Download Model-Based Methods for Measuring Binding Potential in SPECT Imaging and more Study notes Neurology in PDF only on Docsity!
been used as a PETradiotracer for visualization and quan
tification ofbenzodiazepine receptors in humans (Z12-16).
Recently, an iodinated analog of fluinazenil, iomazenil (Ro
16-0154), has been introduced as a SPECT radiotracer (17).
SPECT studies in nonhuman primates (18) and healthy
human subjects (19) have shown that [‘@IJiomazenilhad a
high brain uptake (iO%—i2%of the injected dose). About
90% of this brain activity is displaceable and therefore,
associated with specific binding to benzodiazepine recep
tors.
Analysis of SPECT neuroreceptor imaging studies have
been typically restricted to empirical, semiquantitative
methods such as the ratio of activities in regions of interest
(RO!s) or the RO! washout rates. Since these empirical
measures do not control for factors such as peripheral
clearance, binding to plasma proteins and cerebral blood
flow, the ability of these measures to provide quantitative
information about the receptors under study is question
able (20). The goal of the studies reported here was to
develop model-based methods for SPECT measurement of
[‘@!]iomazenilbinding to benzodiazepine receptors in the
human brain.
Model-based methods for in vivo quantification of neu
roreceptors can be broadly divided into kinetic and equi
librium methods. Kinetic methods yield quantitative inior
mation about the receptorsby estimatingthe rate constants
which characterize the transferbetween plasma, brainand
receptor compartments (21). Equilibrium methods derive this information by analyzing the activity distribution at
equilibrium, i.e., when the receptor-ligand association and
dissociation rates are equal (22). With both approaches,
the outcome measure of experiments performed at tracer
doses is a unitless number, the binding potential (BP),
which equals the product of the receptor density (Bm@,
nM) and affinity (i/Ks, @J@1)(21). The BP is also the
equilibrium distribution volume of the receptor compart
ment, i.e., the ratio of specific binding in the brain-to
Iodine-123-iomazenilbindingto benzoduazepinereceptorsin hu
manbrainwasmeasuredwithSPECTusingkineticandequilib
nummethods.Mthods: In the kineticexperiments(n = 6),
regionalthne-activitycurvesaftera singlebolusinjec@onof the
tracerwere fit to a three-compartmentmodelto provideesti
matesofthe rateconstantsK1to k. Thebindingpotential(equal
to the productof the receptordensityand affinity)was derived
fromthe rateconstants.In the equilibriummethod(n = 8),the
tracerbolusinjectionwasfollowedby a constanttracerinfusion
toinduceasustalnedequilibriumstate.Theregionalequilibrium
volumeof distributionwascalculatedasthe ratioof the regional
brainCOnCentrabOn-ID-thefreeparenttracersteady-stateplasma
concentration.Inthreeexperiments,a receptor-saturatingdose
offlumazenil was injectedfor direct measurementofthe nondis
placeablecompartmentdistributionvolume.Results: The Id
neticandequilibriummethodresuftswereingoodagreementin
all regions investigated. Iodine-125-iomazenilbinding potential
measuredin vitroin 12 postmortemsampleswasfoundto be
consistent wfth SPECT in @vomeasurements. Conclusion:
Thesestudiesdemonstratedthe fea@brIftyof quantificationof
receptorbindingwithSPECT.
KeyWords:SPECT;brain;benzodlazeplnereceptors;10-
dlno-123-Iomazsnll
J Nuci Md 1994; 35:228-
terations of central benzodiazepine receptors have
been described in several neuropsychiatric conditions, in
cluding epilepsy (1—3),Alzheimer's disease (4,5), Hunting
ton's chorea (6—8)and schizophrenia (9—11).Carbon-li
flumazenil (Ro 15-1788),a benzodiazepine antagonist, has
ReceivedJul. 6, 1993;revisionaxepled Nov.4, 1993. Forcorrespondenceand reprirdscon@ As@ssaAbi-Dar@wn,MD,De@ of Psychiatry,Y@eUniversityand West Haven VA MedicalCerter/116A2, CaIT;@bstAve.,West Haven,CT06516.
228 TheJournalofNudearMedicine•Vol.35 •No.2 •February^1994
SPECT Measurement of Benzodiazepine
Receptors in Human Brain with
Iodine-123-Iomazenil: Kinetic and
Equilibrium Paradigms
Anissa Abi-Dargham, Marc Laruelle, John Seibyl, Zachary Rattner, Ronald M. BaldWin, Sami S. Zoghbi,
Yolanda Zea-Ponce, J. Douglas Bremner, Thomas M. Hyde, Dennis S. Charney, Paul B. Hoffer and
RobertB. !nnis
DepartmenL@ ofPsychiatiy and Diagnostic RadiOlOgy, Yale UniVeTSÜySchool ofMedicine, New Haven, Connecticut and
West Haven VA Medical Center, West Haven@Connecticut; Clinical Brain Disorders Branch, IRP, NIMH Neuroscience
Center at St. Elizabeths, Washington, DC
yieldof the [@!Jiomazenilpreparationsaveraged66.4%±8.2%
(with these and subsequent data expressed as mean ±s.d., n = 14) and the radiochemical purity averaged 97.6% ±1.7% (n = 14).The specific activity was determined by comparh@gthe UV absorbance
of thelabeledpmductwitha standardcurvegeneratedfromknown
concentrations of nonradioactive iomazenil. The specific activity of E'23!liomazenilin these preparations was found to be greater than 5000 Ci/mmole and may be estimated to be on the order of 180,
CL'mmole(37).Sterilitywas assuredby lackof bacterialgrowthin
twomediaandapyrogenicitywasconfirmedusingthelimulusame
bocyte lysate (LAL)test. The specific activity of [1@I]iomazenilwas estimated by UV detection to be 1090 Ci/mmole on the day of
labelin@
@ He&thy
Fourteen healthy subjects (age 26 ±6 yr. weight 75 ±7 kg, 11 males and 2 females) were recruited for these studies. Inclusion
criteria were: (1) absence of current medicalconditionsand (2)
absence of neumpsychiatric illness, alcohol or substance use. A physical examination, EKG and routine blood and urine tests were performed in the screening procedure. All subjects gave written informed consent. Protocols were approved by the insti tutional human investigation committee. Subjects received potas sium iodide (SSK! solution 0.6 g in the 24 hr period prior to imaging).
Data Acquisition
SPECF datawere acquiredwith the multislice brain-dedicated
CERASPECF camera (Digital Scintigraphics,Waltham, MA)
(38). The transaxial and axial resolution in air are 7.7 and 5.9 mm
FWHM,respectively(39).Inwater,theresolutionis 10—12mmin
thethreeplanes.
Four fiducialmarkersfilled with 10 j@Ciof [@Tc]sodium
pertechnetatewere attachedon both sides of the subjects's head
at the level of the cantho-meatalline to control positioningof the
head in the gantrybefore tracerinjectionand to identifythe
cantho-meatal plane during image analysis.
Singlebolus experimentswere performedin six subjects(age
23 ±0.7yr,weight78 ±11kg).Iodine-123-iomazenil(12.8±4.
mCi)was injectedas a single bolus over 30 sec. Scans were
acquired in continuous mode every 2 mm for 145 ±5 mm. Arterial blood samples (1—2ml) were collected every 20 sec for the first 2
mis with a peristalticpump(Harvard2501-001,SouthNatick,
MA)andthenmanuallyat3, 4, 6, 8, 12,16,20,30,45and60mm.
After the first hour, samples were drawn every 30 mis until the
endof theexperiment.
Constant infusion experiments without flumazenil displace ment were performedin five subjects (age 26 ±9 yr, weight 75 ±
4 kg).Thedosewasdividedinabolusof3.89 ±0.39mCifollowed
by aconstantinfusionof 1.08±0.12mCi/hr(IMEDpump,Jemini
PC-i, San Diego, CA) given over 450 miii, resultingin a bolus-to
infusionratioof3.61 ±0.27hr.Scanswereacquiredincontinuous
modeevery 5 ruinfrom250 to 450 ruinpostinjection.In three
subjects, data were also acquired from 0 to 150mis postinjection.
Arterialsampleswere collectedin a pattern similarto the single
bolus experiments in the first three experiments, and every 15 mm starting 2 to 4 hr postradiotracer injection in the last five experi
ments.
Humazenildisplacementduring[‘@I]iomazenilconstantlain
sion was performedin threesubjects(age 27 ±2.6 yr. weight
75 ±12 kg). The injection and acquisition protocols were similar to the previous constant infusion studies (bolus 4.1 ±0.54 mCi, infusion 1.12 ±0.16 mCi/hr resulting in a bolus-to-iniusion ratio of
plasma level of free parent tracer at equilibrium.Benzodi
azepine receptors have been measured with [11C]flumazenil
and PET with kinetic (1423,24) and various “equilibrium―
methods (2,25-28).
Among equilibrium methods used to measure [“C]flu
mazenil BP, it is important to distinguish the “peakup
take―equilibriummethod (2) and the “transient―equiib
rium method (25,27,29). “Peakequilibrium― methods measure the BP at peak uptake of the specific binding, usually obtained by subtracting the activity in the white
matter (2) or the pons (2428) from the activity in the RO!.
At peak uptake, the BP is calculated as the ratio of activity
in the ROl-to-activity in a reference region. The difficulty
of this method is the proper identificationof peak uptake
for ligands such as [1@I]iomazenilwhich exhibit a pro
tracted plateau phase (19). The “transient―equilibrium
method, also called quasi-equilibrium(27) or pseudo-equi
librium (25), refers to the measure of the BP when the
specific-to-nonspecific ratio or the specific binding-to
plasma tracer concentration ratio becomes constant over
time (i.e., when both are decreasing at the same rate).
Carson et a!. (30) proposed the term “transient―equilib
rium to describe these conditions. As opposed to the peak
uptakeequilibriummethod, the transientequilibriummeth
ods do not satisfy the Michaelis Menten equilibriumequa
tion and can result in an overestimation of the BP (30). To overcome the technical and theoretical difficulties associ ated with the use of equilibrium methods after single bolus
injection of the tracer, we implemented in humans the
constant infusion/sustained equilibrium method (30—32)
that we previously described in baboons (33—35).
In the study reported here, benzodiazepine receptor BP
was measured in 14 healthy subjects with [‘@!Jiomazenil
and SPECT using both a kinetic (i.e., bolus only) and a
sustained equilibrium (i.e., bolus plus constant infusion)
paradigm. Plasma clearance of the tracer estimated from
the single bolus experiments was used to design the tracer
administration protocol of the bolus plus constant infusion
experiments. A direct measure of the nonspecific compart
ment was obtained in three of the eight constant infusion
experiments by injecting a receptor saturating dose of flu
mazenil (0.2 mg/kg) at the end of the study. In addition, the
accuracy of SPECT measurements was confirmed with
[‘@!]iomazenilbinding studies to postmortem human brain
samples.
METhODS
Radlolab.llng Iodine-123-iomazenil and [‘@!Jiomazenilwere prepared by io dodestsnnylation of ethyl 7-(thbuty@tannyl)-5,6-dihydro-5-meth yI-6-oxo-4H-hnidazo[1,5.a][i,4jbenzodiazepthe-3-carboxylate with
chloramine-Tin methanolicaceticacidat 120°C(36).Radiolabeled
productswere purifiedby reversed-phaseHPLC(C-18column,
3.9x 300mm,55%CH3OH/H20,0.7mI/rain;R@9.2mm),formu
latedin 5%ethanolin normalsalinecontaining0.1mML-ascorbic
acid, pH 5-6, and filtered through a O.2-@membrane filter into a sterile 10-mI serum vial or sterile 10-mi syringe. The radiochemical
SPECTMeasurementofBenzodiazepineReceptors•Abr-Darghametal. 229
@@ K1 = FE = F(1—e (ml. g ‘. miii Eq.
15k3 k2 K1/V2f1(min@ 1),Eq.
16k4=k0ff(miii@'),Eq.17 kcn@mJ2(mm ‘),Eq.
FiGURE1. Thethree-comparb@nentmodelusedforbothkinetic
and equilibriumanalyses.C, = arterialplasmaconcentration,me
taboiftecorre@ed;C2= tra@erconcentrationinIntr@erebralnondis
@abIecompartment(freeandnonspecificbinding);C, = specif
Icallyboundconcentration;K, to k, = fractionalrateconstants
describingthekineticsoftracertransferbetweencompartments.
compartment (f') is constant over time. The vascular activity present within an RO! was estimated assuming a blood volume equal to 5% of the ROI volume (21) and subtracted from the total
RO! activity prior to analysis.The activity concentrationin the
RO! at time t, C@01(t),is the sum of the concentrations in the
second and third compartments:
CRo@(t)= C,Jt)+ C@(t). Eq. 7
The equilibriumvolume of distributionof compartmenti is the ratio of the tracer concentration in compartment i to the free tracer concentration in the arterial plasma at equilibrium, i.e.,
when no net transferbetweenthe plasmaand compartmenti
exists:
vi = @a.
f1C
V2is theequilibriumvolumeofdistributionofthe nondisplaceable
compartment and V3 is the equilibrium distribution volume of the specifically bound compartment. The total tissue equilibrium vol. ume of distribution,V.@.,is the sum of both compartments:
CR0!
VT j@ = V2+ V3.
Passive diffusion is assumed to be the mechanism of transfer of the tracer across the blood-brainbarrier(BBB). When the non displaceable compartment is at equilibrium with the plasma, the free concentration is assumed to be equal on both sides of the BBB,
f1C5= f2C@, Eq.
which yields the inverse relationship between f2 and V2:
f2= i/V2.
Kinetic Anc4ySiS. The tracer concentration over time in each compartment is given by:
dC@(t)
.—@;.—=K1C@(t)—k2C@(t)—k3C@(t)+k@C@(t),Eq.
dC@(t)
—@i--=k3C@(t)—k@C@(t), Eq. 13
where F is the regionalblood flow (ml •g'. min1); E the
unidirectional extraction fraction; PS the permeability surface area productof the tracer (ml •.g'. @i@@');and k@,and k.@the association and dissociation rate constants for ligand-receptor binding. Equation 15 can be derived from Equation 12 by setting k3, k4 and the derivates to zero and multiplyingboth sides by f1.
Attracerdoses,Equations12and13haveconstantcoefficients
and can be solved analytically (43,44), and BP can be derived from the rate constants using Equations 15, 16 and 17:
BP Bmax koaBmax k3 K1k3 i — KD 1(01! k42k2k41 Eq.
Equilibnum Ana@YSiS.At equilibrium, the rate of association and dissociation of the tracer-receptor complex are equal:
Thus,
@ @f2C@B@= Eq. 19
Bmax C3 C
BP-@=@=@-@, Eq.
1@ 8 which demonstrates that BP is equal to the equilibrium volume of £.d@1. distribution of the bound compartment (V3, Equation 8).
At equilibrium,VTwas calculatedwithEquation9. For con
stant infusion experiments followed by an injection of a saturating
dose of flumazenil,V2was calculatedas the ratio of the nondis
placeableactivity-to-theplasma-freeparentcompound(from15to
45miiipostflumazenilinjection)andBP(V@)wascalculatedasthe
differencebetweenVTandV2(Equation9).
Eq. 9 Cwve-Fitting Plvcedwe. Rate constants for arterialclearance and brain uptake ofthe tracerwere estimated by nonlinear regres sion using a Levenberg-Marquartleast-squaresminimizationpro cedure (45) implemented in MATLAB (The Math Works, !nc., South Natick, MA) on a Macintosh Quadra 950. The number of exponential terms used to describe the plasma clearance was
determinedby the F-test (46) and the AIC criteria (47). The
standard error of the parameters was given by the diagonal of the covanance matrix (48) and expressed as percent of the parame
ters (coefficientof variation,%CV).Thus, %CVindicatesthe
identifiabilityofthe parameterby the least-squaresprocedureand
should not be confused with the Standard deviation (s.d.) of the Eq. ii distributionof the parameteramong subjects.
inVitror@'iiom.@anIIBk@dingExpsrlm.nts Twelve brains were collected from the District of Columbia Medical Examiner's Office. The brains were examined by a fo rensic pathologistor a neuropathologistfor the detection of gross abnormalities and were cut into i-cm thick coronal slices. Sam
pleawere dissectedon ice and immediatelyfrozenat -70°Cuntil
assayed. Inclusion criteria were absence of gross pathological
abnormalitieson examinationof the brain and absenceof neuro
psychiatricdisordersas noted by reviewof the medicalrecords.
wherethe kineticparametersK1to k, aredefinedas follows: Ages varied from 22 to 87 yr (50 ±22 yr). Brains from four
SPECTMeasurementofBenzodlazeplneReceptors•Abr-Darghamatal. 231
SubjectV@CLno.Ofter/kg)(lfter/hr/kg)
VboI initial volume of distribution In plasma CL = P@riP'@' dEW
ance;andf1= freefractionInplasma.
females and eight males were used. Time between death and freezingof tissue was 24 ±10 hr. Preservationtime at —80°Cwas 76 ±12 mo.
Onthedayof theassay,brainsampleswereweighed,thawed
and homogenizedwith a Polytron (setting6 for 10 5cc) in 1/40wet weight in mg/vol in ml (wAr)of buffer(25 mM KH2PO4'150 mM NaCl, pH 7.4). After homogenization, tissues were centrifuged (20,000g, 4°C,10min). The supernatantwas discardedandpellets were resuspended and recentrifuged. This procedure was re
peated twice. Incubation(22°C,45 mm)was initiatedby the suc
cessive additionof 100 @tlof [‘@!]iomazenil,100 @dof bufferor
unlabeled iomazenil and 800 @dof tissue solution. A final tissue dilution of 1/2,400 w/v was selected so that total binding was between 5% and 10% of total ligand concentration. Incubation was terminated by rapid filtration though OF/B ifiters on a 48- channelCell Harvester (Brandel,Gaithersburg,MD). Filterswere rapidly washed three times with 5 ml of ice-cold buffer and counted in a COBRA 5010 gamma counter (Packard, Menden, CT) with an efficiency of 80%. Saturationexperiments (n = 12) were performedby the isotopic dilution method (“cold―satura lion) on the occipital cortex from each brain using one concentra tion of [‘@!]iomazethl(0.02 nM) and 15 concentrations of unla beled iomazenil ranging from 10_14 to 106 M. Saturation experiments were analyzed by weighted nonlinear regression analysis using the programLIGAND (NIH, Bethesda, MD) (49).
RESULTS
SingleBolusEXperimentS After single bolus injection of [‘@!]iomazenil,the six subjects exhibited similar kinetics of plasma clearance of
the parentcompound (Table 1). In all cases, a sum of three
exponentials provided a statistically significant improve ment in the fit as compared to a two-exponential fit. A four-exponential fit provided a statistically significant im
provement of the fit in only two of six subjects. Therefore,
a three-exponential model was chosen (Fig. 2A). !mtial
volume of distributionwas 0.44 ±0.14 liter/kg and clear
ance was 7.4 ±1.1 liter/hr/kg.The average A3(0.013 ±0.
min') corresponded to a terminalhalf-life of 97 mm. The
free fraction ofthe parent compound was between 20% and 30% of the total parent compound (f1 = 0.23 ±0.02). The occipital region showed the highest and latest up take, peaking at about 30 mm (Fig. 2B). Other investigated
40 80 120 160
i
C.)
C 900
F
@ 300
80
TiME(mln)
160
FiGURE2. (A@Arterialtlme-actMtycurve of free parent
[1@lJiomazenilfollowinga bolusInjectionof 12.5mCIIn a 22-yr-old male.Valueswerefittoatriexponentlalmodel(solIdlIne)toderive
the peripheralclearance(Cl = 8.5 lIter/hr/kg).(B)Regionalbrain
time-actMtycurvesfromthesamesubjectasInpanelA Measured
values (drcles,trianglesand squares) were fitto a three-comport
mentmodel(solIdlines).BPvalueswere272,174and58 in the
occipitalcortex,frontalcortexandcerebellum,respectivaly.
RO!s peaked earlier and at lower values (in decreasing
**order temporal > frontal > striatum, thalamus, cerebellum
pons). RO! time-activity curves were fit to the three** compartment model (Fig. 2B). The iteration procedure converged for all regions, providing estimates of the re
gional rate constants (K1 to k@)and values for the outcome
measures (‘@2@f2, BP and VT, Table 2).
K1 rangedfrom 0.385 nil. g@ •@Ji@@1(pons) to 0.546 ml
. g'. min' (occipital), with an average value for all
regions of 0.467 ml. g@. piJ1@1.‘flij@parameter was
reasonably identified in the frontal, occipital, temporal and
cerebellarregions (%CVbetween 5%and 6%)but was less
well identified in the striatum (%CV ±s.d.; 16% ±9%),
thalamus (%CV ±s.d.; 18% ±15%) and pons (%CV ±
s.d.; 19% ±7%). The rate constant k2 showed a large
variation across regions, from 0.094 ±0.118 min1 (tem
poral) to 0.300 ±0.214 (pons)with a mean regional value of
0.151 ±0.150 min', and was poorly identified in all re
gions, with %CV ranging from 26.6% ±8.9% in the thal
amus to 127% ±159% in the striatum. The value of k
varied from 0.175 ±0.095 min' (occipital) to 0.104 ± 0.037 min@' (cerebellum) but was as poorly identifiable as
TABLE 1
Peripheral Clearance Parametersin Singie-BolusExperiments
10.296.80.2220.528.20.2330.487.30.2340.418.50.2350.295.60.2760.648.40.21Mean±s.d.0.
±0.147.4 ±1.10.23 ±0.
232 TheJournalof NuclearMedicine•Vol.35 •No.2 •February 1994
A 1000
iii C.) @ £
z
@ 10
Ef 2
B 1500
Parameters Frontal Occiplal Temporal Striatum Thalamus Cerebellum Pore
Valuesaremean±s.d.of sixexperiments.
mctoi@,= transferrateconstants;BP= bindingpotentialasdertvedfromkineticanalysis;@2= freefr@IonInthenondispleceablecompartment
V2- nondlspIaceai@Ieequilbiumvolumeofdl@rIbutIon;VT= t@aItissueequiblumvolumeofdistribution;and% NSB= percentnonspecific
t@ndIng
TABLE 2
KineticAnalyaisof Single BolusExperiments
mc@@jn@1)0.4630.5460.4950.4460.5340.3980.385±(ml. g
s.d.0.2190.1670.2220.2000.3550.1710.172k (m1n1)0.1560.1360.0940.1230.1430.1060.300± s.d.0.1940.0980.1180.1340.2120.0800.214k3(mln-1)0.1450.1750.1080.1340.1180.1040.114±
s.d.0.1540.0950.0610.0800.1330.0370.064k (m1n1)0.0170.0140.0170.0340.0290.0180.024± s.d.0.0050.0040.0060.0270.0100.0040.008BP161.00240.00189.0081.0078.00101.0030.00±
s.d.15.0041.003.0012.0027.0023.0011.00f20.070.050.040.050.050.060.17±
s.d.0.090.030.030.030.040.030.09V237.0031.0054.0027.0041.0020.008.00±
.004.00VT196.00271.00242.00108.00119.00121.0037.00±s.d.25.0030.0054.0019.0030.001 1
.0011.00%s.d.16.0029.0055.00180034. .0019.0024.0033.0016.0022.00± NSB18.001 1 s.d.12.0011.0015.0013.0020.006.0013.
InVitroraswom@eniIBindingin Postmortem HumanBrain
In the occipital cortex, [‘@!JiomazemlBm@was 162±
nM and KD was 0.59 ±0.14 n.M (n = 12). These values
corresponded to a BP of 287 ±95. No significantcorrela
tion was observed between age and [‘@!]iomazenilB@,, or
KD.
DISCUSSION
These studies demonstrated the feasibility of measuring
benzodiazepine receptors regional binding potential with
SPECF. Two methods were tested: kinetic analysis of a
single bolus injection and equilibrium analysis ofbolus plus
constant infusion. With the exception of the pons (where
the low level of counts may have contributed to a higher
level of noise), both methods gave similar regional results.
For example, the occipital BP was 240 ±41 (n = 5) and 248
±53 (n = 3) for kinetic and equilibrium experiments, respectively, and veiy close to the in vitro homogenate
binding value of 287 ±95.
The three-compartment kinetic analysis applied here
was comparable to the kinetic analysis developed for re
versible PET tracers such as [‘1Cjflumazenil(16,24). As
previously observed with PET, the rate constants K1 to k@
were poorly identified (high standard errors due to covari
ance) when unconstrained kinetic analysis was applied
(30,44). Furthermore, the values of the kinetic parameters wereinconsistentwiththeirphysiologicalinterpretation. Regional differences in K1 and k2 should reflect regional
differences in blood flow. Their ratio, V2 should reflect the
ments. Therefore, in subsequent experiments, arterial
plasma samples and scan datawere acquiredonly from 250
to 450 min.
To quantifythe stability of the plasma and brainuptake,
a linear regression was performedon plasmavalues (last 4
hr, mean number of points = 12) and on RO! values (last 2
hr of the experiment, mean numberof points = 22). !n the
plasma, the mean change was 2.5% ±2.1%/hr. With the
exception of the pons, all RO!s showed changes of less
than 5%/hr.
RegionalVTwas measuredby calculatingthe ratioof the
averaged RO! activity and the free unmetabolized plasma
activity over the last 2 hr (Table 3).
Bolus Plus Infusion Plus Dlsplacsmsnt Experimsnts
Flumazenil injection was well tolerated by all three sub
jects with no side effects reported at the dose of 0.2 mg/kg.
Flumazenil induced 88% displacement of the tracer in the
occipital, 87% in the temporal, 86% in the frontal, 84% in
the thalamic and cerebellar, 81%in the striataland 60%in
the pontine RO!s (n = 3). Plasma tracer activity exhibited
a slight (11% ±3%)and transient (5—15mm) increase after
flumazenilinjection. At the time ofthe measurementof V2,
the plasma tracer concentration was back to baseline. Re
gional V2 varied from 34 (occipital) to 23 (pons, Striatum
and thalamus; Table 3 and Fig. 6). The value of f2, calcu
lated as 1/V2, varied from 0.043 (pons) to 0.030 (occipital
and temporal).An importantobservation was the presence
of 60% displaceable binding in the pons, indicating that the
pons cannot be used as an area devoid of benzodiazepine
receptors sites.
234 TheJournalof NucisarMedicine•Vol.35 •No.2 •February
A
wC@ ILl 10.
4W
@ 1.
t
A
C)
C
Cl)@ -J
0
• •
B
w
**-J 0
**
z
B 600 OCCIPITAL
•IdL!s!@• ••%.•-w•? • • ••••••S••. •?
• •
8 o@ @.E. @- 300 4 mg 4
• * •• .• ••.• •• •••• @s TEMPORAL
@@ a
CEREBELLAR I 00
TIME (mm)
TiME (mln)
FiGURE4. (A)Simulatedplasmafree[1@l)momazenilconcenta
tionaftera bolusinjectionof3.4 mCIfollowedbya constantInfusion
of 1 mCI/hrfor420 mm.Thesoftdlinerepresentstheprojected
time-ectivitycurvein a 70-kgsubjectwith [1@IJiomazenilclearance
of553lfter/hr(= averageclearancemeasuredInsixsubjects).The
dottedline(.... ) andthe brokenline(———)areprojectedactMtles insubjectsw@,slow(398Ilter/hr)andhigh(708llter/hr)clearance, respectivaly.In all cases,Steady-Stateis achlevedat 300 mm.Sim ulatlonswere generatedusingthe followingparameters:V@ = @ lfterf01 = 1.0;f@= 0.11;f@= 0.026;A1= 1.18m1n1; 0. min1; A3 = 0.013 mlrr1, 0.007 min@ and 0.022 mWr1 In mean, sk@wand fast clearancescenarios,respectively.(B) Simulatedvol umes of distribution(regionalactivity/freeplasma activity)in the occlpftalcortex and cerebellum.In both regions and In all three clearancescenarios,the volumeof distributionreechedan equ@Ib
numvalue(‘IT)at 360 mm.Parametersusedforthissimulation
were:f1 = 0.231; K, = 0.546ml. mmn@. g1; k@= 0.136m1n@;k@ = 0.175 min1 (ooclpltaJ) and 0.104 min1 (cerebellum), k, = 0.0141 m1n1; V2 = 17; BR = 215 (occipital),91 (cerebellum).
nonspecific distribution volume and is not expected to vaiy
much between regions. In fact, V2. as derived from kinetic
analysis (Table 2), showed more regional variation than V
as measured directly by flumazenil displacement (Table 3).
According to the model, k3 is the parameter that includes
the receptor density (B@, Equation 16), leading theoret
ically to a correlation between VT and k3. No such corre
lation was observed (Table 2). The molecular dissociation
constant, k4 (k@) is expected to be similar across regions,
which was not the case (Table 2). Thus, the rate constants
did not reflect the physiological processes ascribedto them
by the model. However, the regional distribution of the
outcome measure VT was better identified and in accor
FIGURE5. (A)Arterialtime-activitycurveoffreeunmetabolized (1@qioru@i after a bolus of 3.68 ma followed by a constant mnfuslonof 1.12 m@libr(B/I ratio of 3.68 hr) in a 21-yr-oldmale. Valueswere fit to a tilexponentlalmodel (solidlIne)to derivethe peripheralclearance(CI = 6.7 liter/hr/kg).(B) RegionalAOl time activitycurves.Total AOl activityat equllibdum(averageof the last 180 mW@)was 0.61 @CWrnlin the occipital cortax@0.54 @CWmlin the temporalcortex and 0.19 @&CIfmlIn the cerebellum.The average leveloffree parentradlotrecerIn plasma(2.2nCl/mI)duringthe last
2hrwasassumedtobeequaltothefreeredictrecerinthebrain.The
ratiooftotal brainactMtyto thefreegaveVTvaluesof 276,242and 89 Inoccipital, temporal and cerebellar ROIs, respectively.
dance with the known distribution of BDZ receptors in
human brain (50—52).
A similar situation is observed in plasma. Whereas it is
unclear ha defined physiological process is associated with
each exponential, the overall outcome measure, the clear
ance of the compound from the plasma, is adequately de
scribed by a three-exponential model. Similarly, the total
tissue volume of distribution is a well defined outcome
measure.
Theidentifiabilityandphysiologicalmeaningof therate
constants can be improved by constraining the regression
process to values obtained in a region devoid of receptors
(16,44,53).K1,k2ortheirratiocanbederivedfromthese
regions and used in fittingreceptor-richregions. This strat
egy could not be implemented with [‘@I]iomazemldue to
the absence of a region devoid of receptors. Flumazenil
displacement (n = 3) clearly showed that 60% of the activ ity measured in the pons was displaceable. This displace
SPECTMeasurementof Benzod,azepneReceptors•Ab@Darghamat al. 235
6. No arterial sampling. At steady-state, the arterial and
venous plasma concentrations equilibrate, allowing
the measurement of tracer C@ concentration from
one venous blood sample, which further facilitates
the experimental procedure.
Several disadvantagesofthe constant infusiontechnique
must be noted. Although scanning time is reduced, the
total experimental time is increased from 120 miii to 500
min. There is a potential for pump failure and finally, if
subject peripheralclearance is extremely fast or slow, equi
librium may not be reached by the time of scan acquisition.
Low specific activity [‘@I}iomazenilconstant infusion
experiments were used in primates to derive in vivo
[‘@I]iomazenilK@and Bm@(33,35). Derivation of these
parameters with kinetic analysis is complex since the
model becomes nonlinear at significant levels of receptor
occupancy. In contrast, Scatchard analysis can easily be
performed on equilibriumdata. We are in the process of
extending these low specific activity experiments to hu
mans.
Assuming an in vivo K@of 0.59 nM (value measured in
primates with low specific activity SPECT experiments)
(33), the occipital BP value of 244 reported here would
correspond to a Bm,, of 141 nM. This value is close to the
Bm@ measured in vitro in 12 subjects (Bm@ of 162 ± 52 nM). In vitro and in vivo values were thus consistent. In conclusion, both kinetic and equilibrium methods pro
vide adequate measures of regional total equilibriumdis
tribution volume of [@I]iomazenil. Following establish
ment of equilibrium tracer conditions, the injection of
receptor-saturating doses of flumazenil directly measures
the equilibriumdistributionvolume of the nonspecific com
partment. The reproducibility of each method is currently
being tested.
In comparison to PET, quantitation of SPECT activity is
less well developed and more vulnerable to scatter and
attenuation. In lightofthese limitations,the SPECT results
reported here compare favorably with PET measurements
of benzodiazepine receptors in terms of absolute values
and identifiabilityof parameters. The decreased cost and
wider availabilityof SPECF technology may facilitate the
applicationof these research methods to clinical studies.
ACKNOWLEDGMENTS
TheauthorsthankWalterHunkeler,PhD(Hoffman-LaRoche,
Basel, Switzerland) for providing samples of flumazenil and iomazenil; E.O. Smith, 0. Wisnlewski and Quinn Ramsby for theirexcellent technical assistance in collecting and analyzingthe in vivo data; and S. Giddings for performing the in vitro experi
ments.
REFERENCES
1. Roy A, BakayE, HarrisA. Neurotransmitter,receptorandbiochemical changesin monkeycorticalepilepticfoci.BrainRes 1980;206:387-404.
- SavicI, RolandP, SedvallG, PerssonA, PauliS, WidenL In vivo demonstrationof reduced benZodiaZepinereceptor bindingin human epi lepticfoci. Lance: 1988;2:863-866. 3. JohnsonEW,de LancrolleNC, KimJH,et al. “Central―and“penpheral―
benzodiazepinereceptors:oppositechangesin humanepileptogenictissue. Neurology 199242:811—815.
4. OwenF, PoulterM,WaddingtonJL,MaShaIRD,CrowTJ.[3H]Ro05. and[3Hjflunitrazepambindingin kainate-lesionedratstriatwnandtemporal cortex of brains frompatientswith seniledementiaof the Alzheimertype. &ainRes 1983;278:373—375. 5. ShimohamaS, TaniguchiT, FujiwaraM, KameyamaM. Changeinbenzo diazepinereceptorsin Alzheimertype dementia.Ann Newvl 1988;23:404- 4@. 6. PenneyJB, YoungAB. Quantitativeautoradiographyof neurotransmitter receptorsin Huntingtondisease.NeuroIogj@198232:1391-1395.
- WalkerFO, YoungAB, PenneyJB, Dovorini-ZisK, ShoulsonI. Benzodi. azepine and GABA receptors in early Huntington's disease. Neurology 198434:1237—1240.
- Whitehouse PJ, Trifiletti RR, Jones BE, et at. Neurotransmitter receptor alterationsin Huntington'sdisease:autoradiographicand homogenatestud icswithspecialreferenceto benwdiazcpinereceptorcomplexes.AnnNeu _ivi 1985;18:202—210.
- KiuchiY, Koba@ T, TakeuchiJ, ShimizuH, OgataH, Toni M. Ben._ zodiazepinereceptorsincreaseinpostmortembrainofchronic schizophren ice.EwA,th P3ychNewvl Sd 1989;239:71-78.
- BenesFM, VincentSL, Alsterberg0, BirdED, SanGiovanniJP. Increased GABASreceptorbindinginsuperficiallayersofcingulatecortexinschizo phrenics.JNeWV,Sci1992;12:924-929. 11. Squires RF, Lajtha A, SaesderUp E, Palkovits M. Reduced [3Hjflunitrazepambindingin cingulatecortex and hippocampusof postmor tern schizophrenic brains: is selective loss ofglutamatergic neurons associ ated with majorpsychoses?NeurochemRes 1993;18:219-223. 12. HantrayeP, KaijimaM; PrenantC, et al. Centraltype benzodiazepine bindingsites: a positionemissiontomographystudy in the baboon's brain. Neurosci Let: 1984;48:115-120. 13. SamsonY, HantrayeP, BaronJC, SoussalineF, ComarD, MaziereM. Kineticsanddisplacementof [11CJRo 15.1788,a benzodiazepineantago tiist, studiedin humanbrain in vivo by positron tomography.EurJ Phar macol 1985;110:247—251. 14. PerssonA, EhlingE, EriksonI, Ctat. Imagingof[@q-laheledRo 15- to benzodiazepinereceptors in the human brain by positron emissionto mography.IPsychiat Res 1985;19:609-622. 15. ShinotohH, YamasakiT, Inouc0, Ctal. VisUalizatiOnof specificbinding sites of benzodiazc@ in humanbrain.JNuclMed 1986;27:1593-1599. 16. KoeppeR@A,HoithoffVA, FreyKA, KilbournMR. KuhIDE. Coinpart mentalanalysisof['1CJflumazenilkineticsforthe estimationofligandtrans. port rate and receptor distributionusingpositron emissiontomography.I Ce,@bBlood Flow Metab 1991;11:735-744. 17. BeerH.F,BlauensteinPA,HaslerPH,atal.Invitroandinvivoevaluation of iodine-123Ro 16-0154:a new imagingagentfor SPECTinvestigationsof benzodiaze@ receptors.INuciMed 1990,31:1007-1014. 18. InnisR, ZoghbiS, JohnsonE, Ctal. SPECFimagingofthe benzOcliaZepine receptorin non-humanprimatebrainwith [1@'I]Ro16-0154.EurlPharma col 1991;193:249—252. 19. WoodsSW,SeibylJP,GoddardAW,et a!.DynamicSPECFima@ngafter injectionof the benzodiazcpinereceptor ligand [‘@IJiomazenilin healthy humansubjects.Ps@vhüwyRes Newoimaging 1992;45:67—77. 20.CarsonRE.Precisionandaccuracyconsiderationsofphysiologicalquanti ficationin PET.I CerebBloodFlowMetab 1991;11:A45-A50. 21. MintunMA,RaichleME,KilboumMR,WootenOF,WelchMi.A quan. titativemodelfor the invivo assessmentofdrug bindingsiteswith positron emissiontomography.Ann Neurol 1984;15:217—227. 22. FardeL,HallH,EhrinE,Sedvall0. QUantitatiVeanalysisofD2dopamine receptor binding in the living human brain by PET. Science 1986;231:258-
- HolthoffVA, Koeppe R.A,Frey KA, KilbournMR, Kuhi DE. Differentia tion of radioliganddelivety and bindingin the brain: validationof a two compartmentmodelfor [‘1Cjflumazenil.I CerebBloodFlowMetab 1991; 11:745—752.
- PriceJC, MaybergHS, DannalsRF, et al. Measurementof benzodiazepine receptor number and affinityin humans using tracer kinetic modeling, positron emissiontomography,and [“q.flumazenil.I Cei@bBloodFlow Metab 1993:13:656—667.
- Iyo M, Itoh T, YamasakiT Ctal. Quantitativein vivo analysisof benzodi azepinebindingsites in the humanbrain usingpositronemissiontomogra. plw. Neurophannacology1991;30:207-215.
- PauliS, Liljequist5, Farde L, CtaL PET analysisof alcoholwith the brain @ dispositionof Psychophw@nacol1992;107:180-185.
- Pappata S, Samson Y, Chavoix C, Prenant C, Maziere M, Baron JC.
SPECTMeasurementofBenzodiazepineReceptors•Ab@.Darghamatal. 237
Regionalspecificbindingof[―C]Ro15-1788to centraltype benzodiazepine receptors in human brain: quantitativeevaluationby PET. I Cei@bBlood Flow Metab 1988;8:304—313.
- Littoni-E, NeimanJ, PauliS, Ctal. PEr analysisof [llCjflumazenilbinding to benzodiazepinereceptorsin chronicalcohol-dependentmen and healthy controls.PsychiaziyRes Neurounaging 199250:1-13.
- BrouilletE, ChavoixC, Khalili-VarastenM, et at. Quantitativeevaluation of benzodiazepinereceptors in live PappioAnubisbaboonsusingpositron emissiontomography.McI Phwmacol 199038:445-451. 30. CarsonRE, @hanningMA,BlasbergRG,CtaLComparisonof bolusand infusionmethodsfor receptorquantification:applicationto [18fl@y@f@y and positronemissiontOmO&Japhy.ICerebB1OOdFkYWMetab1992;13:24- **42.
- FreyICA,EhrenkauferLE, Beaucage5, AgranoffBW.Quantitativeinvivo** receptor binding I. Theory and application to the muscarinic cholinergic receptor.INeurosci 19855:421-428. 32. KawaiR, CarsonRE, DunnB, NewmanAR, Rice KC, BlasbergRO. Regionalbrain measurementof B@ and K@with the opiate antagonist cyclofoxy:equili1@riumstudies in the conscious rat. I Cereb Blood Flow Metab 1991;11:529—544. 33. LaruelleM, Abi-DarghamA, RattnerZ, Ctal. SPECFmeasurementof benzodiazepinereceptor number and affinityin primate brain: a constant infusionparadigmwith[@‘I]iomazeai1.EurlPhwmacol 1993;230:119-123.
- Laruelle M, Baldwin RM, Rattner 7, at al. SPECF quantificationof [1@iomazenilbindingto benzodiazepinereceptorsin nonhumanprimates. I. Kineticmodelingofsinglebolus @erimenta.ICerebBloodFlowMetab 1994:in press. 35. LaruelleM,Abi-DarghamA,Al.TikritiMS.Ctal.SPECFquantificationof [‘23ljiomazenilbindingto benzodiazepinereceptorsin nonhumanprimates. II. Equilibrium analysis of constant infusion experiments and correlation with in vitro parameters.I Ce,@bBloodFlowMetab 1994:in press.
- Zea-PonceY, BaldwinR, ZoghbiSS, InnisRB.Formationof 14'@I]iodob utane in indodestannylationwith [‘@‘IJioinazcnil:implicationfor the reac lion mechanism. IntIAppI Radlat Isot 1993;45:63—68. 37. McBrideBJ, BaldwinRM, KerrJM, Wu JL. A simplemethodfor the @ preparationof 1231. 1251.labeledIOdObCnZOdiaZepineS.ApplRadiatIso 1991;42:173—175.
- HolmanBL, CaivathoPA, ZimmermanRE, et at. BrainperfusionSPECf using an annular singleciystal camera: initialclinicalexperience.I Nun Med 199031:1456—1461. 39. Zubal10,HarrellC,WoodsSW,etal.Comparisonofquantitationlinearity ofthreebrain SPECrimaginginstrumentsuaing@Fc and 1@'I.INunlMed 1990;31:769—770. 40. InnisRB,Al-TikritiMS.Zoghbi55, et al.SPECFimagingofthebenzodi. azepinereceptor:feasibilityofinvivopotencymeasurementsfromstepwise displacementcuwes.JNuclMed 1991;32:1654-1761. 41. MalisonRT, MillerEG, GreeneR, HofferPB, McCarthy0, InnisRB. Computer-assistedcoregistrationof multisliceSPECF and MRI imagesby fixedexternalfiducials.I ConjpAssirt Tonzogr1992;17:952-960. 42. Zoghbiss, aaid@@inRM,SeibylJP,CtaLPharmacokineticsof theSPECF benzodiaze@ receptor radioligand[‘@‘I]iomazendin human and nonhu man primates.NuciMed Bid 199219:881—888. 43. PhelpSME,HuangSC,HoffmanEl, Scm C, SokoloffL,KuhiDE. Tomo graphicmeasurementof localcerebralglucosemetabolicrate in humans with (F-18)2-fluoro-2-decxy-D-glucose:validationof method.Ann Neural 1979;6:371—382.
- FrostJi, DouglassKH, MaybergHS,Ctal. Multicompartimentalanalysisof “C-carfentanilbinding to opiate receptors in human measured by positron emissiontomography.I CerebBloodFlowMetab 1989,9:398-409. 45. LevenbergK. A methodfor the solutionof certainproblemsin least squares. QUWtApplMath 1944;2:164—168.
- Landlaw EM, DiStefano JJ Ill. Multiexponential, multicompartmental, and noncompartmentalmodeling. II. Data analysis and statistical consider ations.AmIPhysiol 1984;246:R665-R677. 47. AkaikeH. A new look at the statisticalmodelidentification.IEEETrans Auwmat Conir 1974AC19:716-723.
- Carson RE. Parameters estimation in positron emission tomography. In: PhelpsME, MazziottaJC, SchelbertHR. eds. Positronemissionto.nognz phi'.P?inciplecandapplicatlonsforthebrainandthehewi. NewYork: RavenPress; 1986347—390.
- Munson P, ROdbard D. UGAND: a versatile computerized approach for characterizationofligandbindingsystems.AnnBiochem198t@,107:220-239. 50. MohlerH,OkadaT.BenZOdiazepinereceptor:demonstrationinthecentral nelvous system. Science 1977;198:849-851. 51. BraestrupC, AlbrechtenR, SquiresRF. Highdensitiesof bCnZOdiaZCpinC receptorsin humancorticalareas. Natw@1977;269:702-704. 52. MaloteauxJM,LuabeyaMAX,VanisbergMA,et al.Benzodiazepinere ceptors in normalhumanbrain, in Parkinson's disease and in progressive supranuclearpalsy.BrainRes 1988;446:321—332. 53. FardeL, ErikssonL, Blomquist0, HalldinC. Kineticanalysisof central [“CJ@k@ridebindingto D2dopaminereceptorsstudiedby PEr—Acorn parison tothe equih1@riurnanalysis.JCerebBloodFlowMetab 1989;9:696-
- Dubey l(K, McAllisterCB, Inoue M, WilkinsonOR. Plasmabindingand transport of diazepamacross the blood-brainbarrier. No evidence for in vivo enhanceddissociation.I Cli,,Invest 1989;84:1155-1159.
238 The Journal of Nuclear Medicrne•Vol. 35 •No. 2 •February 1994