a reversible tracer analysis approach to the study of effective dopamine turnover

A Reversible Tracer Analysis Approach to the Study ofEffective Dopamine Turnover

*Vesna Sossi, †Doris J. Doudet, and ‡James E. Holden

*University of British Columbia/TRIUMF, †University of British Columbia, Vancouver, British Columbia, Canada; and‡University of Wisconsin, Madison, Wisconsin, U.S.A.

Summary: Changes in dopamine turnover resulting from dis-ease states such as Parkinson’s disease may be reflected incorresponding changes in the kinetics of the positron emissiontomographic tracer [18F]fluorodopa. The authors had previ-ously refined the conventional irreversible-tracer graphical ap-proach to determine both the uptake rate constant Ki and therate constant kloss that describes the slow loss of the trappedkinetic component. Because these parameters change in theopposite sense with disease, their ratios may be more power-fully discriminating than either one alone. The ratio kloss/Ki isindicative of effective dopamine turnover. Its inverse, Ki/kloss,can be interpreted as the effective distribution volume (EDV)

of the specific uptake compartment referred to the fluorodopaconcentration in plasma. Here the authors present a new ap-proach to the estimation of EDV based on reversible-tracergraphical methods. When implemented with a plasma inputfunction, the method evaluates EDV directly. When imple-mented with a tissue input function, the outcome is proportionalto the ratio of the distribution volumes of the specific uptakeand precursor compartments. Comparison of the new and pre-vious approaches strongly validates this alternative approach tothe study of effective dopamine turnover. Key Words: Dopa-minergic system—Effective dopamine turnover—Distributionvolume ratio—Modeling.

The role of the positron emission tomographic tracer[18F]fluorodopa (FD) in the assessment of the integrity ofthe dopaminergic system has been firmly established(Brooks et al., 1990; Garnett et al., 1983). FD is trans-ported across the blood–brain barrier by means of thelarge neutral amino acid system. It is subsequently de-carboxylated by dopa decarboxylase into [18F]fluorodo-pamine, which is then trapped in the presynaptic vesiclesand/or further metabolized into l-3,4-dihydroxy-6-[18F]fluorophenylacetic acid and [18F]6-fluoro-homovanillic acid, which are capable of leaving thebrain. FD in plasma and brain is also metabolized bycatechol-O-methyl-transferase into 3-O-methyl-[18F]fluoro-DOPA, which crosses the blood–brain bar-rier by the same large neutral amino acid system. Theaccumulation of [18F]fluorodopamine in the synapticvesicles is responsible for most of the striatal radioactiv-ity for the first 90 to 120 minutes after FD administration

because [18F]fluorodopamine, like dopamine, cannotcross the blood–brain barrier. During this time, FD thusappears as an irreversibly bound tracer.

This apparent irreversibility is the basis of the graphi-cal approach to the evaluation of FD uptake developedby Patlak et al. (1983) and further refined by Martin et al.(1989). The basic principle of this method is that accu-mulation of the tracer into an irreversible compartmentcan be reliably identified even if the intermediate revers-ible components of the tracer kinetics are not fully iden-tified or known. If steady state between compartments isreached and if an irreversible compartment exists, a lin-ear relationship is obtained between the ratio of the spe-cific compartment and plasma radioactivity and the ratiobetween the running integral and the instantaneous valueof the plasma radioactivity (stretch time). The slope ofthe straight line yields the FD uptake rate constant Ki (Ki

� K1k3/(k2 + k3), where K1 and k2 are the clearancerates from plasma into tissue and from tissue into plasmaand k3 is the rate constant describing the trapping ofbrain FD. Although Ki does not provide separate infor-mation about tracer transport, decarboxylation, and stor-age, it has been proven to discriminate between healthand disease states (Ishikawa et al., 1996; Morrish et al.,1998; Vingerhoets et al., 1996).

Received July 31, 2000; final revision received December 14, 2000;accepted December 15, 2000.

Supported by a grant from the British Columbia Health ResearchFoundation and Vancouver Health Science Center and a TRIUMF LifeScience grant.

Address correspondence and reprint requests to Vesna Sossi, 4004Wesbrook Mall, Vancouver, B.C., V6T 2A3 Canada.

Journal of Cerebral Blood Flow and Metabolism21:469–476 © 2001 The International Society for Cerebral Blood Flow and MetabolismPublished by Lippincott Williams & Wilkins, Inc., Philadelphia

469

The graphical approach has also been extended to in-clude a tissue input function (Patlak and Blasberg, 1985).The time activity course of a region essentially devoid oftrapping, typically the occipital cortex or the cerebellum,is used as input. In this case the slope of the straight line(Kocc) is equal to k2k3/(k2 + k3). An added complicationwith this approach is the presence of 3-O-methyl-[18F]fluoro-DOPA in the reference region, which artifi-cially increases the input function and therefore biasesthe Kocc toward smaller values. Nevertheless, thismethod correlates well with the plasma input approach(Takikawa et al., 1994) and is commonly used. Itpresents two significant advantages over the plasma in-put method: it simplifies the scanning procedure, becauseno blood samples need to be taken, and it eliminates thenoise caused by the plasma radioactivity measurement(Morrish et al., 1998; Vingerhoets et al., 1994, 1996).Such noise may be even more relevant in these studiescompared with those examined in the above references,because in this case the scans are extended to 4 hoursafter injection: at that time, the FD fraction in the plasmais only approximately 2%, and consequently a smallmeasurement error can cause a large variation in the FDfraction estimate, and significant physical radiotracer de-cay has occurred, thus decreasing the statistical accuracyof the radioactivity measurement.

Although FD shows an apparently irreversible behav-ior for the first 90 to 120 minutes after administration innormal subjects, the data obtained from the conventionalgraphical analysis at later times start to deviate from astraight line, exhibiting a progressively decreasing slope.Such a trend is indicative of the slow loss of radioactivityfrom the trapping compartment as a result of [18F]fluo-rodopamine metabolism by catechol-O-methyl-transferase and monoamine oxidase and diffusion of themetabolites out of the brain. An extension of the standardgraphical approach was developed to quantify this re-versibility of the trapping compartment, described by therate constant kloss (Doudet et al., 1998; Holden at al.,1997). The rate constant kloss is a measure of the fre-quency of turnover of the trapped tracer component, andits inverse represents the mean dwell time of that com-ponent in brain tissue. The ratio kloss/Ki is a powerfullydiscriminating indicator of the turnover of the trappedFD compartment (effective dopamine turnover [EDT]).Its inverse Ki/kloss can be interpreted as an effective dis-tribution volume (EDV) of the specific compartmentalone with respect to the plasma tracer concentration andis a similarly discriminating measure of the ability of thetrapping mechanism to store tracer. In Parkinson’s dis-ease, a disease characterized by a progressive loss ofnigrostriatal dopamine terminals in the striatum, the rateof FD uptake decreases and the rate of loss increases.Thus, both EDT and its inverse EDV, because they areratios of these rate constants, are sensitive markers of

disease severity and progression. Nonhuman primatestudies examining animals with varying degrees of1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-induced Parkinsonism have shown the EDT to be a moresensitive marker of disease state than Ki or kloss alone(Doudet et al., 1998). The EDT, or its inverse EDV, maythus be useful in characterizing and staging disease stateswhere an increased [18F]fluorodopamine turnover is ex-pected as a result of loss of storage ability. They mightalso prove useful in understanding and quantifying dif-ferent manifestations of therapy responses, such as treat-ment-induced motor fluctuations in Parkinson’s disease.

Although this modified graphical approach has beenshown to be reliable and robust in a variety of circum-stances, it sometimes fails to converge to a valid solutionwhen the rate constant kloss is high. Given the demon-strated and potential further relevance of the EDV andEDT concepts, we present a new method for their esti-mation that can be implemented using either plasma ortissue input functions. A brief description of the originalmethod and the derivation and validation of the newmethods are presented.

MATERIALS AND METHODS

Modeling methodsThe original method (Holden et al., 1997) is based on the

idea of restoring the apparent irreversible behavior of the tracerby altering the stretch time present in the graphical analysiswith the introduction of a kloss term:

S(t)�Cp(t) = Ki�0

tCp(t�) exp(−kloss(t − t�))dt��Cp(t) + int (1)

where S(t) is the specific tracer concentration in the targetregion, Cp is the tracer concentration in the plasma, and Ki isthe FD uptake rate constant. The specific tracer concentration isobtained after subtracting the radioactivity measured in a re-gion of interest placed on the occipital cortex from the radio-activity measured in a region of interest placed on the striatalregion. The specific tracer concentration is thus determinedunder the assumption that the same 3-O-methyl-[18F]fluoro-DOPA concentration is present in the two regions (Martin etal., 1989). The effect of kloss is to reduce the values of theabscissa, with the largest values undergoing the largest reduc-tion—that is, to shrink the stretch time and to restore thestraight-line behavior of the data. The rate constant kloss isgenerally identified from data obtained between 30 and 240minutes after tracer injection. EDT is then estimated as the ratiobetween kloss and the uptake rate constant Ki obtained fromthe data acquired during the first 90 to 120 minutes aftertracer injection.

In the new method, the deviation of the data from the straightline in the Patlak approach can also be interpreted as an indi-cation of tracer reversibility. Thus, instead of determining kloss

by trying to restore irreversibility into the tracer kinetics, areversible tracer model, similar to the one developed by Loganet al. (1990), can be applied to the data. We have developedsuch a model, which allows for both a plasma and a tissue in-put function.

V. SOSSI ET AL.470

J Cereb Blood Flow Metab, Vol. 21, No. 4, 2001

Reversible plasma input approachFor the sake of simplicity, the following derivation is spe-

cific to the standard two-tissue compartment model (Fig. 1), inwhich the precursor time course P(t) and the specificallytrapped component S(t) are both single compartments:

dP(t)�dt = K1Cp(t) − (k2 + k3)P(t) (2)

dS(t)�dt = k3P(t) − klossS(t) (3)

As above, S(t) is obtained from the radioactivity determinedin a region of interest placed on the striatum after subtraction ofthe radioactivity present in a region of interest placed on theoccipital cortex. This subtraction thus identifies S(t) under twoassumptions: 3-O-methyl-[18F]fluoro-DOPA is present in thesame concentration in the two regions as mentioned earlier, andthe error introduced by the assumption that the precursor con-centration P(t) (free FD) is the same in the two regions is smallcompared with the concentration of the specific striatal activity.The precursor time course P(t) is not directly observable, butthis is not required by the method.

Integration of Eqs. 2 and 3 and rearranging coefficientsyields

�0

TP(t)dt = (K1�(k2 + k3))�

0

TCp(t)dt − (1�(k2 + k3)) P(T) (4)

�0

TS(t) dt = (k3�kloss)�0

TP(t)dt − (1�kloss) S(T) (5)

Substituting Eq. 4 into Eq. 5 yields:

�0

TS(t) dt = (K1k3�(k2 + k3)) × (1�kloss)�0

TCp(t)dt

− (k3�kloss)(1�(k2 + k3))P(T) − (1�kloss) S(T) (6)

The time dependence of the two last terms is eliminated bydividing all terms by S(T):

�0

TS(t) dt�S(T) = (K1k3�(k2 + k3)) × (1�kloss)�0

TCp(t)dt�S(T)

− (k3�kloss)(1�(k2 + k3))P(T)�S(T) − (1�kloss)(7)

The next-to-last term on the right side of Eq. 7 becomes aconstant when a steady state between the precursor and thespecific pools is reached. In this case, a plot of ∫0T S(t) dt/ S(T)vs. ∫0T Cp(t)dt / S(T) yields a straight-line relationship with aslope of (K1k3/(k2 + k3)) × (1/kloss) � Ki/kloss � EDV �1/EDT—that is, the slope of the line is found to be the same

combination of the rate constants that determines the effectivedistribution volume in the original approach.

Reversible tissue input approachA similar approach can also be used to derive the expression

in the case of a reference tissue driving function. Again, thesimplified model in Fig. 1 is assumed. An equation similar toEq. 7 can be derived for a reference region devoid of thespecific process (Logan et al., 1996):

�0

TP�(t) dt�P�(T) = K1��k2��0

TCp(t)dt�P�(T) − 1�k2� (8)

where P�(t) is the tracer time course in the reference regionand K1� and k2� are transfer constants related to that region.Rearranging Eq. 8 to isolate ∫0T Cp(t)dt, substituting this intoEq. 7, and grouping the constant terms into the intercept term(int) yields:

�0

TS(t)dt�S(T) = (K1k3�(k2 + k3)) × (1�kloss)(k2��K1�)

��0

TP�(t)dt + P�(T)�k2�)��S(T) + int (9)

Using the assumption K1�/k2� � K1/k2 (Kuwabara et al.,1993), the coefficient of the first term (graphical slope) can beexpressed as:

(k3K1)�(k2 + k3) × 1�kloss × k2��K1� = (k3k2)�k2 + k3) × 1�kloss= Kocc�kloss = EDVR (10)

where Kocc is the standard graphical uptake rate constant ob-tained with the tissue input method. This ratio, defined here asthe effective distribution volume ratio (EDVR), is the ratio ofthe distribution volumes of the specific and precursor compart-ments, reduced by the factor k2/(k2 + k3). Just as in the case ofthe determination of Kocc, the values of the EDVR determinedby this graphical approach are affected by the presence of 3-O-methyl-[18F]fluoro-DOPA in the reference region. Thus, inanalogy with Kocc (Brooks et al., 1990), empirical evidenceis required to validate the use of the EDVR. Further, Eq. 9includes the reference region efflux constant k2� in the expres-sion used to calculate the values of the graphical abscissa. Forthis to be a truly bloodless method, a population-derived k2�value must be usable, or the term containing k2� must be ofnegligible importance.

Model validationFor the reversible method to be applicable to FD kinetics,

data obtained using Eqs. 7 and 9 must follow a straight linesometime during the extent of the study (linearity). For the

FIG. 1. Model describing [18F]fluorodopa kinetics, including the reversibility of the trapping compartment.

EFFECTIVE DOPAMINE TURNOVER MODELING 471


method to perform reliably, the value of the slope must not beoverly sensitive to the exact starting and ending point of thefitted time range (effect of time range). Because the originalmethod has already been validated (Doudet et al., 1998), it canbe used as reference in the validation of the newly proposedreversible approaches. Therefore, numeric values similar tothose produced by the original method must be obtained withthe reversible plasma input method and proportional valuesmust be produced by the reversible tissue input method. Themethods must be equally able to distinguish various diseasestates (comparison to the original method). In the case of thetissue input function, the method must not be overly sensitiveto the k2� value, because either a population k2� value or no k2�term must be used (effect of k2�).

To investigate these issues, both reversible methods weretested on a study involving nonhuman primate data that ex-plored the sensitivity of EDT to disease severity (Doudet et al.,1998). In addition, the reversible plasma input method wastested with a second nonhuman primate study that investigatedthe effect of pharmacologic interventions on EDT (Doudet etal., 1997).

In set 1, dopamine turnover as a function of disease severity,32 rhesus and cynomolgus monkeys were studied. Sixteen werenormal controls, eight had bilateral lesions of the dopaminenigrostriatal pathway induced by MPTP, and the remainingeight had unilateral lesions. The data derived from these sub-jects were sorted into four groups: normals (16 normal controlmonkeys); an exposed but asymptomatic group containing datafrom the contralateral striata of the monkeys receiving MPTPby unilateral carotid injection (n � 9); an asymptomatic in-jected group containing data from the affected striata of ani-mals not showing Parkinsonism (n � 8); and a symptomaticgroup containing data from the bilaterally lesioned animals andfrom the affected striata of the five unilaterally lesioned ani-mals displaying bradykinesia or hypokinesia and rigidity on thecontralateral body side (n � 9). Three to 5 mCi was injected asa bolus. The scanning sequence consisted of six 30-secondscans, two 1-minute scans, and one 5-minute scan, followed bya succession of 10-minute scans for a total duration of 200minutes (with an interval between 120 and 140 minutes). Datawere acquired on a CTI/Siemens ECAT953B (Knoxville, TN,U.S.A.). Arterial blood samples were drawn throughout the

FIG. 2. Example of the plot obtained using the reversible tracer approach for plasma input (RPI, A) and tissue input (RTI, B). Y-plasmarepresents the integral of S(t) up to the time T divided by S(T). X-plasma is the integral of Cp(t) up to the time T divided by S(T) (see Eq.7). Y-tissue = Y-plasma, and X-tissue is the integral of PN(t) over T divided by S(T) (see Eq. 9, with k2N = 0).

FIG. 3. Example of the effective dopamine turnover calculated using data extending over two different time ranges. (A) Reversibleplasma input method (RPI). (B) Reversible tissue input method (RTI). The line shown is the line of identity.

V. SOSSI ET AL.472


study and 11 samples were analyzed for metabolites. Regionsof interest were placed over the left and right striata on fourconsecutive slices and 12 regions of interest were placed overan area of nonspecific 18F accumulation in the occipital cortexon two consecutive slices. A detailed study description can befound in Doudet et al. (1998).

In set 2, dopamine turnover as a function of intervention,normal juvenile cynomolgus monkeys were studied. Five ani-mals were studied without pharmacologic intervention, sevenafter pretreatment with nitecapone (a peripheral catechol-O-methyl-transferase inhibitor), seven after pretreatment with tol-capone (a catechol-O-methyl-transferase inhibitor that alsocrosses the blood–brain barrier), six after pretreatment with themonoamine oxidase inhibitor deprenyl, and five with a combi-nation of tolcapone and deprenyl. Twenty-four 10-minuteframes were acquired after injection of a 5-mCi bolus of FD.Data were acquired on the University of British Columbia/TRIUMF PETTVI. Arterial samples were acquired throughoutthe duration of the study and five were analyzed for metabo-lites. Data were analyzed similarly as for set 1. A detailed studydescription can be found in Doudet et al. (1997).

RESULTS AND DISCUSSION

Set 1A linear portion of the graph was found in all cases

both for the plasma and tissue input functions, generallybetween 70 and 200 minutes after injection (Fig. 2).

Neither the reversible plasma nor the reversible tissueinput method was found to be greatly sensitive to thestarting or ending time points of the data used for the fitwithin the time range of roughly 70 to 200 minutes afterinjection. Excellent correlations were found between val-ues calculated with the starting points between 60 and 80minutes after tracer injection and ending points between180 and 200 minutes. An example is shown in Fig. 3.

The sensitivity of the EDVR to the value of k2� wasinvestigated by performing the whole analysis for threedifferent k2� values (0.02, 0.05, and 0.08 min-1), as wellas with no k2� term. Minimal differences were observedin the EDVR values obtained with any k2� value (Fig. 4).Even when k2� was as low as 0.02, only a small bias wasobserved when comparing the results with those obtainedwithout the k2�-dependent term in the abscissa. With ak2� of 0.08, the bias was truly negligible. Because thevalues tested were centered over the range of valuessuggested in the literature (Cumming and Gjedde, 1998;Huang et al., 1991; Kuwabara et al., 1993), we chose toignore the term containing k2� in our implementation ofthe reversible plasma input method.

Good numeric agreement was found between the val-ues obtained with the original method and the reversible

FIG. 4. Comparison of the effective dopamine turnover ratio values obtained without the term containing k2N and k2N = 0.02 and withoutthe k2N term and k2N = 0.08. (A) Reversible plasma input method (RPI). (B) Reversible tissue input method (RTI). The line shown is theline of identity.

TABLE 1. Turnover measures in four groups of monkeys with varying degrees ofMPTP-induced Parkinsonism obtained from each method

Original method RPI RTI

Normals (n � 16) 0.25 ± 0.07 (0.27) 0.22 ± 0.07 (0.30) 1.01 ± 0.13 (0.13)Expo (n � 9) 0.34 ± 0.12 (0.36) 0.28 ± 0.11 (0.40) 1.20 ± 0.22 (0.19)Asymp (n � 8) 0.93 ± 0.37 (0.40) 0.80 ± 0.38 (0.48) 2.55 ± 1.05 (0.41)Symp (n � 9) 2.96 ± 1.13 (0.38) 2.60 ± 1.47 (0.56) 5.51 ± 3.08 (0.56)

All values are mean ± SD (fractional SD). Values for the original method are kloss/Ki, and for the RPI methodare 1/EDV. Both have dimensionality cc/mL plasma. The dimensionless values for the RTI method were derivedas 1/EDVR. RPI, reversible plasma input; RTI, reversible tissue input; Expo, exposed; Asymp, asymptomatic;Symp, symptomatic; EDV, effective distribution volume.



plasma model. Table 1 shows mean EDT values, stan-dard deviations, and fractional standard deviations forthe four groups for each method. The fractional standarddeviation for the normal and the exposed group wassmallest for the reversible tissue input method. Thislikely reflects the fact that in this approach, there is noadditional contribution from the plasma FD radioactivitymeasurement to the overall noise. The reversible method

shows a larger spread in the EDT values obtained for thesymptomatic and asymptomatic groups. Because thesegroups are expected to be heterogeneous, no simple con-clusion can be drawn from this comparison.

Figure 5 shows the values obtained for the four groupswith the three methods, and Table 2 contains the corre-lation coefficients between the EDT measures obtainedfrom the original method and those obtained from thenew methods. There was good correlation between thetwo plasma input function methods, both for individualgroups and overall. There was also good correlation be-tween the values obtained from the two new methods inall cases except when the normal and the exposed groupswere considered. This lower correlation is due to the factthat the reversible tissue input method gives a muchtighter distribution of slope values for these two groups,as mentioned above.

Finally, the new methods were examined in terms oftheir ability to separate the four groups. A two-tailed

Figure 5 appearsin the printedissue only.

TABLE 2. Correlation coefficients (r) between EDT valuesobtained from the different methods for all groups together

(Full range) and for each group separately

Original vs. RPI Original vs. RTI RPI vs. RTI

Full range 0.93 0.89 0.96Normals 0.78 0.47 0.44Expo 0.77 0.26 0.46Asym 0.97 0.80 0.83Symp 0.75 0.68 0.94

See Table 1 for explanation of abbreviations.

V. SOSSI ET AL.474


t-test assuming unequal variances was applied to thedata, and the probability values associated with the re-jection of the hypothesis of no difference between thegroups are shown in Table 3. Although the originalmethod was more sensitive in detecting EDT differencesbetween the groups, similar levels of significance wereobtained with all three methods. In particular, the revers-ible tissue input method was the only one that coulddifferentiate the normal from the exposed group.

Set 2Only the results obtained from the reversible plasma

input method and the original method could be comparedin this case. In the case of the tissue input function, acatechol-O-methyl-transferase–inhibiting interventionhad the confounding effect of reducing the presence of3-O-methyl-[18F]fluoro-DOPA in the reference region,thus increasing the Kocc value and decreasing the kloss

value because of a decrease of the input function. There-fore, any observed change would not be entirely due tothe effect of the intervention in the region of interest, butmainly to a change of the input function.

The comparison between the values obtained in theoriginal study and those obtained with the reversibleplasma input method is shown in Fig. 6. There was goodnumeric agreement between the values obtained from the

two methods, and the overall correlation coefficient be-tween the two data sets was 0.92. The ability of the twomethods to separate the effect of interventions was simi-lar: using a two-sided t-test assuming unequal variances,a significant difference was found between baseline andthe group treated with tolcapone with both methods (P� 0.02 for the original method, P � 0.03 with thereversible plasma input method) and between baselineand the group treated with both tolcapone and deprenyl(P � 0.01 with the original method, P � 0.02 with thereversible plasma input method).

ConclusionThe EDT, measured in vivo, has been previously

shown in nonhuman primate studies to be a sensitiveindex of disease severity. The EDT is estimated using FDby quantifying the amount of tracer leakage from thetrapping compartment. A new method to determine EDT(or EDV) and the EDVR has been developed. Themethod is based on the observation that the leakage fromthe trapping compartment that is invariably observed 120minutes after tracer injection (or even earlier in disease)can be viewed as a reversible stage of the FD tracerkinetics. A reversible tracer model can thus be applied tothe data.

The method, based on the graphical approach devel-oped by Logan et al. (1990), can be used with a plasmaor a tissue input function. When the plasma input func-tion is used, the slope obtained from the graph is Ki/kloss.The values obtained from the reversible approach wereproven to correlate well with the values obtained fromthe extended Patlak graphical method, both numericallyand in their ability to separate disease states and effectsof interventions.

When the tissue input is used, the slope of the graph isthe ratio of the distribution volumes of the specific and

TABLE 3. P values associated with the significance level ofthe rejection of the hypothesis of no difference

between groups

Original method RPI RTI

Normal-Asym <0.005 <0.005 <0.005Asym-Symp <0.0005 <0.005 <0.0025Normal-Symp <0.00005 <0.001 <0.002Normal-Expo <0.07 <0.26 <0.05

See Table 1 for explanation of abbreviations.

FIG. 6. Effective dopamine turnover (EDT) values obtained from the original method (A) and the reversible plasma input method (RPI,B) in nonhuman primates after pharmacologic interventions. A, untreated; B, nitecapone; C, tolcapone, D, deprenyl; E, tolcapone +deprenyl. Data in A is reprinted from Doudet et al. (1997) with permission from Elsevier Science.



precursor compartments, reduced by the factor k2/(k2 +k3). As in the case of Kocc, the results obtained from thismethod are biased toward smaller values of the EDVRbecause of the presence of the FD metabolite 3-O-methyl-[18F]fluoro-DOPA in the reference region. Nev-ertheless, the results were shown to correlate well withthose obtained from the reversible plasma input methodand were shown to yield comparable separation betweendisease states. These findings provide empirical supportfor the validity of this approach. In addition, this methodproved to be less susceptible to measurement noise, be-cause no plasma radioactivity measurement is required.The data were found not to be sensitive to the referenceregion efflux constant k2�, thus allowing a truly bloodlessanalysis method.

The extended Patlak method and the reversible tracerapproach presented here both provide an estimate ofEDT, the first one by trying to restore the irreversiblebehavior of the tracer and the other by exploiting thereversible stage of the tracer kinetics. The relative per-formance of the two methods is thus expected to dependon the amount of reversibility and on the rapidity of FDkinetics, and it might vary between nonhuman and hu-man primates. Preliminary results obtained from humandata indicate that this method is well suited to determineEDT in humans.

Acknowledgments: The authors thank the UBC/TRIUMFpositron emission tomography group and the UBC AnimalCare Facility.

REFERENCES

Brooks DJ, Salmon EP, Mathias CJ, Quinn N, Leenders KL, BannisterR, Marsden CD, Frackowiak RSJ (1990) The relationship betweenlocomotor disability, autonomic dysfunction, and the integrity ofthe striatal dopaminergic system in patients with multiple systematrophy, pure autonomic failure, and Parkinson’s disease, studiedwith PET. Brain 113:1539–1552

Cumming P, Gjedde A (1998) Compartmental analysis of dopa decar-boxylation in living brain from dynamic positron emission tomog-raphy. Synapse 29:37–61

Doudet DJ, Chan G, Holden JE, Pate BD, Morrison KS, Calne DB,Ruth TJ (1997) Effect of monoamine oxidase and catechol-O-methyltransferase inhibition on dopamine turnover: a PET studywith 6-[18F]-L-DOPA. Eur J Pharmacol 334:31–38

Doudet DJ, Chan G, Holden JE, McGeer EG, Aigner TA, Wyatt RJ,

Ruth TJ (1998) 6-[18F]Fluoro-L-DOPA PET studies of the turn-over of dopamine in MPTP-Induced Parkinsonism in monkeys.Synapse 29:225–232

Garnett ES, Firnau G, Nahmias C (1983) Dopamine visualized in thebasal ganglia of living man. Nature 305:137–138

Holden JE, Doudet D, Endres CJ, Chan GL, Morrison KS, VingerhoetsFJ, Snow BJ, Pate BD, Sossi V, Buckley KR, Ruth TJ (1997)Graphical analysis of 6-fluoro-L-dopa trapping: effect of inhibitionof catechol-O-methyltransferase. J Nucl Med 38:1568–1574

Huang S, Yu D, Barrio JR, Grafton S, Melega WP, Hoffman JM,Satyamurthy N, Mazziotta JC, Phelps ME (1991) Kinetics andmodeling of L-6-[18F]Fluoro-DOPA in human positron emissiontomographic studies. J Cereb Blood Flow Metab 11:898–913

Ishikawa T, Dhawan V, Chaly T, Margouleff C, Robeson W, Dahl JR,Mandel F, Spetsieris P, Eidelberg D (1996) Clinical significance ofstriatal DOPA decarboxylase activity in Parkinson’s disease. JNucl Med 37:216–222

Kuwabara H, Cumming P, Reith J, Leger G, Diksic M, Evans AC,Gjedde A (1993) Human striatal l-DOPA decarboxylase activityestimated in vivo using 6-[18F]fluoro-DOPA and positron emissiontomography: error analysis and application to normal subjects. JCereb Blood Flow Metab 13:43–56

Logan J, Fowler JS, Volkow ND, Wolf AP, Dewey SL, Schlyer DJ,MacGregor RR, Hitzemann R, Bendriem B, Gatley SJ, ChristmanDR (1990) Graphical analysis of reversible radioligand bindingfrom time-activity measurements applied to [N-11C-methyl-(−)-cocaine PET studies in human subjects. J Cereb Blood Flow Metab10:740–747

Logan J, Fowler JS, Volkow ND, Wang G-J, Ding Y-S, Alexoff D(1996) Distribution volume ratios without blood sampling fromgraphical analysis of PET data. J Cereb Blood Flow Metab16:834–840

Martin WRW, Palmer MR, Patlak CS, Calne DB (1989) Nigrostriatalfunction in man studied with positron emission tomography. AnnNeurol 26:535–542

Morrish PK, Rakshi JS, Bailey DL, Sawle GV, Brooks DJ (1998)Measuring the rate of progression and estimating the preclinicalperiod of Parkinson’s disease with [18F]dopa PET. J Neurol Neu-rosurg Psychiatry 64:314–319

Patlak CS, Blasberg RG, Fenstermacher JD (1983) Graphical evalua-tion of blood-to-brain transfer constants from multiple-time uptakedata. J Cereb Blood Flow Metab 3:1–7

Patlak CS, Blasberg RG (1985) Graphical evaluation of blood-to-braintransfer constants from multiple-time uptake data. J Cereb BloodFlow Metab 5:584–590

Takikawa S, Dhawan V, Chaly T, Robeson W, Dahl R, Zanzi I, MandelF, Spetsieris P, Eidelberg D (1994) Input functions for 6-[fluorine-18]fluorodopa quantitation in parkinsonism: comparative studiesand clinical correlations. J Nucl Med 35:955–963

Vingerhoets FJG, Snow BJ, Schulzer M, Morrison S, Ruth TJ, HoldenJE, Cooper S, Calne DB (1994) Reproducibility of fluorine-18–6-fluorodopa positron emission tomography in normal human sub-jects. J Nucl Med 35:18–24

Vingerhoets FJ, Schulzer M, Ruth TJ, Holden JE, Snow BJ (1996)Reproducibility and discriminating ability of fluorine-18–6-fluoro-l-Dopa PET in Parkinson’s disease. J Nucl Med 37:421–426

V. SOSSI ET AL.476


a reversible tracer analysis approach to the study of effective dopamine turnover

Documents