PET Perfusion Imaging and Nuclear Cardiology

K. Lance Gould

Division ofCardiology and the Positron Diagnostic and Research Center, The University of Texas Health Science Center at Houston, Houston, Texas.

States (1,2). Up to 13% of middle-aged men in the general J NuciMed 1991;32:579-606 population have coronary atherosclerosis, most of it din ically silent (3,4). Silent myocardial ischemia is increas ingly recognized in symptomatic and asymptomatic mdi viduals (5—14).It has an unfavorable prognosis when uclear cardiology has evolved through several im occurring during exercise testing (7) in patients with recent portant stages. Initial methods utilized single or paired unstable angina (8) or in asymptomatic patients with coincident detectors to record a first pass radioangiogram perfusion defects after dipyridamole (11-14). Up to 48% or decay curves reflecting global myocardial perfusion. of asymptomatic subjects with silent ischemia have a The next stage evolved to planar imaging of myocardial cardiac event (angina pectoris, , or perfusion during treadmill exercise with thallium-201, sudden death)within 4—6yr(7,9). Fortyto 60%of patients which became a widespread clinical tool as did gated with sudden death or myocardial infarction pre blood-pool imaging for left ventricular function. Planar sent no previous symptoms (15—18).Therefore, silent imaging evolved to single-photon emission computed to coronary atherosclerosis remains a particular problem in mography (SPECT). Dipyridamole has been documented cardiovascular medicine because there are no warning as a substitute for exercise stress. Positron emission to symptoms until a major cardiac event occurs. mography (PET) of the branched off from earlier The major questions for cardiovascular medicine are developing brainPET at severalcentersutilizingcyclotron how to identify silent (CAD) in produced radionuclides developed originally for brain specific individuals, how to define its severity in either the studies. However,the complexity, expense, and low patient symptomatic or asymptomatic patient, how to decide ob volume of cyclotron-based PET has kept it primarily a jectively among dietary, medical, and mechanical inter researchtool at largeuniversitycenters.Nuclear cardiology ventions, how to assess the results itivasively and nonin is nowwell into its mostadvancedstageof development vasively, and how to reduce unnecessary invasive proce using an economical, generator-produced radionucide, dures by accurately selecting appropriate patients. At the rubidium-82, with fast PET scanners designed to acquire other end of the clinical spectrum, in patients with myo adequate counts for diagnostic images in the brief acqui cardial infarction and myocardial salvage by thrombolysis, sition times available for a short half-life tracer (75 see). percutaneous transluminal coronary (PTCA) This final step in advanced technology brings nuclear or bypass surgeryrequiresreliablemeasures of myocardial cardiology into a phase of widespread, readily available, viability and involvement of other coronary arteries to clinical PET, providing hard, independent data compa determine definitive mechanical interventions having sig rable in reliability and diagnostic value to the coronary nificant risks. arteriogram (1). This review outlines current clinical car diac PET as the mainstream evolution of nuclear cardiol RISK FACTORS FOR CAD ogy into routine, high-volume, generator-based, PET ap With widespread cholesterol testing. many individuals plications for accuratediagnosis ofcoronary arterydisease are identified as having risk factors for coronary athero or assessment of its seventy and extent of viable myocar sclerosis. However, risk factors do not accurately correlate dium as a basis for medical or mechanical intervention. with or identify individuals with symptomatic or asymp tomatic CAD. For example, two-thirds of men aged 40— SCOPE OF THE CLINICAL PROBLEM 55 with high cholesterol and blood pressure remain well Coronary heart disease causes 1.5 million myocardial over the subsequent 25 yr (19), whereasone-third develop infarctions and 520,000 deaths per year (one-third to one coronary heart disease. In asymptomatic males ofthis age halfofdeaths between the ages of35 and 64) in the United group with risk factors, coronary arteriography shows dis ease in 15%—35%(1—4,20—22)that is anatomically severe Received Sept. 15, 1990; revision accepted Jan. 17, 1991. in 7%—35%(1,2,21). Of asymptomatic middle-aged men For reprints contact: K. Lance Gould, MD, Box 20708, Room 4.258, Division of Cardiology,Uiiversity of Texas Medical School at Houston, Houston, Texas who have a positive ECG exercise test, only 30%—43% 77225. have significant CAD by arteriography(2) and 5%—46%

PETPerfusionImagingandNuclearCardiology•Gould 579 (mean 25%) develop clinical disease over the next 13—25radiation, is therefore accomplished electronically with years (2,19). Therefore, approximately two-thirds of men coincidence counting rather than solely with lead colli are resistant to clinical CAD despite having risk factors, mators, as in single-photon imaging. Coincidence counting whereas one-third have or develop it, often with only has severalattributesthat make PET uniquely quantitative modestly elevated or normal cholesterol levels. Conse and accuratefor clinical imaging, including accurateatten quently, high-risk patients commonly undergo coronary uation correction of emission data and electronic colli arteriography at some risk and considerable health care mation that provide higher efficiency, more counts, and costs in orderto confirm normal coronary arteries,partic better statistics, all of which are the basis for better spatial ularly after false-positive exercise testing. A substantial and contrast resolution than single-photon systems. number of patients with risk factors are treated with ex A positron camera has four components: the scintilla pensive lipid-lowering drugs at some risk, despite a large tion detector-photomultiplier tube (PMT) modules, the proportion of such patients who do not have or will not gantry housing these detectors with its patient pallet, the develop coronary atherosclerosis. electronics, and the computer systems. The design of each of these compounds and how they interrelate is essential PET for diagnostic accuracy, patient volume throughput, and for using ultrashort half-life, generator-produced82Rbfor Cardiac PET accurately identifies and assesses severity routine clinical studies without the expense ofa cyclotron. of coronary artery stenosis and myocardial viability as a Specific essential design features of a PET scanner for basis for choosing and following effects of interventions, optimal clinical performancewith 2Rbareoutlined below. including risk factor management, pharmacologic agents, Generically, PET cameras contain 1000-iS® detectors PTCA, thrombolysis, and bypass surgery. Accurate non in 3-8 banks of rings, attached to PMTs in ratios ranging invasive identification and quantitation are particularly from 1 to 8 detectors for each PMT. Scintillations from important in silent CAD as the only criteria for interven the coincidence detectors cause electronic signals from the tion to prevent sudden death or acute myocardial infarc PMTs that are converted to digital information and proc tion. essed in a computer to reconstruct a tomographic image Cardiac PET is not only a reliable guide to managing like a CT scan. However, the source ofradiation is from a CAD in traditional cardiology practice based on symp positron tracer, not an external x-ray tube. In order to toms; it also provides the basis for vigorous medical or optimize spatial sampling, the banks of detectors are wob mechanical management of asymptomatic coronary ath bled in an eccentric path around the subject with each erosclerosis. Such accurate nomnvasive evaluation by PET coincidence decay assigneda spatiallocation in the wobble is intended to identify individuals who need cardiac cath path. A transmission image for attenuation correction is eterization and to avoid unnecessary procedures in patients obtained by placing a ring of activity around the patient with either mild or no coronary artery disease suitable for forimaging the targetorganbefore injection of radiotracer. reversal therapy, as subsequently reviewed. Cardiac PET The positron radiotracer is then injected intravenously interfaces positively with invasive cardiology by making and an emission image obtained by back-projection tech its application more efficient and selective. niques. The emission image is corrected for attenuation Physiologic basis, technology, major clinical appica loss, random coincidences, scattered radiation, dead-time tions, and extensive case studies have been recently de losses, wobble, and variation in detector sensitivity. A scribed in the firsttext of cardiac PET (1). quantitative three-dimensional image of the radiotracer activity is obtained for the imaged organ. BASIC PET PRINCIPLES Quantitative PET imaging refers to reconstruction of A positron is a positively-charged electron emitted by tomographic images that quantitatively reflect or recover unstable atoms in the process of radioactive decay, such the actual or true distribution ofactivity in the targetorgan as generator-producedrubidium-82 or cyclotron-produced or field of view undistorted by tissue attenuation, which is carbon-l 1, oxygen-iS, nitrogen-13, and fiuorine-l8. This corrected by the transmission scan. This type of quantita positron travelsseveralmm through tissue and annihilates tive data recoveryalso depends on the technical character with a negative electron, giving off two 51 l-keV photons istics ofthe camera and the duration ofimaging permitted by the half-life of the positron tracer. in opposite directions. The annihilation photon pair can be detected with a pair of radiation detectors connected through a coincidence-counting circuit, so that one decay POSITRON-EMITtiNG RADIONUCLIDES (uRb) is recorded if both detectors are activated simultaneously Rubidium-86 was the first rubidium radiotracer used by the photon pair. for coronary flow studies and therefore has the longest Radioactive decays occurring outside the sample vol scientific history of all cardiac radiotracers (22). Subse ume between the detectors are excluded from the count quently, thallium-201 and technetium-99m were devel data since an unpaired photon striking only one of the o_ for single-photon imaging. Only relatively recently detectors is not counted. Collimation, or exclusion of stray was cardiac PET in man first carried out with cyclotron

580 The Journal of •Vol. 32 •No. 4 •April 1991 produced ‘3N-ammonia and [‘F]fluorodeoxyglucose eluted volume to be delivered, flow rate, dose to be deliv (FDG). However, as with most mainstream technologic ei-ed,dose rate, and patient volume to be delivered. developments, these important initial radionuclides have been followed by a simpler, more economical, generator Comparison of asRband 13N-ammonia produced tracer, 82Rb.Rubidium radiotracers are analogs Rubidium-82 has several unique characteristicsthat are ofpotassium (and 201Tl),the biologic behavior ofwhich is optimal for cardiac studies. Because of its short half-life, more extensively described than any other cardiac radio several sequential studies may be repeated at one sitting, nudlide.Forwidespread,routinecardiacpractice,a simple such as before-and-afterdipyridamole, PTCA, bypass sur to-use radionuclide source such as the 2Rb generator is gery, drug therapy, episodes of acute ischemia, or serially essential since it does not require a cyclotron (23,24) but following myocardial infarction. A rest-dipyridamole82Rb allows all of the major aspects of for study requires 55 mm compared with 2.50 hr for a study clinical purposes as outlined below. with cyclotron-produced 13N-ammonia.Consequently, pa Rubidium is an alkali metal analog of potassium and is tient throughputis highwith 2Rb,demonstrated in private similar in chemical and biologic properties. It is rapidly practice sites to be nine patients per day. With new atten concentrated by the myocardium with a first-passextrac uation correction methods that shorten the attenuation tion of 50%—60%at resting flow levels, which falls to scan to 10 mm (31), the time required for a rest-dipyri damole pairofscans is 40 mm, therebyallowing 15 patient 25%—30%at high coronary blood flows (25-27) By com _son, first-pass myocardial extraction of ‘3N-ammonia studies in a 10-hrshift. Since 82Rbis generator-produced, is somewhat higher at 70%, falling to approximately 35% it avoids the capital costs of a cyclotron and decreases the at high flows (28-30). With appropriatelylargerdoses of cost of radionuclide per study. Finally, 2Rbprovides data 82Rb(50 mCi), imagesare comparableto thoseof ‘3N-not only on myocardial perfusion but also on myocardial necrosis and viability simultaneously by processing the ammonia (18 mCi) (1). images in different ways during a rest-dipyridamole study Because of its short half-life (75 see), 82Rbis the radio (1). Thus, a rest-dipyridamole-viability study with 2Rb nuclide of choice for repeated or sequential myocardial requires only 55 mm compared with 4—5hr for obtaining imaging. It is particularlyuseful in acute clinical situations comparable data with cyclotron-produced ‘3N-ammonia in which the patient's condition is changing rapidly, or for and FDG. studies before and after an intervention such as dipyrida mole or PTCA. Rubidium-82 is created from its parent, strontium-82, CLINICAL SUMMARY OF CARDIAC POSITRON in a generator that uses hydrous 5n02 as the inorganic EMISSION absorbent. This generator has negligible 2Srbreakthrough Clinical and experimental studies currently supporting level, can be efficiently eluted with isotonic saline of the clinical applications for cardiac PET are listed below physiologic pH, and delivers 2Rbin a small eluate volume. (1). The use of short-lived, positron-emitting 2Rb in cardi ovascular nuclear medicine requires a system for rapid 1. Noninvasive Diagnosis of CAD in Either Sympto elution of an 2Rb generator and on-line injection of the maticorAsymptomaticPatients.The sensitivityand generator eluate. The 2Rb infusion system incorporates specificity of diagnosing CAD by PET (1,32-39) features for efficient elution and precisely controlled deliv compared with automated quantitative coronary ar ery of a uniform, non-pulsating sterile solution of 2Rb in teriography are both 95%—98%in symptomatic or saline. The system has preformed, sterile, disposable plastic asymptomatic individuals, including “balanced― tubing, one-way valves, and a syringe that are electrome three-vessel disease (40,41). False-negative studies chanically-actuated to provide the eluent flow by positive are usually associated with distal coronary stenosis flow pump action. The syringe plunger is bidirectionally or disease of smaller arteries;false-positives are usu actuated by a low-friction, recirculating ball screw jack ally associated with nonatheromatous coronary ar that is coupled to a stepper motor through a reduction tery abnormalities such as spasm or thrombi (1,32, gear. The speed of the motor driving the syringe plunger 35). Table 1 lists the major studies from five different and infusion rate is regulatedautomatically by a radiation centers showing the accuracy of PET perfusion im monitor in line with the infusion tubing to deliver a aging. This application as compared with SPECT is predetermined dose and dose rate. discussed in more detail subsequently. The components of the system—pump,fluid tubing, 2. Assessmentof the PhysiologicSeverityof Coronary filters, one-way valves, electronic control modules, Artery Stenosisor Changesin StenosisSeverity.As shielded vault for the 2Rbgenerator, shield for the waste previously reviewed (1), percent diameter narrowing container, and dosimeter—are self-contained on a mobile is not an adequate standard for quantifying stenosis cart. The elution and injection procedure is fully defined severity of coronary artery narrowing (1,32-35,40- by the settings of the controls on the electronics modules 50), as also confirmed by others (51-54). Percent for the infusion and dosimetry systems. Settings include stenosis does not account for the effects of diffuse

PET Perfusion Imaging and Nuclear Cardiology •Gould 581 TABLE 1 SpecificityDiagnostic UpdatedSensitivityand

testAccuracyPatientsReferenceDiagnosis

of CAD arterydis 83% 47% Train(107) easebythalliumstresstesting 85% 52% 1096 Ranhosky(108) since1983Sn 95% 71% 210 DePasquale(106) 76% 49% 832 Schwartz(113) 82% 62% 461 lskandnan(109) 94% 52% 81 Bungo(114) (111)Average 94%Sp 44%197 242Van VanTrain we@htedfor#cases84% 67.5% 75% (118) (116,117)Diagnosis 65%53% 65%@Corrected@“CorrectedDiamondGould of coronaryartery 00% (32) stenosis using ssRb @j+J@a fast 94% 95% 193 Darner (35) PET scanneror 1@NH395% 97% 100% 32 Schelbert(36) 97% 100% 49 Yonekura (37) (119)Average 98%1 93%50 146Gould Williams weightedfor#cases96%96%

disease, eccentricity, stenosis length, absolute cross ing. By definition, degradingquantitative data into a sectional area, entrance and exit shape on flow, or binary classification is necessarily incorrect because flow capacity. It is also limited by substantial inter the inherent quantitative characteristicsof the data and intraobserver variability. Alternative invasive are eliminated by the arbitrary binary classification approaches providing fluid dynamically correct analysis. Therefore,analyzing PET data according to measurements of graded stenosis severity utilize a binaryclassificationbasedon percentstenosisde quantitative arteriographic methods to calculate ste grades the PET data to a nonquantitative status. nosis resistance (51), pressure-flow curves (49,50), Current literaturedoes not report a quantitative re or coronary flow reserve(32—43). lation of SPECT to stenosis severity by quantitative The quantitative severity of PET perfusion defects arteriography. after dipyridamole reflect the anatomic severity of 3. ImagingMyocardialIschemia,Infarction,and Via the coronary artery stenosis as determined by auto bility. The location and extent of myocardial infarc mated, quantitative coronary arteriography, taking tion (55—57,61—70)and myocardial ischemia or vi into account length, absolute dimensions, shape, and ability (44,55—57,61—71)may be imaged by positron percent narrowing of the stenosis (1,32—35,40—43). emission tomography with or without reperfusion Therefore, noninvasive positron emission tomogra using generator-produced 82Rb or cyclotron-pro phy provides assessment of the physiologic severity duced ‘3N-ammoniaand [‘8FJFDGin order to deter of coronary stenosis as well as changes in severity mine if bypass surgery or PTCA is indicated. This after an intervention such as thrombolysis (55), by application has been recently reviewed (1) as sum pass surgery (56,57), PTCA (58), or risk factor mod marized subsequently. ification (1,59,60). With accurate quantitation of 4.IdentificationandAssessmentofSignificantCollat stenosis severity, binary classification of coronary eral Functionin Man by Imaging CoronarySteal artery disease based on sensitivity and specificity DuringDipyridamole-Handgrip.Coronarystealoc becomes inadequate or misleading for validating or curs under conditions of near-maximum coronary describing diagnostic accuracy. This is one of PET's vasodilation if collaterals provide a significant pro unique strengths,demonstrable by comparison to an portion of resting myocardial perfusion. A fall in appropriate arteriographicgold standard that takes absolute myocardial activity after injection of a per into account all stenosis dimensions. Comparison of fusion tracerduring dipyridamole vasodilation com two noninvasive methods, e.g., PET versus SPECT, paredwith rest indicates coronary steal and the pres using arteriographicpercent diameter stenosis as the ence of significant collaterals (72—74).Because it is standard, simply cannot be done because percent physiologically complex and uniquely assessed by stenosis has little relation to coronary flow reserve PET, myocardial steal is addressed in more detail (32—35,40—43,52—54)except for very severenarrow elsewhere (1).

582 The Journal of Nuclear Medicine •Vol. 32 •No. 4 •April 1991 5. DilatedCardiomyopathy.This conditionisunrelated of normal maximum. This capacity makes PET the most to CAD and may also be diagnosed by positron sensitive and specific nuclear cardiology procedure for imaging as an enlarged, poorly functioning heart with coronary atherosclerosis. no resting or stress perfusion defects typical of is The ratio of radionudide activity after dipyridamole to chemic cardiomyopathy due to coronary artery dis activity at rest reflects absolute flow reserve. The ratio of ease. activity in a perfusion defect to normal maximum activity 6. Left VentricularFunctionand Wall Thickening.By in distributionof normalcoronaryarteriesonthedipyrid gating the positron-emission tomographs with the amole image reflects relative coronary flow reserve. With ECG, left ventricular pumping function and wall quantitation of PET perfusion imaging to account for thickening may be assessed regionally in three di arterial input function and flow-dependent radionuclide mensions(32,75). Sincethereare many other less extraction, absolute and relative coronary flow reserve may expensive, simpler measures ofleft ventricularfunc be quantitatively measured by PET (40). tion, gated PET has not been developed clinically. Absolute and Relative Coronary Flow Reserve However, since PET is recognized as a noninvasive The concept of coronary flow reserve,defined as maxi substitute for diagnostic catheterization, recording mum flow divided (normalized) by resting control flow, first-passgated blood-pool for LV function as a part has evolved into an accepted functional measure of ste of routine perfusion studies will likely become a nosis severity since first proposed by Gould (77—80).Its common method to obtain complete cardiac evalu validity has been confirmed and applied clinically by non ation comparable to catheterization for the basic invasive imaging (1,32—36,40,45—50,76—89)and by inva questions of arterial patency, myocardial viability, sive methods such as coronary sinus thermodilution (90), stressunderperfusion,and left ventricularfunction. Doppler-tip catheters (91—93),and digital subtraction an giography (94,95). These clinical methods measure phar PHYSIOLOGICPRINCIPLESOF PET PERFUSION macologically-induced increases in coronary blood flow, IMAGING most commonly with i.v. dipyridamole for noninvasive With marked increases in coronary flow, functionally studies and intracoronarypapaverine for invasive studies. mild, earlycoronaryarterystenosis can be identified before Coronary flow reserve has also been integrated theoret symptoms of myocardial infarction occur, thereby allow ically with, and experimentally related to, the geometric ing medical therapy aimed at reversal of coronary athero sclerosis. However, with only modest increases in coronary flow as duringexercisestress,only severecoronarystenoses can be identified.An inadequatestimulusfor increasing IE$TN WIE SIF@T coronaryflow or an inadequateperfusionimagingagent @ or technique will limit the ability to detect or quantify soct/@.f1OO @ coronary stenosis by perfusion imaging. Based on animal eo @@iiniioo@m studies (40,45—50,76)and confirmed in man (1,32,35,47, 48), there are three essential principles of an imaging method for detecting and assessing severity of coronary AescFR=PEST@( .2nd4 artery disease: (a) cross-sectional PET to avoid overlapping structures and to obtain depth-independent resolution for accuratequantitative images of coronary blood flow; (b) a 16O@/mei/1OO,m._—. @—32Oa/@i#9ISp myocardialperfusion imaging agent taken up or deposited 4_UN ‘ in myocardiumin proportionto flow at high coronary blood flows up to five times resting control levels; and (c) a potent stimulus for increasingcoronary flow to image @/ the regional disparities in maximum perfusion caused by REt. CFR=@@ —0.5 stenosis. Figure 1demonstrates these principles. In the top IMAGEDEFECTSOSOF MAX panel, resting coronary flow and distribution are normal FIGURE1. Schemademonstratingtheprinciplefordetection despite a severe stenosis. After i.v. dipyridamole, flow ofcoronaryarterydiseasebyperfusionimagingunderconditions increasesfour times base-line (middle panel) in the normal of maximumcoronaryvasodilation.Withan80%diameternar area but is restricted to a two times increase in the diseased rowing,coronaryflow reserveis limitedto approximatelya two area. There must be an average ratio of 2.4:1 or 150% times increase over resting levels as compared to a four times difference between normal and affected areas to visibly the increasein non-stenoticarteries.The abnormalarea, there fore, has 50% less activity, reflectinga 50% decreasein maxi detect a relativeperfusion defect on planarthallium image mumflowcomparedtonormalmaximum.ABSCFRisabsolute in experimentalanimals.With PET, a differenceof only coronary flow reserve and REL CFR is relative coronary flow 15% can be detected, i.e., an abnormal area that is 85% reserve.

PETPerfusionImagingandNuclearCardiology•Gould 583 dimensions of stenosis (32—35,40,82—84).The commonly arteries. When the maximum flow in the stenotic artery is used percent diameter stenosis is poorly related to func normalized by normal maximum flow, the term for mm tional severity of human coronary artery narrowing or imum bed resistance reflecting effects of pressure, (heart coronary flow reserve(32-35,40,52—54,82—84,92—96),de rate or vasomotortone on flow in the numeratorand termined by quantitative coronary arteriography as in denominator of this ratio) are cancelled (40). Therefore, Figure2 (35) or by directly measured flow reserveusing a relative differences in regional maximum flow, or relative Doppler catheter (93). Thus, extensive experimental and flow reserve, are determined primarily by geometric ge clinical literature have substantiated the concept of coro neticstenosisseverity.Relativeflowreserveisthereforea nary flow reserve. measureof stenosisseverityindependentof physiologic However, changes in aortic pressureand heart rate are variables. known to alter cardiac workload and therefore baseline Rather than considering absolute coronary flow reserve coronaryblood flow as well as alteringmaximum coronary to be competitive or antithetical to relative flow reserve, flow underconditions of maximal vasodilation (40). Con our data indicate that these measurementsareindependent sequently, absolute coronaryfiow reserve, as measured by variablesprovidingcomplementaryinformation.Absolute flowmeter, also varies with aortic pressure and heart rate flow reserve is the flow capacity of the stenotic coronary independent ofstenosis geometry due to differentialeffects artery and vascular bed under any conditions of pressure, of these variableson resting and maximal coronary flow. workload, hypertrophy, vasomotor tone, or stenosis. It Under markedlyvaryingphysiologicconditions,or from reflects the cumulative effects of these various factors patient to patient, absolute coronary flow reservemay not without being specific for the mechanism or cause of reliablyor specificallyreflectseverityof coronaryartery alteredflowreserve.Relativecoronaryflowreservereflects narrowing since it may be altered by physiologic factors morespecificallytheeffectsofthe stenosisindependentof unrelatedto stenosis geometry. and unaffectedby the other physiologic variablesif normal In contrast,relativemaximumcoronaryflowor relative maximum flow is high enough. Thus, absolute and relative flow reserve,defined as maximum flow in a stenotic artery coronary flow reserve are complementary. divided (normalized) by the normal maximum flow in the “Balanced―three-vessel coronary artery disease may absenceof stenosis,shouldtheoreticallybe more inde theoreticallycausea false-negativestressperfusiontest pendent of aortic pressure, heart rate, or varying baseline depending upon the stress stimulus and the imaging tech flow caused by changing cardiac workload. Physiological nology, particularly in spatial and contrast resolution. variables unrelated to stenosis severity, such as aortic Based on our experience with PET, most affected pressure, heart rate, metabolic demand, or vasomotor by CAD containsomeunaffectedarterythat servesasa tone, alter distal coronary bed resistance in series with, referencearea.However,small-vesseldisease,left ventric and independent of, proximal stenosis resistance. During ular hypertrophy, or theoretically “balanced―three-vessel maximal coronary vasodilation, distal coronary bed resist coronaryartery diseaseare potentialcausesof diffusely ance is equally minimized for both normal and stenotic impaired flow reserve that must be accounted for in mdi vidual patients. For this purpose, some measure of absolute as well as relative coronary flow reserve is necessary. STENOSISCFR VS. % DIAMETERSTENOSIS 5@ (C@ 79825. 1959) However, relative flow reserve is also necessary since it is not affected by variability in pressureor heart rate within thesamepatientor amongpatients. ClinicalImplicationsof Absoluteand RelativeFlow Reserve @:______A stressmyocardialperfusion image shows relativemax imum perfusion (radiotraceruptake) or relative coronary @ 2 F91$•P0s9i95$•.. @ @i:,,@22% flow reserve. One of the purported limitations of radio nuclide perfusion imaging is its inability to measure ab solute flow and absolute coronary flow reserve. However, perfusion imaging of relative maximal coronary flow or relative flow reserve has been useful for assessing physio S DIAMETER NARROWING logic stenosis severity, despite greatly changing or widely FIGURE2. Graphoftherelationbetweenstenosisflowreserve variant physiological conditions seen between rest condi andpercentdiameternarrowing,bothdeterminedfromauto tions and uprighttreadmill exercise, bicycle stress, supine mated quantitative arteriographic measurements in 100 patients. exercise,or variouspharmacologicstressessuch as dipyrid Becausepercentdiameternarrowing is only one of severalfac amole and papaverine. The results of our study explain torsusedto calculatestenosisflowreserve,thescatterinthis relation indicates the importance of those factors other than this observation. The stress perfusion image shows relative percent stenosis that influence flow impedance of a stenosis. coronary flow reserve, which is independent of physiologic Reproducedwith permissionfrom reference(35). variables of blood pressure and heart rate. Thus, what has

584 The Journal of Nudear Medicine •Vol. 32 •No. 4 •April 1991 long been considered the limitation of stress perfusion True short axis views are shown in Figure 3A; horizontal techniquesisin factitsgreatestadvantage. and vertical long axis views are shown in Figure 3B (41). In order to be optimally effective over the wide range of Rest images are uppermost in each pair of image rows conditionsseenclinically,a methodfor assessingphysio (study 1). The dipyridamole images are in the lower row logic stenosis severity should provide measures of both ofeach pairofimage rows (study 2). The number afterthe relative and absolute flow reserve.For example, measure decimal is the image plane for both studies 1 and 2. In the ment of absolute flow reserve in a coronary artery by a color coding,white indicatesthe highestflow, red next Doppler-tip catheter after adenosine or papaverine mdi highest,yellow intermediate,andgreenandbluethe lowest cates the effects ofsingle-, multiple-vessel, or diffuse CAD. relative flows. The tomographs in the horizontal long axis However, this may change somewhat due to markedly views (top images of panel 3B) are oriented as if looking varying afterload and baseline pressure-rate product or down from above with anterior or apex at the top of each physiological differences among patients that affect flow image, left lateral free wall on the left, and muscular reserve separate from stenosis geometry. On the other septum on the rightwith AV ringand/or inferior myocar hand,relativecoronaryflowreserve,asassessedbycurrent dium at the bottom. In true short axis views of panel 3A standard thallium perfusion imaging, does not adequately the image planes are arranged from the AV ring at the reflect the absolute flow response to vasodilator stimuli or upper left to the apex at the lower rightwith anteriorwall diffuse atherosclerosisaffecting all coronary arteries.Most being up, free LV wall on the left, and septum on the right invasive or noninvasive clinical methods as now used of each tomograph. The open “C―sin the basal short axis provide measurements of either absolute or relative coro views in the upper left images of Figure 3A are caused by nary flow reserve, but not both, with the exception of the membranous septum of the LV cavity, which is avas positron emission tomography. cular and therefore appearsasa defectbut reflectsnormal Therefore, the optimal noninvasive method usesrelative anatomy. coronary flow reserve to assess physiologic stenosis severity The resting tomographs show a small apical defect in and absolute flow reserve to assess response to dipyrida dicating a small myocardial scar. With dipyridamole the mole, small-vesseldisease, left ventricularhypertrophyand anterior, septal, and apical myocardium show a defect in “balanced―three-vessel disease, as well as single or multi both short and long axis views. The inferior and lateral ple coronary artery stenosis of varying severity. myocardium do not show a definite stressperfusion defect Absoluteand relativecoronaryflow reserveare mdc on tomographic views. However, the entire set of data pendent, complementary variables, which together more analyzed together on the polar coordinate maps suggests completely describe physiologic severity ofcoronary artery abnormal relative flow reserve(lower left polar coordinate narrowing than either alone. These conclusions have im map) and absolute flow reserve(right upper polar map in portant implications for clinically assessing physiologic the bottom half of the image). stenosis severity by noninvasive perfusion imaging using Figure 3B shows horizontal (left) and vertical (right) positron emission tomography. long-axis views of the heart. Rest images are shown in the Finally, Figure 2 also shows why a single threshold top row with dipyridamole stress images in the lower row percent diameter stenosis is a poor gold standard of disease ofthe top halfofthe figure.The horizontal long axis views severity for comparison to perfusion imaging. For a hy areorientedasif lookingdownfrom above.The vertical potheticallyperfectimagingtechniquethat measuredcor long-axis tomographs are oriented as if looking at the left onary flow reserve exactly as by a flowmeter, false-positive side of the body cut head to toe; anterior myocardium at and false-negative tests would be 22% and 23% for a the top, inferior at the bottom, apex at the left, and AV threshold of 60% diameter stenosis. Thus, perfect physio ring on the right. logic measures of coronary flow reserve would appear Tomographic data are summarized in a polar display as diagnostically poor with a sensitivity and specificity of if looking at the patient from the outside toward the LV 77% and 78%, dueto the inadequacyof usinga percent apex located in the center of the bull's-eye with the outer stenosis threshold as a gold standardof significant disease. rim of the bull's-eye corresponding to the AV ring. Polar displays on the left (lower half of figure)show the relative INDICATIONSFOR CARDIAC PET BY DIAGNOSTIC activity on a scale of0% to 100% with rest being the upper CATEGORY (Si) anddipyridamolestressbeingthelower(52)of the polar maps on the left side of the panel. The upper right Examples of Physiologic Diagnosis by PET polarmap (in the lower halfofthe figure),labeled absolute Angina Pectoris. Assessing severity, extent, location of S2/Si ratio (ABS 52/51), shows the absolute counts of coronary artery stenosis, and status of collaterals. Figure 3 stress image divided by rest image, displayed on a scale shows rest-dipyridamolePET of generator-produced82Rb from 0 to 2, reflectingabsolute coronary flow reserve. in a 61-yr-oldman with angina pectoris needing evaluation Increase in activity on a dipyridamole image compared for severity/extent of disease and for following regression with rest is shown by warm colors (yellow, orange, red, or progressionof coronaryarterydisease. white) indicating ratios greater than one (blue), which

PETPerfusionImagingandNuclearCardiology•Gould 585 @ A@ @*1 .@ @. -@

ACT' 9* II.

C

@ 1@ Tv@t •@W9511 V$995 - @$T. $19595 1@ ?1@E . ?uPoo@wic*. @I- 9511 . $19595 @ 195111U?195 WU@$95U@f CF 1V@ 09595 95IO@ @95 INS?IIU?195 95@@95$@?Y OF 1E95 95951$ C2@$ cV@?95 @ $nev P951@I 9IWV9 $1,.cv P95I@* $1@0Ya @ @1 • 9595$ . — — @ I. •‘c'@@.@ S. p'c,.-'p. S —@--95 195 ‘aLA?I 95 IN @ @‘ i@\, @• @,. @ c@, @ :?@* S B @ SIWV S 9.9 LA $1J43 @95i S. P9 l9@ @. @95 c@uø@s @s $95 S S 95sauTE @@$$95120 FiGURE3. Rest-dipyridamolePET imagesusing @Rbof a patient with three-vesselcoronary artery disease. See text for 95 ,,‘4$ det@lsof the display,which are identifiedhereas follows. 95 95 1 (A)[email protected] @ SI @t'rj;:@i lower row is the dipyridamole image set. @95W 51595Y S 95 •S 95L*T$@ $$‘S$ 95720 @ 95 •@ p_@_',@ (B)l-Iorizontalandverticallong-axisviewswith polarmapdisplays @ 2$ •4 S ofminimumactivityineachquadrantandwholeheart. @ 1$ $9 p , j@.,4,1T (C)Three-dimensionaltopographicdisplaysOfthe heart in septal, • 9S @1... , @2 95720 $2. $1 95,92 anterior, lateral, and inferior @ñews(see text for detalls). Repro ducedwfth permissionfrom references(1) and (41).

reflect increased radiotracer uptake and perfusion due to in relative activity ofthe anterior, septal, and apical areas, dipyridamole. Beside the scale for absolute 52/51 radio but also a mild decrease in the inferior septal area not traceruptake ratio (stress/rest) is a scale of coronary flow apparent on tomographic views but confirmed by the reserve values derived by a two-compartment model ac minimum activity decreasing on 52 (stress) compared with counting for flow-dependent extraction derived from am 51 (rest). The lateral wall also shows a visual and quanti mal studies. The lower rightpolar display, labeled relative tative decreasein activity reflectinga mild relativedecrease 52/51 ratio (REL 52/51), shows the relative distribution throughout the lateral quadrant not apparent on the to of flow after dipyridamole divided by the relative distri mographic views. On the polar map on the upper right bution of activityon the restimage(relativeinsteadof (Fig. 3B), absolute S2/Si ratio shows that at least one part absolute values), also on a scale from 0 to 2. It therefore ofthe heart located inferior-laterally responded with a flow maps the relative change in activity from rest to stress. reserve of 2.8 (times baseline). Flow reserve in the rest of Letters and numbers beside each polar map show quan the heart was severely depressed, indicating three-vessel titative results. For regions of the heart, A = anterior, S = disease with the LAD proximal to the first septal perforator septal, X = apex, L = lateral, I = inferior. The numbers being most severely affected, and disease ofthe left circum beside each region indicate the minimum activity as a flex and right coronary arteries. The relative flow reserve percentofnormal areas(100%). For letterswith a bar, i.e., defect on 52 is not as apparent as the marked defect of A,thenumbersindicatemeanactivityofallpixels(rather absolute flow reserve on 52/51, consistent with the discus than minimum) for that quadrant of the polar map as sion of the previous chapter. percent ofnormal areas (100%). In addition, partsofthe anteriorseptum and apex show In the example in Figure 3B, visual inspection of the a decrease in absolute counts with an absolute ratio of less polar map display and quantitative analysis show consid than one on the absolute 52/51 ratio polar map. A fall in erably more information than obtained from visual in absolute activity after dipyridamole compared with rest spection of the tomographic views alone. The polar map indicates myocardial steal and hence the presence of col 52 ofrelative flow reserveshows not only a severe decrease laterals to viable myocardium. On the 52/51 absolute ratio

586 The Journal of Nuclear Medicine •Vol. 32 •No. 4 •April 1991 polar map, a large part of the whole myocardium shows hum treadmill test (40). The short-axis rest-dipyridamole myocardial steal and is therefore viable, collateralized views(Fig. 4A) are normal. The polar map of dipyridamole myocardium, mostly in the anterior, septal, and apical study(4B, 52, lowerleft polarmap)reflectsregionalrela regions. Thus, the location, intensity, size, statistical sig tive coronary flow reserveon the same scale of 0%-lOO% nificance, and presence of collateralized viable myocar and is abnormal. Everyquadrantin the dipyridamole polar dium can be automatically quantitated. map shows a fall in relative minimum intensity from 82% In addition to standard tomographs and polar maps, to 73% of normal in the anterior(A) quadrant, from 73% new three-dimensional (3-D) topographic displays of car to 65% in the septal (S) quadrant, from 68% to 50% in diac PET are shown in Figure 3C. These 3-D topographic the apex (X), and from 74% to 67% in the inferior quad displays do not distort the spatial size and shape of defects rant with the high lateral quadrant showing no relative as do polar map displays. They therefore provide more defects, 67% to 66%. However, the relative flow reserve accurate spatial quantitation and visualization of abnor image, 52, does not show a localized regional or sharply malities compared with previously shown polar maps (1). circumscribed relative flow reserve defect. Three-dimensional topographic maps of cardiac PET im The upper right polar map labeled ABS S2/S 1 shows ages are derived from short-axis data. Each panel of the the absolute activity in the stressstudy (52) divided by the topographic map views the left as if looking at absolute activity in the rest study (51) expressed as absolute the septum (first panel), the anterior wall (second panel), coronary flow reserve using a model accounting for radio the left lateral wall (third panel), or the inferior wall (fourth nuclide extraction, dose, and cardiac output changes. In panel). The rest study is shown in the upper row (51) and this instance, the maximum regionalcoronary flow reserve the stress study in the lower row (52). The dashed white located in the anterior-lateral area of the heart in response line marks the upper limit ofautomated quantitative data to dipyridamole was 40 (times resting levels), indicating since the membranous septum often causes a normal that this patient respondedwell to dipyridamole in a small defect at the AV ring. The black dashed lines delineate artery to the high anterior lateral myocardium. Since flow septal, anterior, lateral, and inferior quadrants with the reserve was reduced to every quadrant of the heart except lowerdashed line delineatingthe apex. The colored scaling to this high anterior lateral area, the PET scan was inter and automated quantitation, such as minimum activity preted as showing mild to moderate “balanced―three values, mean activity in each quadrant, and percent of vessel disease without knowledge of clinical data or arte myocardium beyond 1.5, 2.0, and 2.5 standarddeviations riograms. Thus, despite “balanced―three-vessel disease are all the same as previously described for polar maps. affecting all major arteries of the heart, at least some part The rest 3-D topographic display (Fig. 3C) showed a of the myocardium responded well to this vasodilator mild small apical defect suggesting a small apical scar. The stimulus as a normal referencearea. By using measures of dipyridamole scan showed severe defects of large size in both relative and absolute coronary flow reserve, “bal the anterior, septal, apical, anterior lateral, lateral, poste anced―three-vessel disease was identified. Coronary arte nor, and inferior myocardium (1). A small segment of riography confirmed 50%—60%diameter narrowing of the inferiorlateralmyocardiumdemonstratedincreasedup three major coronary arteries. This patient began vigorous take of tracer indicating response to dipyridamole. dietary and pharmacologic cholesterol lowering that would An extensive area, approximately one-third of myocar not have been otherwise started without a PET study. dium in anterior, apical, and inferior apical distributions, Corresponding 3-D topographic displays in Figure 4C demonstrated decrease in activity during dipyridamole show the rest scan in the upper row (51) and dipyridamole compared with rest, suggesting myocardial steal and col scan in the lower row (52), demonstrating mild defects of lateralization. These images were interpreted as showing septal, anterior, apical, lateral, and inferior areas. occlusion of the LAD and right coronary arteries as com Atypical ChestPain. The patient was a 53-yr-old man monly seen with such extensive restingcollateral flow and with 20 yr of nonexertional chest wall pain believed to be a tight stenosis of the left circumflex coronary artery. caused by cervical root compression. He had a normal Coronary arteriography confirmed these findings, dem thallium and ECG treadmill test 1 yr and 10 days priorto onstrating complete occlusion of the proximal LAD and PET scan. His rest PET study is shown in the upper row rightcoronary arteries,which were supplied extensively by (51) and dipyridamole study in the lower row (52) in long collaterals from smaller proximal branches ofthese arteries axis and polar map displays in Figure 5A (1). Three as well as from the left circumflex. dimensionaltopographicmapsare shownin FigureSB. Asymptomatic Patient with .In Each panel ofthe topographic map (Fig. SB) views the left orderto furtherillustratethe applicationof clinicallyas ventricle as if looking at the septum (first panel), the sessing both absolute and relative coronary flow reserve, anterior wall (second panel), the left lateral wall (third Figure 4 shows an example of a PET study at rest and panel), or the inferior wall (fourth panel). afterdipyridamole stressusing generator-produced82Rbin In Figures SA-B, the rest image (upper row) showed a an asymptomatic man with no known cardiac disease or mild small apical defect, suggesting a small, old, apical events who had hypercholesterolemia and a normal thai myocardial infarction despite absence of a clinical event.

PET Perfusion Imaging and Nuclear Cardiology •Gould 587 @ ::@@?@

u@A

@ ‘ .•@ $$4@_@ $@ ,,@ . •_,952$VI@- @T @ •SI11vI$95 @@@_S?f95 @9595@ @ I@ @r@Pt i@o@wic*. @$- $511 . 57955$ @ •o@vI ‘@$ •@@wS • @ ZHST1IIJTIQH @41v95$$7YOF 791195 $@M@W$CIEN@I @V475$ - . — — - . @ $7$@Y I p054,952 hWY 0 p514'tII ,[email protected] . __ ScM.' • I 95 295 • 95 295

$1 N4T $5 2.5? U 295 S A _•‘_frsffi•“:

951$

).DC *15 VI@

@_ @0s95 $95 9S 95SCUAI95.9$95110 95 $9.5 94$ @4 *24 95 $41 S LII) @ ‘4 24 p ip I ii, @ 95 1$ r 95 iii FiGURE4. PETimagesofgenerator-produced@Rbinshort @ 95 S S 95710 @ 95 .* axis (A),horizontal(upperleft) and vertical(upperright) long-axis 95 .4 views(B),andthree-dimensionaltopographicdisplaysoftheheart I. •t @ L•@,• 9I 995 of a paflentwithmild,balancedthree-vesselcoronaryartery disease (C). See text for details. ReprOduCedwith permission from references(1) and (40).

The dipyridamole image (lower row) showed a largesevere controlled on drugs. He had never had chest pain or other anterior and apical defect involving 29% of myocardium symptoms and had no physical limitations. Coronary ar outside 2.5 standarddeviations from normal. There were teriography showed an occluded right coronary artery definitebut lessseveredefectof the anteriorseptumand filling distally through collaterals, an occluded diagonal, a lateral wall. These results were interpreted as showing a 50% diameter stenosis of the obtuse marginal, a 50% very severe mid-LAD, another more moderate severe LAD diameter stenosis of the proximal LAD, and no segmental stenosis proximal to the first septal perforator, and a narrowing of the left circumflex coronary artery. PET moderate stenosis of an artery to the lateral myocardial scans showed a fixed inferior defect without perfusion wall, probably a left circumflex stenosis. Left ventricular abnormalities anterior or laterally (not shown). function was normal. Since the patient had never had angina pectoris as an Accordingly,the patientwasscheduledfor diagnostic endpoint for followinghisdisease,whichpresentedwith catheterization and PTCA of three lesions with surgical myocardial infarction/sudden death, routine follow-up backup. Quantitative coronary arteriography confirmed PET scan was obtained 6 mos later without symptoms or significant stenosis ofthe proximal and mid-LAD, and the change in his clinical status. The rest scan was unchanged mid-left circumflex. PTCA was carriedout on these three from the previous study showing inferior scar. In Figure 6 lesions due to severity of the disease involving a large (1), comparison of the first stress scan (upper row, Si) extent of myocardium. with the follow-up stress scan (lower row, 52) showed a Following CAD in Asymptomatic Patient. The patient new moderate mid-anterior wall perfusion defect and a was a 61-yr-old white male who had a previous inferior severe lateral perfusion defect not present previously. The myocardial infarction, sudden death with successful resus routine follow-up PET scans thereforedemonstrated mod citation, and recurrent inducible ventricular tachycardia erateprogressionof the mid-LAD lesionand appearance

588 The Journal of Nuclear Medicine •Vol. 32 •No. 4 •April 1991 @ I @: @I@'p@

@ L•4 9510 vl@ 9595 — NI5$95 955 @p,-

‘$3

@ I@ 71,1 ?C@I@ @$-119595.119595 @ 19511107195 LIII@11Y 95 9595 @Th @1V@ @095 N@ C9595U951 @0 @ 07$.Y I $25459 sf07 I • @ @.0 $95 0• ...@95?$4 @*A @__ @. S 95 295 I 95 $0 @ S ; 11.110 N‘0• *4II ,@ @ w@ 95 $1 , 95 2• @ 95 0 0 95110 95 94 -—5.— @ :@ 95 .4 11 @ :@.. I. 0$ @ .1― 4@1@ SI. N

FIGURE 6. PET imagesin three-dimensionaltopographicdis plays showing the first stress study (upper row) and a second stressstudy6 molater(lowerrow).Seetext for detalls.Repro ducedwithpermissionfromreference(1).

@ I 1OP@WIC5.@$-91$?.1795*1 (97), and the FATS trial by Brown et al. (98) demonstrate @ 90127171*1 *11*19*2?? 117950$**?W @ID5$@VflD that reversalof coronary arterystenosis in man is feasible. @ hwy I p20*1142 $?L@Y1 p20*0*95 @ I @i w The Ornish-Gould study was the firstrandomized, con L_@ • @• $@ 0 *1 iN trolled, blinded, artenographic trial to determine the ef $12.5, UD1 fects of strict risk factor modification on the geometric dimensions, shape, and fluid dynamic characteristics of arteriographic coronary artery stenosis in man (1,59,60). Using percent diameter stenosis as the end point that is the most widely recognized measure of stenosis severity, FIGURE5. PETImagesInlong-axis@6ewsandpolarmaps(A) this study demonstrated regressionor cessation of progres and three-dimensionaltopographicdisplaysof (B) a patient with sion in approximately 80%-85% of treated patients com two normal thalllum/ECG stress tests. See text for detalls. Re paredto approximately 80%—85%progressionin the con produeadwfth @ionfrom @nce(1). trol group. Brown et al. (98) also demonstrated cessation of progression or reversal of percent stenosis in patients with diet and multiple cholesterol-lowering drugs, as well of a new, severe defect in the left circumflex distribution as a dramatic decrease in myocardial infarction, sudden not present previously. death, and bypass surgery/PTCA in the treated versus Repeat coronary arteriography confirmed progression control group. of the mid-LADstenosisfrom 50% to 70%. The left However, complex changes of different stenosis dimen circumflex, which previously had not demonstrated signif sions in opposite directions or to different degrees may leant localized narrowing. had developed a 90% stenosis cause stenosis shape change with profound effects on fluid in its proximal portion. As a result of PET findings, the dynamic severity unaccounted for by simple percent nar patientwassentto bypasssurgerysincesymptomswere rowing. Accordingly, we analyzed all stenosis dimensions an inadequate guide to progressionof this disease. on coronary arteriograms including absolute proximal, minimal, distal lumen diameter (area), percent stenosis, REVERSALOF CORONARYARTERYSTENOSIS integrated length, and exit effects to determine stenosis The question often arises whether accurate detection of shape and a @ngleintegrated measure of functional sever coronaryatherosclerosisisclinicallyuseful,particularlyin ity and stenosis flow reserve, incorporating the fluid dy asymptomatic subjects. Although overwhelming epidemi namic effects of all these dimensions (1,59,60). In the ologic evidence links coronary heart disease to smoking, controigroup, complex shape change and stenosis molding serum cholesterol, , and family history, large characteristic ofstatistically significant progressing severity trials of modestrisk factormodificationwith mortality occurred with worsening of stenosis flow reserve. In the end points have been somewhat disappointing, as recently treatedgroup, complex shape change and stenosis molding reviewed(1). In contrast with these previous largeclinical characteristicof significantregressingseveritywas ob trials, the randomized arteriographictrial of Ornish and served with improved stenosis flow reserve (1). The im Gould (1,59,60), the CLAS study by Blankenhorn et al. proved geometry was associated with marked reduction in

PETPerfusionImagingandNudearCardiology•Gould 589 angina pectoris such that at the end of the study, most threshold percent diameter narrowing, such as 50% ste patients in the treated group were off cardiac drugs. These nosis, as the criterion for presence of CAD. resultsdocumentfor the firsttime the multidimensional There are three limitations to this use of sensitivity/ characteristics of regressing coronary artery stenosis in specificity analysis for assessing accuracy of noninvasive man associated with risk factor modification. tests for coronary disease. First, coronary atherosclerosis However, only patients with known CAD require is not an all-or-none condition. Binary classification re marked lifestyle change (59,60) or lowering of serum quires arbitrary threshold criteria and creates artificial cholesterolwith multiple drugs(97,98) to prevent progres classifications for a disease that has a continuous spectrum sion or reverse severity of coronary artery stenosis. Such of severity by both arteriographyand perfusion imaging. vigorous lifestyle change or drug interventions may not be Threshold values that yield optimal sensitivity and speci appropriate for those individuals with risk factors who do ficity values for one test may yield optimal sensitivity and not have coronary arterystenosis. Those individuals who specificity values for one test may yield falsely lower values do have coronary heartdisease or a genetic predisposition for a different but more accuratetest if its detection to it need more aggressive risk factor interventions than threshold is different. For example, an imaging test capable are indicated for the general population or for those with ofdetecting40% stenosismayhavelowspecificityaccord modestly elevated cholesterol without coronary arteryste ing to a 60% stenosis criterion but high specificity accord nosis. While community education programs are impor ing to a 40% stenosiscriterion.Second,sensitivityand tant, their targetguidelines for diet or drugtherapyare not specificityare markedly affected by disease severityand as stringent as necessary for a patient with documented distributionin the studypopulation(99-101). A sample CAD or a genetic predisposition. Therefore, the vigor of population with a high frequency of mild disease will be risk factor modification should increase with the probabil distributed centrally near the threshold values where scat ity or certainty of having CAD. Vigorous medical reversal ter is more likely to lower sensitivity and specificity. A therapy should also be considered as an additional step in sample population with a high prevalence ofsevere disease patients undergoing mechanical intervention in so that resulting in good contrast between normal and abnormal progression of disease does not negate the benefits of areas ofan image will produce good sensitivity/specificity. mechanical intervention. Thus, the sensitivity and specificity found in one pop These observations have major implications for con ulation may not apply to a different population. The most trolling health care costs of cardiovasculardisease as dis severe test for a nomnvasive method is a low-prevalence, cussed subsequently. asymptomatic population. In the University ofTexas study (35), 40% of patients were asymptomatic, a much larger QUANTITATIVECORONARYARTERIOGRAPHIC prevalence of asymptomatic individuals than previously GOLDSTANDARDS reported for SPECT and therefore a more realistic test The sensitivity and specificity of diagnosing CAD by population for comparing PET and SPECT. In a subset of PET are both 95%—98%in symptomatic or asymptomatic 46 patients with both PET and SPECT, for moderately individuals(32-39), including“balanced―three-vesseldis severe disease with stenosis flow reserve of <3, PET de ease(40,41). False-negativestudiesare usually associated tected 95%, whereas SPECT detected 58%; for a flow with distal coronary stenosis of moderate severity or dis reserve of 3-4, PET detected 52% and SPECT detected easeof smaller arteriesin which arterial bed sizeis small, none. minimizing the defect intensity or contrast between nor To overcome the limitations of sensitivity/specificity as mal and abnormal areas. False-positives are usually asso end pointsof diagnosticeffectiveness,an analysisof test ciatedwith nonatheromatouscoronaryarteryabnormali results as continuous variables has been proposed; how ties such as spasm, thrombi, or myocardial muscle bridge. ever,thisrequiresquantitativeanalysisof PET scansand Although this diagnostic accuracy surpasses other nonin arteriography (32—35,40--43,45--50,102).In addition to vasive technologies for evaluating CAD, measuring diag these limitations, we have also shown that percent diam nostic accuracy by the end points of sensitivity and speci eter narrowing is not an adequate standard for quantifying ficityfailsto accountfor thequantitativecapacityof PET stenosis severity of coronary artery narrowing. Percent to assess severity and size of involved areas rather than stenosis does not account for the effects of diffuse disease, just presence or absence of disease. eccentricity, stenosis length, absolute cross-sectional area, Most reports on the diagnostic accuracy of myocardial entrance and exit shape, and absolute dimensions on flow perfusion imaging have used sensitivity/specificity to de or flow capacity. It is limited by substantial inter- and scribe the relation between perfusion image defects and intraobservervariability.Alternativeapproachesproviding arteriographically documented disease. This approach re fluid dynamically correctmeasurements ofgraded stenosis quires binary (positive or negative) classification of both severity utilize quantitative arteriographicmethods to cal imaging and arteriographic results. Perfusion images have culate stenosis resistance (51), pressure-flow curves (49, been classified as normal/abnormal and arteriographic 50) or stenosis flow reserve (32—35,40,50). stenosis severity has usually been described in terms of a Quantitative coronary arteriography predicts the func

590 TheJournalof NudearMedicine•Vol.32 •No.4 •April1991 tional pressure-flowcharacteristicsof stenosisdirectly 45 measured by flowmeter if all the dimensions of the lesion @ are taken into account, including relative percent narrow 35 ing, absolute luminal area, shape, and integrated length > effects(32-35,40,49,50). Becausethesemultipledimen 0 b 2.5 sions have cumulative hemodynamic effects and interact 0 with each other, they are combined by fluid dynamic @ 1.5 equations into a single measure of stenosis severity, which 0 @ 1 are applicable clinically to stenosis flow reserve. Stenosis D flow reserve is the same as coronary flow reserve but instead of being measured directly by flowmeter, stenosis flow reserve is derived from all geometric dimensions of 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 10 0.5 0 C0R0NAR@STENOSISFLOWRESERVE the narrowingas a standardized,single, integratedmeasure (ST(NOSJSSEV7RIIY) of its severity for standardizedconditions of pressureand normal reference vasodilator reserve, as described else FIGURE7. Visuallyscoredseverityof PETdefectsfrom0 where(1). (normal)to 5 (severedefect) is related to stenosis flow reserve by automated quantitative coronary arteriography. Reproduced Thus, we use the term stenosis flow reservein reference with permissionfrom reference(1) as adapted from reference to calculatedcoronary flow reservederived from anatomic (35). stenosis dimensions by quantitative analysis of coronary arteriograms.We use the term coronary flow reserve for flow reserve directly measured by a flowmeter, Doppler in symptomatic patients in literatureprior to 1983 (103— catheter, or perfusion imaging. The concept of arterio 105). Studies since 1983 report sensitivity of 70%—85% graphic stenosis flow reserve has been well developed and specificity of 45%—60%in both symptomatic (106— theoretically, validated by comparison with directly meas 111) and asymptomatic subjects (112,113) or those with ured flow reserve experimentally (33,40) and proven din atypical presentations (114), as summarized in Table 1. ically valuable by comparison with regional PET perfusion An average specificity of 53% means that there are 47% defects in patients (32,35). Therefore, stenosis flow reserve false-positive thallium exercise tests in patients whose cor provides the best arteriographic definition of stenosis se onary arteriograms show no significant coronary artery verity, suitable as a gold standard for comparison with stenosis. These arteriograms are unnecessary if a more noninvasive imaging. Using stenosis flow reserve as the accurate noninvasive test is used initially. European cx gold standard, PET has a diagnostic accuracy of 95%- periencewith thallium SPECTconfirms thislow specificity 98%, even in asymptomatic patients (1,32,35). (115). The quantitative severity of coronary artery stenosis One explanation proposed for low specificity in recent correlates with severity of corresponding PET perfusion reports is “referralbias―;i.e., patients with negative thai defecton rest-dipyridamolestudies.Severityof PET per lium stresstests no longer undergo fusion defects read by two readers blinded to clinical or (107,109). The catheterized population in such a study arteriographic data and graded visually on a severity scale would be biased by this exclusion of normals, thereby from 0 (normal) to S (most severe) correlates well with skewing the population toward higher prevalence of dis arteriographic severity of disease.This relation between ease. According to Bayes' Theorem, as the population is defect severity by PET and stenosis flow reserve by quan skewed toward greaterprevalence of disease, the post-test titative coronary arteriography is shown in Figure 7. probability of having no disease with a normal test and Localization of the artery(s)involved is highly specific the observed apparent test specificity decrease in that for PET, as shown in Table 2. Since spatial localization is population (116,117). so good, lesions in series can routinely be identified by However, we have demonstrated that the currently re more severe defects distally, e.g., a septal perfusion defect ported low specificity of S0%—60%is unlikely due to and mild anteriordefect as ifiustratedpreviously in Figures referralbias (116—118).Accounting for referralbias by a 3 and 5. sophisticated model, Diamond (118) estimated an average “corrected―sensitivity ofthallium exercisetesting of 67.5% EXERCISE THALLIUM IMAGING AND and average “corrected―specificity of 75% (Table 1). His conclusion agreed with our estimates of 65% sensitivity DIPYRIDAMOLE PET and 65% specificity accounting for changing patterns of As shown in Table 1, the average sensitivity and speci interpretationsince the early l980s but without correction ficity of thallium exercise testing in the last 3,1 19 cases forreferralbias(116,117). Althoughreferralbiasisa valid publishedsince 1983 are 84% and 53%, respectively, concept, it cannot be invoked to explain away the limited weightedfor numberof casesin eachstudy.Sensitivity sensitivity and specificity of thallium exercise testing as and specificityofexercise thallium imaging was 80%—90% currently used. Diamond also concluded that the use of

PETPerfusionImagingandNuclearCardiology•Gould 591 @ i@1 0 0 0

N ++ ++ .@z ** ** @ bc 0 0 000000 * ****** @ 0 00 0000000 ** *** **

@ Q. 0 00 00 0 0 * * **** *

@ -@- - @o ooco *** - - * @ Hi@ @;H @ 0@30 Al @@oo0 0 0 0000 0 00000•O E @ 00 0 00 000 0 oo•

@ a.(@ 0 00 00

@ 00000 0 0 00000 0 0

I I a @ rE@ -I ti2 a a z @@o•ooooo•o oo 0000 o o•oooo @ o•oooo•o oo•• o oo

.@ 0 0 •oo • o o o ooo•oo I @ jiI@ii

z§ ••o••••••o•.. @ o• oo• oo @ ; •oooo• o oc @ a. 0 o• H @ oo•so•o oo o•oo • @2 o•o

@ !i@@s @go o

@ ***************************** ***** * @ a @ a

@ •oo @1i

u)@1i@i

592 The Journal of Nudear Medicine •Vol. 32 •No. 4 •April 1991 low riskgroups to establish “normalcy―standardswithout and orientation of the heart in the chest. Therefore, as comparison to coronary arteriographyis not correct. sumed constant or “standardized―distribution of atten Table 1 compares currently reported sensitivity and uation corrections is not useful but must be measured. specificity of thallium exercise testing, 84% and 53% re Figure 8 illustrates the importance of attenuation cor spectively,with dipyridamole PET, 96% and 96% respec rection (1,31). A rest perfusion image with N-13 ammonia tively (1,32,35—37,119). In a study population with a high was reconstructedby standardPET techniques with atten prevalence (94%) of severe advanced CAD as reportedby uation correction, upper panel labeled 51. The lower of Tamaki (120), the differences between PET and thallium each image pair, labeled 52, was reconstructed from the SPECT are not apparent (120) because the disease is severe same data without the attenuation correction. Figure 8A enough to cause such high contrast between normal and shows short axis views and 8B shows polar map displays. abnormal areas of images that it is detectable even by The 52 images are therefore ideal SPECT scans collected limited technology. In addition, at a disease prevalence of with 360°ofspatial sampling and high radionucide energy 94%, the poor specificity of any modality is not manifest (51 1 keY), thereby minimizing attenuation losses com because thereare too few normals (3 normals in the study) pared to much lower energy thallium or technetium. The or those with less severe disease in the study population. SPECT image from the same data without attenuation Finally, the Tamaki study used treadmill exercise, which correction shows severe inferior and septal defects typical does not increase coronary flow as much as dipyridamole of false-positive defects associated with tissue attenuation used in other PET reports. The advantages of advanced on thallium or technetium SPECT scans. Therefore, the PET technology are best observed for intermediate disease high number of false-positive SPECT scans, or low speci prevalence(<60% of the study population) and/or mod crate-to-lesssevere CAD in which the question of medical or mechanical intervention is unclear and thallium-stress testing is least accurate. In two other studies directly comparing PET and SPECT (121-123), the diagnostic accuracy of PET was signifi cantly higher than SPEC!' despite inadequate counts (6—7 million counts perwhole-heartdata set) in one study (123) and inappropriatePET softwarein the other (121,122), as previously analyzed in detail (1). Using a faster scanner •!$!@SCaLD collecting adequate counts (30—50million counts per whole-heart data set) and optimized PET software, others have reported much better sensitivity and specificity of 95% or higher (32,35—37,119).A direct comparison of PET and SPECT has been reported using appropriate software, an appropriate study population, and adequate counts in a subset of 46 patients (35). Of those patients

@ with moderately severe disease with stenosis flow reserve I .__ of < 3 by automated quantitative coronary arteriography, @AB blinded readers identified 95% by PET compared with 58% by thallium exercise testing. Of those with milder S@966x disease with stenosis flow reserve between 3 and 4, PET L4@69I 75 detected 52% whereas thallium exercise testing detected V6547411Sr— 61 none. Thus, a total ofthree studies with direct comparison of PET and SPECT showed PET to have better diagnostic accuracythan SPECT (35,121—123). I A675x41L54Iw41 Technical Differences Between PET and SPECT Although thallium treadmill testing and dipyridamole PET are both used for “stress―perfusion imaging, there @ @POST @L3C$1, are major differences between these two procedures. *15 SPECT lacks attenuation correction, has depth-dependent changes in resolution, and limited sampling frequency. Of FIGURE8. Restingmyocardlalperfusionscanusing13N-am these, the lack of attenuation correction is the major monia with standard PET reCOnStruCtiOnusing attenuation cor rection (Si) and without attenuation correction (52). Lack of problem because attenuation of activity from the inferior attenuation correction causes a severe inferior defect that is wall of the heart and septum (in men) and anterior wall artifactual due to tissue attenuation In short-axis views (A,B). (in women) is dependent on highly variable body habitus Reproducedwith permissionfrom reference(1).

PETPerfusionImagingandNuclearCardiology•Gould 593 ficity, can be explained by lack of attenuation correction sensitivity and specificity for detecting and quantifying as illustratedin this example,which cannot be distin severity of coronary artery stenosis (32—35,40,45—50). guished from a real perfusion defect. The three-headed SPECT scanner does not solve this CARDIAC PET IN CLINICAL PRACTICE problem but may make it worse. Artifactual inferior or Clinical cardiac PET utilizing the 82Rb generator (23, septal defects appear to be as common or more common 24,35,36,58,63) is practical in cardiologic practice (119, with the three-headed SPECT system compared with 125,126). Since the first private practice PET study in standard SPECT. The reason is that the better sampling March, 1988 in Atlanta (119), over 4,000 patients have ofthe three-headedSPECT system shows more clearlythe been studied in private practice sites. However, some attenuation of photons from posterior wall and septum. qualifications are important to mention. Based on our Asimagingtechnologyimprovesfromplanartorotational experience in over 1,000 university-based clinical studies SPECT to three-headedSPECT to complete ring SPECT, and the private practice studies, it is necessary to simulta attenuation correction becomes increasingly important. neously utilize all tomographic views available for inter Even with PET, as resolution and sampling frequency pretation—acquisition,short-axis,long-axis,and polar improve in more advanced PET scanners, attenuation map views. Short-axis views alone commonly fail to show correction becomes more critical and more difficult to do abnormalities seen on long-axis views. For “balanced― accurately. three-vesselCAD, thepolarmapdisplaysshowingrelative The claim is often made that attentional losses with and absolute flow reserve of the whole heart demonstrate SPECT are less than with PET because single photons abnormalitiesmore accuratelythan the tomographic views have a shorter path to the detector than the total path of alone. the paired photons of positron decay. This claim is not It is also our observation that PET data cannot be compatible with basic fundamental physics of tomo processed with software designed for SPECT imaging be graphicimaging.Ifa singlephoton arisesfrom the anterior cause the different reconstruction filters, rotational algo myocardium, it travels a shorter path and has less atten rithms, etc. optimal for SPECT are unsatisfactoryfor PET uation than paired photons from the same point detected images. Although the diagnostic accuracy of PET data in coincidence,sincethe pathof pairedphotonsincludes processed by SPECT software is better than that by SPECT the posterior attenuation. For the total path of all single imaging (121,122), PET images processed with SPECT emission photons from inferior and septal areas of myo softwareare overly smoothedwith data displacedinto cardium, attenuation is the same for coincidence- and neighboring pixels (blurring) by multistep rotational single-photon counting, which causes inferior and septal schemes or excessive smoothing. Even with inappropriate defectsunlesscorrectedby transmissiondata.Attenuation software, specificity of PET for diagnosis of CAD was from inferior and septal myocardium is unpredictable higher than for SPECT (121,122), but neither sensitivity depending on cardiac size and orientation in the chest. nor specificity was as high as reported by others with Although overall attenuation by the entire heart is the appropriatesoftware,i.e., Schelbert(36), Tamaki(37), same for single and paired photons for adequate circum Williams (119), Gould (32), and Demer (35). In our ferentialsampling,the differenceis that attenuationis experience, suboptimal processing of PET data by SPECT measuredandcorrectedbyPETbutnotwithSPECT. softwarelowersdiagnosticcontentin comparisonwith The higher energy technetium compounds for SPECT software specifically designed for PET reconstruction and have been thought to decrease the problem of artifactual display (41), which achieves a higher diagnostic accuracy inferiorand septaldefects due to attenuation. However, in (32,35—37,119). our experience, higher energy emission does not reduce Another reason for lower sensitivity (or specificity) is this problem. The images in Figure 8 were obtained with inadequate total counts in the whole-heart image set. Six S1l-keV emission of a positron radionucide. Without to 7 million counts per whole-heart image set, as reported attenuation correction, there is an artifactual severe infe with a blockcameradesignhavinghighdead-timelosses rior and septal defect. There are other causes for false (123), seriously compromises diagnostic content for short positive SPECT scans (124) as well as other technical lived tracers like 82Rbor ‘50-water.In comparison, for details of PET relating clinical diagnostic accuracy to 82Rbwe acquire approximately 15—35million counts per scanner performance (1). whole-heart data set and achieve a greater sensitivity and Exercisestressdoesnotincreasecoronarybloodflowas specificity. In count-poor studies, summing slices makes much as i.v. dipyridamole or adenosine. Using PET to them look better but does not overcome the basic maccu measure absolute coronary flow and coronary flow reserve, racydue to inadequatecounts in the whole-heartdata set. Schwaiger observed the normal increase in coronary flow Therefore, high count rate capacity, good Z-axis sampling with treadmill exercise to be 2.5 times baseline, compared without interplane undersampling (127-130), and quan with 4.1 times baseline for i.v. dipyridamole (Schwaiger titative validated, clinical software (41) are important for M, personalcommunication, 1990). The strongerstimulus achieving high sensitivity and specificity in clinical appli for increasing coronary flow (dipyridamole) increases its cations.

594 TheJournalof NudearMedicine•Vol.32 •No.4 •April1991 ECONOMICSOF PET USING ‘2RbAND BASIS FOR then they should ensure payment for PET, which is more INSURANCEREIMBURSEMENT accurate and provides additional diagnostic certainty on Figure 9 (132) shows the comparative diagnostic costs severity/location of coronary artery stenosis, which is of PET, SPECT, or artenography as a function of CAD crucial for determining the need for PTCA or bypass prevalence in the study population (1,131—133).For this surgery. PET diagnosis can eliminate the need for figure, sensitivity and specificity for SPECT were 80% and additional unnecessary arteriogramsby determining that 50% respectively and, for PET, 95% and 95% respectively, a disease is not severe enough and/or not critical in area as shown in Table 1.Cost ofSPECT was $1,000 per study, or size, e.g., a small defect in the right coronary artery including professional fees, cost for PET using generator distribution, to requirearteriography.Thus, PET provides produced82Rbwas$1,500.Eachpositivenoninvasivetest independent hard data that can determine subsequent by both technologieswasassumedto incur arteriography treatment of patients at cost savings to the insurer. costing $6,000. If the cost of coronary arteriographyfor false-positive tests is counted into the cost of identifying ADDITIONALBENEFITS OF PET WITh @Rb: or ruling out disease, the overall expense of medical care ASSESSINGMYOCARDIALINFARCTION,ISCHEMIA utilizing thallium SPECT is higher than for PET due to AND VIABILITY costly definitive catheterization required for false-positive Distinguishingnecroticfrom viablemyocardiumisim SPECT results. portant for treating patients with acute myocardial infarc Since PET also provides an approximate quantitative tion, particularly following thrombolysis therapy when assessment of severity, it can potentially be used as a or balloon angioplasty may be indicated substitute for arteriography in many patients with mild to for remaining viable myocardium. The extent of myocar moderate disease amenable to medical therapy. If dial infarction and/or viability depends on how it is meas approximately 30%—40%ofpatients undergoing coronary ured. Thallium exercise perfusion imaging in the peri arteriography who need a mechanical procedure (PTCA, infarction period is most commonly used for this purpose bypass surgery) could be identified by PET, the need for (134—141). many arteriogramscould be prevented. This instance is For example, Rozanski et al. (135) reported in 1981 shown by the lower line of Figure 9 where half of the that 35 out of 43 patients (81%) with stress defects that arteriograms have been eliminated on the basis of PET redistributedon 201Tlexercisetolerancetests(ETT) im resultsshowingmild disease.On the other hand, proved left ventricular wall motion after bypass surgery. individuals with asymptomatic severe disease identified by However, 19% with reversible stress defects did not im PET may need mechanical intervention to prevent prove. Although most patients with fixed defects on myocardial infarction or sudden death. Therefore, the use ETT did not show improved wall motion, 4 of 29 did of the 82Rbgenerator instead of a cyclotron for cardiac improve despite fixed defects. In 1989, Tamaki et al. (56, PET reducesthe cost ofPET imaging to a reasonablerange 139) found that 8 of 23 myocardial segments (35%) with for routine clinical use as the basis for vigorous medical transient defects on 2OVflETT SPECT did not improve therapyormechanicalintervention. leftventricularfunctionpostoperatively;14of 33 patients Ifmedical insurersarewilling to pay for thallium, which with fixed defects (42%) did improve left ventricularfunc has a published accuracy of only 50%—60%and leads to tion postoperatively (56). additional costs of unnecessary coronary arteriograms, In 1988, Cloninger et al. (137) reported that of 95 patients with first suspected or documented myocardial infarction and incomplete redistribution on 2OVflETT, 72 (76%) improved redistribution images and 24% did not improve after PTCA. Of 16 patients with prior or acute myocardial infarction and incomplete redistribution, 8 E seoooooo improvedredistributionimagesafterPTCA and8 did not -a $5000000 (137). These reports suggest that identifying myocardial -a viability and postoperative improvement in left ventiricu larfunction by stressthallium imaging without reinjection has limited predictive accuracy for improvement after mechanical intervention and is therefore limited for as C., $2000000• z sessingviability. C.) @ (1) Of the many possiblereasonsfor 201TlETT failing to 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 predict viability, two possible explanations stand out: PREVALENCEOF DISEASE either the imaging technology of 201'flis inadequate or the FIGURE9. Comparativecostsof @°@TISPECTandPETwith basicconceptof identifyingnecrosisby a fixeddefecton generator-produced@Rbas a functionof diseaseprevalence. 201T1ETT is incorrect. With regardto the first alternative, Reproduced with permission from reference (132). exercise perfusion imaging of ‘3N-ammoniaby PET does

PET Perfusion Imaging and Nuclear Cardiology •Gould 595 not improve prediction of postoperative improvement of FIGURE 11. Schematicof left ventricularfunction (56,139). Fourteen of48 segments anenlargingor worseningde fed with stress compared to with transientperfusiondefects by ‘3N-ammoniaand PET the centralrestingdefectof a (30%) had no improvementin left ventricularfunction rest or redistribution thallium postoperatively;5 of27 segments with fixed defects (19%) image. This partially reversible hadimprovedleft ventricularfunction despite fixed defects defectindicatesthe areaof low (56). Therefore, in the same study, the errors of exercise flowreservearounda central resting defect but does not perfusion imaging for predicting viability and postopera provide information on the via tive improvement in left ventricular function was 35%— bllity (or necrosis) of myocar 42% for SPECT and 20%—30%by exercise perfusion PET diumwithintherestingdefect. imaging. Reproducedwithpermission Therefore, the error of 20%—30%in predicting viability fromreference(1). and postoperative improvement in left ventricular func tion based on transientstressperfusion defects may be due in part to a problem with the basic concept of exercise dium by imaging metabolic analogs, especially [18flFDG, perfusion imaging as a means of assessing viability. and/or cell membrane integrity by the potassium analog Figure 10 (1) shows schematically a defect in both stress 82Rb,as outlined below. (left panel) and redistribution (right panel). Due to wash out of thallium from the normal area on the 4-hr redistri PROTOCOLFOR MYOCARDIALVIABILITY AND bution scan, the relative severity of the defect appears to INFARCT SIZE WITh @Rb decreaseasif redistribution into the defect had occurred, suggesting viability. However, the apparent partial redis The leak of potassium from myocardial cells is an tribution may be due to washout of radionucide from immediateearlymarkerof impairedcellmembranefunc normal area with decreasing severity, i.e., a “reversible―tion and necrosis that has been well documented by a defect. Similarly, a rest PET perfusion scan will have a less substantialliterature(1,142,143). Therefore,a quantita severedefectthantheexercisePETperfusionscandueto tive imaging method utilizing a potassium analog reflecting greater uptake of radionucide by normal myocardium. cell membrane function might be useful for assessing Therefore, no information is provided about viability in viability and infarct size. the area of the resting defect since apparent lessening of Rubidium is a potassium analog with a variety of mcd defect severity may be due to washout ofactivity from the ically useful radioactive forms including 81Rb,82Rb,84Rb, normal area or lower resting flow compared with stress and 86Rb.Myocardial cell membrane transport, trapping, flow in normal area. andflow-extractioncharacteristicsofpotassiumandrubid Figure 11 (1) illustrates that a partly reversible exercise ium parallel each other. perfusion defect identifies the area of limited flow reserve The clinical protocol based on rubidium kinetics utilizes around a central injured area of myocardium (left panel). the same resting perfusion protocol described previously It does not provide data on how much viable myocardium with the exception that the image data is acquired in list is left in the area of the defect with potentially partial mode. As illustrated in Figure 12 (144), the resting PET damage. Exerciseor dipyridamole stressperfusion imaging data collected in list mode is divided into an early phase therefore provides a measure of the zone at risk having low flow reserve (right panel) but not of viability within the area of a fixed defect. Positron emission tomography has been developed for MYOCARDIAL VIABILUY/NECROSIS BY RB42 KINETICS Rb.82 Iv hjsctlon @ identifying ischemic, viable, or necrotic/fibrotic myocar sic wait5095

REVERSIBLESTRESS PERFUSION DEFECT

@95:012345@ cc PET ______Il@ LATENAGE1@@C VIABLE OR SCAR BULLSEYE @::IIII@ i::IIII@ FIGURE10. Schematicrepresentationof thalliumwashout fromnormalareasof anexerciseimage(leftpanel),resuftingin FIGURE12. Schematicof the dinicalprotocolutilizingthe relativeredistributionon the 4-hrdelayedimaged(rightpanel), kineticchangesof @Rbafter intravenousinjectionfor assessing which is consistent with either scar or viable myocardium. Repro myocardialviability.Reproducedwith permissionfrom reference ducedwith permissionfrom reference(1). (144).

596 The Journal of Nuclear Medicine •Vol. 32 •No. 4 •April 1991 image (first 15—110 secs) and a late phase image (120—360 tial amount of viable myocardium remaining. As shown secs). A new defect or worsening defect on the late image in Figure l3B, the late rubidium image (51, upper row) compared with the early image indicates washout or failure matched the FDG image (52, lower row) therebyconfirm to traprubidium and thereforenecrosis. Residual myocar ing that substantial viable myocardium remained. Conse dial trappingofrubidium on the late image within a defect quently, arteriographywas carriedout, which confirmed a reflecting partial or limited washout indicates some resid reperfused LAD stenosis followed by successful PTCA. ual viable myocardium. Follow-up PET showed a normal anteriorwall and only a Infarcted myocardium fails to retain rubidium, which residual small apical scar. washes out from cells after initial uptake. A mix of in Figure 14(144) correlatesinfarctsize based on rubidium farcted and viable tissue in the field of view results in an washout compared with FDG images, both with auto intermediate level of washout proportional to the percent mated infarct sizing, in the first 36 patients studied with of viable or infarcted tissue. The activity in the late (52) both 82Rband FDG after glucose loading. The patients rubidium image relative to the early (51) image therefore with very large infarct areas had previous infarctions and reflects the extent of washout and proportion of viable or severe ischemic cardiomyopathy, and were being consid necroticmyocardium. ered for cardiac transplantation. For these very large in Figure13 (41) showsearly(Si) andlate(52) images after a single i.v. injection of generator-produced82Rbin a patient with an evolving acute anterior myocardial in farction in horizontal and vertical long axis views. The A S C earlyimagesare shownuppermostin eachpair of image p rows (study 1). The late images are in the lower row of each pair of image rows (study 2). The number after the

@ decimal is the image plane for both studies 1 and 2. In the 17,.cY C p951't•$iC•- @- S S L@I p 0 OILS 7t@@ —— C color coding, white indicates the highest activity, red next @95f — , .w @ I. Oil 7*03 S highest,yellow intermediate,and greenand blue the lowest relativeactivity. PU@95 C@IiN?E @S 1*0 15 [email protected]?C95'1295110 <*1 I 953 Figure 13A shows horizontal (left) and vertical (right) 95 IS 5 AN 5% S Si *0 11 S LII long axis views ofthe heart.The horizontal long axis views 3 x 0* 75 14 IN 95 are oriented as if looking down from above. The vertical *5 Il L N long axis tomographs are oriented as iflooking at the left A N 5• *5 1 @, 0,, (55 *• 55 951_mIll It'll 95710 ( 953 @ side of the body cut head to toe. Anterior myocardium is *53 •1 S ES at the top, inferior at the bottom, apex at the left, and AV Es 54 I 04 IS St L N S 55 ring on the right. A ,@Si. St RATIOSZ'$I Tomographic data is summarized in a polar display as if lookingat the patientfrom the outsidetowardthe left ventricularapex located in the center ofthe bull's-eye with the outer rim of the bull's-eye corresponding to the AV ring. Polar displays on the left (lower half of figure) show the relative activity on a scale of0%—100%with the early study being the upper (Si) and the late study being the

@ lower (52) of the polar maps on the left side of the panel. I@ I'Sfl . VI@ - @it . @ The lower rightpolar display, labeled relative 52/51 ratio, DS?IIUVISS. SSIs@IW I@iW @IS@ @ SI@ I . 1V@1@- a . @ shows the relativechange in normalized activity ofthe late @s, ii @. @ image divided by the relative distribution of normalized S S •0C5S@NI . S.. •ss**a@ 1 activity of the early image (relative instead of absolute I I@RAIRA-) LP S • @ values), on a scale of 0 to 2. It therefore quantifies the SSL* S@5 I @ 95*1 p 55 95 relativechangein activityfromearlyto lateimages(144). .‘CIS@ p The upper right polar map, labeled absolute 52/Si ratio,

is used for comparing rest-stressimages and has no mean I*1@ S @lII @ ing or use for comparing early-late images. •5 L* @‘ @ 95*.' ‘ p The patient in this figure is a 51-yr-old woman with 55 Pu 95@ acute anterior myocardial infarction 2 days previously. The earlyrubidium image (Si) shows normal perfusion to FIGURE13. PETImagesearly(Si) andlate(52)aftera single intravenousinjectionof @Rbat rest in long-axisviews and polar the anteriorwall indicating a reperfusedLAD that washes maps (A). (B) The late rubidium image (Si) is identical to the FDG out to leave a severe defect at the apex indicating necrosis. image(52).Seetextfordetails.ReproducedwithpermiSSiOn The anteriorwall retained rubidium indicating a substan from reference(41).

PET Perfusion Imaging and Nudear Cardiology •Gould 597 100 AUTOMATED INFARCT SIZING BY PET IN MAN reflected by abnormal kinetics of the potassium analog Rb-82 compared to FDG 82Rb parallels loss of intracellular intermediary glucose 90 metabolism as reflected by lack of FDG uptake. Our > SO O@ observations extend the well-described behavior of potas @ 70 95 sium as a marker of myocardial necrosis into clinical i@ 60 0 applications for infarct sizing and relate it to more recent @ 50 measures of viability based on metabolic imaging. Thus, @ 40 myocardialnecrosisor viability may be identified by meas @30 OO ures of either glucose metabolism or cell membrane integ @20 r . .824. pP.0 @001 @ — I 0 :@ FDG . 872Rb• 4.234 rity.

@ @0 I@O 30 40 @5@0 60 70 SO 90 00 INFARCTSIZEBYRb-82,SOFLV FDG IMAGING IN COMPARISONWITH @Rb FIGURE14. Infarctsizebasedonrubkliumimagescompared to infarctsize by FDG images, both usingautomated computer Metabolic imaging for myocardial viability is often con ized infarct sizing methods. Reproducedwith permissionfrom sideredthe bestvalidatedadvantageofPET overperfusion reference(144). imaging with thallium or positron perfusiontracers.How ever, this claim is based on only two clinical reports involving a total of39 patientsin the world'sPET literature farct areas, the relation between infarct size by rubidium (57,141), Table 3. The accuracy of FDG imaging for and FDG is less good, primarilydue to probable overesti predicting improvement of poorly contracting LV seg mates by the FDG method since some ofthe largestinfarct ments afterbypass surgerywas 78% (141) to 85% (57), as sizes by FDG would appear to be incompatible with sur comparedwith 65% (56) to 81% (135) for thallium.For vival. comparative purposes, Table 3 also lists the major clinical Nine out of 43 patients with acute evolving myocardial imaging trials of PET involving 470 patients reporting a infarctionstudiedwith FDG afterglucoseloading(21%) 95% or greater sensitivity and specificity for the diagnosis had no myocardial uptake anywhere in the heart or large of coronary artery disease. The numbers of studies and areas of normal myocardium that did not take up glucose relativeimprovement in diagnostic power supportsthe use after glucose loading despite normal contraction and per of PET for perfusion imaging as much as for identifying fusion. Eight ofthese nine patients were diabetic (144). viabilityandreversiblewallmotionabnormalities. Our results demonstrate that the size of infarcted myo The initial report on FDG in 17 patients indicates that cardium defined by abnormal rubidium kinetics is com 85% of hypokinetic segments taking up FDG improved parableto thesizeof infarctedmyocardiumdefinedby functionafterbypasssurgery(57). However,ejectionfrac lack of FDG uptake on PET images using appropriate tion improved significantly only in those with three or glucose loaded protocols. They also indicate that in myo more segments involved. Data was not provided on the cardial necrosisthe loss of cell membrane integrity as percent of patients with three or more viable regions that

TABLE 3 Comparison of Clinical lmaqing Trials for Patient Management of Viable Myocardium

TestAccuracYPatientsReferencePosNegPredictingDiagnostic

(57)left improvedpostop85%92%17Tillisch 141)byventricularwallmotion78%78%22Tamaki (139, FDGPredicting (135)left improvedpostop81%86%25Rozanski (56)byventricularwall motion65%58%31Tamaki (136)DiagnosisT120186% Sn78% Sp26lskandrian (32)tery of coronary ar 95%100%50Gould (35)withstenosis using @Rb94%95%193Demer (36)13NH397%a fast PET scanneror97%100%32Schelbert (37) 98%100% 93%49 146Yonekura Williams(119)

598 The Journal of Nuclear Medicine •Vol. 32 •No. 4 •April 1991 would warrant bypass surgery or the percent of patients scans showed fibrosis in these areas suggesting that FDG showing improved ejection fraction. uptake in the setting of acute myocardial infarction does Since FDG imaging is currently used for assessing myo not indicate viability. cardial infarction and viability (61 ), it is important to The mechanism for FDG uptake in recently infarcted emphasize the conditions under which it is applicable and myocardiumis unclear.In experimentalchroniccerebral those under which it is not reliable fo assessing ischemia infarction, elevated FDG uptake is observed in infarcted viability and necrosis. After oral glucose loading in chronic tissue in association with white blood cell phagocytosis of CAD and acute myocardial infarction, myocardial areas cellulardebris(148). In areaswith highlyactivephagocy with a rest perfusion defect and increased FDG uptake tosis, the intensity of FDG uptake in chronic cerebral after glucose loading (flow-metabolism mismatch) are hy infarction may be higher than FDG uptake of normal pothesized to be viable and may demonstrate improved brain.FDG uptakeby whitebloodcellsin myocardial left ventricular function after revascularization (5 7,61). infarction in acute short term experiments (6 hr) is insuf Areas with a perfusion defect and low FDG uptake after ficient (149) to explain these observations.However, glucose loading (flow-metabolism match) identify necrotic longer term experiments on white blood cell uptake of or fibroticmyocardiumshowingno improvementafter FDG in evolving myocardial infarction comparable to revascularization. After fasting and exercise stress in those observed for cerebral infarction or comparable to chronic CAD, FDG uptake identifies ischemic viable myo our observations in man have not been carried out. Con cardiumasa positiveimagewith no uptakein normalor sequently,observedFDG uptake in evolving myocardial necrotic tissue (44,61,65—67,71). infarction after fasting may be associated with phagocytic However, there are several circumstances in which myo activity ofwhite blood cells as found in cerebral infarction. cardial FDG uptake is not consistent or predictable and Myocardial uptake of FDG is also highly dependent on therefore fails to differentiate necrotic from viable tissue substrate availability (150) and catechol levels (151), (1,144—147).Normally perfusedand contractingmyocar thereby reflecting many conditions aside from viability. dium of some diabetic patients may not take up FDG WhenmyocardialuptakeofFDGisunpredictableortaken either with or without glucose loading, even after their up in myocardium by unclear mechanisms perhaps unre usual dose of insulin and/or oral hypoglycemic agents. lated to viability, the application of quantitative models This failure of FDG uptake in diabetics after glucose for deriving absolute glucose utilization in g/mg/min is loading may erroneously suggest necrosis when the myo inappropriate. Even under usual circumstances, absolute cardium is normal. In such instances, perfusion imaging measurements ofmyocardial flow or glucose consumption or measures of cell membrane integrity discussed subse in themyocardiumorbraindonotappearnecessaryfor quently are the major guides to viability or necrosis for patient studies since most clinical reports showing the real clinical decisions on intervention. benefits of PET imaging use end points of relative FDG In the fasting state at rest in normal and diabetic sub uptake or relative uptake of a perfusion tracer. jects,FDG uptakeeitherwith or withoutevolvingmyo Finally,in contrastto chronicallyreducedrestingflow cardial infarction is so variable as to be uninterpretable. to viable myocardium, reperfused myocardium during Additionally,afterfastingandat restin thesettingofacute acute myocardial infarction makes interpretation ofa per evolving myocardial infarction, intense FDG uptake may fusion-metabolism mismatch on FDG images quite com occur in areasof myocardiumthat are necrotic-docu plex. A defect on a perfusion image after reperfusionwith mentedby lackof FDG uptakeafterglucoseloading,left a patent artery suggests a no-reflow phenomenon associ ventricularakinesis,andarteriographicinvolvementofthe ated with necrosis and an FDG defect. Alternatively, per correspondingcoronaryartery(144—146).A similarob fusion may be normal or near normal, either with or servation has been made in experimental animals using withoutanFDG defect.In thiscase,a reversemismatch cardiac autoradiography (147). For these patients, necrotic may occur with no perfusion defect associated with an tissue erroneously appears viable due to intense FDG FDG defect, the significance of which has not been docu uptake.In our earlyexperiencebeforerecognizingthese mented. problems,two suchpatientshad bypasssurgeryon the With reperfusion in acute myocardial infarction, the basis of intense FDG uptake in the occluded artery distri zone of myocardium at risk is not readily defined by a bution but demonstratedno recoveryof function and resting perfusion defect, which, in chronic stable CAD, is remained in chronic heart failure postoperatively despite the basis for the perfusion-metabolism mismatch to define open bypass grafts (144—146).Therefore, we now do not viablemyocardium.Sincemostof ourpatientshadre carry out FDG imaging in the fasting, resting state in acute ceived thrombolysis therapy in this study, we avoided these evolvingmyocardialinfarction. problems by using relative FDG uptake compared with Pierard et al. (146) showedthat a large proportion of normal myocardium (after glucose loading) without ref patients with myocardial infarction and myocardial areas erenceto a metabolism/perfusionratio,whichhasnot of intense FDG uptake did not improve LV function in beendemonstratedas diagnostically useful in the setting these areas after bypass surgery; at 9-mo follow-up, FDG of acute myocardial infarction and/or reperfusion.

PETPerfusionImagingandNuclearCardiology•Gould 599 STRESS PERFUSIONIMAGING FOR ASSESSING has been shown to be comparable to FDG PET imaging MYOCARDIALVIABILITY (152,153).

For thallium stress testing, stress perfusion defects not COMBINED ASSESSMENT OF CORONARY FLOW present at rest or redistribution indicate areas of limited RESERVE, MYOCARDIALINFARCT SIZE, VIABILITY, coronary flow reserve that are measures of zones at risk, AND ZONE AT RISK not viability per se. For severe stenoses that reduce resting As a measurement conceptually separate from infarct flow, the defect severityon stressperfusion images may be size and viability, we define the area of reduced flow more intense and/or larger than at rest simply because reserve or zone at risk by dipyridamole perfusion imaging adjacent areas with normal flow reserve have greaterper (1,32,34,35) rather than rest perfusion imaging. Therefore, fusion tracer activity associated with higher stress flow we conceptually use the term “viability―of myocardium unrelated to viability in the area of the defect. Thus, in reference to a nontransmural, mixed, or incomplete reversiblestressdefectsreflectflow capacityof normal infarction as distinct and different from what is loosely myocardiumarounda restingdefectbut do not provide termed in the literatureas “viability―ofareas with normal information on viability of myocardium within a resting resting perfusion within a zone of reduced flow reserve or perfusion defect. stress perfusion defect due to a stenotic artery. Imaging With reperfusion, the definition of viability becomes analogs of metabolism (FDG) or membrane function more complex since the zone at risk characteristically (82Rb)identifies viabilityand necrosis in a mixed or incom contains a mix of viable and necrotic tissue with a patent plete infarction either with or without a resting perfusion artery and adequate flow. In this case, stress-induced en defect with or without reperfusion. Dipyridamole or exer largement of a perfusion defect may indicate additional cise stress perfusion imaging identifies zones of adequate zones at risk with low coronary flow reserve around the resting flow but low flow reserve at risk ofpotential further damaged area or limited flow capacity in the central dam necrosis. Thus, in the literature, viability is loosely defined aged region due to edema or obliterationof vascular depending on how it is measured. channels. However, it does not provide information on Complete analysis of myocardial infarction requiresas whether there is viable myocardium in the more central sessing viability of myocardium in the central infarcted or reperfused area of injury. Similarly, the newer technetium partiallyinfarctedmyocardiumat restaswellasthelarger imaging agents do not provide measures of viability nor zones at risk around that region or in other parts of the reflect cell membrane integrity. heartwith reducedflow reservesignifyingproximal ste Although the concept of a perfusion-metabolism mis noses. Thus, complete clinical evaluation for a major match is important and often predicts viable myocardium intervention like bypass surgery or PTCA requires imaging in resting defects, its general clinical applicability may be coronary flow reserve and cell membrane integrity by the somewhat limited due to the difficulty of defining a per potassium analog 82Rbor cell metabolism by the glucose fusion zone at risk under resting conditions and/or after analog FDG, and perfusion with ‘3N-ammonia. reperfusion. Consequently, we utilize FDG uptake and/or rubidium washoutat restnot only with referenceto resting PRACTICALCONSIDERATIONSFOR CLINICAL PET perfusion but also compared with dipyridamole defects, in orderto completelyevaluatethe proportionof viableor WITH 2Rb necrotic myocardium mixed together in the resting defect Although FDG is used for assessing myocardial viability, as well as the zone at risk defined by limited flow reserve. it requires a cyclotron and takes approximately 2 hr for From a study of rest-exercise PET using ‘3N-ammonia, the perfusion-metabolism study including the sequence of 34 of 48 segments with transient perfusion defects (71%) imaging ‘3N-ammoniafor perfusion, the time for FIX) improved after bypass surgery (56). Of myocardial seg uptake, and FDG imaging for metabolic activity. This 2- ments with transient reversible perfusion defects on thal hr period for one viability study limits the patient volume hum exercise imaging, 65% (15/23) (56) to 81% (35/43) to approximately four per day, which may be below the (135) respectivelyimproved afterbypasssurgery(56,135). minimum volume for “economicbreakeven―depending Therefore,basedon publishedliterature,predictionof on the clinical charge. Assessing myocardial viability with postoperative recovery in 78% (141 ) to 85% (57) of left generator-produced 82Rb at rest (without dipyridamole ventricularsegmentstakingup FDG is not muchbetter imaging) requires only 35 mm to complete and therefore than the improved postoperative function in 65% (56) to may be more practical from a clinical point of view. 8 1% (135) of left ventricular segments with reversible For clinical applications, 82Rbis well suited for assessing stressdefects by PET perfusion or thallium perfusion both viability/infarctsizeat restand the zoneat risk by imaging. Furthermore, studies ofleft ventricular segments dipyridamole perfusion imaging in a brief (1 hr) single donotreflecttheproportionof patientswith largeenough study. The use of dipyridamole in the peri-infarction viableareasto warrantinterventions.Finally,identifying period is safe and useful for risk stratification (1). A viablemyocardiumby the thalliumreinjectiontechnique rubidiumwashoutstudyfor infarctsizeis carriedout as

600 The Journal of Nuclear Medicine •Vol. 32 •No. 4 •April 1991 part of a rest-dipyridamole sequence by collecting the rest 2. Statement. The best documented application of PET data in list mode.Dipyridamoleperfusionimagesusing is for myocardial viability. Response. Table 3 shows rubidium are then obtained. For the rest-dipyridamole that the validation of FDG by PET showing im comparison to define the zone at risk, all the list mode proved LV function is based on only 39 patients in data are reconstructed into a single rest image and com the world's literature. Furthermore, other university pared to the dipyridamole image. For the resting rubidium centers have demonstrated major problems with washout analysis to determine infarct size-viability, the using FDG for viability (144—147,150,151).These rest data are divided into early and late rubidium images. limitations include 20%—25%of studies being unin Acquiring the entire data set for infarct size-viability and terpretable (144,145), intense uptake of FDG in rest-dipyridamole images with 82Rb requires 1 hr in our necroticmyocardium(144—147)as evidencedby laboratory.The same sequence of viability and stress im failureto improve left ventricularfunction after suc ages with ‘3N-ammoniaand FDG requires4—Shrs due to cessfulbypasssurgery(146), and dependence of FDG longer data collection times for both these tracers, longer uptake on substrate availability (150) and catechol time for decay of ‘3N-ammonia for rest-stress studies in levels (151). In comparison, Table 3 shows that there sequence, and longer uptake period of FDG. are more clinical cases from more centers validating Thus, using rubidium kinetics to evaluate myocardial high sensitivity and specificity for perfusion imaging viability may have several advantages. First, it does not than FDG for assessingmyocardial viability. Finally, require the expense of a cyclotron. Second, the entire rest Bonow (152,153) has demonstrated that PET of rubidium study takes significantly less time than for ‘3N FDG has no advantage over reinjection of thallium ammonia and FDG, thereby allowing higher patient vol 201 for assessing viability. umes at reasonable cost. By combining washout measure ments with dipyridamole stress using rubidium, complete 3. Statement. Only one study reports a direct compari information on viability of injured myocardium and ste son between PET and SPECT, showing them to be nosis severity of other myocardium at risk are obtained. equivalent. Response. In the study by Tamaki (120), Finally, it avoids the problems of inadequate FIX) uptake the prevalence of disease was 94% with only three in some diabetics and variable patient response to fasting normals studied. This small number ofnormals does or glucose loading. not permit determination of specificity, yet it was the main basis of the ACC report. Furthermore, this REVIEW OF THE AMERICAN COLLEGE OF study population had advanced, severe, symptomatic CARDIOLOGY(ACC) POSITION PAPER clinical disease. For a population with such advanced disease, virtually any diagnostic modality is accurate, The ACC report makes a number of factually incorrect including a simple history. Finally, the Tamaki paper statements as follows: used treadmill exercise stress for patients on cardiac drugs, which blunt heart rate response. Other PET 1. Statement. The major data on PET perfusion imag studies have used dipyridamole, which increases flow ing are from only one large center. Response. As more than exercise does. However, even with this shown in Table 1, five papersfrom four centers have advanceddiseasepopulationTamakiconcludedthat all reportedcomparablesensitivity and specificity of PET was betterthan SPECT for regional localization 95—98%,including Gould (32), Schelbert (36), Wil of disease, a conclusion ignored by the ACC report. hams (119), Demer (35) and Yonekura (37). In The study by Go (121,122) comparing PET with comparison, seven papers from six centers published SPECT also utilized a population with advanced since 1983 report that the diagnostic specificity of disease and processed the PET images with SPECT thallium exercise testing in the last 3,1 19 cases is softwaredisplaysthatdegradedthe PETdata. Despite 53% and sensitivity is 84%, as shown in Table 1 these drawbacks the specificity of PET was better (106—114).Thislimitedspecificityis furthercon than SPECT. A direct comparison with an appropri firmed by reports from the most recognized SPECT ate study population and appropriate hardware-soft experts in the United States(11 1) as well as in Europe warewas made by Demer(35), whose study included (1 15). Two independent analyses (116—118) have a subset of 46 patients who had both PET and ruled out referral bias to explain away this low spec SPECT. Of patients with moderately severe CAD ificity, including a report from the most recognized (coronary flow reserve of <3), PET identified 95%, PET center (118). Uncathed “normalcy―rates are whereasthallium stresstestingidentified58%. Of not appropriate substitutes for observed specificity patients with mild disease (flow reserveof 3—4),PET (116—118). Low specificity is primarily due to lack identified 52%, whereas thallium stress testing iden ofattenuation correction (31) and a number of other tified none. Thus, at each level of disease severity, well-defined factors (124) that are not accurately the PET was considerably more accurate than corrected. SPECT. Therefore, 3 studies directly comparing the

PETPerfusionImagingandNuclearCardiology•Gould 601 two modalities (35,121—123)have shown PET to be multiple university and private practice sites for routine, superiorto SPECT. high volume, cardiac diagnosis and management for the 4. Statement. The specificityof PET was 74% in the following indications: single large study reported (35). Response. The ACC 1. Diagnosing and assessing severity and location of reportmisquotes the specificity, which was correctly coronary artery stenosis in patients with angina pec reported in the peer-reviewed paper as 95%. The toris, atypical chest pain, or risk factors for coronary ACC committee members utilized an erroneous re artery disease, to determine the need for arteriog calculation of specificity from peer-reviewed data raphy, medical, or mechanical intervention. based on their incorrect reclassification of patients. 2. Identification of additional myocardial areas at risk Their incorrect modification of this data was NOT and myocardial viability/necrosis in patients with peer reviewed. The committee review also ignored evolving or old myocardial infarction, thereby estab the subset of patients in that report with a direct lishing the need for arteriography,medical, or me comparison of PET and SPECT, as noted above. chanical intervention. Despite this erroneous non-peer-reviewed miscalcu 3. Evaluation of collateral function which, if adequate lation and misquote, several other centers have re under appropriate circumstances, may make me ported specificity of 95% including Schelbert (36), chanical intervention unnecessary. Tamaki (37), Williams (119), and Gould (32), com 4. A moreaccuratesubstitutefor currentstandardex parableto the 95% reportedby Demer (35) but ercise perfusion imaging done for whatever condi misquoted in the ACC report. tions are medically indicated. 5. Statement. The ACC report indicates a problem with 5. Ruling out significant coronary artery disease with a availability of the 82Rbgenerator. Response. Careful high specificity in patients who would otherwise have follow-up has indicated that not one member of the an exercise stress test with poor specificity leading to ACC committee contacted either the supplier unnecessary arteriography. (Squibb) or any rubidium clinical site to determine whether rubidium availability was a problem. The Traditionally in medicine, the patient and physician committee also overlooked Squibb publications and wait until angina pectoris, myocardial infarction, arrhyth news releases establishing a public record of rubid mia, heart failure, or sudden death leads to medical or ium availability that has been confirmed by every mechanicalintervention.However,symptomsareusually clinical site. late manifestations of advanced disease in which reversal 6. Statement. The ACC report suggests that lack of of coronary artery stenosis is difficult and mechanical approval of dipyridamole by FDA limits the use of procedures are often necessary. As a guide to therapy, 82Rbfor PET. However, the ACC report fails to angina pectoris usually indicates severe coronary artery indicate that no NDA for cyclotron-produced radi stenosis, does not predict sudden death or acute myocar otracer has been approved by the FDA for commer dial infarction, and can be improved or eliminated by cial clinical use. However, rubidium now also has medical therapy without affecting progression of disease. been approved. Intravenous dipyridamole for diag By the time angina pectoris develops, the optimal oppor nostic imaging has been approved by the FDA for tunity for reversal therapy has passed. Consequently, di clinical use.It hasbeen and is now usedwidely for agnosing coronary artery stenosis and its severity at the both thallium and rubidium perfusion imaging under mildest, earliest possible stage, particularly before symp IND status in over 4,000 patients prior to FDA toms, is important for instituting vigorous riskfactormod approval. ification. An initial reporton clinicalPET wasapprovedby the Although reactive treatment triggeredby symptoms ap ACC Imaging Committee in 1988 and circulated within propriatelyremainscentralto cardiovascularmedicine, ACC's administrative review committees. It was a careful, advanceddiagnosticand therapeutictechnologyprovide balanced review favorable to cardiac PET for both perfu the opportunity for another major step in the evolution of cardiovascular medicine. PET has sufficient diagnostic sion imaging and viability. However, this initial favorable report was subsequently revised by new and prior com power for a new therapeutic approach based on routine, mittee members, none of whom have clinical experience economicalnoninvasivediagnosisand assessmentof se verityin symptomaticor asymptomaticindividualswith with 82Rband none of whom routinely carry out or have published data on routine clinical PET perfusion imaging intense dietary or medical treatment for reversalof CAD. Such therapy is not appropriatefor the general population for managing cardiac patients. or for individuals without known coronary artery stenosis or a specific genetic susceptibility to it. This approach is CONCLUSION therefore targeted, individually specific, preventive inter Cardiac PET using 82Rbhas proven accurate, nearly as vention. For severe silent disease identified by PET, coro definitive as the coronary arteriogram, and economical in nary arteriographyand mechanical intervention may be

602 The Journal of Nuclear Medicine •Vol. 32 •No. 4 •April 1991 indicatedto prevent myocardialinfarction or sudden Dis l987;56:l—4. 11. Younis LT, ByerS 5, Shaw L, Barth G, Goodgold H, Chaitman BR. death. Prognostic importance of silent myocardial ischemia detected by intra In contrastwith the traditionalapproachbasedon symp venous dipyridamole thallium myocardial imaging in asymptomatic pa toms, this new approach is based on functional and ana tients with coronary artery disease. J Am Coil Cardiol 1989;l4:1635— 1641. tomic characterization ofstenoses and myocardial viability 12. Brown KA, O'Meara J, Chambers CE, Plante DA. Ability of dipyrida with sufficient reliabilityin symptomatic or asymptomatic mole-thallium-201 imaging one to four days after acute myocardial patients to justify medical or mechanical interventions infarction to predict in.hospital and late recurrent myocardial ischemic events. Am J Cardiol l990;65:l60—l67. having significant benefits and risks, or to avoid unneces 13. Hendel RC, Layden ii, Leppo JA. Prognostic value of dipyridamole sary ones. The scientific basis for this approach has been thallium for evaluation ofischemic heart disease. J Am Coil documentedandfeasibilitydemonstrated(1). Cardioll990;lS:l09—l16. 14. Leppo J, Plaja J, Gionet M, Tumolo J, Paraskos JA, Cutler BS. Nonin Economic analysis ofcardiac PET using 82Rbfor symp vasive evaluation of cardiac risk before elective vascular surgery. J Am tomatic or asymptomatic subjects with coronary athero CoilCardiol l987;9:269—276. sclerosis demonstrates that it is less expensive than the cost 15. Midwall J, Ambrose J, Pichard A, Abedin Z, Herman MV. Angina pectoris before and after myocardial infarction. Chest l982;81:681—686. of SPECT stress testing and associated definitive coronary 16. Reunanen A, Aromas A, Pyorala K, Punsar 5, Maatela J, Kneki P. The arteriography in the many normal individuals with false Social Insurance Institution's coronary heart disease study: Baseline data positive thallium stress tests (1). and 5.y@ mortality experience. Acta MedScandSuppl l983;673:67—81. 17. Kannel WB, Abbot RD. Incidence and prognosis of unrecognized myo. It is significant that in over 4,000 clinical cardiac PET cardial infarction. An update on the Framingisam Study. N Engi J Med studies at university and private sites, over 90% of private l984;3l1:1144—1147. health insurers have reimbursed fully for cardiac PET with 18. Lown B. Sudden cardiac death: The major challenge confronting contem porary cardiology. Am J Cardiol l979;43:3l3—328. 82Rbbecauseit is approvedby the FDA and becausethey 19. Oliver MF. Strategies for preventing and screening for coronary heart realizethat it preventsmany unnecessarycoronaryarterio disease.Brlleartf1985;54:l—5. grams and mechanical interventions while correctly iden 20. UhI GS, Froelicher V. Screening for asymptomatic coronary artery dis ease.JAm CoilCardiol 1983;l:946—955. tifying even asymptomatic individuals for whom such 21. Houck PD. Epidemiology oftotal cholesterol to HDL ratio in 11,669 Air procedures are essential. Therefore, PET provides an eco Force personnel and coronary artery anatomy in 308 healthy aviators. J nomical basis for selective, specific medical intervention Am Coil Cardiol l988;l l:222A. 22. Love WD, Burch GE. A study in dogs ofmethods suitable for estimating to reverse coronary atherosclerosis and/or selective me the rate of myocardial uptake of Rb.86 in man, and the effect of norepi chanical intervention for severe disease to prevent myo nephrine and pitression on Rb-86 uptake. J Clin Invest l957;36:468—478. cardialinfarctionor suddendeath. 23. Yano Y, Cahoon JL, Budinger iT. A precision flow.controlled Rb.82 generator for bolus or constant-infusion studies of the heart and brain. J NuclMed 198l;22:l006—l0l0. ACKNOWLEDGMENTS 24. Neirinckx RD, Kronauge iF, Gennaro GP. Evaluation of inorganic absorbents for rubidium-82 generator I. Hydrous Sno2. J NucI Med Supported in part by NIH Grants ROI-HL26862, ROl l982;24:898—906. HL42453, and RO1-HL26885and a Joint Collaborative Re 25. Mullani NA, Gould KL. First pass regional blood flow measurements search Agreementwith the Clayton Foundation for Research, with external detectors. JNuclMed 1983;24:577—581. 26. Mullani NA, Goldstein RA, Gould KL, Fisher Di, Marani 5K, O'Brien Houston, Texas. HA. Myocardial perfusion with rubidium.82. I. Measurement of extrac tion fraction and flow with external detectors. J NucI Med 1983;24:898— REFERENCES 906. 27. Goldstein RA, Mullani NA, Fisher D, Marani 5, Gould KL, O'Brien HA. I. Gould KL. Coronary artery stenosis. New York, Elsevier Scientific Pub Myocardial perfusion with rubidium-82. II. The effects of metabolic and lishing, 1990. pharmacologic interventions. J NuclMed 1983;24:907—915. 2. Michael Fisher, ed. Guide to clinical preventive services. Report of the 28. Schelbert HR, Phelps ME, Huang SC, et al. N.l3 ammonia as an indicator U.S. preventive services task force. An assessment oldie effectiveness of ofmyocardial blood flow. Circulation 198l;63:l259—1272. 169 interventions. Baltimore: Williams and Wilkins; 1989; 3—11. 29. Schelbert HR. Phelps ME, Hoffman El, Huang SC, 5dm CE. Regional 3. Olofsson BO, Bjerle P, Aberg T, Osterman G, Jacobsson KA. Prevalence myocardial perfusion assessed with N-l3 labeled ammonia and positron of coronary artery disease in patients with valvular heart disease. Acta emission computerized axial tomography. Am J Cardiol l979;43:209- Med Scand1985;218:365—371. 218. 4. Langou RA, Huang EK, Kelley Mi, Cohen LS. Predictive accuracy of 30. Yoshida K, Endo M, Himi T, et al. Measurement ofregional myocardial coronary artery calcification and abnormal exercise test for coronary blood flow in hypertrophic cardiomyopathy: Application ofthe first pass artery disease in asymptomatic men. Circulation 1980;62(6):1 196-1203. flow model using N.l3 ammonia and PET. Am J Physio Imaging 5. Campbell S. Barry J, Rebecca GS, ci al. Active transient myocardiaJ l987;4:97—l04. ischemia during daily life in asymptomatic patients with positive exercise 31. Xu EZ, Mullani NA, Gould KL, Anderson WL. A segmented attenuation tests and coronary arterydisease. Am J Cardiol 1986;57:10l0—1016. correction CSAC/for PET. J Nud Med@1990; 31:000-000. 6. Resnekov L. Silent myocardial ischemia: Therapeutic implications. Am J 32. Gould KL, Goldstein RA, Mullani NA, et al. Noninvasive assessment of Med 1985;79:30—34. coronary stenosis by myocardial perfusion imaging during pharmacologic 7. Fleg JL, Gerstenblith G, Zonderman AB, ci al. Prevalence and prognostic coronary vasodilation. VIII. Clinical feasibilityofpositron cardiac imaging significance of exercise induced silent myocardial ischemia detected by without a cyclotron using generator-produced rubidium-82. J Am Coil thallium scintigraphy and in asymptomatic volun. Cardiol 1986;7:775—789. teers. Circulation 1989;81:428—436. 33. Kirkeeide RL, Gould KL, Parsel L. Assessment of coronary stenosis by 8. GottliebSO,WeisfeldtML,OuyangP.MellitsED,Gerstenblith0. Silent myocardial perfusion imaging during pharmacologic coronary vasodila. ischemia as a marker for early unfavorable outcomes in patients with tion. VII. Validation of coronary flow reserve as a single integrated unstableangina.NEngIJMed l986;314:l214-12l9. functional measure of stenosis severity reflecting all its geometric dimen. 9. Cohn PF. Clinical importance ofsilent myocardial ischemia in asympto. sions. JAm Coil Cardiol 1986;7:103—113. matic subjects. Circulation 1989;8l:691—693. 34. Gould KL. Identifying and measuring severity ofcoronary arterystenosis. 10. Cohn PF. Silent myocardial ischemia: present status. Mod Concepts CV Quantitative coronary arteriography and positron emission tomography.

PET Perfusion Imaging and Nudear Cardiology •Gould 603 Circulation1988;68:237—245. l986;3l4:884—888. 35. Demer LL, Gould KL, Goldstein RA, et al. Assessment of coronary artery 58. Goldstein RA, Kirkeeide R, Smalling RW, ci al. Changes in myocardial disease severity by positron emission tomography. Comparison with perfusion reserve after PTCA: Non-invasive assessment with positron quantitative arteriography in 193 patients. Circulation l989;79:825—835. tomography.JNuclMed l987;28:1262—l267. 36. Schelbert HR. Wisenberg G, Phelps ME, et al. Non-invasive assessment 59. Gould KL, Buchi M, Kirkeeide RL, Ornish D, Stein E, Brand R. Reversal of coronary stenosis by myocardial imaging during pharmacologic coro of coronary artery stenosis with cholesterol lowering in man followed by nary vasodilation. VI. Detection ofcoronary artery disease in man with arteriography and positron emission tomography [Abstract]. J Nuci Med intravenous N-13 ammonia and positron computed tomography. Am J l989;30:345. Cardiol1982;49:l197—1207. 60. Ornish DM, Scherwitz LW, Brown SE, et al. Can lifestyle changes reverse 37. Yonekura Y, Tamaki N, Senda M, et aL Detection of coronary artery atherosclerosis? Lancet 1990;336:l29—l33. disease with ammonia and high-resolution positron-emission computed 61. Schelbert HR. Buxton D. Insights into coronary artery disease gained tomography. Am Heart J l987;l 13:645—654. from metabolic imaging. Circulation l988;78:496—505. 38. Selwyn AP, Allan RM, L'Abbate AL, et al. Relation between regional 62. Brunken K, Schwaiger M, Grover-McKay M, Phelps M, Tillisch J, myocardial uptake of rubidium-82 and perfusion: absolute reduction of Schelbert H. Positron emission tomography detects tissue metabolic ac cation uptake in ischemia. Am J Cardiol l982;50:1l2—l21. tivity in myocardial segments with persistent thallium perfusion defects. 39. Deanfleld JE, Shea Mi, Wilson RA, Horlock P. de Landsheere CM, JAm Coil Cardiol l987;lO:557—567. Selwyn AP. Direct effects of smoking on the heart: silent ischemic 63. Goldstein PA, Muilani NA, Wong WH, et al. Positron imaging of disturbances ofcoronary flow. Am I Cardiol l986;57:l005—1009. myocardial infarction with rubidium-82. I Nuc! Med 1986;27:1824— 40. Gould KL, Kirkecide R, Buchi M. Coronary flow reserve as a physiologic 1829. measure ofstenosis severity. I. Relative and absolute coronary flow reserve 64. Geltman EM, Biello D, Welch MJ, et al. Characterization of transmural during changing aortic pressure. II. Determination from arteriographic myocardial infarction by positron.emission tomography. Circulation stenosis dimensions under standardized conditions. J Am Coil Cardiol l982;65:747—755. l990;15:459—474. 65. Schwaiger M, Brunken R, Grover-McKay M, et al. Regional myocardial 41. Hicks K, Ganti G, Mullani N, Gould KL. Automated quantitation of 3D metabolism in patients with acute myocardial infarction assessed by cardiac PET for routine clinical use. JNuclMed l989;30:l787—l797. positron emission tomography. JAm Coil Cardiol l986;8:800—808. 42. Gould KL. Percent coronary stenosis: Barteredgold standard, pernicious 66. Marshal RC, Tillisch JH, Phelps ME, Huang SC, Carson R, Henze E, relic, or clinical practicality? JAm Coil Cardiol l988;l 1:886—888. Schelbert HR. Identification and differentiation of resting myocardial 43. Goldstein PA, Kirkeeide R, Demer L, et al. Relations between geometric ischemia and infarction in man with positron computed tomography. dimensions of coronary artery stenosis and myocardial perfusion reserve “F-labeledfluorodeoxyglucose and N-ammonia. Circulation l983;67: in man. I C/in Invest l987;79:l473—l478. 766—778. 44. Camici P, Araujo LI, Spinks T, et al. Increased uptake of F-l8 fluoro 67. Brunken R, Tiilisch J, Schwaiger M, Child JS, Marshall R, Mandelkern deoxyglucose in postischemic myocardium of patients with exercise M, Phelps ME, Schelbert HR. Regional perfusion, glucose metabolism, induced angina. Circulation l986;74:281—292. and wall motion in patients with chronic electrocardiographic Q-wave 45. Gould KL, Lipscomb K, Hamilton GW. A physiologic basis for assessing infarctions evidence for persistence ofviable tissue in some infarct regions critical coronary stenosis instantaneous flow response and regional dis by positron emission tomography. Circulation 1986;73:95l—963. thbution during coronary hyperemia as measures of coronary flow re 68. Ter-Pogossian MM, Klein MS. Markham J, Roberts R, Sobel BE. Re serve. Am J Cardiol l974;33:87—94. gional assessment of myocardial metabolic integrity in vivo by positron 46. Gould KL. Noninvasive assessment of coronary stenosis by myocardial emission tomography with @C-labeledpalmitate. Circulation l980;6l: imaging during coronary vasodilation. I. Physiologic principles and cx 242—255. perimental validation. Am J Cardiol l978;4l:267—278. 69. Goldstein RA. Kinetics of rubidium-82 after coronary occlusion and 47. Gould KL, Westcott JR. Hamilton GW. Noninvasive assessment of reperfusion.I C/inInvest1985;75:l131—I137. coronary stenosis by myocardial imaging during coronary vasodilation. 70. Garza D, Sease DR, Merhige ME, Hicks K, MUHaniN, Gould KL. Non II. Clinical methodology and feasibility. Am I Cardiol l978;4l:279—287. invasive identification of viable versus infarcted myocardium by auto 48. Albro PC, Gould 1(1, Westcott Ri, Hamilton GW, Ritchie JL, Williams mated three-dimensional cardiac positron emission tomography with DL Noninvasive assessment ofcoronary stenosis by myocardial imaging generator-produced rubidium-82 [Abstract). J Nuc/Med l989;30:865. during pharmacologic coronary vasodilation. III. Clinical trial. Am J 71. Grover-McKay M, Schelbert HR. Schwaiger M, et al. Identification of Cardiol l978;42:75l—760. impaired metabolic reserve by atrial pacing in patients with significant 49. Gould KL, Kelley 1(0, Bolson EL Experimental validation of quantita coronary artery stenosis. Circulation l986;74:28l—292. tive coronary arteriography for determining pressure-flow characteristics 72. Demer L, Gould KL, Kirkeeide R. Assessing stenosis severity: coronary ofcoronary stenosis. Circulation l982;66:930—937. flow reserve, collateral function, quantitative coronary arteriography, 50. Gould KL. Quantification of coronary artery stenosis in vivo. Circ Res positron imaging, and digital subtraction . A Review and 1985;57:341—353. Analysis. Prog Dis l988;30:307—322. 51. Brown BG, Bolson E, Frimer M, Dodge HT. Quantitative coronary 73. Demer LL, Gould KL, Goldstein R, Kirkeeide RI. Non invasive assess arteriography: Estimation of dimensions, hemodynamic resistance, and ment ofcomnaiy artery collaterals in man by PET perfusion imaging. J atheroma massofcomnaryarterylesionsusingthe artenogram and digital NuciMed l990;3l:259—270. computation. Circulation l977;55:329—337. 74. Gould KL. Coronary steal—Isitclinicallyimportant? Chest 1989;96:227— 52. Marcus ML, Skorton Di, Johnson MR. CollinsSM, Harrison DO, Kerber 229. RE. Visual estimates of percent diameter coronary stenosis: “Abartered 75. Kehtarnavaz N, DeFigueiredo RiP. A novel surface reconstruction and goldstandard.―JAm CoilCardiol1988;4l:882—885. display method for cardiac PET imaging. IEEE Trans , 53. White CW, Wright CB, Doty DB, et al Does visual interpretation of the 1984;MI-3:l08—l15. coronary arteriogram predict the physiologic importance of a coronary 76. Gould KL. Assessment of coronary stenosis by myocardial perfusion stenosis? N Eng/JMed l984;3l0:8l9—824. imaging during pharmacologic coronary vasodilatation. IV. Limits of 54. Marcus ML, Harrison DO, White CW, McPheeson DD, Wilson RF, stenosis detection by idealized, experimental, cross-sectional myocardial KerberRE. Assestingthe physiologic significance of coronary obstructions imaging.Am J Cardioll978;42:76l—768. in patients Importance of diffuse undetected athemsclerosit Frog Car 77. Gould KL, Lipscomb K. Effects of coronary stenosis on coronary flow diovascDisl988;3l:39—56. reserve and resistance. Am J Cardiol l974;34:48—55. 55. Sobel BE, Geltman EM, Tiefenbrunn AS, ci al. Improvement of regional 78. Gould 1(1, Lipscomb K, Calvert J. Compensatory changes of the distal myocardial metabolism after coronary thrombolysis induced with tissue coronary vascular bed during progressive coronary constriction. Circula type plasminogen activator or streptokinase. Circulation l984;69:983— tion l975;5l:1085—1094. 990. 79. Gould KI, Hamilton GW, Lipscomb K. Kennedy JW. A method for 56. Tamaki N, Yonekura Y, Yamashita K, et al. Value ofrest-stress myocar assessing stress-induced regional malperfusion during coronary arteriog dial positron tomography using N-l3 ammonia for the preoperative raphy: experimental validation and clinical application. Am J Cardiol prediction ofreversible asynergy. J Nuc! Med 1989;30:l302—l310. l964;34:557—564. 57. Tillisch J, Brunken R, Marshall R, et aL Reversibility of cardiac wall 80. Gould KL, Schelbert HR. Phelps ME, Hoffman EJ. Noninvasive assess motion abnormalities predicted by positron tomography. N Engi J Med ment of coronary stenosis with myocardial perfusion imaging during

604 The Journal of Nuclear Medicine •Vol. 32 •No. 4 •April 1991 pharmacologic coronary vasodilation. V. Detection of 47% diameter arteriography and resting and stress electrocardiography. Circulation coronary stenosis with intravenous N.l3 ammonia and positron emission l977;56:66—78. tomography in intact dogs. Am I Cardiol 1979;43:200—208. 104. Ritchie JL, Zaret BL, Strauss HW, et al. Myocardial imaging with 81. Gould KL. Quantification of coronary artery stenosis in vivo. Circ Res thallium-201: A multicenter study in patients with angina pectoris or 1985:57:341—353. acute myocardial infarction. Am J Cardiol 1978;42:345-350. 82. Gould KL, Kelly KO, Bolson EL Experimental validation of quantitative 105. Maddahi J, Garcia EV, Berman DS, Waxman A, Swan HJC, ForresterJ. coronary arteriography for determining pressure-flow characteristics of Improved non-invasive assessment ofcoronary artery disease by quanti coronary stenosis. Circulation l982;66:930—937. tative analysis of regional stress myocardial distribution and washout of 83. Gould KL, Kelly KO. Physiological significance ofcoronary flow velocity thallium-201. Circulation 198l;64:924—935. and changing stenosis geometry during coronary vasodilation in dogs. 106. DePasquale EE, Nody AC, DePuey EG, ci al. Quantitative rotational Circ Res l982;50:695—704. thallium-201 tomography for identifying and localizing coronary artery 84. Gould KL Positron tomography for heart disease: clinical update. In: disease. Circulation l988;77:3l6—327. Braunwald E, ed. Heart disease, a textbook ofcardiovasczdar medicine. 107. Van Train KF, Berman DS, Garcia EV, et al. Quantitative analysis of Third edition. 10th update. New York: W.B. Saunders; 1990. stress thallium-201 myocardial scintigrams A multicenter trial. J Nuc! 85. Josephson MA, Brown 8G. Hecht HS, Hopkins .1, Pierce CD, Petersen Med l986;78:l7—25. R. Noninvasive detection andlocalization ofcoronary stenosisin patients 108. Ranhosky A, Gerlag DM. Quantitative interpretation provides no advan comparison ofrestingdipyndamokand exercisethallium-20l myocardial tage over qualitative interpretation in intravenous dipyridamole thallium perfusion imaging. Am Heart) l982;103:l008—1018. imaging. Circulation l988;87:II—432. 86. Leppo J, Boucher CA, Okada RD, Newell JB, Strauss W, Pohost GM. 109. Iskandrian AS, Heo 3, Kong B, Lyons E. Effect ofexercise level on ability Serial thallium-20l myocardial imaging after dipyridamole infusion. Cir ofthallium-20l tomographic imagingin detectingcoronary artery disease: culation1982;66:649—657. Analysis of46l patients. JAm Coil Cardiol 1989;14: 1477—1486. 87. lskandrian AS, Heo J, Askenase A, Segal BL, Auerbach N. Dipyridamole 110. Kahn JK, McGhie I, Akers MS. et al. Quantitative rotational tomography cardiac imaging Am Heart J l988;1 15:432-443. with @°‘Tland @Tc2-methoxy-isobutyl-isonitrile. A direct comparison 88. Leppo J. Dipyridamole-thallium imaging the lazy man's stress test. J in normal individuals and patients with coronary artery disease. Circula Nuci Med 1989:30:281—287. tion 1989;79:1282—293. 89. Verani MS. Mabmarian .JJ, Hixson JB, Boyce TM, Staudacher RA. 111. Van Train KF, Maddahi J, Berman DS, ci al. Quantitative analysis of Diagnosis ofcoronary artery disease by controlled coronary vasodilation tomographic stress thallium-201 myocardial scintigrams: A multicenter with adenosine and thallium-20l scintigraphy in patients unable to exer trial. JNuclMed l990;3l:l 168—79. cise. Circulation 1990;82:80—87. 112. Nolewajka Ai, Kostuk Wi, Howard J, Rechnitzer PA, Cunningham DA. 90. Brown G, Josephson MA, Peterson RB, et al. Intravenous dipyridamole thallium-201 stress myocardial imaging an evaluation of fifty-eight combined with isometric handgrip for near maximal acute increase in asymptomatic males. C/in Cardiol 1981;4: 134—138. coronary flow in patients with coronary artery disease. Am I Cardiol 113. Schwartz RS, Jackson WG, Ceio PV, Hickman JR. Exercise thallium 18148:1077—1085. 201 scintigraphy for detecting coronary artery disease in asymptomatic 91. Wilson RF, Laughlin DE, Ackell PH, et al. Transluminal subselective young men. JAm CoilCardiol 1988;ll:80A. measurement of coronary artery blood flow velocity and vasodilator 114. Bungo MW, Leland OS. Discordance of exercise thallium testing with reserve in man. Circulation 1985;72:82—92. coronary arteriography in patients with atypical presentations. Chest 92. Harrison DG, White DW, HiratZkaLF, et al. The value oflesion cross l983;83:l12—I16. sectional area determined by quantitative coronary angiography in assess 115. Koistinen MH, Huikuri HV, Pirttiaho H, Linnaluoto MK, Takkunen iT. ing the physiological significance of proximal left anterior descending Evaluation of exercise electrocardiography and thallium tomographic coronary arterial stenosis. Circulation l984;69:llll—lll9. imaging in detecting asymptomatic coronary artery disease in diabetic 93. White CW, Wright CB, Doty DB, et al. Does visual interpretation of the patients. Br Heart J l990;63:7—Il. coronary arteriogram predict the physiological importance ofa coronary 116. Gould KL. How accurate is thallium exercise testing? JAm Coil Cardiol stenosis?N EnglJMed 1984;3l0;819—824. l989;l4:1487—l490. 94. Vogel RA, LeFree M, Bates E, et al. Application ofdigital techniques to 117. Gould KL. Agreement on the accuracy of thallium stress testing. J Am selective coronary arteriography: use of myocardial contrast appearance Coil Cardiol 1990;16:1022—23. time to measure coronary flow reserve. Am Heart J 1984;107:153-l64. 118. Diamond GA. Suspect specificity. JAm CoilCardiol l990;l6:l017-2l. 95. NissenSE,ElionJL,BoothDC, EvansJ, DeMariaAN. Valueand 119. Williams BR, Jansen DE, Wong LF, Fiedotin AF, KnopfWD, Toporoff limitations ofcomputer analysis ofdigital subtraction angiography in the Si. Positronemissiontomographyforthediagnosisof coronaryartery assessment ofcoronary flow reserve. Circulation l986;73:562—57l. disease: A non-university experience and correlation with coronary an 96. Zijlstra F, Fioretti P, Reiber JHC, Serruys PW. Which cineangiographi giography tAbstracti. J NuclMed 1989;30:845. cally assessed anatomic variable correlates best with functional measure 120. Tamaki N, Yonekura Y, Senda M, et aL Value and limitation of stress ments of stenosis severity?:a comparison of quantitative analysis of the thallium-20l single photon emission computed tomography: Comparison coronary cineangiogram with measured coronary flow reserve and cxci with N-l3 ammonia positron tomography. J NucI Med l98&,29:l 191- cisc/redistribution thallium-201 scintigraphy. I Am Coil Cardiol 1988; 1188. 12:686—691. 121. Marwick TH, Go RT, Maclntyre Wi, et al. Disparities between myocar 97. Blankenhorn DH, Nessim SA, Johnson RL, Sanmarco ME, Awn 5P, dial perfusion imaging using Rb-82 PET and Tl-20l SPECT after dipyr. Cashin-Hemphill L Beneficial effects ofcombincd colestipolniacin ther idamole stress lAbstract]. J NucI Med l989;20:861. apy on coronaryathemsclerosisand coronary venous bypassgrafls. JAMA 122. Go RT, Marwick TH, Maclntyre Wi, et al. Initial results of comparative l987;257:3233—3240. rubidium-82 and thallium-20l myocardial perfusion imaging in diagnosis 98. Brown GB, Alberts ii, Fisher LD, et al. Niacin or Lovastatin, combined ofCAD [AbstractJ. JNuclMed l989;30'.759. with Colestipol regresses coronary atherosclerosis and prevents clinical 123. Kalus ME, Stewart RE, Gacioch GM, et al. Comparison of Rb-82 PET events in men with elevated apolipoprotein B. N Engi I Med l990;323: and Tl-20l SPECT for the detection ofregional coronary artery disease 1289—98. (Abstract]. JNuclMed l989;30.829. 99. Turner DA. An intuitive approach to receiver operating characteristic 124. Friedmani, Van Train K, Maddahii,etaL “UpwardCreep―oftheheart: curve analysis. JNuclMed 1978;19:2l3—220. a frequent source of false-positive reversible defects during thallium.20l 100. Turner DA, Battle WE, Deshmukh H, et aL The predictive value of stress-redistribution SPECF. JNuclMed 1989;30:l718—l722. myocardial perfusion scintigraphy afterstress in patientswithout previous 125. Gould KL, Gordon DO. In: Zipes DP, Rowlands Di, ads. PET and myocardial infarction. JNuclMed 1978;19:249—255. dinical cardiology in progress in cardiology. Philadelphia Lea and Febi 101. Massof R, Emmel T. Criterion-free parameter-free, distiibution-inde ger, 1990. pendent index of diagnostic test performance. Applied Optics 198726: 126. Lyle FM, Gordon DO. The importance of timing with Rb-82 PET 1395—1408. imaging ofthe heart (Abstract]. JNuclMed l989;30:87l. 102. Hlatky MA, Mark DB, Hand FE, Calif RM, Piyor DB. Rethinking 127. Senda M, Yonekura Y, Tamaki N, et a). Axial resolution and the value sensitivity and specificity. Am J Cardiol 1987;59:l 195—1198. ofinterpolating scan in multislice positron computed tomography. IEEE 103. Ritchie JL Trobaugh GB, Hamilton GW, et al. Myocardial imaging with Trans on Medicallmaging 1985; MI-4;44—5l. thalhium-20l at rest and during exercise. Comparison with coronary 128. Mullani NA, Gould KL, Hartz RK, et al. Design and performance of

PETPerfusionImagingandNuclearCardiology•Gould 605 Posicam 6.5 BOO Positron Camera. J Nuc/Med l990;31:6l0—6l6. 142. Hill JL, Gettes LS. Effect of acute coronary artery occlusion on local 129. Raylman RR, Hutchins GD, Schwaiger M, Paradise AH. The effect of myocardial extracellular K+ activity in swine. Circulation l980;6l:768— axial sampling and motion on three-dimensional quantification of myo 777. cardial defects with positron emission tomography. J Nuc/Med l989;30: 143. Conrad GL, Rau EE, Shine KI. Creatine kinase release, potassium-42 892. content and mechanical performance in anoxic rabbit myocardium. J 130. Bendriem B, Dewey SL, Schlyer Di. Dependence on the recovery coeffi C/inInvest1979;64:l55—161. cient on axial sampling in multislice positron emission tomography. J 144. Gould KL, Haynie M, Hess Mi, Yoshida K, Mullani NA, Smalhing RW. Nuc/Med 1989;30:892. Myocardial metabolism of fluorodeoxyglucose compared to cell mem 131. Gould KL, Goldstein PA, Mullani NA. Economic analysis of clinical brane integrity for the potassium analog Rb-82 for assessing viability and positron emission tomography ofthe heartwith rubidium-82. JNucl Med infarct size in man by PET. I NucI Med 1990:31:1—9. 1989:30:707—717. 145. Sease D, Garza D, Merhige ME, et al. Does myocardial uptake of F-18- 132. Gould KL. Goals, gold standards and accuracy of non-invasive myocar Fluoro-deoxy-glucose by positron emission tomography reliably indicate dial perfusion imaging for identifying and assessing severity of coronary myocardial viability in acute myocardial infarction? [Abstract]. Circula artery disease. Current Opinion in Cardiology l989;4:834—844. tion l989;80:II—378. 133. Gould KL, Mullani NA, Williams B. PET, PTCA and economic priorities. 146. Pierard LA, DeLandsheere CM, Berthe C, Rigo P. Kulbertus HE. Iden C/in Cardiol 1990:13:153—164. tification ofviable myocardium by during dobutamine 134. Bodenheimer MM, Banka VS. Fooshee C, Hermann GA, Helfant RH. infusion in patientswith myocardial infarction afterthrombolytic therapy: Relationship between regional myocardial perfusion and the presence, comparison with positron emission tomography. J Am Coil Cardiol severity and reversibility of asynergy in patients with coronary heart l990;l5: 1021—31. disease. Circulation l978;58:789—878. 147. Bianco JA, Bakanauskas J, Carbon M, ci al. Augmented uptake of 2-C- 135. Rozanski A, Berman DS, Gray R, et al. Use ofthallium-20l redistribution l4-D.deoxyglucose in reversibly-injured myocardium. Eur J NucI Med scintigraphy in the preoperative differentiation of reversible and nonre l988;l3:557—562. versible myocardial asynergy. Circulation 198l;64:936—944. 148. Komatsumoto 5, Greenberg JH, Hickey WF, Reivich M. Local cerebral 136. Iskandrian AS, Hakki A-H, Kane SA, God IP, Mudth ED, Segal BL. glucose utilization in chronic middle cerebral artery occlusion in the cat. Rest and redistribution thallium-20l myocardial scintigraphy to predict J Cereb Blood Flow Metab 1989;9:535—547. improvement in left ventricular function after coronary arterial bypass 149. Wijns W, Jacque AM, Leners N, et al. Accumulation of polymorphonu grafting.AmJCardio/1983;Sl:l3l2—l3l6. clearleukocytes in reperfused ischemic canine myocardium: relation with 137. Cloninger KG, DePuey EG, Garcia EV, ci al. Incomplete redistribution tissue viability assessed by fluorine.l8-2-Deoxyglucose uptake. J Nucl in delayed thallium-20l single proton emission computed tomographic Med l988;29:1826—1832. (SPECT) images: an overestimation of myocardial scarring. J Am Coil 150. Mody F, Buxton D, Krivokapich J, Hansen H, Scm C, Schelbert H. Cardiol 1988;l2:955—963. Attenuated response ofglucose metabolism in reperfused canine myocar 138. Galli M, Bencivelli W, Pardo NF, Tavazzi L Underestimation of residual dium to changes in substrate levels. JAm Coil Cardiol l990;15:80A. ischemia by thallium-201 scintigraphy after myocardial infarction. Chest 151. Merhige ME, Ekes RD. Mossberg K, Taegtmeyer HT, Gould KL. Cate 1988;94:876—878. chol stimulation, substrate competition and myocardial glucose uptake 139. Tamaki N, Yonekura Y, Yamashita K, et al. Relation ofleft ventricular in conscious dogs assessed with positron emission tomography. Circ Res perfusion and wall motion with metabolic activity in persistent defects on l987;6l(supp II):l24—l20. @°‘T1tomography in healed myocardial infarction. Am J Cardiol 152. Bonow RO, Bacharach SL, Cuocolo A, Dilsizian V. Myocardial viability 1988;62:202—208. in coronary artery disease and left ventricular dysfunction: thallium-20l 140. Brunken RC, Kottou 5, NienaberCA, ci al. PETdetection ofviable tissue reinjection vs fluorodeoxyglucose [Abstract]. Circulation l989;80: (suppl) in myocardial segments with persistent defects at Tl.20l SPECT. Ra 11—377. diology l989;l72:65—73. 153. Bonow RO, Dilsizian V, Cuocolo A, Bacharach SL. Myocardial viability 141. Tamaki N, Yonekura Y, Yamashita K, et al. Positron emission tomog in patients with chronic coronary artery disease and left ventricular raphy using fluorine-18 deoxyglucose in evaluation of coronary artery dysfunction: thallium-20l reinjection versus “F-fluorodeoxyglucose.Cir bypass grafting. Am J Cardiol 1989;64:860—865. culation (In Press).

EDITORIAL The ClinicalRoleof PositronEmissionTomographyfor Cardiology inthe 1990s and Beyond

A lthough positron emission to required equipment ($5-7 million joint venturesbetweenclinicaland/or mography (PET) has been per with camera, cyclotron and support research centers with radiopharma formedin patientsfor more than 15 ing equipment), absence ofU.S. Food ceutical groups that share a cyclotron. years, it has only recently begun to and Drug Administration (FDA) ap By sharing or leasing the cyclotron, emerge as a diagnostic modality for proval, the lack of widespread reim the capital equipment and operating use by clinicians. Implementation of bursementfrom federaland private costs should be reduced while making clinicalPET hasbeendelayedby sev insurers,and the paucity oflarge din PET tracers available to sites with eral factors, including the high cost of ical trials (including outcome data) cameras but without cyclotrons. The from multiple sites. Some solutions to regulatorybarriersarealsostartingto these limitations appear to be near. resolve. In November 1989, the FDA ReceivedJan. 17, 1991. issued a position statement on PET For repnnts contact: Richard A. Goldstein, MD, The entry of major manufacturers Director, Nuclear Cardiology, Cardiology Division into PET imaging should decrease the radiopharmaceuticalsindicatingthat Room 1.246, The Lhiiversity of Texas Medical PETcenterscould continue to operate School at Houston, 6431 Fannin, Houston, TX price of cameras due to increased 77030. competition. Other recent changes are eventhoughNew Drug Applications

606 TheJournalof NuclearMedicine•Vol.32 •No.4 •April1991