic or viable and hibernating, with the remaining 55% of myocardium having severe diffusely reduced stress flow and CFR but above ischemic levels.

In the transmural scar, minimum rest and stress perfusion was 0.26 cc/min/g and minimum CFR was 0.65, indicating myocardial steal that may be seen in the border zones of scar. Gated PET perfusion images showed abnormal LV contraction with anterior and lateral akinesis with the LVEF of 41% at rest falling to 24% during adenosine stress. These results indicated severe dysfunction resulting from severe regional abnormalities on the relative images and severe diffuse CAD on the perfusion images (not shown in Fig. 19-46), more apparent than on the relative images. Based on the PET perfusion and ¹⁸F-FDG viability images, coronary angiography and revascularization were recommended (1) to salvage approximately 30% of the myocardium that was viable, ischemic, and underperfused and (2) to improve CFR in the remaining approximately 55% of LV with flow-limiting stenosis or diffuse disease.

Case 18: Scar Replaced by Endogenous Stem Cells with Recovery of LV Function

A 69-year-old man whose images are shown in Fig. 19–47 had an anterior MI in 1996 at age 49 years while hunting in the Alaska wilderness. An angiogram 1 week later showed an occluded LAD proximal to the first septal perforator and first diagonal with an LVEF of 25% to 30% without significant obstructive disease of the LCX or RCA. Follow-up PET perfusion and ¹⁸F-FDG viability imaging showed a large anterior, septal, and apical myocardial scar with insignificant FDG uptake. Treated for heart failure and with vigorous risk-factor management, he remained stable with an LVEF increasing to 42% in 2002. Follow-up PET imaging in 2006 showed normal ¹⁸F-FDG uptake throughout the LV, a normal resting LVEF of 61%, and a large severe stress-induced perfusion defect during dipyridamole perfusion imaging. PET imaging remained unchanged up to 2013. He received PCI of the LAD, and he underwent implantable cardioverter-defibrillator (ICD) placement because similar patients with recovered left ventricular function had experienced sudden death. Several years later, while lifting 300-lb farm equipment, the man had wide complex tachycardia that electrophysiology determined was probably aberrantly conducted atrial fibrillation; he had an internal shock with no sequelae and no recurrence.

This spontaneous left ventricular recovery prior to any procedures and without exogenous stem cell therapy was so striking that heparinized blood was analyzed for endogenous stem cells. The results showed massive numbers of circulating CD133–/CD34+ stem cells comprising 5% of mononuclear cells, compared to 0.03% in a normal control. Repeat blood analysis for circulating stem cells 6 months later showed the same finding. The electrophoretic migration of these cells showed them to be “sticky,” indicating a subtype of CD 133–/CD34+ cells not previously identified in the literature. This unknown endogenous CD34+ cell type was associated with striking left ventricular recovery of myocardial scar replaced by normally functioning myocardium over 10 years.

Clinical Outcomes Related to Myocardial Viability

Segmental perfusion-metabolism patterns accurately predict improved postrevascularization segmental contractile function. Based on a pooled analysis of 24 clinical investigations in 756 patients, perfusion-metabolism imaging identified myocardial viability with a weighted average sensitivity of 92% and specificity of 63%.¹⁰⁷ Corresponding positive and negative predictive values are 74% and 87%, respectively. Although postrevascularization segmental wall motion provides a measure (or gold standard) of the diagnostic accuracy of perfusion-metabolism imaging, changes in the global left ventricular function are more powerful clinical measures of benefit. Pooled data from three investigations indicate average sensitivities and specificities of 83% and 64%, respectively, for at least a 5% postrevascularization increase in LVEF with positive and negative predictive values of 68% and 80%, respectively.¹⁰⁷

The degree to which left ventricular function recovers after revascularization depends on the amount of “viable myocardium.” To achieve at least a 5% increase in the LVEF requires the presence of “viability” in about 15% to 25% of the LV.^108,109,110 These thresholds are based on ¹⁸F-FDG and SPECT Tc-99m perfusion data. These thresholds may be lower when the extent of viable myocardium is determined from the higher spatial resolution PET perfusion-metabolism images. Postrevascularization improvements in left ventricular function correlate linearly with the extent of viable myocardium (Fig. 19–48).^111,112 A larger perfusion-metabolism mismatch is associated with a correspondingly greater postrevascularization improvement in left ventricular function.

Available evidence suggests that the risk of cardiac death relates to the amount of viable myocardium.^{108,109,110,111,112} This finding has been corroborated by two key observations from a prospective study of 260 patients with ischemic heart disease and severely diminished left ventricular function over a mean follow-up period of 2.1 years after a PET perfusion-metabolism study.¹¹³ First, the incidence of cardiac death in patients with clinically significant perfusion-metabolism mismatch of greater than 20% of the LV was significantly lower in those patients who received revascularization than in those treated with medical therapy alone (28% vs 15%; P < .05). Second, the risk of death rose in proportion to the amount of viable myocardium in patients with a mismatch greater than 20%. Importantly, small amounts of viable myocardium of less than 7% to 10% of the LV were not associated with an increased risk of death, implying that small amounts of viable myocardium insufficient for a postrevascularization improvement in global left ventricular function do not require revascularization for risk reduction (Fig. 19–49).¹¹⁴

Another study of 648 consecutive patients with reduced left ventricular function demonstrated a direct relationship between extent of viable myocardium and the benefits of revascularization.¹¹⁵ Revascularization conferred a statistically significant benefit for all-cause death in patients with PET a perfusion-metabolism mismatch greater than 10% of the left ventricular myocardium. The survival benefit depended on the extent of viable myocardium and was greatest for mismatch greater than 15%.

The only randomized trial to date, the Positron emission tomography And Recovery following Revascularization (PARR-2) trial,¹¹⁶ showed no overall mortality advantage of ¹⁸F-FDG–guided revascularization compared to standard of care, including aggressive medical therapy and revascularization.¹¹⁶ However, this negative result may have been the result of 30% nonadherent crossovers, wherein revascularization decisions were made contrary to what would have been recommended by ¹⁸F-FDG PET results (Fig. 19–50). In fact, in a per-protocol treatment analysis, the patients who underwent ¹⁸F-FDG PET–guided revascularization had a significantly better composite outcome comprised of cardiac death, MI, heart transplant, heart failure, or cardiac-driven hospitalization within 1 year) than did patients with non-FDG PET–guided revascularization (see Figure 19–49).¹¹⁴ For the 182 patients with LVEF less than 35% and mismatch of 10% or greater, those who received revascularization had a composite outcome that was 40% or lower than the composite outcome for patients without revascularization (see Fig. 19–49).⁹⁸ Nonetheless, for resting perfusion defects comprising small regions of the LV, ¹⁸F-FDG imaging–guided decisions are not likely to reduce mortality.

Global LV Function

Most studies have reported statistically significant postrevascularization increases in LVEF in patients with viability as compared to no or only minimal improvements in LVEF in patients without viable myocardium.¹⁰⁷ Only a few investigators did not observe such significant improvements in left ventricular function following revascularization.^117,118 Yet these studies still report significant improvements in the response of the LV to exercise and improvements in overall exercise capacity following surgical revascularization. Distinct from the more conventional perfusion-metabolism imaging performed at rest only, these studies identified myocardial viability and its extent from stress perfusion images (when the rest perfusion images were normal) and combined these results with ¹⁸F-FDG images at rest to report stress-induced myocardial ischemia. Successful revascularization rectifies the cause of stress-induced ischemic myocardial dysfunction or “stunning” and thus restores the normal functional response to stress.

Revascularization of viable myocardium may lead to reversal of left ventricular remodeling.¹¹⁹ Ventricular remodeling is known to occur after a non-Q wave MI or, in the absence of a prior infarction, in the presence of dysfunctional, but viable myocardium.¹²⁰ The end-diastolic and end-systolic volumes of the LV progressively increase, LVEF declines, and the LV becomes more spherical. Revascularization of reversibly dysfunctional viable myocardium can reverse the remodeling process; follow-up studies 7.6 months after revascularization revealed significant decreases in left ventricular volumes (both in end-diastole and end-systole), an increase in LVEF, and a less spherical shape of the LV.¹¹⁹

It is important to emphasize that despite the presence of myocardial viability, even involving an adequate amount of the LV, surgical revascularization may not always be followed by a functional improvement or reversed remodeling of the LV. Fixed subendocardial scar with normal adequate perfused subepicardium as after PCI for acute coronary syndromes may show reduced perfusion with good ¹⁸F-FDG uptake but fail to improve function after revascularization. left ventricular function may also fail to improve after revascularization in markedly enlarged LVs with advanced remodeling.^121,122 Increased left ventricular end-systolic volumes or dimensions predicted failure to recover left ventricular function.¹²¹ Large myocardial regions with severely reduced perfusion require longer times for improvement in contractile function after revascularization, sometimes as long as 12 to 18 months.^119,123,124 Finally, myocardial viability may not persist indefinitely but may progress to scar tissue formation.^125,126 In patients with PET demonstrated myocardial viability, for example, left ventricular function failed to improve when revascularization was delayed by an average of 35 days.¹²⁶ Hence, there may be a basis for revascularization as soon as possible to avert irreversibility that precludes improvement of left ventricular function after late revascularization.

Heart Failure Symptoms and Physical Activity

Revascularization of viable myocardium in ischemic cardiomyopathy was also associated with relief of congestive heart failure symptoms.¹²⁷ In one study, 81% of patients in New York Heart Association (NYHA) heart failure classes 3 and 4 had improved to class 1 or 2 one year after CABG as compared to an improvement in heart failure symptoms in less than 20% of patients without viable myocardium.¹²⁸ Postsurgical improvements in physical activity are related to the amount of viable myocardium with improved physical exercise during daily life prior to and 24 months after CABG.¹²⁸ Expressed in metabolic equivalents (METs), the level of physical activity increased by an average 107% in patients with a greater than 18% mismatch as compared to an only 34% increase in patients with a mismatch less than 5% in size.

Myocardial Viability in Early Postinfarction and Reperfused Myocardium

Patterns of myocardial perfusion and glucose utilization differ early after an acute MI. In early postinfarction patients without reperfusion, studied within 72 hours of onset of acute symptoms, 50% of regional flow defects were associated with concordantly reduced ¹⁸F-FDG uptake; ¹⁸F-FDG uptake was preserved or enhanced in the remainder of flow defects.¹²⁹ Viability in these patients was associated with residual anterograde flow through the infarct vessel.¹³⁰ Contractile function in segments with matching flow and glucose metabolism defects failed to improve on follow-up, whereas some improvement in contractile function was observed in half of the segments with initially persistent metabolic activity.

Persistent ¹⁸F-FDG uptake with subsequent improvement in regional contractile function has been reported in some early postinfarction patients, whereas in other patients, regional ¹⁸F-FDG activity concentrations in postischemic myocardium were lower than regional flows, a pattern referred to as reversed mismatch.^131,132 The size of regions with diminished ¹⁸F-FDG uptake exceeds that of contrast enhancement on MRI.¹³¹ Importantly, oxidative metabolism rates determined with C-11 acetate in reverse mismatch regions were not significantly lower than in remote myocardium, and thus were normal, suggesting a possible shift in substrate selection as the reason of the reduced ¹⁸F-FDG uptake. Of interest, relative flows in reversed mismatch regions were similar to those in remote myocardium while relative flows in mismatch regions were significantly reduced. This observation raises the question of whether flow in reversed mismatch regions had recovered more rapidly than glucose utilization rates and thus accounted for the observed flow metabolism pattern, whereas in mismatch myocardial regions, the recovery of myocardial blood flow was substantially delayed.

Postinfarction Inflammation and Regeneration

Because ¹⁸F-FDG uptake is not specific to myocardium, increases in radiotracer uptake in reperfused myocardium may also reflect accumulation of leukocytes in postinfarction myocardium as the result of recruitment of inflammatory cells as active participants in infarct healing and regeneration.^133,134 Indeed, regional increases in ¹⁸F-FDG uptake in reperfused myocardium in early postinfarction patients have been demonstrated when radiotracer uptake in remote myocardium had been effectively suppressed by prolonged fasting.^133,135 Although the findings did not fully exclude the possibility of increased uptake in injured myocardium, studies with a novel radiotracer targeting the chemokine receptor CXCR4 as a “central regulator of infarct healing and regeneration” unequivocally demonstrated the postinfarction inflammatory process in early post infarction patients.¹³⁶ Thus, increased ¹⁸F-FDG uptake in acutely reperfused myocardium may reflect the inflammatory process, the presence of viable myocardium, or both.

Cardiac Disorders Potentially Mimicking Patterns of Perfusion and Metabolism in CAD

Transient apical ballooning, called Takotsubo (stress) cardiomyopathy, may mimic an acute coronary syndrome; it affects predominantly postmenopausal women and accounts for 1% to 2% of patients with troponin-positive acute coronary syndromes but with angiographically normal coronary vessels. It has been referred to as neurogenic myocardial stunning¹³⁷ because it frequently follows a stressful event. Wall motion is typically impaired and perfusion severely diminished in the apical portion of the LV.^137,138 MRI frequently does not show abnormal enhancement, whereas ¹⁸F-FDG uptake in the affected myocardial regions is typically reduced in proportion to blood flow (Fig. 19–51). Regionally ¹⁸F-FDG uptake may be increased when patients are studied after suppression of ¹⁸F-FDG uptake in normal myocardium with prolonged fasting.¹³⁷ The changes observed during the acute phase are fully reversible when wall motion, perfusion, and metabolism were examined several weeks later (see Fig. 19–51).^139,140 Finally, isolated left ventricular noncompaction may also cause regional myocardial perfusion defects that may or may not be associated with relative increases in regional ¹⁸F-FDG uptake.^137,141

Risk Assessment and Stratification

Assessment of myocardial viability has become clinically important in patients with ischemic cardiomyopathy with severely impaired left ventricular function because of its independent prediction of cardiac risk and on outcomes of treatment. A meta-analysis of long-term follow-up data in 3088 ischemic cardiomyopathy patients in 24 clinical investigations with an average LVEF of 30 ± 8% found the presence of myocardial viability to be associated with a 16% annual mortality in patients treated medically as compared to an only 6.2% annual mortality of patients without viable myocardium (Fig. 19–52).¹⁴²

Surgical revascularization of patients without viable myocardium remained without significant effect on annual mortality (7.7%) but was associated with a significantly lower annual mortality of only 3.2% in patients with viable myocardium. Revascularization thus substantially reduced the annual mortality of patients with viable myocardium in these studies by as much as 80% in the meta-analysis. Although most studies in the meta-analysis were retrospective, the findings nevertheless strongly suggest that viable myocardium, in ischemic cardiomyopathy with severely depressed left ventricular function, signifies a high risk of cardiac death that is significantly reduced by revascularization.

In response to shortcomings of that meta-analysis, especially the inclusion of retrospective data, the STICH (Surgical Treatment of IsChemic Heart failure) trial attempted to prospectively answer the question whether “CABG surgery improves survival in patients with ischemic left ventricular dysfunction compared to aggressive medical therapy.”¹⁴³ In a subgroup of 601 of 1212 patients enrolled in the STICH trial, viability was assessed with SPECT Tc-99m perfusion imaging or with low-dose dobutamine stress echocardiography.¹⁴⁴ Although the viability substudy revealed a significantly lower mortality for patients with than without viability (37% vs 51%; P < .003), this difference was no longer significant after adjustments were made for baseline variables. The failure of the STICH substudy to demonstrate survival benefit of patients with myocardial viability and surgical revascularization may be ascribed to the limitation of SPECT imaging and lack of radionuclides specific for and proven to identify viable myocardium and its quantitative extent. In contrast, ¹⁸F-FDG PET has shown essential for predicting benefit from revascularization.^145,146

Indications for Perfusion-Metabolism Imaging

Patients with ischemic cardiomyopathy, severe LV dysfunction, and congestive heart failure symptoms will benefit most from the assessment of myocardial viability. Presence or absence of myocardial viability can substantially contribute to treatment planning. Therapeutic options in these patients include aggressive medical management, revascularization, and cardiac transplantation. Presence of viable myocardium in amounts sufficient for a possible postrevascularization improvement in LV function favors stratification of patients to revascularization provided, of course, that the coronary anatomy is suitable for CABG surgery. Demonstration of viability in these patients not only indicates the high risk of cardiac death if treated medically, but also favorably affects the surgical risk-to-benefit ratio, considering the high perioperative risk of these patients.

In patients considered initially for cardiac transplantation, surgical revascularization based on demonstration of extensive amounts of viable myocardium resulted in substantial improvements in left ventricular function and congestive heart failure symptoms together with long-term survival rates that were comparable to those of patients following cardiac transplantation.¹⁴⁷ Even if the coronary anatomy is unsuitable for CABG surgery, the presence of viability is predictive of a more favorable outcome with beta-blocker treatment or with cardiac resynchronization therapy.^{148,149,150,151,152} Finally, evaluation of early postinfarction patients for myocardial viability can also aid in deciding on interventions seeking to salvage myocardium compromised by ischemia.¹³⁰

Rest-Stress Quantitative Perfusion, Myocardial Scar, and Viability

Large severe relative resting perfusion defects with uptake below 20% to 30% of maximum are most commonly transmural scar by FDG PET. Quantitative perfusion in such defects is commonly 0.2 cc/min/g or less as in the case examples above that may increase somewhat with stress, the hyperemia of scar tissue, or a thin layer of subepicardial myocardial cells. For such very large severe resting perfusion defects, if clinical circumstances indicate, ¹⁸F-FDG PET may show substantial viability or more commonly scattered subepicardial foci of viability. For less severe to moderate relative resting perfusion defects, quantitative perfusion may range to near normal, indicating a corresponding range of nontransmural scar. Resting defects with perfusion of about 0.4 cc/min/g or higher suggest sufficient viable myocardium that ¹⁸F-FDG PET may not add substantial new information. However, with the high risk of bypass surgery, definitive ¹⁸F-FDG imaging is indicated.

With resting perfusion in the range of 0.2 to 0.4 cc/min/g, the clinical question is whether this average transmural perfusion represents viable under perfused transmural hibernating myocardium or fixed nontransmural scar comprised of fixed subendocardial scar with nonischemic adequately perfused subepicardium at resting conditions. Similarly, the mean transmural uptake of ¹⁸F-FDG may reflect either mean transmural viable myocardium or the average of no uptake in subendocardial scar with intense subepicardial uptake. In many patients, these possibilities are mixed heterogeneously with transmural scar, border zone nontransmural scar, hibernating, stunned, ischemic, normally perfused myocardium, and primary diabetic or hypertensive cardiomyopathy in the same patient. In these circumstances, combined quantitative rest-stress perfusion and ¹⁸F-FDG imaging may be necessary as described above for comprehensively assessing the proportion of viable ischemic, scarred, or normally perfused myocardium in specific arterial distributions for guiding management as in the cases above.

Myocardial Cellular Potassium Space, Injury, and Viability

Myocardial cell membranes normally maintain high intracellular potassium concentrations compared to the extracellular fluid, thereby maintaining the resting membrane potential of the myocardial cells (Fig. 19–53).¹⁵³ Myocardial injury causes potassium to leak out of the cell with an injury current reducing the resting membrane potential manifest as ST-segment elevation on the AC-coupled ECG. However, some intracellular potassium may be retained with a reduced membrane potential associated with potential cell recovery. Restoration of oxygenated coronary blood flow restores normal potassium trapping, normal resting membrane potential, and normal contraction. However, with ongoing injury caused by low blood flow ischemia, cell necrosis follows with irreversible loss of all potassium trapping.

The positron emitting potassium analog Rb-82 follows the distribution of the cellular potassium space of viable myocardium. It is, therefore, not trapped in myocardial scar associated with severe perfusion defects.^{154,155,156,157} Residual activity in resting perfusion defects indicates residual myocardium with intact potassium space that traps Rb-82 as a perfusion tracer. Initial Rb uptake may leak back out or wash out of injured myocardium to some extent while ¹⁸F-FDG is still trapped, indicating some viability. Therefore, the rate of potassium washout may not reflect viability because myocardium taking up ¹⁸F-FDG may show some Rb-82 leaking out of the myocardial cells. However, myocardium that fails to trap Rb on standard static 5-minute images is usually scar, the size of which relates to the LVEF (Fig. 19–54) and adverse outcomes.^155,156 However, ¹⁸F-FDG uptake may identify viable myocardium in select cases and remains the definitive modality for assessment of myocardial viability.

As for all perfusion radionuclides, the clinical question is whether residual average transmural trapping of perfusion tracers and perfusion indicates the average of subendocardial scar with subepicardial perfusion or low mean transmural perfusion to viable myocardium. Of course, ¹⁸F-FDG images pose the same question. Both experimental infarction and pathology of human infarction indicate that the extent of subendocardial infarction or ischemia is nearly always greater than the extent of subepicardial infarction or injury. For large regions of viability, 15% to 20% of LV, associated with benefit from revascularization, the fine distinction between subendocardial and subepicardial extent of infarction or injury is likely less important than the overall proportion of the LV at risk.

Alternative Non-PET Viability Imaging

Identification of Viable Myocardium with Perfusion Imaging at Rest with SPECT and Tc-99m-Labeled Flow Tracers

Perfusion in functionally impaired myocardial regions contains information on potential reversibility of wall motion. Blood flow at rest closely follows the distribution of normal myocardium; accordingly, it decreases in proportion to the amount of fibrosis and scar tissue. Transmural tissue concentrations of flow tracers correlate inversely with the transmural amount of fibrosis and scar tissue on histopathologic analysis.¹⁵⁸ Accordingly, regional flow reductions closely and quantitatively correlate to the amount of fibrosis and scar tissue. For assessments of potentially reversible contractile dysfunction, myocardial segments with flow reductions of less than 50% (ie, < 50% transmural scar tissue) are defined as “viable,” whereas more severe flow reductions (> 50%) are defined as “nonviable.” The approach identifies potentially reversible regional wall motion abnormalities with a sensitivity and specificity of 82% and 57%, respectively.¹⁰⁷ Assessment of perfusion alone does not identify a key feature of viable myocardium (ie, the enhanced extraction of glucose) and defines regions with severe flow reductions as nonviable even though comparison studies with PET have shown that myocardial viability can be associated with more severely reduced flows.¹⁵⁹ This limitation likely accounts for the only moderately high negative and predictive accuracies of SPECT imaging (ie, 72% and 71%, respectively).¹⁰⁷

Contrast-Enhanced MRI for the Assessment of Myocardial Scar Tissue

Late gadolinium contrast enhancement on MRI indicates the presence of scar tissue. In patients with ischemic cardiomyopathy, the approach identifies the transmural extent of scar tissue. Nonenhancing myocardial tissue is considered to be normal or, less precisely, to be “viable.” Generally, myocardial segments with less than 50% transmural scar tissue improve contractile function after restoration of blood flow, whereas the likelihood of a postrevascularization improvement in contractile function declines with increasing transmurality of late contrast enhancement.¹⁶⁰ Analogous to the resting perfusion approach, the late enhancement does not differentiate between normal and ischemic or viable myocardium. Sensitivity and specificity are 84% and 64%, respectively.¹⁰⁷

However, it is possible to distinguish between normal and viable myocardium in nonenhancing regions by hybrid ¹⁸F-FDG PET/MRI.¹³⁵ In early postinfarction patients, wall motion was more likely to improve in nonenhancing myocardial segments on MRI with persistent ¹⁸F-FDG uptake than without (Fig. 19–55). Hybrid PET/MRI could thus enhance the diagnostic accuracy of the late contrast enhancement MRI approach because it would identify presence and amounts of viable and of normal myocardium in nonenhancing regions as well presence and extent of irreversibly injured (scar) tissue. As an additional and important aspect of MRI, its high spatial resolution offers predictive information based on wall-thickness measurements. Severe reductions in end-diastolic wall thickness (ie, < 7 mm) were highly accurate in predicting irreversible wall motion impairment with a 92% negative predictive accuracy.¹⁰⁷

Thallium-201 Rest Redistribution for Cell Membrane Integrity

Cell membrane integrity, and thus, preserved transmembrane ion concentration gradients in functionally compromised myocytes serve as another measure of reversibility that can be evaluated with thallium (Tl)-201 rest and redistribution imaging. Like Rb-82, Tl-201 is a potassium analog reflecting potassium trapping in the myocyte. The initial, early postinjection myocardial images depict the distribution of myocardial blood flow while delayed (4 hours or more) “redistribution images” reflect the equilibrium of the cation Tl-201 between myocardium and blood and, thus, the myocardial potassium pool. Although highly predictive in patients with only moderately reduced left ventricular function,¹⁰⁷ its accuracy is limited in patients with severely depressed left ventricular function, because of low signal-to-noise ratios resulting from low myocardial and high blood radiotracer activities, limiting the certainty with which redistribution can accurately be identified.^{161,162,163,164}

Contractile Reserve by Low-Dose Dobutamine Stimulation

Persistence of contractile reserve serves as another marker of potentially reversible contractile dysfunction. Wall motion responses to inotropic stimulation with intravenous low-dose dobutamine infusion, assessed by 2D echocardiography, cine CT, or gated MRI, accurately predict postrevascularization outcomes of regional myocardial wall motion and global left ventricular function, including changes in the left ventricular volumes, both in end-diastole and the end-systole and thus, in left ventricular remodeling.¹⁰⁷

Identification of Myocardial Viability by Reversibility of Stress Perfusion Defects

Stress-induced perfusion defects on myocardial perfusion imaging with SPECT have been considered as indicators of reversible contractile dysfunction. However, in patients with ischemic cardiomyopathy and severely impaired left ventricular function, stress-rest myocardial perfusion imaging has proved to be of limited accuracy in distinguishing between reversible and irreversible contractile dysfunction (Fig. 19–56).¹⁶⁵ Possible reasons for this limitation include impairments in CFR in “remote” myocardium resulting from multivessel macrovascular and/or microvascular disease, so that pharmacologic stress fails to selectively induce more severe perfusion defects. Rather, assuming that the perfusion reserve is uniformly diminished throughout the left ventricular myocardium, the relative distribution of myocardial blood flow during rest and stress remains relatively constant so that a perfusion defect appears “fixed” despite the presence of viable myocardium.¹⁶⁶

Cardiac Sarcoid

Sarcoidosis is a granulomatous disease that most often affects the lymphoid and respiratory systems. The cardiovascular system is the third most often affected organ system, although its involvement frequently remains undiagnosed. Several etiologies have been proposed and explored, although the underlying etiology of sarcoidosis still awaits full clarification; sarcoidosis has been defined as a “multisystem disorder of unknown cause(s).”¹⁶⁷

The presence of sarcoidosis is histologically confirmed by demonstration of noncaseating epithelioid granulomas. Involvement of the cardiovascular system is frequently underestimated because cardiac sarcoidosis remains clinically silent in as many as 50% of patients. Different from the 5% to 10% incidence reported previously, more recent use of advanced diagnostic imaging techniques such as CT and MRI has raised the incidence to about 40%¹⁶⁸; nearly 50% of patients with cardiac involvement are asymptomatic. Importantly, however, cardiac complications account for about 50% of all deaths in sarcoidosis patients.

According to autopsy findings in patients with cardiac sarcoidosis, the LV free wall, including the papillary muscles, together with the immediately adjacent myocardium, the interventricular septum, and the right ventricular myocardium are affected most frequently.¹⁶⁹ The inflammatory regions preferentially reside in the more basal portions of the left ventricular myocardium and typically exhibit granulomatous noncaseating lesions; with progression of the active inflammatory process, there is myocardial repair with increased fibrosis and formation of scar tissue. Clinicopathologic correlations have found no statistically significant associations of the site of inflammation and clinical findings. However, small sarcoid lesions appear to be associated with fewer clinical symptoms or cardiac findings, whereas more extensive involvement tends to be associated with left ventricular dysfunction and heart failure.¹⁶⁹

Perfusion and Metabolism Imaging in Cardiac Sarcoid

The diagnosis of cardiac sarcoidosis has remained challenging because it depends on the histologic demonstration of noncaseating granulomas in myocardium. In the absence of myocardial biopsy data, a diagnostic algorithm is employed that includes presence of histologically confirmed systemic sarcoidosis together with clinical signs of cardiac involvement. The now modified Japanese Ministry of Health and Welfare (JMHW) 2006 guidelines consist of major and minor criteria.¹⁷⁰ Major criteria include advanced atrioventricular block, basal thinning of the interventricular septum, positive myocardial ¹⁸F-FDG uptake, and depressed LVEF to less than 50%. Minor criteria include abnormal ECG, abnormal echocardiogram (regional wall motion abnormality), myocardial perfusion defect (Tc-99m SPECT or Rb-82 or N-13 ammonia PET), gadolinium-enhanced MRI, and an abnormal endomyocardial biopsy (interstitial necrosis or moderate monocyte infiltration). Two or more major criteria or only one major and at least two minor criteria are required for the diagnosis of cardiac sarcoid.

Detection of inflammatory sarcoid lesions strongly depends on adequate suppression of ¹⁸F-FDG uptake in normal myocardium. It requires careful patient preparation prior to PET/CT imaging in order to shift the myocardium’s substrate selection from glucose to free fatty acid, thereby reducing accumulation of ¹⁸F-FDG in normal myocardium. Among several approaches, prolonged fasting for 16 to 20 hours has been proven most effective in consistently suppressing ¹⁸F-FDG uptake.^171,172,173 The fasting approach can be supplemented with intravenous unfractionated heparin 15 minutes prior to ¹⁸F-FDG injection in order to further raise circulating free fatty acid levels,¹⁷⁴ although a true incremental effect on the suppression of myocardial ¹⁸F-FDG uptake remains uncertain.

The study protocol typically includes assessment of myocardial perfusion at rest and of myocardial and whole-body ¹⁸F-FDG uptake (the latter for identifying extracardiac sarcoid as a major criterion). Myocardial perfusion is preferably assessed with PET/CT either with N-13 ammonia or Rb-82 but can also be evaluated with SPECT myocardial perfusion imaging with Tc-99m labeled radiotracers of blood flow. Perfusion defects indicate the presence of scar tissue as evidence of the postinflammatory myocardial repair process. ¹⁸F-FDG images are typically acquired 60 to 120 minutes after the radiotracer administration. They include dedicated cardiac and whole-body or only chest ¹⁸F-FDG images for identifying the inflammatory activity as well as the presence or absence of systemic sarcoidosis. As an alternative clinically more practical approach, especially when searching for active inflammation as the target of treatment, metabolism imaging is performed first for identifying presence or absence of acute inflammatory lesions in the myocardium and evidence of systemic sarcoidosis; myocardial perfusion images are then obtained the following day as needed.

Patterns of Myocardial Perfusion and Glucose Imaging in Sarcoid

Patterns of myocardial ¹⁸F-FDG uptake are defined as (1) no uptake, (2) mild diffuse uptake, (3) focal on diffuse uptake, and (4) focal uptake (one or more foci of increased uptake; Fig. 19–57). Absence of focally enhanced ¹⁸F-FDG uptake effectively rules out the presence of moderate- to large-sized inflammatory sarcoid granuloma, although small lesions as reported from autopsy studies in patients with systemic sarcoid may remain undetected. Diffuse myocardial ¹⁸F-FDG uptake poses several diagnostic challenges. It may be related to insufficient suppression of ¹⁸F-FDG uptake in normal myocardium, but it has also been attributed to a diffuse nonsarcoid inflammation of the myocardium.

Most consistent with cardiac sarcoid are one or more myocardial foci of intensely increased ¹⁸F-FDG uptake (see Fig. 19–57). Their location does not follow the distribution of coronary vessels; they preferentially locate in the more basal portions of the left ventricular myocardium—that is, the basal segments of the inferior ventricular septum and the inferior and the lateral walls, as well as the right ventricular wall. Occasionally, focally enhanced ¹⁸F-FDG uptake may be observed selectively in papillary muscles; while possibly related to cardiac sarcoid, such selective uptake may also represent a normal variation.

While the diagnosis of cardiac sarcoid is predicated on the presence of systemic sarcoid, recent observations have provided evidence for sarcoidosis limited to the cardiovascular system. Necropsy findings in patients with sudden death from cardiac sarcoid have failed to detect gross extracardiac disease in 10 of 25 (40%) with cardiac sarcoid lesions.¹⁷⁵ Observations with ¹⁸F-FDG PET in patients with suspected cardiac sarcoid, but without apparent clinical evidence of systemic sarcoid, are similar¹⁷⁶; in 25 patients with focally enhanced myocardial ¹⁸F-FDG and flow defects, only 13 revealed involvement of mediastinal lymph nodes on whole-body ¹⁸F-FDG PET, whereas the remaining 12 (48%) patients were without evidence of extracardiac sarcoidosis.

Addition of myocardial perfusion imaging increases the sensitivity of identifying cardiac sarcoid.¹⁶⁸ Perfusion defects correspond to scar tissue after healing of the acute inflammatory process. The incidence of regional perfusion defects varies between clinical investigations—possibly for technical reasons (SPECT vs PET perfusion imaging)—but also because of differences in patient populations. Early acute inflammatory stages of the disease exhibit abnormally increased ¹⁸F-FDG, whereas more advanced stages are associated with tissue fibrosis and scar formation and, consequently, with perfusion defects. Therefore, depending on the stage of the granulomatous disease, cardiac sarcoid may be associated with focally increased ¹⁸F-FDG uptake only, reflecting the acute inflammatory process; with flow defects only, reflecting myocardial scar tissue; or with both, regionally increased ¹⁸F-FDG uptake and reduced flow, reflecting a combination of both.

Diagnostic Accuracy of PET for Sarcoid

Estimates of the diagnostic accuracy of perfusion-metabolism imaging are limited because histopathologic evidence of myocardial noncaseating inflammatory lesions as the gold standard is frequently missing. The diagnostic yield of endomyocardial biopsies has remained low (in the range of 30% to 35%).¹⁷⁷ The diagnosis of cardiac sarcoid is further complicated because it is predicated on histopathologic proof of systemic sarcoid as defined by the JMHW criteria.¹⁷⁰ For example, in 118 patients without CAD, 38 patients met the JMHW criteria for cardiac sarcoid, although only 16 (or 42%) had abnormal cardiac findings on PET imaging.^178,179 Among 20 patients with perfusion defects and increased ¹⁸F-FDG uptake, endomyocardial biopsy findings were positive in only 9 (45%) patients.

Overall, the presence of cardiac sarcoid as established by JMHW criteria correlates poorly with that by cardiac PET/CT as well as that by endomyocardial biopsy findings. Some of the disparity among diagnostic strategies also relates to shortcomings of the JMHW criteria, which require the presence of systemic sarcoidosis while sarcoid disease may in fact be limited to the cardiovascular system in some patients. As another reason for the diagnostic disparities is that the current diagnostic criteria do not consider the possibility of focally increased myocardial ¹⁸F-FDG uptake due to nonsarcoid inflammatory processes.

Focal Myocardial ¹⁸F-FDG Uptake Unrelated to Cardiac Sarcoid: Differential Diagnosis

The pattern of focally increased myocardial ¹⁸F-FDG uptake is not specific for cardiac sarcoid but also occurs in other types of granulomatous and nongranulomatous inflammatory processes of the myocardium as, for example, in tuberculous myocarditis,¹⁸⁰ Epstein-Barr virus myocarditis,¹⁸¹ or Chagas disease (Fig. 19–58).¹⁸² In a retrospective study of 29 patients with suspected myocarditis, focally increased ¹⁸F-FDG uptake was observed in nearly half of the patients with histopathologically confirmed myocarditis (lymphocytic cell infiltrates associated with myocyte necrosis).¹⁸³

Apart from myocardial inflammatory processes, interpretation of perfusion and ¹⁸F-FDG images should include consideration of potentially alternate causes of focally enhanced myocardial ¹⁸F-FDG uptake. Regional increases in myocardial ¹⁸F-FDG uptake may be related to postischemic myocardium after strenuous exercise in patients with epicardial CAD^184,185 or following an ischemic injury (Fig. 19–59). Stress-related, postischemic regional increases in ¹⁸F-FDG uptake may persist for as long as 24 hours or more but typically correspond in location to the distribution of a coronary artery, whereas sarcoid lesions typically reside in the more basal portions of the LV and are unrelated to coronary vascular territories.

The known physiologic heterogeneity of myocardial ¹⁸F-FDG uptake can complicate the image interpretation. Myocardial ¹⁸F-FDG uptake in normal individuals is more prominent in the posterior lateral wall^186,187; activity concentrations are on average about 10% higher than in the anterior and anterior septal wall after standard glucose loading prior to PET/CT imaging, and may be as high as 20% when patients are imaged after prolonged fasting.

Primary and secondary cardiac tumors may also account for focal or regional increases in myocardial ¹⁸F-FGD uptake. Primary cardiac tumors are rare, and most are benign. Secondary cardiac tumors are 40 to 100 times more frequent than primary tumors.¹⁸⁸ They are seen in cancer of the lung, breast, and kidneys and in melanoma, lymphoma, and leukemia. As compared to the mild to moderate ¹⁸F-FDG uptake in benign cardiac tumors, malignant cardiac tumors characteristically exhibit intense ¹⁸F-FDG activity.^174,189 Differentiation of cardiac tumors from inflammatory processes may be possible with radiolabeled amino acids, as has been shown for cardiac metastases of serotonin-producing neuroendocrine tumors with ¹⁸F fluoro-L-dopa.¹⁹⁰ However, it is probably clinically accomplished best by structural characterization with contrast CT or MRI.¹⁹¹

Cardiac MRI contributes to the diagnosis of cardiac sarcoid.^177,192 The myocardial distribution of late gadolinium enhancement is characteristically “noncoronary” and primarily involves the subepicardial and mid-myocardial layers with sparing of the endocardium, a pattern distinctly different from that of ischemia and infarction (Fig. 19–60). Based on comparison studies, MRI appears to be more specific but less sensitive than ¹⁸F-FDG PET for the detection of cardiac sarcoid.¹⁹³ Furthermore, findings on PET and MRI can differ. Of 21 patients, 8 patients were positive on either PET or MRI, while 8 patients were positive on both, yet with frequently disparate locations of PET and MRI lesions. Late contrast enhancement of MRI reflects scar tissue and edema as compared to active inflammation on ¹⁸F-FDG PET so that interpatient and intrapatient disparities likely reflect different stages of the inflammatory process. Use of MRI may, however, be limited in patients with cardiovascular implanted electronic devices (CIEDs).

It is uncertain whether the intensity of regional ¹⁸F-FDG uptake can be used as a diagnostic markers of cardiac sarcoid. Besides the level of inflammation, the apparent intensity of ¹⁸F-FDG uptake depends on the size of the inflammatory lesion with lower activity in smaller lesions. Some evidence suggests that standard uptake values (SUVs) or coefficients of variance, an index of the heterogeneity of the myocardial ¹⁸F-FDG uptake, improves the interpretive certainty and, possibly, yield associations with specific electrocardiographic abnormalities. It is also possible to assess the abnormal, sarcoid-related ¹⁸F-FDG uptake more quantitatively. Analogous to the metabolic tumor volume and the total lesion glycolysis, widely used with oncologic PET studies, the cardiac metabolic volume and the cardiac metabolic activity can be estimated from the total volume of increased ¹⁸F-FDG uptake (using threshold values for defining the extent and, thus the total volume of abnormal ¹⁸F-FDG uptake) and the intensity of ¹⁸F-FDG uptake, derived from average SUV (intensity of ¹⁸F-FDG uptake corrected for the dose of activity injected and body weight).^179,194

Predictive Value of FDG PET/CT and Therapeutic Implications in Sarcoid

PET perfusion-metabolism imaging in cardiac sarcoid contains predictive information on cardiac death and risk of sustained ventricular tachycardia. One investigation reports an annual event rate of 7.3% in patients without focal ¹⁸F-FDG uptake that increases to 18.4% with either a flow defect or focal ¹⁸F-FDG uptake and, even further to 31.9% when both flow defects and focally increased ¹⁸F-FDG uptake are present.¹⁷⁸ Semiquantitative indices of the total ¹⁸F-FDG uptake such as cardiac metabolic volume and cardiac metabolic activity similarly contain predictive information and appear to be useful for estimating treatment responses.

¹⁸F-FDG PET/CT defines the target of therapy and aids in monitoring treatment responses.¹⁹⁵ Besides antiarrhythmic treatment and ICD implantation, immunosuppression with corticosteroids is the treatment of choice. It is associated with decreasing lesional ¹⁸F-FDG uptake on repeat imaging (Fig. 19–61).¹⁹⁶ In one study, the cardiac metabolic volume decreased in five of seven patients by an average of 88% during steroid treatment¹⁹⁴; in another study, therapy-induced decreases in the cardiac metabolic volume were associated with moderate improvements in the LVEF.¹⁷⁹

Infective Endocarditis

The modified Duke criteria remain the diagnostic gold standard of infective endocarditis¹⁹⁷ (see Chap. 67 on endocarditis). Major criteria include positive blood cultures, evidence of endocardial involvement, echocardiographic evidence for infective endocarditis, and new-onset valvular regurgitation. Minor criteria include a predisposition for infective endocarditis, persistent fever despite antimicrobial treatment, vascular phenomena, immunologic phenomena, and microbiological evidence. The criteria are highly specific for infective endocarditis, but their sensitivity is low, so that infective endocarditis and its downstream complications may remain undetected or are recognized only late in the course of the disease. Early diagnosis is, however, critical for initiating aggressive and prolonged antibiotic treatment and, if indicated, valvular surgery in order to reduce morbidity and improve overall survival.

There is now substantial evidence that ¹⁸F-FDG PET/CT contributes to the identification of infective endocarditis (including infection of prosthetic valves, CIEDs, pacemakers, and ICDs), endovascular devices and prosthetic vascular grafts and, importantly, detection of downstream septic emboli.

FDG PET/CT Imaging of Infective Endocarditis

Transthoracic or transesophageal echocardiography detects valvular infections and abscesses with a sensitivity of 75% to 85%; both are considered “major criteria.”^197,198 Vegetations and abscesses of native and prosthetic valves can be visualized with ¹⁸F-FDG PET/CT, though with a lower and clinically insufficient sensitivity (about 40%) and a specificity of about 70% to 100%.^199,200 They are typically seen as small foci of mildly to intensely increased ¹⁸F-FDG activity in the region of the mitral or aortic valve (Fig. 19–62) and do not invariably correspond to findings on echocardiography. In one investigation in 30 patients with definite prosthetic valve infection, for example, findings by echocardiography and ¹⁸F-FDG PET/CT agreed in only half of the patients; in nearly half of the remaining patients, focally increased ¹⁸F-FDG uptake was seen in the valve area without corresponding vegetations on echocardiography, possibly because of an early stage of disease²⁰⁰ and when echocardiography may still be negative. Conversely, absence of abnormal ¹⁸F-FDG uptake effectively ruled out the presence of prosthetic valve infection in that study so that the authors proposed to include PET/CT as a major criterion for the diagnosis of infective endocarditis. The feasibility of ¹⁸F-FDG PET/CT for visualizing infected endovascular leads has also been explored but has remained of limited accuracy because of low signal intensity.²⁰¹

Clinically more promising has been the use of ¹⁸F-FDG PET/CT for diagnosing infections of cardiovascular implantable electronic devices. According to the Consensus on Transvenous Lead Extraction,²⁰² removal of devices is indicated when one of three criteria is present: (1) infection of device (or generator) pocket, (2) lead endocarditis, and (3) persistent or recurrent bacteremia. Several studies have reported sensitivities ranging from 76% to 100% for detecting device infections with ¹⁸F-FDG PET/CT.^201,203 Abnormally increased ¹⁸F-FDG uptake in patients with device infections is frequently seen in the pocket around the generator, over intravascular leads, superficially in skin and subcutaneous tissues, and within the region of the heart (Fig. 19–63). Evidence of infection on ¹⁸F-FDG PET/CT led to device extraction in 24 of 27 patients in one study.²⁰³ Negative findings on ¹⁸F-FDG PET/CT may also prove useful as they may aid in deciding against device extraction.

Limitations of Endocarditis Imaging

Despite the promise of ¹⁸F-FDG PET/CT for visualizing the site of endocarditis (native and prosthetic valves), limitations remain. One is the frequently small size of infectious lesions, at least relative to the spatial resolution of the PET imaging system. A second one is the low intensity of the radiotracer activity relative to background activity (ie, ¹⁸F-FDG activity concentrations in blood and myocardium) and, thus, a low signal-to-noise ratio. Approaches to enhance the signal strength include delayed image acquisition after radiotracer administration in order to allow more time for clearance of radiotracer from blood and suppression of radiotracer uptake in normal myocardium by prolonged fasting in order to suppress myocardial glucose uptake.¹⁷¹ Delaying the time of image acquisition from the standard 1 to 3 hours substantially after the radiotracer administration improved the detection of device infections (especially of infected endovascular leads).²⁰³

As a third limitation, increased radiotracer uptake in noninfected tissues or structures adjacent to valves or prostheses may lead to falsely positive interpretations, especially in patients with recent device or prosthesis implants when postsurgical reactive tissue inflammation is associated with increased ¹⁸F-FDG uptake. A comparison study found no abnormal ¹⁸F-FDG uptake in patients with remote device implants without evidence of infection, only mild ¹⁸F-FDG uptake in patients with recent device implants (4-8 weeks) without device infection but focally intense ¹⁸F-FDG uptake in patients with suspected infection of device implants.²⁰³ Prosthetic vascular grafts, when combined with valve replacements, may also represent sites of endocarditis and demonstrate intense ¹⁸F-FDG uptake^204,205 but may also be associated, even when not infected, with persistent but only mildly to moderately increased radiotracer uptake.²⁰⁶

Assessment of Septic Emboli and Metastatic Infections

Investigations have established the clinical value of ¹⁸F-FDG PET/CT for identifying septic emboli and downstream infectious metastases. Their detection with conventional approaches (ultrasound, CT, and MRI) has remained challenging and is of limited effectiveness. Although frequent in infectious endocarditis, as many as 30% to 40% of downstream infections persist without localizing signs or symptoms but will significantly affect clinical outcomes. Timely detection of septic emboli and metastatic infection is thus critical for initiating aggressive prolonged antibiotic treatment together, if needed, with draining of infected abscesses, in order to avoid relapses of infectious endocarditis and to reduce mortality. In one investigation in 111 patients with gram-positive bacteremia, whole-body ¹⁸F-FDG PET/CT uncovered focally abnormal radiotracer uptake in 65 patients (58.6%).²⁰⁷ Of these, 53 were true positives for metastatic infections as verified by ultrasound, CT, or MRI. Forty percent of the patients had more than one lesion. There were 46 true negative patients overall; the study reported a sensitivity and specificity of 100% and 87%, respectively. Other investigators have confirmed the high sensitivity and specificity of ¹⁸F-FDG PET/CT for the detection of septic emboli and metastatic infections in patients with definite infectious endocarditis by the Duke criteria.^{199,200,208,209} Sites most often affected include the lung (in right-sided infective endocarditis), the bones, the colon, the spleen, and the brain. Metastatic infections of brain, myocardium, and kidneys may remain undetected because of physiologically high ¹⁸F-FDG uptake and, thus, high background activities.

Vascular Inflammation and Atherosclerosis

Inflammation is a hallmark of several vascular disease entities and, because of the avidity of inflammatory cells for ¹⁸F-FDG, can be identified with PET. Vascular imaging with PET in arteritis of the large vessels is clinically useful while its clinical role in atherosclerotic disease remains to be fully determined.

Large-Vessel Vasculitis

Among inflammatory diseases of the great arteries, giant cell arteritis and Takayasu’s arteritis have been studied most with PET/CT. Monocytes and macrophages avidly accumulate and retain ¹⁸F-FDG²¹⁰ and represent the “biological” substrate of enhanced ¹⁸F-FDG uptake in the walls of the great arteries in patients with inflammatory vascular diseases (Fig. 19–64). Patients with suspected arteritis are studied in the fasting state with acquisition of whole-body PET and PET/CT, beginning in most investigations at 45 minutes after intravenous ¹⁸F-FDG administration.²¹¹ Longer time intervals are preferable to allow more time for vascular uptake and for clearance of radiotracer from blood, producing higher signal-to-background ratios with improved detection of vascular ¹⁸F-FDG uptake.²¹² Uptake in the arterial wall is graded visually on a three-point scale, where 1 is minimal or no uptake; 2, mild but definite; and 3, intense. A total vascular score, as the sum of all regional scores throughout the arterial system, can be obtained and be used for estimating progression or regression of vascular inflammation on serial PET imaging.²¹³

Diagnostic Performance

Case reports and investigations with only small patient numbers limit an accurate assessment of the diagnostic performance of PET and PET/CT ¹⁸F-FDG imaging in patients with arteritis. Pooled findings in 90 patients with clinically documented giant cell arteritis or Takayasu’s disease (based on the presence of least three of the classification criteria established by the American College of Rheumatology),^214,215 in four investigations, indicate a weighted average sensitivity of 81% and specificity of 86%.^211,213,216

Some, but not all studies, report a correlation between the vascular ¹⁸F-FDG uptake and serum inflammatory markers. In one study, the intensity of vascular ¹⁸F-FDG activity was found to correlate with serum C-reactive protein (CRP) and the erythrocyte sedimentation rate (ESR).²¹¹ The diagnostic sensitivity of vascular imaging appears to be related to levels of serum markers of inflammation and, thus, to the inflammatory activity. As suggested in one study, the sensitivity declines when CRP is less than 10 mg/L or the ESR is less than 12 mm/h.²¹¹

Clinical Indications

Vascular imaging with ¹⁸F-FDG PET/CT may prove useful for establishing the diagnosis of vasculitis, especially in patients with nonspecific symptoms as, for example, fever of unknown origin or polymyalgia rheumatica, which frequently coexists with giant cell arteritis. Furthermore, it delineates the spread of active disease throughout the great vessels. Although relatively infrequent, aortic aneurysms and dissections are complications of giant cell arteritis; complications of Takayasu’s arteritis include arterial stenoses, but also aneurysms, most often involving the aortic root and leading to valvular insufficiency.²¹⁷ Delineation of the extent of vascular involvement may therefore draw attention to possible sites of downstream vascular complications. Finally, the intensity of vascular ¹⁸F-FDG uptake serves as a measure of the degree of inflammation and of its response to therapy. The vascular ¹⁸F-FDG uptake declines with successful corticosteroid treatment, parallels symptomatic improvement (see Fig. 19–64),²¹⁸ and, as suggested by one investigation, reflects the treatment response more accurately than contrast-enhanced MRI.²¹⁸

Atherosclerosis and Plaque Imaging

General Considerations

Increased ¹⁸F-FDG uptake in the large arteries, including, in declining order, the ascending aorta and the aortic arch, the abdominal aorta, and the carotid arteries, but also the iliac and femoral arteries, is frequently observed in patients undergoing whole-body PET and PET/CT imaging for cancer diagnosis and staging. Unlike the diffuse vascular uptake seen in arteritis, ¹⁸F-FDG uptake in atherosclerosis is characteristically more heterogeneous and often focal.^{219,220,221,222} Studies combining PET with contrast CT or even MRI angiography have reported increased ¹⁸F-FDG accumulation in the carotid arteries of patients with recent cerebral vascular ischemic events^223,224 or in patients with carotid artery stenoses scheduled for endarterectomy.²²⁵

Consistent with findings in animal experiments,²²⁶ vascular ¹⁸F-FDG uptake correlates quantitatively with lesional concentrations of macrophages, especially of the activated proinflammatory type.^224,225 Potentially vulnerable or unstable atherosclerotic plaques are typically bulky in size, have a thin fibrous cap, exhibit a large necrotic core (≥ 25% of the plaque area) and are rich in macrophages.²²⁷ Rupture appears related to the severity of plaque inflammation and, thus, the degree of macrophage invasion. The magnitude of ¹⁸F-FDG (ie, macrophage concentration) might thus serve as a “marker” of potential plaque instability.²²⁵

Methodological Aspects of Imaging of Vascular ¹⁸F-FDG Uptake

Fusion of PET and CT images allows for easy localization of vascular radiotracer uptake and its extent and intensity. Regions of vascular ¹⁸F-FDG uptake typically coexist with regions of vascular calcification but rarely overlap.²²⁰ Focal increases in ¹⁸F-FDG uptake are more prevalent than are calcifications in younger patients, whereas vascular calcifications are more prevalent in older patients.²²⁰ Vascular radiotracer uptake is identified by its higher than blood activity and can be quantified by peak activity (counts/pixel/min) or by SUVs.^225,228,229 The target-to-background ratio (TBR) is a more quantitative measure of ¹⁸F-FDG uptake in the arterial wall because it is corrected for activity in blood.²³⁰

Repeat measurements over longer time periods reveal considerable changes in focal vascular ¹⁸F-FDG uptake. When reexamined after an 8- to 26-month period, vascular calcifications had remained relatively stable, but 48% of the initially observed regions of increased ¹⁸F-FDG had resolved while 32% of all lesions at follow-up had newly developed.²³¹ Although attributable to some extent to methodology-related differences, the lesion variability likely reflects the dynamic nature of atherosclerosis with healing of inflamed lesions and development of new inflammatory lesions. The incidence of vascular ¹⁸F-FDG uptake in patients with atherosclerotic disease is still undetermined.

Aortic aneurysms, especially of the abdomen, frequently reveal intense ¹⁸F-FDG activity. On histopathologic comparison studies, ¹⁸F-FDG activity concentrations correspond to accumulation of adventitial inflammatory cells often associated with decreased numbers of vascular smooth muscle cells, suggesting that ¹⁸F-FDG–rich abdominal aortic aneurysms might be prone to rupture.²³² Observations in a relatively small number of patients with abdominal aortic aneurysms appear to support this possibility; intense ¹⁸F-FDG uptake was associated with unfavorable outcomes including rupture, death, and surgical repair.²³³ Other studies have pointed to cyclic patterns of ¹⁸F-FDG uptake in aortic aneurysms where diameter increases are preceded by low ¹⁸F-FDG uptake.²³⁴ Future studies are thus needed to clarify the potential value of ¹⁸F-FDG uptake in guiding therapeutic strategies, especially in asymptomatic patients with abdominal aortic aneurysms that by size criteria might qualify for surgical (or endovascular) repair.^234,235

Coronary Arterial Plaques

Visualization of ¹⁸F-FDG uptake in atherosclerotic plaques of the coronary circulation faces formidable technical challenges for several reasons: the limited spatial resolution of PET relative to the size of the coronary vessels, cardiac and respiratory motion, and high background activity (ie, accumulation of ¹⁸F-FDG in the myocardium). Gated image acquisition for both cardiac and respiratory motion may thus become necessary for visualizing coronary arterial plaques and for quantification of ¹⁸F-FDG activity. Despite the inherently limited spatial resolution of PET, visualization of coronary plaques has become possible, especially because they are most frequently located in the larger sized proximal segments of the coronary arteries. Indeed, recent observations have confirmed such a possibility and have shown ¹⁸F-FDG avid lesions in the coronary vessels.²²⁰ In 25 patients with CAD, PET/CT images in patients with acute coronary syndrome revealed high ¹⁸F-FDG activity in the culprit coronary lesion (Fig. 19–65).²³⁶

Clinical Implications

Vascular ¹⁸F-FDG uptake has been associated with an increased incidence of cardiovascular events. Investigations in cancer patients undergoing whole-body ¹⁸F-FDG PET/CT have reported statistically significant associations among large artery ¹⁸F-FDG uptake, traditional risk factors for cardiac disease, and adverse cardiovascular events.^237,238 More definitive support for a predictive value of vascular ¹⁸F-FDG uptake comes from an investigation in 513 patients who had undergone whole-body ¹⁸F-FDG PET/CT for cancer surveillance.²³⁹ Patients enrolled were free of cancer, were at least 30 years of age and were without known cardiovascular disease. Over a median 4.2-year follow-up, 44 cardiovascular events occurred (ischemic stroke or transient ischemic attacks, acute coronary syndrome, coronary, carotid or peripheral revascularization, new-onset angina, peripheral arterial disease, heart failure, and cardiac death). The incidence of adverse events was related to the intensity of vascular ¹⁸F-FDG uptake (TBR in the ascending aorta). Abnormally high ¹⁸F-FDG uptake (TBR > 2.2) was associated with a 4.71-fold higher risk of a cardiovascular event than was low uptake (TBR ≤ 1.84). Of interest, cardiovascular events occurred earlier in patients with higher and more intense vascular ¹⁸F-FDG uptake.

¹⁸F-FDG uptake in large arteries can serve as imaging biomarker of the atherosclerotic inflammatory activity. In fact, local vessel inflammation as visualized with ¹⁸F-FDG correlates with serum inflammatory biomarkers.²⁴⁰ Vascular ¹⁸F-FDG imaging may therefore prove useful for monitoring responses to anti-inflammatory and plaque-stabilizing interventions. The enhanced ¹⁸F-FDG uptake in atherosclerotic vessels also serves as a surrogate end point for assessing anti-inflammatory effects of drugs. Lipid-lowering with statins has been associated with significant reductions in vascular ¹⁸F-FDG uptake,²³⁰ an effect that is dose dependent without, however, directly and significantly correlating with changes in plasma low-density lipoprotein cholesterol levels, corroborating a pleotropic anti-inflammatory effect of statins.

Active Mineral Deposition and Calcification of Atherosclerotic Plaques

¹⁸F sodium fluoride (¹⁸F-NaF) has recently emerged as another tool for probing the pathophysiology of atherosclerosis. Initial observations with this bone-imaging agent in oncologic patients reported focal increases in ¹⁸F activity throughout the large arteries. They were most prominent to in the femoral arteries, and, in declining order, the abdominal aorta and the thoracic aorta. Almost invariably, they were associated with vascular calcifications on CT.²⁴¹ In another study, vascular ¹⁸F-NaF lesions were found to be more intense than ¹⁸F-FDG lesions; increased ¹⁸F-NaF uptake colocalized with vascular calcifications as compared to only 14.5% of ¹⁸F-FDG sites. Importantly, ¹⁸F-FDG sites generally did not colocalize with ¹⁸F-NaF sites; in only 6.5% of all sites were both radiotracers present.²⁴² Vascular ¹⁸F-NaF uptake appears to be associated with cardiovascular events, CAD and the Framingham cardiac risk score.²⁴³

The findings reported in both the large arterial vessel and the coronary vessels studies imply that ¹⁸F-FDG and ¹⁸F-NaF target atherosclerotic plaques but different aspects and, probably, different phases of dynamic plaque biology. Lesional ¹⁸F-FDG uptake likely precedes uptake of ¹⁸F-NaF as an indicator of active calcification.²⁴⁴ Vascular uptake of either radiotracer may reflect potential plaque instability—one as a consequence of high inflammation and the other as a consequence of plaque rupture and active mineral deposition.

Future Developments of PET/CT for Inflammation and Infections of the Cardiovascular System

The clinical utility of metabolic imaging of the cardiovascular system with PET and, especially with hybrid PET/CT, has considerably grown in recent years. Initially confined to the identification and characterization of myocardial viability, metabolic imaging now includes assessments of inflammatory processes of the heart and the vasculature and monitoring of response to various therapies. Much of the knowledge gained through metabolic imaging with PET has remained in the domain of basic and clinical research; if applied to human cardiovascular disease, it will likely lead to important insights into mechanisms of human cardiovascular disease and meet specific diagnostic needs and challenges. Judging from recent developments in PET radionuclide imaging, much of the current and future growth will be driven by progress in cancer research and through translation of basic science findings to human cardiovascular disorders. Moreover, the ever-increasing availability and use of multimodality imaging will likely accelerate further current developments in PET radionuclide imaging.

KLG received internal funding and is salaried solely from the Weatherhead PET Center for Preventing and Reversing Atherosclerosis. KLG is the 510(k) applicant for HeartSee, a software package for quantifying absolute flow using cardiac positron emission tomography. All royalties will go to a University of Texas scholarship fund. The University of Texas has a commercial, nonexclusive agreement with Positron Corporation to distribute and market HeartSee in exchange for royalties. However, KLG retains the ability to distribute cost-free versions to selected collaborators for research. Additionally, KLG has financial research support from Volcano Corporation (maker of FFR pressure wires) for the multicenter study DEFINE_FLOW, Clinical Trials number NCT02328820.

HS has no financial relationships to disclose.

JN has no financial relationships to disclose.