Although important clinical advances in the prevention and treatment of atherosclerosis have been made over the past 20 years, atherosclerotic disease remains the primary cause of mortality in industrialized countries (1). Studies have shown that more than two-thirds of acute coronary syndromes occur in patients who are classified as being at intermediate risk according to traditional Framingham Score risk-stratification methods. Based on these findings, it is clear that noninvasive diagnostic strategies are required to accurately assess the extent of cardiovascular disease in order to predict which patients need the most aggressive management. Recently a clear link between plaque vulnerability and plaque composition was established (2–6). Most vulnerable plaques contain a high content of functionalized macrophages, oxidized low-density lipoproteins, large lipid cores, thin fibrous caps, areas of necrosis or thrombus, and neovascularisation (5, 7–12). High-resolution magnetic resonance imaging (MRI), which allows for submillimeter resolution of the arterial wall, has emerged as one of the most promising techniques for the direct evaluation of plaque composition (13–27). Despite the promise of MRI for plaque detection and characterization, issues related to sensitivity, partial voluming effects, and motion have limited its clinical utility.
Recent studies using commercially available extracellular (ECF) contrast agents, such as GdDTPA, have shown that ECF agents increase the sensitivity and efficacy of MRI for detecting and characterizing plaque composition (19, 28, 29). Imaging is generally performed 6–20 min after administration of the ECF agent. Since the excretion kinetics of the agent within the plaque depends on the composition of the lesion, it is possible to accurately identify different plaque structures, such as the fibrous cap and lipid-rich core, based upon delayed MR enhancement (19, 29). Although ECF agents may allow for increased detection and characterization of atherosclerotic plaque, they are still not optimal for a total evaluation of the vasculature. Intravascular contrast agents allow for evaluation of vessel morphology and function, and may also be useful for characterizing atherosclerotic plaque. By definition, intravascular agents should not leak or permeate out of the normal vessel wall (30–32). Consequently, these materials usually have long circulation times (relative to GdDTPA) that are defined by the rate of clearance from blood (i.e., the blood halflife). These long circulation times allow for a complete evaluation of vessel morphology and have even been used in imaging of the coronary arteries (33–36). As a result of long circulation times, intravascular agents normally exhibit a steady-state phase in which the longitudinal relaxation times in blood remain constant over defined time intervals. Since the concentration of the contrast agent is well defined within the steady-state phase, it is possible to accurately assess functional parameters (e.g., permeability, blood volume, and blood flow) that may be related to neovascularization or vessel density within tumors and potentially other lesions, such as atherosclerotic plaque (30, 37–41). Studies using iron-oxide particles and lipophilic micellular contrast agents have shown that long-circulating nanoparticles may penetrate the abnormal vessel wall of hyperlipidemic rabbits or ApoE−/− mice, and thus allow improved detection and characterization of atherosclerotic plaque (26, 42–49).
Lipids are naturally occurring amphiphilic molecules that contain a hydrophilic head group and a hydrophobic tail. Because of the dual character of the lipids, and the energetically unfavorable contact between the lipid tails and water, amphiphiles self-associate into aggregates of different sizes and geometries. The length of the hydrophobic chains and the size of the head group (in relation to the chain) determine the curvature of the aggregate, and whether a micelle-like structure (single layer) or a bilayer (liposome) structure will be formed (44, 50). Gadolinium (Gd) mixed-micelles are formed by mixing amphiphiles with a Gd chelate linked to a similar lipid, and a non-ionic surfactant used to stabilize the resultant nanoparticle (51–56). Previous studies have indicated that Gd mixed-micelles may be effective intravascular contrast agents for MRI (52, 57). Following injection studies have indicated that 70% of a 0.075-mmol Gd/kg dose of C18GdDOTA mixed-micelle was still present in the blood of rats 30 minutes after i.v. injection (52). Approximately 20% of the injected dose was found in the liver 7 days p.i.. The biodistribution data suggest that liver uptake or retention of the nanoparticles may be problematic from a safety standpoint (52). Studies have shown that Gd chelates may not be stable within intracellular environments (58). As a result, uptake of the Gd mixed-micelles within liver cells may induce the release of Gd ions. Liver uptake should therefore be minimized to allow for the development of clinically relevant Gd mixed-micelles. Preliminary studies indicate that the pharmacokinetics and biodistribution of mixed-micelles are largely dependent on the amphiphile used to generate the micelle (52, 57). By manipulating the amphiphile, it should be possible to prepare Gd mixed-micelles with acceptable blood clearance and liver retention. As a result, Gd mixed-micelles may provide a platform for the complete diagnostic assessment of vessel morphology and wall composition, and may eventually enable accurate evaluation of plaque neovasculature.
The primary aim of the current study was to prepare, characterize, and evaluate in vivo the efficacy of two Gd mixed-micelle formulations using apolipoprotein E knockout (ApoE−/−) mouse models of atherosclerosis. Gd mixed-micelles were prepared using two different amphiphiles but similar GdDTPA lipids, surfactants, and fluorescent labels. Emphasis was placed on the development of a mixed-micelle with long circulation times and low liver retention. As a result, the micelles were characterized with respect to size, structure, flexibility, relaxivity, blood clearance, and liver retention. MRI of the abdominal aorta was performed at 9.4T before, immediately after, and 24 hours after the administration of the two micelle formations. GdDTPA also was administered to both ApoE−/− mice and WT mice as a positive control. Histology and confocal microscopy were performed 24 hr p.i. to locate the micelles within the vessel walls of ApoE−/− and WT mice.
MATERIALS AND METHODS
Preparation of Mixed-Micelles
Gd mixed-micelles were prepared according to established methods (54, 55). In short, Gd-DTPA-bis(stearylamide) (7 w/w%; Gateway Chemical Technology, St. Louis, MO, USA) was mixed with 1,2-Dipalmitoyl-sn-glycero-3-Phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) ammonium salt (DPPE-NBD, 2 w/w%; Avanti Polar Lipids, Alabaster, AK, USA). The phospholipid 1,2-dipalmitoyl-sn-glycero-phosphatidylcholine (DPPC; 79 w/w%, Avanti Polar Lipids) was used to prepare small micelles, and 1-palmitoyl-2-oleoyl-sn-glycero-phosphatidylcholine (POPC; 79 w/w%, Avanti Polar Lipids) was used to prepare large micelles. Chloroform/methanol (3:1) was added to dissolve all salts. The solvent was evaporated using a rotavap at 65°C, and the resulting film was dried under nitrogen. The film was rehydrated in 1:1 sterile water/PBS and sonicated for 15 min at 70 W at a 90% cycle duty. After sonication, the surfactant (Tween 80®, 14 w/w%) was added and the solution was resonicated for an additional 15 min. Following sonication the micelles were centrifuged and the supernatant was concentrated to 3.5 mM Gd with the use of ultracentrifugation techniques.
Hydrated Particle Size and Micelle Structure
The hydrated mean diameter of the resultant micelles was determined using a Malvern light-scattering spectrophotometer (Malvern Instruments, Malvern, UK). All samples were analyzed at 25°C in filtered (cutoff = 30 nm) water solutions. The weighted z-average (based on intensity, volume, and number averages) and subsequent polydispersity values were reported.
In addition to light-scattering, the diameter and structure of the mixed-micelle formulations were evaluated using transmission electron microscopy (TEM, model CX-100; JEOL, Tokyo, Japan). Prior to TEM analysis, 1% osmium tetraoxide was added to 1 ml of both the small- and large-micelle formulations. To avoid aggregation during sedimentation, the samples were centrifuged within 1 hr after incubation in osmium. The resulting pellet was isolated, washed in PBS, and fixed according to established methods (59).
The total Gd concentration of the stock solutions of the micelle formulations was obtained by inductively coupled plasma mass spectrometry (ICP-MS) measurements (Elemental Research Inc.). From these results, the number of Gd ions per micelle was determined (60).
Longitudinal Relaxivity in Aqueous Solution
Mixed-micelles were prepared in water at six different concentration levels ranging from 0 to 0.5 mM Gd (where 0 mM Gd reflects the matrix without the addition of micelles). Longitudinal relaxation times (T1) were determined at 60 MHz using a Bruker Minispec (Bruker Medical BmbH, Ettingen, Germany) operating at and 40°C. Inversion-recovery sequences (15 inversion times) were used to obtain all raw data, and T1 values were calculated from a monoexponential fit of SI vs. time. The longitudinal relaxivity (r1) was determined from a linear fit of Gd concentration (mM Gd) vs. 1/T1 (1/T1 = R1, s−1), as described by
where R1 is the relaxation rate of the micelle sample containing Gd (1/s), x is the concentration of Gd present (in mM Gd), r1 is the slope of the linear correlation (equal to the relaxivity in s−1mM−1), and R1m is the relaxation rate of the matrix (not containing the micelles; 1/s). All analyses were performed in Microsoft Excel (Microsoft Corp., Seattle, WA, USA).
The ability of the mixed-micelles to permeate through a 25-nm pore was evaluated by subjecting the micelles to vacuum filtration using filters (Millipore, Billerica, MA, USA) with a 25-nm cutoff size. Undiluted micelles (1 ml) were placed under vacuum filtration. The ease of filtration was noted and the hydrated mean diameter of the micelles was determined both pre- and postfiltration.
In Vivo MRI in ApoE−/− Mice
Thirteen-month-old ApoE−/− mice (N = 35) on a high-cholesterol diet, and wild-type (WT) mice (N = 47) on a normal diet underwent in vivo MR microscopy (MRM) of the abdominal aorta using a 9.4T MR system (Bruker) equipped with both a 30-mm transmit/receive birdcage coil and a thermocouple/heater system to keep the body temperature constant at 37°C. The local ethics committee approved all of the animal experiments. The ApoE−/− and WT mice were divided into two groups: the small-micelle group (N = 13 ApoE−/−, N = 13 WT) and the large-micelle group (N = 10 ApoE−/−, N = 10 WT). The animals were administered either 0.015 mmol Gd/kg or 0.038 mmol Gd/Kg of the small- and large-micelle formulations via intravenous injection into the tail vein. No adverse clinical signs indicative of toxicity were observed during or after micelle administration. ApoE−/− (N = 6) and WT mice (N = 6) were administered a dose of 0.1 mmol Gd/kg as a positive control, and imaging was performed at multiple time points up to 24 hr p.i.. All of the mice were anesthetized continuously with inhaled isofluorane (1.5–2%) and imaging was performed using a T1-weighted black blood (spin-echo (SE)) sequence (TR/TE/flip = 600 ms/8.6 ms/30°, NEX = 14, FOV = 2.6 cm × 2.6 cm), and 16 contiguous 500-μm-thick axial slices were acquired (pixel size = 101 × 101 × 500 μm3). The total scan time was approximately 37 min. At each time point the slices were matched to the baseline precontrast scan by using unique vertebral anatomy and paraspinous muscular anatomy as anatomic landmarks. Since the abdominal aorta was retroperitoneal, it was stationary in the axial and longitudinal planes relative to the vertebrae, spinal cord, and paraspinous muscles. As a result, no respiratory or cardiac gating was necessary.
Signal intensity (SI) measurements were taken using regions of interest (ROIs) on the aortic wall with four points in four quadrants of the aorta on slices exhibiting signal changes postcontrast. SI measurements of the aortic lumen, muscle, and noise were also obtained on each slice. The contrast-to-noise ratio (CNR) of the aortic wall (CNRWL) relative to the lumen was calculated for each slice using the following equation:
where W1 is the SI of the aortic wall in quadrant 1, W2 is the SI of the aortic wall in quadrant 2, W3 is the SI of the aortic wall in quadrant 3, W4 is the SI of the aortic wall in quadrant 4, L is the SI of the lumen, and SDN is the standard deviation associated with the noise.
The percent increase in the CNR value was determined according to:
The error associated with the %CNR value was determined via error propagation methods (61).
For each matched slice we determined the percent enhancement (%ENH) in order to evaluate the enhancement of the aortic wall relative to muscle 24 hr p.i. according to:
where SNRwallpost is the signal-to-noise ratio (SNR) of the aortic wall 24 hr p.i., SNRmusclepost is the SNR of the muscle 24 hr p.i., SNRwallpre is the SNR of the aortic wall precontrast, and SNRmusclepre is the SNR of the muscle precontrast.
Histology, Immunohistochemistry, and Confocal Fluorescence Microscopy
Immediately after imaging at the 24-hr time point p.i. was completed, the animals were sacrificed and the aorta from the thoracic descending aorta to the iliac bifurcation was isolated and frozen in optical cutting temperature compound (Tissue Tek, Sakura Finetech). Sections (8 microns thick) were cut and mounted on slides and stained with RPE-labeled anti-CD68 (excitation wavelength = 532–561 nm, emission wavelength = 578–610 nm; macrophage staining), DAPI (excitation wavelength = 385–400 nm, emission wavelength = >450 nm; cell nuclei staining), and imaged with confocal laser-microscopy. The NBD-labeled micelles had an excitation wavelength of 480 nm and emission wavelength of 534 nm.
Pharmacokinetics and Biodistribution
The blood half-life and liver uptake of each mixed-micelle formulation were determined in 1-year-old WT mice (N = 18) and ApoE−/− mice (N = 6, liver uptake only). The mice were administered 0.038 mM Gd/Kg of the micelle formulations via tail-vein injection. Stock solutions were diluted with PBS so that equivalent volumes (0.400 ml) of the various micelle formulations were injected. Blood was drawn (0.250 ml) into heparinized blood collection tubes via heart puncture at seven different time points p.i. (ranging from 5 min to 24 hr p.i.). Because of the large volume of blood required, the animals were killed after heart puncture at each time point p.i. At the 24-hr time point, the liver was excised, cleaned of excess fat and blood, and homogenized to allow for determination of Gd content. The local ethics committee approved all of the animal experiments.
The T1 values of all blood samples and liver homogenate were determined at 60 MHz (Bruker Minipsec) within 1 hr after sample collection. The blood halflife values were determined based on the resultant R1 vs. time (p.i.) curves. From the curves of R1 vs. time, kinetic parameters were obtained using standard noncompartmental biexponential pharmacokinetic analysis according to:
where t1/2a and t1/2b are the halftime clearance of the two components, and fa and fb represent the fractional volumes of the two compartments. All analyses were performed using Matlab (Mathworks version 7.0.1).
Prior to the determination of mouse liver Gd concentration, the relaxivities of each of the mixed-micelle formulations were obtained in control ex vivo liver homogenate. For each of the formulations investigated, dose-response curves were generated by adding known concentrations of micelle to 0.49 g of control mouse liver homogenate. The concentration range studied was 0–0.3 mM Gd, with at least five concentration levels in each dose-response curve. The longitudinal relaxivities were determined according to Eq. , and the quantification limit (QL, defined as the lowest Gd concentration significantly (P = 0.05) greater than the background) of the method was determined according to:
where SEy is the standard error associated with the y-intercept (1/s), and r is the slope or relaxivity (s−1mM−1).
Immediately following the 24-hr scan, the livers were excised, weighed, and homogenized. The time between the death of the animals and the relaxation measurements of the ex vivo liver homogenate did not exceed 45 min. All homogenate samples were analyzed within 5 min after tissue excision and exposure to 40°C. The average liver R1 values, along with the SDs, were determined for each of the formulations studied. The corresponding Gd concentration found in the liver samples was calculated according to:
where x is the concentration of iron-oxide particles present, r is the slope or r1 relaxivity value obtained from the dose-response study, and b is the R1 obtained for control liver.
The percent injected dose (%ID) was determined based on 1) the total amount of Gd administered and the total Gd content obtained in the liver, and 2) the dose administered per gram body weight and the dose obtained in liver per gram of liver. The %ID values obtained for the WT mice were compared with the results obtained in ApoE−/− mice.
To identify potential excretion pathways and uptake into other organs, the WT mice used in the pKa study were imaged at 9.4T both before and 24 hr p.i. according to equivalent methods used for the ApoE−/− mice. Images from the aortic bifurcation to the liver were obtained, and significant changes in signal were identified and reported.
Two-tailed unpaired t-tests were performed to evaluate differences among the longitudinal relaxivity values obtained for the various formulations in water, blood, and liver homogenate matrices. In addition, t-tests were used to evaluate the change in the mean hydrated particle size pre- and postfiltration through a 25-nm filter. P-values less than 0.05 were considered statistically significant.
A one-way analysis of variance (ANOVA) with Bonferroni post hoc multiple-comparison tests were used to compare the %ENH and CNR values obtained for the small-micelle formulation in ApoE−/− mice, the large micelles in ApoE−/− mice, large micelles in WT mice, and small micelles in WT mice. Additionally, an ANOVA test was performed to evaluate the differences between the percent injected dose found in the liver for the small micelles, large micelles, control WT mice, and ApoE−/− mice.
Table 1 summarizes the physical and chemical properties of the small and large mixed-micelle formulations relative to GdDTPA. The polydispersity indexes associated with the mean hydrated particle size revealed relatively large size distributions associated with both micelle formulations. However, both batches exhibited only one significant peak (>99%) on intensity-, volume-, and number-weighted chromatograms. From the ICP-MS results, it was evident that the large micelles were capable of loading a significantly greater number of Gd ions per micelle (15 times the loading capacity). The r1 values obtained in a water matrix at 60 MHz corresponded well with previously published results for Gd mixed-micelle formulations evaluated under equivalent testing conditions (54, 55). Although the r1 values obtained in liver homogenate were slightly lower than the values obtained in water, no significant differences were observed between the two matrices. The QL associated with the determination of Gd concentration in liver homogenate (by T1 methods) was approximately 0.07 mM Gd for both mixed-micelle formulations. Assuming that the total blood volume in the mouse is 2 ml, the QL of the method was approximately 2% of the total dose administered for both formulations. As a result, liver concentrations values must be greater than 2% of the injected dose (%ID) to allow for determination of Gd content in the liver by T1 methods.
Table 1. Summary of the Effect of Amphiphile on the Physical and Chemical Properties of the Resultant Gadolinium Mixed-Micelles Formed, Relative to GdDTPA
Mean hydrated size (nm) (z-average ± disp.)
Number of Gd/micelle
r1 60 MHz water (s−1mM−1)
r1 60 MHz liver s−1mM−1
QL liver (Gd) mM
disp. reflects the polydispersity associated with the mean, SE = the standard error associated with the longitudinal relaxivity (r1) obtained in aqueous solution at 60 MHz and 40°C.
26.7 ± 0.17
7.1 ± 0.2
7.0 ± 0.2
105.7 ± 0.15
7.8 ± 0.2
7.4 ± 0.2
3.9 ± 0.2
The TEM results (shown in Fig. 1) indicate that both mixed-micelle formulations were composed of single phospholipid layers (Fig. 1b and c). As a result, both formulations were clearly micelles (not bilayers or liposomes), with the hydrophobic tails of the phospholipids directed inward (away from the solvent) and the polar heads (containing the Gd chelate) directed outward, as shown in Fig. 1a. The TEM results also confirmed that the mean hydrated diameter of the large micelles was significantly greater than that of the small-micelle formulation. Additionally, the TEM results clearly show that there was a great deal of polydispersity associated with the size of both formulations, as previously indicated by light-scattering techniques (Table 1).
The effect of filtering (25-nm pore size) on the mean hydrated diameter of the various micelle formulations is shown in Table 2. No significant differences in the mean hydrated diameters of the both micelle formulations were observed pre- or postfiltration. In this study the mean hydrated sizes based on the intensity weighting (dI) were shown. Intensity values were chosen because the error reflects the true SD associated with the measurement, and not the polydispersity index that is obtained when z-average results are used. The results of this study strongly suggest that the large micelles are able to permeate through a 25-nm pore without significant changes in molecular shape or size.
Table 2. The Effect of Filtering (25-nm pore size) on the Mean Hydrated Diameter of the Various Micelle Formulations*
Mean hydrated size <d>I (nm)
Mean hydrated size <d>I (nm)
The SD value reflects the SD associated with the mean hydrated size that was determined based upon the intensity weighting. In all cases the percent of the intensity associated with the measured peak was greater than 90%. p = NS.
23.4 ± 10.3
22.2 ± 5.3
140.7 ± 89.2
143.8 ± 60.5
The efficacy and uptake of the Gd mixed-micelle formulations in the vessel wall of ApoE−/− and WT mice, relative to GdDTPA, are summarized in Table 3. Relative to muscle (as described by the %ENH), the administration of small or large micelles resulted in a significant enhancement of the vessel wall of ApoE−/− mice. For both doses tested, no significant enhancement of the vessel wall was observed in WT mice following administration of small or large micelles. Administration of the high dose (0.038 mmol Gd/kg) resulted in significantly greater enhancement of the vessel wall of ApoE−/− mice relative to the lower dose. For a given dose, no significant variation on the vessel wall enhancement (of ApoE−/− mice) was observed between the small and large micelles tested. The enhancement of the vessel wall, relative to lumen, was expressed by the %CNR values. The %CRR values obtained were similar to the %ENH values. For a given dose, no significant difference in the %CNR values was observed between the two micelle formulations. Additionally, the micelles did not significantly increase the %CNR values of WT mice at the doses tested.
Table 3. Summary of the Effect of Amphiphile on the In Vivo Efficacy in ApoE−/− Mice and WT Mice 24 Hours after the Administration of 0.015 mmol Gd/kg and 0.038 mmol Gd/kg Doses, Relative to GdDTPA (0.1 mmol Gd/kg at 1 and 24 Hours Postinjection)*
The normalized signal enhancement (%ENH) reflects the signal increase of the vessel wall relative to muscle. The contrast-to-noise ratios (CNR) describe the contrast of the vessel wall relative to the lumen both prior to and 24-hours after the administration of the various contrast agents. All imaging was performed at 9.4T using T1-weighted black blood (SE) sequences with TR/TE/flip = 600 ms/8.6 ms/30°, NEX = 14, FOV = 2.6 cm × 2.6cm. Bold values are statistically significant (p < 0.05).
p.i. = postinjection.
42 ± 15
39 ± 11
7 ± 10
9 ± 11
64 ± 9
62 ± 5
8 ± 11
10 ± 11
20 ± 12
−2 ± 7
9 ± 2
8 ± 2
6 ± 2
7 ± 2
8 ± 3
7 ± 2
8 ± 2
11 ± 2
8 ± 2
8 ± 2
14 ± 2
13 ± 2
7 ± 6
8 ± 7
25 ± 2
23 ± 3
9 ± 11
8 ± 7
10 ± 3
8 ± 2
36 ± 10
38 ± 15
14 ± 15
13 ± 12
68 ± 20
70 ± 22
11 ± 14
−3 ± 9
18 ± 5
0 ± 1
Following administration of GdDTPA, transient signal enhancement was observed in the vessel wall of ApoE−/− mice 1 hr p.i. (Table 3). No significant enhancement of the vessel wall of WT mice was observed at this time point. These results are consistent with previous studies that showed that GdDTPA may be used to delineate and characterize atherosclerotic plaque at delayed time points p.i. (19, 28, 29). However, no significant enhancement of the vessel wall of ApoE−/− or WT mice was observed 24 hr p.i.
The increase in the SI of the vessel wall relative to the lumen was expressed as the CNR. No significant differences in aortic CNR values were observed between the ApoE−/− mice and WT mice in the precontrast scans. However, 24 hr p.i., significant increases in the CNR values of plaque bearing ApoE−/− mice were observed for both dose levels relative to the vessel wall of control WT mice. The CNR values obtained for the small- and large-micelle formulations in ApoE−/− mice were statistically equivalent when compared with each other.
Figure 2a and b show representative images obtained in ApoE−/− mice using the small (a) and large (b) micelle formulations (both precontrast and 24-hr p.i. images are shown) following administration of the low dose. Figure 2 illustrates the heterogeneous signal enhancement observed in the vessel wall of the ApoE−/− mice. In Fig. 2a the plaque is relatively advanced, with an identifiable lipid-rich core. The lipid-rich plaque generally had less enhancement than the smaller, less advanced plaques, as shown in Fig. 2b. However, despite differences in enhancement due to plaque composition, both formulations caused a significant increase in the signal enhancement of atherosclerotic plaque in ApoE−/− mice (Table 3).
Figure 3 shows a laser-scanning confocal microscopy image demonstrating atherosclerotic plaque localization of the NBD-labeled micelles (green) 24 hr p.i. of a 0.015-mmol Gd/Kg dose, relative to cell nuclei (blue). It is evident from this figure that NBD-labeled micelles were predominantly found within the extracellular matrix of atherosclerotic plaque on the luminal side. NBD-labeled micelles were also present in the draining lymphatic vessels/microvessels surrounding the aorta. Colocalization of the micelles was not observed within cell nuclei, indicating that untargeted micelles are not internalized by cellular components of atherosclerotic plaque. No significant enhancement (caused by the presence of micelles) was observed in the vessel wall of any of the control WT mice studied.
The blood halflives of the small and large Gd mixed-micelle formulations in WT mice are shown in Table 4. Both formulations exhibited biphasic clearance with a fast component (fb) that exhibited half-lives of 0.6 and 0.09 hr for the large and small micelles, respectively. The slow-decaying component (fa), however, exhibited halflives of 27 and 10 hr for the large and small micelles, respectively. For both formulations a 1-hr steady-state phase was observed in which the T1 values of blood remained between 250 and 300 ms following administration of a 0.038-mmol Gd/Kg dose at 60 MHz.
Table 4. The Effect of Amphiphile on the Blood Clearance, as Reflected by the Blood Half-Lives (t1/2), in WT Mice Following Administration of a 0.038 mmol Gd/Kg Dose
Blood-half life in WT mice (hours)
t 1/2 a
t 1/2 b
Sum of squares
The sum of squares reflects the error associated with the two-compartment fit (t1/2a and t1/2b).
Small (N = 5)
Large (N = 5)
5.6 × 10−13
Table 5 summarizes the liver uptake of the Gd mixed-micelles 24 hr p.i. (dose = 0.038 mmol Gd/Kg) in control WT mice and ApoE−/− mice. The uptake in the liver, based on the total Gd concentration injected, was 8.7% for the small micelles and 8.1% for the large micelles. No significant differences in the %ID values were observed between the two formulations and between the ApoE−/− mice and control WT mice (based on ANOVA results). These findings suggest that both mice species may exhibit similar biodistribution and excretion pathways for the mixed-micelle formulations. The %ID values found in liver, based on the weight of the mouse and the weight of the liver (w/w%), were 6.7% and 6.4% for the small and large micelles, respectively. No significant differences were observed between any of the w/w% values obtained in the liver (micelle formulations and WT vs. ApoE−/−). For all animals tested, the concentration found in the liver was greater than the QL associated with the method (Table 1).
Table 5. The Effect of Amphiphile on Liver Uptake in Both ApoE−/− and Wild-Type Mice 24 Hours After the Administration of a 0.038 mmol Gd/kg Dose*
% ID (total Gd content)
ApoE−/− (N = 6)
WT (N = 6)
The percent injected dose (%ID) was determined based upon the total amount of gadolinium administered and the total gadolinium content obtained and upon the dose administered per gram body weight and the dose obtained in liver per gram liver (w/w).
8.7 ± 1
8.7 ± 0.9
6.7 ± 0.8
6.7 ± 0.8
8.1 ± 0.9
8.0 ± 0.5
6.4 ± 0.3
6.2 ± 0.4
Figure 4a and b illustrate the lymphatic drainage and lymphatic uptake of the micelles 24 hr following administration of a 0.038-mmol Gd/Kg dose. The MR images confirm the laser-scanning confocal microscopy results (Fig. 4) that revealed substantial lymphatic drainage of the micelles. For all formulations and animals studied (control WT and ApoE−/−), significant lymphatic involvement was observed. Additionally, signal enhancement of the kidneys was observed at early time points p.i. with residual enhancement for up to 24 hr p.i. (Fig. 2a).
Two different amphiphiles were used in the current study to generate Gd mixed-micelles of various sizes. Large micelles (106 nm) were prepared using a lipid (POPC) containing a double bond in the oleoyl chain. The presence of the double bond reduces the ability of the micelle to pack, thereby increasing the overall size of the micelle. The small micelles (26 nm) were prepared using a lipid that did not contain a double bond (DPPC), which allowed for tight packing of the lipids. The GdDTPA lipid, surfactant, and fluorescent label were similar for both formulations, so the only variation was related to the amphiphiles used. Characterization of the resultant micelles revealed that both formulations were made up of polydisperse nanoparticles, as reflected by both the light-scattering (Table 1) and TEM (Fig. 1) results. Additionally, the TEM results strongly suggest that the nanoparticles formed as single-layer phospholipids with the polar heads (containing the Gd chelate) directed toward the solvent, as shown in Fig. 1a. These findings are consistent with the results obtained for other Gd mixed-micelle formulations (44, 50–55, 57, 62).
The relaxivity of the micelle formulations at 60 MHz was consistent with values reported for other Gd mixed-micelles at similar field strengths (54, 55). However, the relaxivity was low when compared to C18-GdDOTA and GdPCTA- formulations (51, 53). Additionally, the r1 values obtained for the small-micelle formulation were approximately 10% lower than those observed for the large-micelle formulation in both aqueous and liver homogenate matrices (Table 1). Based on the Gd loading (shown in Table 1), one would expect the large micelles to exhibit much larger relaxivities relative to the small micelles. However, the structure and assembly of the POPC lipid used to generate the large micelles may limit water exchange between water protons and the electrons associated with Gd. Studies have indicated that the length and type of lipid used to generate the micelle may greatly influence the water exchange rate (54, 55, 63). As a result, increased Gd loading and increased micelle size may not necessarily result in higher r1 values, as observed here. One limitation of the current study was that the relaxivities were not measured at the imaging field (400 MHz or 9.4T). Previous studies have indicated that the longitudinal relaxivities of Gd mixed-micelles typically reach maximum values at approximately 20–30 MHz (51, 53–55). As result, the relaxivity of most Gd mixed-micelles are reported at 20 MHz (0.47T). At this field the r1 values typically range from 12.8 to 28.5 s−1mM−1, with GdPCTA- derivatives exhibiting the highest relaxivity (51, 53–55). After the peak relaxivity, the r1 values decline as the applied magnetic field increases and eventually approaches zero at extremely high field. As a result, the r1 values obtained for the micelle formulations at 60 MHz will be significantly greater than the relaxivities obtained at the imaging field used (400 MHz or 9.4T). The enhancement observed in the vessel wall of ApoE−/− mice is therefore expected to increase with decreasing field, with maximum enhancement occurring at an imaging field of 0.5T.
The results from this study show that the choice of amphiphile greatly influenced the blood clearance in WT mice, as shown in Table 4. These results are consistent with previous studies that showed a direct link between the lipid used to create the micelle and the pharmacokinetics (52, 57). In the current study the large micelles prepared with POPC exhibited slower blood clearance than the small micelles prepared with DPPC. Based on experience with iron-oxide particles, one might expect faster clearance of the larger POPC micelles relative to the smaller DPPC micelles (64, 65). However, the excretion pathway of the micelles is presumably different from that of iron-oxide particles that are predominantly cleared by the mononuclear phagocytic system (MPS) and/or the reticular endothelial system (RES). Whereas a fraction of the micelle dose may be cleared by the MPS/RES, the majority of the dose is expected to be renally cleared (52). As a result, the findings of this study suggest that the kidneys may clear the smaller particles more quickly than the larger particles. To evaluate this in future studies, urine must be collected and analyzed in order to obtain information related to renal clearance.
The 24-hr p.i. liver uptake by ApoE−/− and WT mice was limited, as shown in Table 5. This limited uptake may be related to the time point studied. Due to the long circulation times, a fraction of the dose is still in the blood 24 hr p.i. As a result, the actual liver uptake may be greater than the values presented. However, at 24 hr p.i., <9 w/w% of the injected dose (or approximately 0.000103 mmol Gd) was found in the liver. Comparatively, if we assume 0.1% of a GdDTPA dose is not renally excreted, then the concentration of Gd potentially present in the body of a mouse would be approximately 0.000003 mmol Gd. Clearly, the liver retention of the Gd mixed-micelles should be reduced further in order to create a clinically useful diagnostic agent. In contrast to previous findings that the choice of amphiphile influences liver retention, the results of the current study show similar lever retention for both formulations of micelles studied (52). In this study it was apparent that the micelle size (based on the amphiphile used) did not influence liver uptake. In the current study the surfactant and GdDTPA lipids were kept constant so that global charge and surface composition was similar for both particles. In previous studies that showed variations in liver retention based on the lipid used, the micelles were formulated with varying surfactants (52). Therefore, factors other than hydrated micelle size may be modulating liver uptake.
One of the main findings of the current study is that the amphiphile did not influence the in vivo efficacy with respect to uptake in the vessel wall of ApoE−/− mice. In the current study no significant difference in the enhancement of the vessel wall was observed between the two formulations. Although the hydrated mean size of the micelles formulations was different, the large micelles were able to pass through a 25-nm pore, as shown in Table 2. This suggests that the flexibility of the micelles may enable the larger particles to pass through the abnormal endothelium associated with plaque. Additionally, similarities in the surface properties of the large and small micelles may also contribute to the observed similarities in efficacy. From the microscopy, it was apparent that the uptake mechanism of the micelles in plaque may be related to passive diffusion of the micelles through the abnormal endothelium. This hypothesis is further strengthened by the following findings: 1) no significant mixed-micelle uptake was observed in normal vessel walls of WT mice (Table 3, Fig. 3); 2) histology (Fig. 4) showed a predominant uptake of the micelles into the extracellular space of the plaque on the laminar side of the lesion in ApoE−/− mice; 3) colocalization of the micelles with cell nuclei (DAPI) revealed that micelles were not internalized by any cellular components associated with plaque (Fig. 3); and 4) maximum signal enhancement of the plaque occurred at relatively long time points (24 hr p.i.), suggesting slow uptake and accumulation of the micelles within the plaque. Due to slow excretion of the micelles out the plaque (via lymphatic drain and possibly reperfusion through the vessel wall), we were able to detect plaque at late time points p.i.
For both micelle formulations, heterogeneous enhancement of the vessel wall was observed in ApoE−/− mice 24 hr p.i., as shown in Fig. 2. The heterogeneous enhancement was limited to plaque-laden regions that were associated with the vessel wall, as confirmed by histology (Fig. 3). Additionally, there appeared to be a correlation between plaque composition and the signal enhancement observed in the vessel wall. Figure 2a shows the representative enhancement pattern observed for advanced plaques that exhibit an identifiable lipid core and fibrous cap. Post micelle injection, the non-lipid-rich areas of the plaque (or extracellular matrix, as confirmed by histology) enhanced, leaving the lipid core hypointense. On the other hand, the enhancement of smaller lesions (with limited lipid-core formation) was homogeneous, with increased overall signal gain, as shown in Fig. 2b.
In WT mice, limited signal enhancement of the normal vessel wall (%ENH < 10% and CNR increase < 1.5%) was observed 24 hr p.i. for both doses studied (Table 3). Additionally, confocal microscopy confirmed limited uptake of micelles within the normal vessel wall. Following injection of GdDTPA, enhancement of the vessel wall of ApoE−/− mice was only observed 1 hr p.i. We might have been able to increase the enhancement by GdDTPA by imaging the mice at earlier time points; however, due to limitations related to the scan times, this was not possible. At a dose of 0.1 mmol Gd/kg, the enhancement observed in the vessel wall of ApoE−/− mice 1 hr p.i. was approximately half that observed for the 0.015-mmol Gd/kg dose of micelles 24 hr p.i.
Lymphatic drain and uptake into thoracic lymph nodes was observed in all animals injected with both micelle formulations, as shown in Fig. 5. As a result, Gd mixed-micelles may potentially be useful for identifying and assessing the lymphatic system. Currently, ultrasmall iron-oxide particles, such as Ferumoxtran-10, are being used for evaluation of the lymphatic system by MRI (66–68). These particles exhibit a hydrated particle size of 20 nm and a blood halflife of >200 min in rats (64, 69). Ferumoxtran-10 has also been used for the detection of atherosclerotic plaque following administration of relatively high doses of iron (26, 43, 45, 47). Like the ultrasmall iron oxides, the micelles will allow for the evaluation the lymphatic system and plaque detection. However, an advantage of these particles over iron oxides is related to the fact that iron generates signal loss when compartmentalized in plaque and lymph tissue, and micelles generate signal gain. Because of issues related to partial voluming and other artifacts, it is difficult to evaluate signal loss by MRI (43). As a result, contrast agents that induce signal gain (such as Gd micelles) may be desirable. Additionally, because of the long circulation times associated with the micelles, it is possible to obtain detailed morphologic information about the entire cardiovascular system.
The results of the current study show that the choice of amphiphile used to prepare Gd mixed-micelles may influence the hydrated particle size, relaxivity, and blood clearance of the resultant nanoparticles. However, the in vivo MR efficacy with respect to uptake in the vessel wall of ApoE−/− mice was not affected by the amphiphile used. Additionally, the biodistribution with respect to liver uptake was not influenced by the choice of amphiphile. The similarities in the in vivo efficacy and biodistribution with respect to liver uptake may be explained by the flexibility of the large micelles formed and/or by the similarities in the surface properties of the micelles. In the current study the flexibility of the large POPC-based micelles was demonstrated. Large 100-nm micelles were able to permeate 25-nm pores. Therefore, the uptake into the vessel wall of ApoE−/− mice may not be hindered by micelle size. Additionally, in the current study the surfactant, Gd lipid, and fluorescent label of the micelle formulations were held constant. As a result, the surface properties of the micelles may be similar despite differences in the amphiphiles used. The results of this study strongly suggest that the liver uptake and vessel wall enhancement may be regulated by the surface properties of the micelle and not by other factors, such as micelle size. The Gd mixed-micelles evaluated in the current study exhibited long circulation times as well as significant enhancement of the vessel wall of ApoE−/− mice. Consequently, Gd mixed-micelles may become an important platform for the development of intravascular MR contrast agents that allow for a complete evaluation of the vasculature. These materials may allow for evaluation of vessel morphology, atherosclerotic plaque detection, and plaque characterization, including functional properties such as permeability and blood volume. The results also suggest that these agents may be useful for evaluating the lymphatic system. In order to develop clinically relevant micelles, however, future work will need to focus on reducing the liver uptake of these particles.
We are grateful to Edward A. Fisher, M.D., Ph.D., and Michael Lipinski for their help and comments during this project. This study was supported in part by the Advanced Imaging Program, Department of Radiology, Mount Sinai School of Medicine (Z.A.F.), and the Stanley J. Sarnoff Endowment for Cardiovascular Research, Inc. (V.A.). The MSSM-Microscopy Shared Resource Facility is supported by funding from NIH-NCI shared resources grant R24 CA095823, and NSF Major Research Instrumentation grant DBI-9724504.