such as frontalhypoperfusionandasymmetricinvolvementof the cerebral cortex have been observed (7,8), the presence of hypoperfusion in the temporo-parietal areas demonstrated by SPECT and @“@Tc-hexamethylpropylenamine oxime (@Tc-HMPAO) has a high predictive valuefor the presence of AD in patients with memory and cognitive impairment (8).

The use of SPEC!' in clinical practice is facilitated by itslower costs and its wide diffusioncompared to other techniques such as PET. On the other hand, the reliability ofSPECT findings has always been limited by the relativelypoor spatial resolution of circular shaped rotating gammacameras and the low brain uptake values of the tracersused (9). With the development of new high-resolutionSPEcT systems, both spatial resolution and detection efficiency oftomographic imaging are improved compared to

single-head rotating gamma camera systems (10), thus allowing a better image quality and improving the quantification of regional radioactivity. However, the improvement in image quality obtained by high-resolution SPECTdoes not diminish the limitations related to the kinetics ofthe tracers used (11).

PET is a more accurate technique than SPECT withregardto resolution and quantitativeaccuracy (12,13) andit should be regarded as the reference technique for theassessment of regional brain function. A reduction of glucose metabolism measured with PET and 18F-fluoro-2-deoxy-D-glucose (‘8F-FDG)in the temporo-parietal cortexhas been consistently shown in patients with mild to modcrate probableAD (14—16),althoughother patternsof glucose metabolism such as frontal (17) or asymmetric involvement of the cerebral cortex (18—21) have also been

observed. Although PET has been successfully used toassess the metabolic function in patients with probableAD, its applicability to clinical practice is limited by thehigh costs; indeed PET has now been mostly used forresearch purposes.

The aim of this study was to carry out a systematicevaluationof PETversus SPECT in patientswith early AD

SPECT studies of regionalcerebral perfusionwItha high-resoIutlonsystem were comparedto PETstudies of regionalcerebralglucoseutilization(rCMRgk@)in21patientswfthprobableAlzheimar'sdisease(AD).Ten normalsubjectswerealsoeValUatedwfthSPECTand10w@iPET.Methods:rCMRgIc(forPET)andcounts(forSPECT)in theassociativecor@ceswerenormalizedtotheaveragerCMRgIc,andcountsinthecalcatinecortexandbasalgangliawereconsideredasa “referencearea―to obtainaratio.The ratiodifferencesbetweenpatientsandcontrolsweretestedw@iANOVAperformedseparatelyfor PETandSPECT.Results:ThedifferencebetweenprobableADpatientsandcontmls was significantfor bofhPET (p < 0.00001)and SPECT(p < 0.005);thisdifferencewassignificantfor the frontal,ternporaland parietalcortices(p < 0.0001)for PET,and for thetemporal(p < 0.005)and patietal (p < 0.001)cor@cesforSPECT.Temporo-pailetaldefectsweredetectedin all subjectswith PET and in 90% with SPECT.Conclusion: PET andSPECTareabletodetecttharacteristicternporo-parietalabnormalitiesinprobableAD.However,thepresenceofabnormalitiesinotherassociativeareasisbetterevaluatedwithPET.

KeyWords:single-photonemlulon computedtomography; positron emission tomogmphy; AWielmer's diaaaae;fluodna-18-fiuorodeoxyglucoee

J NuciMed1994;35:210—216

he assessment of cerebral perfusion with SPECT isbeing increasinglyused in the routine clinical evaluation ofneurological and psychiatric disorders (1).

The clinical relevance of SPECF studies in the evaluation of patients with probable Alzheimer's disease in theearly phase is related to the evidence of an abnormal pattern of perfusion (Le., hypoperfusion of the temporo-parietal cortex) consistently detected by SPECT studies ofblood flow in these patients (2—6).Although other patterns,

ReceivedJun. 14, 1993;revisionacceptedOct 20, 1993.For repdntsand correspondencecon@ @r@inalinens. MD,Medk*iaNu

dears, IstitutoHSanRaltaele,ViaOlgettina60,20132Milano,Italy.

210 TheJournalofNtidearMedicine•Vol.35•No.2•February1994

High-Resolution Technetium-99m-HMPAOSPECT in Patients with Probable Alzheimer'sDisease : Comparison with Fluorine- 18-FDGPETCristina Messa, Daniela Perani, Giovanni Lucignani, Arturo Zenorini, Felicia Zito, Giovanna Rizzo,Franco Grassi, Angelo Del Sole, Massimo Franceschi, Maria Carla Gilardi and Ferruccio Fazio

INB-CNR, Unive,@rityofMilano, Institute H.S. Raffaele, Milano, Italy

Duration@tPET!ofhigh-resolutionPatientAgediseaseSPECTno.

Sex (yr)(mo) MMSEMRI (days)

resolution SPECF study with @Tc-HMPAO.The average timebetween PET and high-resolution SPECF examinations was 11.7±21.7 days (Table 1) and the order of the exams was random.

Normal controls for the PET study were 3 males and 7 females(mean age 47 ±13yr), while normal controls for the high-resolutionSPECFstudywere4 malesand6 females(meanage53 ±13yr). All controlswerefreefromknownneuroloajcaldiseases,includingcerebrovasculardiseaseas assessedby clinicalhistory,standardclinicalandneurologicalexaminationsandneuropsychologicaltesting(MMSE).

The protocolwas approvedby the EthicsCommitteeof theScientificInstitute H. San Raffaele.Informedconsent was obtamed from each subject.

PETImagingPET studies were performed with a PET tomograph (931fi14-12

CPS/Siemens, Erlangen, Germany)with a spatial resolution equalto 6.3 mm fullwidth at half maximum (FWHM) in the axial imageplane.Thetomographhadanaxialfieldof view of 5.4 cmwhichyieldedsevenaxialimageslices675 mmthick(23).

Thepatientsrestedon thebed in a dimroom,eyes openandears unplugged,with their head positioned in a special headholderwhichusesthesubject'sdentalmorphologyforpositioningand fixation(24). In some patients, a personalmoldedfoamwasused to fix the head.

Thesynthesisandqualitycontrolof ‘8F-FDGwas carnedoutaccordingto standardprocedures(25). Fourteen images,parallelto theorbito-meatalline,wereacquired45 mmafterthe intravenous injection of approximately 250-300 MBq of ‘8F-FDOwithtwo spatially contiguous emission scans, each lasting 10 mm, inorderto examinethe whole brain.Duringscanning,approximately 14millioncounts were acquired.

Axial slices were reconstructed employing a Hann filter with acut-offfrequencyof 0.5 cycles/pixel(matrix 128x 128).Photonattenuationwas correctedby use of coefficients determinedfromtwo transmissionscans, one for each bed position,obtainedwitha @Gef@Gaexternalringsource,lasting10mmeach,priorto the18Fp@ injection.

Plasma 18F-FDO and glucose time-activity curves (input functions) were obtained by blood sampling from the radial artery atapproximately3-sec intervalsfor the first minute after injectionandatprogressivelylongerintervalsforthedurationof theentirestudy.

Valuesof re@onalcerebralmetabolicrateof glucose(rCMRglc)were calculated according to the method based upon the autoradiographictechniqueofSokoloffet al. (26)andrevisedby Reivichet al. (27)forhumanPETstudies.

High-RISOIUtIOnSPECTImagingHigh-resolutionSPECF imagingwas performedwith an annu

lar SPECF system (CERASPECF—Digital Scintigraphic Inc.,Maynard, MA) (28) with a spatial resolution equal to 8.4 mmFWHM in the center of the axial plane. The system covered anaxial field ofview of 10.5cm. As with PET, the patients rested ona bed in a quiet room at the time of injection. The patient's headwas positioned in a head-holderspecificallybuilt for theCERASPECFmachine.

Images parallel to the orbito-meatal line were acquired beghining 15mm after intravenous injection of approximately 740-800MBq of@Fc-HMPAO. In 30 mm, 7 to 9 million total counts wereacquired for the entire head, using a square pixel size equal to 1.66mm and an angular step of 3°over 360°(matrix size 128 x 64 and120projections).

TABLE IPatients'clinicalData

I F 59 12 20 Atrophy2 M 57 24 19 Atrophy3 M 57 24 16 Norm@4 M 67 36 22 Atrophy+wm5 M 62 24 20 Atrophy6 M 60 12 15 Atrophy7 F 70 5 14 Atrophy8 F 77 24 25 Atrophy9 F 73 29 29 Atrophy

10 F 74 24 21 Wm11 M 57 24 24 Atrophy12 M 66 36 19 Atrophy13 F 52 12 22 Atrophy+wm14 N 64 36 14 Atrophy15 F 60 24 23 Normal16 M 63 24 23 Atrophy17 F 62 6 29 Atrophy18 F 44 24 14 Normal19 M 68 36 24 Atrophy20 F 69 24 12 Atrophy21 F 58 36 12 Atrophy

MMSE= minimentalstateexamination;@tt= timeelapsedbetweenPETand SPECTstudIes; Atrophy - m@ddegree ofatrophy wm = smallesionsof thewhitematter.

with the purpose of evaluating, in the same patients, therelative ability of high-resolutionSPECF with @Tc-HMPAO in comparisonwith PET and ‘8F-FDGto differentiatepatients in the early phase of the disease from normalcontrols and to detect abnormalities that are considered

characteristic of probable AD patients, such as temporoparietal impairment.

MATERIALS AND METHODS

Sub@tsTwenty-one patients with probable AD (10 males, 11 females,

mean age 62.8 ±7.8 yr) were studied. The main clinical andradiological features of these patients are reported in Table 1.

Allpatientswere clinicallydiagnosedas probableAD (averagedurationof illness 23.6 ±9.6 mo) according to standardcriteria(22). Other causes of dementia were excluded through neurological examinationand laboratoryblood tests. The severityof thedisease, assessed with the Mini Mental State Examination(MMSE), was mild to moderate (MMSE mean score 19.8 ±5;Table 1).

An MRI study was performed in all patients with a 1.5 Tscanner (Magnetom, Siemens, Enrlangen, Germany). A mild degree of generalizedatrophy was observed in 16t21patients andsmalllesionsofthe whitematter,probablydue tovascularlacunarlesions, were present in 3t21patients (Table 1).

At the timeof the PETandhigh-resolutionSPECTexaminalions, patients were free from pharmacological treatments knownto alter cerebralperfusion and metabolism.

Allpatientsunderwenta PET studywith ‘8F-FDGand a high

2576

281226

26

48

88

116

211PETandSPFCTin ProbableAlzheimer'sD@eese•Messaat al.

Region PET AD PET N@ SPECT ADSPECTNAssociative

frontal 0.796 ±0.092k 0.962 ±0.024 0.828 ±0.1020.887 ±0.034Meslalfrontal 0.792 ±0.071@ 0.933 ±0.043 0.871 ±0.0920.889 ±0.054Temporal

0.708 ±0.061k 0.830 ±0.057 0.839 ±O.O76@0.910 ±0.034Pailetai0.704 ±0.080k 0.943 ±0.030 0.838 ±0.101@0.960 ±0.040Lateral

occipital 0.826 ±0.116 0.918 ±0.073 0.983 ±0.1091 .026 ±0.046Thalamus0.958 ±0.074 0.951 ±0.054 1.031 ±0.0860.995 ±0.080Cerebellum0.830 ±0.082 0.856 ±0.042 1.198±0.121.154 ±0.075(0)

@ft and right hemispheres ratios are averaged. AD = Alzheimer's disease; N = normalcontrols.*

= p <0.0001.t

= p <0.005.*

= p < 0.001.

TABLE 2PETandHigh-ResolutionSPECTRalioValues(°)in ProbableAlzheimer'sDiseasePatientsandNormalControls

Sixty-fouraxialslices, 1.66mmthick,were reconstructedusinga Harm filter with a cut-off frequency of 0.5 cycles/pixel. Thereconstructed images were corrected for attenuation with theChang's first order method (29) with an attenuation coefficientequal to 0.1 cm'. Four consecutive axial slices were addedtogether in orderto obtain a slice thickness of 6.64 mm, similartothatof thePETimages.

DataAnalysisPET andhigh-resolutionSPECF imageanalysiswas performed

on a SUN (SPARC) workstation. Circular regions of interest(ROIs,diameter= 1.5theFWHMof therespectivesystem)weremanuallypositionedon allthecorticalandsubcorticalstructuresand on the cerebellar hemispheres. Mean rCMRglcvalues (forPET)andmeancounts (forSPECF)were calculatedforeachROIand these values were simply averaged to obtain values of nineanatomo-functional areas identified according to the Damasio andDamasio atlas (30). This atlas is particularly suitable to examine

cortical activity of associative areas (typically damaged in AD)and it has been previously used to evaluate PET and SPECTpatterns in patients with cognitive deficits (31,32). The areas cxamined were the associative frontal, mesial frontal, temporal,parietal, lateraloccipital and calcarinecortices, the thalamus,thebasal ganglia (head of caudate and putamen) and the cerebellarcortex. Since the calcarine cortex and basal ganglia are generallyspared in probableAD (33), mean rCMRglcand meancounts ofthe calcarine cortex and the basal gangliawere taken as the leastaffected areas and averaged to obtain a reference value (Aref) for

normalizationof brain activity. Data were then analyzed in theform of ratios determined as follows:

Ratio = Ar/Aref,

where Ar equals rCMRglcfor PET studies and counts for highresolutionSPECF studies in each anatomo-functionalregion, andAref equals average rCMRglc for PET studies and average countsfor high-resolutionSPECT studies in the referenceareas, i.e., thecalcarine cortex and the basal ganglia.

StaUsticalAnalysisPET and high-resolutionSPECT ratiovalues obtainedin prob

able AD patientswere comparedwith the respective ratiovaluesobtained in normalcontrols.

The groupdifferenceswere evaluatedusing a factorialanalysisof variance (ANOVA) with two between-level factors (probableAD and control subjects) and two within-subject factors (cerebral

regions, i.e., the ratios obtained for the anatomo-functional areasidentifiedandlisted in Table2, andcerebralhemispheres, i.e., leftand right). This analysis was repeated separately for PET andhigh-resolution SPECT.

An individual analysis was also performed considering ratiovalues of rCMRglcand counts in each patientas abnormalwhenbeyond the 99% confidence interval of the control group ratiovalues.

RESULTSThe mean ratio values and standard deviations of PET

and high-resolution SPEC!' regional values of probable ADand normals are reported in Table 2.

For PET studies, the ANOVA showed that both thegroup effect (F = 49.9, df = 1, p < 0.00001) and the regioneffect (F = 21.7, df = 6, p < 0.00001) were significant,whereas the hemisphere effect was not significant. Thegroup x region interaction was significant (F = 12, df = 6,p < 0.00001). This was further analyzed by simple maineffects which showed high significant difference (p <0.0001) for the associative and mesial frontal cortex, theparietal and temporal cortices, whereas the difference was

not significant for the lateral occipital cortex, the thalamusand the cerebellum (Table 2).

For high-resolutionSPECT studies, there was a significant difference between groups (F = 9, df = 1, p < 0.005)and between regions (F = 48, df = 6, p < 0.00001),whereas the difference between sides was not significant(F = 3.2, df = 1, n.s.). The group x region interactionwassignificant (F = 3.2, df = 6, p < 0.005) and the simple maineffects analysis showed significant differences for the panetal (p < 0.001) and the temporal (p < 0.005) cortices(Table 2). A typical PET and high-resolution SPEC1' pattern of bilateral temporo-parietal abnormalities is shown inFigure 1.

The results obtained with the individual analysis arerepresented in Figure 2 where the ratio values of the panetal, temporal and associative frontal contices of the leftand right hemispheres are reported. Metabolism or perfusion were considered abnormal when the regional values(either of the right or left hemisphere or both) were out of

212 The Journalof NuclearMedicine•Vol.35 •No. 2 •February1994

R LR L

f IIIf I

A PET hrSPET

1.2@

1.1

‘2 0.82i 0.71. 0.6

0.5

0.4

B PET hrSPET1.2

@ 1.:

@ 0.9

,@ 0.8

.@ 0.7

8.@ 0.6

0.5

0.4

C1.2

@ 1.1@

@ 1@

,@ 0.9

.@ 0.8

,@ 0.7

I 0.68

@ 0.5

0.4

R L@ R

FiGURE I. Cerebral glucose metabolismusing PET with[18flp@ (toprow)and perfusionusinghigh-resolutionSPECTwith

@rc-HMPAOrespectively(bottomrow)InonepatientwithprobableAD (Patient3, Table 1).Two tomographklevelsparallelto theorbito-meatallinethroughthe midandcranialpartsof the brainarerepresented. Images are displayed with respect to the madmumvalueofeechstudy(PETandSPECT).Bothstudiesshowmarkedreductionof metabolismin the temporaland parietalcortices,moremarkedinthenghtside.Notethepreservationof metabolismandperfusioninthecalcarinecortexandInthebasalganglia,whicharewell depictedInbothstudies.

the mean ±2 s.d. of the control group. In the parietalcortex (Fig. 2A) abnormal values were found in all patientsby PET and in 16/21 patients by high-resolution SPECT'. Ofthe five patients who did not show areas of hypoperfusionin the parietal cortex at high-resolution SPEC!', threeshowed hypoperfusion in the temporal cortex and the remaining two in the frontal cortex. Areas of hypometabolism and hypoperfusion were also found in the temporalcortex (12121with PET and 13,21 with SPECF, Fig. 2B)and in the associative frontal cortex (18/21with PET and11/21with SPECT, Fig. 2C).

The percentageof patientswith PET andhigh-resolutionSPECT abnormalities is shown in Table 3. When temporaland parietalregions of the left and righthemispheres werecombined, all patients showed hypometabolism with PETand 19/21(90%)patients showed hypoperfusionwith highresolution SPECT. The two patients who had normal perfusion in the temporo-parietal cortex (#19 and #20; Table1) showeda selectivefrontaldeficit.PET studiesin thesepatients showed marked abnormalities in the frontal cortex, but also in the parietal lobes (Fig. 3).

As for the presence of interhemispheric asymmetries,unilateral involvement of the temporo-parietal cortex wasshown only in one patient (#9) by both PET and highresolution SPECT, although bilateral abnormalities, moremarkedon one side, were present in several patients. Highresolution SPEC!' unilateralinvolvement of the temporoparietal cortex was found in two patients (#5 and #12) inwhom PET demonstrated bilateral hypometabolism moremarkedon one side.

S

I

IS

PET hrSPET

R L R L

@;f: @j@f

213PET and SPECT in ProbableAlzheimer's Disease•Messa at al.

DISCUSSION

The results of this study demonstrate that areas of significant reduction of glucose metabolism and blood flow

can be detected by PET and high-resolution SPECT, respectively, in the same population of patients with earlyprobableAD. Although the possible relationship betweenPET metabolic changes and SPECF perfusion changes observed in several neurological disorders, including demen

FIGURE2. ComparieonoftherelatWevaluesofglucosemetabolism(PET)andperfusion(high-resolUtiOnSPECI)of eechsingleprobableADpatient(filledsymbols)withthecorrespondingmeanvalue ±2s.d. obtained In normal controls (empty symbols) for thetwohemispheres.R = righthemisphere;L = lefthemisphere.TheratiosbetweenmeanrcMRglc(forPEt) and meancounts(forhigh-resolutionSPECT)inthepadetal(A@,temporal(B)andassodativefrontalcortices(C),andmeanrCMRglcandcountsInthe“reference―areas(averageof calcarinecortexandbasalganglia)axerepresented.

RegionPETSPECTAssociative

frontal86%52%Meslalfrontal62%24%Temporal62%62%Pañetal100%76%Lateral

occipItal33%28%TernpOro/parletai100%90%

TABLE 3Percentof PatientswithAbnormalRegions

In this study, abnormalities in the temporo-parietal cortex were also detected by high-resolutionSPEC!' in a highpercentage of patients (90%), which is in the range ofpreviously published data (from 70% to 100%)(35). Theonly two patients with a prevalent frontal hypoperfusion onSPECT showed cortical hypometabolism more marked inthe frontal areas, but diffuse also on the temporo-parietalareas on PET. In these two cases, Pick's disease or frontaltype dementia cannot be excluded, and the differentialdiagnosis between Pick's andAD is based only on autopsyfindings.

The high sensitivity observed in this report might berelated to the availability ofhigh spatial resolution PET andSPECT systems and to the use of semiquantitative analysis, such as the ratio analysis reported here.

The anatomical structure most frequently used for thesemiquantitative ratio analysis of SPEC!' studies is thecerebellum (3—5)In this study, the reference area used inthe ratio was the average activity of the basal ganglia andthe calcarine cortex. We did not choose the cerebellum asa reference area because of an apparent uncoupling between perfusion and glucose metabolism in the cerebellarhemispheres. In fact, an overestimation of perfusion ohtamed with @Tc-HMPAOin the cerebellum comparedwith blood flow estimatedwith ‘8F-fluoromethaneandPEThas been previously reported and it has been attributedtothe high capillaritydensity in the cerebellum (36). Furthermore, a cerebralblood flow (CBF) and cerebral metabolicrate for oxygen (rCMRO@)higherthanglucose metabolismin normal cerebellum measured with PET has also beenreported by Sasaki et al. (37). Although a comparativeanalysis between PET and high-resolution SPECT valuesin normal regions was not performed because out of thepurposes of the present work, an indication of higher values of blood flow estimated with @‘@‘Fc-HMPAOin the

r@i cerebellumcanbeinferredbythedatashowninTable2where the ratio value of cerebellum m controls measuredwith high-resolutionSPECF was 26%higher than the corresponding value measured with PET

@@:i Thecalcarinecortexandbasalgangliawerechosenbecause they are usually spared in AD (33) and because the

L use of these structures m the ratios has been recently testedin a PETIFDG study of AD and nonAD patients (38,39)and its diagnostic power has been found to be high (sensitivity of 92% and specificity of 80%)

A discrepancybetween PET andhigh-resolutionSPECFfindmgs was observed m the detection of abnormalities mthe frontal associative cortex. Whereas in the group analysis, PET metabolic values showed that both the lateralandmesial frontalcortex was significantlyaffected in probable AD, high-resolutionSPED.' regionalvariations in thefrontal cortex, although present in some ofthe patients, didnot reach a statistical significancein the group analysis Onthe other hand, in the parietal cortex, abnormalities were

detected by high-resolution SPECF in a smaller number ofpatients than by PET (Table 3). Although the defect in thetemporal and the parietal cortex is severe and constant

tia, has been raised (13,34), the present study reports asystematic comparison of the same series of patients withprobable AD examined with PET and high-resolutionSPECT techniques.

Significant differences between probable AD and controls were found in the parietal and temporal cortices bothwith PET and high-resolution SPECT. The finding of abnormalities in the temporo-parietalcortex, present in allpatients by PET, demonstrates the abilityof this techniqueto detect a metabolic pattern which is characteristic ofprobable AD, consistent with previous reports (14—16).Recently, the sensitivity of PET with ‘5O-waterto detectbilateral abnormalities in the temporo-pa.rietal cortex hasbeen reported to be 38% by Powers et al. (21) in a population of probable AD patients similar or even more compromised than the one in the present report. However, inthe study by Powers et aL, the spatial resolution of theimages (18mm FWHM) was lower in comparison with thatof the present PET (63 mm FWHM) and high-resolutionSPEC!' (8.4 mm FWHM) studies because offactors relatedto the tracer and PET technology. Furthermore, resultsand conclusions were drawn only from visual analysis.

(18F]@G PET

R

214 The Journal of Nudear Medicine•Vol. 35 •No. 2 •February 1994

@@@Tc]HM—PAOhrSPET

FIGURE3. ReglonalglucosemetabolismobtainedwithPETand[18fl@ (toprow)andperfusionobtainedwithhigh-resolutionSPEcT ani @c1HMPAO(bottomrow) In Patient19 (Table1).Twotomographiclevelsparalleltothe orbito-meatalIhiethroughthemid and cranial parts of the brain are represented.Imagesaredisplayedwithrespectto themwdmumvalueof eachstudy(PETand SPECT).Both studiesshow hypometabolismand hypoperfusionbilaterally@mthefrontalcoitex.PETaISOshowshypometabolismof the pailetalcortexbilaterally.

study of regionalcerebral perfusionin early Alzheimer'sdisease.I NuciMed 198829:1507—1514.

5. BurnsA, PhilpotMP, Costa DC, EUPJ, Levy R. The inves@gationofAlzheimer's disease with single phOtOnemission tomography.I News!Neumsur and Ps@h 1989,52:248-253.

6. GemmelHG,SharpPF,SmithFW,Bessoni,EbmeierK, DavidsonJ,etal.CerebralbloodflowmcasuredbySPECT as a diagnostictool in the studyofdementia.P.sychiabyRes 198929:327—329.

7. BonteFJ, HornJ, TmtnerR, WeinerMF. SinglephotontomographyinAlzheüner'sdiseaseand the dementias.SemEnNuc Med 1990;20:342-352.

8. Hobnan BL, Johnson KA, Gerada B, Carvalbo PA, Satlin A. The scintigraphicappearanceofAlzheimer'sdisease:a prospectivestudy usingtechnetium-99m-FIMPAO SPECF. INuciMed 1992;33:181-185.

9. ElI P3.Mappingcerebralbloodflow[Editorial.INuciMed 1992;33:1843-1845.

10.KourisK,JarriuPH,CostaDC,EliPJ.PhysicalassessmentoftheGE/CORneurocam and comparison with a single rotatinggamma camera. EuriNuciMed1992;19:236—242.

11. Lucignani0, RossettiC, FerrarioP, et al In vivo metabolismand kineticsof@Fc-FDvWAO.EurlNuciMed 1990;16:249-255.

12.SorensonJA, PhelpsME.Nuclearmedicinetomography:systemsanddevices. In: SorensonJA, Phelps ME, eds. Physics in nuckar medicine,second edition. Orlando, Florida: Grime & Stratton, Inc.; 1987:424-451.

13. Jagust W, Eberling J. MRI, CF. SPECI@,PET: their use in diagnosingdementia.Geiiatiic.c1991;46:28—35.

14. Alavi A, Dann R, Chawluk 3, Alavi J, Kushner M, Reivich M. Positronemission tomographyimagingof regional cerebral glucose metabolism.Semin NUClMed 1986;16:2—34.

15. Haxby iv, Grady CL, Duara R, Schlageter N, Berg 0, Rapoport SI.NeocorticalmetabolicabnormalitiesprecedenonmemorycognitivedefectsinearlyAlzheimer'stypedementia.AivhNews! 1986;43:882-885.

16. Kuhl DE, Small GW, Riege WH, Ct at. Cerebral metabolic patterns beforethe diagnosisof probableAlzheimer'sdisease.I CerebBloodFlowMetab1987;7:5406.

17. Mielke R, Herholz K, Kessler 3, Helm D. Differences of reglonal cerebralglucosemetabolismbetween presenileand senile dementiaof Alzheimertype. Neurot@oIogjofAging 1991;1393-98.

18. Haxby iv, Duara R, Grady CL, Cutler NR, Rapoport SI. Relations betweenneuropsychologicalandcerebralmetabolicasymmetriesinearlyAlzheimer's disease.I CerebB1o4 FlowMetal, 1985;5:193-200.

19. McGeer PL, Kamo H, Harrop R, Ctal. Positron emission tomography inpatients with clinicallydiagnosedAlzheimer's disease Can Med Ass I1986;134-.597-607.

20. LoewensteinD,YoshiiF,BarkerWW,Ctal.Predominantlefthemisphericmetabolicdeficitpredicts early manifestationof dementia.I Cei@bBloodFlow Metal, 1987;7:5416.

21. PowersWJ,PerlinutterJS,VideenTO,Ctal.Blindedclinicalevaluationofpositron emissiontomographyfor diagnosisof probableAlzheimer'sdisease. Newvlog@'199242:765—770.

22. McKhannG, DrachmanD, FoisteinM, et al. Clinicaldiagnosisof Alzheiiner's disease: report of the NINCDS-ADRDAWork Group under theauspiciesof Depaitment of Health and Human Services Task Force onAlzheimer'sdisease.Neurology1984;34:939—944.

23. SpinksTJ, Jones 1, GilardiMC, Heather JD. Physicalperformanceof thelatest generationof commercialpositron scanner. IEEE Tmns Nuci Sd1%835:fll-fl5.

24. BettinardiV, ScardaoniR, GilardiMC,Ctal. Head bolderfor PET, CT andMR.!studies.I ComputAssLctTornogr1991;15:886-892.

25. HamacherK, Cohenen FIN, Stocklin 0. EffIcient stereospecific synthesis ofno-carrier-added 24l8FJfluoro-2.deoxy-D.gluooseusing aminopo@yethersupportednucleophilicsubstitution.INUC!Med 1986;27:235-238.

26. SOkOIOffL,ReivichM, KennedyC, et al. The ‘@Cdeoxygjucosemethodforthemeasurementoflocalcerebralgjucoseutilization:theory,procedureandnormal @raIiiesin the consciousand anesthetizedalbinorat. I Newvxhe,n197728:897—916.

27. Reivich M, Alavi A, Wolf A, et al. Glucose metabolic rate kinetic modelparameter determinationin humans: the lumpedconstants and rate conslants for I1Flfluorodeoxyglucoseand (‘1qdeoxygluco@e.I Cereb BloodFlow Metab 1985;5:179—192.

28. GennaS, SmithAP.ThedevelopmentofASPECr, anannularsingleciystalbraincamerafor highefficiencySPECF.IEEE TransNuclSci 1988;NS-35:654—658.

29. Qiang LT. A methodfor attenuationcorrectionin radionucide computedtomography.IEEE TransNuc!Sd 1978;NS-25:638-643.

215

enough to reach a statistical significance with both techniques, the defect in the frontal cortex is less frequent anddoes not reach a statistical significance with high-resolutionSPECF. This finding of a slightly lower sensitivity of highresolution SPECT compared to PET can be caused byseveral factors, including physical limitations, such aslower resolution of high-resolution SPECF compared toPET, andbiological constraintsdue to the differentkineticsof the tracers used (12,13).

PET images were obtained with 18F-FDG, a tracerwhose metabolic fate and kinetics are well known Thus,PET analysis can be done based on data of glucose consumption and the information obtained is directly repre

sentative of regional functional activity. For high-resolution SPEC!' studies, we used @‘@Tc-HMPAO,a tracerthatis known to be distributed in the brain in proportion tocerebral blood flow. Although models to quantify bloodflow with HMPAO have been proposed (40—42),they require arterial blood samples and metabolite assays, resulting in a limitation for wide clinical application of SPECFstudies of demented patients.

In spite of these discrepancies, our findings of similarpatterns of glucose metabolism and perfusion in the sameprobableAD patients seem to indicate that a tightrelationship between blood flow and glucose metabolism is maintamed in this disorder, which is consistent with the coupling between blood flow and oxygen metabolism

previously demonstrated with PET (43).

CONCLUSIONThe results of this study confirmthat: (1) both PET and

high-resolution SPECF identify the characteristic temporoparietal pattern in the same series of probable AD patients,with PET being only marginally more accurate thanSPEC!'; and (2) other abnormalities such as frontal involvement are better evaluated with PET. The high resolution of the systems used as well as the use of eitherquantitative or semiquantitative analysis of the emissiontomography data may contribute to the high sensitivity ofthe methods. Further development of methods for bloodflow quantificationmay improvethe diagnosticaccuracyofhigh-resolution SPECF.

ACKNOWLEDGMENTSTheauthorsthankMr.GiuseppeStrianoof theDepartmentof

Nuclear Medicine, H. San Raffaelefor expert technical assistance. This work was supported by a National Research Councilgrant INV 93-2-365.

REFERENCES1. Hohnan BL, Devous MD. Functional brain SPECF: the emergence of a

powerfulclinicalmethod JNuclMed 199233:1888-1904.2. JohnsonI(A, MuellerST,WalscbeTM, EnglishRi, HolmanBL Cerebral

perfusionimagingin Alzheimer'sdisease: use of sinaje photon emissioncomputed tomographyand iofetaminehydrOChlOride1-123.Arch Neurol1987;44:165-168.

3. Jagust AJ, Budinger iT, Reed BR. The diagnosis of dementia with singlephotonemissioncomputedtomography.Arch Neumi 1987;44:258-262.

4. PeraniD, Di PieroV, Vallar0, Ctal.Technetium-99mHMPAO-SPECF

PETandSPECTinProbableAlzheirner'sDiaease•Masseatal.

30. DamasioH,DamasioAR.L,sionan4,sis inneumps@tholo@j@.NewYork:Oxford University Press, 1989.

31. Cappa SF, Perani D, Bressi 5, Paulesu E, Franceschi M, Fazio F. Crossedaphasia:a PET follow-upstudy of two cases. JNeumlNeumswg Psychiaby 1993;56:665.-671.

32. PantanoP, DiP1cmV, RicciM,FieschiC, BozzaoL, LenziGL Motorstimulation response by technetium-99m-hexamethylpropyleneamineoxime split-dose method and single photon emission tomography. Eur INuci Med 1992;19:939-945.

33. BrunA, EnglundE. Regionalpatternof degenerationin Alzheimer'sdisease: neuronal loss and histopathological grading. HistOpathOlOgy 1981;5:549—564.

34. Maurer MI. Nuclear medicine: SPECIEcomparison to PET. Radio! ClinNorth Am 1988;26:1059-1074.

35. Dewan M, Gupta S. Toward a definite diagnosis of Alzheimer's disease.Comprehensive Psychiatiy 1992;33:282-290.

36. Heiss WD, Herholz K, POdrekaI, Neubauer I, Pietrzyk U. Comparison of@Tc-HMPAOSPECT with 18F-fluoromethane PET in cerebrovascular

disease. I Cereb Blood Flow Metal, 1990;1O:687-697.37. Sasaki H, Kanno I, Murakami M, Shishido F, Uemura K. Tomographic

mapping of kinetic rate constants in the fluorodeoxyglucose model using

dynamic @tronemissiontomography.I CerebBloodFlowMetab 1986;6:447—454.

38. Herholz K, Adams R, Kessler J, Szeiers B, Grond M, Heiss WD. Criteriafor the diagnosis of Alzheimer's disease with positron emission tomography. Dementia 1990;1:156—164.

39. HemltzK,PeraniD,SalmonE,FranckG,FazioF,HeissWD,ComarD.ComparabilityofFDGPETstudies inprobableAlzheimer'sdisease.INuclMed 1993;34:1460-1466.

40. Lassen NA, Andersen AR, Friberg L, Paulson GB. The retention of@'Tcd,l-HMPAOin the humanbrainafter intracarotidbolus injection:a kineticanalysis.I CenthBloodFlowMetab 1988;8:S13-S22.

41. Matzuda H, Oba H, Seki H, Ctal. Determination offlow rate constants in akinetic model of @‘Tc-hexamethyl-propileneamine oxime in the humanbrain.I CerebBloodFlowMetab 1988;8:S61-S68.

42. Pupi A, De Cristofaro MTR, BacciOttini L, Ct al. An analysis of the arterialinput curve for technetium-99m-HMPAO: quantification of rCBF usingsingle-photon emission computed tomography. INucI Med 1991;32:1501-1506.

43. FrackowiackPSi, PozzilliC, LeggNNJ,et al.Regionalcerebraloxygensupplyand utilizationin dementia:a clinicaland physiologicalstudy withoxygen-15 and positron tomography. Brain 1981;104:753-778.

216 TheJournalofNudearMedicine•Vol.35•No.2 •February1994