WO1999052564A1 - Method - Google Patents

Abstract

This invention relates to a method of measuring magnetic properties of water-insoluble particulate materials, in particular particulate magnetic resonance (mr) contrast agents. The method comprises disposing said sedimentable agent in a flowable aqueous solution of a polyalkyleneoxide-based matrix forming agent, optionally cooling the resulting dispersion to form a non-flowable aqueous dispersion of said sedimentable agent, measuring a magnetic property of said dispersion, and if desired calculating from the measured magnetic property a magnetic property of said agent or said dispersion. The method is useful in mr imaging applications.

Description

Method

This invention relates to a method of measuring magnetic properties of water-insoluble particulate materials, in particular particulate magnetic resonance (mr) contrast agents.

In magnetic resonance imaging, image contrast enhancement may be achieved by administration to the patient of a contrast agent which affects the water proton relaxation times and thus alters the mr signal intensity for tissues, organs or body cavities into which it distributes. Several such mr contrast agents are available commercially, e.g. Omniscan^®, Magnevist^® and ProHance^®.

Since the contrast effect of an mr contrast agent is related to its relaxivity (a measure of its effect on the relaxation times of water protons) , it is clearly important to be able to study the relaxivity of mr contrast agents and candidate mr contrast agents for development. This is a relatively straightforward task achievable by NMRD investigation of contrast agent solutions where the contrast agent is a water-soluble molecule, e.g. as is the case for the gadolinium chelate compounds of Omniscan, Magnevist and ProHance.

However, for insoluble particulate agents and other materials which have a tendency to sediment out from aqueous solutions or dispersions, it is necessary to suspend the agent in an aqueous matrix to perform any type of T_x or T₂ measurement, for example Nuclear Magnetic Relaxation Dispersion (NMRD) measurements. (NMRD involves the measurement of the l/T_x or l/T₂ values for solvent (usually water) protons as a function of magnetic field strength. From these measurements the relaxivity profiles of the solute or any material otherwise dispersed in the solvent may be determined) . Such water-insoluble particulate agents and agents which show a tendency to sediment out of aqueous solution or dispersion are referred to herein as "sedimentable agents" . Typically, for NMRD studies with sedimentable agents, agarose media (gels) have been used (see for example Josephson et al . in Magn. Res. Imaging 6_.: 647- 653 (1988) and Jung et al . in Magn. Res. Imaging 13 : 661-674 (1995)) . Agarose media have been used because they are readily available, relatively simple to handle, non-toxic, and, unlike many other media (such as acrylamide-based media) , do not require a chemical cross-linking reaction.

However we have found that the use of agarose media in this context leads to certain drawbacks which can be avoided or reduced by the use of a polyalkyleneoxide- based matrix forming agent. Two particular problems encountered with agarose media relate to the preparation of a dispersed sedimentable agent containing matrix and to the extraction of the contribution of the sedimentable agent to the NMRD profile (l/T_l or l/T₂ profile) of the sedimentable agent-containing matrix. These problems are avoided by the use of a polyalkyleneoxide-based matrix forming agent.

Two other commonly used suspension media are polyvinylalcohol (PVA) and polyvinylpyrrolidone (PVP) . For these (and indeed all) suspension media, the ability to suspend particles for any length of time will have to be tested on a case-by-case basis because this property is particle-dependent as particles will differ in size, density, and tendency to aggregate. Once the ability of a medium to satisfactorily suspend particulates is confirmed, the suitability of these media can be tested with respect to their applicability for relaxation rate measurements . PVA and PVP both have the disadvantage that they each cause an increase in microviscosity compared to water. An increase in the microviscosity causes a lengthening of the rotational and translational correlation times of the agent, leading to an increase in relaxivity. This causes the relaxivity of the agent to be suspension medium-dependent , thus complicating any comparison of relaxivity (as well as the correlation times that determine the relaxivity) to that of other agents measured in a different medium (typically water, which is the most relevant and frequently used medium for MRI) .

Thus viewed from one aspect the invention provides a method of determining magnetic properties of a sedimentable agent, e.g. a water-insoluble particulate material, said method comprising dispersing said sedimentable agent in a flowable aqueous solution of a polyalkyleneoxide-based matrix forming agent, optionally cooling the resulting dispersion to form a non-flowable aqueous dispersion of said sedimentable agent, measuring a magnetic property of said dispersion, and if desired calculating from the measured magnetic property a magnetic property of said agent or said dispersion.

Viewed from a further aspect the invention provides an aqueous dispersion, e.g. a non-flowable dispersion, comprising a sedimentable agent (e.g. a water-insoluble particulate material) and a polyalkyleneoxide-based matrix forming agent .

Viewed from a yet further aspect the invention provides the use of polyalkyleneoxide-based matrix forming agent for the preparation of a dispersion for use in the determination of magnetic properties of sedimentable agents (e.g. particulates) .

In the method of the invention, the magnetic property measured or calculated may for example be a water proton relaxation rate (e.g. 1/T_X or l/T₂) , T_x or T₂ relaxivity of the sedimentable agent or of the dispersion, NMRD profiles, magnetization, etc. Measurement may be effected by conventional means using equipment which is available commercially or has been described in the literature, e.g. a field cycling relaxometer, a Minispec PC- 12Ob (from Bruker GmbH, Rheinstetten, Germany), etc. Measurement of magnetization may be performed conventionally by measurement of magnetic susceptibility or hysteresis curves (magnetization as a function of applied magnetic field) . For a detailed description of field-cycling relaxometry as a means of determining NMRD profiles see Koenig et al . Prog. NMR Spectr. 22: 487-565 (1991). The sedimentable agent investigated using the method of the invention may be any water-soluble precipitable or agglomeratable material (e.g. a material which is subject to precipitation at or near ambient or physiological temperatures or pH's or a material which although initially water soluble agglomerates or precipitates over time) or water- insoluble particulate material, e.g. solid or deformable (ie. semi-solid) particles, liquid droplets, gas bubbles or vesicles (e.g. microbubbles, microballoons, liposomes, micelles, etc) . The particles may have a simple uniform structure (e.g. microcrystals or liquid droplets) or they may have a more complex structure, e.g. porous particles or membrane-containing particles. Since the sedimentable agent will generally be an mr contrast agent or a material under investigation for possible use as an mr contrast agent, it will generally contain paramagnetic centres or comprise materials exhibiting or capable of exhibiting cooperative magnetic behaviour, e.g. ferromagnetism, ferrimagnetism and superparamagnetism. Paramagnetic centres as mentioned above may for example be metal ions (e.g. transition metal, lanthanide or actinide ions), in particular chelated metal ions, or free radicals. Materials exhibiting cooperative magnetic behaviour or capable of exhibiting such behaviour will typically be iron oxides, e.g. magnetite. Such magnetic materials may constitute the whole or only part of the particles; thus for example such magnetic materials may be loaded into the pores of a porous support (e.g. a zeolite) , or they may be conjugated to the surface of the particle, or they may be contained within a membrane, or they may be particles provided with a coating or with surface attached materials (e.g. polyethylene glycol groups attached to prolong the blood residence time of the particles) .

Where the sedimentable agent is a particulate material, it will generally have a mean particle size (determined for example with a Coulter Counter) in the range 0.005 to 100 μm, preferably 0.010 to 20 μm, more particularly 0.030 to 10 μm, e.g. 0.030 to 0.2 μm or 1 to 10 μm. The smaller particles will tend to comprise materials exhibiting or capable of exhibiting cooperative magnetic behaviour while the larger particles will tend to be ones containing paramagnetic centres .

The polyalkyleneoxide-based matrix forming agent is preferably an amphiphilic block copolymer comprising at least one polyalkyleneoxide block in which the alkylene moieties are lower alkylene (e.g. C₂_₄, preferably C₂) and at least one relatively hydrophobic block (e.g. a higher alkylene moiety, for example a C₈_₃₀, preferably C₁₂_₂₅, more preferably C₁₅.₂₀ alkylene moiety) .

Typically the matrix forming agent will have a molecular weight in the range 5 kD to 20 kD, preferably 8 to 12 kD. Such materials may for example be esters of alkylcarboxylic acids or alkylcarbonyloxyalkylcarboxylic acids with polyethyleneglycols . One suitable compound is

OCH₂(CH₂OCH₂)_nCH₂OCH₃

PEG-BC or α- (16-hexadecanoyloxyhexadecanoyl) -ω-methoxy- polyoxyethylene ester. This has a molecular weight of about 10 kD and has typically been used as an emulsifying or stabilizing agent. The preparation and use of PEG-BC are described in WO96/07434 (see Example 2k) .

The polyalkyleneoxide-based matrix forming agent is preferably a material which in aqueous solution produces a flowable liquid at elevated temperatures, e.g. up to 60°C, more preferably up to 45°C, yet which on cooling to ambient temperature (e.g. to 15°C, more preferably to 20°C) yields a non-flowable material. In this way, it is possible to produce a non-flowable dispersion of the sedimentable agent by heating a mixture of water and matrix forming material, e.g. to 35-45°C, to produce a non-viscous, free-flowing liquid, dispersing the sedimentable agent in this liquid, e.g. with agitation, stirring or sonication, and cooling to allow a non- flowable dispersion to form. During cooling, it will generally be desirable to treat the dispersion to prevent settling out, especially if the sedimentable agent comprises particles of micron or larger size. Such treatment may for example involve rotation of the liquid about a horizontal axis and in a gas-free container, or continuation of sonication.

In this way a dispersion can be produced without thermal degradation of heat-sensitive or thermally labile sedimentable agents. With some agarose gels, a high temperature is required for its dissolution and the sedimentable agents have to be added to this hot solution before the medium sets.

However, some low gelling-temperature agarose gels - 7 - do exist, and these gels typically do not set above a temperature of about 40°C. This implies that it may be possible to add particles at about the same temperature as is done with the PEG-BC medium (see Example 2 below) . However, these low gelling-temperature agarose gels are still inferior to the PEG-BC with respect to the field dependence of the 1/T_X (and l/T₂) of the suspension medium. While the PEG-BC medium has a l/Ti (and 1/T₂) that is independent of field strength with the same value as for water, the low gelling-temperature agarose gels have about the same field dependence as the higher gelling-temperature agarose gels such as those used in the following Examples.

In general, to form the dispersion, the matrix forming agent will be present at from 0.1% by weight up to its water solubility limit, e.g. from 2 to 10% by weight, preferably 3 to 7% by weight relative to the total weight of water and matrix forming agent .

The aqueous matrix forming agent solution is preferably a material with a 1/T_X or 1/T₂ NMRD profile, especially in the proton Larmor frequency range of 0.1 to 10 MHz, more preferably the range 0.001 to 100 MHz, which is similar to that of water, e.g. with a variation of no more than 20%, more particularly no more than 10% in proton T_x relaxation rate beyond that of water in this frequency range and at temperatures in the range 15 to 25°C, preferably within the range 5 to 35°C.

With such water-like NMRD profiles it is much more straightforward to calculate the relaxivities for the sedimentable agent from the measured relaxation rates than is possible with agarose which makes a relatively large and magnetic field dependent contribution to the proton relaxation promoting effect of the dispersion as a whole. For agarose media, in order to determine the contribution to the NMRD profile of the sedimentable agent, it is necessary to subtract the NMRD profile of the agarose medium itself. This can lead to large Ω 3 P- 0 P⁾ 3 TJ 3 pj TJ P- 3 pi m cr a hi rt CQ P- a a Hi P- ft⁾ H P- Φ TJ a ft⁾ ra ft φ ft⁾ Φ 3 ø 0 Φ 0 Φ , t⁾ P- ^•g hi h-¹ Φ rr 0 Φ 3; Φ r 0 0 Φ P- 0 ø ø X 0 P- LQ PJ rr h|

P¹ a TJ a ft⁾ Φ h-¹ a Φ ra » P¹ Φ CQ < It CO 0 0 CD a ft⁾ ra Φ 3 o^j hi

Ω P- 0 ^ f 1 O Φ TJ 0 P- TJ 3 α ft⁾ "< TJ φ Φ Ω ≤ CQ 0 It 0 3 ^ TJ 0 Φ tt 0 >

0 0 hi LQ 0 0 0 PJ CD 0 Ω Φ Φ X Φ 0 hi 0 ft⁾ H rr Ω J ft⁾ φ It hi ( 1

H 3_^ rt ft Φ 3 φ hi ø hi P> rt 3 H ø rr 3 P- 3 0 > ft 3 CO cr 3 0 p- ft P" P" hi - rjd 3 rr CD o 4-\- ft⁾ P⁾ rr 0 hi Φ CQ Φ ra H P⁾ a Φ < Φ a P- P- CQ P¹ ft⁾ ft CQ P- φ ? ra Φ PJ PJ ft 0 Φ (t P- 3 a rt ft⁾ ra ra Φ P- OJ P- Φ P- 0 0 0 Φ φ ^ P- < ; P- 3 CO LQ < CD

P- Ω rt ^ Ω φ CQ rt ft⁾ CQ a 0 CQ ft rt ø m 0 ft⁾ a 0 h{ Φ ft⁾ 0 Φ P- ø Φ P-

0 0 . Φ P- rr 0 0 cr Ω Φ r 0 LQ φ ft ft⁾ cr ø Φ ft Φ ø rt a Φ 0 ø 0 hh O P- CQ rt M ^ hj a P- o PJ ft⁾ hi < hh P- ft! sQ Φ LQ ^ 0 0 ra OJ Φ it TJ Ω rt π hh CO 0 CD < P- Q Hi rt a Φ ft⁾ 3 0 0 cr ø o ft⁾ 3 Φ P¹ CO P- hi Φ

0 ft⁾ 0 Φ Φ cr - Φ a ft ft⁾ Hi a 3 hi Φ P¹ 0 0 Φ ft⁾ Hi hi Ω p- 0 Ω Φ hh p- ø Ω X P- CQ 0 P- Φ ra P- CQ P- φ *τ P- a < 0 a a 0 Ω 0 Hi P- rt Ω P- ø CQ ft TJ m ft P- ft ra H φ 0 Φ ft 0 0 0 P- ≤ Hi Φ ø a ft 0 X 0 rr hh P- (t hj CQ Φ hi P⁾ ft • CO P- Hi CO O ft⁾ P- ft hi 0 ft> ft⁾ 0 . CO Φ rr TJ ft⁾ P- ft⁾ 0 Φ P- ra

Φ Λ O Φ p. ø 0 0 0 TJ CD 0 O CD ^ hi a Φ ft H σ a Φ Φ P-

P^J m 0 Hi CQ φ Ω P- 0 H PJ φ ft ø _^ ft⁾ Hi 3 CO rr Φ O Φ 3 Ω 3 P¹ o ft⁾ J Φ CQ P¹ φ CD 1 < P¹ 0 Ω 0 ft CQ ι-3 0 &⁾ 3 Φ a 0 1 Φ 0 ft⁾ a ^

X φ 0 rt Φ P⁾ 0 P- 1 0 a H rr ft CO hi hi 0 rr P¹ CQ Φ hi σ It 0 LQ a

P- Ω rt rr a rt P⁾ 0 hi ø £ P- 0 0 rr 0 P- ft⁾ P- 0 rt 3 TJ TJ ft^j ft⁾ Ω 0 a rt P-

< P- P¹ Φ Φ CQ Φ LQ > ra rt 0 ø Φ 03 ft 2 Ω CQ ft o) ft⁾ ft^{) 1} ft D h-¹ Φ φ Φ rr Hi

P- φ hi

^ a Ω 2 •• rr LQ LQ rt hi TJ Φ * 0 hh P- P¹ It ft⁾ P- Φ 0 ft TJ Φ hh

(t CO ft⁾ Φ J Φ § ft⁾ ra hj rr Φ Φ h{ 50 3 ø 3 0 TJ hi CQ 0 a P- ft P- Φ P-

^ LQ h-¹ rt 0 ra J 0 φ 3 φ Φ 0 P- α TJ hi φ 0 PJ P- P- 3 0 0 hi Ω ø Q Ω

0 P- Φ ft⁾ 0 ra CD α J O φ φ 0 PJ a PJ ft⁾ It rt hi 0 X ft⁾ 3 0 ft⁾ a ft⁾ 0

P- rt hh ø rt O PJ 0 TJ a 0 3 hi P- PJ hi rr rr 0 r 0 1 CD ft⁾ ft TJ Φ 3 P¹

CD r rt P- ft p- hi ⁾ H φ 0 P- 0 Φ φ ø hi J Φ φ 0 Hi P- CQ Hi ft⁾ TJ rt P- P- hi ø Φ It φ ft <! rr cr ^ ø 0 hh ft⁾ LQ a LQ Φ TJ hi a Ω 0 ft Φ hi 0 0 0 Ω oo

Ω hi rr 0 φ Φ a O rr P- ω It ft 0 0 hi 0 Ω P- 0 TJ ^ h{ It

0 Φ 0 Φ ft φ P- ft hi 0 3 J P- rr ft⁾ 0 rr ^ hi 3 3 ft X PJ Φ Φ 0

0 rt rt hh 0 0 hi 0 rr Φ ft 3 ra Φ rt Ω Φ ra Hi φ ft⁾ P- P- hi 0 Ω hi r P¹

CO rr CO 0 PJ rt rr P- 0 Φ J 0 ø a P- ft⁾ TJ ft 3 ø Ω 0 hh ft⁾ ft Φ ft⁾ TJ

P- P⁾ 0 ft⁾ Ω 0 P⁾ rr Ω ra hi Ω P- 0 ft CD φ 3 3 φ 0 LQ CO 0 ft⁾ 0 P- ø X hi a 0 rt CO rt a Φ 3 Φ Φ 3 Φ rr 0 LQ P- 0 Ω Φ ft⁾ hj Φ Ω hi Λ Φ Φ ft⁾ 0

Φ < Φ PJ J a h{ CD φ CQ ft⁾ Φ 3 0 P- rt It rt 0 Φ ft⁾ 3 0 3 0 cr CQ PJ It a p. ss φ hi rt hj 0 a Ω Φ rV TJ Ω rt ø (t ft⁾ ι-i hi LQ P- J P- Φ Φ It P- 0

P⁾ PJ 0 TJ rr Φ ø⁾ hi σ P- hi 0 Φ P- hi rt 0 CQ P⁾ 0 P¹ P- LQ φ ra ^ ø 0 0 rr 0 Ω cr ft ft TJ P¹ PJ J CQ Φ CO 0 P- ft 0 P- 0 cr hh X CQ Φ 0 Q LQ 0 a hh 0 Φ

P" φ hi φ rt P⁾ P¹ rt 3 rt Φ ft LQ rt 3 P- P- P- 0 0 rt P- _^-_^ CD Φ H h{ . — . 0 hi P¹ PJ ft^j Φ a rr Φ cr Φ CQ -> It 0 Hi Φ 0 ft⁾ It ft⁾ O hi t⁾ ft⁾ g P⁾ O rt <! hi Φ P- X ø P- hi φ ft hi Φ J P- rr 0 0 0 hi 3 ø LQ O Φ ft⁾ LQ

Ω TJ 0 0 0 φ rt 0 Ω CQ P- cr rr P- to ft ιt Φ hi ft f Φ O hi CQ 2 ft PJ

0 hi CQ 0 M ft⁾ rt 3 Φ P⁾ ^ 3 Φ ft⁾ ø 0 0 >< Ω 3 P- •<; CQ 0 P¹ 3 Φ X Φ hi

3 0 J CD 0 P⁾ H Φ 0 Φ ra hi TJ 0 P- 0 P" J 0 It 0 P- a N CQ 0

TJ ft Φ 3 X CO Φ rt P- m ft⁾ ft⁾ J ft φ Φ Ω 0 hi 0 0 ø >< cr Φ hi » ft 0 P- • CO

P" 0 0 P- φ P- 0 i 0 0 ra pj rr Hi 3 0 hh φ Ω LQ Φ Ω P- φ 3 ft⁾ φ

P- 0 CD CD <; CQ 0 P- ft⁾ TJ 0 hi ft⁾ 0 Φ 3 ra Φ rt ft ft 0 0 a φ 0

Ω CQ P- P- TJ Hi Ω ft⁾ M hi rt rt ø TJ φ 0 ft⁾ 0 rr PJ hi CD 0 0 a 3

P⁾ - 0 cr ft Φ 0 Ω φ P- rt ⁾ ø rt LQ Φ a 0 P- hh rin φ H ft 0 0 ^ 0 rt *< 1 3 Ω 3 Hi h{ It hi φ ft⁾ t⁾ ft

(t a CQ 0 0 hi f a

Φ rt CD rr ω Φ P¹ ø 0 0 Φ ft⁾ 0 3 Φ Ω LQ Hi 0 cr rr P- rr 03 a 0 P- Φ n ø Φ CO hi Hi ft It Φ Φ hh a 0 3 p^j φ ft⁾ 0 rt CQ ft rr P- rt J Hi ft⁾ Φ 0 0 0 ft⁾ ft⁾ It 0 ≥

0 CQ H 0 hi

- 9 - because the volume fraction of water in the sample is not easily evaluated. Typically, one assumes that the mass fraction is equivalent to the volume fraction; for example, a 5% gel is assumed to cause the relaxivities to be 5% higher than would be the case in water. Consequently, the relaxivities measured using the PEG-BC medium must be multiplied by 0.95 (0.95 = 1 - 0.05) to obtain relaxivities equal to what would be measured in water.

2nd criterion - The water proton relaxation rates (1/Tχ and 1/T₂) of the suspension medium must be independent of magnetic field strength and should be equal to those of water. This means that the polymeric substance, dissolved in water, cannot affect the water proton relaxation rates in any manner; the polymeric substance, or the effect of its presence in the solution, should be "invisible" to water protons.

3rd criterion - The polymeric substance must have no influence on the relaxivity of the agent . This implies that the polymeric species cannot influence the microviscosity - the viscosity of the medium immediate to the agent. For example, an increase in the microviscosity would cause a lengthening of the rotational and translational correlation times of the agent, leading to an increase in relaxivity. This would cause the relaxivity of the agent to be suspension medium-dependent , thus complicating any comparison of relaxivity (as well as the correlation times that determine the relaxivity - see Table 1 in the Examples attached hereto to that of other agents measured in a different medium (typically water, which is the most relevant and frequently used medium for MRI) .

It is important to note that the second and third criteria are not necessarily equivalent : the presence of a polymeric agent in water can cause the relaxation rates to increase without requiring any significant increase in the microviscosity (Koenig, Brown III, - 10 -

Ugolini, Magn. Reson. Med. 1993;29:77-83).

The present invention seeks to achieve a suspending medium which meets all of the 3 relaxivity measurement criteria, as will be demonstrated in the following Examples attached hereto.

Agarose meets all of the above criteria except the second (see Figs. 5, 7, 8 and 9 attached hereto) . Indeed, recent work has shown that the relaxivity of water soluble contrast agents was identical in water and in agarose gel. Furthermore, the mass fraction of agarose was only 0.8% and could therefore be neglected. On the other hand, the mass fraction of the PEG-BC was 5%, which is significantly higher than that used for agarose gels. At first glance, this may appear to be an advantage for the agarose gels. However, the correct (volume fraction adjusted) relaxivities of GdDTPA (Magnevist^®) in the PEG-BC medium could be trivially calculated by multiplying the measured relaxivities by 0.95, as was verified (see Fig. 1 and Table 1 hereto) . Also, the microviscosity of agarose gel is analogous to that of water (see Fig. 9 hereto, above 10 MHz) . Figure 5 shows that the magnetic field dependence for 1/Tχ (and therefore l/T₂ by extension) of agarose gels is only significant at field strengths corresponding to proton Larmor frequencies below 1 MHz. Clinical MRI is performed at field strengths higher than 1 MHz, so determining the relaxivities for clinical concentrations of agent in agarose gels is acceptable, but only for the T_: relaxivity. Concerning the T₂ relaxivity, it is well known that the value of 1/T₂ at MRI field strengths is related to the value of l/T_x at low field strengths (0.01 MHz and below - see Koenig, Brown III, Ugolini, Magn. Reson. Med. 1993;29:77-83). Because the determination of relaxivity involves subtraction of the l/Ti (and l/T₂) of the "pure" suspension medium, the disadvantage of agarose gels is that the low-field values of l/T_x are not reproducible. Figure 7 shows the NMRD profiles for 3 - 11 - different preparations of a 0.8% agarose gel. Although the high-field (greater than 1 MHz) 1/Tj_. values are reproducible, the low-field values are not. Therefore, the importance of the second criterion comes down to:

(1) the accuracy of the measurement of the T₂ relaxivity at MRI -relevant fields; and (2) to consequences of not being able to accurately determine the value of the T_x relaxivity at low-fields.

These "low-fields" represent a field strength range where some of the relaxation mechanisms that ultimately determine or influence the relaxivity are revealed. One consequence of inaccurate low-field relaxivities relates to an uncertainty in the evaluation of correlation times that determine the electron spin correlation time, τ_v and τ_s0 (see Table 1 attached hereto) ; correlation times that are especially important to assess for macromolecular contrast agents with long τ_R values (see Kellar et al., Magn. Reson. Med. 1997; 38: 712-716, and Koenig, Invest. Radiol. 1994; 29: S127-130) . As another example, any uncertainty in low-field relaxivities for the DyDTPA-SP

(see Fig. 2 hereto) would have made it impossible to determine the significant diamagnetic contribution of the starch matrix (see Fig. 3 hereto) of the particles to the relaxivity of the DyDTPA-SP. If agarose gels were used, the diamagnetic contribution would have been difficult, if not impossible, to identify and quantify. This is true regardless of whether the agarose is of a low gelling-temperature type or not.

In addition to allowing reproducibility of low- field values of l/T_l7 PEG-BC shows unexpected advantages over two other commonly used suspension media, polyvinylalcohol (PVA) and polyvinylpyrrolidone (PVP) . PVA and PVP both failed to satisfy the third criterion, as both caused an increase in microviscosity compared to water. NMRD profiles were measured for GdDTPA in these media analogous to what was done for PEG-BC in Fig. 1 hereto. Figs. 10-12 hereto show the NMRD profiles of - 12 -

GdDTPA in 5% PVA, 5% PVP, and 15% PVP, respectively (the open symbols are the relaxivities of GdDTPA in water, reproduced from Fig. 1.). Clearly, especially for the 5% PVA and 15% PVP, the relaxivities are significantly higher than those in water, and not by a constant factor at all field strengths. This indicates that these polymeric substances cause an increase in the microviscosity of the suspension medium. Such a feature has also previously been demonstrated at 5°C for PEG BC both qualitatively (see Fig. 1 hereto) and quantitatively (see Table 1 hereto) . This interpretation can be extended to the results in Figs. 10-12 hereto. Consequently, the PVA and PVP media fail with respect to the third criterion, even at the lowest concentration of 5%, and are inferior to PEG-BC as suspension media for relaxometric measurements.

The invention will now be described further by reference to the following non-limiting Examples and the accompanying drawings, in which:

Figure 1 shows a set of NMRD profiles (T_x relaxivity vs proton Larmor frequency) for a water-soluble gadolinium chelate in a PEG-BC medium;

Figure 2 shows a set of NMRD profiles (^rT₁ relaxivity vs proton Larmor frequency) for a water-soluble dysprosium chelate and for a particulate-bound dysprosium chelate in a PEG-BC medium;

Figure 3 shows a set of NMRD profiles { _λ relaxation rate vs proton Larmor frequency) for a non-paramagnetic particulate in water;

Figure 4 shows a set of NMRD profiles (T_x relaxivity vs proton Larmor frequency) for a particulate-bound gadolinium chelate in a PEG-BC medium;

Figure 5 shows a set of NMRD profiles (T_: relaxation rate vs proton Larmor frequency) for an aqueous agarose medium; and

Figure 6 is a diagrammatic representation of a starch particulate bound gadolinium chelate. - 13 -

Figure 7 shows a set of NMRD profiles (T_: relaxation rate vs. Proton Larmor frequency) for three replicate preparations of aqueous agarose medium;

Figure 8 shows a set of NMRD profiles (T_λ relaxation rate vs. Proton Larmor frequency) for a water-soluble gadolinium chelate in an aqueous 0.8% agarose medium;

Figure 9 shows a set of NMRD profiles (T_x relaxivity vs. Proton Larmor frequency) for a water-soluble gadolinium chelate in an aqueous 0.8% agarose medium, 5% PEG-BC medium and water;

Figure 10 shows a set of NMRD profiles (Υ_λ relaxivity vs. Proton Larmor frequency) for a water- soluble gadolinium chelate in an aqueous 5% PVA medium;

Figure 11 shows a set of NMRD profiles {T_λ relaxivity vs. Proton Larmor frequency) for a water- soluble gadolinium chelate in an aqueous 5% PVP medium;

Figure 12 shows a set of NMRD profiles (T_t relaxivity vs . Proton Larmor frequency) for a water- soluble gadolinium chelate in an aqueous 15% PVP medium; EXAMPLE 1 Preparation of DTPA-SP. GdDTPA-SP and DyDTPA-SP

The starch particles consisted of swellable, epichlorohydrin cross-linked hydrolyzed potato starch non-labelled (empty) or labelled with GdDTPA or DyDTPA, referred to as DTPA-SP, GdDTPA-SP and DyDTPA-SP (see Fig. 6) . Particulate formulations were prepared and lyophilised until further use (Virtis Benchtop Lyophilisator BT-5L, Virtiscomp, Gordimer, NJ) , as described by Rongved et al . in Carbohydrate Research 214: 325-330 (1991) .

The metal content of the starch particles was determined by inductively coupled plasma atomic emission spectrophotometry (ICP-AES, Perkin Elmer Plasma 2000, Norwalk, CT) - see Fossheim et al . Magn. Res. Med. 35 : 201-206 (1996). The metal content of the DyDTPA-SP was 4.9% (w/w) Dy and for the GdDTPA-SP, the metal content was 4.0% (low loading) and 6.7% (high loading). For - 14 - particle size analysis, particles were suspended in deionized water and sonicated for ten minutes. The mean volume-weighted particle diameter was determined by the Coulter Counter technique (Coulter Counter Multisizer II, Coulter Electronics Inc., Luton, England), with constant mechanical stirring. The mean volume diameters of the DyDTPA-SP, low and high loading GdDTPA-SP were 3.2, 2.3 and 2.4 μm, respectively.

EXAMPLE 2

Preparation of the PEG-BC Medium for Suspending Particles

A medium containing 5% (w/w) of the amphiphilic PEG-BC, α- (16-hexadecanoyloxyhexadecanoyl) -ω- methoxypolyoxyethylene ester (Nycomed Amersham Imaging, Wayne, PA) , was prepared by adding 5g of polymer to 95g distilled water. The aqueous mixture was heated, with stirring, until a clear solution resulted. The solution was allowed to cool to room temperature, where it remained a free-flowing, but somewhat viscous, solution. It was not until the solution was cooled below about 20°C that a gel formed.

Relaxation analyses of this medium were performed at 0.47 T (Minispec PC-120b, Bruker GmbH, Rheinstetten, Germany) . The T_x was obtained using the inversion recovery pulse sequence and T₂ was measured by the Carr- Purcell-Meiboom-Gill method. The values of l/T_x obtained at 5, 25 and 35°C were 0.700, 0.401 and 0.294 s^"1, respectively. The value of l/T₂ was measured at 35°C, but was so long (equal to that of l/T_x within experimental errors) that it could not be determined accurately. The values of l/Υ_γ at 25 and 35°C for the medium were identical to those of water, and were so low (at all temperatures) that measuring the NMRD profiles would be impractical. However, determining the rates as - 15 - a function of field strength was not necessary since they were expected to be field independent . This assumption was substantiated as the l/T_: of the medium was similar to that of water and the l/T_x and l/T₂ values of the medium were similar (the high-field value of l/T₂ reflects the low-field value of l/T_x - see Koenig et al . Magn. Res. Med. 29_: 311-316 (1993).

EXAMPLE 3

Sample Preparation for NMRD Studies

Particles were suspended in the PEG-BC medium to prevent aggregation and sedimentation during the time course of the relaxation measurements. The suspension was prepared by slowly adding particles to a mildly heated solution (about 40°C to obtain a non-viscous, free-flowing liquid) , followed by sonication. For the low and high loading GdDTPA-SP, 47.2 mg and 39.4 mg were suspended in 30g of the PEG-BC medium, respectively. For the DyDTPA-SP, 152 mg were suspended in 5g PEG-BC. A sample containing GdDTPA (Magnevist^®, Schering AG, Berlin, Germany) was prepared by diluting the commercially available product with PEG-BC to a concentration of 1.5 mM Gd. A sample of DyDTPA-BMA (Sprodiamide, Nycomed Imaging AS, Oslo, Norway) was prepared by dissolving 62.5 mg in lOg of the PEG-BC medium. The metal ion concentration in the samples was verified by ICP-AES (see below) . A 0.8% (w/w) agarose medium sample (Agar, Kebo Lab, Oslo, Norway) was also prepared .

The l/Ti NMRD profiles of the samples were measured on a field-cycling relaxometer. For all samples, NMRD profiles were typically recorded for temperatures ranging from 5 to 35°C.

For ICP-AES analysis, the PEG-BC samples were initially dissolved in concentrated nitric acid and 30% - 16 -

(v/v) hydrogen peroxide. After addition of an internal scandium standard, the samples underwent a 2 hour heating cycle. The metal ion concentration was determined using a multipoint standard calibration curve. The metal ion concentrations were corrected for the solid content of the 5% PEG-BC medium.

EXAMPLE 4

NMRD Profiles

Figure 1 shows the l/T_x NMRD profiles of GdDTPA in 5% PEG-BC medium at 5 , 25 and 35°C (filled symbols) . Also shown are the NMRD profiles of GdDTPA in water, reproduced from results published by Kellar et al . (in Magn. Res. Med. 37: 730-735 (1997)), at these temperatures (open symbols) . The NMRD profiles of GdDTPA in 5% PEG-BC were fitted to relaxation theory by using the procedure used previously (see Kellar et al . , supra) , treating τ_R (the rotational correlation time) , τ_so (the electronic relaxation time at zero field) , the τ_v (the correlation time for electron spin relaxation arising from fluctuations in the spin-orbit interaction) as independent and unknown. The results of the fits are shown as solid curves. The values for the fitted parameters and values obtained previously for GdDTPA in water, are given in Table 1.

Table 1

Results of fitting the NMRD profiles of GdDTPA in 5% PEC-BC medium to relaxation theory. The values in parenthesis are the relaxation parameters obtained previously for GdDTPA in water . 17 -

T° (C) τ_R (ps) τ_v (ps) τ _so (PS )

5 183 32 99 ( 140 ) ( 38 ) ( 89 )

25 84 27 76 ( 80 ) ( 38 ) ( 85 )

35 65 25 77 ( 62 ) ( 35 ) ( 84 )

At 25 and 35°C, the parameters obtained in PEG-BC medium agree well with those previously obtained in water. That the parameters obtained for GdDTPA in PEG- BC agree well with those obtained in water at those two temperatures is consistent with their respective NMRD profiles being indistinguishable. The results also verify that the values of 1/T_X for the PEG-BC medium are independent of field strength.

Figure 2 shows the l/T_x NMRD profiles of DyDTPA-BMA (open symbols) and DyDTPA-SP (filled symbols) in PEG-BC medium at 5 and 35°C. The relaxivity of DyDTPA-BMA is independent of magnetic field strength and shows only a slight temperature dependence, as expected - see Bertini et al. J. Phys Chem 97: 6351-6354 (1993). The relaxivity of the DyDTPA-SP is significantly greater than that of DyDTPA-BMA, and displays a strong dependence on both magnetic field strength and temperature, all of which are unexpected findings.

Figure 3 shows the l/T_: NMRD profiles of empty DTPA- SP in water (8.5% w/w) at 5, 15, 25 and 35°C. The qualitative form of the NMRD profiles is similar to those obtained previously in cross-linked bovine serum • albumin solutions and in tissue - see Koenig et al . Magn. Res. Med. 29_: 77-83 (1993) . The qualitative form, as well as temperature dependence, is also similar to that of the DyDTPA-SP (see Fig. 2) .

Figure 4 shows the l/T_x NMRD profiles of high loading (6.7%, open symbols) and low loading (4.0%, - 18 - filled symbols) GdDTPA-SP at 5 , 25 and 35°C. Although qualitatively similar in form, the relaxivity of the low loading GdDTPA-SP is significantly higher than that of the high loading GdDTPA-SP at any particular temperature. Moreover, the relaxivities for both types of particles decrease at all field strengths with decreasing temperature.

Figure 5 shows the 1/Tχ NMRD profiles of 0.8% agarose medium at 5, 25 and 35°C. The l/Υ_x showed a strong field dependence below 1 MHz, increasing with increasing temperature at field strengths below 0.1 MHz, and decreasing with increasing temperature above 0.1 MHz.

That the NMRD profiles of GdDTPA in the PEG-BC medium are at the higher temperatures indistinguishable from those of the metal chelate in water shows that the PEG-BC is an ideal matrix for suspending particles for relaxometric measurements. The results of the best-fit analysis to the data of Fig. 1 also demonstrates this (see Table 1) ; in comparison to previously determined values of τ_R for GdDTPA in water, the τ_R determined in PEG-BC becomes significantly larger only at low temperatures. This is also evidenced by the significantly larger relaxation rates at 5°C and high field strengths (corresponding to proton Larmor frequencies of 10 MHz and above) obtained in the PEG-BC medium as compared to those in water. The increase in τ„ at 5°C is significant (31%) , and most likely reflects an increase in viscosity of the PEG-BC medium with decreasing temperature that is greater than that of water; the PEG-BC medium is solid at this low temperature, but is liquid at 20°C and above. At ambient and physiological temperatures, the PEG-BC medium, while maintaining the particles suspended indefinitely, can still be regarded as pure water from a relaxation point of view. Additionally, because there is no dependence of l/T_: on magnetic field strength, the - 19 - l/T₂ should also show no magnetic field dependence at conventional field strengths. Therefore, the PEG-BC is ideal as medium for l/T₂ measurements as well .

In a previous study, the relaxivities of the GdDTPA-SP measured in agarose media (at 20 MHz and 37°C) were significantly lower (about 30%) than those described here under similar conditions. The lower relaxivities in agarose media can be attributed the procedure of preparing the agarose medium suspensions of GdDTPA-SP, where the particles had to be added at very high, near boiling, temperatures. The GdDTPA-SP has been shown to degrade in aqueous solution, the degradation pathway being ester hydrolysis of GdDTPA from the starch particle - see Rongved et al . Carbohydrate Research 287: 77-89 (1996) . In fact, the 1/T_X of the suspension of the low-loading GdDTPA-SP in the PEG-BC medium decreased by over a factor of two after only one month of storage at ambient temperature, which would be consistent with the rapid degradation of these particles at elevated temperatures. The results demonstrate another advantage of the PEG-BC matrix as a suspending medium. Since high temperatures are not required to prepare particulate suspensions, particles of low stability are much less likely to degrade, allowing more accurate relaxivity determinations to be made .

At all field strengths, the higher and strongly temperature dependent Υ_λ relaxivities for DyDTPA-SP, in comparison to DyDTPA-BMA (see Fig. 2) , are unexpected results because the T_x relaxation efficacy of Dy- containing compounds is modulated by a very short ι^~ _s - see Bertini et al . supra. Any increase in τ_R, due to attachment of DyDTPA to the starch particle, would not have any measurable influence on the T_: relaxation properties. Therefore, the T_± relaxivities of DyDTPA-SP should be identical to those of DyDTPA-BMA at all temperatures and field strengths . The increase n the T_λ - 20 - relaxivities cannot be explained by a lengthening of τ_s either. Unlike Gd, the τ_s of Dy ions is independent of magnetic field strength and is not modulated by collisional processes that are characterised by τ_v - see Bertini et al . supra. One explanation for the elevated Tx relaxivity values, and their strong temperature dependence for the Dy DTPA-SP is related to the strong decrease in the T_x relaxivity with increasing field strength (the dispersion of the T_± relaxivity) below 10 MHz. Such a dependency is not possible for Dy agents due to the short τ_s that dominates the T_λ relaxivity; the NMRD profile should be flat in this region like it is for DyDTPA-BMA. Consequently, the dispersion of the _x relaxivity in this region must be due to the starch matrix of the particle. The NMRD profiles of starch particles containing no paramagnetic species (DTPA-SP) confirm this (see Fig. 3) . Furthermore, the NMRD profiles are similar in form to those obtained for tissue and for cross-linked bovine serum albumin solutions - see Koenig et al . Magn. Res. Med. 23.: 77-83 (1993) . The similarity between the current NMRD profiles and those of tissue and cross-linked bovine serum albumin solution indicates that there must be sites on the starch matrix that bind water molecules by at least three hydrogen bonds, resulting in residence lifetimes of water molecules on the surface of at least 10^"7 seconds. It is this residence lifetime that becomes the correlation time responsible for the shape of the NMRD profile of the empty DTPA-SP and DyDTPA-SP, below 10 MHz. This additional diamagnetic contribution, also causes the T₂ relaxivities of the DyDTPA-SP to be greater than those of DyDTPA-BMA. There is another possible explanation for the higher T_x relaxivities of DyDTPA-SP which is related to the probability that a water-binding site on the starch matrix is also in the vicinity of a Dy ion due to the three-dimensioned structure of the particle. Consequently, there would be a significant - 21 -

Dy-wate proton magnetic dipolar interaction, analogous to a second-sphere effect.

The identification of a diamagnetic contribution to the relaxivity of the DyDTPA-SP demonstrates yet another advantage of using the PEG-BC medium for suspending the particles. For example, this verification would have been difficult if agarose medium were used. The NMRD profile of a 0.8% agarose medium (see Fig. 5) shows a strong magnetic field dependence, particularly at low fields where the diamagnetic contribution is rather large, for essentially the same reasons as for the empty DTPA-SP. Furthermore, the relaxation rates for the 0.8% agarose medium at low fields (less than 0.1 MHz proton Larmor frequency) where the diamagnetic contribution due to hydrogen bonding of water molecules to the medium is most evident, increase with increasing temperature. This behaviour is opposite to that of the empty starch particles (see Fig. 3) , and would cause even further complications if agarose medium were used instead of the PEG-BC medium. Obtaining the diamagnetic contribution to the relaxivities of DyDTPA-SP would require subtraction of the NMRD profile of the agarose medium from that of the DyDTPA-SP suspended in the medium. The accuracy of the result of this subtraction depends on the reproducibility of making the agarose medium and whether the presence of the DyDTPA-SP (2% w/w) would have any influence on the three-dimensional structure of the agarose medium which could change its intrinsic relaxation rate. In the presence of such concerns, the use of a medium that has relaxation rates identical to those of pure water (low rates with no significant field dependence at temperatures of 25°C and above) leads to both a simplification of the procedure and an increase in the accuracy of the relaxivity determinations.

Due to the similarity in chemical structure to DyDTPA-SP, diamagnetic and possibly second sphere contributions must also contribute to the ^rT₁ relaxivities - 22 - of GdDTPA-SP. However, because the values of the T_: relaxivities are larger than those of the DyDTPA-SP, the magnitude of their contribution to the total relaxivity will be negligible. In comparison to GdDTPA, the value of t_R increases considerably upon binding to the starch matrix, as evidenced by the presence, and position, of a peak in the NMRD profile above 10MHz (see Fig. 4) - also see Kellar et al Magn. Res. Med. 3_7: 730-735 (1987) . The decreasing T_λ relaxivities of the GdDTPA-SP with decreasing temperature are typical for systems having a slow exchange of water molecules between the inner coordination sphere of the Gd³⁺ ion and the bulk. Such a slow exchange from the inner sphere is not expected for GdDTPA. Rather, the slow exchange is likely due to a long residence time of a water molecule within a starch particle (τ as a result of diffusional, rather than chemical exchange, limitations. Within a starch particle, the distance a water molecule can diffuse within a given time is less than would be the case outside the particle. A starch particle can be regarded as a gel, and the movement of water molecules is sterically restricted within the pores of the gel, slowing its diffusional progress. Additionally, starch- water interactions (through hydrogen bonding) also serve to hinder the motion of water molecules and lengthen the time a water molecule will reside within the starch particle .

Based on their similar particle size, the value of τ_x should be similar for both the high loading and low loading particles. The qualitative shape of the NMRD profiles, including the position of the peak (around 30MHz) , is similar for both GdDTPA loadings at a given temperature. This indicates that the T_λ relaxivity per Gd³⁺ ion of the GdDTPA within the starch particle {r_λl) is the same for both GdDTPA loadings . The relationship between the measured T_x relaxivity (x_λ) , τ _l the Gd³⁺ ion concentration ([Gd]₁) and relaxivity (r_Xl) within the - 23 - starch particle, is given by an expression analogous to liposome systems where paramagnetic species are located within the interior of the liposome (see Pύtz et al J. Liposome Res. 4.:771-808 (1994)).

r = ^ (1)

This equation shows that, if slow exchange conditions apply (l+r_u [Gd] if^l) , an increase in the Gd³⁺ ion concentration within the starch particle will result in a decrease in the measured relaxivity. Consequently, the relaxivity of the high loading particles would be lower than that of the low loading particles, at all magnetic field strengths, as observed (see Fig. 4) . The increase of the relaxivities with increasing temperature, at all field strengths, for the starch particles is also consistent with slow exchange conditions . An increase in temperature results in an increase in the self-diffusion constant of water, and possibly shortens the time a water molecule remains hydrogen bonded to the starch matrix, thereby decreasing the time it takes for a water molecule to diffuse a given distance. Consequently, τ_± decreases, and the T_x relaxivity increases, with increasing temperature. This behaviour is opposite to what was observed for the DyDTPA-SP, where the relaxivities decreased with increasing temperature (see Fig. 2) . As the relaxivity of Dy-containing compounds is much less than that of Gd- containing compounds (see Fossheim et al J. Magn. Res. Imaging 7.-.251-257 (1997)), slow exchange conditions do not apply for the DyDTPA-SP (r_n [Dy] _iτ_i<<l,Eq. (1) ) , and the T_: relaxivity increases with decreasing temperature mainly because the diamagnetic contribution increases with decreasing temperature (see Fig. 3) . - 24 - EXAMPLE 5

Preparation of Polwinylalcohol (PVA) ,

Polwinylpyyrolidone (PVP) Suspension Media

Suspension media containing 5% (w/w) of PVA (MW 70000-100000, Sigma, St. Louis Missouri), 5% and 15% of polyvinylpyrrolidone PVP (MW 30000, NMD, Oslo, Norway) were prepared by dissolving 5 or 15 g of polymer in 95 or 85 g of distilled water, respectively. The aqueous mixtures were heated, with stirring, until a clear solution resulted. The solutions were allowed to cool down to room temperature .

EXAMPLE 6

Preparation of 5% PVA. 5% and 15% PVP samples containing GdDTPA for NMRD Studies

Samples were prepared by diluting GdDTPA (Magnevist^®, Schering AG, Berlin, Germany) with the appropriate suspension medium to give a final concentration of 1 mM Gd. The Gd concentration was verified by ICP-AES and corrected for the mass fraction of the polymeric substances.

EXAMPLE 7

Preparation of Agarose Suspension Medium for NMRD Studies

A suspension medium containing 0.8% (w/w) of agarose (Kebo Lab, Oslo, Norway) was prepared (in triplicates) by dissolving 0.4 g of agarose in 49.6 g of distilled water. The aqueous mixture was heated to about 80°C, with stirring, until a clear solution resulted and 0.6 ml of each replicate was withdrawn into 3 separate - 25 - NMR tubes that were cooled down in a water-ice bath.

EXAMPLES 8

Preparation of Agarose Suspension Medium for Relaxometric Studies

A suspension medium containing 0.8% (w/w) of agarose (Kebo Lab, Oslo, Norway) was prepared by dissolving 80 mg of agarose in 9.92 g of distilled water. The aqueous mixture was heated to about 80°C, with stirring, until a clear solution resulted and 2 ml were withdrawn into 3 separate NMR tubes that were cooled down in a water-ice bath.

EXAMPLE 9

Preparation of 0.8% Aσarose Samples Containing GdDTPA for NMRD Studies

Duplicate samples were prepared by adding GdDTPA (Magnevist^®, Schering AG, Berlin, Germany) to a heated agarose solution, as described in Examples 5,6 to give a final concentration of about 1.5 mM Gd. The solution was shaken horizontally and 0.6 ml was withdrawn into NMR tubes that were cooled down in a water-ice bath. The Gd concentration was verified by ICP-AES.

EXAMPLE 10

NMRD Profiles

Figure 7 shows the l/T_x NMRD profiles of three aqueous samples of 0.8% agarose at 25°C, whose preparation is described in Examples 7,8. As observed, the reproducibility is very good above 1 MHz, but poor below 1 MHz. The likely reason for this is related to - 26 - the difficulty of preparing gels with identical three- dimensional structures . This lack of reproducibility will lead to errors in determining the relaxivity of any contrast agent, as shown in Fig. 8.

Figure 8 shows the l/Tj_. NMRD of GdDTPA in 0.8% agarose gel at 25°C (given as TOTAL) . This NMRD profile is the sum of two contributions, which are also shown: the BACKGROUND (0.8% agarose gel, Fig. 7.), and the DIFFERENCE, which represents the paramagnetic contribution arising from the metal chelate. It is important to note that only the NMRD profiles of the TOTAL and BACKGROUND can be directly measured; the DIFFERENCE profile can only be obtained by subtraction of the BACKGROUND from the TOTAL profiles. The NMRD profile of the DIFFERENCE is important, since this represents the paramagnetic contribution that gives the relaxivities, once divided by the millimolar Gd concentration. Accurate relaxivities are dependent on having an accurate NMRD profile of the BACKGROUND. Any variations in the low-field (less than 1 MHz) rates between the sample used to represent the BACKGROUND and the actual background contribution from the sample containing the contrast agent can cause errors in relaxivities. This is apparent in Fig. 8; a "rise" in the NMRD profile of the DIFFERENCE exists, a "rise" which is not due to the contribution of the contrast agent (see Fig. 1) and only exists due to the lack of reproducibility of the BACKGROUND contribution. This will lead to inaccurate relaxivities at low fields, as shown in Fig. 9., which could result in an erroneous interpretation of the data. Also, the "aesthetic" quality of the paramagnetic contribution is poor.

Figure 9 shows the field dependence of the T_: relaxivity for GdDTPA in 0.8% agarose gel at 25 °C. This was accomplished by dividing the DIFFERENCE contribution (see Fig. 8) by the millimolar Gd concentration (1.60 mM) . Also shown, for comparative purposes, are the - 27 - relaxivities in water and 5% PEG-BC. Above 10 MHz, the relaxivities in all three media are essentially the same, confirming that the microviscosity is not affected by the agarose. The "aesthetic" or "cosmetic" value of the relaxivities in 0.8% agarose is poor at fields below 10 MHz. However, it is only a "cosmetic" problem with GdDTPA, where the relaxivities in water are known. If the compound were not GdDTPA but some other agent where the relaxivities could not be measured in water (for example, for particles that settle), this would be a critical, and not simply "cosmetic", drawback.

Figure 10 shows the field dependence of the T_x relaxivities for GdDTPA in 5% PVA at 25 and 35°C. The relaxivities were adjusted for the 5% content of PVA. Also shown are the corresponding relaxivities in water, for comparative purposes (open symbols) . As can be seen, there is no "cosmetic" problem visible at low fields (below about 1 MHz) caused by the 5% PVA medium (the relaxivities are flat) , but the relaxivity values are considerably greater than in water. This is true also above 1 MHz, showing that microviscosity of the 5% PVA medium is increased compared to that of water.

Figure 11 shows the field dependence of the T_: relaxivities for GdDTPA in 5% PVP at 25 and 35°C. The relaxivities were adjusted for the 5% content of PVP. Also shown are the corresponding relaxivities in water, for comparison purposes (open symbols) . As can be seen, there is no "cosmetic" problem visible at low fields (below about 1 MHz) caused by the 5% PVA medium (the relaxivities are flat) , but the relaxivities are considerably greater than in water. This is true also above 1 MHz, showing that the 5% PVP medium has an increased microviscosity compared to that of water, though the increase is not as substantial as is the case for the 5% PVA.

Figure 12 shows the field dependence of the T_: relaxivities for GdDTPA in 15% PVP at 25 and 35°C. The - 28 - relaxivities were adjusted for the 15% content of PVP. Also shown are the corresponding relaxivities in water, for comparative purposes (open symbols) . There is no "cosmetic" problem visible at low fields (below about 1 MHz) caused by the 15% PVA medium (the relaxivities are flat) , but the relaxivities are considerably higher than in water. This is true also above 1 MHz, showing that the microviscosity of the 15% PVP medium is increased compared to that of water.

EXAMPLE 11

Practical Considerations of Microviscosity and Field- Strength Independence of the Suspension Medium

Concerning the three criteria for a suitable suspension medium for relaxometric analysis, it is apparently difficult to determine whether the medium has an increased microviscosity relative to that of water and if the medium has a T_x (and T₂) that is independent of magnetic field strength. This determination ultimately requires an analysis by NMRD, and the problem is that only a few of these instruments are available. Therefore, it would be useful to do a preliminary evaluation of whether these two criteria are fulfilled before sending a sample for NMRD analysis. A simple way of performing this evaluation is to measure the T_x and T₂ of a matrix at a single field strength and temperature; instruments operating at 20 MHz are very common and this is a particularly good field strength for this evaluation. Table 2 shows the T_x and T₂ values for various suspension media at 25°C and 20 MHz. 29

Table 2

Sample Ti (ms) T₂ (ms)

5% PVA 2160 1600

5% PVP 2462 2100

15% PVP 1927 1590

5% PEG-BC 2500 2000

0.8% Agarose 2561 270

Water 2785 3000^a

"It is theoretically impossible for T₂ to be greater than T_L in any aqueous solution, so this result simply shows that it is difficult to measure long T₂ values experimentally.

For the 5% PVA, and 15% PVP media, the T_± and T₂ values are significantly shorter than those of pure water. These results indicate that the microviscosity of these media is higher than that in water, with the highest microviscosity for the 15% PVP medium. Also, because the T_λ and T₂ values are not much different from each other (in comparison to the case for the 0.8% agarose gel) , the T_x and T₂ values would be expected to be field-independent (0.01 MHz up to field strengths accessible on NMRD and MRI spectrometers) . This is because T_x and T₂ have to be equal at low fields because the low-field value of T_x reflects the high-field T₂ as discussed previously. Furthermore, the T_λ relaxivities of any agent in these polymeric media would be expected to be higher than those in water, as exemplified for GdDTPA in 15% PVP with the highest relaxivity at all fields (Figs 10-12) .

For the 5% PVP, the T_x and T₂ values are longer than those of the 5% PVA, indicating a lower microviscosity and, hence, a lower T_λ relaxivity of GdDTPA compared to those in 5% PVA and 15% PVP (Figs 10-12) . In fact, the T_L relaxivity of GdDTPA in 5% PVP at 25°C was very - 30 - similar to that in water at field strengths above 2 MHz. This may be expected from the results in Table 2, where the T_x and T₂ values of 5% PVP are similar to those of 5% PEG-BC. However, over the entire field-strength range, the data of GdDTPA in 5% PVP do not compare as well with the GdDTPA in water data as well as GdDTPA in 5% PEG-BC does. There is no obvious explanation for this result; it just shows that the 5% PVP is not as good as a suspension medium as 5% PEG-BC. However, from the results of Table 2, one would expect 5% PVP to be as good of a suspension medium (based on the three criteria) as 5% PEG-BC. This illustrates a problem in selecting a suspension matrix without collecting NMRD data, or, the need for collecting NMRD data before any suspension medium is ultimately selected. Based on the results in Table 2, it would be fairly simple to eliminate 5% PVA and 15% PVP as potential suspension media, as their T and T₂ values are similar to each other but significantly lower than those of water, and even 5% PVP and 5% PEG-BC. However, there would be no reason not to choose 5% PVP over 5% PEG-BC based on the results in Table 2. Regardless, the PVA and the PVP- based media will not be suitable as suspension media, as comparison of relaxivity (and therefore the correlation times that determine the relaxivity) to that/those of other agents suspended in different media (typically water) will be difficult.

In the case of agarose 0.8%, its similar T_x to that of water and its substantially shorter T₂ are reflected in the field-dependency of both T₁ (Figs 5,7) and T₂ below 1 MHz . This illustrates that the agarose medium does not influence the microviscosity of water and that the water protons are relaxed by a diamagnetic mechanism that does not involve microviscosity (see Koenig, Brown III, Ugolini, Magn. Reson. Med.. 1993; 29: 77-83). Also, the low-field T_x values and the reproducibility of the field-dependency at field strengths below 1 MHz might - 31 - differ among various agarose preparations, which may prevent accurate T_x relaxivity determination at low fields (Fig. 8) as well as T₂ relaxivity assessment at MRI fields.

Claims

- 32 -Claims

1. A method of determining magnetic properties of a sedimentable agent, said method comprising disposing said sedimentable agent in a flowable aqueous solution of a polyalkyleneoxide-based matrix forming agent, optionally cooling the resulting dispersion to form a non-flowable aqueous dispersion of said sedimentable agent, measuring a magnetic property of said dispersion, and if desired calculating from the measured magnetic property a magnetic property of said agent or said dispersion.

2. A method as claimed in claim 1 wherein said sedimentable agent is a water-insoluble particulate material .

3. A method as claimed in claim 1 or claim 2 wherein said measured or calculated magnetic property is a water proton relaxation rate T_x or T₂, relaxivity of the sedimentable agent or of the dispersion, NMRD profiles or magnetization.

4. A method as claimed in any one of claims 1 to 3 wherein said sedimentable agent contains paramagnetic centres .

5. A method as claimed in any one of claims 1 to 3 wherein said sedimentable agent comprises an iron oxide.

6. A method as claimed in any one of claims 2 to 5 wherein said particulate material has a mean particle size in the range 0.005 to 100 ╬╝m.

7. A method a claimed in any. one of claims 1 to 6 wherein said polalkyleneoxide-based matrix forming agent is an amphiphilic block copolymer. 33

8. A method as claimed in claim 7 wherein said amphiphilic block copolymer comprises at least one polyalkyleneoxide block in which the alkylene moieties are C₂-C₄ and at least one relatively hydrophobic block in which the alkylene moieties are C₈-C₃₀.

9. A method as claimed in any one of claims 1 to 8 wherein said matrix forming agent is a compound of formula

OCH₂(CH₂OCH₂)_nCH₂OCH₃

10. An aqueous dispersion comprising a sedimentable agent and a polyalkyleneoxide-based matrix forming agent .

11. An aqueous dispersion as claimed in claim 10 wherein said sedimentable agent is a water-insoluble particulate material.

12. The use of polyalkyleneoxide-based matrix forming agent for the preparation of a dispersion for use in the determination of magnetic properties of sedimentable agents particulates.