myocardium. The underlying metabolic abnormalities responsible for hibernatingmyocardiumare unknown. Therehas not been direct evidence that human myocardiumcanremain in a hypoxic state chronically. Furthermore, the

concept of hibernatingmyocardium has been questionedrecently (7). A method for direct in vivo identificationofhypoxic myocardium would enable us to address this issue.

Strategies for detecting viable myocardium based on thereversibility of stress perfusion defects using @°‘Tl(8) orincreased [‘8F]FDGuptake relative to myocardial bloodflow (FDG/flow mismatch) (3,4) have shown promise butare not entirely satisfactory. Both methods have positivepredictive values of 75%—85%with similar negative predictive values (8). Tamakiand et al. (4) found that 20%ofsegments preoperativelypredictedto be scar by PET FDG/flow matching demonstrated improved function postoperatively. Furthermore, of the hypoperfused segments thatdemonstratedFDG uptake preoperatively, 24%continuedto have FDG uptake postoperatively in spite of patentsaphenous vein grafts. Technical considerations in relatingventricular function studies to PET images might accountfor some of the discrepancies. It is also possible, however,that the FDG method is not sensitive to the level of myocardialischemia/hypoxiapresent in these cases, or that themethods of analysis used were suboptimal. Neither Tillischet al. (3) nor Tamaki (4) quantitatedFDG uptake or flow.Furthermore, the FDG method is sensitive to plasma glucose levels. Prellwitz et al. (9) suggested that diabetic patients are more likely to have a suboptimal study thannondiabetic patients, unless careful attention is paid to thepatients' metabolic state at the time of the PET study.

We, and others, have demonstrated that the 2-nitroimidazole, fluoromisonidazole (FMISO), is selectivelytrapped in hypoxic myocardium (10—14).Following intravenous injection, FMISO is initiallydistributedthroughoutthe total body water space, with subsequent clearancefrom the normal myocardium directly as a function ofplasma clearance (10). Because its retention is dependenton active electron transportenzymes, it is not trapped innecrotic tissue (14). When labeled with the positron emitter, ‘8F,the trappedFMISO can be imaged with PET and

Ruohne-i8-fluoromaonidazole (FMISO) is trapped in hypoxiobut viable canine myocardium.Because of the potentialfor itsuse as a marker of myocerd@v@lfty, we compared FMISOactivityto [18flfl@ygl@j@@ (FDG)activityin the samemyocard@samples from eight dogs subjected to 3 hr of moderate regional myocardial ischema Methods: Tritiated FMISOwas injected 15-30 mmafter onset of regionalischemia (40%-70% reductionin sysdic wallthickening)wt*h was maintainedfor 3 hr. FDG was injected after 2 hr of ischemia. Myocardialblood flow (MBF)was measured by the radiolabeled microsphere technique atthe timeofeai@hradiotracerinjection.At3 hrof ischemia,the heartwas excisedandcutintoshort-wdsslices.One slice encompassing both ischemicand normaltissue wascut into 64 samples. FMISOand FDG actMty in each samplewere normalized to the mean normal zone activity and furtherexpressed as a function of regional MBF. Results: FMISO uptake was consistendy greater than FDG uptake, although thiswas significantlydifferentonlyfor MBF,between 40%-60% ofnormal.When analyzed relativeto endOcardiaJ-epicardiallocalion, endocardial FMISOuptake was significantlygreater in allhypoperfusedsamples. Conclusion: These resultssuggest thatFMISOis as sensitiveas FDGfordetectingmyocardialischemiaand could be used for identificationof viable myocardium.

Key Words: myocardial viability;fluodne-l84luorodeoxyglucose; myocardialbloodflow;fldne-18-fluoromisonidazde

J NuciMed1995;36:1633-1638

everal studies indicate thatchronic left ventricularwallmotion abnormalitiesmay be reversed by coronary revascularization (1—3).Recent data (4) suggest that 30%—50%of dysfunctional myocardial segments are potentially viable. From these observations, the concept of persistentischemia as a basis for chronic regional dysfunction hasemerged (5,6). This concept has been termed hibernatingmyocardiumand has promptedthe search for noninvasivemethods for detecting chronically hypoperfusedbut viable

ReceivedJune 17, 1994;revisionacceptedApr.12, 1995.For correspondenceor reprintscont@ James H. Caldwel,MD,Divisionof

Card@ogy(ill-C), DepartmentofVeteransAffalrsMe@calCenter,1660S. Co.lumblanWay,Seattle,WA98108.

1633FMISOversus FDG Uptake in lschemic Myocardium•Caidwell at al.

Comparison of Fluorine- 18-Fluorodeoxyglucoseand Tritiated Fluoromisonidazole Uptake duringLow-Flow IschemiaJ.H. Caldwell, J.R. Revenaugh, G.V. Martin, P.M. Johnson, J.S. Rasey and K.A. Krohn

Division ofCardio1o@, Depamnent of Veterans Affai@r Medical Center; and Departments ofRadioIo@ and RadiationOncology, University of Washington, Seattle, Washington

kiI.ct FMISO&@ FOG & EtShanIzeM — Mm@

4 11 3L!!@@

Hsmodynamlc Coos@mds@w@thdunlngS@on

TIme(mm)

wallthickeningwhichseemedstableafter20mmandthusFMISOand microsphereswere injected. Over the next 15—30mm, however, wall functionreturnedto controlvalues despite of completeocclusion of the LAD coronaryartery.Presumably,this occurredas collateral vessels opened. Two hours after the initial microsphere injection, a second microspherewas injectedfor measurement of regionalmyocardialblood flow. This was followed by aninjection of 2—4mCi of [‘8FJFDGintravenously over a 1-mmperiod. The reduced level of wall thickening was maintained foran additionalhour, at which time the animalwas euthanizedby aoverdose of potassium chloride.

The heartwas rapidlyexcised, rinsedfree of blood and the leftventricular cavity was filled with an expandable polyurethanefoam which rapidly hardened to maintain its shape. A 1.5-cm thickslice was cut perpendicularto the long-axis of the heart andencompassing the sites of both echo crystals. This slice was thencut into 16equal radialsectors andeach of these sectors was thencut into four epicardialto endocardialsamples for weighing andcounting. The averageweight for 576 samples was 0.55 ±0.22 g.

Data AnalysIsRRinterval,systolicbloodpressureandsystolicwallthicken

ingwere measuredfromthe calibratedstripchartrecordings.Thepeak of the R wave was used to define end-diastole and endsystole at the time of the aorticpressure diacroticnotch. Systolicwallthickeningwas calculatedby determiningtheabsolutethickening differencebetween end-diastole and end-systole and dividing that difference by the end-diastolic thickness. Five consecutive beats were averaged at each time point for wall thickening,heart rate and blood pressure determinations. Percent systolicwall thickeningat each time pointwas subsequentlyreferencedtothe baseline value and expressed as a percentageof baseline.

The tissue samples were initiallycounted in a gamma counterwith a 3-inch Na! crystal, along with standards for each of themicrospheres and ‘8F.After the ‘8Fhad decayed for 10 halflives, the sampleswere recounted for gammaactivity from themicrospheres. All counts were corrected for spillover using amatrix inversion technique (18). To accurately count tritium, itwas necessary to separate the microspheresfrom the tissue samples. To accomplishthis separation,tissue sampleswere frozen at—70°Cand crushed with an aluminum mortar and stainless steelpestle chilled to —70°C.Aliquots (30—40mg) were placed inchilled, preweighedmicrofugetubes and reweighed. The sampleswerehomogenizedin distilledwaterat 1:2(w/v),microfugedfor10 mm and 0.5 ml of their supernatantwas transferredto scintillationvials. One milliliterof Soluene 350 (Packard)was added toeach vial and the vials were vortexed and allowed to stand for 30mm.

Hydrogen peroxide (0.05 ml 30%) was added to each vial andthe vials swirled and allowed to stand for 24 hr. Hionic-fluorscintillation cocktail (Packard) was added to each of the vialsbeforetheywerecountedforgammaradioactivity(PackardAutogamma) and beta decay (Packard 1600TR) because leaching ofradioisotope from the microspheres interfered in tritium counting.Thismicrospherespilloverintothe liquidscintillationspectrumwas determinedby back calculatingthe contributionof eachgammaradionuclideto eachof thebeta regions.A set ofefficiencystandardswas prepared for “3Snand ‘°3Ruusing Soluene 350treated blood as a quenching agent. These efficiency standardswerecountedin thesameliquidscintillationcounterandanefficiencycurvewas calculatedfor each of the gammaradionuclidesin three beta windows. The quantity of each of the gamma

FiGURE1. Graphicalrepresentabonofthestudyprotocol.

its uptake occurs in myocardial regions with reversibledysfunction (12,13).

Because the uptake and retention mechanisms ofFMISO and FDG differ, we postulated that importantdifferences might exist in the relative sensitivities of the twoagents for the detection of myocardial ischemia. The purpose of this study was to compare the trapping of FMISOto FDG uptake in an animal model of moderate ischemianot associated with infarction. This model demonstratesimmediately reversible regional dysfunction followingreperfusion (12).

METhODS

Animal Surgery and InstrumentationAfter a 12-hr fast, nine mongrel dogs (20—35kg) were initially

anesthetized with thiamyl sodium (15 mg/kg), intubated, mechanically ventilated to maintainarterialblood gases within a physiologic level and maintained on halothane as an anesthetic agent.Througha leftthoracotomy,theanimalswereinstrumentedwitha left atrial line and a mechanical or hydraulic occluder on themidportionof the leftanteriordescendingcoronaryartery(LAD).Doppler crystals were sutured to the epicardiumin the distribution of the LAD and circumflex coronaries for measurementofsystolic wall thickening(15).

ExperImental ProtocolAfter hemodynamic stabilization(15—30mm), wall thickening,

heart rate and systolic blood pressure were recorded. To standardize the recording of ventricular function and eliminate theeffect of respiratoryvariation, the stripchartrecorderwas startedevery 15 mm by a timer to run at 100 mm/sec. As soon as therecorderhadstarted, the respiratorwas stoppedat end expirationand 10—12cardiac cycles were recorded and the respiratorrestarted. The fifth beat after the respirator stopped was used as thestarting point to determine EKG RR interval, systolic blood pres

sure and wall thickness. Heparin(2000—3000units) was administered intravenouslyand the coronaiy occluder was then adjustedso as to reduce anteriorwall thickeningby approximately40%—70%.Theprotocolis summarizedin Figure1. Plasmaglucosedeterminations were made during the control period and at hourlyintervalsduringthe ischemic period.Afterstabilizationof reducedwall thickening(15—30mm), 2—4x 10@radiolabeledmicrospheres(15p.mdiameterlabeledwith‘‘3Snor ‘°3Ru)wereinjectedintothe left atrium,alongwith simultaneouswithdrawalof blood fromthe ascending aorta, for 1 mm to measure regional myocardialblood flow (16). This was followed by an intravenous injection oftritiatedFMISO(approximately0.01mCi/kganimalweight)(17).

In seven of the nineanimals,wall thickeningwas maintainedwithin the 40%—70%reduction range for 2 hr by occasional adjustmentof the occluder. In two animals, therewas a decrease in

1634 The Journalof NudearMedicine•Vol.36 •No.9 •September1995

ParameterControlIschemlaHeart

rate (bpm)101 .6 ±12.497.9 ±II.9Systolicblood pressure (mmHg)101 .6 ±13.196.3 ±9•3*SystolicwallthickeningLAD(%ofcontrol)—65.3±20.1Circumflex

(% of control)—82.8 ±l9.l@PlasmaGkicose(mgfcl)64±1468±11*p

< 0.05vs.control.tp0.05 vs. LAD.

0 0.2 0.4 0.6 0.8 I 1.2 1.4

radionuclides in each sample was measured with the gammacounterandthenthecontributionof eachof thegammaradionuclides to each of the beta regions was calculated using the efficiency curvesdescribedabove. Those contributionswere subtracted from the counts in the correspondingbeta regions. Thequantity of each of the beta radionuclides in each sample was

calculated from the remainingbeta spectrum and the beta radionuclide efficiency curves for each of those regions.

Microsphereactivitywas convertedto myocardialblood flow(ml min ‘@ g 1) using the standard reference organ technique(16).Fluorine-18-FDGand3H-FMISOactivitywere expressedascounts per minute per gram and disintegrations per minute pergram, respectively, in each of the tissue samples.

To facilitatecomparisonbetweenthe animals,flow, FMISOand FDGvalues in each samplewere normalizedas a decimalfraction of the mean @‘normalmyocardial―activity for each animal.Averagenormalzonebloodflowwascalculatedbyaveragingthe respective values for the two separate microsphereinjectionsfrom12 myocardialtissuesampleslocatedin the centerof theanatomicallynormalzone. “Normal―FMISOandFDGactivitywas calculatedby averagingthe activityin eachof the same 12samplesusedforthenormalbloodflowdetermination.To examinc the effect ofendocardial or epicardiallocationon the uptakeofthetwocompounds,thesampleweightsandactivitiesforthetwoinnermost and outermost samples from each sector were addedtogether and averaged; the data were expressed as above.

Intheclinicalsetting,absoluteFDGactivitymaybelowerinanischemic but viable myocardial region than in a normal region inthe same heart. The activity may be relatively greater, however,thanthecorrespondingflowto thatregion.Inthenormalregion,FDG and flowwill be balanced. To achieve the greatest sensitivityand specificity for identificationof viable myocardiumin humanstudies, regional PET FDG activity is frequentlyexpressed as aratioto the FDG in a known normalregionor to a normalfile andthenthisratiocomparedtothecorrespondingregionalmyocardialblood flow [either in absolute terms or normalizedto a normalregion or to a normal file (19,20)]. To express our results in acomparable manner, we divided the normalized FDG activity ineach piece by its correspondingabsoluteflow and plottedtheresults as a function of the normalized (ischemic-to-normal ratio)myocardial blood flow. For comparison, the FMISO data havealso been expressed in the same manner, although this couldpotentiallyintroduceapparentFMISOuptakein normalregions.This artifactoccurs because FMISO concentrationin normaltissue at late times (morethan 1 hr) is drivenby the partitioncoefficient of FMISO, resulting in relatively low concentrationandvery littlevariation(approximately20%)comparedwiththeapproximatelyfivefoldheterogeneityin myocardialbloodflowinnormal myocardium (21). This would result in a fivefold range ofFMISO-to-flowratios.

Statistical AnalysIsDifferencesbetweencontrolhemodynamicandplasmaglucose

levels andthe averageforthe ischemicperiodforeachof thesevariables were compared using a paired t-test (two-tailed). Systolicwallthickeningduringischemiaas a percentof controlwascompared between the LAD and circumflex regions using theWilcoxonSigned-Ranktest.Becauseof thesmallnumberof samples with low flow, the datawere groupedfor each 0.2 intervalinflow from zero to maximalflow for the comparisonbetweenFMISO and FDG uptake. FMISO and FDG activity in each of the

TABLE IPh@ogic Measures

myocardialblood flow ranges were compared using analysisofvariance.

RESULTS

Control heart rate, systolic blood pressure, systolic wallthickeningand plasmaglucose levels are shown in Table 1along with the average values for the ischemic period.

The differencebetween LAD and circumflexwall thickening is at the limits of statistical significance (p = 0.05,Table 1). The lack of a largerdifferenceis likely due to twoanimals (as noted in the Methods section) in which therewas an initial reduction in wall thickening in the ischemic(LAD) region, however, thickening returned to controllevels with time, despite complete occlusion of the coronary artery. Presumably, this occurred because of theopening of collaterals.

Uptake of FMISO and FDG for all of the samples (n =576)normalizedonly to their respectivevalue in the normalregion is shown as a function of the normalized myocardialblood flow in Figure 2. The greater FMISO activity in the

FIGURE 2. The indMdualdata pointsfor FMISOand FDGInAgure 2 have been groupedaccordingto myocardialt@oodflowintervalsof 0.20 units. The statisticalcomparisonsare betweenFMISOand FDG for a given flowinterval.

I.@

Normalized Flow

FMISOversus FDG Uptake in lsthemic Myocardium•Caldwellat al. 1635

: • .0 FDG

12.

10@

8• oD•

6

4. !.cIlJ

:@ 0.2‘0:4NormalIzed Flow

FiGURE5. FMISOandFOGactMtyina singleexperiment.Thsue actMtyineachsamplehasbeennormalizedtothecorresponding absolute myocardlal I@oodflowin each sample. This method isthe same as that used for dinical FDG via@ktystudies.

This is shown for a single animal in Figure 5 and for allanimals in Figure 6. There was a trend for the FMISOuptake to be greater than FDG, but this was statisticallysignificant only for one range of flows (0.4—0.6)(Fig. 6).

When the data were examined in terms of endocardialversus the epicardialhalf of the myocardialwall, we foundthe same increased uptake of both tracerswith decreasingflow. Uptake of both compounds, however, was higher inthose samples lying within the endocardialhalf, suggestingthe ischemic insult was greater in the endocardium.

DISCUSSION

Following injection, FMISO is initially distributedthroughout the total body water space, with subsequentclearance from the normalmyocardiumdirectly as a function of plasma clearance (10). Because its retention is de

L+1

1-. I:1

0-0.2 0.2.0.4 0.4.0.6 0.6-0.8 0.8-1.0 1.0-1.2 1.2-1.4

Nomiallzid Flow 0

FDG41492FiGURE3. FMISOandFOGuptakein576myocardialsamplesnormalizedto the acdvityin12samplestakenfromthe centeroftheanatomically normal zone. Uptake is plotted as a function of myocardial blood flow in the same samples normalized to the averageflow in the same 12 normal samples.

low flow regions relative to the FDG uptake is apparent.Figure 3 shows the normalized FMISO activity and FDGuptake grouped according to increments in myocardialblood flow. Again, the greater FMISO uptake is apparent,although this was significantly different from FDG only forthe 0.4—0.6flow range. The relationshipbetween normalized FMISO and FDG uptake in the same samples at threelevels of myocardial blood flow is shown in Figure 4. Asshown, in the most ischemic regions (normalized MBF <0.5), the uptake of FMISO was greaterthan FDG uptake in14 of 16 samples. Mean FMISO uptake in these 16 sampleswas 2.52 ±1.01 compared to 1.52 ±0.82 for FDG (p <0.002).

When normalized FMISO and FDG uptake in each samplc were expressed as ratios to the correspondingabsoluteflow, in the mannerfrequently used for analysis of clinicalFDG-PET images, uptake increased as flow decreased.

Comparison of FMISOto FDG in Same Sample

.@MBF0 <0.50 @0.5<0.6

. @O.6<0.7

0.0.2 0.2@0.4 0.4@0.6 0.6.0.8 0.8-1.0 1.0-1.2 1.2.1.4

@ R@

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

NormalizedFDGUptakeFiGURE 6. Row-normalizedFMISOand FDG activityas described for Figure 5 for all 576 tissue samples are expressed intermsof0.20intervalsofflow.Statisticalcomparisonsare betweenFMISOand FDGfora given flow interval.

FiGURE4. NOrmalIzedFMISOand FDGuptakein the samesample are compared for all samples whose normalized myocardialbloodflow (MBF)was withinthe indicatedranges.

1636 The Journalof NuclearMedicine•Vol.36 •No. 9 •September1995

pendent on active electron transport enzymes, it is selectively trapped in hypoxic myocardium (10—14)but not innecrotic tissue (14). Thus, the increased FMISO uptake inareas of moderate myocardial ischemia observed in thisstudy is consistent with previous work by our group andothers (12,13). The finding of most interest is the trendtoward greateruptake of FMISO than FDG. This suggeststhat FMISO might be an alternative to FDC for determining myocardial viability. FMISO may be particularlyadvantageous in patientswith diabetes in whom FDG imagestend to be of poorer quality (9).

Another potential advantage of FMISO over FDG toidentify chronically ischemic myocardium is that FMISOuptake, expressed in absolute terms, is elevated even inareas of very low flow, whereas absolute FDG uptake isnot. In practical terms, this suggests that FMISO imagingstudies may not require a separate flow study, as is routinely done for FDG to identify areas of FDG flow mismatch (19,20) or to derive quantitative estimates of glucoseutilization (7).

It is not surprisingthat both the FMISO and FDG activity was greatest in the endocardial half of the left ventricularwall. It has long been recognized that the endocardiumis more sensitive to hypoperfusion than the epicardium.This greater endocardial uptake also supports the wallthickening data that the ischemic burden was relativelymild and would have been reversible.

There are potential limitationsto this study. Myocardialblood flow was only measured at two time points and theaverage of these two were assumed to represent flowthroughout the 3 hr of ischemia. The validity of this assumption cannot be tested. There was little change, however, in the hemodynamic parameters or wall thickeningduring this period, suggesting that there should have beenlittle change in flow. If change occurred, it was probably adecrease in flow along with the trend for decrease in metabolic demands in both the normaland ischemic zones, assuggested by the trend in wall thickening changes. Furthermore, since FMISO and FDG were measured in the samepieces, any changes in absolute flow would not affect therelative differences in tracer trappingor uptake. The diiference between FMISO and FDG uptake may have beenunderestimated because the experiment was terminatedafter only 3 hr of ischemia. We have previously shown aprogressive increase in FMISO uptake in ischemic myocardium over a 4-hr period (10, 12). Thus, by sampling at3 hr, we may haveunderestimatedFMISO activityin theischemic samples, whereas FDG sampling for a similartime used for clinical imaging protocols resulted in optimized FDG data. Also, because of the 2-hr time differencebetween injectionof FMISO and FDG, differentanaerobicconditions may have existed. There was no significantdiiference in MBF between the time of the FMISO and FDGinjections, althoughtherewas a trendfor MBF to be lowerat the time of the FDG injection. This could have resultedin greater uptake of FDG than would have occurred hadFDG been injected at the time of the FMISO injection. On

the other side of the ischemia issue, there was an insignificant decrease in blood pressure with a constant heart rateover this time which would reduce myocardialwork, thuspotentiallydecreasingischemia and lessening FDG uptake.This could also reduce FMISO uptake.

FDG uptake may have been underestimated because ofa potential inhibitionof glucose uptake througha heparininduced release of lipoprotein lipase, with subsequent increase in the plasma free fatty acids. Nuutila et al. (22)recently reported that a 20-fold increase in plasma fattyacids produced by an infusion of heparin (3000 IU infusedover 3.5 hr) and Intralipid(200 ml of 20%)resulted in onlya 30% decrease in myocardial glucose uptake as measuredby PET. It seems unlikely that heparin alone, as used inour study (2000—3000IU given 2 hr before the FDGinjection),would increase free fatty acids to the level notedabove. We did not, however, measure plasma fatty acidlevels and thus cannot exclude the possibility that FDGuptake was artificially depressed relative to what wouldhave been expected for the level of ischemia present.

The animals used in this study were fasted for 8—12hrwhich clearly suppressed plasma glucose levels and presumably the insulin levels. Both of these conditions wouldhave the tendency to decrease FDG uptake. We used theaverage FDG activity in anatomically normal myocardialregions, however, to normalize the values in the individualsamples both in the ischemic and normal regions. Unlessthere is an unrecognized relative greaterinhibitionof FDGuptake iii ischemic compared to normal regions because ofthe fasting state, there should not have been an underestimation of FDG uptake.

Independentof the effect of basal glucose levels on FDGuptake, if flow had been reduced to the point of preventingwashout of inhibitory metabolic intermediates, FDG uptake could have been underestimated. Given the relativemild nature of the ischemia, it seems unlikely that thisphenomenon occurred to any great extent, although thisissue cannot be addressed by our data. Similarly, highplasma lactate levels also could have suppressed FDGuptake. Althoughplasmalactate levels were not measured,given the relatively mild ischemia and the small ischemicregions, the amount of lactate spilling over into the systemic circulation should have been small and the lactateeffect minimalor nonexistent.

CONCLUSION

FMISO trappingduringmoderate myocardial ischemiais at least equal to, if not greater than, that for [‘8F]FDG.FMISO is a potentially attractive alternative to FDG forseveral reasons:

1. FMISO is taken up only by cells that are hypoxic butstill retain their viability.

2. FMISO probably does not require normalization tomyocardial blood flow for determination of tissue viability.

FMISOversus FDG Uptake in IschemicMyocardium•Caidwellat al. 1637

3. FMISO isprobablyunaffectedby tissuepH or basalmetabolic conditions such as plasma fatty acid orglucose.

4. FMISO requires only a simple analysis of tissue-toblood ratios. Finally, we observed FMISO uptake inclinically stable patients (23); however, studies cornparing FMISO and FDG uptake in the same patientpopulation prior to coronary revascularization arenecessary to determine the relative role of the twocompounds in identifying viable myocardium.FMISO is unlikely to be useful in clinical conditionsof transient ischemia, such as might occur duringexercise, because of the need for ischemia/hypoxia tobe sustained during the interval from radiopharmaceutical injection through imaging to observe trapping.

ACKNOWLEDGMENTS

Thetechnicalassistanceof MarilouGronkaandJohnChenginthe experimental animal lab, Jodi Stewart for [18F]FDGsynthesisandZdenkaGrunbaumfor3H-FMISOlabelingaregreatlyappreciated. This studywas supportedby the Departmentof VeteransAffairs Medical Research Service and National Institutes ofHealth grants HL38736, HL 50239 and CA34570.

REFERENCES

1. BnmdageBH, MassieBM, BotvinickEH. Improvedregonalventricularfunctionaftersuccessfulsurgicalrevascularization.JAmC&Cardiol 1984;3:902—908.

2. Rankin iS, Newman GE, Mulhbaier UI, Behar VS, Fedor JM, SabistonDC. The effect of coronary revasCUlariZatiOnon ventricular function inischemicheart disease lAbstracti.I ThonicCwviiovascSw@1985;90:818.

3. Tillisch JH, Brunken R, Marshall R, Ct at. Reversibility of cardiac wallmotion abnormalitiespredicted by positron tomography.N Engi I Med1986;314:884—888.

4. Tamaki N, Yonekura Y, Yamashita K, Ctal. Positron emission tomographyusing fluorine-18 deoxyglucose in evaluation of onronaiy artery bypassgrafting. Am I Cart/Aol 1989;64:860-865.

5. BraunwaldE, RutherfordJD. Reversibleischemicleftventriculardysfunction: evidence for the “hibernatingmyocardium.―JAm Coil Caniiol 1986;8:1467—1470.

6. RahüntoolaSH.Thehibernatingmyocardium[EditorialJ.AmHeartl 1989;117:211—221.

7. vanoverschelde 3-U, Wijns W, Depre C, et al. Mechanisms of chronicregionalpostischemicdysfunctionin humans:new insightsfrom the study

of noninfarctedcollateral.dependentmyocardium.Circulation1993;87:1513—1523.

8. BonowRO,Dilsizianv, CuocoloA, BacharachSL Identificationofviablemyocardium in patients with chronic comnaiy artery disease and left yentricular dysfunction: comparison of thallium scintigraphy with reinjectionandPETimagingwithF-18fluomdecxygluonse.Circulation1991;83:26-37.

9. PrellwitzJA, vastu MC,Sunderlandii, ShiueCY, GuptaMC,FrickMP.Investigationof factors influencingFDG myocardial image quality [Abstracti. I Nuci Med 199132:1039.

10. MartinGV,CaldwellJH,CerqueiraMD,RaseyiS, KrohnKA. Enhancedbinding of the hypoxic cell marker [H-3Ffluoromisonidazole in ischemicmyocardium. I NuciMed 1989;30:194-201.

11. MartinGV,CerqueiraMD,CaldwellJH,RaseyiA, EmbreeL, KrohnKA.Fluoromisonidazok:a metabolic marker of myocyte hypoxia. Circ Res1990;67:240—244.

12. MartinGV, CaldwellJH, GrahamMM,et al. Noninvasivedetectionofhypoxicmyocardiumusing[‘8FJ.fluoromisonidazoleand positronemissiontomography. JNucI Med 1992;33:2@l2-2208.

13. SheltonME,DencecS, HwangD-R,HerreroP.WelchMi, BergmannSR.In vivo delineation of myocardial hypoxia during coronaty occlusion usingF-18-misonidazoleand positronemissiontomography:a potentialapproachfor identificationofjeopardizedmyocardium.JAm CoilCardiol199016:477—485.

14. SheltonME,DenceCS,HwangDR.WelchMi, BergmannSR.MyOCardiaIkineticsof fluotine-18-misonidazok:a marker of hypoxicmyocardium.INuciMed 198930:351-358.

15. HartleyCi, LatsonLA, MichaelLH, SeidelCL, LewisRM,EntmanML.Doppler measurement of myocardial thickening with a single epicardialtransducer.AmI P@sioI 1983;251:H1044-H1055.

16. HeymannMA, PayneBD, HoffmanilE, RudolphAM. Bloodflow measurements with radionuthde-labeled particles. Plvg Cardiovasc Dic 1977;20:55—79.

17. GnmbaumZ, FreauffSi, KrohnKA, WilburD5, MageeS, Rasey iS.Synthesisand characterizationof congenersof misonidazolefor imaginghypoxia.INuciMed 1987;28:68-75.

18. Universityof WashingtonAcademic ComputerCenter. Pivot invertmethod. Document W00042.

19. SchelbertHR. Metabolicalterationsin reversil$ydysfunctionalmyocardium as identified with positron emission tomography. Am I Cardiac Imaging 19fl;6:219—227.

20. TamakiN,OhtaalH,YamaslathK,etal.Metabolicactivityintheareasofnew fill-in after thallium-201 reinjection: comparison with positron emissiontomography using fiuorine-18-deoxyglucose. INuciMed 1991;32:673—678.

21. KingRB, BassingthwaighteiB, HalesiRS, RowellLB. Stabilityof heterogeneity of myocardialblood flow in normal awake baboons. Circ Res1985;57:285—295.

22. Nuutila P, Koivisot VA, Knutti J, et al. Glucose free fatty acid cycleoperates in human heart and skeletal muscle in vivo.J Clin Inwst 1992;8%1767—1774.

23. Revenaugh iR, Caldwell iH, Martin GV, Grierson JL, Krohn KA. PositronemissKMl tomography (PET) imaging of myocardial hypoxia with ‘8F-fluoromisonidazole(FMISO) in postmyocardialinfarction patients [Abstracti. Circulation1991;84:II-415.

1638 The Journalof NudearMedicine•Vol.36 •No. 9 •September1995