Resonance Imaging,

Vol. 8, pp. 467-48 1, I990

Printedin the USA. All rightsreserved.

0730-725x/90$3.00 + .oa Copyright 1990Pergamon 0 Pressplc

l Original Contribution

WILLIAM P. CACHERIS, STEVEN C. QUAY, AND SCOTT M. ROCKLAGE Salutar, Inc., Sunnyvale, California 94086, USA
The suitability of gadolinium complexes as magnetic resonance imaging contrast agents depends on a number of factors. A thermodynamic relationship to toxicity exists if one assumes that the chemotoxicity of the intact complex is minimal but that the toxicity of the components of the complex (free metal and uncomplexed ligands) is substantial. Release of Gd3+ from the complex is responsible for the toxicity associated with gadoliium complexes; this release appears to be a consequence of Zn2+, Cu2+, and Ca2+ transmetallatiou in vivo. This hypothesis is supported by acute toxicity experiments, which demonstrate that despite a 50-fold range of LDse values for four Gd complexes, all become lethally toxic when they release precisely the same quantity of Gd3+, and by subchronic rodent toxicity experiments, which demonstrate a set of gross and microscopic findings similar to those known to be caused by Zn2+ deficiency. Finally, this hypothesis predicts that subtle changes in formulation can further enhance the intrinsic safety of these complexes. Ke_~words: Thermodynamics; Toxicity; Gadolinium; Contrast.


The evaluation of paramagnetic complexes as contrast agents for magnetic resonance imaging has focused upon the safety and efficacy of these agents.’ Safety has been evaluated by acute toxicity (LD,,), subchronic toxicity, local tissue tolerance, cardiovascular pharmacology, mutagenic potential, absorption, distribution, metabolism and excretion studies. These complexes are composed of relatively toxic components, i.e., metal ion and ligand, bonded together by ionic forces and subject to dissociation. An understanding of the in vivo affinity, i.e., stability, of these constituents for each other is important to the future development of magnetopharmaceuticals. One mechanism to explain the toxicity of these complexes is in vivo dissociation and/or metabolism to yield toxic metal ions and free ligands. Many workers have used the in vitro thermodynamic stability constant of the complex as a measure of the in vivo affinity of metal ion and ligand. The relationship between thermodynamic stability constant and in

vivo toxicity, however, does not hold for GdDTPABMA, GdDTPA, GdDTPA-BP and GdEDTA. A more complete thermodynamic evaluation is needed to ascertain in vivo stability and to explain the observed toxicity of these complexes.

Synthesis of GdDTPA-BA4A A full description of ligand (DTPA-BMA) and complex (GdDTPA-BMA) synthesis is given in Ref. 2. Synthesis of N,N”-2-Bis(2-pyridylrnethyl) diethylenetriamine Diethylenetriamine (76 g, 0.74 mol) and pyridine2-carboxaldehyde (174 g, 1.62 mol) in 2.5 L of absolute ethanol were heated for 2 hr at 50°C with stirring. After cooling to room temperature, 25 g of 10% palladium on charcoal was added and the schiff base hydrogenated at slightly greater than 1 atm of hydrogen, over a 48-hr period. The catalyst was removed by filtration, the filtrate adjusted to pH 4 with HCl gas

RECEIVED 9/30/89;

ACCEPTED 2/17/90.

Acknowledgments-The authors thank William Dow and David Love for synthesizing and formulating GdDTPABMA, Joan Carvalho for synthesizing and formulating 467

GdDTPA-BP and Dilip Worah, Debra Kesler and Dean Kessler for performing the animal studies. Address all correspondence to Scott M. Rocklage, Salutar, Inc., 428 Oakmead Parkway, Sunnyvale, CA 94086.


Magnetic Resonance Imaging 0 Volume 8, Number 4, 1990

and then lowered to pH 1 using 12 N HCl. The resulting precipitate was removed by suction filtration, washed with absolute ethanol until the washings were colorless, and recrystallized from 95% ethanol. This HCl salt was then dissolved in 500 mL of water and neutralized with 5 N NaOH, then raised to pH 12.5, and the free base extracted with methylene chloride (4 x 500 mL). The methylene chloride solution was dried to give 136 g (65%) of a pale yellow oil. NMR (D,O): 6 2.46 (s, 8H), 3.55 (s, 4H), 7.10 (m, 4H), 7.55 (t, J= 12.5 Hz, 2H), 8.25 (d, J= 10 Hz, 2H), 7.50 (t, J = 10 Hz, 2H), 8.40 (d, J = 10 Hz, 2H). Synthesis of N,N”-2-Bis(2-pyridylmethyl)-N,N,WTris(t-Butylcarboxymethyl)Diethylenetriamine To a solution of N, N”-2-bis(2_pyridyl)diethylenetriamine (23.6 g, 82.6 mmol) and diisopropylethylamine (53.4 g, 0.4 mol) in 1.2 L of methylene chloride at room temperature was added dropwise t-butylbromoacetate (50 g, 0.2 mol) in 300 mL of methylene chloride. After stirring for 24 hr, the solution was evaporated to dryness and placed under vacuum for 2 hr to remove excess diisopropylethylamine. The crude solid was taken up in 1.5 L of methylene chloride, washed with 0.2 N NaOH, water (2 x 250 mL), brine (200 mL) and dried (MgS04). The methylene chloride was removed, 200 mL of ethyl acetate added and this solution passed through 300 g of silica gel in a buchner funnel using EtOAc to elute the product. The pure fractions (TLC: MeOH/CH2C12: 3/7) were combined to give 40.6 g (79.5%) of the ester, NMR (CDC&) 6 1.26 (s, 9H), x 1.31 (s, 18H), 2.62 (s, 8H), 3.12 (s, 2H), 3.17 (s, 4H), 3.78 (s, 4H), 7.02 (t, J = 10 Hz, 2H), 7.35 (d, J = 10 Hz, 2H). Synthesis of N,N”-2-Bis(2-Pyridylmethyl) Diethylenetriamine-N,W,N”-Triacetic Acid The tris(t-butylcarboxymethyl)ester (24.8, 0.1 mol) was dissolved in a solution of 600 mL of methylene chloride containing 380 mL of trifluoroacetic acid. The solution was stirred for 48 hr, evaporated under reduced pressure and diluted with 50 mL of water. This solution was applied to 200 mL of AG50-X8, H+ form, 100-200 mesh and after washing with water until neutral, the product was eluted with 1 N NH40H. After removal of NH40H solution, the product was taken up in 24 mL of water, adjusted to pH 10 and the solution applied to AGl-X8, acetate, 100-200 mesh. The column was washed with three bed volumes of water and the product eluted with 2 N HOAc to give 12.0 g (69%) of product after several lyophilizations. NMR (D,O) 6 3.02 (t, J= 6 Hz, 4H), 3.08 (t, J = 6 Hz, 4H), 3.14 (s, 4H), 3.41 (s, 2H), 4.08 (s, 4H),

7.52 (m, 4H), 8.05 (t, J= 10 Hz, 4H), 8.40 (d, J= 10 Hz, 2H). Potentiometric Measurements Potentiometric titrations, to determine the acid protonation constants as well as the metal ion stability constants of DTPA-BMA and DTPA-BP, were carried out with an automatic titrator system.3 The system was controlled by a BASIC computer program which displays the data in tabular form concurrent with a high resolution plot. Components of the autotitrating system included a Fisher digital pH meter, Corning glass and AgCl reference combination electrode, and a Metrohm digital autoburette. In each experiment, temperature was maintained at 25.O”C + 0.1 and ionic strength was kept constant at 0.10 M with NaCl. The concentration of metal ions and ligand was maintained between 3 x lop3 M and 5 x 10m3M, and 0.1000 M HCl was used as the titrant to minimize ionic strength changes during the course of a titration. Addition of a small amount of NaOH prior to the titrations was used to raise the solutions to pH 11. Stability constants for Gd3+, Zn2+, Cu2+, and Ca2+ complexes of DTPA-BP and the Zn2+, Cu2+ and Ca2+ complexes of DTPA-BMA were determined by direct titration. For these systems the complexes were found to be greater than 25% dissociated at pH 2, while the data for the GdDTPA-BMA system revealed the complex to be dissociated less than 1% at pH 2. Thus GdDTPA-BMA was studied using a ligand-ligand competition titration. In this experiment a 1: 1: 1 molar ratio of Gd3+, DTPA-BMA, and EDTA was titrated. EDTA forms a complex with Gd3+ whose stability constant is accurately known. At high pH, Gd3+ is primarily bound to EDTA. The Gd3+ ion was readily transferred to DTPA-BMA as the pH was lowered to 2. The rate of metal transfer was sufficiently fast that equilibrium pH measurements could be made in a reasonable time period after each addition of acid. Computations Proton association constants for DTPA-BMA were calculated using a BASIC computer program written for polyprotic weak acid equilibrium calculations. The stability constants of the metal ion complexes were calculated using a BASIC program designed for metal ion, ligand and proton systems containing a variety of species. Both programs employed a modified NewtonRaphson algorithm which solved the simultaneous nonlinear mass balance equations, yielding -log[H+] values. The evaluated equilibrium constants were varied by a combined Simplex/Marquardt nonlinear regres-


and toxicity 0 W.P. CACHERIS AL. E?


sion algorithm. The average difference between observed and calculated -log [H+] was co.04 throughout all titrations. Biospeciation calculations were performed with a BASIC program which employed the modified Newton-Raphson algorithm. The non-linear equations resulted from the mass balance equations for an eight-component system consisting of Gd3+, Zn’+, Cu2+, Ca2+, ligand, monoaminomonocarboxylate amino acids (which were combined since their stability constants are very similar), citrate, and human serum albumin. The program required equilibrium constants which were taken from the Iiterature,4-6 at 37°C and p = 0.15 M. Values that were not available at p = 0.15 M were replaced by values at p = 0.10 M. Equilibrium constants that were not known at 37°C were calculated from the enthalpy and entropy of formation7 (measured at 25°C). The enthalpy and entropy of formation are relatively temperature independent over small temperature ranges7 and were relied upon for accurate equilibrium constants at 37°C. In the case of GdEDTA it was necessary to take into account ternary complexes of GdEDTA with amino acids and citrate.8 It can be shown from the relatively small value for the equilibrium ML + M = M2L (log K = 4.48 for ZqDTPA) that the binuclear complexes of DTPA are not formed in significant concentrations in vivo, and these were excluded from the biospeciation calculations. A distribution volume of 180 mL/kg was used to convert dosage to in vivo concentration. The in vivo concentrations of the components used were: Ca2+, 2.5 mM: Zn2+, 50 PM: Cu2+, 1 PM: amino acids, 2.76 mM: citrate, 0.11 mM: and albumin, 0.4 mM.9 The root of the set of equations yielded the free concentrations of all components. These concentrations were used along with the stability constants to determine full speciation for each dosage of the complexes. Transmetallation Kinetics A measurement of the rate of Gd3+ release from GdDTPA-BMA, GdDTPA and GdDTPA-BP was attempted by a method used by Margerum for GdEDTA and GdCDTA.‘O A solution of 1.288 x 10e5 M GdDTPA-BMA (or GdDTPA) was prepared at pH 5.0 with an acetate buffer (0.01 M buffer, p = 0.1 M). A lo-fold excess of Cu2+ was added and the formation of the Cu2+ complex was followed by UV absorption at 268 nm. Animal Studies Acute toxicity (LD,,) studies were conducted in Swiss-Webster mice. Five groups of mice, each

containing four males and four females, were injected intravenously with either GdDTPA-BMA, Na2 [GdDTPA] , GdDTPA-BP , Na [ CaDTPA-BMAI , Na[CaDTPA-BP] or formulations containing either GdDTPA-BMA + 5 mole % Na[CaDTPA-BMAI , Na2 [GdDTPA] ] + 5 mole % Na3 [CaDTPAl , GdDTPA-BP + 5 mole % Na[CaDTPA-BP] or GdDTPA-BMA + 5 mole % Na[CaDTPA-BMA] + 123 mole % NaCl. For each LD,, determination, each group of eight mice was given a different dosage of compound. Dose volumes larger than about 40 mL/kg body weight were administered as divided doses, at one-half hour apart. Animals were observed, once daily, for mortality or morbidity for 14 days and the LDso value was calculated. The calculations were made using a weighted probit method.” Osmolarity of the various formulations were measured on a Wescar model 550 vapor pressure osmometer. A subchronic toxicity study was conducted in Sprague-Dawley rats. Three groups of rats, each consisting of 10 male and 10 female rats, received GdDTPA-BMA intravenously three times a week for three consecutive weeks at dosage levels of 0.1, 2.0, and 5.0 mmol/kg. A fourth group received normal saline and served as the negative control. All animals were observed twice daily for morbidity and mortality. Clinical observations for obvious toxicologic effects were also performed at least once daily. Gross and microscopic evaluations were conducted at study termination.

Potentiometry The potentiometric titration curves for DTPABMA and ZnDTPA-BMA are shown in Fig. 1. The DTPA-BMA curve shows a very sharp decrease between pH 9 and 5.5. This is due to the large separation of the pK, values of the two most basic groups of the ligand, 9.37 and 4.38 (shown by NMR to be the amines). The third amine of the ligand has a pK, of 3.31 and the carboxylates all have pK, values below 2. The ZnDTPA-BMA curve drops rapidly from the initial pH of 10 to pH 5, since the only protonation of the complex takes place with a pK, of 3.99. At pH 2, the protonated ZnHDTPA-BMA and CuHDTPABMA complexes are ca. 35% and 25% dissociated, respectively. The CaDTPA-BMA complex undergoes a similar protonation with a pK, of 4.45, but becomes completely dissociated at pH 2.7. The results of the potentiometric measurements for DTPA-BMA, DTPA, DTPA-BP and EDTA and the Gd3+, Zn2+, Cu2+, and Ca 2+ thermodynamic stability

470 12 11 10 9

Magnetic Resonance Imaging 0 Volume 8, Number 4, 1990

constant with Gd3+ while DTPA-BMA (CpK, = 17) is less basic than EDTA (CpK, = 22) and has the second lowest Gd3+ thermodynamic stability constant. Thermodynamics The ligands DTPA-BMA, DTPA, DTPA-BP and EDTA are polyprotic. A comparison of thermodynamic stability constants with the stability at pH 7.4 is an important first step in understanding in vivo stability. Each ligand has a different response to proton competition for Gd 3+ binding which is based on its intrinsic basicity. Specifically, the thermodynamic stability constant (Ktherm) is defined as

7 6 5 4 3 2 1 0



4 of base


M-+LPML K therm
The conditional

Moles of acid/moles

= IMl ILl

Fig. 1. Potentiometric titration curves of DTPA-BMA in the presence and absence of equimolar Znzf. Concentration of metal ion and ligand = 5.00 x 10e3 M (~1= 0.10 M NaCl, 25°C).


stability constant (Kcond)specifies the degree of metal chelation at a given pH, and is given by

constants are given in Table 1. The value obtained for the thermodynamic stability constant of the GdDTPABMA complex was obtained by a competitive titration technique with EDTA. The most basic ligand, DTPA (CPK, = 30), has the highest thermodynamic stability
Table 1. Protonation

K co”d= +

([L] + [HLI + [z&L] + . . . )
for free ligand species in variforms. This expression correctly de-

where Kcond accounts
ous protonated

constants and metal chelate stability constants (25”C, p = 0.10 M (NaCl)). Uncertainty (u) in Log K values are given in parentheses Log K



DTPA-BPa 9.53 6.46 4.76 3.41 2.28 (0.11) (0.07) (0.08) (0.08) (0.10)

DTPAb 10.49 8.60 4.28 2.64 2.0 1.6 22.46 2.39 10.75 6.11 18.70 5.60 21.38 4.81

EDTAb 10.17 6.11 2.68 1.95 1.56 17.27 1.53 10.61 c 16.86 3.0 18.86 3.0


9.37 (0.01)
4.38 (0.01) 3.31 (0.04) 1.43 (0.12)

[GdHL]/[GdL][H] [CaHL]/[CaL][H] [ZnL]/[ZnL][L] [ZnHL]/[ZnL](H] [ZnH2L]/[ZnHL][H]

16.85 (0.05) c 7.17 (0.04) 4.45 (0.02) 12.04 (0.03) 4.04 (0.04) 13.03 (0.03) 3.36 (0.02)

16.83 (0.11) c 7.97 (0.05) 5.30 (0.03) 14.02 (0.14) 5.36 (0.07) 3.67 (0.04) 17.50 (0.17) 3.74 (0.03) 2.92 (0.03)



[CuHL]/[CuL][H] [CuH,L]/[CuHL][H] aThis work. bReferences 4 and 6. ‘Not measured.


and toxicity 0 W.P.



scribes the affinity of the metal ion for the ligand but is not equivalent to the thermodynamic expression. The conditional stability constant, at a given pH, can be calculated from the thermodynamic value if the protonation constants of the ligand are known
K cond- Ktherm

more stable than GdDTPA-BMA by only a factor of 102-9. The complexes are equally stable at pH 4.


where LT is the total concentration of the uncomplexed ligand, i.e. ([L] + [HLI + [H2L] + . . . ). Thus,
K cond
&xxm (1 + &I

[H+l + KHIKw[~+I~

+ . . . )-’ =

where KH values are stepwise proton association constants for the ligand and the sum of these terms is (~6’. Figure .2 describes the variation of the conditional stability constants for GdDTPA-BMA and GdDTPA with pH. At a pH high enough to ensure complete deprotonation of the ligands (pH > 1l), the conditional stability constants of the complexes are equal to their thermodynamic values and GdDTPA is found to be more stable than GdDTPA-BMA by a factor of 105.2. However, at pH 7.4, GdDTPA is

An important consideration in understanding the stability of such complexes in vivo is their solubility with respect to precipitating anions such as PO:-, OH-, and CO:-. Ligands that have too weak an affinity for gadolinium to prevent precipitation in vivo would be expected to be toxic. For example, simple gadolinium salts have been demonstrated to precipitate in vivo.12 Martell has defined a solubility constant for chelating agents that predicts the solubility of metal ions in the presence of chelating agents and precipitating ion13. This solubility constant was initially designed to predict the ability of a ligand to diminish metal ion overload when the metal may be present in some insoluble form. The solubility constant is defined as KSO, m =

which describes the amount of metal, as the complex, in the presence of the ligand. Alternatively, KsOl may be expressed as &,I = &erm [Ml [Ll

?of ;

where LT is the total concentration of the ligand not bound to the metal ion M, i.e., (1L] + [HL] + [l&L] + . . . + [CaL] .+ [ZnL] + [CuL] + . . . ) . In biological systems the ligand can exist in different protonated forms and also be bound to other metal ions:
LT = [Ll + + . . . KHI + [H+l [Ll + KHIKHZ[H+I~[LI

KCaL [ Ca2+] [ Ll + KZnL [ Zn2+] [L] [L] + . . .

+ KcuL[Cu2+] thus

KsO, = Ktherm [Ml ((.y~l + a& + (~2: + a& + . . . )-I

2 3 4 5 6 7 PH 0 9 10 11 12

where a;’ = 1 + KH, [H+]
+ K,,K,,[H+]’ + ...

Fig. 2. pH dependence of conditional stability constant for
GdDTPA-BMA and GdDTPA. GdDTPA has a higher conditional stability constant than GdDTPA-BMA at pH 11 by a factor of 105.6. At pH 7.4 this factor is reduced to 103.4 and at pH 4 these complexes have the same conditional stability constant values.

a& = KCaL [Ca2+]

cr.& = KZnL [Zn’+] CX& Kc-“,/ [Cu2+] =


Magnetic Resonance Imaging 0 Volume 8, Number 4, l!XKl

KSolis often reported as a logarithm log&l1 = 1ogKtlerm + h%[Ml

A negative value of log KS,,, predicts that the ligand will not be able to solubilize GdP04 while a positive value indicates the ligand has the ability to dissolve the precipitate. The amount of free metal derived from the solubility of GdP04, in the presence of a ligand, can be calculated by14 log[M] = -logaM - logan (PO:-) + log&, - log [PO:-]

-10 -15 ’
Increasing Thermodynamic Stability Constant



and an(POj-) is the proton competition factor for phosphate at pH 7.4. Log[M] values were calculated from the above expressions for GdP04 in the presence of DTPA-BMA, DTPA, DTPA-BP, and EDTA. These calculations were performed, at a [PO:-] of 1.13 mM, by determining the amount of [L] [Zn2+], , [Cuzf] , and [Cat+] in vivo via a biospeciation model (vide infra). Table 2 shows the results of these calculations for a 0.1 mmol/kg dosage of these complexes (distribution volume = 180 mL/kg). Values of free ligand and endogenous metal ions are shown along with free metal (log[M]) . The value of log[M] and free endogenous metal ions were then used to calculate KSol for these complexes. Log KSolvalues of 6.7, 10.2, 5.7, and 2.0 were calculated for GdDTPA-BMA, GdDTPA, GdDTPA-BP and GdEDTA, respectively. These values for KSol predict stability towards phosphate precipitation in vivo, and this has been confirmed by biodistribution data obtained for GdDTPA-BMA, GdDTPA”, and GdEDTA”. Figure 3 illustrates the variation in the value of

Fig. 3. Values of log Ksol for several Gd3+ complexes. The complexes are arranged in order of increasing thermodynamic stability constant. Complexes with positive log Ksol values are expected to be soluble in vivo with respect to GdP04 precipitation while those with negative Ksol values are not. The right hand legend estimates the percent of gadolinium as the indicated complex (the remainder is insoluble GdP04).

KSolwith complexing strength of the ligand. GdC& is known to precipitate in vivo,r2 consistent with a very large negative value of KSO,.Gd(OAch and GdNTA would also be expected to be unstable towards PO:precipitation in vivo. Figure 4 shows the results of similiar calculations for GdDTPA-BMA, GdDTPA, GdDTPA-BP and GdEDTA collected in the urine. All four complexes show stability towards PO:- precipitation from pH 5 to 8 when collected and concentrated in the urine. The concentration of the complexes in the urine was estimated to be 30 mM based on a 7 mmol dose (0.10 mmol/kg x 70 kg), assuming that 50% of the complex is excreted in 1.58 hrr6 and that urinary output is 75 mL/hr. GdCl,, Gd(OAc)3 and GdNTA were not included since these

Table 2. Constants obtained for solubility prediction of gadolinium complexes in the presence of phosphate. K SP = 1O-22.26, [PO:-] = 1.13 mM, dosage = 0.1 mmol/kg, volume of distribution = 180 ml/kg GdDTPA-BMA GdDTPA-BP 3 42 -84 2.34 2.20 1 78 517 x LO-” M x 1O-3 M x 1O-8 M x 10-l’ M GdDTPA 2.84 -8.9 2.33 2.37 1 78 10:2 x lo-l3 M x lo-’ M x lo-’ M x 10-l’ M GdEDTA 1 40 -714 2.40 1.34 1.84 2.0 x 10-r’ M x 1O-3 M x lo-* M x lo-” M

LogtMl [Ca2+] [Zn2+] [Cu2’] Log &or

1.35 -9.3 2.33 2.46 1 78 6:7

x 1O-9 M x 1O-3 M x 10-s M x 10-l’ M

Thermodynamics and toxicity 0 W.P. CACHERIS ET







pH 50

pH 65

pH 80

GdDTPA, and GdEDTA in urine at pH 5.0, 6.5, and 8.0. The complex concentration was assumed to be 30 mM based on a 7-mmol dose (0.1 mmol/kg x 70 kg) with 50% of the complex being collected in 1.58 hours with urinary output of 75 mL/hr. All three complexes are expected to be stable toward GdP04 precipitation under these conditions.

Fig. 4. Values of log Kso, for GdDTPA-BMA,

are insoluble in plasma and would not be expected to be present in the urine. Biospeciation In order to better understand the toxicity of gadolinium complexes we have employed a broad thermodynamic approach. The Gd3+ stability constant is not the only stability constant relevant to a consideration of the toxicity of Gd3+ complexes. Based on the assumption that most of the toxicity of such complexes arises from release of the highly toxic Gd3+ ion in vivo, it is important to know the stability of the ligand with endogenous metal ions such as Zn’+, Cu’+, and Ca2+ since these metal may displace the Gd3+ from the complex. In addition, the stability constants of Gd3+ with biological ligands such as amino acids, citrate and albumin must be known. Figure 5 shows the variation of Gd3+ released from GdDTPA-BMA with concentration of the complex, as a function of the conditions of the experiment. The in vitro calculations show that very little Gd3+ is released if the equilibrium M + L Ft ML is considered in isolation. Proton competition at pH 7.4 increases the quantity of gadolinium ion released from about 0.1 mM to 1 nM. Additional equilibria with endogenous metal ions and ligands are very important, increasing the amount of

Gd3+ released from the complex by a factor of 103-104. Recognition of these additional equilibria is important in predicting the amount of dissociation of a complex in vivo and in explaining the subsequent differences in toxicity amongst complexes. The difference in stability constant between Gd3+ and other endogenous metal ions for the same ligand is termed selectivity. In a thermodynamic context, the different stability constants for these endogenous metal ions provide an explanation of the toxicity of GdDTPA-BMA, GdDTPA, GdDTPA-BP, and GdEDTA. EDTA shows very little selectivity between metal ions. The Gd3+ stability constant for EDTA is greater than that for Zn2+ only by a factor of 10°.4. DTPA-BP has a greater selectivity for Gd3+ over Zn2+ by a factor of 102.8. In DTPA, this factor increases to 103.8, however, with DTPA-BMA the Gd3+/Zn2+ selectivity is even greater than for DTPA, with Gd3+ being preferred by 104.*. The selectivity of these ligands with Gd3+, Zn2+, Cu’+, and Ca2+ is illustrated in Fig. 6. Among the endogenous metal ions, Zn’+, Cu2+ and Ca2+, Zn2+ was expected to have the greatest significance in effecting Gd3+ release from the complexes in vivo. Cu 2+ has a favorable stability constant with respect to all three ligands, but is present in such small concentrations in the blood (l-10 PM) that it cannot displace much Gd3+. Ca2+ is not expected to displace Gd3+ from these complexes, despite its high

11 t


10 I\-




5 i,,,,...,,‘,,,,,....‘.,.,,,...,...,.....,.........J 0 1 2 3 [GdDTPA-BMA]



Fig. 5. Comparison of in vitro and in vivo -log[Gd3+] versus concentration of GdDTPA-BML. The in vitro values were obtained by consideration of thermodynamic and conditional (pH 7.4) equilibria. The in vivo values were determined by biospeciation calculations.


Magnetic Resonance Imaging 0 Volume 8, Number 4, 1990










Fig. 6. Three-dimensional plot of the thermodynamic Gd3+, Zn2+, Cu2+ and Ca’+.

stability constants




and EDTA with

concentration in plasma (2.5-4 mM), since DTPABMA, DTPA, DTPA-BP, and EDTA have relatively low stability constants with this metal ion. Zn’+, however, has moderately high stability constants with DTPA-BMA, DTPA, DTPA-BP and EDTA and it is present in concentrations (lo-50 PM) that can displace a significant amount of Gd3+. However, as indicated above, DTPA-BMA, DTPA, DTPA-BP, and EDTA are very different in their selectivity for Gd3+ over Zn2+. This can be seen in Fig. 7, which shows the ratio of stability constant for these ligands with Gd3+ with Zn2+. This illustrates that of all the ligands considered herein GdDTPA-BMA has the largest selectivity for Gd3+ over Zn2+. The consequences of Gd3+ selectivity over other endogenous metal ions for a specific ligand can be understood through in vivo speciation calculations. A computer calculation of the speciation of Gd3+ complexes was performed incorporating the metal ions Zn2+, Cu2+, and Ca2+. Fe3+ was not considered since it is very tightly bound by the storage proteins ferritin and hemosiderin, and is essentially unavailable for interaction with Gd3+ complexes. The in vivo “ligands” used in the biospeciation calculations included amino acids, citrate and human serum albumin. Figure 8 shows the result of such calculations. The amount of Gd3+ released from the complex as a function of dosage is shown for GdDTPA-BMA, GdDTPA, GdDTPA-BP, and GdEDTA. GdEDTA releases a relatively large amount of Gd3+ at low dos-

ages. GdDTPA-BMA releases about one-half as much Gd3+ as does GdDTPA at any given dosage. The results of the biospeciation model are consistent with the observed LDso values (IV, mice) of GdDTPA-BMA (14.8 mmol/kg), Na2 [GdDTPA] (5.6 mmol/kg), GdDTPA-BP (2.8 mmol/kg), and NMG[GdEDTA]’ (0.3 mmol/kg). The LD5,-,dosage for each of these




Fig. 7. Selectivity of DTPA-BMA, DTPA, DTPA-BP and . . . . EDTA for Gd3+ over Zn’+. Selectrvity is given as KG& K =“r. GdDTPA-BMA has the largest selectivity of the three ligands.


and toxicity l W.P. CACHERIS AL. ET


complexes is indicated in Fig. 8 and relates administered dosage to the quantity of Gd3+ released. Despite a 50-fold range in LDso values based on administered dosage, all four complexes become lethally toxic to half the population of mice when they release ca. 13-15 PM Gd 3+. Since Gd3+ has been shown to inhibit Ca2+ binding to mammalian cardiac sarcoplasmic reticulum at about 20 PM Gd3+, the actual mechanism of toxicity could involve hemodynamic disruption.r7 The conditional stability of such complexes, in the form considered for in vivo speciation of the complexes, thus provides an excellent correlation with observed toxicity. Figure 9 shows that the Gd3+ rel .~sed from GdDTPA-BMA is found primarily as the &rate complex. The correlation between Gd3+ selectivity of a ligand and toxicity can also be understood by definition of a Gd3+ selectivity constant. The Gd3+ selectivity constant accounts fur Gd3+ selectivity of the ligand by modifying the thermodynamic stability constant of a Gd3+ complex to incorporate ligand equilibria with H+, Zn2+, Ca2+ and Cu2+. More explicitly,








Fig. 9. Speciation of Gd3+ released from GdDTPA-BMA as a function of dosage. The Gd3+ released is primarily found as the citrate complex.

where CYH’ 1 + KH, [H+] + Km Krr2 [H+]’ + . . . =
30 ,


a~:= KCaL[Ca2+]
cc&= KZnL[Zn2+]
a& = KCuL[Cu2+] .
Table 3 indicates the calculated Gd3+ selectivity constants at pH 7.4 for DTPA-BMA, DTPA, DTPABP and EDTA as well as the LDso values for the Gd3+ complexes. The concentrations of Ca2+, Zn2+, and Cu2+ used were 2.5 mM, 50 PM and 1 PM, respec-




1 Dosage (mmole/kg)


Fig. 8. Results of the biospeciation calculations for GdDTPA-BMA, GdDTPA, GdDTPA-BP and GdEDTA. The amount of Gd3+ released from the complex as a function of complex dosage is shown (distribution volume = 180 mL/kg). GdEDTA is expected to release a relatively large amount of Gd3+ at small dosages. At each dosage, GdDTPA-BMA is expected to release about one-half of the Gd3+ as compared to GdDTPA. All four complexes become lethally toxic to half the population (mice when ca. 13-15 PM Gd3 is released.

Table 3. Thermodynamic and selectivity stability constants for gadolinium complexes and LDso values (mice, iv). Uncertainty (a) in LD,, values are given in parentheses. Complex GdDTPA-BMA Na*[GdDTPA] GdDTPA-BP NMG[GdEDTA] ‘mmol/kg. Log Krherm 16.85 22.46 16.83 17.27 Log &et 9.04 7.04 5.32 423 LD50a 14.8 5.6 2.8 0.3 (0.7) (0.2) (0.1) (0.1)


Magnetic Resonance Imaging 0 Volume 8,

Number 4, 1990

tively. These represent the total in vivo concentrations of these metal ions, since equilibria with endogenous ligands were not considered. Figure 10 shows that unlike the thermodynamic Gd3+ stability constants, a correlation can be drawn between the Gd3+ selectivity constant and LDSOfor each of these four complexes.

15 1








5 .

EDTA 0 L...,....,r...,....,....,...,l....,....~..,...,’

15 16 17 16 19 20 21 22 23 24 25 Log KWm+u.,c with Gd”



10 -


Kinetics A potential criticism of such a biospeciation model is the need to assume that thermodynamic equilibrium of the complex is reached in vivo before the compound can be excreted from the body. Brucher has shown that the kinetics of Gd3+ displacement from GdEDTA by Cu*+ is indeed very fast and occurs within the stopped-flow time frame (~1 sec).18 Similar experiments have been performed for GdDTPABMA, GdDTPA and GdDTPA-BP at pH 5, and the formation of CuDTPA-BMA, CuDTPA and CuDTPA-BP is complete within a similar time frame. The rate of transmetallation, however, seems to be important for sterically rigid complexes, like GdCDTA and GdDOTA. Both of these complexes do not fit into the thermodynamic correlation discussed here. CDTA has selectivity comparable to EDTA for Gd3+ over Zn*+ Ca*+ and Cu*+. l9 Yet its LDSo value, ca. 2 mmol/kgl: is significantly more favorable than that of GdEDTA. GdCDTA fits the thermodynamic correlation only if one assumes that the amount of Gd3+ released before clearance from the body is around 35% of the amount expected if the transmetallation reaction were to reach equilibrium. Margerum has shown that the reaction of Cu*+ with GdCDTA occurs with a fl,* of 40-100 min at pH 4-5 (35”C).” This reaction rate would certainly place the amount of Gd3+ released from GdCDTA into the predicted range if this complex is cleared from the body at the normal rate for an extracellular agent ( tl,* = 95 min).16 GdDOTA is expected to be very toxic based on the biospeciation model, due primarily to the high stability constants for the Ca*+, Zn*+, and Cu*+ complexes of DOTA*’ (reflecting its poor selectivity). As with GdCDTA, GdDOTA conforms to the thermodynamic correlation if the amount of Gd3+ released within its biological lifetime is less than 5% of the amount expected at transmetallation equilibrium. This is not unreasonable, by analogy with the observed slow dissociation kinetics of CdDOTA.*l Toxicity and Formulations Improvements in safety can be predicted from slight modifications in formulation if the primary mode of toxicity is Gd3+ release. A small excess of ligand added to formulations of the complex should be capable of minimizing transmetallation reactions in vivo. Figure 11 shows the effect of the amount of Gd3+ released (due to Zn*+ transmetallation) from the addition of small quantities of DTPA-BMA salts to GdDTPA-BMA. Even a 1 mole % addition of DTPA-BMA salts causes a marked decrease in the amount of Gd3+ released. ’ Figure 12 indicates that the addition of small

EDTA 0 ’








Log KYnlb,,, with Gd’+
Fig. 10. Correlation between Gd3+ selectivity constant and LD50 values for Gd3+ complexes. The thermodynamic stability constants of GdDTPA-BMA, GdDTPA, GdDTPABP and GdEDTA show no correlation with LDso (left graph). However, LDso appears to be correlated to the gadolinium selectivity constant (right graph) with GdDTPABMA having the largest Gd3+ selectivity constant and the highest observed LDso. Data for all complexes are given in Table 3.


and toxicity 0 W.P. CACHERI~ ET




Ilgand is

30 -



= 0.1 mmoie/kg

,’ 10 -/<
Expected trend


. . ..___ ___
















1 04



Mole % DTPA-BMA Added

to Formulation


Fig. 11. The effect of adding excess DTPA-BMA to formulations of GdDTPA-BMA. Even a 1Vo addition of DTPABMA significantly decreases amount of Gd3+ expected to be released in vivo.

amounts of free ligand to the formulation initially yields an improvement in the LDse value of more than a factor of two. Addition of more than 1% excess of free ligand, however, decreases the LDSo value and a moderately toxic formulation exists when 5 mole % free ligand is added. The free ligand (at physiological pH), Na2HDTPA-BMA (LDso = 0.16 mmol/kg, iv, mice), is as toxic as the Gd3+ ion. This toxicity presumably predominates in the toxicity of the formulation when more than 1 mole Vo excess free ligand is added. Na[CaDTPA-BMA] has a more favorable LD,e value of 16.6 mmol/kg (iv, mice). Since in vivo transmetallation is expected to occur with Zn2+ and not with Ca2+, the addition of small amounts of

Fig. 12. The variation of LD,o values for formulation of GdDTPA-BMA with the stoichiometry of the Gd3+ and DTPA-BMA. A 1% excess of DTPA-BMA increases the LDso value from 14.8 mmol/kg to 34 mmol/kg by minimizing Zn2+ transmetalation in vivo. Quantities of excess DTPA-BMA above 1% (added as the free ligand) decrease the LDso value due to the toxicity of the free ligand. If the excess ligand is added as CaNaDTPA-BMA, then sustained high LD,, values (38.3 mmol/kg) are observed up to 5% excess ligand.

Na [ CaDTPA-BMA] to GdDTPA-BMA formulations should enhance the safety of GdDTPA-BMA formulations. Since available endogenous Zn2+ will displace Ca2+ from Na[CaDTPA-BMA] , Zn’+ will no longer be available for Gd3+ transmetallation. An increase in the LDSo value from 14.8 to 38.3 mmol/kg (Fig. 13) is observed with a 5% addition of Na[CaDTPABMA] to the formulation, as opposed to simply adding DTPA-BMA itself. Table 4 summarizes the LDSo values found for

Table 4. Osmolarity and LD,, valus (mice, iv) for gadolinium complexes in various formulations. Uncertainty (a) in LD,, values are given in parentheses Complex/Formulation GdDTPA-BMA Na,[GdDTPA] GdDTPA-BP GdDTPA-BMA + 5 mole% Na[CaDTPA-BMA] Na2[GdDTPA] + 5 mole% Na3[CaDTPA] GdDTPA-BP + 5 mole% Na[CaDTPA-BP] GdDTPA-BMA + 5 mole% Na[CaDTPA-BMA]
amOs/kg, normalized to a complex concentration bmmol/kg. of 500 mM.

Osmolaritya 645 1983 534 707 2060 594 1977

LD,ob 14.8 5.6 2.8 38.3 4.5 2.9 6.6 (0.7) (0.2) (0.1) (1.4) (0.1) (0.1) (0.3)

+ 123 mole% NaCl

478 15 , 14 13 12 /

Magnetic Resonance Imaging 0 Volume 8, Number 4, 1990 I

p, ! “\ i ! ‘.\,

Dosage 0.1mm&/kg =



Fig. 13. Change in -log[Gd3+] released from GdDTPABMA, GdDTPA, GdEDTA, and GdDOTA with consideration of stability factors. The largest increase in the amount of Gd3+ released occurs when considering the selectivity of the ligands for Gd3+ over other endogenous metal ions. GdDTPA-BMA show the smallest amount of Gd3+ released, after consideration of all the stability factors.

GdDTPA-BMA, Naz [GdDTPA] , GdDTPA-BP and formulations of these complexes. Unlike GdDTPABMA, the LDso of GdDTPA-BP remains virtually unchanged with addition of 5 mole % Na[CaDTPABP]. This is a result of the intrinsic high toxicity of Na[CaDTPA-BP] which was found to have a LDso value of 0.9 mmol/kg. While addition of Na [ CaDTPA-BP] may reduce the toxicity of GdDTPABP due to in vivo Zn2+ transmetallation, the addition of an inherently toxic component to the formulation will increase the toxicity of the formulation. Furthermore, in the case of Na2[GdDTPA] a decrease in LDSO value from 5.6 mmol/kg to 4.5 mmol/kg was seen by adding 5 mole % Na3 [CaDTPA] . In order to understand this result a formulation of GdDTPA-BMA + 5 mole % Na[CaDTPA-BMA] (LDSo = 38.3 mmol/kg) was corrected to the osmolarity of a formulation of Na2 [GdDTPA] by the addition of NaCl. A 500 mM formulation of GdDTPA-BMA + 5 mole % Na[CaDTPA-BMA] has an osmolarity of 707 mOs/kg. A 123 mole % addition of NaCl to this formulation raised its osmolarity to 1977 mOs/kg, comparable to that of a 500 mM solution of Na2 [GdDTPA] (1983 mOs/kg). The LD,, value for this “high salt”version of GdDTPA-BMA + 5 mole % Na[CaDTPA-BMA] was found to be 6.6 mmol/kg, not much greater than the LD5,, value found for

GdDTPA (5.6 mmol/kg). Thus, it seems that Gd3+ release may #not be the only mode of toxicity in Na2 [GdDTPA] . The toxicity of a Na2 [GdDTPA] formulation may in part result from osmotic shock. Hence the addition of 5 mole % of Na3 [CaDTPA] to Na2 [GdDTPA] would reduce in vivo transmetallation, but raise the osmolarity of the formulation 20% (5% x 4 ions). If the transmetallation theory of toxicity is correct for Gd3+ complexes then animals receiving multiple high doses of such compounds would be expected to show signs of zinc deficiency or depletion. Signs of zinc deficiency can include anorexia, retarded growth, testicular atrophy, hyperkeratotic dermatitis, loss of hair, parakeratosis of tongue and esophagus, gastric ulceration and diarrhea in animals.21*22 A subchronic toxicity study of GdDTPA-BMA with 5 mole 070Na[CaDTPA-BMA] in rats was performed by administering 0.1, 2.0 and 5.0 mmol/kg intravenously 3 times per wk for 3 wk. At the end of the third week, the rats receiving 5.0 mmol/kg GdDTPABMA (50 times the expected clinical dose) exhibited hair loss, skin lesions, pale stomach, and an average testicular weight loss of 57%. Histopathological findings included gastric mineralization and subacute inflammation, testicular giant cell degeneration and cutaneous mineralization with ulceration and eschar formation. The rats receiving 0.1 and 2.0 mmol/kg demonstrated none of these gross necropsy observations. The findings in animals receiving 5.0 mmol/kg dosages are consistent with the symptom spectrum of zinc depletion, and give further support to the hypothesis that zinc transmetallation is the primary mode of both acute and subchronic toxicity.

Thermodynamics There are several factors that must be considered when assessing the in vivo stability of Gd3+ complexes. Gd3+ and the ligand must have a strong affinity for each other, reflected by the thermodynamic stability constant for the complex. The effect of physiological pH on the stability of a complex can be readily determined if the pK, values of the ligand are known. The pK, values of the ligand along with the thermodynamic stability constant can then be used to determine a conditional stability constant which is a more accurate measure of the stability of the metal complex at physiological pH. Solubility Another important factor in determining in vivo stability of gadolinium complexes is the possible pre-

Thermodynamics cipitation

and toxicity l W.P. CACHERIS ET



of Gd3+ as GdPO+ KS,,,values of -0.4 and -5.1 have been incorrectly calculated’ for GdDTPA and GdEDTA respectively. These values imply that 71.5% of GdDTPA is converted to GdP04 in vivo, and that GdEDTA is virtually completely insoluble. In fact, clinical studies in humans indicate GdDTPA is nearly quantitatively excreted intact in the urine after intravenous administration,16 i.e. GdP04 precipitation does not occur. Even GdEDTA is sufficiently stable to undergo rapid, substantial (> 50% in 1 hr) renal excretion in animals. l5 Therefore, these log KS,,,value calculations are inconsistent with in vivo excretion studies. The negative values determined’ for log Ksol arise from two sources. The log[M] term for GdP04 used for the above log Ksol values was 4 x lo-l5 M, according to the solubility limit, or Ksp, of GdP04 at pH 7.4 and 2 mM total phosphate concentration. This is valid if the medium is saturated with GdP04 and no solubilizing l&and is present. The presence of such a ligand greatly enhances the solubility of the precipitate and hence increases the concentration of free Gd3+, which will make log Ksol more positive. Secondly, the metal ion interference terms, aCaL, aZnL and acUL must be calculated using thefree, not total, metal ion concentrations. These metal ions never exist completely in their dissociated form in vivo and will certainly bind to the ligand if Gd3+ is precipitated from the complex, decreasing their free concentrations. This will also make log Ksol more positive. Consideration of these factors reveals that complexes with stability constants comparable with GdEDTA should be completely stable with respect to phosphate precipitation. These calculations for in vivo stability suggest that any complex with moderately high conditional stability constants (>1014) will be soluble in vivo. Selectivity The most important thermodynamic criterion is the selectivity of the ligand for Gd3+ over other endogenous metal ions, particularly Zn2+. Ligands that have similar thermodynamic stability constants for Zn2+ and Gd3+ form complexes which are very toxic, due to Gd3+ release as a result of transmetallation of the complex in vivo with Zn 2+. Ligands that have high selectivity for Gd3+ over Zn2+ (and Ca2+) form complexes which are much less toxic. Once this selectivity criteria is satisfied, the acute toxicity of the complex may be reflective of the osmolarity of the formulation in addition to the amount of Gd3+ released in vivo. Kinetics The above criteria include only thermodynamic considerations, and can explain the observed toxicity

of GdDTPA-BMA, GdDTPA, GdDTPA-BP, and GdEDTA. Complexes whose structural features make in vivo transmetallation reactions much slower than their renal clearance rates have significantly improved toxicities than would be predicted by thermodynamics. GdDOTA is an example. GdDOTA releases Gd3+ so slowly that, under normal conditions, it is cleared from the body before the equilibrium concentration of Gd3+ can be attained via transmetallation. Factors that may slow the clearance of Gd3+ complexes from the body such as changes in renal function, binding to plasma proteins or conjugation of complexes to monoclonal antibodies would place a much greater importance on the kinetic stability of the complex.23 Slower clearance from the body would be likely to significantly increase the toxicity of GdDOTA or any other Gd3+ complex. Summary Figure 13 summarizes the variation between -log[Gd3+] released and the effects of stability constant, pH, solubility, metal ion selectivity and transmetallation kinetics for GdDTPA-BMA, GdDTPA, GdEDTA and GdDOTA. At a dosage of 0.1 mmol/kg, gadolinium salts (uncomplexed Gd3+) would produce a -log[Gd3+] value of 3.25. Complexation dramatically reduces the free gadolinium concentration, i.e. the -log[Gd3+] value increases. GdDTPA and GdDOTA release smaller amounts of Gd3+ compared with GdDTPA-BMA and GdEDTA when considering only their thermodynamic stability constants. The conditional stability constants predict a greater change in Gd3+ released for GdDTPA and GdDOTA as compared to GdDTPA-BMA and GdEDTA. All four complexes are expected to be stable toward phosphate precipitation in vivo: this effect does not change the amount of Gd3+ released. Consideration of the selectivity of the ligands for Gd3+ greatly alters the amount of Gd3+ released. GdDOTA shifts from being the complex with the least Gd3+ released to the one with the greatest metal released. GdEDTA also shows a large increase in the amount of Gd3+ released. GdDTPA-BMA shows the lowest Gd3+ released, consistent with the highest observed LD50 value. A consideration of the transmetallation kinetics only affects GdDOTA, raising the -log[Gd3+] value for this complex to a more favorable value. The factors which influence in vivo stability in gadolinium complexes highlight considerations which are important in ligand design. An increase in the basicity of the ligand will increase the thermodynamic stability constant of the complex in cases where steric effects do not predominate. An increase in the basic-


Magnetic Resonance Imaging 0 Volume 8, Number 4, 1990

ity of the oxygen donors (carboxylates) is most important since their intrinsic basicity always results in pK, values less than 6. Thus the carboxylate groups, unlike amine groups, will not experience proton competition at physiological pH values. In addition, a conditional stability constant of approximately 1Or4 determines the lower limit for the prevention of precipitation of GdP04 from a given complex. Selectivity of the ligand for Gd3+ over Zn*+, Cu2+, and Ca2+ is an extremely important factor in ligand design. Multidentate and macrocyclic ligands with chelate rings, or ligands with cavity sizes appropriate for Gd3+, ought to maximize Gd3+/Zn2+ selectivity. Donor sets that prefer the high charge-to-radius ratio of Gd3+ over Ca2+ should also help to maximize the desired Gd3+/Ca2+ selectivity. Complexes with reduced selectivity and reduced stability towards precipitation can be useful if they have slow kinetics of dissociation and if clearance is sufficiently rapid. However, a prolonged biological residence time might result in increased toxicity if a complex releases a relatively large amount of Gd3+ at equilibrium.

the addition of a toxic component to the formulation. Secondly, the addition of the Ca2+ salt of the ligand must not significantly increase the osmolarity of an already high osmolar formulation or the observed LDso value may reflect toxicity from osmotic shock. GdDTPA-BMA satisfies both of these conditions and formulations containing added Na[CaDTPA-BMA] are found to have LD,o values >30 mmol/kg in mice. Further studies directed at verifying these results with other polyaminocarboxylate complexes are ongoing.


trans-1,2-diaminocyclohexane-N, N’, N”, N”‘-tetraacetic acid 1,4,7,10-tetraazacyclododecane-N, N’, N”, N”-tetraacetic acid diethylenetriaminepentaacetic acid diethylenetriaminepentaacetic acid bis(methylamide) N, N”‘-bis(2-pyridylmethyl)diethylenetriamine-N, N’, N”-triacetic acid ethylenediaminetetraacetic acid nitrilotriacetic acid acetate N-methylglucamine

This study describes a detailed thermodynamic characterization of the relationship between in vivo stability and measured acute toxicity for four gadolinium polyaminopolycarboxylate complexes. The current concept of a direct relationship between the thermodynamic stability constant of the complex and observed toxicity is demonstrated to be incomplete. Ligands that exhibit enhanced selectivity for gadolinium over competing endogenous metal ions such as zinc, copper and calcium yield highly stable, and hence relatively nontoxic, complexes in vivo. Precipitation from competing biological anions does not appear to be a significant contributor to complex dissociation because of the relatively high stability constants of the complexes, and hence this factor is not anticipated to be important in the overall toxicity of the complexes discussed herein. The thermodynamic concepts discussed in this study have led to the development of formulations that minimize the amounts of gadolinium released under in vivo conditions. These formulations yield safer compositions when two basic criteria have been satisfied. The added Ca2+ salt of the ligand must have a favorable toxicity so that the decrease in toxicity from in vivo transmetallation is not offset or overcome by

0 II 0 II










HO& HO,'2


and toxicity 0 W.P. CACHERIS AL. ET


1. Lauffer, R.B. Paramagnetic metal complexes as water proton relaxation agents for NMR imaging: Theory and design. Chem. Rev. 87901-927; 1987. 2. Quay, S.C. Diamine-DTPA paramagnetic contrast agents for MR imaging. U.S. Patent No. 4,687,659; 1987. 3. Rocklage, SM.; Sheffer, S.D.; Cacheris, W.P.; Quay, S.D.; Hahn, E.F.; Raymond, K.N. Structural and thermodynamic characterization of manganese(E) N,N’dipyridoxyl-ethylenediamine-N,N’-diacetate. A novel manganese(H). Znorg. Chem. 27:3530-3534; 1988. 4. Wright, D.L.; Holloway, J.H.; Reilley, C.N. Heats and entropies of formation of metal chelates of polyamine and polyaminocarboxylate ligands. Anal. Chem. 37:
884-892; 1965. 5. Yang, B.; Yang, P.; Song, L. The binding of gadolinium(II1) to human serum albumin. I(exue Tongboa

14. Kelly, J.J.; Sutton, D.C. Prediction and measurement of effect of chelating selectivity on precipitation reactions. Talanta 13:1573; 1966. 15. Fornasiero, D.; Bellen, J.C.; Baker, R.J.; Chatterton, B.E. Paramagnetic complexes of manganese(E), iron(III), and gadolinium(II1) as contrast agents for magnetic resonance imaging. The influence of stability constants on the biodistribution of radioactive aminopolycarboxylate complexes. Invest. Radiol. 22:322-327;
1987. 16. Weinmann,

R.J.; Laniado, M.; Muetzel, W. Pharmacokinetics of gadolinium-DTPA/dimeglumine after intravenous injection into healthy volunteers. Physiol.
Chem. Phys. Med. NMR 16:167-172; 1984.

29:1502-1505; 1984. 6. Smith, R.M.; Martell, A.E. Critical stability constants, Vols. 1-4. New York: Plenum; 1974; 1975; 1976; 1977. 7. Hartley, F.R.; Burgess, C.; Alcock, R. Solution equilibria. New York: R. Halsted Press; 1980. 8. Spaulding, L.; Brittain, H.G. Complexation of amino acids by terbium(II1) ethylenediaminetetraacetate. ZnJ.C.; Weiss, P.; Motekaitis, R.J. Inorganic chemistry in biology and medicine. ACS Symposium Series, 391; 1980. 10. Nyssen, GA.; Margerum, D.W. Multidentate ligand kinetics. XIV. Formation and dissociation kinetics or rare earth-cyclohexylenediaminetetraacetate complexes. Znorg. Chem. 9:1814-1820; 1970. 11. Rosiello, A.P.; Essigmann, J.M.; Wogan, G.N. Rapid and accurate determination of the median lethal dose (LD,,) and its error with a small computer. f. Toxicol.
Environmental 12. Weinmann, Health 3:797-809; 1977. org. Chem. 24:3692-3698; 1985. 9. Rubin, M.; Martell, A.E.; Gohil, R.; Penhos,

17. Krasnow, N. Effects of lanthanum and gadolinium ions on cardiac sacroplasmic reticulum. Biochem. Biophys. Acta. 282: 187-194; 1972. 18. Brucher, E.; Szarvas, P. Kinetic study of the isotopeexchange reactions of the central ions of the lanthanide ethylenediaminetetraacetate and trans-1,2-diaminohexaminetetraacethe complexes. Znorg. Chem. Acta. 4:632; 1970. 19. Delgado, R.; Frausto da Silva, J.J.R. Metal complexes of cyclic tetraazatetraacetic acids. Talanta 29:815-822; G.; Makra, Z. Studies on the kinetics of formation and dissociation of the cerium(III)DOTA complex. Znorg. Chim. Acta. 139:141-142; 1987. 21. Baker, H.J.; Lindsey, J.R.; Weisbroth, S.H. The laboratory rat. Orlando, FL: S.H. Academic Press; 1979:~. 138. 22. Cho, C.H.; Fong, L.Y.Y.; Ma, P.C.C.; Ogle, C.W. Zinc deficiency: Its role in gastric secretion and stressinduced gastric ulceration in rats. Pharmacol. Biochem.
Behav. 26~293-297; 1987. 23. Cole, W.C.; DeNardo, S.J.; Meares, C.F.; 1982. 20. Brucher, E.; Laurenczy,

H.J.; Brasch, R.C.; Press, W.R.; Wesbey, G.E. Characteristics of gadolinium-DTPA complex: A potential NMR contrast agent. Am. J. Roent. 142:619624; 1984. 13. Pitt, C.G.; Martell, A.E. Inorganic chemistry in biology and medicine. ACS Symposium Series 285; 1980.

McCall, M.J.; DeNardo, G.L.; Epstein, A.L.; O’Brien, H.A.; Moi, M.K. Serum stability of copper-67 chelates: Comparison with indium-111 and cobalt-57. Nucl. Med. Biol. 13:363-368; 1986.

Sign up to vote on this title
UsefulNot useful