plex biological fluids. In this article, we describe the chemicalspeciation of Ga(III) and Gd(III), two metal ions commonlyused to prepare imaging agents and MRI contrast agentsrespectively. We hope to show how a knowledge of chemicalspeciation can help to explain existing clinical data and how itcan be used in the design of new and more effective agents.

ThEORYANDMEThODOLOGYMetal Ion Speciation

Speciation modeling ofblood plasma was pioneered by Pemn ina series of papers (1). At the same time Sillen (2 ), Morel (3),Martell (4) and others worked on natural waters and urine. Theseearly models were restricted to one or two metal ions with a limitednumber of ligands. The development of faster computers withlarger memories enabled these early models to be expanded intomulticomponent systems. In the blood model developed by May etal. (5), 7 metal ions and 40 ligands were considered simultaneously.

An essential process in the development ofany speciation modelis to define a series of chemical equilibria which represent thesystem under investigation. In defining the equilibria all thechemical species involved and the equilibrium constants for thereactions have to be specified (Eq. 1). This is the area of greatestdifficulty in speciation modelling as accurate equilibrium constantsare not always available. Indeed, in some cases, all the possibleequilibria have not been identified.

Si —13j@ X@((t.J),

Metal chelate ions are commonly used in medical diagnostic imaging as MRI contrast or imagingagents.The efficacy of these metalsdepends on their in vivo behavior, which in turn depends on their invivo speciation. Methods: A computer model has been used tosimulate the speciation of Ga3@ and Gd3@ in blood plasma. Themodel has been tested against known clinical data and then used toinvestigate Ga3@ uptake by tumor cells. The iatrogenic effect of agadopentetic acid enhanced MRI scan upon the biodistilbution of67Gacitratehasalsobeencalculated.Results:Thespeciationof

@3±calculated using the computer model is concordant with

clinical data. The results support transfernn mediated uptake ofGa3@ by tumor cells but also account for GaQII)biodistributionobserved inhypotransfemnemicsubjects. Ina study of the effectofgadopenteticacidupon67Gagalliumcitrate,neitherresidualDTPAnor [Gd(DTPA)J2 cause signfficant changes in the speciation ofGa(III).The calculations show that dissociation of 4% of the administered gadopentetic acid results in the formation of a mixed, Gd(III)and Ga(III),metal transfernn complex and a 100-fold increase in theconcentration of [Ga(OH)@J.Conclusion: Computer simulation is avaluable tool which can be used to explain/understand in vivobehavior of radioactive metal ions.Key Words: gallium-67;Gd(III);computersimulation;metalspeciation

J Nuci Med 1996;37:379-386

I@ recentyears,diagnosticpatientimaginghasrapidlyexpanded from traditional radiological techniques to a host of newmodalities. These include CT scanning, MRI and radionuclideimaging, both two-dimensional and SPECT. Many of theradiopharmaceuticals used as imaging agents in the lattertechnique are coordination complexes of metal ions (e.g.,99mTcDTpA 99mTc..DMSA and 67Ga-citrate), as are severalcontrast media used in MRI are (e.g., Gd-DTPA and GdDOTA). Thus, the chemical equilibria in solution ofboth thesetypes of agents is important as it can affect their behavior invivo. Most inorganic complexes in solution are labile, undergoing rapid chemical exchange. The possibility therefore existsthat, upon injection, the imaging agent or contrast agent maydissociate and interact with ions present in plasma to give riseto a totally different set of chemical species in solution. Thebehavior of these agents is then a function of this newspeciation.

Speciation describes the concentration and composition ofevery species in a chemical sample including ions and undissociated molecules. It is clearly important in determining thebiodistribution, excretion and toxicity of an element. In dynamic systems such as blood plasma, however, it is often verydifficult, if not impossible, to determine the speciation of anelement. Therefore, powerful computer models have to be used.These permit extrapolation from simple systems to more com

Recalved Oct. 31, 1994; revision accepted Aug. 16, 1995.For correspondence or repnnts contact: G.E. Jackson, PhD, Department of Chemis

try, University of Cape Town, Rortdebosch 7700, South Africa.

Eq.1

where S@is the species concentration, /3@the equilibrium constant,xi thefreecomponentconcentrationandk(ij) thecomponentstoichiometric coefficient.

The series ofequilibria, together with the total or free componentconcentrations constitute the computer model or database of thesystem. From these data, a series of simultaneous, mass-balanceequations can be set up for the total component concentrations (Tn)as a sum of all the individual species concentrations:

T@= X@+ ISj k(i,j). Eq.2

The mass-balance equations can then be solved iteratively for thefree component concentrations. There are many computer programs that will do this. In this study we have used ECCLES (6).Substitution back into Equation 1 yields the individual speciesconcentration.

Metal ions in blood plasma can exist in four distinct forms: themetal ion could be irreversibly bound to protein (e.g., copper boundto ceruloplasmin), reversibly bound to protein (e.g., copper boundto serum albumin), complexed to low-molecular-mass ligands, andbound to water as the aqua ion. The first form is not of interest inspeciation modeling as the metal ion is not part of the equilibriabeing considered unless released from the protein by chemicaldegradation.The aqua-ion is a low-molar-masscomplex but

METALION SPECIATIONIN BLOODPLASMA•Jackson and Byrne 379

Metal Ion Speciation in Blood Plasma:Gallium-67-Citrate and MRJ Contrast AgentsGraham E. Jackson and Michael J. ByrneDepartments of Chemistry and Nuclear Medicine, University of Cape Town, Rondebosch, South Africa

L p qr@g@3tLpqrlogf3Phosphate

1 1118.8Oxalate1106.4511010.012012.3811219.513017.86Citrate*

1 1111.611—17.111010.02Malonate1102.412015.31203.6Hydroxide

1 0—1—3.31Transfernn11018.110—2—6.7621035.210—3—11.16Salicylate11016.110—4—17.17Tartrate11018.5Cysteinate

1 1016.1Serinate1109.011118.411110.511220.5DTPA11024.3DFO

1 1028.1711128.511 —117.9EDTA1

11 10 121.022.8*N@f@

ligandH3citrate.tpGA+ qL + *1 = GaplqHr.

TABLE ITotal Ugand and Free Metal-Ion Concentrations Used as Blood

Plasma Model

Com@ Concentrst@n@ Component Concentra@n(@

TABLE 2Binary Forrnatkn Constants Used as Bkod Model Database for

GalliumQll)

Setum albumin7.2 x iO@Tyrosinate5.8 xiO@Transfernn2.5x 1O@Valinate2.3 xio@Alanate3.7x iO@Carbonate2.5 x1O_2Aminobutyrate2.4x105Phosphate1.6x103Arginine9.5

x iO@Thiocyanate1.4 xiO@Asparaginate5.5x iO@Silicate1.4 xio@Aspartate5.0x 1O_6Sulphate2.1 xiO@Cysteinate2.3x io@Ammonia2.4 xlO@Cystinate4.0x iO@Citrate1.1 xio@Citrullinate2.7x iO@Lactate1.8 xi0@Glutamate4.8x iO@5Malate3.5 xiO@Glutaminate5.2x iO@Oxalate1.2 xiO@Glycinate2.4x iO@Pyruvate9.5 x1O@Histidinate8.5x 1O@Salicytate5.0 x1O_6Histamine1.0x 108Succinate4.2 xio@Hydroxyprolinate7.0x 1O_6Ascorbate4.3 xiO@Isoleucinate6.5x iO@OW1.2 x1O_6Leucinate1.2x iO@Ca@@xio@Lysinate1.8x104Mg@@5.2x104Methionate2.9x105Cu@1.Ox1O@°Omithinate5.8

x iO@Fe@@1.0 x10_liPhenylalanate6.4x105Fe3@1.Ox1O@Prolinate2.1

x iO@Pb@1.0 x1O_14Setinate1.2x iO-@Mn2@1.0 x1O_12Threoninate1.5x iO@Zn@1.0 xio@Tryptophanate1

.0 x iO@

species at equilibrium. To the original database we have added theappropriate constants for Gd(III) (7) and Ga(III). Constants forGa(III) were, as far as possible, extracted from the literature (8,9).In cases where the same system had been studied by severalauthors, the results were evaluated critically. Where a constant waspotentiallyimportantbut for whichno datawereavailable,valueswere estimated either by using a linear free-energy relationshipbetween Fe(III) and Ga(III) or by chemical analogy. The final setof Ga(III) equilibrium constants used in this study are given inTable 2 and the component concentrations in Table 1.

Mixed-ligand complex formation occurs widely in biologicalfluids and has been shown to be important (10). Since little data isavailable on Ga(III) ternary complexes, binary constants were usedto estimate the stability of these mixed complexes (6).

RESULTS AND DISCUSSION

Validation of ModelUnfortunately the measurement of the in vivo speciation of

Ga(III) at nM concentrations is beyond the limit of modernanalytical chemistry. Hence, it is not possible to test ourcomputer model directly. Instead, the calculated speciation hasto be tested, by inference, against clinical observations. This hasbeen done using the normal biodistribution of 67Ga as well asdesferrioxamine (Desferal, DFO) and Fe(III) induced changes.The effect of bicarbonate has also been calculated.

GalliumThe chemistry of gallium has been comprehensively re

viewed by Moerlein and Welch (11 ) and Hayes and Hubner(12). In aqueous solution, the only oxidation state of importance is +3. The binding in Ga(III) is mainly electrostatic andhence size is important. The ionic radius of Ga(III) is 76 pm,which is quite similar to the ionic radius ofhigh-spin Fe(III) (79pm) (13) andresultsin a goodcorrelationbetweenthe stabilityof high-spinFe(III) complexesand Ga(III) complexes.

Gallium(III) is extensively hydrolyzed in solution (14). Thishydrolysis canbe representedby Equations3—6in which, forconvenience, waters of coordination have been omitted.

[Ga]3@+ H2O = [Ga(OH)]2@ + H@, Eq. 3

because of its importance is treated as a separate form of the metalion. It must be present in all aqueous equilibrium systems and isoftenreferredto as the free (uncomplexed)ion eventhoughit iscomplexed to solvent.

In blood plasma, the total metal ion concentration is much lessthan the total protein concentration and the free metal ion concentration is very much lessthanthe concentrationofthe metal-proteincomplex. This meansthat until a substantialportion ofthe metal isstripped off the protein, the free metal ion concentration remainsconstant, i.e., is effectively buffered by protein binding. However,if an exogenously administered metal ion is used in high concentrations (e.g., gadolinium(III) or carrier gallium(III)) the proteinbinding capacity may easily be exceeded. In this case the free metalconcentrationwill not remainconstant.Also, metal uptakebyproteins is often slow and so a higher than thermodynamicallyexpected free metal ion concentration may exist before equilibriumis established. Notwithstanding these problems, the buffering of thefree metal concentration by the protein can be used to simplify thesimulation calculations. During the calculation, the free metal ionconcentrationis fixed ratherthanthetotal concentration,but careis taken that the total concentration does not exceed known plasmalevels. The free metal concentrations used are given in Table 1.

Because of the slow complexation kinetics of proteins, twoextreme simulation conditions exist, one in which protein bindingis neglected and one in which protein binding is explicitly includedin the calculation. These two simulation conditions serve toillustratethe speciationimmediatelyafter injectionand at sometime later when equilibrium with the protein has been established.

B@od Piasma D@The number of possible chemical equilibrium reactions that can

occur in blood plasma are legion. The medium containsa numberof different metal ions and a myriad of low molar mass ligands.May et al. (5) have constructed a computer model ofblood plasmain an attempt to calculate the concentration of all the possible

380 THEJOURNALOFNUCLEARMEDICINE•Vol. 37 •No. 2 •February1996

Eq. 7

c@D)

FIGURE 2. Calculatedspeciationofgallium(III)inbloodplasmaat pH 7.4 inthe presence and absence of Tf.Total Ga(IlI)= 1O@,Tf = 25@ M. Othercomponent concentrationsas given inTable 1.

leaves typically 50 p.M of sites available for the coordination ofother metal ions. Fe3@binding is too strong (log K1 = 20.7 andlog K2 = 19.4)(15) for it to be displacedby Ga3t The resultof introducing transferrin into our model of blood plasma are

Eq. 4 shown in Figure 2. Most of the Ga3@ is bound to Tf (99.9%),while the remaining metal ion is distributed among the same

Eq. 5 low molecular mass ligands as before.Eq. 6 The above results, calculated using our model of blood

plasma, can be compared with observationsin the literature.Clinically, it has been observed that l0%—20% of 67Ga isexcreted through the kidneys within the first 24 hr (19).Another lO%—20%(20) is excretedthroughthe gastrointestinaltract via the bowel wall rather than through the biliary system.The remaining Ga(III) is concentrated in the skeleton (includingmarrow) (24%) and other soft tissue (34%). The same resultsare found for the injection of either Ga3@ or its citrate complex(21 ). Our simulation results are in accord with these clinicalobservations in that, irrespective of the presence of citrate, ourresults show that, the major species present in the injectingsolution is [Ga(OH)4]. Complexation of metal ions by Tf isrelatively slow (1 7,18). Thus, uponinjection,Ga(III) is initiallydistributed in plasma as charged low molar mass complexes butat equilibrium, would become incorporated into Tf, accordingto our calculations. This change in speciation with time is inaccordance with observed changes in tissue distribution withtime and the time variation of 67Ga scans.

Desferrioxamine(Desferal,DFO) hasbeenusedto improvelesion contrast in humans (33) and animals (34). Within 12 hrof 67Gaadministration, DFO is able to remove the radionuclidethat has localized in tissue. After 24 hi, however, it is no longerable to mobilize the metal ion. If DFO is administered prior to67Ga-citrate, the metal ion is rapidly excreted in the urine andno transferrin binding occurs (35). Repeat 67Ga scans afterdiscontinuation of the DFO treatment result in normal diagnostic images.To test our model of blood plasma against theseobservations, the possibility of Ga(III) binding to lactoferrmn(LF) has to be considered. Figure 3 illustrates these results andshows how the amount of Ga3@ bound to Tf or Lf changes withthe concentration ofDFO. At a concentration of3.2 X lO_6 M,DFO can remove 50% of the gallium from Tf but is able toremove the same amount of gallium from Lf only at aconcentration of l0@ M. This last concentration is unrealisticfor clinical use in humans, but it illustrates the result that DFOincreases the excretion of Tf-bound gallium but not Lf-boundgallium. Also, the results show that if DFO is administered prior

pHFiGURE1.(A)Speciationofa iO@M gallium(JII)aqueoussolutionand(B)inthe presence of 4 x iO@ M citrate.

[Ga(OH)]2@ + H2O = [Ga(OH)2]@ + H@,

[Ga(OH)2]@ + H20 = [Ga(OH)3] + H@,

[Ga(OH)3] + H20 = [Ga(OH)4] + H@.

The equilibrium constant for each of the steps above is givenby:

- [Ga(OH)jH@]

Kn [Ga(OH)n_i]

The hydrolysis constants of Ga(III), together with the totalconcentration of the metal ion, can be used to calculate thespeciationin solution. This is shown in Figure 1A for a 1 Xio-9 Msolutionof GaCl3.Thisconcentrationof galliumistypical for a —5mCi injection of [67Ga]Ga-citrate. As can beseen, even for this relatively simple system, a number of speciesexist in equilibrium. As the pH of the solution is raised thespeciation changes with the concentration of [Ga(OH)3] increasing and then decreasing.

To prevent hydrolysis at physiological pH, many radiopharmaceuticalpreparationsof 67Ga(III)includecitrate.Figure lb showsthe effect that citrate has upon Ga(III) speciation. At pH values<6.3 the major species present in solution is [Ga(citrate)OH]. AtpH values >6.3, however, the [Ga(OH)41 species predominates.These calculations serve to illustrate the complexity of speciationin equilibrium situations.

Upon intravenousinjectionofa gallium radiopharmaceutical,the number of possible complexes in solution are numerous.This is because blood plasma is a highly coordinating mediumcontaining many potential ligands. Not only are there a numberof low molar mass ligands like phosphate,carbonate,aminoacids and carboxylic acids, but there are also a number ofproteins which are potential coordinators of gallium. Figure 2shows the speciation ofgallium calculated using our transferrinfree model of blood plasma. At pH 7.4, the major speciespresent is [Ga(OH)4]. At blood plasma concentrations, citrateand phosphate complexes are not important.

Transferrin(TO, which is presentin blood plasma,is anavidbinder of Ga3@ (log K1 = 18.1, log K2 = 17.1) (15). Theprotein has two binding sites per molecule and under normalconditions 30% of these sites are occupied by Fe3@ (16). This

METAL ION SPECIATIONIN BLOOD PLASMA•Jackson and Byrne 381

Species NBPtHP subjectsNBP + 50@ M UNBP+ 0.014M

@2+NBP, pH6.4[Ga@TfJ*

4.0x1094.0x1094.5x10@4.0x1094.0x109[Ga]@e5.4 x 10_255.4 x 10_246.0 x 10275.4 x 10_258.8 x 1022[Ga(OH)@]

1.4x10@21.4x10@1.6x10141.4x10@22.4x10@3[Ga(OH)@]5.8x10145.8x10@36.5x10@65.8x10@49.7x10@4[Ga(citXOH)]

4.5 x 10154.5 x 10_145.0 x 10_174.5 x i0@@7.1 x10_13[Ga-Lf]——4.0 xiO@——•Total

[6a3÷] = 4.05 x iO@M.tNBP= normalbloodplasma;Tf = 25pM.*1@f

= 2.5 pM.

decreased liver activity is seen (36—38). A similar picture isseen in iron-overload patients (39). When our computer modelis tested against these observations the results indicate that theFe(III) displaces Ga(III) from its transferrmn binding site therebyincreasing its excretion and lowering the background activity.The [Ga(Tf)] + [Ga(Lf)] already accumulated at the lesion siteis less accessibleto the Fe(III) and so improved contrast isobtained. The resultant biodistribution is consistent with thecomputer simulation model excluding Tf binding (Fig. 2).

Finally, the model was tested against the effect of changes inbicarbonate concentration upon the speciation of gallium.Staker et al. (51 ) found that when using gallium citrate,97%_99% Of the gallium is bound to molecules with a MW> 10,000 if the bicarbonate concentration is above 13 mM. Asthe bicarbonate concentration decreases, the high molecularweight retention decreases. At a bicarbonate concentration of 7mM, only 83% of the gallium is bound in this way. Changingthe bicarbonate concentration in our model gives similar resultsexceptthat the cut off is —2mM bicarbonaterather than the 7mM found by Staker et al. (51 ). These figures, however, aredependent on the pH of the solution which is not specified byStaker et al. (51).

Application of ModelHaving validated the computer model of gallium speciation

in blood plasma in a qualitative manner, we then applied it toproblems associated with gallium biodistribution.

It has been proposed (22,23) that gallium is taken up intocells as low molar mass complexes or localization is mediatedby Tf-specific receptors on cell membranes with the galliumbeing bound to the Tf. Weiner et al. (24) showed in man andSohn et al. (25) showed in mice that in hypotransferrinemic(HP) subjects that have severely depressed Tf levels (decreasedfrom 25 to 2.5 x l0_6 M), uptake of67Ga by nonosseous tissueis depressed relative to normal subjects. Increased renal excretion and decreased liver uptake is seen. Computer simulationusing the total component concentrations typical of humanblood plasmabut with the available metal binding site concentration ofTfdecreased from 50 to 5 X 106 M shows (Table 3)that while most of the gallium will still be bound to Tf, the[Ga(OH)4] concentration would increase from 0.03% to 0.3%of the total Ga3@. This 10-fold increase in low molar massgallium could explain the increased excretion and hence lowertissue uptake of °7Gawhich is seen clinically. If the results ofWeiner et al. (24) and Sohn et al. (25) are expressed as%(retained dose)/g then similar tissue distribution of 67Ga innormal and HP subjects is seen (blood-to-liver ratio in normaland HP subjects: man, 1:9 and 1:12; mice 1:14 and I : I2). This

A

E

0)00

B

100

= 80

@ 60

c@0) 4000

20

4.5 5 5.5 6 6.5 7 7.5 8

-log(DFO CONG.)

,-% — [Ga-DFOI

4 4.5 5 5.5 6 6.5 7 7.5 8

-log(DFO CONC.)

FIGURE3 DependenceofGa(lll)speciationon DFOconcentra@oninbloodplasma at pH 7.4. (A)TotalGaQII)= iO@M,Tf = 25 .@M;(B)TotalGa(lIl)=iO@ M, Tf = Lf = 2.5 x 10@ M. OthercomponentconcentrationsareasgiveninTable 1.

to 67Ga-citrate no [Ga(Tf)] is formed and so the metal ion israpidly excreted.

Metal ions can be used to enhance the contrast of 67Gaimages. Thus Fe(III), when administered 24 hr after a tracerinjection of 67Ga-citrate, lowers the whole-body retention of67Ga resulting in lower background activity (22).

If, however, Fe(III) or carrier levels of Ga(III) are administered prior to the 67Ga-citrate injection, then the transferrmnbinding sites are saturated and elevated kidney and spleen and

TABLE 3CalculatedSpecies Distributionof GaQII)Complexes in Blood Plasma*

382 THEJOURNALOFNUCLEARMEDICINE•Vol. 37 •No. 2 •February1996

ever, this correlation fails. Kaplan et al. (28) and Sturrock et al.(29) demonstrated a Tf-independent mechanism in which therate of iron uptake correlates with the low molar mass concentration of the metal ion. If such a mechanism is postulated for67Ga, then the increased uptake in XS63 tumor bearing HP mice(25) is accounted for by the speciation calculations. Thetumor-to-liver 67Ga uptake ratio in normal mice is 1.04 while inHP mice it is 9.04 (25). This correlates well with the 10-foldincrease in the low molar mass 67Ga concentration of HPsubjects (Table 3).

Other explanations have been put forward to account for thevaried 67Ga uptake of different tumors. One proposal is thatgallium is displaced from Tf by the high concentration ofintracellular Ca2@ (30). The present model cannot be applied tointracellular fluids as the component concentrations are different to blood plasma, but the plasma results (Table 3) show thatTf-mediated uptake ofgallium is a viable mechanism as most ofthe metal ion circulating in the plasma is bound to this protein.On the other hand, the results also show that a 10-fold increasein calcium plasma concentration does not affect gallium speciation.

Another hypothesis is that a local decrease in pH results indissociation of gallium followed by binding to tumor protein(31 ). In general, equilibrium constants between a metal ion anda protein are measured as conditional constants KC0fld,whichcan only be used under the same conditions at which they weredetermined. The equilibrium is therefore represented by:

M + Protein = [M . Protein],

C0

.04-.

U)•0

•00

1O@@ 1O@TotalGd(lll) (mol dm3)

FIGURE4. Calculateddisttibutkxiof GdQII)in t@oodplasma at pH 7.4, as afunction of total metal concentration in the presence and absence of Tfbinding.The coordinated ligands and their stoichiometryare as indicated.If = transferrin;cit = citrate;sal = salicylate.

is consistent with [Ga(Tf)] being the major gallium complex inboth groups.

Sohn et al. (25) reported increased uptake of67Ga by bone inHP mice. The bone-to-liver uptake ratio increases from 1.4 to8.8, a six-fold increase which is similar to the predicted 10-foldincrease in [Ga(OH)4} concentration. The high affinity ofGa(III) for phosphate is consistent with its localization in bone,which is thought to be Tf-independent (25).

There is much controversy surrounding the mechanism bywhich 67Ga accumulates in tumors (23,26). That gallium canbind to Tf and enter tumor cells is well documented (22). Thereis a good correlation between gallium uptake and Tf(23) and Tfreceptor (TfR) (23,27) concentration. For some tumors, how

Eq. 8

where no account is taken of possible involvement by otherligands. In the case of Tf, the metal is bound by two tyrosineresidues, two histidine residues, a water molecule and a bicarbonate anion. At the same time there is a loss of three protonsfrom the protein (16). The equilibrium can therefore be represented by the equation:

M3@+ H3Tf + HCO@ = [M . Tf HCO3] + 3H@, Eq. 9

A

•0C!,

B

•0

FiGURE 5. CalCUlated speciation ofGdQII)contrast agents inblood plasma atpH 7.4 as a functionof concentration.(A)[Gd(DTPA)]2-,(B)[Gd(EDTA)].Thecoordinated ligands and their StOiChkxnetryare as indicated. Tf = transfenin; cit =citrate;sal = salicylate.

1@ I 0.8 10@ 10.6 10@'

Total concentration/mol dm-3

383METALION SPECIATIONIN BLOODPLASMA•Jackson and Byrne

Bkod plasma, 10 nM

Species Conc.(M)DTPA,

9.4 nM [@J(Ijl@pA)J2@pM GdQll)2nMDWATotal

Ga@ (%)Conc. (M)TotalGa3@(%)3.0x iO@ 7.6 x 10@0

2.2 x iO@1.4 x 10@

7.0 x 10_145.8 x 10_131.2 x io@@3.0 x 10_22c ‘@@.-

25.973.60.50.0

0.00.00.0

1.0 x io@@2.5 x io@@4.3 x io@@1.0 x 10_142.7 x io@@4.0x 10_25

0.10.00.00.00.00.0

Essentials ofnuclear medicine science. Baltimore: Williams and Wilkins; 1987:165—188.

37. Sephton R, Martin I. Modification of distribution of gallium-67 in man by administration ofiron. BrfRadiol I980;53:572—575.

38. Ohkubo Y, Araki S, Abe K, Takasu S, Kohno H, Kubodera A. The effect of FeCl3 onthe accumulation of gallium-67 into inflammatory and normal tissue. Ann Nuci Med1988;2:59—62.

39. Loesberg AC, Martin WB. Altered biodistribution of 67Ga-citrate in an ironoverloaded patient with sickle cell disease. Clin Nuci Med I993; 19: I57—I59.

40. Moeller T. The lanthanides. In: Bailar IC, Emeleus Hi, Nyholm R, TrotmanDickenson AF, eds., (‘omprehensive inorganic chemistry, vol. 4. Oxford: PergamonPress; 1973:1-104.

41 . Lauffer RB. Paramagnetic metal complexes as water proton relaxation agents for NMRimaging: theory and design. Chem Rev 1987;87:901—927.

42. Reuben I. Gadolinium(lII) as a paramagnetic probe for proton relaxation studies ofbiological macromolecules: binding to bovine serum albumin. Biochemistry 1971 ;l0:2834—2838.

43. Zak 0, Aisen P. Spectroscopic and thermodynamic studies on the binding ofgadolinium(III) to human serum transferrin. Biochemistry I988;27: I075—1080.

44. Harris WR, Chen Y. Difference ultraviolet spectroscopic studies on the binding oflanthanides to human serum transferrin. lnorg Chem 1992;31 :5001—5006.

45. Harris WR. Binding constants for neodymium(III) and samarium(lII) with humanserum transferrin. Inorg Chem l986;25:2041—2045.

46. Weinmann H-i, Brash R, Press W, Weskey G. Characteristics of GdDTPA complex:a potential nmr contrast agent. Am J Roentgenol 1984;l42:619—624.

47. Cacheris W, Quay 5, Rocklage S. The relationship between thermodynamics and thetoxicity of gadolinium complexes. Magn Reson Imaging l990;8:467—48l.

48. Hattner RS, White DL. Gallium-67/stable gadolinium antagonism: MRI contrast agentmarkedly alters the normal biodistribution of gallium-67. J Nuci Med 1990;31 :1844—1846.

49. Wiggins I, Goldstein H, Weinmann H. Gallium-67/stable gadolinium antagonism. JNuciMed l99l;32:1830—183l.

50. Weinmann H-i, Press WR, Raduchel B, Platzek I, Schmitt-Willich 1-I,Vogler H.Characteristics of Gd-DTPA and new derivatives. In: Bydder G, Felix R, Bucheler E,et al., eds. Contrast media in MRI. Intemational Workshop Berlin. Bussum, TheNetherlands: Medicorm Europe; 1990:19—30.

21 . Nelson BM, Hayes RL, Edwards CL, Kniseley RM, Andrews GA. Distribution ofgallium in human tissues after intravenous administration. J Nuci Med 1972;13:92—100.

22. Hoffer P. Gallium: mechanisms.J Nuc! Med I980;2I:282—285.23. Weiner R. The role of transfemn and other receptors in the mechanism of 67Ga

localization. Nuci Med Biol I990;I7:141—149.24. Weiner RE, Spencer RP, Dambro Ti, Klein BE. Gallium-67 distribution in a man with

a decrease in both transferrin and hepatic gallium-67 concentration. J Nuci Med1992;33:1701—I703.

25. Sohn M-H, iones B, Whiting i, Datz F, Lynch R, Morton K. Distribution of 67Ga innormal and hypotransferremic tumor-bearing mice. J Nuci Med I993;34:2135—2143.

26. Hayes RL, Rafter ii, Byrd BL, Carlton iE. Studies of the in vivo entry of 67Ga intonormal and malignant tissue. J Nuci Med I98 1;22:325—332.

27. Eckelman WC. The applicationof receptortheory to receptor-bindingand enzymebinding oncological radiopharmaceuticals. Nuci Med Biol 1994;21:759—769.

28. Kaplan J, iordan I, Sturrock A. Regulation ofthe transferrin-independent iron transportsystem in cultured cells. J Biol Chem 1991;266:2997—3004.

29. Sturrock A, Alexander J, Lamb J, Craven CM, Kaplan I. Characterization of atransferrin-independent uptake system for iron in HeLa cells. J Biol Chem 1990;265:3139—3145.

30. Anghileri L, HeidbrederM. On the mechanismof accumulationof 67Gaby tumors.Oncology 1977;34:74—77.

3 1. Vallabhajosula SR. Harwig iF, Wolf W. Effect of pH on tumor cell uptake ofradiogallium in vitro and in vivo. Eur J Nuci Med 1982;7:462—468.

32. Harris WR. Thermodynamicsofgallium complexationby humanlactofemn.Biochemislrv I986;25:803— 808.

33. Koizumi K, Tonami N, Hisada K. Deferoxamine mesylate enhancement of 67Gatumor-to-blood ratios and tumor imaging. Eur J NucI Med 1982;7:229—233.

34. Oster ZH, Som P. Sacker DF, Atkins HL. The effects of deferoxamine mesylate ongallium-67 distribution in normal and abscess-bearing animals. J Nuci Med I980;21:42 1—425.

35. Baker DL, Manno CS. Rapid excretion of gallium-67 isotope in an iron-overloadedpatient receiving high-dose intravenous deferoxamine. Am J Hematol I988;29:230—232.

36. Hodges R. latrogenic alterations in the biodistribution ofradiotracers as a result of drugtherapy: theoretical considerations. In: Hladik W, Saha G, Study K, Friedman B, eds.

TABLE 4Calculated Species Dist,ibution of Ga(lll)Complexes*

[Ga(T1@HCO@][GdGa(rf)(HCO@J[Ga(OF@[Ga@(rfXHCO3),J[Ga(OH)@][Ga(citrateXOH)1[Ga(DTPA)r[Ga@

with the clinical observation that gadopentetic acid is excretedintact while the Gd(III) is lost from [Gd(EDTA)] (46).

One of the side effects predicted by our simulation results isthat when Gd3@ is released from a DTPA or EDTA complex,significant amounts ofzinc complex (—48% ofplasma zinc) areformed with the liberated ligand. These zinc complexes enhancethe excretion of zinc so that animals receiving multiple highdoses ofcontrast agent may show signs ofzinc deficiency. Suchsigns have been reported for rats receiving 5.0 mmole . kg 1a modified DTPA gadolinium(III) complex, intravenously threetimes a week for 3 wk (47).

latrogenic InteractionsRecently, Hattner and White (48) reported the interference of

[Gd(DTPA)]2 upon the biodistributionof 67Ga-citrateobservedin a child at 96 hr. These authors noted that an injection of67Ga-citrate 4 hr after a [Gd(DTPA)}2-enhanced MRI scanresulted in unexpected renal uptake of the radioisotope togetherwith skeletal uptake emulative ofbone perfusion. In addition, littleof the expected liver or bowel activity was seen. The authorspropose that some dissociation of the çGd(DTPA)}2 complexoccurs in vivo and that the released Gd@ @vesrise to a carriereffect on the 67Ga biodistribution. Wiggins et al. (49) reject thisargument, suggesting instead that the change in biodistribution of67Ga is due to excess free ligand present in all commercialpreparationsofMRI contrastagents.Typically, 500 @gfree DTPAare administered and Wiggins et al. (49) estimate that this wouldresult in a plasma concentration of —2X i0@ M DTPA at thetime of67Ga administration to the patient described by Hattner andWhite.

We used a computer model of blood plasma which incorporates both Ga(III) and Gd(III) to calculate what effect agadopentate-enhanced MRI should have on the speciation ofGa(III). Typically, [Gd(DTPA)]2 is used as a contrast agent inMRI at an administered dose of —0.1mmole . kg I@ Within 30mm postinjection, the in vivo plasma concentration falls to 0.3mM and should have dropped to 0.06 mM 2 10 mm later (50)(this may be an overestimate for an 11-yr-old boy). At the sametime, the concentration of DTPA should be 2 X l0@ M (49).On the other hand, 67Ga as Ga(III) citrate is used at aconcentration of 5.2 X 10 I I M. In practice, 67Ga-citrate istypically used at a specific activity of > 10 mCi4tg gallium atthe reference date. This implies that the maximum carrier levelsof Ga(III) are 58.9 times the 67Ga concentration resulting in amaximum Ga(III) concentration in vivo of 3.0 X l0@ M. Atthe same time, the citrate concentration will be elevated to 0.34mM. We have used these concentrations, together with normalblood plasma component concentrations, to calculate the speciation of Ga(III) and Gd(III).

The results (Table 4) of our calculations show that, contraryto the suggestion of Wiggins et al. (49), residual DTPA doesnot cause a significant change in the speciation of gallium. Insupport of this claim, we can find no reported instances ofDTPA interferenceof 67Gascans following the use of 99mTc@DTPA even though kits of this radiopharmaceuticalhave asubstantial excess of DTPA (4 X l0@ moles). The reason forDTPA not affecting the gallium speciation is that it preferentially binds to Ca(II) which is present in high concentrations invivo. Similarly, at the concentrations used in the calculation,[Gd(DTPA)]2 - does not significantly affect the speciation ofGa(III).

In the absence of Gd(III), Ga(III) in blood plasma existsmainly as the [GaTf(HCO3)] complex. In the presence of 0.06mM Gd(III) (the calculated concentration of residual Gd(III)(50 ), the mixed metal [GdGaTf(HCO3)2)] predominates (see

Table 4). The Gd(III) occupies the C-terminal site of Tf, whilethe Ga(III) is complexed to the N-terminal site (44). At thesame time, there is a 100-fold increase in the concentration of[Ga(OH)4]. The 67Ga image obtained by Hattner and White(48) post-Gd-DTPA appears consistent with an increase in theproportion oflow molecular mass complexes of Ga(III) relativeto [Ga(Tf)], bearing in mind the changes seen in hypotransferremic subjects. Thus, the calculated increase in [Ga(OH)4]may be sufficient to account for the observed clinical change inthe 67Ga image observed 4 days later. Alternatively, the cellularuptake of Ga(III) from [GdGaTf(HCO3)@)J may be different tothat of [Ga(Tf)] resulting in a different 7Ga image.

CONCLUSIONOur simulation results support the original postulate of

Hattner and White (48) in that the different radio imageobtained with 67Ga-citrate 4 hr after a [Gd(DTPA)]2-enhancedMRI scan is due to dissociation ofthe [Gd(DTPA)]2 complex.Our calculations show that only 4% of the injected dose ofgadopentate need dissociate to cause substantial changes inGa(III) speciation. On the other hand, our calculations are alsoclear that neither residual DTPA nor [Gd(DTPA)]2 wouldaffect the biodistribution of 67Ga(III).

ACKNOWLEDGMENTSupported by the Foundation for Research Development and the

University of Cape Town.

REFERENCESI. Haliman PS, Perrin DD, Watt AE. Computed distribution of copper(Il) and zinc(ll)

ions among seventeen amino acids present in human blood plasma. Biochem J197I;121:549—555.

2. Sillen LG. How have sea water and air got their present compositions? Chem BriiI967;3:291—297.

3. Morel FM, Westall I, O'Melia C, Morgan I. Fate of trace-metals in Los-AngelesCounty wastewater discharge. Environ Sci Tech l975;9:756—761.

4. Rubin M, Gohil R, Martell AE. Motekaitis R, Penhos I, Weiss P. Computer-predicteddrugs for urolithiasis therapy. In: Inorganic Chemistry in Biology and Medicine. ACSSymp. Series. 1980:381—406.

5. May P. Linder P. Williams D. Computer simulation ofmetal ion equilibria in biofluids:models for the low-molecular-weight complex distribution of calcium(1l). magnesium(ll), manganese(Il), iron(lI). copper(ll), zinc(ll), and lead(lI) ions in human bloodplasma.J Chem Soc DaltonTransl977;588—595.

6. May PM. Computer applications in bioinorganic chemistry. PhD Thesis, University ofSt. Andrews, Scotland, 1980.

7. iackson GE, Wynchank S. Woudenberg M. Gadolinium(lII) complex equilibria: theimplicationsforGd(III)MRIcontrastagents.MagnResonMed1990;I6:57-66.

8. Martell AE, Smith RM, Motekaitis Ri. NIST critical stability constants of metalcomplexes database. NIST Standard Reftrence Database 46, Gaithersburg, MD.

9. May PM, Murray K. lEss, A ioint Expert Speciation System-Il. The ThermodynamicDatabase.Talantal991;38:1419—1426.

10. Berthon GW, Hacht B, Blais M-i, May PM. Copper-histidine ternary complexequilibria with glutamine. asparagine and serine. The implications for computersimulated distributions ofcopper(ll) in blood plasma. Jnorg Chim Acta 1986; 125:219 —227.

I I. Moerlein SM, Welch Mi. The chemistry of gallium and indium as related toradiopharmaceutical production. mt i NucI Med Biol 19818:277—287.

12. Hayes RL, Hubner KF. Basis for the clinical use ofgallium and indium radionuclides.In: Sigel H, ed. Metal Ions in Biological Systems, vol. 16, New York: Marcel Dekker;I983;279—315.

13. Shannon RD. Revised effective ionic radii and systematic studies of interatomicdistances in halides and chalcogenides. Ada Crs's:allogr l976;A32:751—767.

14. Brown PL. The hydrolysis of metal ions. Part I I. The ionic strength dependence ofgallium(lll). J Chem Soc Dalton Trans 1989;399—402.

15. Hams WR. Pecoraro VL. Thermodynamic binding constants for gallium transfemn.Biochemistry I983;22:292—299.

16. Baldwin DA, Egan Ti. An inorganic perspective of human serum transfemn. S Afr JScience 1987;83:22—3I.

17. Loomis LD, Raymond KN. Kinetics of gallium removal from transfemn andthermodynamics of gallium-binding by sulfonated tricatechol ligands. J Coord Chem1991 ;23:36l—387.

18. Welch Mi, Welch Ti. Solution chemistry of carrier-free indium. In: Subramanian G,Rhodes BA, Cooper iF, SOdd Vi, eds. Radiopharmaceuticals. Reston, VA: Society ofNuclear Medicine; 1975:73—79.

19. Halpem 5, Hagan P. Gallium-67 citrate imaging in neoplastic and inflammatorydisease. NucI Med Ann l980;2l9—259.

20. Taylor A, Chafetz N, Hollenbeck I, Hooser W. The source of fecal gallium—clinicalimplications: Concise communication. J Nucl Med 1978;l9:l214—12I6.

METAL ION SPECIATIONIN BLOOD PLASMA•Jackson and Byrne 385

21 . Nelson BM. Hayes RL, Edwards CL, Kniseley RM. Andrews GA. Distribution ofgallium in human tissues after intravenous administration. J Nuci Med l972;l3:92—100.

22. Hoffer P. Gallium: mechanisms. J Nuci Med l980;21:282—285.23. Weiner R. The role of transferrin and other receptors in the mechanism of 67Ga

localization. Nucl Med Biol l990;l7:l4l—149.24. Weiner RE, Spencer RP, Dambro Ti, Klein BE. Gallium-67 distribution in a man with

a decrease in both transfemn and hepatic gallium-67 concentration. J Nucl Medl992;33:llOl—l703.

25. Sohn M-H, iones B, Whiting I, Datz F, Lynch R, Morton K. Distribution of 67Gainnormal and hypotransferremic tumor-bearing mice. J NucI Med l993;34:2l35—2l43.

26. Hayes RL, Rafter JJ, Byrd BL, Carlton JE. Studies of the in vivo entry of 67Ga intonormal and malignant tissue. J Nuci Med 1981 ;22:325—332.

27. Eckelman WC. The application of receptor theory to receptor-binding and enzymebinding oncological radiopharmaceuticals. Nuci Med Biol 1994;21:759—769.

28. Kaplan i, iordan I, Sturrock A. Regulation ofthe transferrin.independent iron transportsystem in cultured cells. J Biol Chem 199 l;266:2997—3004.

29. Sturrock A, Alexander i, Lamb i, Craven CM, Kaplan i. Characterization of atransfemn-independent uptake system for iron in HeLa cells. J Biol Chem l990;265:3139—3145.

30. Anghileri L, Heidbreder M. On the mechanism of accumulation of 67Ga by tumors.Oncology I977;34:74—77.

3 1. Vallabhajosula SR. Harwig iF, Wolf W. Effect of pH on tumor cell uptake ofradiogallium in vitro and in vivo. Eur J Nucl Med l982;7:462—468.

32. Harris WR. Thermodynamics ofgallium complexation by human lactofemn. Biochemistrvl986;25:803—808.

33. Koizumi K, Tonami N, Hisada K. Deferoxamine mesylate enhancement of 67Gatumor-to-blood ratios and tumor imaging. Eur J NucI Med l982;7:229—233.

34. Oster ZH, Som P, Sacker DF, Atkins HL. The effects of deferoxamine mesylate ongallium-67 distribution in normal and abscess-bearing animals. J Nucl Med 1980;21:421—425.

35. Baker DL, Manno CS. Rapid excretion of gallium-67 isotope in an iron-overloadedpatient receiving high-dose intravenous deferoxamine. Am J Hematol I988;29:230—232.

36. Hodges R. latrogenic alterations in the biodistribution ofradiotracers as a result of drugtherapy: theoretical considerations. In: Hladik W, Saha G, Study K, Friedman B, eds.

Essentials ofnuclear medicine science. Baltimore: Williams and Wilkins; 1987:165—188.

37. Sephton R, Martin i. Modification of distribution of gallium-67 in man by administrationofiron.BrJRadiol l980;53:572—575.

38. Ohkubo Y, Araki 5, Abe K, Takasu S. Kohno H, Kubodera A. The effect of FeCl3 onthe accumulation of gallium-67 into inflammatory and normal tissue. Ann Nucl MedI988;2:59—62.

39. Loesberg AC, Martin WB. Altered biodistribution of 67Ga-citrate in an ironoverloaded patient with sickle cell disease. Clin NucI Med 1993;19: 157—159.

40. Moeller T. The lanthanides. In: Bailar iC, Emeleus Hi, Nyholm R, TrotmanDickenson AF, eds., Comprehensive inorganic chemistry, vol. 4. Oxford: PergamonPress; 1973:1—104.

41 . Lauffer RB. Paramagnetic metal complexes as water proton relaxation agents for NMRimaging: theory and design. Chem Rev 1987;87:90l—927.

42. Reuben I. Gadolinium(IIl) as a paramagnetic probe for proton relaxation studies ofbiological macromolecules: binding to bovine serum albumin. Biochemistry 197 l;I0:2834—2838.

43. Zak 0, Aisen P. Spectroscopic and thermodynamic studies on the binding ofgadolinium(lII) to human serum transferrin. Biochemistry 1988;27:1075—1080.

44. Harris WR, Chen Y. Difference ultraviolet spectroscopic studies on the binding oflanthanides to human serum transferrin. lnorg Chem l992;3l :5001—5006.

45. Harris WR. Binding constants for neodymium(IlI) and samarium(lll) with humanserum transferrin. lnorg Chem 1986;25:2041—2045.

46. Weinmann H-i, Brash R, Press W, Weskey G. Characteristics of GdDTPA complex:a potential nmr contrast agent. Am J Roentgenol l984;l42:6l9—624.

47. Cacheris W, Quay 5, Rocklage S. The relationship between thermodynamics and thetoxicity of gadolinium complexes. Magn Reson Imaging l990;8:467—48l.

48. HaWserRS, White DL. Gallium-67/stable gadolinium antagonism: MRI contrast agentmarkedly alters the normal biodistribution of gallium-67. J Nucl Med I990;3 I :1844—1846.

49. Wiggins J, Goldstein H, Weinmann H. Gallium-67/stable gadolinium antagonism. JNuciMed l991;32:l830—l83l.

50. Weinmann H-i, Press WR, Raduchel B, Platzek J, Schmiu-Willich H, Vogler H.Characteristics of Gd-DTPA and new derivatives. In: Bydder G, Felix R, Bucheler E,et al., eds. Contrast media in MRI. International Workshop Berlin. Bussum, TheNetherlands: Medicorm Europe; 1990:19—30.

386 THEJOURNALOFNUCLEARMEDICINE•Vol. 37 •No. 2 •February1996