608

Clinical Significance of Reduced Regional Myocardial Glucose Uptalkwein Regions With Nor-mall Blood Flow in Patients With ChronicCoronary Artery DiseasePASQUALE PERRONE-FILARDI, MD, PHD,* STEPHEN L. BACHARACH, PHD,

VASKEN DILSIZIAN, MD, FACC, JOSE A. MARIN-NETO, MD, SIMONE MAUREA, MD,

JAMES A. ARRIGHI, MD, ROBERT O. BONOW, MD, FACC

Bethesda, Maryland

Objectives . The objective of this study was to assess the clinicalof reduced rQual fl

10 ('OF) fluoroideoxyglucoseoronaryuptake With 13=111111 flow in

is with el" c

arteryo'"ortatdlwm.

BmAgmwmL In patients with imbentic left ventricular dysfunc-sm, uptuake may be reduced in some0,wardild regions despite normal flow. The significance of this

is unclear and has not been Investigated systematically .Mwa*Wds . Twenty-three patients with coronary artery disease

mW Impaired ventricular function (mean ejection fraction[*I 8D] 23 * 10%) underwent positron emission tomographywith ' OF-fluomleoxyglucose and oxygen-15-labeled water at rest,exercise thafflum-201 tomographic Imaging with rest reinjectionmatl pMA nmpdk rnamoanauKae Imaging to measure end-diastolicwall thickness and systolic wall thickening .

Resub. Of 168 regions with normal flow (z!0 .7 mUg per min),125 (74%) had normal ' IF-flu orodeoxyglucose uptake (98 :t10%), and the remaining 4.1(26%) showed moderately reduced"F-fluorWeOxyglacm uptake (69 * 8%) . Systolic wall thicken-

Metabolic imaging with positron emission tomography using'"F-fluxrodeoxyglucose has been reported to distinguishamong normal, ischemic and fibrotic myocardium in patientswith chronic coronary artery disease and left ventriculardysfunction (1-5). This differentiation is based on the rela-tion between regional blood flow and glucose metabolism . Inparticular, myocardial regions with normal blood flow areconsidered to represent normal myocardium regardless ofthe magnitude of "aF .fluorodeoxyglucose uptake in thoseterritories. In contrast, myocardial regions with reduced

From the Cardiology Branch, National Heart, Lung, and Blood Instituteand the Department of Nuclear Medicine, Clinical Center, National Institutesof Health, Bethesda, Maryland .

Manuscript received May 13, 1992 ; revised manuscript received October4, 1993, accepted October 14, 1993 .

*Present address : Dr. Pasquale Perrone-Filardi, Cattedra di Cardiologia,University di Napoli "Federico II," Via S . Pansini 5,1-80131 Napoli, Italy .

Address for correspondence: Dr. Robert 0 . Bonow, Division of Cardiol-ogy, Northwestern University Medical School, Wesley Pavilion, Suite 524,250 East Superior Street, Chicago, Illinois 60611 .

01994 by the American College of Cardiology

JACC Vol. 23, No . 3March 1 . 1994:608-16

MYOCARWAL ISCHEMIA

lug was absent at rest in 14% of regions with normal 18F-fluorodeoxyglucose uptake compare with 32% of regions withreduced 'IF-fluorodeoxyglucose uptake (p < 0 .01). Reversiblethallium abnormalities were observed in 45 (36%) of 125 regionswith normal 181E-fluorodeoxyglucose uptake compared with 27(63%) of 43 regions with reduced 'O F-fluorodeoxyglucose uptake(p < 0.01). This difference was accounted for by a higherproportion of partially reversible defects in regions with reduced'IF-fluorodeoxyglucose uptake compared with regions with nor-mal ' OF-fluorodeoxyglucose uptake (42% vs. 18%, respectively,p < 0.01).

Conclusions . Thus, regions with moderately reduced 'OF-fluorodeoxyglucose uptake with normal flow occur commonly inpatients with Ischemic left ventricular dysfunction . The majorityof these regions show impaired systolic function at rest andexercise-induced thallium abnormalities that are only partiallyreversible. These observations suggest that such regions rep nten Admixture of fibrotic and reversibly ischemic myocardium .

(J Am Coll Cardiol 1994,23 .608-16)

blood flow are considered to represent either ischemicmyocardium, when ' 8F-fluorodeoxyglucose uptake is in-creased relative to flow ( 18 F-fluorodeoxyglucese-blood flowmismatch), or nonviable myocardium, when both blood flowand ' 8F-fluorodeoxyglucose uptake are proportionately re-duced (1-5) . However, in patients with coronary arterydisease, myocardial territories with normal blood flow mayoccasionally manifest a reduction in ' OF-fluorodeoxyglucoseuptake, which is apparent even in the glucose-loadedstate when the physiologic inhomogeneity in glucose uptakeis minimal (6) . Although these territories are convention-ally assumed to represent normal myocardium, becauseblood flow is normal, the causes of the reduced 18F-fluorodeoxyglucose uptake are unknown and its functionalcharacteristics have not been investigated . Therefore, theaim of this study was to assess the functional characteristicsand the response to exercise of myocardial regions withnormal blood flow and reduced 18F-fluorodeoxyglucose up-take and to compare these characteristics with those ofmyocardial regions with normal blood flow and normal

0735-1097/94/57.00

JACC Vol. 23, No . 3March 1, 1994 .608-16

MYOCARDIAL BLOOD FLOW AND REDUCED GLUCOSE UPTAKE

' 8F-fluorodeoxyglucose uptake ii, nctients with chronic cor-onary artery disease and left ventricular dysfunction .

MethodsPatient selection. We studied 23 patients with angiograph-

ically proved coronary artery disease and left ventriculardysfunction . There were 22 men and 1 woman ranging in agefrom 45 to 78 years (mean 59 ± 9) . Mean left ventricularejection fraction by radionuclide angiography was 28 ± 10%at rest. Although all patients had evidence of previousmyocardial infarction, we studied only patients with chronicstable coronary artery disease . No patient had an acutemyocardial infarction or unstable angina within 6 months ofthe study, and no patient had diabetes mellitus . The glucoseplasma level before the ' 8F-fluorodeoxyglucose study (ob-tained in 20 of the 23 patients) was in the normal fastingrange (96 ± 19 mg/WY Coronary arteriography demonstratedsignificant stenosis (~5O% reduction in lumen diameter) ofall three major epicardial coronary arteries in 14 patients andof two coronary arteries in 6 patients . All patients underwentradionuclide angiography, positron emission tomography,thallium-201 single-photon emission computed tomographyand gated magnetic resonance imaging . in 19 patients allcardiac medications were discontinued for >48 h beforeradionuclide angiography, thallium-201 positron emissiontomography and magnetic resonance imaging were per-formed. An average of 27 ± 27 days elapsed betweenperformance of the first and the last of these tests . Positronemission tomography and thallium data in relation to re-gional systolic function have been reported previously inthese patients (7) . Written informed consent for the protocolwas obtained in each case. The protocol was approved bythe National Heart, hung, and Blood Institute ClinicalResearch Subpanel on March 1989 (protocol 86-H-209) . Theestimated radiation exposure did not exceed 1 .3811 totalbody dose .

Study protocol . Gated blood pool cardiac scintigraphy .Gated equilibrium radionuclide angiography was performedat rest with the patients in the supine position using redblood cells labeled in vivo with 20 to 25 mCi of technetium-99m. High temporal resolution (20-ms/frame) time-activitycurves were generated, from which the left ventricularejection fraction and time to end-systole were computed, aspreviously described (8) .

Thallium imaging, All patients underwent exercise thal-lium single-photon emission computed tomography, as pre-viously described (9) . In brief, after an overnight fast,patients underwent treadmill exercise according to a stan-dardized multistage exercise protocol, with continuous mon-itoring of heart rate and rhythm, blood pressure and symp-toms. At peak exercise, 2 mCi of thallium-201 was injectedintravenously, and the patients continued exercise for anadditional 45 to 60 s . Approximately 10 min after terminationof exercise, thallium images were obtained using a wide fieldof view rotating gamma camera equipped with a high sensi-

PERRCNE-FILARDI ET AL

609

tivity, medium resolution, parallel hole collimator centeredon the 68-keV photo peak with a 20% window . The camerawas rotated in a 180° arc in an elliptic orbit about thepatient's shoulders across the thorax from a right anterioroblique angle of 40° to a left posterior oblique angle of 40° to6° increments for 30 s each (9) .

Redistribution images were obtained 3 to 4 h after stresswhile the patients were resting . Immediately thereafter allpatients received an additional I mCi of thallium, andtomographic images were acquired 10 to 15 min later . Theexercise, redistribution and reinjection data were recon-structed as a series of whole body transaxial tomograms fordirect comparison with the corresponding positron emissionand magnetic resonance tomograms, as described later .Single-photon emission computed tomography in plane andz axis resolution was approximately 12 mm . Data werereconstructed with 6 .88 mm per pixel sampling and with a6.88-mm separation between slices .

Positron emission tomography . Positron emission to-mography was performed, as previously described (7,10), toassess regional myocardial perfusion with oxygen-15-labeled water (H 2 15 0) and exogenous glucose utilizationwith 18F-fluorodeoxyglucose using a whole body positronemission tomography camera producing 21 contiguous tomo-grams spaced 5 .1 mm apart with a slice thickness of 13 mmand an in-plane reconstructed resolution of 6 .5 mm. Imageswere obtained perpendicular to the long axis of the body tocreate a series of transaxial tomograms . After an overnightfast, all patients were pretreated with 50 g of oral glucose I hbefore the ' 8F-fluorodeoxyglucose study . An attenuationscan was performed, and then two separate bolus injectionsof 12 to 15 mCi of H 2 150 were administered intravenously12 min apart, followed by the administration of 5 mCi of"F-f,uorodeoxyglucose 15 min later . Data were acquiredfor 5 min in list mode after each water injection and for 60to 75 min in list mode after the ' 8F-fluorodeoxyglucoseinjection. The data beginning at 30 min after ' 8F-fluorodeoxyglucose injection, corresponding to the final 30to 45 min of data acquisition, were reconstructed to cre-ate tomographic images of regional myocardial ' 8F-fluorodeoxyglucose uptake . To prevent chest movement,two seat belts were fastened around the patient's thoraxduring the acquisition of the study .

Magnetic resonance imaging . Electrocardiographicgated magnetic resonance imaging (MRI) was performedusing a 1 .5 tesla scanner . A 15- to 20-min scan allowedacquisition of four to five slices at four to five time points inthe cardiac cycle from end-diastole to end-systole using spinecho imaging (echo time = 20 ms ; repetition time = R waveto R wave time, two excitations) . Each slice was 10 mmthick, with a center to center slice distance of 20 mm .Immediately after this acquisition, a second acquisition wasbegun, again consisting of four to five slices at four to fivetime points to fill in the gaps between slices . The final imagesequence therefore consisted of 8 or 10 contiguous slices(10 mm thick, 10-mm center to center interstice distance) at

610

PERRONE-FILARDI ET AL .MYOCARDIAL BLOOD FLOW AND REDUCED GLUCOSE UPTAKE

March 1, 1994:608-16

four to five time points in the cardiac cycle from end-diastoleto end-systole. Total

time was 30 to 45 min .The time to end-systole was determined before MR11

studies from the left ventricular volume curve obtained fromthe radionuclide ventriculogram that was acquired at asimilar heart rate. To obtain the end-systolic image, theintersequence delay of the MRI study (i .e ., the time intervalbetween successive time points) was adjusted so that the last(fourth or fifth) time point of the MRI study occurred atend-systole, as determined from the radionuclide study .

Nuta aunlysis . In each patient, corresponding transaxialtomograms from the three sets of thallium images represent-ing the exercise, redistribution and rcir&ction studies werevisually aligned for direct comparison (7,10) . These in turnwere aligned with the corresponding transaxial tomographicimages of myocardial "F-fluorodeoxyglucose uptake fromthe positron emission tomography study . To avoid errorsstemming from the reslicing process of three different imag-ing approaches, only transaxial images were analyzed .Therefore, considering the difficulty of assessing the inferiormyocardial wall using the transaxial plane, only correspond-ing midventricular positron emission, thallium and mag-netic resonance myocardial tomograms were used for theanalysis. To objectively compare relative regional ' 8F-fluorodeoxyglucose and thallium activity, five myocardialregions of interest representing the posterolateral, anterolat-eral, anteroapical, anteroseptal and posteroseptal myocar-dium were drawn on each 18F-fluorodeoxyglucose tomogramand on each of the three corresponding thallium images .Regional "F-fluorodeoxyglucose and thallium activities(nCi/ml) were then computed within each region of interest(7,10) .

Wall thickness measurements. To assess regional end-diastolic wall thickness and absolute wall thickening, corre-sponding transaxial end-diastolic and end-systolic magneticresonance images were analyzed by two independent oper-ators unaware of the thallium and '"F-fluorodeoxyglucoseresults. Alignment of the positron emission tomograms withthe corresponding magnetic resonance tomograms was donevisually using both the positron attenuation images (thatshow the lung outlines, the heart shadow and the slice inwhich the liver first begins to appear) and the positronemission images. Because of the different interslice distance(5.1 mm for positron emission tomography and 10 mm forMRI studies), each magnetic resonance tomogram wasmatched with two contiguous positron emission tomograms .An average of three magnetic resonance slices and 6 18F-fluorodeoxyglucose and thallium tomographic planes wereanalyzed for each patient . As described earlier, eachpositron emission tomography and thallium slice had beendivided into five regions of interest . By appropriate scalingand resampling, the positron emission tomography and mag-netic resonance images were made to be of identical size(i .e ., the same number of millimeters per pixel) . The fiveregions of interest drawn on the positron emission tomo-grams could then be superimposed visually on the magnetic

JACC Vol. 23, No. 3

resonance tomograms, with appropriate rotations and trans-lations also determined visually . Because each region ofinterest encompassed a relatively large amount of myocar-dial tissue, minor misalignments using this visual techniquewould not significantly alter the results . Thickness measure-ments were made from the MRI data at the center of eachregion of interest by manually identifying a point on theepicardial and endocardial borders perpendicular to the twosurfaces. The length of the line segment joining these twopoints was calculated and considered as the wall thickness .Systolic wall thickening was defined as the difference be-tween end-systolic and end-diastolic wall thicknesses . Ac-cording to the extent of systolic wall thickening (11), myo-cardial regions were considered normokinetic (systolic wallthickening ~2 mm) . hypokinetic (systolic wall thickening<2 mm) or akinetic-dyskinetic (absent wall thickening orsystolic wall thinning) .

Regional myocardial thallium activity . In each patientthe myocardial region of interest with the maximal counts onthe exercise thallium study was used as the normal referenceregion for that patient (7,10). The corresponding anatomicregions in the redistribution and reinjection studies wereidentified and used as the reference regions for those studies .The thallium activity in all other myocardial regions wasthen expressed as a percent of the activity measured in thereference region for each of the exercise, redistribution andreinjection image series .

For each exercise study, thallium activity in any myocar-dial region that measured <85% of the normal referenceregion was considered reduced and defined ass a thalliumperfusion defect (7,10) . On the basis of previous reproduc-ibility studies from our laboratory (12), a regional thalliumabnormality during exercise was considered reversible if therelative thallium activity increased by ? 10% on the subse-quent redistribution or reinjection images . A defect wasconsidered completely reversible if this increase resulted inthallium activity that was >85% of the activity in thereference region . Similarly, thallium abnormalities duringexercise were considered irreversible if the relative thalliumactivity was unchanged or increased < 1001 on the subse-quent redistribution and reinjection images. In addition, adefect on exercise images was also considered irreversible ifthe activity in that region remained <50% of the activity inthe normal reference region after redistribution and reinjec-tion . On the basis of the relative regional thallium activity onthe exercise images, thallium defects that were irreversibleon redistribution and reinjection images were classified asmild to moderate (51% to 85% of peak activity) or severe( :55en of peak), as previously described (7,10,12) .

Regional myocardial blood flow. We computed absoluteregional myocardial blood flow from the dynamic H2'50 data(13). The myocardial regions of interest previously con-structed on the 18F-fluorodeoxyglucose images for measure-ment of regional 18F-fluorodeoxyglucose activity were ap-plied to the tomographic H 2150 data to derive regionalmyocardial H2'50 time-activity curves . Absolute regional

JACC Vol . 23 . No . 3March 1 . 1994 :608-16

MYOCARDIAL BLOOD FLOW AND REDUCED GLUCOSE UPTAKE

where the parameters of the fit are PV (a partial volumecorrection factor), F (flow in mI/g per min) and SO (aspillover correction factor, indicating the fraction of countsspilling from the left ventricular cavity into the myocardi-um); x is the co9volutior, operation, and t is time . Thepartition coefficient p was assumed fixed at 0 .92, and thearterial input function, LV(t) was measured directly fromthe left ventricular blood pool using the positron emissiontomographic data . Contamination from the myocardium wascarefully avoided by inspecting both the end-diastolic andend-systolic corresponding magnetic resonance tomograms .This approach is similar to the method of lida et al . (14) .Bergamlin et al . (15) and Herrero et al . (16) except that ourfitting was weighted with the data's inverse variance, asdetermined from total counts . deadtime and random correc-tion factors. The computation of blood flow values was

performed twice, one for each 1-1 2 150 study, and the resultsof the two measurements were then averaged together . Theaverage SD of the blood flow measurements was 0.220.18 ml/g per min .

The equation used to calculate blood flow also yields therecovery coefficient for correction for partial volume ef-fects. The recovery coefficient obtained form the waterdata is also valid for the 18 F-fluorodeoxyglucose data ex-tracted from the same regions of interest . The ' 8 F-fluorodeoxyglucose data were analyzed both with and with-out the correction for partial volume effects. Because asimilar correction could not be made for the thallium data .the 18F-fluorodeoxyglucose and thallium comparisons weremade with neither data set corrected for recovery coefficientin order not to bias the results .

Regional myocardial 18F-fluorodeoxyglucose uptake .The myocardial region on the 18 F-fluorodeoxyglucose seriesthat corresponded to the normal reference region on thethallium exercise image series was used as the normalreference region for relative ' 8 F-fluorodeoxyglucose uptake .The 18F-fluorodeoxyglucose uptake in all other myocardialregions was expressed as a percent of the activity in thisreference region (7,10) . The 23 regions defined as normalreference regions for the 23 patients all showed normal bloodflow (1.0 ± 0.2 ml/g per min) and normal systolic wallthickening (3 .7 ± 2.7 mm) .

Regional "F-fluorodeoxyglucose uptake relative to bloodflow. Myocardial regions with normal blood flow were de-fined as those in which blood flow was >O .7 ml/g per min(17) . In light of the variability reported by Gropler et al . (6)in normal persons, "F-fluorodeoxyglucose uptake in eachpatient was considered reduced when it measured <80% ofthat of the normal reference region for that patient . Myocar-dial regions with normal blood flow were then divided into agroup with normal 18F-fluorodeoxyglucose uptake and a

8

0 6

am

0 11

0 11Normal V FLIG

1 Y

PERRONE-FiLARDI ET AL

611

, I-

Normal 4 FDG

Normal 4 FDGF D G

FOG

FDG

Figure 1 . Blood flow, "F-fluorodeoxyglucose (FDG) activity andend-diastolic wall thickness in myocardial regions with normalregional blood flow and either normal or reduced "F-fluorodeoxyglucose uptake .

group with a moderate reduction in "F-fluorodeoxy&coseuptake (50% to 79% of that of the normal region) .

Statistical analysis. The two-tailed unpaired Student t testwas applied for comparisons between the two groups . Thechi-square test was applied to compare percent data . Thereported correlations were analyzed by linear regressionanalysis . All data are expressed as mean value ± SD . Ap value < 0.05 was accepted as the minimal level ofsignificance .

ResultsOne hundred sixty-eight myocardial regions showed nor-

mal blood flow at rest (~!0 .7 ml/g per min) . Of these, 125regions (74%) had normal "'F-fluol -odeoxyglucose uptake(98 ± 10%), and the remaining 43 regions (26%) showedreduced 18F-fluorodeoxyglucose uptake (69 ± 8%) . Amongthe 43 regions with reduced "F-fluorodeoxyglucosc uptake,19 (44%) were localized in the lateral wall, 12 (28%) in theseptum and 12 (28%) in the apex. Myocardial regions withreduced "'F-fluorodeoxyglucose uptake and normal bloodflow were observed in 15 (65%) of the 23 patients . Ten ofthese 15 patients had three-vessel coronary artery disease, 4had two-vessel and I had one-vessel coronary artery dis-ease . Of the 43 regions with normal blood flow and reduced18 F-fluorodeoxyglucose uptake, 41 (95%) were localized inthe territory of a significantly stenosed vessel . Myocardialblood flow was similar between regions with normal orreduced 18 F-fluorodeoxyglucose uptake (Fig . 1) . In all myo-cardial regions with normal blood flow there was no cor-relation between blood flow and the amount of ' 8 F-fluorodeoxyglucose activity (r = 0 . 19, p = NS) .

Regional anatomy and systolic function . Fluorine-18-fluorodeoxyglucose activity did not correlate with end-diastolic wall thickness (r = -0.04) . End-diastolic wallthickness did not differ between myocardial regions withnormal or reduced 18F-fluorodeoxyglucose uptake (Fig . 0 .

myocardial blood flow was calculated by fitting the myocar-dial H,`O time-activity curve, M(0, to the formula :

M(t) = RVHFHLVU) X e"M + 00)(A00,

1 -1

I ()

612 PERRONE-FILARDI ET AL .MYOCARDIAL BLOOD FLOW AND REDUCED GLUCOSE UPTAKE

Normolvnetic Hypokinelic Akinelic/02ntyyet, c

Normal FDG

P <0.01I

Normokinetic Hypokinelic Akinetic/Olsfrigic

I FOG

When systolic wall thickening was evaluated in individualregions, akinesia or dyskinesia was observed in 14 (32%) of43 regions with reduced '"F-fluorodeoxyglucose comparedwith only 18 (l4%) of 125 regions with normal '"F-fluorodeoxyglucose uptake (p < 0.01, Fig . 2) . No correlationbetween '"F-fluorodeoxyglucose activity and systolic wallthickening was observed in all myocardial regions withnormal blood flow (r = 0.10, p = NS) or in the group ofregions with reduced '8F-fluorodeoxyglucose uptake (r =0.21, p = NS). In both groups of myocardial regions withnormal or reduced "F-fluorodeoxyglucose uptake, no differ-ence in '"F-fluorodeoxyglucose activity was found betweenregions with normal or impaired wall thickening (95 ± 7% vs .95 ± 7% in regions with normal '"F-fluorodeoxyglucoseuptake and 68 ± 8% vs . 68 ± 8% in regions with reduced'"F-fluorodeoxyglucose uptake .

When the partial volume correction factor derived fromthe En"O data was applied to the '"F-fluorodeoxyglucosemeasurements, '"F-fluorodeoxyglucose activity was con-firmed to be reduced in 29 (67%) of the 43 regions withreduced uncorrected '"F-fluorodeoxyglucose activity, and11(38%) of these 29 regions showed absent wall thickening .

RegWW thallium activity. Fifty-one (41%) of 125 myo-cardial regions with normal '"F-fluorodeoxyglucose uptakehad normal thallium activity during exercise (Fig . 3) com-pared with only 6 (14%) of 43 regions with reduced '"F-fluorodeoxyglucose uptake (p < 0 .01). Reversible thalliumdefects were observed in 45 (361%) of 125 regions with normal'"F-fluorodeoxyglucose uptake and in 27 (63%) of 43 regionswith reduced '"F-fluorodeoxyglucose uptake (p < 0.01).Systolic wall thickening was impaired in 17 (63%) of these27 regions. An example of the finding of reduced '"F-fluorodeoxyglucose in a region with normal blood flow butreduced wall thickening and a reversible thallium defect isshown in Figure 4 . The difference in the frequency ofreversible thallium defects was accounted for by a higher

Figure 2. Systolic wall thickening at rest in myocardialregions with normal regional blood flow and either nor-mal or reduced ' OF-fluorodeoxyglucose uptake. Thepercent of regions with akinesia or dyskinesia wassignificantly higher in the group with reduced "F-fluorodeoxyglucose uptake. Abbreviations as in Figure 1 .

Figure 3 . Exercise thallium-201 findings in regions with normal regionalblood flow Lt rest and either normal or reduced "F-fluomdeoxyglucoseuptake. Myocardial regions with reduced "F-fluorodeoxyglucose uptakehad a lower likelihood of normal thallium uptake on single-photon positronemission computed tomography (SPEC" and a greater likelihood ofpartially reversible (Rev) thallium deflects, as indicated by the asterisks(p < 0 .01) . Other abbreviations as in Figure 1 .

100

88

20

NormalFOG

ReversibleDetects

Thallium-201 SPECT

JACC Vol. 23, No. 3March 1 . 1994:608-t6

IrreversibleDefects

JACC Vol . 23 . No. 3March 1 . 1994 :608-16

Figure 4. Example of matched positron emission(PET), magnetic resonance (MRI) and thallium-201positron emission computed (SPECT) tornogramsfrom a patient with a dilated left ventricle . The flowimage (H 2 `0 [oxygen-labeled water]) at top left isa functional image of the blood flow measurementsthat were calculated in the five myocardial regionsof interest . The functional flow image shows ananterolateral region with normal blood flow (arrow)that manifests reduced "F-fluorodeoxyglucose(FDG) uptake . The sarnie territory shows severelyreduced systolic wall thickening in the MRI lomo-grams and corresponds to a reversible thalliumdefect on the SPECT tomograms . Redist = redis-tribution .

number of partially reversible thallium defects in regions

with reduced '8F-fluorodeoxyglucose uptake compared withregions with normal "F-fluorodeoxyglucose uptake (42% vs .18%, respectively, p < 001 : Fig. 3), whereas the number oftotally reversible thallium defects was similar between thetwo groups . Twenty-nine regions (23%) with normal "4F-fluorodeoxyglucose uptake showed mild to moderate irre-

versible thallium defects, with no region manifesting asevere irreversible thallium defect (Fig . 3). Similarly, 10

regions (23%) with reduced "F-fluorodeoxyglucose uptakeshowed irreversible thallium defects ; the irreversible thal-lium defect was mild to moderate in 9 regions and severe in

only I region (2%) (Fig . 3). Finally, the level of thalliumactivity, expressed as the percent of maximal activity mea-

sured in either the redistribution or the reinjection study,was significantly higher in regions with normal than inregions with reduced ' 8F-fluorodeoxyglucose uptake (86

13% vs . 73 ± 15%, respectively, p < 0 .01) .

DiscussionThe results of the present study indicate that myocardial

regions with normal blood flow and moderately reduced' 8F-fluorodeoxyglucose uptake occur commonly at rest in

patients with coronary artery disease and left ventriculardysfunction . This finding was observed in 65% of patients inour series. These regions correspond to myocardial territo-

ries with a greater impairment of systolic function at restthan that of regions with normal blood flow and ' 8F-

fluorodeoxyglucose uptake and with a greater extent ofinducible ischemia during exercise .

PERRONE-FILARDI ET AL .MYOCARDIAL BLOOD FLOW AND REDUCED GLUCOSE UPTAKE

PET

Thallium-201 SPECT

Stress

Redict

Regional systolic function . Although the percent of myo-cardial regions showing normal systolic wall thickening atrest was similar between regions with normal ' xF-fluorodeoxyglucose uptake and those with reduced lt F-fluorodeoxyglucose uptake, the percent of regions showingakinesia or dyskinesia at rest was greater among the regionswith reduced 18F-fluorodeoxyglucose uptake . Thus, the re-duction in '"F-Ruorodeoxyglucose uptake is frequently as-sociated with an impairment of regional systolic function

despite the presence of normal blood flow (Fig . 2) . It isnoteworthy that regional systolic function was also impairedin 45% of the regions with normal blood flow and normal

' 8F-fluorodeoxyglucose uptake, which were either hypoki-netic or akinetic, indicating that the presence of normalblood flow at rest may be inaccurate in predicting normal

systolic function in a given myocardial territory . This findingis not surprising considering the transmural nature of blood

flow measurements obtained by current positron emissiontomographic techniques, and it is in agreement with experi-mental studies showing a good correlation between suben-docardial, but not transmural, flow and regional systolic

function (18) . Thus, the presence of very different regionalsystolic function among territories with normal blood flow

may be explained by the possibility that subendocardial orsubepicardial ischemia or fibrosis might still exist despitenormal transmural flow, considering the relative contribu-tions of the subendocardium and the subepicardium to

regional wall thickening (19) .Regional thallium uptake . The percent of regions with

normal ' 8F-fluorodeoxyglucose uptake showing normal thal-

lium uptake during exercise (Fig . 3) was significantly greater

MRI

613

614 PERRONE-FiLARDI ET AL .MYOCARDIAL BLOOD FLOW AND REDUCED GLUCOSE UPTAKE

than that of regions with reduced ' 8F-fluorodeoxyglucoseuptake (42% vs . 14%). Although totally reversible thalliumdefects occurred with similar frequency between regionswith normal and those with reduced t8F-fluorodeoxyglucoseuptake, the percent with only partially reversible defectswas significantly greater in regions with reduced 18F-fluorodeoxyglucose uptake . The frequency of irreversiblethallium defects was similar between the two groups of18F-fluorodeoxyglucose regions, and all but one of the irre-versible defects were only mild to moderate in severity .Previous data indicate that such mild to moderate irrevers-ible thallium defects correspond to hypoperfused but viable,rather than nonviable, myocardium (7,10,12,20) .

Mechanism of reduced '$F-fluorodeoxygl

uptake inwith normal b flow. Although the cause or

causes of reduced regional f 8F-fluorodeoxyglucose uptakedespite normal blood flow cannot be elucidated from thepresent study, some hypotheses can be made. First, thehigher frequency of partially reversible thallium defects inregions with reduced versus normal 18 F-fluorodeoxyglucoseuptake, su ests the possibility that the former regionsrepresent a mixture of fibrotic myocardium associated withjeopardized myocardium capable of developing ischemiaduring exercise . The normal transmural blood flow in theseterritories is not surprising . Previous observations in dogsundergoing coronary occlusion and reperfusion have dem-onstrated that postischemic myocardium, consisting of anadmixture of necrotic tissue interspersed with normal myo-cardium, shows normal blood flow values (21) . In addition,as recently reported by Yamamoto et al. (22), the H 2 ' sOapproach used to measure blood flow only measures bloodflow per gram of perfused tissue . This means that, in amyocardial region consisting of an admixture of necrotic andnormally perfused myocardium, this approach only mea-sures blood flow to the residual normal myocardium . Thiswould yield a normal blood flow value in that territory . Inaddition, the value measured is independent of the actualamount of perfused myocardium in that territory and istherefore not accurate in predicting the functional outcomeafter revascularization . This may explain why, as reportedby Yamamoto et al . (22), myocardial blood flow in asynergicregions showing improved wall motion after revasculariza-tion did not differ from that measured in irreversibly asyn-ergic regions . Our data confirm the observations of Ya-mamoto et al . indicating that normal regional myocardialblood flow, as measured by the current H, 15 0 approach, isnot always associated with completely normal myocardiumin a given territory. In this regard, the reduced 18F-fluotodeoxyglucose uptake in myocardial territories withimpaired systolic function, but with "normal" blood flowvalues, is consistent with the hypothesis that these territo-ries represent an admixture of fibrotic myocardium withseverely reduced perfusion and viable myocardium withnormal perfusion.

However, the coexistence of viable and necrotic myocar-dium in regions with normal blood flow and reduced 18F-

3ACC Vol. 23, No.March L 1994 :608®1

fluorodeoxyglucose uptake cannot entirely explain oupresent findings because 32% of such myocardial regiondemonstrated normal thallium uptake or completely reversible thallium defects, suggesting the absence of necrotimyocardium in these territories . In fact, other alternativhypotheses can be made to explain the occurrence of normsflow with reduced B8F-fluorodeoxyglucose uptake . First, thpresence of myocardial stunning would also explain thecoexistence of normal blood flow with impaired systolicfunction (23) . Although no patient had a history of recenmyocardial infarction or unstable angina before the stud,and none reported angina during the study, the possibility osilent ischemic episodes before the study cannot be rulesout. The available information regarding the metabolicchanges characterizing stunned myocardium are controversial : Clinical data suggest that myocardial stunning is alsociated with enhanced glucose uptake (24-26), whereas experimental data indicate that glucose consumption isreduced in reperfused stunned myocardium (27) . Thereforemyocardial stunning cannot be ruled out as an explanatioifor the present observation .

Although in the glucose-loaded state, glucose become:the predominant myocardial fuel (28,29), accounting for uIto 1 0 of the myocardial oxygen consumption (29),preferential use of fatty acids in some myocardial territoriemight represent an additional explanation for the reduces' 8F-fluorodeoxyglucose uptake in regions with normal bloodflow. However, the association with impaired regional funstion and the scintigraphic findings, as well as the nonlocalized distribution of these territories throughout the leftventricle, are evidence against this hypothesis . Likewise, atrue depression of metabolism in these territories cannot beruled out from the present data . Future studies to assess theoverall rate of regional oxidative metabolism, perhaps using' IC-acetate, would be helpful to further address this issue .

Because end-diastolic wall thickness was similar betweenregions with normal or reduced ' SF-fluorodeoxyglucoseuptake and no correlation between regional 18F-fluorodeoxy-glucose uptake and end-diastolic wall thickness was observed,a partial volume effect is unlikely to have significantly affectedi8F-fluorodeoxyglucose measurements. Nonetheless, the pres-ence of a significantly more akinetic and dyskinetic regionsin the group with reduced 18F-fluorodeoxyglucose mighthave contributed in part to the reduced recovery of 18F-fluorodeoxyglucose activity in those regions (30) . The find-ing that regional 18F-fluorodeoxyglucose activity did notcorrelate with systolic wall thickening in the current study isevidence against this possibility. Moreover, in both groupsof regions, with normal and with reduced 18F-fluorodeoxy-glucose uptake, no difference in 18F-fluorodeoxyglucoseuptake was found between regions with normal or impairedwall thickening .

The lack of a normal data base to define 18F-fluoro-deoxyglucose uptake might represent a limitation of thestudy. A reduction in 18F-fluorodeoxyglucose uptake in theseptum, compared with the lateral wall, has been reported in

JACC Vol . 23, No . 3 PERRON E-FELARD1 ET AL .Marcia 1 . 1994 :608-16

MYOCARDIAL BLOOD FLOW AND REDUCED GLUCOSE UPTAKE

normal persons under fasting conditions (6) . 111 wever, thisphenomenon is unlikely to explain the present observation .Under glucose-loaded conditions, the inhomogeneity in glu-cose uptake becomes minimal, and the septal reduction in"F-fluorodeoxyglucose uptake does not exceed 13% of thevalues observed in the lateral wall (6,31), which is well abovethe 20% reduction in "F-fluorodeoxyglucose uptake em-ployed as a threshold for defining abnormal myocardialregions in the present study . Moreover, in the present study,myocardial regions with reduced ' 8F-fluorodeoxyglucoseuptake were localized throughout the entire myocardium,and only 28% were localized to the septum .

The definition of normal blood flow in the current study isarbitrary, and this might represent a limitation . In addition,the use of a fixed threshold value to define normal blood flowmay not entirely reflect the physiology of normal perfusion .However, the blood flow cutoff value employed was derivedfrom a previous study (17) reporting a mean blood flow of0.88 ± 0.08 ml/g per min in a group of normal persons usingpositron emission tomography with ' 5O-labeled-carbon di-oxide . which is almost instantaneously converted in vivo to1112"0 . In addition, Bergmann et al . (15), who measuredregional blood flow in a group of normal subjects usingH2150 and dynamic positron emission tomography, reporteda lower limit of normal blood flow value of 0 .67 mllg permin, similar to our threshold value of 0 .70 ml/g per min .Finally, in the present study regions with normal ' 8F-fluorodeoxyglucose uptake showed blood flow similar to thatof regions with reduced "F-fluorodeoxyglucose uptake (Fig .1) . Hence, the potential inaccuracy caused by an arbitrarychosen lower limit for normal blood flow and the possibilitythat a fixed cutoff may not reflect the physiologic variabilityof normal perfusion do not affect the conclusion that thesame blood flow value may be associated with differentfunctional, metabolic and scintigraphic behavior .

Conclusions. In patients with chronic coronary arterydisease, normal regional blood flow at rest is not alwaysassociated with normal systolic function and with normal' IF-fluorodeoxyglucose uptake. Compared with findings inregions with normal blood flow and normal ' 8F-fluorodeoxy-glucose uptake, a moderate reduction in "F-fluorodeoxy-glucose uptake in territories with normal blood flow is associ-ated with a greater frequency of impaired regional systolicfunction at rest and a greater frequency of exercise-inducedthallium defects that are only partially reversible. These obser-vations suggest that the majority of such regions represent anadmixture of fibrotic and reversibly ischemic, jeopardizedmyocardium . Consequently, in the individual patient the as-sessment of myocardial perfusion alone may be inaccurate indefining myocardial viability because the presence of normalblood flow in a given myocardial territory does not necessarilyindicate completely normal myocardium in that region .Whether impaired systolic function associated with normalblood flow and reduced "F-fluorodeoxyglucose uptake repre-sents a reversible phenomenon cannot be determined from ourstudy . Therefore, further studies assessing the effects of revas-

615

cularization in such myocardial territories are needed to definethe impact of the present findings on the clinical managementof patients .

Rehreces1 . Marshall R, Tillisch 111, Phelps ME, et al . Identification and differentia •

tion of resting myocardial ischemia and infarction in men with positroncomputed tomography, "F-labeled fluorodeoxyglucose and N-13 ammo.nia . Circulation 1983 ;67 :766-78 .

2 . Tillisch J, Brunken R, Marshall R, et al . Reversibility of cardiac wall .motion abnormalities predicted by positron tomography . N Engl J Med1986;314:884-8 .

3 . Brunken R, Schwaiger M, Grover-McKay M, Phelps ME, 'Fillisch J,Schelbert HR . Positron emission tomography detects tissue metabolicactivity in myocardial segments with persistent thallium perfusion de,fects . J Am Coll Cardiol t987 ;10 :557-67.

4 . Schelbert HR, Buxton D . Insights into coronary artery disease gainedfrom metabolic imaging . Circulation 19W78 :4%-505 .

5 . Brunken RC, Kouou S, Nienaber CA, et al . PET detection of viabletissue in myocardial segments with persistent defects at TL 201 SPECTRadiology N89 :172 :65-73,

6 . Gropler R), Siegel BA, Lee KJ, et al . Nonuniformity in myocardiaaccumulation of fluorine- 18-fluorodeoxyglucose in normal fasted humansJ Nucl Med 199( , ;31 :1749-56 .

7 . Perrone-Filardi V Bacharach SL . Dilsizian V, Maurea S, Frank JABonow RO. Regional left ventricular wall thickening : relation to regionauptake of "fluorodeoxyglucose and 21 'Tl in patients with chronic coronary artery disease and left ventricular dysfunction . Circulation 1992 :861125-37 .

8 . Borer JS . Bacharach SL, Green MV, Kent KM, Epstein SE, JohnstorGS. Real-time radionuclide cineangiography in the non-invasive assessment of global and regional left ventricular function at rest and durintexercise in patients with coronary artery disease. N EngI J Med 1977 ;296839-44.

9. Dilsizian Y Rocco TV Freedman NMT, Leon MB, Bonow RO . Thalliunreinjection after stress-redistribution imaging : enhanced detection oischemic and viable myocardium . N Engi J Med 1990 ;323 :141-6 .

10 . Bonow RO, Dilsizian V, Cuocolo A . Bacharach SL . Identification oviable myocardium in patients with chronic coronary artery disease amleft ventricular dysfunction : comparison of thallium-201 scintigraphy willreinjection and PET imaging with 'IF-fluorodeoxyglucose . Circulatiot

1991 -.83 :26-37 .It . Sechtem U, Sommerhoff BA, Markiewicz W, White RD, Cheitlin MB

Higgins CB . Regional left ventricular wall thickening by magnetic rest)nance imaging: evaluation in normal persons and patients with global aniregional dysfunction . Am J Cardiol 1987 ;59 :145-51 .

12 . Dilsizian V . Freedman NMT, Bacharach SL, Perrone-Filardi P . Bono%RO. Regional thallium uptake in irreversible defects : magnitude of changin thallium activity after reinjection distinguishes viable from nonvialalmyocardium . Circulation 1992 ;85 :627-34 .

13 . Bacharach SL, Cuocolo A, Bonow RO, et al . Arterial blood concentratiocurves by cardiac PET without arterial sampling or image reconstructiorin : Computers in Cardiology 1988, Washington, D.C . : IEEE ComputeSociety Press, 1989:219 .

14 . lida H, Kanno 1, Takahashi A, et al . Measurement of absolute myocardkblood flow with H 2 ' 5 0 and dynamic positron-emission tornographlstrategy for quantification in relation to the partial-volume effect . CirCL

lation 1988 ;78 :104-15 .15 . Bergmann SR. Herrero P, Mark .harn J . Weinheimer CJ, Walsh MI'

Noninvasive quantitation of myocardial blood flow in human subjectwith oxygen-l5-labeled water and positron emission tomography . J AiColl Cardiol 1989 ;14 :639-52,

16 . Herrero V Markham J, Bergmann SR . Quantitation of myocardial blot)flow with 1-1 2 "0 and positron emission tomography : assessment and err"analysis of a mathematical approach . i Comput Assisted Tomogr 198 1

13 :862-73 .17 . Araujo LI, Lammertsma AA, Rhodes CG, et al . Noninvasive quantilic ;

tion of regional myocardial blood flow in coronary artery disease wit

616

PERRONE-FILARDI ET AL .MYOCARDIAL BLOOD FLOW AND REDUCED GLUCOSE FAKE

oxygen-15-labeled carbon dioxide inhalation and positron emission to-mography . Circulation 1991 ;83 :875-85.

18 . Ross J Jr. Myocardial perfusion-contraction matching : implications forcoronary artery disease and hibernation . Circulation 1991,83:1076-83 .

19 . Gall ;~4cr KP, 0sakada G. Matsuzaki M, Miller M, Kemper WS, Ross JJr . N : ; ,iuniformityof inner and outer systolic wall thickening in consciousee - Am J Physiol 1985 ;18:H241-8 .

20. ~- in RS, Watson DD, Taylor GJ, et al . Prospective assessment ofm . ,,malmyocardial perfusion bvion and after coronary revascularizationsurgery by quantitative thallium-201 scintigraphy . J Am Coil Cardiol191111 :81 4 45.

21 . Ambrosba G, Weisman HF, Mannisi IA . Becker LC . Progressive impair-ment of regional myocardial perfusion after initial restoration of postisch-emic blood &w . Circulation 1999,80 :1846-61,

21 Yamamoto Y. de Silva R. Rhodes 00 . et al . A new strategy for theassessment of viable myocardium and regional myocardial blood flowusing "O-water and dynamic positron emission tomography . Circulation1992 .86167-78 .

23 . Braunwald E, Kloner RA . The stunned myocardium: prolonged poslisch-cmic ventricular dysfunction . Circulation 1982 ;6 :1146-9 .

24, Schwaiger M, Schelbert HR, Ellison D, et al . Sustained regional abnor.

25 .

26.

27.

28 .

29.

30.

31 .

JACC Vol. 23, No . 3March 1 . 1994:608-16

malities in cardiac metabolism after transient ischemia in the chronic dogmodel . J Am Coll Cardiol 1985 -.6:336-47 .Camici P, Araujo LI, Spinks T, et al . A. Increased uptake of"fluorodeoxyglucose in postischemic myocardium of -patients with exer-cise-induced ischemia. Circulation 1986:1 :81-8 .Nienaber CA, Brunken RC. Sherman CT, et al . Metabolic and functionalrecovery of ischemic human myocardium after coronary angioplasty .J Am Coll Cardiol 1991,18:%6-78.Renstrom B, Nellis SH, Liedtke AJ . Metabolic oxidation of glucoseduring early myocprdial reperfusion, Circ Res 1989-,65 :1094-101 .Camici P . Ferrannini E . Opie LH . Myocardial metabolism in ischemicheart disease : basic principle and application to imaging by positronemission tomography . Progr Cardiovasc Dis 1989,32 :217-38 .Opie LH. Physiology and metabolism . In : Opie LH, editor . The Heart .New York: Raven, 1991 :10 :208-46 .Parodi A Schelbert H114 Schwaiger M, Hansen H . Seha C . Hoffman EJ .Cardiac emission computed tomography : underestimation of regionaltracer concentrations due to wall motion abnormalities . J Comput AssistToniogr 1984 :8 :1083-92 .Hicks RJ. Herman WH. Kalif V . et al . Quantitative evaluation of regionalsubstrate metabolism in the human heart by positron emission tomog-ra-phy . J Am Colt Cardiol 1991 :18 .101-11 .