Assessment of myocardial perfusion and function with PET and PET/CT

INTRODUCTION

Over the past decade, there has been a growing interest in cardiac imaging with Positron Emission Tomography (PET). However, PET has been used for more than 35 years as a powerful tool to study cardiac physiology. PET started as an investigative tool used at select academic medical centers equipped with cyclotrons, to probe physiologic processes such as myocardial perfusion, metabolism, neuronal innervation, and receptor function. At the outset, myocardial perfusion imaging (MPI) with PET was primarily used in research applications, or as an adjunct to ¹⁸F-fluoro deoxy glucose (FDG) imaging for viability assessment, or to guide clinical management in high-risk patients. Over the past 7-8 years, we have witnessed a paradigm shift in the use of MPI with PET. It is now being increasingly used for routine clinical evaluation of patients with known or suspected coronary artery disease (CAD). Also, it is being used not only at large academic institutions, but also at community hospitals and large private practice groups. There are several factors contributing to this shift in the use of PET MPI, including the exponential growth and availability of combined PET + computed tomography (CT) systems which were driven primarily by oncology applications, FDA approval of an easily available generator produced radiotracer, ⁸²Rubidium (Rb), changes in reimbursement, and the increasing clinical evidence supporting the value of PET/CT MPI.

There are several excellent review articles that detail the clinical applications, PET radiotracers,¹ quantitative PET,² viability assessment,³ and utility of hybrid PET MPI applications.^3-6 The focus of this article is to discuss the evolution of PET MPI over the course of the years to highlight some of the major achievements in PET that have culminated in the present day applications of PET MPI.

THE EVOLUTION OF PET AND PET/CT MPI

Technical Developments

Scanners

The initial concept of tomographic emission imaging was introduced by Kuhl and Edwards in the late 1950’s and the early 1960’s. By the mid-1960’s these investigators had also developed an early tomographic device using monoenergetic radionuclides as sources of gamma photons. Soon, thereafter prototype PET devices were developed at the University of Pennsylvania, Washington University (Michael Ter-Pogossian and Michael Phelps), and Massachusetts General Hospital (Brownell and Sweet).^7-10 The development of devices continued and PET scanning was introduced to the medical community in the 1970’s,¹¹ as a novel technique to study pathophysiological processes. The scanners evolved in the 1980’s to more commercial models, with improved image resolution, resulting in the clinical use of cardiac PET. The discovery of the scintillators, bismuth germinate (BGO) (late 1960’s to 70’s),¹² and subsequently lutetium oxyorthosilicate (LSO) (1990’s) led to further improvements in original image quality. The next major breakthrough in technology was the concept of PET/CT (in the early 1990’s). A prototype PET/CT scanner was developed (1998) and commercial scanners became available in 2000 (Nutt, Beyer and Townsend). These systems were based on a single-slice CT scanner and enabled faster imaging with greater anatomic information with a single slice CT. This was followed soon thereafter by hybrid devices with enhanced multi-detector computed tomography (MDCT) scanners capable of calcium scoring and CT coronary angiography. Finally, the enormous growth in cardiac PET has fueled the development of a dedicated small foot print PET scanner (Positron corporation, attrius PET scanner) utilizing with radionuclide-based attenuation correction specifically for cardiac applications. Also, although proposed several decades ago,^13,14 only recently have some of the PET systems incorporated annihilation-photon time-of-flight (TOF) information during image acquisition; the value of TOF imaging for MPI is unclear at this time.

Simultaneously, several software enhancements have become available. Software was developed for detecting and correcting for misregistration of the transmission and emission images.¹⁵ Motion frozen imaging, a concept introduced by Slomka et al¹⁶ for SPECT is currently being adapted to PET imaging. Also, novel resolution recovery techniques to improve contrast and contrast noise ratio are under development.¹⁷ Finally, software for quantitative PET and myocardial blood flow (MBF) assessment is rapidly evolving^18-21 and will soon be commercialized for clinical use.

Radiopharmaceuticals

Positron radiopharmaceuticals have been used since the late 1950’s.^22,23 In the 1960’s, Ter-Pogossian and Powers measured blood flow in the brain using ¹⁵O water.²⁴ In the mid-1960’s, quantitative estimation of coronary blood flow was performed with ⁸⁴Rb, a positron emitter, using a coincidence counting system.^25,26 With the development of biomedical cyclotrons,^{27 15}O water and later ¹³N-ammonia became available for research use in the mid-1970’s. ¹⁵O water was initially used by Parker et al²⁸ in 1978 to quantify regional myocardial blood flow (MBF) and regional fractional extraction of oxygen. While myocardial perfusion can be assessed using ¹⁵O water, ¹³N-ammonia, ⁸²Rb, and the new ¹⁸F-BMS compound, currently, ⁸²Rb and ¹³N-ammonia are the only two FDA approved radiotracers.

The initial studies of ¹³N-ammonia MPI were reported from Dr. Schelbert’s²⁹ and from Dr. Gould’s laboratories. Gould et al. demonstrated in intact dogs, that using ¹³N-ammonia, 47% diameter coronary stenosis can be detected. ¹³N-ammonia is a cylotron-produced radiotracer (half-life of 9.96 minutes) that was approved by the FDA for assessment of myocardial perfusion in 2000. ¹³N-ammonia enters the myocyte (passive diffusion or active transport) and is rapidly metabolized to glutamine ¹³N and retained in the myocyte or can diffuse back into the blood pool.^30,31 Gated ¹³N-ammonia imaging can provide accurate assessments of both regional and global cardiac function.³² However, the use of this imaging agent is limited to centers with a cyclotron and also it is also not well suited for peak stress gated imaging, due to the 3-4 minutes time interval between radiotracer injection and blood pool clearance.

The initial studies with ⁸²Rb MPI were reported by Gould et al.^{33-35 82}Rb is a generator produced radiotracer (half-life 76 seconds) that was approved by the FDA for assessment of myocardial perfusion in 1989. It is the most widely used radiotracer for clinical PET MPI.¹ The main advantages of ⁸²Rb are its availability at sites without a cyclotron, rapid imaging protocols, and the ability to obtain peak stress gated imaging. ⁸²Rb has a relatively long positron range (the distance the positron travels before colliding with an electron to release two 511 keV photons), which contributes to its somewhat worse spatial resolution in comparison to ¹⁸F or ¹³N. Also, due to the reduced first-pass extraction of ⁸²Rb, its myocardial uptake is not linear throughout the flow range, i.e., it is characterized by a plateau of uptake at high flow values, resulting in a potential underestimation of flow at maximal hyperemia.³⁶

Imaging protocols

Myocardial perfusion PET began as a research tool for quantitative blood flow assessments using dynamic image acquisition (multi frame imaging). The image frames were summed for relative perfusion assessment. Some of the initial PET studies of function included, assessments of lung water content using ¹⁵O-water,³⁷ as well as first pass imaging for computing stroke volume, lung water content (a measure that correlated with restrictive filling pattern on echocardiography), and ejection fraction.³⁸ Gated PET imaging necessitated a separate acquisition with extra time, radiation burden, and imposed greater demands on the processors for image processing and storage. Also, with PET, gated MPI is not necessary for distinguishing attenuation artifacts from scar (due to accurate attenuation correction). Hence, in order to expedite image acquisition and analysis, PET MPI was performed as a nongated study, and echocardiography or radionuclide angiography were used for the assessment of left ventricular ejection fraction (LVEF) and wall motion, when needed. This led to two issues, firstly, LVEF assessment at a temporally different time from the metabolic assessment and next, anatomic co-registration of the perfusion and metabolic image with the wall motion abnormality was difficult.³⁹ Miller et al,³⁹ described the use of inhaled ¹⁵O carbon monoxide labeled red blood cell blood pool imaging (used to delineate the vascular pool to correct the ¹⁵O-water myocardial perfusion images), and demonstrated that simultaneous assessment of perfusion and function is feasible with ¹⁵O-water MPI.

Presently, simultaneous assessment of perfusion and function has become routine practice. The latest PET scanners and computers are very powerful and allow for list mode imaging, i.e., accrual of a list of coincidence-photon detected events along with a recorded ECG and time synchronization signal. List mode acquisitions have greatly enhanced the utility of PET MPI, allowing a comprehensive simultaneous assessment of perfusion and LVEF, and anatomy (with CT when performed) in a single sequence (Figure 1). The current clinically used image acquisition protocols have been well described in the ASNC PET guidelines.⁴⁰ All the current clinical imaging protocols are sequential with stress following the rest MPI, with time in between the tests for myocardial radiotracer activity to be decay or diminish by biological pathways. Simultaneous research protocols are being developed wherein rest/stress imaging is done in a single scan to increase throughput, reduce radiation dose, and improve co-registration of the rest and stress emission images.⁴¹

Figure 1.

Demonstrates the comprehensive imaging data (relative perfusion, rest and stress gated data, absolute perfusion, calcium score, CT coronary angiogram, and fused PET and CT coronary angiogram images) obtained by a list mode acquisition in a case of combined PET and CT coronary angiography.

Attenuation correction

In contrast to SPECT, PET images are always corrected for attenuation. Conventional dedicated PET scanners use radionuclide transmission imaging, whereas, hybrid scanners use CT transmission imaging for measurement and correction of soft-tissue attenuation. Most hybrid scanners do not offer the option of radionuclide transmission imaging. The most commonly used CT protocol is a non-gated low-dose CT in shallow tidal breathing, although end-expiration, end-inspiration, and slow CT protocols have been tried.⁴² The accuracy of low-dose CT (10 mA) appears to be comparable to Ge transmission imaging for attenuation correction.⁴³ However, inaccurate registration of transmission (radionuclide or CT) and emission images from respiratory or patient motion can lead to artifactual defects in 20-40% of cases.^15,44-49 Most commercial PET/CT systems now include software tools providing reasonable correction for transmission-emission misalignments from breathing differences or patient movement between the transmission and emission images. However, patient motion during either the transmission or emission imaging limits image quality and cannot be easily corrected using software.⁵⁰

Stress protocols

Stress testing with PET MPI has been used since the early 1970’s. Exercise stress with treadmill or bicycle, pharmacological stress with adenosine, dipyridamole, dobutamine have all been used. Supine bicycle testing was used with PET MPI,⁵¹ but did not gain popularity due to patient motion. Treadmill exercise testing^52,53 with ¹³N-ammonia and ⁸²Rb were used since the early 1980’s. Exercise stress is somewhat cumbersome with ⁸²Rb due to its short half-life, and breathing motion from immediate post exercise imaging. Also, with ¹³N-ammonia, close coordination with the cyclotron is necessary making it less appealing for routine clinical use in high volume centers. Lastly, the routine use of exercise stress with PET tracers, and close patient contact may significantly increase the radiation exposure to the staff.⁵⁴ Therefore, vasodilator stress, dipyridamole, is the most widely used stressor for PET MPI. Handgrip exercise was initially used in conjunction with dipyridamole.^55-57 Subsequently, this practice was abandoned as studies showed that hand grip was associated with higher coronary vascular resistance and reduced peak stress MBF.⁵⁸

Cold pressor testing with PET MPI has been used for the noninvasive assessment of coronary endothelial function and formed the basis for several seminal investigations. Typical protocols include immersion of the foot/hand in ice-cold water (2°C) for 2 minutes, with injection of radiotracer at 1 minute and continued immersion for an additional 1 minute.⁵⁹ Unlike vasodilator stressors, which produce a predominantly endothelium independent coronary vasodilator response, cold pressor testing increases MBF indirectly by sympathetic activation, release of norepinephrine from cardiac sympathetic-nerve terminals, which leads to vasodilation via an endothelium-dependent mechanism mediated by NO.^60-62

Adenosine has been used with PET since it became available in the early 1990’s. It produces maximal hyperemia and is short-lived, but it necessitates two intravenous lines for use with ⁸²Rb, and is hence cumbersome to use. More recently, Regadenoson, an adenosine A2a receptor agonist has been approved for SPECT MPI.⁶³ Its main advantages include bolus administration with standardized dosing regardless of body weight (single bolus administration of 0.4 mg from a prefilled syringe) and better tolerability. Regadenoson when used in conjunction with ⁸²Rb PET enables ultra-short imaging protocols with complete stress and rest MPI completed in ~17 minutes (Figure 2). The diagnostic accuracy and flow quantitation of PET MPI with Regadenoson are currently being studied.

Figure 2.

A sample protocol for clinical cardiac PET/CT imaging with ⁸²Rb takes ~25 minutes. The use of Regadenoson stress makes the protocol ultrashort with completion of rest and stress imaging in ~17 minutes. CT, CT scan for attenuation correction; CTCA, CT coronary angiography; ⁸²Rb, ⁸²Rubidium.

Quantification of myocardial blood flow

Myocardial blood flow has been quantified in humans for over 40 years.² Initially, measurements were based on thermodilution or Doppler techniques during angiography. Radionuclide assessments of regional blood volume,²² fractional extraction, and blood flow measurements were reported as early as the 1950’s using ⁸⁵Krypton,^{64 84}Rb,^65,66 and ⁴²potassium.⁶⁷ Over the next two decades,^28,68 these techniques evolved to the use of labeled microspheres (albumin microspheres labeled with ⁶⁸Ga⁶⁹ and ¹¹C-labeled microspheres⁷⁰) to measure MBF. These investigations led to further development and validation of techniques to quantify MBF with PET.^71-75

Currently, ¹³N-ammonia, ¹⁵O water, and ⁸²Rb are used for noninvasive MBF quantitation. These tracers have been validated in animal models using microsphere techniques, and their reproducibility ascertained.^76-81 The short half-life of these tracers enable repeat measurement of blood flow for research applications. ¹⁵O-labeled water is an ideal tracer for flow quantitation as its uptake is linearly related to flow and a single compartment model is used for flow quantitation.² A 3-compartment model has been developed for blood flow quantitation with ¹³N-ammonia,⁸² while, a two compartment model is used for ⁸²Rb.^2,83 The advantage of ⁸²Rb is that it is widely available at sites without cyclotrons, enabling widespread clinical applicability of absolute blood flow in large number of patients. However, as described earlier, the lower extraction fraction of ⁸²Rb may limit its value for flow quantitation at high flow rates.

DIAGNOSTIC ACCURACY

The coronary circulation is comprised of the large epicardial vessels and the smaller coronary microvascular or resistance vessels. Some of the early investigations in ischemic heart disease with PET, characterized the metabolic changes in the myocardium induced by ischemia.⁸⁴ Subsequent studies focused on evaluation of relative and absolute myocardial perfusion assessments. While absolute PET MPI blood flow assessment is helpful to identify coronary microvascular flow abnormalities from a variety of cardiovascular diseases, relative PET MPI blood flow assessment is useful to diagnose epicardial CAD and more widely used clinically.

Evaluation of Coronary Microvascular Dysfunction

Direct imaging of the coronary microvasculature in humans is presently not feasible due to resolution limits of the existing imaging techniques.⁸⁵ Abnormalities in endothelial and coronary microvascular function are among the earliest manifestations of atherosclerotic vascular damage and precede overt atherosclerosis. Coronary microvascular function was conventionally assessed by studying MBF changes detected by thermodilution or intracoronary Doppler flow wires, at rest and during intracoronary provocation testing [adenosine (endothelium-independent flow), acetyl choline, papaverine (endothelium-dependent flow)]. The invasive nature of this evaluation limited the diagnosis to a select group of symptomatic individuals requiring invasive coronary angiography. Absolute PET (vasodilator and cold pressor testing) is a direct and precise technique to study microvascular function that is increasingly being used in lieu of intracoronary provocation testing. With the advent of CT coronary angiography, CT can be used to exclude obstructive epicardial CAD and absolute PET MPI to diagnose coronary microvascular flow abnormalities, allowing for non-invasive evaluation of early microvascular dysfunction.

Absolute PET MPI is a powerful tool to evaluate the integrated effects of risk factors on the health of the microvasculature.⁸⁵ The magnitude of microvascular dysfunction is directly related to the individual and combined risk factor burden.⁸⁶ Indeed, individuals with diabetes,^60,87 dyslipidemia,^88,89 hypertension,⁹⁰ and smoking,⁹¹ manifest abnormalities in coronary micro-vascular function even in the absence of underlying obstructive epicardial CAD.⁸⁵ Microvascular dysfunction by PET has been described by Dr. Camici’s laboratory in primary and secondary left ventricular hypertrophies, hypertrophic⁹² and dilated cardiomyopathies,⁹³ and infiltrative heart diseases.⁸⁵ Also, Zeiher et al⁹⁴ demonstrated that endothelial and microvascular dysfunction (from risk factors, and myocardial diseases) contributes to the pathogenesis of myocardial ischemia and predisposes individuals with mild atherosclerosis to exercise-induced ischemia. For a detailed review of coronary microvascular function, the readers are referred to a comprehensive review on this topic.⁸⁵

Diagnosis of Obstructive CAD

In 1974, Gould et al⁹⁵ proposed the use of coronary flow reserve (CFR) as a physiologic measure of coronary artery stenosis severity. Subsequent reports⁵⁶ extended this concept to the use of PET imaging to assess the functional significance or coronary artery stenoses. Relative coronary perfusion reserve decreased linearly with ~50% diameter stenosis or ~70% area stenosis of the coronary arteries (Figure 3). In 1975, ⁸¹Rb MPI was used by Berman et al⁹⁶ to diagnose myocardial ischemia, using planar imaging with a scintillation camera.

Figure 3.

Relative PET perfusion reserve as a tool to assess physiological significance of coronary stenoses. Reproduced with permission from Goldstein et al.⁵⁶

Studies of PET and PET/CT MPI

Several studies have since evaluated the diagnostic accuracy of relative PET MPI in the diagnosis of obstructive CAD (Table 1).⁹⁷ It is important to note that most of the available data has been obtained with dedicated PET scanners with radionuclide attenuation correction, and using vasodilator stress rather than exercise stress. The average weighted sensitivity of PET MPI for detecting at least one coronary artery with >50% stenosis is 90%, whereas the average specificity is 89%. The corresponding average positive and negative predictive values for the diagnosis of obstructive CAD are 94% and 73%, respectively, and the overall diagnostic accuracy is 90%. The sensitivity of PET for detecting obstructing CAD appears to be equally high in patients with single and multi-vessel (≥2 vessels) disease (92% and 95%, respectively) as well as in overweight and obese individuals (mean BMI>30 kg · m⁻²), and in men and women.⁹⁸

Table 1.

Summary of published literature regarding diagnostic accuracy of PET MPI

Author	Stress agent	Patients	Women	Prior CAD	PET Radiotracer	Sens	Spec	PPV	NPV	Accuracy
Sampson*98	Dipyridamole, adenosine,  dobutamine	102	0.42	0	82Rb	0.93	0.83	0.80	0.94	0.87
Bateman103	Dipyridamole	112	0.46	0.25	82Rb	0.87	0.93	0.95	0.81	0.89
Marwick130	Dipyridamole + Hand grip	74	0.19	0.49	82Rb	0.90	1	1	0.36	0.91
Grover-McKay131	Dipyridamole + Hand grip	31	0.01	0.13	82Rb	1	0.73	0.80	1	0.87
Stewart57	Dipyridamole	81	0.36	0.42	82Rb	0.83	0.86	0.94	0.64	0.84
Go127	Dipyridamole	202	NR	0.47	82Rb	0.93	78	0.93	0.80	0.90
Demer34	Dipyridamole	193	0.26	0.34	82Rb/13NH3	83	0.95	0.98	0.60	0.85
Tamaki99	Supine Bike	51	NR	0.75	13NH3	0.98	1	1	0.75	0.98
Gould33	Dipyridamole + Hand grip	31	NR	NR	82Rb/13NH3	0.95	1	1	0.90	0.97
Weighted summary		877	0.29	0.35		0.90	0.89	0.94	0.73	0.90

Sens, Sensitivity; Spec, specificity; PPV, positive predictive value; NPV, negative predictive value.

^*Study using PET/CT (where CT is used for attenuation correction only). Adapted with permission from Di Carli MF et al.⁶

Comparative studies of PET vs. SPECT

Several studies have compared the diagnostic accuracy of PET vs. SPECT MPI in separate groups of patients.^55,57,99 Go and colleagues compared ⁸²Rb PET and ²⁰¹Thallium SPECT in 202 patients and demonstrated a higher sensitivity with PET than with SPECT (93% vs. 76%, respectively), without significant changes in specificity (78% vs. 80%, respectively). On the other hand Stewart et al compared ⁸²Rb PET and ²⁰¹Thallium SPECT in 81 patients and observed a higher specificity for PET than for SPECT (83% vs. 53%, respectively), without significant differences in sensitivity (86% vs. 84%, respectively). The differences between these two studies are likely to be attributable to patient selection resulting in differences in pre-scan likelihood of CAD. More recently, Bateman et al¹⁰³ compared ⁸²Rb PET and ^99mTc Sestamibi SPECT in two matched patient cohorts undergoing clinically indicated pharmacologic-stress perfusion imaging using contemporary technology for both SPECT and PET. Overall diagnostic accuracy was higher for PET than for SPECT (89% vs. 79% with a 70% angiographic threshold). The sensitivity of PET MPI to detect obstructive CAD was similar to SPECT MPI, however, the specificity of MPI to exclude obstructive CAD appears was higher for PET compared to SPECT. To directly compare the diagnostic value of PET compared to SPECT MPI, the same patient cohort should undergo both SPECT and PET MPI, both with attenuation correction, especially in obese patients. But such studies are not available yet.

Evaluation of Diffuse CAD

Apex to base gradient

Gould et al¹⁰⁰ described an apex to base perfusion gradient in subjects with mild diffuse angiographic CAD, using relative ¹³N-ammonia MPI. This gradient was not seen in volunteers and patients with severe obstructive CAD demonstrated discrete perfusion defects without the apex to base gradient (Figure 4). This apex to base gradient was also demonstrated by quantitative ¹³N-ammonia MPI.¹⁰¹ The findings of these studies suggest that the apex to base perfusion gradient may indicate underlying diffuse CAD.

Figure 4.

Schematic demonstrating a discrete defect from a segmental coronary stenosis (top), in comparison to the gradual apex to base gradient that may be evident in cases of diffuse CAD. Reproduced with permission from Gould et al.¹⁰⁰

Evaluation of Multi-Vessel CAD

Absolute perfusion

The diagnosis of multi-vessel or left main CAD with balanced flow reduction remains a challenge with relative perfusion imaging. In a recent study, nearly 13% of patients with angiographically significant left main stenosis had normal SPECT imaging.¹⁰² As with SPECT, if relative perfusion imaging is used, typically the coronary territory supplied by the most severe stenosis is uncovered by PET and the extent of disease is underestimated in almost 30-50% of patients with multi-vessel obstructive CAD.^103-105

PET measurements of MBF (in mL · min⁻¹ · g⁻¹ of myocardium) may also help overcome the limitations of relative perfusion assessments with PET to uncover the presence of multi-vessel CAD. Myocardial blood flow and coronary vasodilator reserve (the ratio between peak hyperemic and rest MBF) are inversely and nonlinearly related to stenosis severity. Quantitative estimates of MBF by PET allows for a better definition of the extent of obstructive CAD.^72,106 If absolute perfusion assessments are used, more global reductions in flow (i.e., balanced ischemia) could be identified (Figure 5).¹⁰⁵ In a study of 23 patients,¹⁰⁵ Rb-82 net retention was quantified as an estimation of absolute perfusion at rest and with dipyridamole stress by use of dynamic PET MPI. Defect sizes were larger with absolute MPI in patients with 3-vessel disease (Table 2). While most of the experience with quantitative perfusion imaging is with ¹³N-ammonia, quantitative approaches with ⁸²Rb based on factor least squares cardiac factor analysis. Correction for ambiguous solutions in factor analysis using a penalized least squares objective¹⁰⁷ have been developed and validated.¹⁰⁸ Also, software to quantify absolute MBF is being incorporated into commercial packages and should soon be available for routine clinical use.

Figure 5.

Relative myocardial perfusion images demonstrate inferior and inferoseptal ischemia, while absolute myocardial perfusion is globally reduced, suggesting balanced ischemia.

Table 2.

Percent abnormal myocardium in standard method vs. absolute perfusion reserve

	Standard method	Absolute method
Group	Stress	Stress	Stress/rest absolute perfusion ratio	Stress-rest absolute perfusion reserve
3-vessel disease  (% abnormal LV)	44% ± 18%	41% ± 29%	55% ± 25%	69% ± 24%†
1-vessel disease  (% abnormal LV)	18% ± 17%	4% ± 7%*	9% ± 13%	10% ± 12%

^*P = .01

^†P = .008 compared with corresponding defect size by standard method. Table reproduced with permission from Parkash et al.¹⁰⁵

Left ventricular function

ECG-gating of MPI allows for the assessment of left ventricular volumes and ejection fraction which have proven diagnostic and prognostic utility.¹⁰⁹ Gated PET imaging provides a unique opportunity to assess LVEF at rest and during peak stress (as opposed to post-stress with gated SPECT). Recent data suggest that in normal subjects, LVEF increases during peak vasodilator stress.^104,110 In the presence of CAD, however, changes in LVEF (from baseline to peak stress) are inversely related to the magnitude of perfusion abnormalities during stress (reflecting myocardium at risk) (Figure 6A) and the extent of angiographic CAD (Figure 6B). An LVEF reserve (stress minus rest LVEF) of ≥5% has a negative predictive value of 97% to exclude the presence of three-vessel and/or left main CAD.¹⁰⁴

Figure 6.

Bar graphs demonstrating the relationship between LVEF reserve (peak stress minus rest LVEF) and the magnitude of stress-induced perfusion abnormalities (A) and the extent of angiographic CAD (>70% stenosis) (B). Reproduced with permission from Dorbala et al.¹⁰⁴

Finally, CT coronary angiography and calcium scoring are important adjuncts for identifying more extensive disease than is evident on relative PET MPI.

RISK STRATIFICATION

Semi-Quantitative Relative Perfusion

Given its recent adoption in the routine clinical imaging services, studies documenting the prognostic value of PET MPI in predicting patient’s outcomes are few, but beginning to emerge (Table 3). Overall, all the studies demonstrated an excellent outcome with normal MPI; increases in the extent and severity of stress perfusion defects translated into proportional increases in predicted mortality. Of note, the patient cohort in the study by Marwick et al¹¹¹ was comprised of high risk patients, with 84% of patients having had prior coronary angiography, and reflecting the utilization of PET during the early 1990’s. Yoshinaga et al¹¹² studied a more contemporary cohort of patients, but their study was limited by the smaller patient cohort, and limited number of cardiac events. Subsequent studies^113,114 confirmed the prognostic value of ⁸²Rb PET MPI in larger patient cohorts. The incremental prognostic value of ⁸²Rb MPI over demographic variables, stress variables, and rest LVEF (Figure 7A and B) was also demonstrated.¹¹³

Table 3.

Summary of published literature regarding prognostic value of PET MPI

First author	Year	Stress agent	Tracer	Patients N	Events N	Event type	Prior CAD %	Percent normal scans (%)	Event/ yr in normal MPI (%)	Event/yr in abnormal MPI
Marwick111	1997	Dipyridamole	82Rb	685	81	Cardiac death	Prior MI 48%, Prior Revas 37%	24	0.9	Mild 2.6%, moderate 5.1%, severe 5.1%
Yoshinaga112	2006	Dipyridamole	82Rb	367	17	Cardiac death  or MI	40.3%	70.5	0.4	2.3%, mild, 7.0% moderate- severe
Lertsburapa114	2009	Dipyridamole	82Rb	1,441	132	All cause  mortality	53.6%	64.8	2.4	4.1% mild, 6.9% moderate- severe
Dorbala113	2009	Dipyridamole Adenosine	82Rb	1,432	140	Cardiac death  or MI	30.6%	54	0.7	5.5% mild, 5% moderate, 11% severe
Herzog118	2009	Adenosine	13N ammonia	256	29	Cardiac death	66%	45	0.5	3.1%
Chow132	2009	Exercise   and   dobutamine	82Rb	124	16	Cardiac   death, MI   revasculari-   zation	MI 40%, PCI 29%, CABG 15%	37	1.7	13%

⁸²Rb, ⁸²Rubidium; MI, non-fatal myocardial infarction

Figure 7.

(A) Risk adjusted survival curves demonstrating event free survival based on percent myocardium abnormal. Survival was excellent in patients with normal ⁸²Rb MPI (0% abnormal), and progressively worse survival was noted for patients with mild (1-10% abnormal), moderate (11-20% abnormal), or severely abnormal (≥20% abnormal) scans. (B) Risk adjusted survival curves demonstrating worse event free survival for patients with LVEF reserve <0%. Reproduced with permission from Dorbala et al.¹¹³

Left Ventricular Function

Regional wall motion and LVEF obtained using echocardiography or radionuclide angiography have been used in conjunction with PET viability assessment for the past several decades to improve predictive accuracy for identification of myocardial segments that may improve function following revascularization. For instance, Yoshida et al demonstrated that large infarct size and absence of viable myocardium as assessed by ⁸²Rb, as well as low LVEF determine adverse outcomes in patients with CAD.¹¹⁵ However, since LVEF was not routinely assessed with PET MPI, unlike SPECT,¹⁰⁹ data about the prognostic value of LVEF in conjunction with MPI are just beginning to emerge. The study by Lertsburapa et al,¹¹⁴ was the first to demonstrate that the addition of stress LVEF to clinical and perfusion variables significantly enhanced the value of the ⁸²Rb MPI to predict all cause mortality. Also, in the study by Dorbala et al, in 985 patients with peak stress gated data, the annualized rates of cardiac events (2.1% vs. 5.3%, P<.001) and all-cause death (4.3% vs. 9.2%, P<.001) were higher in patients with an LVEF reserve <0% compared with those with an LVEF reserve 0%. Multivariable risk adjusted analysis demonstrated independent and incremental prognostic value of LVEF reserve compared to clinical variables, rest LVEF, and MPI.

Absolute Perfusion

A few studies have shown that coronary vasodilator reserve assessed by rest and vasodilator PET can assess progression of CAD and stratify risk of future cardiovascular outcomes. Gould et al¹¹⁶ demonstrated that a modest regression of coronary artery stenoses after risk factor modification is associated with decreased size and severity of perfusion abnormalities on rest-dipyridamole PET. These results suggest that progression or regression of CAD in response to therapy can be followed non-invasively by absolute PET MPI. Also, in 51 subjects with hypertrophic cardiomyopathy followed for 8 years, those with the lowest tertile of ¹³N ammonia dipyridamole MBF at baseline had significantly worse long-term clinical outcomes.⁹² Similarly, the degree of coronary microvascular dysfunction has been shown to be an independent predictor of death and progressive heart failure in patients with dilated cardiomyopathy.⁹³ Lastly, Herzog and colleagues¹¹⁷ reported that coronary flow reserve (CFR) could add incremental prognostic value to semi quantitative MPI in predicting hard and soft endpoints in a study of 256 patients who underwent ¹³N-ammonia PET. In patients with normal perfusion, abnormal CFR was independently associated with a higher annual event rate over 3 years compared with normal CFR for cardiac events (1.4% vs. 6.3%; P<.05) and cardiac death (0.5% vs. 3.1%; P<.05).¹¹⁸ Although this study was limited by the small event rate, it is the first study to report the prognostic value of absolute PET MPI in patients with known or suspected CAD.

All of the aforementioned studies demonstrated the prognostic value of vasodilator perfusion reserve, a measure of predominantly endothelium independent mechanisms of MBF. The prognostic value of endothelial-dependent flow abnormalities, were studied by Schindler et al⁵⁹ using rest and cold pressor stress MPI in 72 subjects without epicardial CAD. This study demonstrated that impaired or decreased MBF response to cold pressor test may stratify risk of future cardiovascular events. However, on multivariable analysis, cardiac events were independently related to hypercholesterolemia, hypertension, smoking, increases in body mass index, but not to impaired MBF response to cold pressor test, suggesting that impaired CFR may be a marker of risk reflecting the effects of atherogenic risk factor burden.

COMBINED PET AND CT APPLICATIONS

A detailed discussion of combined PET and CT applications is beyond the scope of this review. The ability to image myocardial perfusion along with anatomic atherosclerosis (calcium scoring or CT coronary angiogram) is a major advantage of the hybrid PET/CT scanners compared to dedicated PET scanners. While these offer several interesting clinical applications, the costs of combined imaging and radiation burden need to be considered. Also, the challenges of interpreting the non-cardiac ancillary findings on the low-dose CT scan are not trivial. At this time there is no consensus on whether these non-diagnostic CT scans should be routinely reviewed for ancillary findings.

Calcium Score

Recent data suggest that quantification of coronary artery calcium (CAC) score at the time of stress MPI with PET/CT adds incremental prognostic information to MPI.¹¹⁹ In a consecutive series of 621 patients undergoing stress PET imaging and CAC scoring in the same clinical setting, there was an increase in events (death and myocardial infarction) with increasing levels of CAC score for any given degree of perfusion abnormality. Indeed, the annualized event rate in patients with normal MPI and no CAC was substantially lower than among those with normal MPI and a CAC ≥ 1,000 (Figure 8). The results of this and other studies suggest the utility of the knowledge of anatomic atherosclerosis in conjunction with physiologic data while performing PET MPI.¹²⁰ Some investigators propose routine calcium scoring with PET/CT MPI in patients without known CAD, but, the effectiveness of this approach needs to be further evaluated.

Figure 8.

Risk adjusted survival curves in patients with nonischemic relative PET scans demonstrating worse event free survival in patients with a calcium score (CAC) of ≥1000 compared to CAC <1000. Reproduced with permission from Schenker et al.¹¹⁹

CT Coronary Angiography

Advances in cardiac CT have accelerated in the past few years. With the use of multi-detector CT systems, coronary CT angiography has been shown to have high negative predictive value for ruling out CAD. However, cardiac CT is limited by its modest positive predictive value.¹²¹ With the use of CT dose-reducing techniques (step and shoot imaging (prospective triggering), ECG dose modulation, low KVp imaging etc.) hybrid PET and CT angiography imaging could be performed (on the same or separate scanners) with a relatively low radiation exposure.^122,123 Hybrid PET and CT coronary angiography may be useful in identifying more extensive CAD, balanced ischemia, microvascular dysfunction, coronary anomalies, structural abnormalities of the coronary arteries and their functional consequences and at clarifying equivocal findings on either the CTA or the PET MPI. Hybrid imaging may also be useful in defining the vessel responsible for a perfusion defect seen on PET MPI (Figure 9). Further research is underway focusing on identifying the patient population that would benefit from combined anatomical and perfusion imaging.¹²⁴ Combined PET/CT imaging offers great potential for molecular imaging applications with targeted radiotracers, with the CT image serving as the anatomic roadmap for localization of the radiotracer uptake.

Figure 9.

A hybrid PET CTA examination in a patient with significant LAD ischemia. CT coronary angiogram image is overlaid on volume rendered stress ⁸²Rb perfusion image. CT coronary angiogram was performed to evaluate the patency/caliber of the distal LAD (beyond a known chronic total occlusion of the mid LAD) for consideration of coronary artery bypass surgery.

RADIATION EXPOSURE WITH PET AND PET/CT

Radiation exposure from cardiovascular imaging has become a major public health concern given the increase in the utilization of diagnostic testing. The estimated effective dose estimates for various cardiac PET/CT procedures are listed in the Table 4.^125-127 Multiple steps should be taken to reduce radiation exposure in PET MPI including ensuring the appropriateness of the study, reducing the amount of injected dose, using low-dose CT for attenuation correction, using 3D PET imaging with a smaller dose of radioactivity, and possibly considering stress-only imaging.¹²⁸

Table 4.

Estimated effective radiation dose of common PET perfusion imaging studies

Test	Radiation dose (mSv)
82Rb (rest + stress)125	12.6–13.5
13N-ammonia (rest + stress)125	2.0–2.3
15O-water (rest + stress)125	1.6–2.3
Scout/localizing CT133	0.04
CT attenuation correction133	0.35
Ca Scoring (Prospective  gating)134	1.5–3

FUTURE DIRECTIONS

New Perfusion Agents

It has been a long time since a new myocardial perfusion agent for PET imaging has been introduced. However, studies are ongoing with new novel agents. Recently, Nekolla et al¹²⁹ reported on a new ¹⁸F-labeled perfusion agent, ¹⁸F-BMS-747158-02 (¹⁸F-BMS) in a porcine transient ischemia model, with comparisons to myocardial uptake with ¹³N ammonia and radioactive microspheres. Compared with ¹³NH₃, ¹⁸F-BMS showed higher activity ratios between myocardium and blood (rest 2.5 vs. 4.1; stress 2.1 vs. 5.8), liver (rest 1.2 vs. 1.8; stress 0.7 vs. 2.0), and lungs (rest 2.5 vs. 4.2; stress 2.9 vs. 6.4). Regional MBF assessed with ¹⁸F-BMS PET showed excellent correlation (r = .88). In addition, ¹⁸F-BMS showed homogeneously high and stable cardiac uptake and its image quality was rated superior to that of the ¹³N ammonia images by blinded observers (Figure 10). In the isolated perfused rat heart, ¹⁸F-BMS demonstrated a high first-pass extraction (above 90%). These results are encouraging and are currently being tested in Phase 2 clinical studies. If successful, this new agent holds promise for significantly expanding the clinical use of PET MPI, by providing the option of unit dose PET radiotracers (akin to SPECT tracers).

Figure 10.

Myocardial perfusion images using F-18 BMS showing comparable or better image quality compared to ¹³N ammonia images. Reproduced with permission from Nekolla et al.¹²⁹

CONCLUSIONS

We have reviewed the evolution of PET and some of the major developments in myocardial perfusion and function imaging by PET. PET is a robust and mature technique to non-invasively study myocardial perfusion and function. The rapidly accumulating investigative evidence to support the utility of PET MPI, combined with the developments in scanners, software, and novel radiotracers, will lead to greater clinical use of cardiac PET imaging. The availability of absolute perfusion measurements and investigations into targeted molecular imaging will likely further enhance the clinical applications of PET MPI in the future.