Radionuclide myocardial perfusion imaging (MPI) currently

Transcription

1 Contemporary Reviews in Cardiovascular Medicine Integration of Quantitative Positron Emission Tomography Absolute Myocardial Blood Flow Measurements in the Clinical Management of Coronary Artery Disease Henry Gewirtz, MD; Vasken Dilsizian, MD Abstract In the >40 years since planar myocardial imaging with 43 K-potassium was introduced into clinical research and management of patients with coronary artery disease (CAD), diagnosis and treatment have undergone profound scientific and technological changes. One such innovation is the current state-of-the-art hardware and software for positron emission tomography myocardial perfusion imaging, which has advanced it from a strictly research-oriented modality to a clinically valuable tool. This review traces the evolving role of quantitative positron emission tomography measurements of myocardial blood flow in the evaluation and management of patients with CAD. It presents methodology, currently or soon to be available, that offers a paradigm shift in CAD management. Heretofore, radionuclide myocardial perfusion imaging has been primarily qualitative or at best semiquantitative in nature, assessing regional perfusion in relative terms. Thus, unlike so many facets of modern cardiovascular practice and CAD management, which depend, for example, on absolute values of key parameters such as arterial and left ventricular pressures, serum lipoprotein, and other biomarker levels, the absolute levels of rest and maximal myocardial blood flow have yet to be incorporated into routine clinical practice even in most positron emission tomography centers where the potential to do so exists. Accordingly, this review focuses on potential value added for improving clinical CAD practice by measuring the absolute level of rest and maximal myocardial blood flow. Physiological principles and imaging fundamentals necessary to understand how positron emission tomography makes robust, quantitative measurements of myocardial blood flow possible are highlighted. (Circulation. 2016;133: DOI: /CIRCULATIONAHA ) Key Words: coronary circulation coronary disease fractional flow reserve, myocardial microvessels myocardial perfusion imaging myocardium regional blood flow Radionuclide myocardial perfusion imaging (MPI) currently is dominated by single-photon emission computed tomography (SPECT), which uses 99m Tc-labeled tracers to map the relative distribution of myocardial blood flow (MBF) both at rest and with stress. An extensive literature has demonstrated the diagnostic accuracy of the methodology for both coronary artery disease (CAD) diagnosis and prognosis. 1 3 Furthermore, recent efforts have focus on technological approaches aimed at improving SPECT image quality. These include solidstate cadmium zinc telluride detectors and gamma cameras; optimized cardiac collimators; and hardware and software advances to correct for patient motion, photon attenuation, and scatter, to enhance energy resolution, and to improve image reconstruction while reducing patient radiation exposure. 1,4,5 Thus, a substantial effort has been and continues to be directed at improving a technology (ie, SPECT MPI) that is limited by its fundamental physics to an inherently qualitative, at best semiquantitative methodology. 6 Measurement of true tissue concentrations (nci/ml) of 99m Tc-labeled tracers or thallium with SPECT has not been proven possible to date and faces many technical barriers, 7 including the current state-of-the-art gamma camera technology, currently available software, and recent experimental work that suggests it may be possible in the future possibly with a better 99m Tc-labeled MBF tracer. 7,8 Positron Emission Tomography MBF: Physics In contrast to SPECT, the fundamental physics of positron emission tomography (PET) allow true quantification and tracer kinetic modeling to permit accurate measurements of absolute MBF. 13 N-ammonia and 15 O-water require an onsite cyclotron because of their very short half-lives (10 and 2 minutes, respectively). In contrast, 82 Rb (half-life, 75 seconds), a decay product of 82 Sr (half-life, 25 days), is eluted from a commercially available 82 Sr generator system, obviating the requirement for an onsite cyclotron. PET permits absolute quantification of tracer concentration in blood and myocardium principally because an annihilation event occurs when a tracer label positron collides with its antiparticle, a negative electron (negatron), with resulting production of two 511-KeV photons that travel at a 180 From Department of Medicine (Cardiology Division), Massachusetts General Hospital, Harvard Medical School, Boston (H.G.); and Department of Diagnostic Radiology and Nuclear Medicine, University of Maryland School of Medicine, Baltimore (V.D.). Correspondence to Vasken Dilsizian, MD, Department of Diagnostic Radiology and Nuclear Medicine, University of Maryland Medical Center, 22 S Greene St, Room N2W78, Baltimore, MD vdilsizian@umm.edu 2016 American Heart Association, Inc. Circulation is available at DOI: /CIRCULATIONAHA

2 Gewirtz and Dilsizian Quantitative PET MBF Measurement and CAD 2181 from one another. Accordingly, the PET camera is able to very closely localize where the annihilation event occurred by counting only those photons that nearly simultaneously strike its 180 opposing detectors. Because positron range in matter is proportional to its energy, the locus of the annihilation event differs from that of positron origin, on average by 1.0 to 2.0 mm, which imposes a physical limit on spatial resolution. A computed tomography (CT) scan is used to make an attenuation map of the chest to correct for soft-tissue attenuation of photons. Corrections for random coincidences, scatter, and dead time also must be made. The response time of the instrument is much faster than that of a cadmium zinc telluride detector SPECT device; thus, higher count rates are possible with PET, which favors improved statistical accuracy for tracer kinetic modeling, along with enhanced spatial resolution resulting from the higher energy and greater abundance of photons imaged. Furthermore, with the PET camera operating in the 3-dimensional (3D) mode, sensitivity is increased. Accordingly, although tracer dose must be reduced to prevent overwhelming the response capabilities of the camera, patient radiation exposure also is reduced, an increasingly important clinical consideration. 9 A typical rest/stress PET examination (3D mode) with 13 N-ammonia or 82 Rb imparts a nominal radiation dose of 2 msv compared with 14 msv for SPECT rest/ stress 99m Tc-MIBI. 10,11 Thus, even operating in 3D mode, which sacrifices some spatial resolution compared with the 2-dimensional mode, PET technology provides substantially better image resolution, counting statistics, and, most important, true absolute quantization of tracer concentration (nci/ml) in blood and myocardium compared with SPECT. These features enable robust dynamic data acquisition and analysis with well-validated tracer kinetic models, thereby making accurate absolute measurements of MBF 12 possible (Figure 1A and 1B) Tracers Used to Measure Absolute MBF, Present and Future 13 N-Ammonia and 15 O-Water, Require Onsite Cyclotron Onsite cyclotron permits use of 13 N-ammonia and 15 O-water for measurements of absolute MBF. The principal advantage of 15 O-water over 13 N-ammonia is its extraction fraction, which approaches 1 and thereby renders it a near-perfect MBF tracer. However, 15 O-water lacks US Food and Drug Administration (FDA) approval, has a short physical half-life ( 2 minutes), and fails to accumulate in the myocardium, so image quality is poor. In contrast, 13 N-ammonia (FDA approved) has a physical half-life 10 minutes, accumulates in the myocardium (glutamine synthase reaction), and thus provides high-quality images. 13 Its first-pass extraction fraction (nominal 0.8), however, is flow dependent, so accurate tracer kinetic MBF measurements depend on Renkin Crone correction (Figures 1B and 2). Data acquisition in list mode permits retrospective binning and ECG gating for determination of left ventricular (LV) volumes and ejection fraction with 13 N-ammonia and possibly from first-pass dynamic 15 O-water data. Although full-cycle, summed 15 O-water images have been attempted for visual assessment of relative uptake defects, the focus of recent studies has been on quantitative measurement of absolute MBF and has omitted measurements of LV volumes and EF. However, diagnostic-quality gated blood pool images for LV volumes and EF can be obtained with 15 O-CO. 19 Variability Among Computer Software and Quantitative Measures of 15 O-Water and 13 N-Ammonia MBF Tracer kinetic models in computer software for quantitative measurements of absolute MBF differ not only between 15 O-water and 13 N-ammonia but also for each Software may be institution specific, especially for 15 O-water, 25 although commercial applications for each are now widely available. Although detailed discussion of the various models is beyond the scope of this review, certain points are addressed. The first issue is comparability of the absolute MBF values across institutions and studies not only with the same tracer and similar models but also with different tracers and models. In the physiology laboratory, PET measurements of MBF with 15 O-water and 13 N-ammonia were compared in the same animal with simultaneously injected radiolabeled microspheres. 26 In Figure 1. A, Block diagram of 2 compartment tracer kinetic model (2TCM) used for 13 N-ammonia quantitative measurements of myocardial blood flow (MBF). K1 reflects MBF but requires correction (Renkin Crone) for the myocardial extraction fraction, which is flow dependent (Figure 2), to obtain an accurate measurement of MBF (ml min 1 g 1 ). The rate constants k2 and k3 reflect back-diffusion of the tracer and its trapping in the myocardium by incorporation in glutamine (glutamine synthase reaction), respectively. B, Shown here are 13 N-ammonia positron emission tomography time-activity curves used by the 2TCM model (A) for determination of MBF in a human subject. The diamond symbol curve represents the left ventricular (LV) blood pool and is the arterial (ART) input function. The square symbol curve represents myocardial (MYO) tracer uptake (response function). Note that the symbols are the actual data points, whereas the solid lines (redrawn) reflect the actual fit of the function to the data (R 2 =0.94). The value of K1 is 0.84 (1.03 ml min 1 g 1 after Renkin Crone correction).

3 2182 Circulation May 31, 2016 brief, the data from the study indicated excellent correlations between absolute MBF by both 15 O-water and 13 N-ammonia and microspheres (R=0.97, P<0.001; and R=0.94, P<0.001, respectively). 26 In humans, MBF values for 15 O-water and 13 N-ammonia in healthy volunteers have been reported. 27 Excellent agreement between the methods for both rest and stress MBF was noted (y = x, r =0.99, P<0.001, SEE=0.023 for average and y= x, r =0.97, P<0.001, SEE=0.025 for regional values; rest and stress, respectively). 27 However, the coefficients of variation for 15 O-water (14±11% at rest and 16±9% adenosine stress) were significantly greater than for 13 N-ammonia (9±4% at rest and 12±7% adenosine stress; P<0.004 and P<0.03, respectively). 27 The greater variability in the 15 O-water measurement notwithstanding, almost identical critical stress (adenosine) MBF values for identification of an anatomically and, when available functionally significant, coronary stenosis (measured fractional flow reserve [FFR] <0.80) have been reported by 2 different laboratories, 1 using 15 O-water (1.86 ml min 1 g 1 ) 14 and 1 using 13 N-ammonia (1.85 ml min 1 g 1 ). 28 Thus, the 2 tracers generally provided very comparable albeit not identical MBF results for both rest and adenosine stress. 14,28 Clinical studies with 15 O-water or 13 N-ammonia may yield varying results for similar end points (eg, stress MBF threshold for flow-limiting stenosis) for technical or clinical reasons or both. 25 Nonetheless, the literature suggests that the absolute values of maximally stimulated MBF (adenosine) of 2 ml min 1 g 1 will prove useful in identifying a flow-limiting coronary stenosis with 15 O-water or 13 N-ammonia. However, differences will be encountered, depending on a variety of factors. Thus, by default, a commercial program for 13 N-ammonia reports K1 values as MBF without Renkin Crone correction of K1, an important problem that results in exponential underestimation of MBF with increasing flows (Figure 2). Entry of Renkin Crone constants for correction of K1 to MBF, however, is possible. Comparisons of various modeling approaches and commercial software for 13 N-ammonia measurements of MBF are available. 21,24,27,29 In a study of multiple modeling approaches compared with radiolabeled microspheres in a canine model, it was found that a 2 tissue compartment model (2TCM) with a Renkin Crone correction of K1 and corrections for partial volume and spillover effects provided the best correlations with microspheres (r= , depending on methodology) and that metabolite correction did not improve results. 21 The initial 2 minutes of data acquisition was used for model fitting. Graphical (Patlak 30 ) analysis performed well, but dose uptake index (similar to standard uptake value but corrected for recovery coefficient and measured from dynamic data at 2 minutes post tracer injection) did not. 21 Thus, the 2TCM approach with geometric regions of interest for spillover and partial volume correction was recommended. 21 Human studies have compared data obtained in the same patient (rest and vasodilator stress) with 2 different tracer kinetic models (2TCM and 3 tissue compartment model) 24 or in the same patients with 3 commercial software packages, 2 with a 2TCM model and the other with a 1 tissue compartment model. 29 The precise details of ostensibly the same tracer kinetic model commonly vary in configuration and implementation and thus have an important impact on absolute values of MBF, particularly at the higher flows observed with vasodilator stress (eg, Figure 2). Thus, although the study comparing MBF obtained with 2TCM and 3 tissue compartment model reported statistically significant correlations between the 2 methods for rest and stress MBF and myocardial flow reserve (MFR) ratio, values of R 2 for rest and stress MBF were 0.61 and 0.48, respectively, and 0.48 for MFR. 24 Accordingly, given the clinical importance of the stress measurement of MBF and MFR, 14 17,27,28,31,32 the need for familiarity with specific software details and how data are acquired and processed cannot be understated. 82 Rb: Generator Produced 82 Rb is eluted from a commercially available 82 Sr generator system, which typically must be replaced every 6 weeks, given the 25-day half-life of the parent isotope. The short halflife of 82 Rb (75 seconds) translates to low radiation burden and dosimetry that may vary from a 1.1- to 3.5-mSv total effective dose for a maximal allowable activity of 60 mci injection at both rest and stress in an adult subject. 10,11 The effective dose is well below levels of most radiation-based cardiac assessments and particularly relevant to younger patients with longer time frames for stochastic effects to develop. Moreover, the short acquisition time of completing an entire study in <1 hour is an advantage for ill or higher-risk patients and those who find it difficult to remain still for more than a few minutes under the PET camera. Figure 2. Myocardial blood flow (MBF; ml min 1 g 1 ) is shown on the ordinate as a function of K1 (abscissa). Note the exponential nature of the relationship such that values of K require progressively greater, nonlinear correction to accurately reflect MBF (eg, K1=2.5 corresponds to MBF=4.0 ml min 1 g 1 ).

4 Gewirtz and Dilsizian Quantitative PET MBF Measurement and CAD 2183 Details of the data acquisition protocol will vary, depending on whether only relative, qualitative retention images are used for interpretation, which is most often the case, or actual tracer kinetic modeling is intended for absolute measurements of MBF. 33 Because list-mode dynamic acquisition (3D mode) with ECG gating will permit both qualitative and quantitative MBF image analysis and assessment of LV function, it is recommended. There are a number of issues to be considered when 82 Rb is used for quantitative measurements of absolute MBF. Its short half-life and high positron energy (3.15 MeV, with mean range in tissue of 2 4 mm 34 ) tend to limit spatial resolution compared with 13 N-ammonia and 15 O-water and make correction for partial volume effect essential. 34 The mathematical models for addressing this issue are numerous and are not discussed further, except to note that results may vary appreciably, depending on choice. Likewise, the myocardial first-pass extraction fraction of 82 Rb is low (nominal 0.5 at MBF of 1 ml min 1 g 1 and declines substantially with increasing MBF; nominal 0.3 at 4 ml min 1 g 1 ). 35 Accordingly, tracer kinetic models, typically 1 tissue compartment 36,37 but other more complex models have been used, 38 must take into account partial volume effects and flowrelated extraction fraction. An excellent comparison of models is available and documents the potential for substantial differences in MBF values obtained, depending on the basic model used and how it is implemented. 39,40 Another report indicates that when rest and stress 82 Rb MBF results were compared for the same 1 tissue compartment model, but as implemented in 3 different commercial software packages, results were highly correlated (R 2 = for pairwise comparisons), although the mean coefficient of reproducibility among them was 0.26 ml min 1 g Although the difference may seem small, the ischemic threshold for dipyridamole stress 82 Rb MBF has been reported at 0.91 ml min 1 g 1, 42 so software-related variability of 0.26 ml min 1 g 1 reflects >25% potential error in detecting a critical physiological end point. The relative error will be even larger when rest MBF is measured, which in turn has the potential to have a substantial impact on the MFR ratio, a parameter that has gained considerable attention in relation to CAD prognosis Furthermore, the same study that observed an ischemic threshold of 0.91 ml min 1 g 1 also demonstrated that myocardial ischemia with dipyridamole stress rarely occurred if stress MBF exceeded 1.12 ml min 1 g 1, 42 which is within the margin of error observed in the comparative study 41 noted above. Accordingly, given the importance of absolute values of MBF, both rest and stress, and the potential for methodology-based variation in these values, ideally it would be best for individual laboratories to establish their own normal limits for stress and rest MBF, particularly for 82 Rb. 39,40 18 F-Flurpiridaz: Availability as Unit Dose Without the Requisite of an Onsite Cyclotron Each of the tracers discussed above presents practical barriers to widespread adoption. 15 O-water requires onsite cyclotron and is not FDA approved. Although 13 N-ammonia is FDA approved, it also requires an onsite cyclotron, and the 10-minute physical half-life is a relative barrier for quick patient throughput (because of the 1-hour time delay between acquiring rest and stress images), which also restricts the efficiency of the cyclotron. Although the generator delivery system of 82 Rb is practical, it is rather expensive, and the radiotracer has less than optimal imaging and absolute MBF measurement characteristics. Accordingly, an 18 F-labeled tracer that is available in unit doses from regional medical cyclotrons, particularly one suitable for making measurements of absolute MBF, would represent a definite improvement over currently available 82 Rb generator technology. Moreover, the longer physical and biological half-lives of an 18 F-labeled tracer myocardial perfusion tracer would allow adequate image acquisition time to perform a treadmill test with certain limitations (see below) and pharmacological vasodilation stress studies, as well as a complete transition of nuclear laboratories from SPECT- to PET-based myocardial perfusion studies. 49 Currently, 18 F-flurpiridaz is in stage 3 clinical trials and, if approved by the FDA, has the potential to make PET quantitative measurements of absolute MBF more widely available. It should be noted, however, that the inability to easily obtain an arterial input function in the setting of exercise treadmill testing will effectively preclude the use of 18 F-flurpiridaz for obtaining absolute measurements of stress MBF in this setting. Although an arterial line could be used to obtain the input function, it would not be practical for routine clinical work. Data indicate, however, that MFR can be obtained as the standard uptake value ratio of stress/ rest 18 F-flurpiridaz based on images obtained 5 to 10 minutes after tracer injection 53 and thus would provide this important parameter in the context of treadmill exercise without the need for an arterial input function. Flurpiridaz is a derivative of the insecticide pyridaben, a rotenone-like compound that binds irreversibly to electron transport complex I on the inner mitochondrial membrane The tracer is lipophylic and rapidly taken up by myocytes Importantly, it has a high and flow-independent first-pass extraction fraction estimated at >90%, 54 along with prolonged retention in the myocardium. 54,56 Accordingly, it has the potential to be an excellent tracer for absolute quantitative measurements of MBF. A study in a porcine model of CAD supports this notion. 55 Thus, compared with radiolabeled microspheres, 18 F-flurpiridaz measurements of MBF (range, ml min 1 g 1 ) correlated well (R 2 =0.77, slope = 0.84). 55 The authors also observed a high target-to-background ratio and excellent delineation of a left anterior descending artery infarct zone. In general, it also was noted that 18 F-flurpiridaz and 13 N-ammonia measurements of MBF (ml min 1 g 1 ; both with the 3 tissue compartment model) in the same animal without left anterior descending artery stenosis agreed reasonably well at rest (median, 0.66; 25th and 75th percentiles, 0.64 and 0.69) versus 0.56 (25th and 75th percentiles, 0.51 and 0.58) for 18 F-flurpiridaz versus 13 N-ammonia, respectively. 55 Similar results were obtained with combined adenosine plus dobutamine stress (median 1.8 (25th and 75th percentiles, 1.7 and 1.9) versus 1.9 (25th and 75th percentiles, 1.7 and 2.0), 18 F-flurpiridaz versus 13 N-ammonia, respectively, although absolute values of stress MBF were for the most part <2 ml min 1 g Accordingly, the tracer appeared superior to 13 N-ammonia in terms of overall image quality and provided

5 2184 Circulation May 31, 2016 comparable levels of absolute MBF in swine with intact coronary circulation. However, a comparison, was not performed in the same animal of both 13 N-ammonia and 18 F-flurpiridaz measurements of rest and stress MBF and microspheres. In any case, additional work with either 13 N-ammonia or 15 O-water as the reference standard is required in human subjects to better document the utility of 18 F-flurpiridaz for quantitative MBF measurements. Given its reported very high and flow-independent extraction fraction, 54 one would anticipate quantitative absolute MBF results very similar to that of 15 O-water over a wide range of flows and to 13 N-ammonia. 26 A study of 7 healthy volunteers and 8 patients with documented CAD reported 18 F-flurpiridaz measurements of rest and adenosine stress MBF. 57 The study used the initial 90 seconds of dynamic data acquisition to measure tracer uptake rate as a function of arterial blood concentration and vascular volume within a given myocardial segment. Although corrections were made for partial volume and spillover effects, a tracer kinetic model was not used, and there were no reference MBF data (eg, 13 N-ammonia). Values of MBF obtained in healthy volunteers were plausible for rest (0.73±0.13, mean±sd; n=21 segments) and stress (2.53±0.48; same 21 segments). Compared with healthy subjects, MBF in segments in the distribution of coronary vessels with 50% stenosis had higher rest MBF (0.86±0.21; 12 segments in 8 patients) but lesser stress MBF (1.86±0.59; 12 segments in 8 patients). Additional work is required to validate 18 F-flurpiridaz methodology for absolute quantification of MBF with in humans. Nonetheless, the initial results hold promise and provide high-quality images (Figure 3A and 3B) 49 and incentive for definitive clinical trials. Coronary Physiology and Absolute Measurements of MBF Physiological studies of the coronary circulation and regulation of MBF are optimally performed with measurements of absolute MBF (ml min 1 g 1 ). Coronary flow velocity measurements, however, have been made in human subjects both invasively 58 and noninvasively by echocardiography. 59 More recently, efforts to assess maximal coronary vasodilator capacity and flow reserve as indicators of coronary stenosis severity and microvascular function have focused on coronary pressure based measurements such as FFR, index of microvascular resistance (requires coronary flow estimate, typically thermodilution mean transit time for the Ohm law model of resistance), 67 coronary diastolic wave free interval, 68 and a hybrid measurement of instantaneous middiastolic stenosis pressure gradients (dp) versus corresponding hyperemic coronary flow velocities (dp at 50 cm/s is optimal for diagnosis of hemodynamically significant stenosis). 69 The fact that FFR is an invasive measurement and that complete evaluation of the coronary circulation may require instrumentation of >1 and even all 3 major coronary vessels notwithstanding, the mean pressure ratio (distal to stenosis/ aorta) has gained wide acceptance as a decision-making tool for coronary revascularization because it is an excellent predictor of clinical outcomes However, FFR has its limitations, which include potential confounding factors such as hemodynamic conditions, 70,71 diffuse atherosclerosis, and microvascular dysfunction, any or all of which may affect the ratio importantly. 72 Furthermore, measurements of MFR and FFR may differ in opposite directions in a substantial percentage of patients for a variety of physiological reasons, including those noted above. 72,73 Such discrepancies may obfuscate rather than facilitate clinical decision making unless one has available absolute measurements of rest and stress MBF and thus is able to recognize, for instance, that MFR (absolute reserve) is critically reduced (ie, <1.7X) and FFR (relative reserve, based on assumption of normal reference value of 1) appears noncritical (ie, >0.8) because stress MBF is relatively reduced. This scenario commonly occurs in the context of diffuse CAD often with mild/moderate focal stenosis, which in the setting of impaired stress MBF fails to result in a pressure drop sufficient to reduce FFR to the ischemic level (ie, <0.8). 72 The heart has quantifiable energy requirements for fulfilling its vital function. It depends on MBF to supply the oxygen required to meet the prevailing level of myocardial oxygen demand. Experimental data from both human and animal studies provide insight into just what those levels are in milliliters per minute per gram. Although efforts have been made recently to address this issue with invasive thermodilution measurement of regional coronary flow in humans in the cardiac catheterization laboratory, 74 the methodology remains invasive and is hampered by the fact that no simple real-time method is available to determine the mass of the perfusion bed studied. Thus, MBF in milliliters per minute per gram is not obtainable. It is possible that 3D contrast echocardiography may assume this role in the future, although its real-time practicality remains to be demonstrated. Review of quantitative PET measurements of absolute MBF demonstrates the importance of such data for physiological assessment of the coronary circulation, particularly in the setting of CAD. In an important human study, the threshold was determined for vasodilator-stimulated MBF, which optimally predicted clinical evidence of severe life-threatening myocardial ischemia as manifest by diagnostic ST-segment depression, severe angina requiring medical treatment, or both. 42 PET with 82 Rb was used to measure MBF at rest and with dipyridamole stress. Stress MBF <0.91 ml min 1 g 1 was highly predictive of the occurrence of myocardial severe ischemia. Indeed, 40% of patients manifesting ischemia in that study had evidence of coronary steal and thus may be especially at risk for cardiovascular death or myocardial infarction (MI). Furthermore, evidence of severe ischemia was rarely encountered with stress MBF >1.12 ml min 1 g 1, and mean heart rate (HR), systolic blood pressure, and rate-pressure product (RPP) in patients manifesting ischemia (87±14 bpm, 131±21 mm Hg, and mm Hg/min, respectively) were very similar to those in individuals without ischemia (90±15 bpm, 118±16 mm Hg, and mm Hg/min, respectively). 42 The argument has been made that patients manifesting ischemia with dipyridamole stress in whom MBF cannot be increased >0.91 ml min 1 g 1 may be at high risk for cardiovascular death or MI and would benefit most from coronary revascularization, a hypothesis that has yet to be tested although a clinical trial has been proposed. 44

6 Gewirtz and Dilsizian Quantitative PET MBF Measurement and CAD 2185 Figure 3. Representative examples of normal and abnormal 18 F-flurpiridaz positron emission tomography (PET) myocardial perfusion images. A, Consecutive tomograms of paired stress (top row) and rest (bottom row) images show normal distribution of the radiotracer in all myocardial regions with excellent target background ratio. B, Paired stress (top row) and rest (bottom row) images show extensive reversible regional perfusion defects in all 3 coronary artery vascular territories associated with transient ischemic cavity dilation. If the initial promising absolute myocardial blood flow data with 18 F-flurpiridaz (or other 18 F-labeled PET perfusion tracers) are confirmed in larger clinical trials, they may facilitate the transition of nuclear laboratories from single-photon emission computed tomography to (Continued )

7 2186 Circulation May 31, 2016 Table 1. Condition Canine Exercise Treadmill Test MBF HR, bpm MAP, mm Hg HR MAP MBF, ml min 1 g 1 (TM) Rest 76±3 86± ±0.09 Moderate exercise 152±4 102± ±0.11 Severe exercise 193±53 116± ±0.22 Maximal exercise 240±6 124± ±0.12 HR indicates heart rate; MAP, mean arterial pressure; MBF, myocardial blood flow; and TM, transmural. Data taken from Ball et al. 75 In contrast, other experiments indicate high levels of physical activity in the range of 10 metabolic equivalents (METs) or more require MBF in the range of 2 to 3 ml min 1 g 1 and possibly as much as 4 ml min 1 g 1. An ischemic threshold in this range will be useful in the identification of functional significant CAD (see below). Although it may not be predictive of cardiovascular death or MI, it will provide a very helpful measure for more general and more broadly clinically relevant physiological assessment of the coronary circulation. Thus, in a canine study of healthy animals with intact coronary circulation (Table 1), treadmill exercise to an RPP of (HR mean arterial pressure [MAP]; HR=152±4 bpm, mean±sem) was associated with microsphere-measured MBF of 1.83±0.11 versus 0.94±0.09 ml min 1 g 1 at rest at an RPP of Exercising more vigorously to an RPP of (HR, 193±53 bpm) increased MBF to 2.75±0.22 ml min 1 g 1. Maximal exercise (6.4 mph at 11% grade, steady hemodynamics for 45 seconds at an RPP of and HR of 240±6 bpm) further augmented MBF to 3.90±0.12 ml min 1 g A study in swine undergoing 10 minutes of atrial pacing with simultaneous norepinephrine infusion (Table 2) demonstrated that MBF (microspheres) in normally perfused segments increased from 1.52±0.07 ml min 1 g 1 (mean±sd; RPP= ) at baseline to 2.06±0.60 ml min 1 g 1 at an HR of 149±9 bpm and MAP of 149±17 mm Hg (RPP= ). 76 In humans with CAD given maximally tolerated intravenous dobutamine infusion (7±3 minutes at 34±7 μg kg 1 min 1 ), MBF (PET 13 N-ammonia; Table 3) in regions perfused by 1 functionally normal coronary vessels (MBF 2 ml min 1 g with adenosine infusion) increased from a baseline level of 0.99±0.45 ml min 1 g 1 (HR=79±22 bpm; systolic arterial pressure=124±14 mm Hg; RPP=10± ) to 2.16±0.99 ml min 1 g 1 (HR=115±28 bpm; systolic arterial pressure=135±25 mm Hg; RPP=15± ). 77 Finally, in young, healthy humans performing supine bicycle exercise while in the PET scanner (Table 4), exercise to 150 W ( 9 METs) for 10 minutes was associated with MBF ( 15 O-water) of 2.02±1.01 ml min 1 g 1 in endurance-trained subjects and 3.10±0.73 ml min 1 g 1 in untrained individuals, who at that level of effort had exceeded their anaerobic threshold. 78 More intense effort to 70% of maximal for 10 minutes for a given individual caused MBF to increase further to 3.46±1.31 ml min 1 g 1 in trained subjects (at HR 135 bpm and MAP 100 mm Hg at Table 2. Porcine Atrial Pacing Plus Norepinephrine Microsphere MBF Condition HR, bpm MAP, mm Hg HR MAP NZ MBF, 1 ml min 1 g (TM) Rest 100±16 119± ± ±0.27 Atrial pacing+ norepinephrine (10 min) 149±9 149± ± ±0.60 HR indicates heart rate; MAP, mean arterial pressure; MBF, myocardial blood flow; NZ, normal zone; and TM, transmural. Data taken from Berman et al W= 13METs) and 4.11±2.08 ml min 1 g 1 in untrained individuals (at HR 175 bpm and MAP 100 mm Hg at 180 W= 10METs). 78 It should be noted that baseline MBF was similar in endurance-trained (0.72±0.20 ml min 1 g 1 ) and untrained (0.79±0.19 ml min 1 g 1 ) individuals although as expected trained subjects had lower resting HR (44±6 bpm) versus untrained subject (58±6 bpm) but similar MAP (78±6 versus 76±4 mm Hg, respectively). 78 In summary, therefore, across a range of experimental designs and species, including humans (Tables 1 4), exercise levels to 10 to 13 METs have been associated with measured MBF of 2 to 4 ml min 1 g 1. In this regard, it is particularly noteworthy that an empirically derived, quantitative PET absolute MBF threshold for the detection of a functionally significant coronary stenosis in humans has been reported to be in the range of 1.9 to 2.4 ml min 1 g ,28,31 Accordingly, a physiological basis is evident for the interpretation of quantitative PET absolute measurements of stress (vasodilator) MBF over a broad range of flows. Inability to increase MBF >0.9 ml min 1 g 1 is likely to be accompanied by clinical evidence of severe myocardial ischemia, 42 especially if associated with coronary steal (as many as 40% of patients), and appears to be a marker of increased risk for cardiovascular death or MI in patients who may benefit from coronary revascularization. Furthermore, although patients who are able to increase MBF in all vascular territories to levels >1.12 ml min 1 g 1 (and typically have MFR 1.7X) may not show clinical evidence of myocardial ischemia in the context of vasodilator stress, 42 unless they are able to mount responses at least 2 ml min 1 g 1 in the bulk of all major coronary territories, they are unlikely to be able to perform workloads in the range of 10 to 13 METs and remain free of a lesser degree of ischemia, which may not be life-threatening but will be clinically limiting and thus very important to the patient and his/her physician. Thus, at MBF >1.12 but <2 ml min 1 g 1, patients may be at risk for ischemia at workloads such as, for example, making beds (3 5 METs) 79 or somewhat higher such as shoveling snow (4 8 METs). 79 Decisions about the need for cardiac catheterization and coronary revascularization in such patients, as in all with ischemic heart disease, must be individualized and cannot be made based solely on a single stress MBF or MFR number or even both. In this regard, it is recognized there will be individual variation in MBF required to achieve a given level of external work ischemia free. Thus, Figure 3 Continued. PET-based myocardial perfusion studies. This would be justified on the basis of high-quality images of PET at lower radiation exposure. Reproduced from Dilsizian and Taillefer 49 with permission from the publisher. Copyright 2012 American College of Cardiology.

8 Gewirtz and Dilsizian Quantitative PET MBF Measurement and CAD 2187 Table 3. Condition Humans PET 13 N-Ammonia MBF HR, bpm SBP, mm Hg HR SBP NZ MBF, 1 ml min 1 g (TM) Rest 79±22 124± ± ±0.45 Dobutamine (34 1 µg kg 1 min for 7 min) 115±28 135± ± ±0.99 HR indicates heart rate; MBF, myocardial blood flow; NZ, normal zone; PET, positron emission tomography; SBP, systolic blood pressure; and TM, transmural. Data taken from Skopicki et al. 77 for clinical decision-making purposes, ischemia must be considered on a spectrum from severe life-threatening (low workload, low stress MBF, eg, <0.9 ml min 1 g 1 ) and low MFR (eg, <1.7X) to milder grades, which may occur at workloads in the range of 9 to 13 METs if MBF fails to reach or exceed 2 to 3 ml min 1 g 1 and MFR at least 2X, and all variations of both stress MBF and MFR in between. Factors that may play a role include physical fitness 78 and genetic and other factors (eg, myocyte oxidative stress) that may affect the efficiency of myocardial oxygen utilization both in mitochondrial production of ATP 80 and in generation of external mechanical work. 81 Individual variation notwithstanding, stress MBF levels <1.9 ml min 1 g 1 in a given coronary territory are likely to render it vulnerable to effort-induced ischemia, perhaps not life-threatening or severe but ischemia nonetheless (and potentially life-threatening if engendering ventricular tachycardia/ventricular fibrillation), particularly as effort exceeds 9 METs. Stress MBF >2.5 ml min 1 g 1 likely will be adequate to meet work in the range of 10 to 12 METs with higher MBF (3 to perhaps 4 ml min 1 g 1 ) as METs approach or exceed 13. Finally, it should be emphasized the MBF values should not be used in isolation or as rigid, immutable cut points. Rather, the absolute levels of stress MBF discussed above should be Table 4. Ref. 78). Humans PET 15 O-Water MBF (Data taken from HR, bpm MAP, mm Hg HR MAP MBF, 1 ml min 1 g (TM) Condition, trained Rest 46±6 78±6 3375± ± W 110±12 94± ± ±1.01 (10 min) 70% max W (10 min) 143±9 106± ± ±1.31 Condition, untrained Rest 58±6 76±4 4421± ± W 136±12 104± ± ±0.73 (10 min) 70% maximum W (10 min) 162±10 105± ± ±2.08 HR indicates heart rate; MAP, mean arterial pressure; MBF, myocardial blood flow; PET, positron emission tomography; and TM, transmural. Data taken from Laaksonen et al. 78 seen as guidelines for clinical decision making and must be set in the context of the patient s overall clinical cardiac and general medical condition. Inability to exceed 0.9 ml min 1 g with stress or coronary steal is likely to be a marker of severe 1 ischemia, especially if encountered at relatively low workloads, and may require more aggressive intervention. Clinical Applications of PET Absolute Measurements of MBF: Stress Hyperemic MBF and MFR Detection of CAD: Conduit Vessels Validation of PET methods for quantitative measurements of absolute MBF with each of the 3 tracers used clinically ( 82 Rb, 13 N-ammonia, and 15 O-water) is available. 20,21,26,35,36,82 84 It was appreciated from the outset that the inherently quantitative nature of PET methodology held the potential for greatly improving the assessment of patients with known or suspected CAD and other cardiac conditions. An integrated approach to CAD assessment based on quantitative PET 82 Rb measurements of absolute MBF, which takes into account MFR, relative tracer uptake, and absolute MBF at rest and with dipyridamole stress, has been reported. 42,44,46 Important details concerning cut points of absolute stress MBF for the likelihood of the presence (<0.91 ml min 1 g 1 ) or absence (>1.12 ml min 1 g 1 ) of clinical or ECG evidence of severe, potentially life-threatening ischemia that may benefit from coronary revascularization are noted above. 42,46 Because in patients with known or suspected ischemic heart disease MFR 2X is considered normal 46,61 and is associated with favorable prognosis, even though on average MFR is greater in healthy volunteers (4 5X), 85,86 it is important to note in the setting of vasodilator stress, sectors with rest MBF of 0.2, 0.5, 1.0, and 2.0 ml min 1 g 1, that stress MBF of 0.4, 1.0, 2.0, and 4.0 ml min 1 g 1, respectively, would qualify as having normal MFR with flow reserve classed in this schema as mildly (stress MBF, 0.4 and 1.0 ml min 1 g 1 ) to minimally (stress MBF, 2.0 ml min 1 g 1 ) reduced or equivalent to that of healthy volunteers (stress MBF, 4.0 ml min 1 g 1 ) and, on average, none would demonstrate severe, life-threatening ischemia. 44,46 The MFR ratio, however, is heavily dependent of the level of rest MBF and thus may be misleading in certain cases. Data reviewed above indicate that myocardial oxygen demand in the range of 9 to 10 METs clearly cannot be met by stress MBF in the range of 0.4 to 1.0 ml min 1 g 1, often regarded as within the range of normal rest MBF. 85,87 Although it has been reported anecdotally that such patients may do well clinically if MFR is 1.7X, 42 this likely reflects the combination of sedentary lifestyle and β-blocker therapy, which limit MVo 2 and make such low levels of stress MBF adequate to the patient s needs. It is highly unlikely, however, that such patients would be able to perform 5-MET workloads such as walking at 7 km/h or biking at 10 km/h 79 and remain free of ischemia notwithstanding MFR 1.7X. In contrast, stress MBF in the range of 2 to 4 ml min 1 g 1 with MFR 2 likely will be adequate in most cases to meet myocardial oxygen demand in the range of 9 to 13 METs (eg, squash, pool swimming at 3 km/h, backpacking at 7 8 km/h with a 20-kg pack and 5% grade). 79

9 2188 Circulation May 31, 2016 Because MFR obviously requires measurement of rest MBF, it will be helpful in distinguishing a region of very low stress MBF, which may reflect coronary steal from a region of scar in which both rest and stress MBFs are very low 82 and do not differ. Moreover, coronary microvascular disease has been associated with heterogeneity of rest MBF, 88 which may be identified with quantitative rest MBF measurement. Although it is true that MFR smooths method variation in the rest and stress MBF measurements by canceling of the errors in each, this is correct only if stress and rest MBFs are computed simultaneously and the division is made in the mathematical model. More commonly, however, the ratio is computed after each component has been determined separately (ie, stress MBF/ rest MBF), in which case it can be shown mathematically that the error of the ratio is greater than that of its individual components. Accordingly, if MFR is planned as an important parameter in the analysis of data from a multicenter clinical trial, it should be computed as noted above and not as a simple ratio after the individual components have been determined. A related but somewhat different approach for using quantitative PET measurements of absolute MBF for the diagnosis of CAD has been reported by others ,28,31 In 1 study using 13 N-ammonia for the diagnosis of CAD (invasive coronary angiography, 70% stenosis), a receiver-operating characteristic/area under the curve analysis was performed to compare relative myocardial tracer content at stress (adenosine infusion of 140 µg kg 1 min 1 for 6 minutes) versus MFR ratio versus absolute level of stress MBF for CAD detection. 28 Optimal cut points based on a sample of 144 vessels in 48 patients were identified as tracer content <70% of maximum; MFR <2X, and absolute stress MBF <1.85 ml min 1 g 1. The diagnostic accuracy of absolute stress MBF was significantly greater than that of tracer content (0.85 versus 0.72, respectively; P<0.005) and marginally greater than that of MFR (0.79; P=0.06). However, the sensitivity of absolute stress MBF exceeded that of MFR (0.81 versus 0.62, respectively; P<0.01) and tracer content (0.81 versus 0.48, respectively; P<0.001). The specificity for each parameter was comparable (0.85 versus 0.82 versus 0.85; stress MBF versus tracer content versus MFR, respectively). Accordingly, stress MBF overall was superior to tracer content for CAD diagnosis. Stress MBF also demonstrated significantly greater sensitivity for CAD diagnosis than MFR with marginally greater accuracy. 28 Subsequently, the above findings, in particular the superiority of stress MBF for detection of CAD vis-à-vis MFR or tracer content, were confirmed by several larger studies that importantly included FFR data when available to define hemodynamically significant CAD (FFR 0.80) ,31 These studies were performed with 15 O-water as the MBF tracer. Nonetheless, the optimal level of stress MBF for identification of hemodynamically significant CAD in 1 of the studies 14 agreed almost identically with an earlier study 28 that used 13 N-ammonia (1.86 and 1.85 ml min 1 g 1, respectively). More recent reports indicate that the optimal threshold may be higher, at 2.4 ml min 1 g ,31 1 or <1.5 ml min 1 g increase in stress MBF above rest MBF level. 31 Thus, for an MFR cut point of 2.5X, the sensitivity, specificity, and positive and negative predictive values were 0.81, 0.87, 0.66, and 0.94, respectively, with an area under the curve of An absolute stress MBF cut point of 2.4 ml min 1 g 1 demonstrated sensitivity, specificity, and positive and negative predictive values of 0.95 (P=0.03 versus MFR), 0.90, 0.73, and 0.98, respectively, with an area under the curve of 0.94 (P=0.02 versus MFR). Accordingly, the authors concluded that absolute stress MBF was superior to MFR for CAD detection. They noted that the examination could be completed with a single tracer injection, which would reduce patient radiation exposure and enhance laboratory efficiency. 31 The superiority of absolute stress MBF measurement to tracer content analysis for CAD diagnosis also has been confirmed by this group (Figure 4). 17 Finally, these authors also noted that the combined use of absolute stress MBF measurement and CT coronary angiography may be very helpful in improving CAD diagnosis, especially in circumstances in which both stress MBF and MFR are reduced in all major vascular territories. In such cases, diffuse but nonobstructive CAD in association with coronary microvascular disease (or microvascular disease alone) must be distinguished from severe, focal, triple-vessel conduit artery disease, which may be accomplished by combining CT coronary angiography with the PET examination. 14,16 Omission of the rest MBF measurement, however, would preclude not only the use of the stress MBF increment criteria (ie, <1.5 ml min 1 g 1 increase in stress MBF above rest MBF) but also the ability to diagnose coronary steal and, as noted above, assist with the recognition of myocardial scar, microvascular disease, and diffuse CAD, any of which if Figure 4. The superiority of absolute measurement of stress MBF to myocardial tracer content analysis is demonstrated. Note the normal tracer uptake with stress (A; all territories >80% maximal uptake; yellow to red) in the face of what in fact is impaired stress myocardial blood flow response in all coronary territories (B; all territories in shades of blue, <2.5 ml min 1 g 1 ; yellow to red, 2.5 ml min 1 g 1, within normal limits). Reproduced from Kajander et al. 17

10 Gewirtz and Dilsizian Quantitative PET MBF Measurement and CAD 2189 identified would enhance the physiological information of the examination and may help identify a clinically high-risk subset of patients. 42 Accordingly, an individualized approach likely is best. For a patient with known, relatively simple coronary anatomy but of uncertain functional significance, a single stress MBF measurement may be adequate. In contrast, a full rest/stress MBF examination may be required in a patient being considered for repeat coronary revascularization who has known complex coronary anatomy (eg, previous coronary artery bypass graft surgery and possibly chronic total occlusion) in whom the correct target vessel is uncertain, as is the optimal procedure to be used (eg, complex percutaneous coronary intervention versus repeat coronary artery bypass graft surgery). Systematic comparison of absolute PET MBF measurements with SPECT MPI for the diagnosis of CAD has not been reported, although there are numerous reports of recognition of triple-vessel CAD based on absolute PET MBF measurements, which would have been missed with the use of uptake images alone 14,17,28 (Figure 4). A study comparing static PET myocardial uptake images with SPECT MPI for the detection of CAD (gold standard, coronary angiography) in a cohort of 112 patients who had both SPECT and 82 Rb PET MPI indicated that the diagnostic accuracy for PET was 89% compared with 79% for SPECT (P=0.03). 89 Both image quality and interpretive certainty were superior for PET than SPECT. A qualitative phase II study comparing PET 18 F-flurpridaz with SPECT 99m Tc-MIBI for the diagnosis of CAD 50 demonstrated the following: In 86 patients who underwent ICA, sensitivity of PET was higher than SPECT (78.8% versus 61.5%, respectively, P=0.02). Specificity was not significantly different (PET: 76.5% versus SPECT: 73.5%). Receiver-operating characteristic curve area was 0.82±0.05 for PET and 0.70±0.06 for SPECT (P=0.04). Normalcy rate was 89.7% with PET and 97.4% with SPECT (p = NS). 50 In a subsequent phase III study 18 F-flurpridaz again had superior sensitivity to SPECT 99m Tc-MIBI for the diagnosis of CAD but failed to meet its noninferiority target for specificity, although it was superior by receiver-operating characteristic/ area under the curve analysis for CAD diagnosis. 52 Again, it should be emphasized that none of these studies compared actual PET measurements of absolute stress MBF (or MFR) with SPECT for CAD diagnosis. Furthermore, the potential advantages of PET-CT combined with CT angiography for CAD diagnosis compared with SPECT MPI have not been reported. PET Absolute MBF, MFR, and Prognosis in CAD A number of quantitative PET absolute MBF studies have retrospectively examined the prognostic value of coronary flow reserve (generally used synonymously with MFR, although strictly not the same), static tracer uptake images, or both in patients with known or suspected CAD. Thus, in 1 study, 13 N-ammonia uptake images were subjectively scored per the standard 17-segment model, 45 and absolute values of MBF were used to calculate MFR in regions with defects relative to regions with normal radiotracer uptake, referred to as remote regions. Major adverse cardiac events (MACEs) were defined as cardiac death, nonfatal MI, late revascularization, or hospitalization for cardiac reasons. Absolute values of rest and stress MBF were neither used nor reported except to compute MFR (<2X considered abnormal). The patients were analyzed in a number of ways. First, prognosis was compared in those with abnormal versus those with normal perfusion (ie, no uptake defects). The incidence of MACEs and cardiac death was significantly greater in those with abnormal perfusion (P<0.001 and P<0.05, respectively, versus normal perfusion). 45 Moreover, adding MFR to perfusion results refined prognostication. Over 3 years of follow-up, compared with those with normal MFR and perfusion, those with normal perfusion but abnormal MFR had a higher incidence both of MACEs (1.4% versus 6.3%; P<0.05) and cardiac death (0.5% versus 3.1%; P<0.05). However, over 10 years of follow-up, MFR had no added value in the prediction of MACEs or cardiac death over and above normal perfusion, although it did enhance the prediction of both MACEs and cardiac death in those with abnormal perfusion. Thus, in those with abnormal perfusion but normal MFR, the incidence of both MACEs and cardiac death (36% and 7%, respectively) was less than in those with both abnormal perfusion and abnormal MFR (56% and 26%, respectively). 45 The authors recognized the retrospective analysis and use of global MFR in the data analysis as weaknesses of the study. 45 Reliance on relative tracer uptake as opposed to absolute values of MBF, particularly stress MBF, to categorize perfusion as normal or abnormal was not commented on but clearly is a weakness (Figure 4). 17 Nonetheless, the authors concluded that the combination of normal perfusion and MFR offers a 3 year warranty period of lower risk for MACEs and cardiac death compared with abnormal MFR and that the combination of abnormal perfusion and MFR was associated with a particularly unfavorable prognosis. 45 In a considerably larger multicenter (n=4) cooperative study using 82 Rb PET for qualitative (17-segment model) assessment of myocardial perfusion, >7000 patients with known or suspected CAD were followed up for a median of 2.2 years with a primary outcome of cardiac death (n=169) and secondary outcome of all-cause mortality (n=570). 90 The authors reported, Risk-adjusted hazard of cardiac death increased with each 10% myocardium abnormal with mildly, moderately, or severely abnormal stress PET (hazard ratio [HR]: 2.3 [95% CI (confidence interval): ; P=0.001], HR: 4.2 [95% CI: ; P<0.001], and HR: 4.9 [95% CI: ; P<0.0001], respectively [normal MPI: referent]). 90 The study did not use MFR as a risk predictor, nor did it use absolute measurements of MBF in the analysis. The authors concluded that the extent and severity of ischemia and scar on PET MPI by qualitative analysis of 82 Rb images were powerful predictors of both cardiac death and all-cause mortality and provided added information beyond that provided by traditional cardiac risk factors, which also were considered. 90 This report, similar to the one discussed above, 45 was limited by the failure to use quantitative measurements of absolute MBF, particularly stress, as a risk parameter; instead, it used

11 2190 Circulation May 31, 2016 subjective visual analysis of the tracer retention images as the principal metric for risk stratification. A single-center study of almost 2800 patients with known or suspected CAD that used quantitative measurements of absolute MBF ( 82 Rb PET) to calculate MFR and to assess its value as a prognostics indicator of cardiac death also assessed the value of absolute stress MBF for the same purpose. 47 Patients were followed up for a median of 1.4 years. Both MFR and absolute stress MBF were considered in tertiles and collapsed to reflect global, not individual, coronary vascular territories. Multiple statistical models and parameters were used to analyze the data. In brief, in multivariate analysis, the adjusted hazard ratio for the lowest tertile of MFR (<1.5X) was associated with 5.6-times (95% CI, ; P<0.001) increased risk of cardiac death compared with the highest tertile (>2X). 47 The authors noted in this regard that increments in global MFR tertiles were driven by progressive increments in global absolute stress MBF (absolute global rest MBF remained constant at 1 ml min 1 g 1 across tertiles), which also was a predictor of cardiac death in multivariate analysis although not as strong in the models used as MFR, 47 the highly correlated nature of the parameters notwithstanding. They also reported that the addition of MFR data to a risk model consisting of clinical information, results of visual analysis of the MPIs (combined ischemia and scar), rest LV ejection fraction, and LV ejection fraction reserve demonstrated statistically significant improvement in risk assessment, particularly in those at intermediate risk (annual cardiac mortality, 1% 3%), with 17% being reassigned to high risk (>3%) and 34% to low risk (<1%) and the remainder (49%) unchanged 48 (Figure 5). 47 The effect on risk reassignment of the addition of global stress MBF in place of MFR to the risk model was not reported. The authors concluded, however, that assessment of coronary vasodilator function with positron emission tomography is a powerful, independent predictor of cardiac mortality in patients with known or suspected coronary artery disease. 47 It is noteworthy that each tertile reduction in MFR was driven by a reduction in absolute stress MBF, which the authors correctly stated was indicative of the fact in this study (with constant absolute rest MBF across tertiles) that the decline in MFR reflected a reduction of coronary vasodilator function. 47 Accordingly, the key determining physiological parameter of interest was absolute stress MBF, optimally assessed on a regional, not global, level, in the setting of known or suspected CAD. Detection of Coronary Microvascular Disease In the application of quantitative PET measurements of absolute MBF for the assessment of coronary microvascular function in a variety of clinical contexts, including heart failure, idiopathic and hypertrophic cardiomyopathy, coronary artery vasculopathy after cardiac transplantation, and endothelial dysfunction in the setting of ischemic heart disease, risk factors such as smoking, diabetes mellitus, dyslipidemia, hypertension, obesity, and chronic kidney disease have been considered in some detail in several recent reviews 87,91 93 and are not discussed further here. Figure 5. The effect of adding myocardial flow reserve (MFR) to a cardiac risk model in terms of low-, intermediate-, and high-risk tertiles is shown. In the intermediate-risk category (blue bar, before MFR), addition of MFR information (pie chart below) demonstrates that 49% of patients remain in the same risk category, whereas 17% are reassigned to high risk and 34% to low risk. Reassignment to other levels of risk in pre-mfr low- and high-risk groups (green and red bars, respectively) is more modest. Bar graphs at the bottom depict annualized cardiac mortality in the post-mfr reassignment groups. Note that the postreassignment annualized cardiac mortality in the high-risk group (10.5%) remains considerably greater than that of intermediate-risk group (4.4%). Reproduced from Murthy et al. 47

12 Gewirtz and Dilsizian Quantitative PET MBF Measurement and CAD 2191 One potential application that has gained considerable attention in recent years concerns CAD in women. Thus, the Women s Ischemia Syndrome Evaluation (WISE) study 94 and a more recent review 95 have emphasized the not uncommon clinical presentation of woman with atypical chest pain who are found by either invasive or CT coronary angiography to have anatomically normal or apparently only mildly diseased conduit coronary arteries. Such patients often represent a diagnostic challenge. Furthermore, in a recent, large, randomized, clinical trial of patients with non ST-segment elevation acute coronary syndrome (Metabolic Efficiency With Ranolazine for Less Ischemia in Non-ST-Elevation Acute Coronary Syndromes- Thrombolysis in Myocardial Infarction 36 [MERLIN-TIMI 36]), women more often than men were found to have nonobstructive conduit vessel CAD (19.4% versus 8.6%; P<0.001) yet more likely to have an ECG-documented ischemic episode during continuous monitoring in the initial week after presentation (22.5% versus 19.3%; P=0.0025). 96 The cause of ischemia in these patients often is unclear and may be multifactorial. 95 Microvascular dysfunction, however, is thought to play an important pathophysiological role, 94,95 which clinically is very similar, if not identical, to syndrome X (angina with objective evidence of myocardial ischemia notwithstanding normal or only minimally diseased conduit coronary vessels ). Quantitative PET measurements of absolute MBF are playing an increasingly important role in the evaluation and management of such patients, in whom qualitative SPECT MPI commonly is uninformative, if not misleading. Furthermore, a stress-only study will be adequate in most cases to answer the clinical question and will minimize radiation exposure, a particularly important consideration in women who may also require screening mammography. Thus, the inability to increase stress MBF >2.0 to 2.4 ml min 1 g 1 in the absence of obstructive epicardial vessel disease is indicative of coronary microvascular disease. 17,87,92 The presence of coexistent diffuse epicardial CAD without apparent flow-limiting, focal stenosis, which may accompany and contribute to the impaired MBF response to vasodilator stress in this setting, must be considered and may be recognized by a pattern of progressive base to apex decline in stress MBF 100 and a pattern of heterogeneous rest MBF, which obviously would require an additional tracer dose but would permit determination of MFR. It is recognized, however, that the diagnosis of coronary microvascular disease may be difficult for many reasons, including 1 or more of the following; (1) variability in clinical signs (eg, ECG evidence of ischemia) and symptoms; (2) absence of a uniformly accepted definition of heterogeneous rest MBF 101 ; (3) individual variation in rest MBF not explained by differences in RPP in the context of a functional disorder with a pathogenesis is not well defined but potentially includes hypothesized mechanisms that would allow for either relatively elevated or reduced MBF 88 ; and (4) use of MFR as a metric for making the diagnosis given the potentially highly variable levels of rest MBF that may be encountered both within and between patients. Future Applications Once quantitative PET measurements of absolute MBF have been integrated into routine evaluation and management of patients with CAD, it is easy to see how this information could be integrated into other related areas. This is particularly true once an 18 F-labeled tracer such as flurpiridaz, which has been shown to be suitable for making quantitative, absolute MBF measurements, 51,55,57 is approved for clinical use. Application to heart failure management at various stages of the syndrome has been reviewed recently and ranges from diagnosis to exclude obstructive CAD in patients at either low risk for CAD or high risk for renal failure with invasive coronary angiography 93 to more detailed assessment of coronary vasodilator function in clinical decision making for coronary revascularization in complex cases often involving previous coronary artery bypass graft surgery or chronic total occlusion or both. In this regard, it is worth recalling that although there is considerable overlap between viable and nonviable myocardium at relatively reduced levels of rest MBF, 102 viable myocardium is only rarely, if ever, present at very low rest MBF (<0.25 ml min 1 g 1 ). 82 Accordingly, although one would not order a quantitative PET MBF study for the sole purpose of determining myocardial viability, in the context of heart failure evaluation, for instance, the information would be available if the study were obtained for physiological assessment of the coronary circulation. In such cases, rest MBF data could obviate the need for a separate 18 F-FDG viability study in the event that a very low value is observed, for example, in a region subtended by chronic total occlusion for which complex percutaneous coronary intervention was under consideration. Monitoring of coronary endothelial function in response to statin therapy (Figure 6) 32 or other intervention already has been shown to be feasible 32,85,92, and has been reviewed in detail by others. 87,92 Application to the assessment of coronary endothelial function of newly FDA-approved PCSK9 inhibitors is an obvious extension of previous work in this area. 32,85,92, The toxic effects of obesity, diabetes mellitus, hypertension, and tobacco abuse on coronary endothelial function are abundantly documented in the literature. New therapies aimed at treating any of these very common clinical entities would be appropriate candidates for baseline and follow-up PET measurements of absolute MBF to document improvement (or lack thereof) and for use in conjunction with clinical outcomes research. The latter will play an increasingly important role in the assessment of new therapies, which often are expensive, in some cases extremely so, and require justification in a socioeconomic environment appropriately concerned with quality improvement and cost containment. Finally, the current revolution in DNA editing (CrispR-Cas9 technology) with its application to Duchene muscular dystrophy, 107 a disease that commonly results in a fatal cardiomyopathy at an early age, is just one example in which absolute measurements of MBF may prove helpful in providing not only advance indication of cardiomyopathy onset but also, even more important, early indication of responsiveness to dystrophin gene editing therapy. Conclusions The current era in the management of CAD has witnessed a paradigm shift from an anatomic gold standard (ie, coronary angiogram) for determining the extent and severity of disease and thus the basis for clinical decision making to a

13 2192 Circulation May 31, 2016 Figure 6. Polar plots of myocardial 13 N-ammonia uptake (left) during adenosine vasodilation are shown in a patient with coronary artery disease and corresponding stress myocardial blood flow (MBF; right) at baseline (top) and after 12 months of treatment with pravastatin (bottom). At baseline, the relative distribution of the radiotracer (top, left) suggests single-vessel disease in the territory of the left anterior descending (LAD) artery, as would be the case when single-photon emission computed tomography studies are visually interpreted. Note, however, that the quantitative stress MBF (top, right) in all coronary territories is <2 ml min 1 g 1, which is consistent with hemodynamically significant focal LAD stenosis, in addition to a combination of diffuse coronary and microvascular disease in all 3 vascular territories. Normal vasodilator stress MBF is >2.0 to 2.4 ml min 1 g 1. The response to pravastatin therapy (bottom, left) shows that the extent of the stress-induced defect in the LAD vascular territory declined from 51% to only 3%. Stress MBF (bottom, right) shows a dramatic increase in each of the 3 coronary artery territories and, importantly, that stress MBF is normalized in the region with a previous stress-induced defect. Only measurements of absolute MBF, not the evaluation of the relative distribution of the radiotracer uptake in the myocardium, demonstrate the improvident in MBF in remote or normal-appearing myocardium. LCx indicates left circumflex artery; and RCA, right coronary artery. Reproduced from Schindler and Schelbert 32 with permission from the publisher. Copyright 2013 Springer. functionally oriented one. A number of invasive coronary pressure and flow measurements (alone or combined) are increasingly being used not only to judge CAD extent and severity but also to guide clinical decision making, which in turn is being evaluated on the basis of clinical outcomes. The invasive measurements have a number of well-recognized limitations, not least being their invasive nature and attendant risk, particularly when instrumentation of multiple coronary arteries is required for comprehensive, physiological evaluation of CAD severity. In brief, therefore, it makes sense to use a quantitative method for the noninvasive, comprehensive, physiological assessment of the coronary circulation both to guide clinical management and to better inform clinical and basic research in the field. The widespread availability of PET instruments and commercially available software for advanced image display (Figure 4) 17 and quantitative measurement of absolute MBF make this goal feasible now and, we hope, more so in the near future when an 18 F-labeled MBF tracer (eg, flurpiridaz) is approved for clinical use. Limitations to widespread implementation include administrative concerns (eg, instrument time and reimbursement) and the time and experience required for clinicians to move from a qualitative, defectno-defect, subjective analysis of myocardial perfusion images to an understanding and appreciation of quantitative physiological measurements of absolute MBF. It will be essential to do so to bring noninvasive MPI in line with coronary interventional practice, which is increasingly focused on objective, functional parameters of CAD severity, to guide management and to predict and improve patient outcomes. None. Disclosures References 1. Berman DS, Shaw LJ, Hachamovitch R, Friedman JD, Polk DM, Hayes SW, Thomson LE, Germano G, Wong ND, Kang X, Rozanski A. Comparative use of radionuclide stress testing, coronary artery calcium scanning, and noninvasive coronary angiography for diagnostic and prognostic cardiac assessment. Semin Nucl Med. 2007;37:2 16. doi: /j. semnuclmed Berman DS, Shaw LJ, Min JK, Hachamovitch R, Abidov A, Germano G, Hayes SW, Friedman JD, Thomson LE, Kang X, Slomka P, Rozanski