Myocardial Blood Flow Measurement by PET: Technical Aspects and Clinical Applications Philipp A. Kaufmann, MD1; and Paolo G. Camici, MD2 1Nuclear Cardiology Section, Cardiovascular Center, University Hospital, Zurich, Switzerland; and 2Faculty of Medicine, Medical Research Council Clinical Sciences Centre, Imperial College, London, United Kingdom the symbol F/W is also used, have units of volume per time The availability of plastic microspheres labeled with dif- per unit weight of myocardium (i.e., mL/min/g). ferent ␥-emitting isotopes has allowed the quantification of To achieve the accurate quantification of tracer uptake, blood flow in different organs, including the heart (1). In the which is a characteristic of PET, correction for photon past 3 decades this has made it possible to study the changes attenuation is crucial. At present, a rotating 68Ge (half-life ϭ in myocardial blood flow (MBF) under different physio- 287 d) source is commonly used for photon attenuation logic and pathophysiologic conditions in experimental ani- correction in oncology but also in cardiac viability and mals. perfusion scanning. 68Ge sources have only a low photon In a similar fashion, the development of PET has allowed flux, which necessitates long acquisition times of up to 20 the noninvasive quantification of regional MBF in healthy min (6,7), although with current methods transmission times humans and patients with different cardiovascular diseases. much shorter than 20 min can be used. This results in This article summarizes how PET measurement of MBF has increased chance of motion artifacts and poor anatomic contributed to advance our understanding of cardiac phys- image quality as well as patient discomfort and economic iology and pathophysiology. loss because of lower patient throughput (8). Introduction of combined PET/CT systems, where the CT scan can be used for attenuation correction, is expected to allow improve- METHODOLOGIC AND TECHNICAL CONSIDERATIONS ment of these limitations since a high-end CT scanner is Several techniques, including Doppler catheterization able to acquire images with far higher spatial resolution in and coronary sinus thermodilution, are available for mea- a much shorter time (Ͻ4 s) compared with a conventional suring coronary blood flow (CBF) in humans. These tech- 68Ge transmission scan (68Ge attenuation correction). The niques, which are fundamentally variations of indicator feasibility of adequate photon attenuation correction with dilution methods, are invasive and affected by serious lim- the CT scan of a hybrid PET/CT scanner has been demon- itations (2). Although CBF, for which the symbol F is often strated for static PET scans using 18F-FDG (9,10) and, more used, has units of volume per time (i.e., mL/min), Doppler recently, also for dynamic scanning with 13N-labeled am- 13 measurements usually allow assessment of flow velocity monia ( NH3)(11). (cm/s) and only few techniques provide volumetric flow (3,4). Measurement of inert tracer clearance, which can be TRACERS AND CAMERAS invasive (based on arteriovenous sampling) or semiinvasive Various tracers have been used for measuring MBF by (e.g., based on external detection of radionuclide washout, PET, including 15O-labeled water (H 15O) (12–17), 13NH after intracoronary injection 133Xe) provides estimates of 2 3 (18–23), the cationic potassium analog 82Rb (24,25), 62Cu- regional perfusion, although the mass of tissue subtended by pyruvaldehyde bis(N4-methylthio-semicarbazone (62Cu- the artery under study is unknown (2). SPECT allows the PTSM) (26–30) and 11C as well as 68Ga-labeled albumin noninvasive assessment of directional changes in regional microspheres (31,32), 94mTc- teboroxime (33), and 38K(34). tissue perfusion, but its physical limitations do not permit 13 82 Early PET studies used NH3 (18) and Rb (35) for qual- quantification of MBF (5). PET has been shown to allow 13 itative assessments of regional MBF. Currently, NH3, noninvasive and accurate quantification of regional MBF if 15 82 H2 O, and Rb are the most widely used PET perfusion suitable tracers are used and appropriate mathematic models 13 82 tracers. NH3 and Rb are given intravenously as boluses. are applied. These PET measurements of MBF, for which 15 In the case of H2 O, the tracer can be administered as an intravenous bolus injection (6,12,15,36), an intravenous 15 Received Apr. 15, 2004; revision accepted Sep. 13, 2004. slow infusion (6,37), or by inhalation of O-labeled carbon 15 15 For correspondence or reprints contact: Paolo G. Camici, MD, Medical dioxide (C O2), which is then converted to H2 Obycar- Research Council Clinical Sciences Centre, Imperial College, Hammersmith 82 Hospital, Du Cane Rd., London W12 0NN, United Kingdom. bonic anhydrase in the lungs (14). Generator-produced Rb E-mail: [email protected] is a very appealing MBF tracer because it does not require MYOCARDIAL PERFUSION QUANTIFIED BY PET • Kaufmann and Camici 75 a cyclotron on site and has a very short half-life (78 s). estimates obtained with the 2 tracers in healthy human Although several models have been proposed for quantifi- subjects. However, in a highly heterogeneous tissue (e.g., cation of regional MBF using 82Rb (25,38–40), they are jeopardized myocardium in patients with previous infarc- 15 limited by the heavy dependence of the myocardial extrac- tion), the diffusion/extraction and final uptake of H2 O and 13 tion of this tracer on the prevailing flow rate and myocardial NH3 are determined by the flow rates in each tissue metabolic state. Therefore, quantification of regional MBF compartment—that is, higher in viable tissue and lower in 82 13 with Rb may be inaccurate, particularly during hyperemia scar tissue. NH3 uptake (on which the model for the or in metabolically impaired myocardium. In addition, the computation of MBF is based) in a given region of interest high positron energy (3.15 MeV) of this radionuclide results will reflect the average uptake and, hence, average flow in in relatively poor image quality and in a reduced spatial this mixture of viable and fibrotic tissue. On the other hand, 15 resolution due to its relatively long positron track. since the uptake of H2 O in scar tissue is negligible, wash- 15 Several tracer kinetic models for quantification of MBF out of H2 O (on which the model for the computation of have been successfully validated in animals against the MBF is based) will mainly reflect activity in better-perfused radiolabeled microsphere gold standard over a wide flow segments and the resulting flow can therefore be higher than 15 13 13 range for both H2 O and NH3. The latter has also been that obtained with NH3 in the same region (54). validated against the argon gas technique in humans (41). The PET cameras used for the quantification of MBF, as Single-compartment models, based on Kety’s model for an well as for other cardiac PET applications, work in 2-di- inert freely diffusible tracer (42), are used for estimation of mensional (2D) mode with collimating septa between the 15 MBF using H2 O(12,14,15,43). A 3-compartment model detector rings to reduce the number of interplane scattered describing the kinetics of the myocardial metabolic trapping photons (55). A new generation of 3-dimensional (3D)-only 13 and whole-body metabolism of NH3 has been used for tomograms has become available with potential benefits calculation of MBF using this tracer (21,23,44–46). The over the 2D systems, particularly a higher efficiency. Re- 15 models have to include corrections for underestimation of cently, absolute quantification of MBF with H2 O and 3D radiotracer concentration due to the partial-volume effect PET has been validated in experimental animals (53). and spillover from the left chamber onto the ventricular myocardium (44,47), which result from the limited spatial resolution of the PET camera and the motion of the heart. APPLICATIONS OF PET TO STUDY CARDIAC PHYSIOLOGY AND PATHOPHYSIOLOGY Additional corrections have been developed to account for 13 the impact of flow (20) on myocardial extraction of NH3 Coronary flow reserve (CFR), the ratio of MBF during 13 and for the radiolabeled metabolites (48)of NH3, which near-maximal coronary vasodilation to basal MBF, is an accumulate in blood. integrated measure of flow through both the large epicardial 15 13 The equivalence of H2 O and NH3 as perfusion tracers coronary arteries and the microcirculation and has been has been demonstrated in experimental animals (49) and in proposed as an indirect parameter to evaluate the function of humans (50), but the proof of congruence of the tracers in the coronary circulation. An abnormal CFR can be due to ischemic and infarcted segments requires further investiga- narrowing of the epicardial coronary arteries or, in the 15 tion. The use of H2 O as a perfusion tracer is potentially absence of angiographically demonstrable atherosclerotic 13 15 superior to NH3 because H2 O is metabolically inert and disease, may reflect dysfunction of the coronary microcir- freely diffusible across capillary and sarcolemmal mem- culation. The latter can be caused by structural (e.g., vas- branes. Thus, it equilibrates rapidly between the vascular cular remodeling with reduced lumen-to-wall ratio) or func- and extravascular spaces and its uptake by the heart does not tional (e.g., vasoconstriction) changes, which may involve vary despite wide variations in flow rate. The short half-life neurohumoral factors or endothelial dysfunction. Further- (123 s) of 15O allows repetitive MBF measurements at short more, an abnormal CFR may also reflect changes in coro- intervals (10 min, equivalent to 5 half lives of 15O). Previ- nary or systemic hemodynamics as well as changes in 15 ously, an important shortcoming of the H2 O technique was extravascular coronary resistance (e.g., increased intramyo- its need for additional 15O-carbon monoxide (C15O) blood- cardial pressure).