Simplified PET Quantitation of Myocardial Glucose Utilization Keiichi Nakagawa, Hiroki Namba, Masaomi Iyo, Kiyoshi Fukushi, Toshiaki Irie, Masato Yamanouchi, Nobuaki Shikama, Toshiharu Himi, Katsuya Yoshida and Yoshiaki Masuda Third Department of Internal Medicine, Chiba University; Division of Clinical Sciences, National Institute of Radiological Sciences, Chiba, Japan; Department of Neurosurgeiy, Chiba Cancer Center Hospital, Chiba, Japan; and National Institute of Mental Health, National Center of Neurology and Psychiatiy, Chiba, Japan
The purpose of this study was to validate experimentally a
simple method to quantify tissue glucose utilizationwith the braln referenceindex (BRI)using 14C-deoxyglucoseand assess
its clinicalfeasibilityfor myocardialPET. Methods To validate the GAl method, glucose utilization in myocardial and skeletal
muscle was studied inrats with14C-deoxyglucoseafter increas ing doses of oral glucose loading.To assessclinicalfeasibilityof the method, the BRI was applied to nine patients undergoing myocardial PET and compared to rMGU measured by the de
glucose but several other substrates, such as free fatty acids, lactate and ketone bodies, for its energy source (9). Con sequently, myocardial utilization of these substrates de
pends on systemic metabolic conditions (10), which there fore affect clinical utilization of cardiac PET. In the fasting state, free fatty acids are major substrates for cardiac energy metabolism and glucose metabolism is reduced (11,12). After oral glucose loading, glucose is a major substrate and oxidation of free fatty acids is inhibited
oxyglucose model of Sokoloff at al. and by Patlak graphical
(10). Therefore, FDG-PET myocardial images obtained in
analyals. The normal range of myocardial FDG uptake ex pressed as the BRI was estimated with four normal volunteers. Results: In skeletal muscle, a dose-dependent increaseof glu cose utilizatkn was observed during oral glucose loading with
the fasted state are often difficult to analyze because myo cardial glucose uptake is low (13—15). Moreover, significant heterogeneity in regional myocardial glucose utilization in
the fasted state has been reported (13). After oral glucose
doses up to 4 mg/g. In the myocardium, glucose utilization loading (16,17), or with the insulin clamp method (17,18), increasedwith a glucose loading dose of up to 1 mgfg without increasingfurther at greater glucose doses. Ratios of maximal glucose utilization in glucose-loaded rats to 19-hr fasted rats
myocardial glucose uptake is increased and homogenous
throughout the heart, thereby providing high quality myo
(controls), expressed as the BRI for left and right ventricular
cardial PET images. Therefore, PET imaging of myocardial
rMGU; four of these patients had a constant plasma glucose
deoxyglucose model of Sokoloff et al. (19) and Patlak graphical analysis (20,21), these methods require constant plasma glucose concentrations during data acquisition and measurement of the arterial input function. Study subjects
myocardium and skeletal muscle were 4.16, 3.74 and 7.39, glucose uptake is now most commonly done after glucose respectively.Glucose utilizationof rightventricularmyocardium loading or the insulin clamp (16—18). was approximately 70% of left ventricular myocardium for all Although regional myocardial glucose utilization rates glucose-loadedconditions. For patients,the BRIcorrelatedwith have been quantified noninvasively in humans using the concentration. Conclusion: MyocardialBRIis a sensitive mdi cator of rMGUthat does not requiredynamicdata acquisitionor constant plasma glucose concentrations. Key Words myocardium; oral glucose kad; positron emisalon
frequently show varying plasma glucose concentration after
tomography; carbon-14-deoxyglucose;fluorine-i8-FDG
oral glucose, thereby failing to satisfy the requirement of
steady-state glucose concentration necessary for applica tion of these models. Euglycemic hyperinsulinemic clamp establishes constant blood glucose levels for standardizing metabolic conditions and stimulating myocardial glucose utilization rates (22), to measure tissue glucose utilization rates in vivo and to but this method is too complex for routine clinical studies assess myocardial viability in ischemic heart disease (1—4) and is not a physiological condition seen in daily life. As carried out clinically without the insulin clamp method, or abnormal myocardial glucose metabolism in cardiomy opathy (5—8).The myocardium, however, uses not only there are several problems with quantifying glucose utiliza
J Nuci Med 1995;36@94-21O2
PT with [18F]fluorodeoxyglucose (FDG) has been used
tion as it is usually measured. In many normal subjects, glucose levels are relatively constant 60 mm after oral
glucose loading and remains almost unchanged. At that ReceiVed Nov.14,1994;revisionacceptedJun.6, 1995. Forcorrespondence orreprints cont@t: K@N Nakagawa, MD,D@sionof time, however, when myocardial PET is usually performed, cardk@ogy, TheUnivers@y ofTexasMed@School atHouston, MSB4.258,6431 Fannin, Houston, TX 77030.
2094
varying glucose levels are common, especially in patients
The Journal of Nuclear Medicine • Vol. 36 • No. 11 • November 1995
Transmission Transmission Scanfor Heart Scanfor Brain (6 mm) (6 mm)
@
PETacquisitIonfor Heart (63mm)
jo 759 glucose p.o.
I
BloodSampling 1
BloodSampling2
PETacquisition for Brain (6 mm)
110 BloodSampling3
FIGURE 1. Studyprotocolof [18F]FDG PETinsix patientswho had both dynamk@ myocard@and bralndata acquisition. with diabetes mellitus, which is associated with ischemic
cose, the rats were killed by decapitation and the cerebral cortex,
heart disease. All tissue uptake of labeled glucose analogs increases or decreases in direct proportion to plasma glu coseconcentration.In addition, myocardialglucoseutiliza tion is largely accelerated by insulin and affected by some
cerebellum,myocardiumand skeletal muscle (iliopsoasmuscle) were removed and weighed.These tissues were dissolvedwith
other factors such as fatty acids or catechols. Therefore, the
tissue solvent and the radioactivities were counted with a liquid scintillation counter. Plasma glucose concentrations were mea sured by Glucose Analyzer 2 (Beckman Instruments), plasma insulin by radioimmunoassay and free fatty acids by the enzymatic
clinicalinterpretation of absolute myocardial glucose con method. sumption is questionable. On the other hand, visual interpretation of standard myocardial uptake images of FDG after oral glucose load
ing may be misleading and are not quantifiable without accounting for arterial input functions and plasma glucose concentration and, potentially, its variation after oral glu coseloading.A simplemethod of quantifyingrelative FDG
Human Study
Subjects. The feasibility of the method was assessed in 10 sub jects. Nine patients had previously diagnosed heart disease and
fourwerenormalvolunteers.To showthe feasibilityof the method over the spectrumof cardiacdisease;two patients had hypertro
uptake that accounts for arterial plasma glucose concentra
phiccardiomyopathy, two had dilatedcardiomyopathy, one had aortic stenosis, two had previous myocardial infarction, one had
tion and its variation but does not require arterial blood
hypertensive cardiac disease and the other had nonspecific chest
sampling and the insulin clamp would be clinically useful.
pain with normal coronaiy angiogramresults. No subject had a
Based on the operational equation derived by Sokoloff et
al. (19), we have previously described the basis for a method (23) to measure organ tissue glucose utilization relative to the cerebellum, which maintains constant glu
cose utilization over a wide range of metabolic conditions and stimuli: therefore, it is termed the brain reference index (BR!). The aims of the present study were to validate experimentally
the BR! method
for quantifying
myocardial
diagnosis of diabetes mellitus. Written informed consent was ob
tamed from each subject.
The patientsfastedat least5 hr after a low-fatbreakfastand the four normalvolunteersfastedovernight(at least 15hr) before the PETstudy.On arrivalin the PETlaboratory,all 10subjectswere given a 75-g oral glucose load 60 mm before FDG administration. Plasma insulin and free fatty acid concentrations were measured at the time of FDG injection,at 30-mm postinjectionand at the end of dynamicdata acquisition.
glucose utilization after fasting and oral glucose loading in rats using ‘4C-deoxyglucoseand to demonstrate the con cept and clinical feasibility of the method in myocardial
FDGPreparation.FDGwassynthesizedby the acethylhypofluo rite method with a “CYPRIS― small cyclotron,and a “CUPID― automatictracer synthesizer(SumitomoHeavyIndustries,Tokyo,
[‘8F]FDG PET in nine human subjects to assess its poten
Japan). The radiochemicalpurity of FDG was more than 95%.
tial clinical usefulness.
The preparationwastested in accordancewiththe standardof the cyclotron committee of Chiba University Hospital.
Study Pmtocol for the BR! Method and Patlak's
[email protected]
MATERIALS AND METhODS
scanningwas performed using a whole-body,multislicepositron
Rat Study
emission tomograph (HEADTOME
Male wistarrats (300—410 g, 11wk old) fasted for over 19 hr before the study with only water provided ad libitum. Thirty rats were randomly divided into five groups (six rats per group). Four groups of rats were orally administered 0.5, 1, 2 and 4 mg of glucose per gram of body weight in a 50% glucose solution.
The remaining six rats were administered 1 ml of physiological
III, Shimazu, Kyoto, Japan).
This PET scanner can acquire three slices simultaneously with a slice thickness of 16.5 mm and spatial resolution of 10.4 mm FWHM. The study protocol is shown in Figure 1. In each study, after oral
administrationof 75g glucose,a transmissionscanof the brainwas obtained for 6 mm using @Ge ring to measure attenuation cor rection.Thispositionand that of the lightbeam of the tomograph
saline as controls. Thirty minutes after administration of the solu tions, a 27-gauge catheter was inserted into a tail vein and 1 ml of
was marked on the skin with a felt pen to relocate the patient at
blood was withdrawn through the catheter to measure plasma
the samepositionfor emissionscanningcarriedout later.A trans
glucose concentrations,
mission scan of the heart was then obtained for 6 mm using the same ring. Sixty minutes after oral administration of glucose, dynamic PET acquisition of heart was started with intravenous
insulin and free fatty acids. Immediately
after blood sampling 15 mCi/kg ‘4C-deoxyglucose (specific activity: 50—55mCi/mmole)
were injected
through
the same catheter
which
was then removed. Thirty minutes after injection of ‘4C-deoxyglu administration of FDG (4 mCi, 148 MBq) and images were ob
Simplified Quantificationof Myocardial FOG-PET• Nakagawaat al.
2095
tamed for 63 mm at 3 frames of 1 mm duration followedby 5
where C(t) is the decay-correctedmyocardialactivity,Cb(t)is the
frames of 2 mm duration, 11 frames of 4 mm duration and 1 frame
decay-corrected blood-pool activity at any given time t, f0 C(s) ds
of 6 mm duration. Immediatelyafter dynamicPET acquisition, is the integral of decay-corrected blood-pool activity from time each subject was carefully repositioned on the PET camera in the zero to time t, which is an index of arterial input function of same position used for the brain transmission scan according to the skin marks and the light beam. PET brain images were ob
[‘8FIFDGin the myocardium. This relation becomes linear at later
times
tamedfor 6 mm.Each PET imagewascorrectedfor deadtimeand
with slope = K = (k1 x k2)/(k2+ k3)
physical decay of FDG.
Patlak Analysis ofNormal Volunteers.The rMGU of the septal, anterior and lateral walls in the midventricular section on the
when the dephosphorylationrate constant (k4)is assumedto be zero and k1_@are the rate constantsof FDG in a three-compart transverseimageswas calculatedusing Patlak graphicalanalysis. ment model. Myocardialglucose utilization rate (rMGU) was In each volunteer(n = 4), the averageddata of three regionsof interest (ROIs) drawn in each segment were used to determine the
calculated as follows:
rMGU.
rMGU=C@,XK/LC,
MRJ. In the normal volunteers, MR images of the heart were obtained to measure left ventricular wall thickness in the trans verse view,which is the same view in the PET image. Regional left ventricular wall thicknesses were measured on the end-diastolic
where C@is a plasma glucose concentration, K is a rate constant of FDG calculated from the slope on the Patlak plot and LC is the
were used for partial volume effect correction of the PET emission
at 0.67 (29). Myocardial BR! (BRIm) was calculated by:
lumpedconstant.Plasmaglucoseconcentrationat the beginningof MRI imagesand its correspondingrecoverycoefficientvalues the dynamicacquisitionwasused and the lumpedconstantwasset
data. BRIm
Data A@ Cakulation
of Glucose Utilization Rate in Tissues: Rat Study.
where Cm(T) @5 regional myocardial activity during the last 6 mm
Namba et al. (23). BRI is a ratio of ‘4C concentration of tissue to
in the dynamicacquisitionand C@@(T) is cerebellaractivityduring the 6 mm of brain data acquisition.
that of the brain. For example,myocardialBR! (BRIm)is:
S@caI
Glucoseutilizationin tissuewasexpressedas the BRI describedby
BRIm
@
Cm(T)/Cce(T),
C@m
LC,,,
@.= GUm X ,@ T
‘-“b
Ana@
Data are expressedas mean ±s.d. or s.e. One-wayanalysisof variancefollowedby a ScheffeF-test was used for comparisons
I r' = GUm X Constant,
‘@@Ub X @-“--b
among groups. A p value < 0.05 was the minimal level of signifi
where CImand Ci,, are ‘4C concentrations of the myocardium and brain, GUm and GUb glucose utilization rates of the myocar dium and brain and LCmand LC@, are the lumped constants of the myocardium and brain.
The rate of cerebral and, particularly, cerebellar glucose utili
cance.
RESULTS Rat Study Plasma glucose, insulin and free fatty acids concentra tions in blood 30 mm after oral glucose administration are
zation,quantitativelymeasuredbythe methodof Sokoloffet al. is not affectedby plasmaglucoseor insulinconcentrations(24,25) except under conditionsof profound hypoglycemia(26). There shown in Table 1. As loading doses of oral glucose in fore, BR! has a valuethat reflectsglucoseutilizationof the tissue creased, plasma glucose and insulin concentrations in relative to the cerebellum, including effects of arterial input func tion, without requiring steady-state plasma glucose or insulin con
centration and arterial input measurements.The cerebellum is
creased and free fatty acids concentrations decreased. Figure 2A shows myocardial and blood activity (% dose/g of tissue) after various loading doses of glucose. In the
used as the reference region because it has been shown to be constant and relatively insensitive to sensory inputs such as noises
control rats, myocardial and blood activity were similar. As
in the ordinarylaboratotyenvironment(27).
creased, whereas myocardial activity increased, thereby
Quantitative Analysis: Human Study. In each patient, the mid
ventricularslicewas selectedfor quantitativeanalysisfrom three transverse slices. Thirteen 6 x 6 mm2 ROIs were drawn in the left ventricular wall on the selected slice, incorporating all of the
myocardiumin the slice.A square ROt
loading doses of oral glucose increased, blood activity de causing an increase of the myocardium-to-blood 2B).
ratio (Fig.
Deoxyglucose uptake for different tissues expressed as the percent dose per gram of tissue are shown in Figure 3. 18 x 18mmwasdrawn
in the blood pool of the left atrium, which was used for the arterial
As the oral glucose dose increased, deoxyglucose uptake of
input function. The SET-120W computer system (Shimazu) was
the cerebral cortex and the cerebellum decreased, when
used to perform data analysis.
uptake is expressed as the percent dose of injected tracer. Deoxyglucose uptake in the left and right ventricular myo cardium increased with increasing glucose loading doses of
Quantification ofMyocardial
Glucose Utilization: Human Study.
In each subject, Patlak graphic analysis was performed to calculate
regionalmyocardialglucoseutilizationusingthe serial ‘8F activi ties in myocardial and blood-pool ROIs (21,28). The Patlak plot describes the relation between
C(t)/Cb(t) and f C(s)ds/C@@,(t), Jo
2096
up to 1 mg/g, which reflects increased utilization with greater substrate availability. At loading doses greater than 1 mglg glucose, there was no further increase in myocardial
uptake and uptake per unit loading dose fell. Deoxyglucose uptake in skeletal muscle increased as the oral glucose dose increased up to the highest dose of 4 mglg. For all glucose
The Journal of Nuclear Medicine • Vol. 36 • No. 11 • November1995
TABLE I
Plasma Glucose, Insulinand Free FattyAcids (FFA)Concentrations in Rats loading0
Dose of oralglucose
mg/g0.5
mg/gGlucose
mgfg1
19.5 ±16.0
Insulin 0.97mean FFA1
mg/g2
±28.5 4.5 ±0.5
5.2±1.9 1.52157.8
mg/g4
.3 ±51.2
±22.0
1.02 ±0.08167.2
7.9 ±2.9
11.0 ±9.4
1.0 ±0.06181
0.85 ±0.01184.2
±46.4
7.6 ±3.2
±s.d. mg/dl (glucose),rig/mIfinsulin),and mEq/liter(FFA).
when rMGU calculation by Patlak's analysis is performed
loading doses, brain uptake of radiolabeled deoxyglucose was higher than muscle. The BR! in the cerebral cortex (BRIc@), left ventricular myocardium (BRILV), right ventricular myocardium
using plasma glucose concentrations
(BRIRV)and skeletal muscle (BRIs@J,usingthe cerebellum as a reference, are shown in Figure 4. The BRI@@was similar for all glucose loading doses. The BRILV increased incrementally up to an oral loading dose of 1 mglg glucose
at the time of FDG injection (60 mm), in the middle of the
and then plateaued. There were no significant differences in myocardialBR! for 1, 2 and 4 mg/gloadingdosesof oral glucose. The response of BRIRV to the loading dose of oral glucose was similar to that of BRILV, although BRIRV were
approximately 70%of BRILVin eachgroup.The BRISK increased as the loading dose of glucose increased, reaching
maximum uptake after 4 mg/g glucose loading. Human Study
Biochemical Data. Table 2 shows the biochemical data for each patient during FDG injection (60 mm), in the middle of the dynamic PET acquisition (90 mm) and at the end of the dynamic PET acquisition (120 mm). Patients 1,
3, 5 and 6 showed relatively stable plasma glucose concen
dynamic PET acquisition (90 mm) and at the end of the dynamic PET acquisition (120 mm). No volunteer had di abetes mellitus or impaired glucose tolerance. Correlation between BR! and rMGU. All nine patients had myocardial dynamic and static brain imaging. Myocardial
BR! (BRIm), using the cerebellum as a reference, and rMGU, measured by Patlak graphical analysis, were calcu lated in 13 ROIs of each midventricular slice of each pa tient (Fig. 5A). The correlation between BR! and rMGU of nine subjects was r = 0.85 (Fig. 5B). When we used the data
of Patients 1, 3, 5 and 6, who had relatively constant plasma glucose concentrations (s.d./average < 10%), the correla tion between BR! and rMGU was 0.98 (Fig. 6). For these
four patients, the relation between BR! and rMGU was
trations, in which the standard deviation of three measure ments for each individual was less than 10% of the average.
derived by singlecurve fitting as follows:
Five patients (Patients 2, 4, 7, 8, 9) had varying plasma glucose concentrations
at the beginning of
PET imaging, whereas the BR! is not affected by these variations. Based on glucose tolerance testing with 75 g oral glucose loading, Patients 1 and 7 had impaired glucose tolerance and Patients 3, 6 and 9 had diabetes mellitus (30). Table 3 shows the biochemical data of normal volunteers
BR! = (5.37 x 102) x rMGU + 0.048.
with a s.d. that was more than 10%
of the average value for each individual. These five patients
Eq. 1
Normal Range of Myocardial BR!. The wall thickness
illustrate thatvariableplasma glucose concentrations after values measured on the end-diastolic MR images ranged glucose loading with a s.d. of plasma glucose concentration is common. Varying plasma glucose concentrations may be associated with uncertainty in calculated absolute myocar dial glucose utilization rates that are difficult to interpret
from 6 to 9 mm (average 6.9 mm), which corresponded to the recovery coefficient from 0.50 to 0.69 (average 0.56). In the 12 regions from the normal volunteers, averaged rMGU in all regions was 0.73 ±0.15 @mole/min/g and the
A0.4—0-H@B7I •
a0.3—•—bbod26
50.?ae0.2@4
•>00.1go
@
2 [
[email protected] Oral Gluco1Se 010.01
Dos
e [mg/gJ
4
control
0.5 Oral
4 ucoe Do...2
Simplified Quantification of Myocardial FDG-PET • Nakagawa et al.
FiGURE 2. @A) Myocardialandbloodactiv ities after oral glucose loading, expressed as the percent dose per gram oftissue. (0) Myo cardial activityand (•) blood actMty. (B)Ratio
of myocard@-to-bloodactivityafteroralglu cose loading. Values are mean ±s.e.m.
2097
0@@.@
data collection to correct for plasma glucose concentration effects on tissue deoxyglucose uptake. Steady state is achieved in principle only by the insulin clamp method. In the present study, deoxyglucose uptake of the cerebral
U Omg/g
a 0.5m@/g . lmg/g
0.8
cortex, expressed as the percent dose per gram of tissue,
0 2rnglg 0 4mg/g
0
decreased
0.6 U
2w o@ii) >,
0
vui@In@ FIGURE3. Effectsof graded doses of oral glucose loadingon deoxyglucose (DG)uptake intissues expressed as the percent dose per gram of tissue. Values are mean ±s.d. of six animals for each dose. LV = left ventricular myocardium, RV = right ventricular myocardium,SK = skeletalmuscle,CX = cerebralcortex.
differences between these regions were not statistically sig nificant (Table 4). When these data were applied to the above equation, the normal range of myocardial BR! was calculated to be 3.93 ±0.84.
and plasma glu
cerebellum, it accounts for arterial input function effects and varying plasma glucose concentrations (23), which therefore do not have to be directly measured. Conse quently, tissue glucose uptake can be semiquantitatively determined even after oral glucose loading and variable plasma glucose concentrations. Thus, the BR! method is, in principle, applicable not only to animal experiments, as in this study, but also to human FDG-PET studies as a simple method that does not require arterial input function mea surement and serial dynamic images.
In this study, myocardial activities after deoxyglucose were higher than those of the blood in all glucose loaded conditions. This is consistent with the fact that PET cardiac images
DISCUSSION Carbon-14-Deoxyglucose
Study
The methods of Sokoloff et al. (19 ) and Patlak graphical (20,21 ) to calculate
glucose
utilization
rates after
glucose loading require stable glucose concentration
during
after FDG
administration
are markedly
improved
with high myocardium-to-blood ratios in the glucose loaded state. Our results show that the BRI@@was constant at all glucose
analysis
doses increased
plasma glucose concentrations. Since the BR! method pro vides quantitation of tissue deoxyglucose uptake relative to
0.4
0.2
@
as oral glucose
cose concentrations increased (Fig. 3) because plasma-spe cific activity of 14C-deoxyglucose decreased due to raised
loading
doses (Fig. 4). Since glucose
utilization
of
the brain is not affected by plasma glucose or insulin con centrations (24,25), it is reasonable that the cerebral-to cerebellar ratio of glucose utilization (BR!@@) dose not change due to increased plasma glucose or insulin concen trations.
The BRILV and BRIRV reached a plateau after glucose loading dosesof greater than 1 mg/g (Fig. 4). In the per fused rat heart at low work loads, insulin accelerates the membrane transport of glucose; during maximal insulin stimulation, the membrane transport rate exceeds the phos phoiylation rate of intracellular glucose and phosphoryla tion becomes limiting for glucose uptake (32,33). Our re
*
sults suggest
£2
that myocardial
glucose
utilization
becomes
maximal at the 1 mg/g loading dose.
00 0 .0
In contrast, BRISKincreases as the loading dose of glu cose increased up to the maximum loading doses studied (Fig. 4). The difference between the myocardium and the skeletal muscle in BR! response to oral glucose loading may be related to different metabolisms of red and white muscle. In red muscle, such as the myocardium, there is no synthesis of glycogen from pyruvate and oxaloacetic acid due to absence of the appropriate enzymes, such as fructose 1,6-diphosphatase and phosphoenolpyruvic carboxylase. In contrast, these enzymes in white muscle explain why skel
.@(-)
Cu)
-a i'?
LV
RV
5K
OC
FiGURE 4. Effectsof graded doses of oral glucose loadingon
etalmuscle hasa largercapacity forglycogen storage than
tissue glucose utilization(GU)expressedas the BRI. Valuesare mean ±s.d. of six animals for each dose. LV = left ventricular myocardium,RV = right ventricularmyocardium,SK = skeletal muscle, CX —cerebralcortex. *p < 0.05 compared to 0 mg/g glucose load. tp < 0.05 compared to 0.5 mg/g glucose load. @p < 0.05comparedto 1 mg/gglucoseload.
the myocardium after glucose loading (31 ). Glycogen in
2098
skeletal muscle may even increase while glycogen in heart muscle decreases in the fed state (34).
The ratio of maximal BRILVafter a glucose loading dose of 2 mg/g to that of fasted controls was 4.16, and the ratio
The Journal of NuclearMedicine• Vol.36 • No. 11 • November1995
TABLE 2 Plasma Glucose, Insulinand Free Fatty Acids (FFA)Concentrationsin Patients Insulin(J.LU/liter)
Glucose (mg/do Patientno.
60 mm
90 mm
120mm
Avg.
s.d.
s.diavg.
60 mm
90 mm
FFA(mg/dl)
120mm
60 mm
90 mm
120mm
117616714016115.30.1068.9970.7960.180.110.080.0521531381
1413516.10.1248.4648.1528.910.250.190.14319723021321313.50.0618.5125.0328.180.080.060.0341259111210914.00.1313.9422.4323.910.140.090.0551421391
.8343.5840.810.050.030.0562062092192115.60.0326.1832.3437.110.320.220.09717113112314221.00.15132.8054.4423.250.080.070.06813155738632.40.38193.
.3523.060.130.080.0560
7021935.50.1633.5041
mm =the time of FDGinjection; 90 mm=
middleof
the dynamic study; 120 mm=
end
of
the
dynamicstudy.
was similar at 3.74. The ratio
using compartment
of maximal BRISK (after a glucose loading dose of 4 mg/g) to that of fasting controls was 7.39. Hom et al. reported the
sis (20,21 ) because
response of deoxyglucose uptake in the brain, heart, skel etal muscle and other tissues after a single injection of submaximal doses of insulin (33). When we apply the BR!
FDG uptake highly depends, is changing.
of BR!RV to fasting controls
method to their results (35), the ratios of BR! of the heart and skeletal muscle after a single injection of insulin to the
plasma
glucose
models (19) or Patlak graphical analy the metabolic
concentration,
administration
study.
In the present study, BRIRVwas approximately 70% of BRILVfor all glucose loading doses. Recently, Kuschinsky et al. (36 ) also reported
greater
glucose
utilization
of the
left ventricle compared to the right ventricle using quanti tative
autoradiography
and ‘4C-deoxyglucose in rats. The
upon which myocardial
Euglycemic hyperinsulinemic clamp is a useful method to produce metabolic steady-state conditions stimulating FDG uptake by insulin-sensitive tissues (17,22). It is too complex, however, for routine clinical PET studies. For cardiac PET
BR! of fasting controls were 4.37 and 7.08, respectively, studies, we selected which are similar to those in the present
state is not steady and the
lization
75 g glucose
as the loading
dose for
to humans based on maximum glucose uti
in rat hearts
after
oral administration
of 1 mg/g
body weight glucose. BR! was calculated using the cerebel lum as a reference and compared to rMGU calculated by Patlak graphical analysis. There was a good correlation between the BR! and
difference between glucose utilization of the left and right
rMGU, especially in patients with constant plasma glucose
ventricles may reflect their differing work loads due to left ventricular contraction against aortic pressure compared to right ventricle contraction against the pulmonary artery pressure.
concentrations.
Fluonne-18-FDG For cardiac PET studies, oral glucose loading has been commonly
used before FDG injection
to increase
its uptake
and obtain myocardial images of good quality (4,14). Met abolic conditions after oral glucose loading, however, are not always suitable for quantitative measurement of rMGU
Since rMGU calculated by Patlak graphical
analysis requires a stableplasmaglucose concentration, it may be that BR! is more accurate than rMGU under conditions of varying glucose concentration. Thus, rMGU was underestimated in Patients 2 and 4 and overestimated
in Patients 8 and 9 because of varying glucose concentra tions during the study (Fig. 5). Also, for Patient 2, the cerebellar FDG uptake was quite unusual in that it was heterogenous for unknown reasons. This may be one of the reasons why this patient's data did not correlate with that in Figure 6. By using Equation 1, we can derive a rMGU value
TABLE 3 PlasmaGlucose, Insulinand Free Fatty Acids (FFA)Concentrationsin Normal Volunteers Glucose (mg/dI) Volunteer no.
60 mm
90 mm
120 mm
Avg.
Insulin(PU/liter) s.d.
s.diavg.
60 mm
71 151 1382103.318.50.181 7.8418.736.280.180.330.29889106106100.39.80.1027.4437.5031 70.560.2491 1310980100.718.00.1845.4342.8222.390.050.430.18101 191 101091 12.75.50.0531 .9822.3620.150.060.130.0760
mm = time of FDG injection; 90mm = middie ofthedynamicstudy;
90 mm
FFA(mg/do
120 mm
60 mm
90 mm
120 mm
.120.1
120 mm = end of dynammcstudy.
SimplifiedQuantificationof MyocardialFDG-PET• Nakagawa et al.
2099
B
A.. •• • 0
ROll
. ROI2 5
@
B
,••s..•B
ROI3
.
a R014 4
@
B
R015
0
R016
BE B
___
. R017
m
3
@
a R018 B R019
2
ha
a
@o 0f@B B @B B BB B
a
0
0.0
0.2
0.4
0.6
0.8
1.0
rMGU (jimol/min/g)
0.0
0.2
0.4 0.6 rMGU (pmol/min/g)
0.8
1.0
FIGURE 5. (A)Datapointsof 117 ROlsconsistsofthe BRIand regionalrMGUinninepatients.(B)CorrelationbetweenrMGUand BRIm in ninepatients. from a BR! value without consecutive data acquisition or
restless for prolonged
constant glucose concentration measurements.
brain scan to obtain the cerebellar reference for relative
The rMGU calculated with Patlak graphical analysis in
studies. Although
BR! requires
a
myocardial uptake, the time for this additional static image
the four normal volunteers demonstrated homogenous is shortcompared to theprolonged protocolof dynamic FDG uptake throughout the heart. These data are similar imaging required by the Sokoloff model or Patlak's analysis. Quantitation of relative myocardial FDG uptake by BR! is to those reported by Knuuti et al. (17). Umitations The BR! is a standardized measure of FDG uptake relative to the cerebellum
that does not require arterial
blood sampling or the insulin clamp to ensure constant plasma glucose concentrations. This method cannot be used in patients who have cerebellar infarction. It is a relative index that is not useful for measuring absolute myocardial glucose utilization. It is, however, useful for
45 mm shorter than our standard protocol for these models
for determining absolute glucose consumption. Table 5 summarizes the advantages/disadvantages of the various models for quantifying myocardial FDG uptake. Since there is no gold standard for measuring absolute regional myocardial glucose utilization in humans, the
standardized relative comparisons between a patient and a group of patients, between different groups of patients or between
repeat
studies
in the same patient
where the mdi
vidual metabolic conditions of each patient or study are normalized in an objective reproducible way. It is therefore also sensitive for intrapatient comparisons under different conditions, such as before and after coronary intervention.
Since BR! dose not require arterial input function and consecutive data acquisition, it shortens patient scanning time, allows more studies per day and decreases the phys ical burden on cardiac patients that are often too ill or
TABLE 4
Values of rMGUin NormalSubjects 0.4
SeptalAntenorLateralMean0.71 0.15mean ±0.200.74
±0.100.74
±s.d. (@mol&mmn/g).
2100
0.6
rMGU (@smol/min/g) ±0.160.73
±
FiGURE 6. CorrelationbetweenrMGUand BRImin four patients
who had relativelyconstant plasmaglucoseconcentrationsduring dynamicdata acquisition. The Journal of NuclearMedicine• Vol.36 • No. 11 • November 1995
TABLE 5 Comparison of Analysis Methods of Tissue Metabolic Rates Using FDG-PET analysisand/orPatlak's Patlak's modeland/or+BRI*Sokoloff analysisSokoloff
uptake@ValueRelativetAbsoluteRelativeRelativeData
modei@insulin
clamp%
Dose uptake@Late
acquisitionStaticDynamicDynamicStaticStaticData analysisSimpleCOmpliCatedComplicatedSimpleSimpieCorrection functionIncludedCorrectableCorrectableUncorrectableUncorrectableCorrection of input glucoseIncludedUncorrectableCorrectableUncorrectableUncorrectableconcentration of plasma
effectCorrection constantIncludedCorrectableCorrectableUncorrectableUncorrectable*PET of lumped loading.t@nv@ble study after oral glucose
to absolute values usingthe correlalionbetween BRIand rMGUbased on patientswith stable glucose levels.
Sokoloff model and Patlak's analysis after oral glucose loading is the only way to acquire absolute but not neces sarily accurate values if plasma glucose concentrations vary.
BR!, however, as a relative index, has advantages over other models for quantifying myocardial glucose utilization
both conceptually and experimentally since it is not depen dent on varying blood glucose levels that invalidate these other models and it is much simpler than other more complex models.
dium with persistent 24-hour single-photon emission computed tomography 20111 defects.
cardiomyopathy.AmI Cardiol1992;69:242—246.
RO.Regionalsystolicfunction,myocardial bloodflowandglucoseuptakeat
Larger clinical studies will be required to demonstrate definitively the clinical role of BR! in cardiac PET. The
BR! method provides semiquantitative measurements of maximal relative myocardial glucose utilization after oral glucose loading without directly determining arterial input function or steady-state glucose concentrations, and is the only method of objectively quantifying myocardial FDG
uptake under conditions of varying plasma glucose concen trations commonly seen after oral glucose loading.
ACKNOWLEDGMENTS The authors thank Dr. K. Lance Gould, Division of Cardiology, University of Texas, for his thoughtful review of this manuscript.
We also thank Keiko Imazeki, PhD and Hiroshi Ito, RT for FDG production and management of the PET system at Chiba Univer Chiba,
1992;86:1357—1369.
7. Nienaber CA, Gambhir SS, Mody P/, et al. Regional myocardial blood flow and glucose utilization in symptomatic patients with hypertrophic cardio myopathy.Circulation 1993;87:1580—1590. 8. Perrone-Filardi P. Bacharach SL, Dilsizian V, Panza JA, Maurea S, Bonow
CONCLUSION
sity Hospital,
Circulation
5. Grover-Mckay M, Schwaiger M, Krivokapich J, Perloff JK, Phelps ME, Schelbert HR. Regional myocardial blood flow and metabolism at rest in mildly symptomatic patients with hypertrophic cardiomyopathy. I Am Coil Cardiol 1989;13:317—324. 6. Kagaya Y, Ishide N, Takeyama D, et al. Differences in myocardial fluoro 18-2-deoxyglucose uptake in young versus older patients with hypertrophic
rest in hypertrophic cardiomyopathy. Am I Cardiol 1993;72:199—204. 9. Bing RI. The metabolism of the heart. In: Han'ey lecture series. New York:
AcademicPress; 1954:27—70. 10. Opie LH. Function of the heart in normal and pathologicalstate. In: Sperelakis N, ed. Substrate and energy metabolism of the hewi. New York: Martinus Nijhoff; 1984:301—336. 11. Opie LH. Metabolism of the heart in health and disease. Am Heart I
1968;76:685—698. 12. Neely JR, Morgan HE. Relationship between carbohydrate and lipid me tabolism and the energy balance of heart muscle. Ann Rev Physiol 1974;36: 413—459. 13. Gropler Ri, Siegel BA, Lee KJ, et aL Nonuniformity in myocardial accu mulation of fluorine-18-fluorodeoxyglucose in normal fasted humans. INuci Med 1990;31:1749—1756. 14. Gould KL. PET perfusion imaging and nuclear cardiology. I Nucl Med 1991;32:579—606. 15. Tamaki N, Yonekura Y, Kawamoto M, et al Simple quantification of regional myocardial uptake of fluorine-18-deoxyglucose in the fasting con dition. I Nucl Med 1991;32:2152—2157. 16. Hicks Ri, Herman WH, Kalif V, et al. Quantitative evaluation of regional
substratemetabolismin the humanheart by positronemissiontomography.
Japan.
REFERENCES 1. Marshall RC, Tillisch JH, Phelps ME, et al. Identification and differentiation of resting myocardial ischemia and infarction in man with positron computed tomography, ‘8F-labeled fluorodeoxyglucose and N-13 ammonia. Circulation
JAm Coil Cardiol 1991;18:101—111. 17. Knuuti Mi, Nuutila P, Ruotsalainen U, et al. Euglycemic hyperinsulinemic
clamp and oral glucoseload in stimulatingmyocardialglucoseutilization during positron emission tomography. I Nucl Med 1992;33:1255—1262. 18. Dahl JV, Herman WH, Hicks Ri, et al. Myocardialglucoseuptake in patients with insulin-dependent diabetes mellitus assessed quantitatively by dynamic positron emission tomography. Circulation 1993;88:395—404.
1983;67:766—778. methodfor 2. CamiciP,AraujoLI, Spinks1, Ctal. Increased uptakeof ‘8F-fluorodeoxy19. SokoloffL, ReivichM, KennedyC, et al. The [‘4C]deoxyglucose the measurementof localcerebralglucoseutilization:theoiy,procedureand glucose in postischemic myocardium of patients with exercise-induced an normal values in the conscious and anesthetized albino rat. I Neurochem gina.Circulation1986;74:81—88. 1977;28:897—916. 3. SchwaigerM, BrunkenRC, KrivokapichJ, et al. Beneficialeffectof residual anterograde flow on tissue viability as assessed by positron emission tomog raphy in patients with myocardial infarction. Eur Heart I 1987;8:981—988. 4. Brunken RC, Mody P/, Hawkins RA, Nienaber C, Phelps ME, Schelbert
HR. Positron emissiontomographydetects metabolicviabilityin myocar
20. Patlak c:s, Blasberg RG, Fenstermacher JD. Graphical evaluation of blood to-brain transfer constants from multiple-time uptake data. J Cereb Blood Flow Metab 1983;3:1—7. 21. Patlak c:s, Blasberg RG. Graphical evaluation of blood-to-brain transfer
Simplified Quantificationof Myocardial FDG-PET• Nakagawaat al.
2101
constants from multiple-time uptake data. Generalizations. I Cereb Blood 29. Ratib 0, Phelps ME, Huang S-C, Henze E, Selin CE, Schelbert HR. Flow Metab 1985;5:584—590. Positron tomographywith deoxyglucosefor estimating local myocardial 22. DeFronzo RA, Tobin JD, Andres R. Glucose clamp technique; a method for glucosemetabolism.I NucIMed 1982;23:577—586. quantifying insulin secretion and resistance. Am I Physiol 1979;237:E214— 30. Wi-JO. Diabetes mellitus. Report of WHO Study Group, Technical Report E223. Series. WHO, Geneva; 1985:727. 23. NambaH, NakagawaK, IyoM, FukushiK, Itie T. A simplemethodfor 31. Opie LH. The heart. In: Physiologyand metabolism, 2nd edition. New York: measuring glucose utilization of insulin-sensitive tissues by using the brain as RavenPress; 1991. a reference. EwJ NucI Med 1994;21:228—231. 32. Morgan HE, Henderson MJ, Regen DM, Park CR. Regulation of glucose 24. Lucignani G, Namba H, Nehlig A, Porrino LI, Kennedy C, Sokoloff L. uptake in muscle.1.The effectsof insulinand anoxiaon glucosetransport Effects of insulin on local cerebral glucose utilization in the rat. I Cereb and phosphorylationin the isolated,perfused heart of normal rat. J Biol Blood Flow Mezab 1987;7:309—314. Chem 1961;236:253—261. 25. OrziF, LucignaniG, Dow-Edwards D,et al.Localcerebralglucoseutiliza 33. Hom FG, Goodner Ci, Beme MA. A [3HJ2-deoxyglucose method for tion incontrolledgradedlevelsof hyperglycemiain the consciousrat.I Cereb comparing rates of glucose metabolism and insulin responses among rat Blood Flow Metab 1988;8:346—356. tissues in vivo. Diabetes 1984;33:141—152. 26. Namba H, Lucignani G, Nehlig A, et al. Effects of insulin on hexose 34. Nguyen VT, Mossberg KA, Tewson Ti, et al. Temporal analysis of myocar transport across blood.brain barrier in normoglycemia. Am I Physiol 1987; dial glucose metabolism by 2-['MF]fluoro-2-deoxy-D-glucose. Am I Physiol 252:E299-E303. 1990;259:H1022—H1031. 27. Kushner M, Tobin M, Alavi A, et al. Cerebellar glucose consumption in 35. Hom FG, Goodner Ci. Insulin dose-response characteristics among individ normal and pathologic states using fluorine-FDG and PET. I Nucl Med
ual muscle and adipose tissues measured in the rat in vivo with [3H]2-
1987;28:l667—1670.
28. GambhirSS,SchwaigerM,HuangS-C,et al.A simplenoninvasive quanti fication method for measuring myocardial glucose utilization in humans employing positron emission tomography and fluorine-18 deoxyglucose. I NucI Med 1989;30:359—366.
2102
deoxyglucose. Diabetes 1984;33:153—159. 36. Kuschinsky W, Bunger R, SchrOck H, Mallet RT, Sokoloff L. Local glucose
utilizationand local blood flowin hearts of awake rats. BasicRes Cardiol 1993;88:233—249.
The Journal of Nuclear Medicine • Vol. 36 • No. 11 • November 1995