High-contrast in vivo visualization of microvessels using novel FeCo/GC magnetic nanocrystals



FeCo-graphitic carbon shell nanocrystals are a novel MRI contrast agent with unprecedented high per-metal-atom-basis relaxivity (r1 = 97 mM-1 sec-1, r2 = 400 mM-1 sec-1) and multifunctional capabilities. While the conventional gadolinium-based contrast-enhanced angiographic magnetic MRI has proven useful for diagnosis of vascular diseases, its short circulation time and relatively low sensitivity render high-resolution MRI of morphologically small vascular structures such as those involved in collateral, arteriogenic, and angiogenic vessel formation challenging. Here, by combining FeCo-graphitic carbon shell nanocrystals with high-resolution MRI technique, we demonstrate that such microvessels down to ∼100 μm can be monitored in high contrast and noninvasively using a conventional 1.5-T clinical MRI system, achieving a diagnostic imaging standard approximating that of the more invasive X-ray angiography. Preliminary in vitro and in vivo toxicity study results also show no sign of toxicity. Magn Reson Med, 2009. © 2009 Wiley-Liss, Inc.

Noninvasive in vivo visualization of morphologically small vascular structures is of pivotal importance for the diagnosis of cardiovascular (1) and cancerous (2) diseases in which collateral, arteriogenic, and angiogenic vessel formation plays a critical role in their etiopathology. Moreover, monitoring the therapeutic effects in patients with vascular diseases necessitates the availability of a tool that is noninvasive and provides high-contrast and high-spatial-resolution angiographic images up to the scale of microvessels (i.e., ∼100 μm). However, noninvasive imaging of microvessels continues to pose technical challenges for present angiographic approaches. For example, X-ray angiography necessitates an invasive insertion of a catheter, and the use of ionizing radiation and iodine contrast adds potential danger and discomfort for the patient. In addition, the contrast level of the intravenously (i.v.) injected iodine contrast agent drops rapidly and can be difficult to time for small vessels where the flow is slow. Likewise, while conventional MR vascular imaging is routinely performed using gadopentetate dimeglumine contrast agents, similar to iodine, there is high contrast during the first pass, but the rapid drop in contrast leaves insufficient time for high-resolution small vessel imaging. Furthermore, with clinical MRI scanners, the imaging resolution is typically in the millimeter scale, which is not suitable to visualize micron-scale collateral and/or angiogenic vessels.

These challenges posed for current vascular imaging techniques provide a high impetus to seek the development of blood pool agents with high sensitivity and low toxicity capable of labeling vessels below millimeter scales. Some recent examples include Fenestra VC® (Advanced Research Technologies Inc., Montreal, Canada) for CT, macromolecular agents (3–5), protein-binding agents (6, 7), and ultrasmall iron-oxide nanoparticles (8) for MRI.

In this study, we report a significant progress we were able to achieve in visualizing microvessels through the use of novel FeCo-graphitic carbon shell (FeCo/GC) nanocrystals coupled with high-resolution in vivo MR vascular imaging technique. A rabbit hind limb peripheral ischemia model (9) was used for demonstration. To this end, our attempts were made possible by the unique properties of FeCo/GC nanocrystals (10, 11): (1) the highest magnetization among all magnetic materials (allowing high relaxivities and thus sensitivity), (2) biocompatibility due to graphite encapsulation and chemical functionalization/pegylation, and (3) positive intravascular T1 contrast capability (12). By utilizing the ultrahigh sensitivity and long blood circulation characteristics of these nanocrystals, we aimed to demonstrate that low-dose (approximately 20-fold reduction metal dose compared to conventional Gd-based contrast agent) FeCo/GC-based MRI (13) of microvessels (∼100 μm) can be achieved using a ubiquitous conventional 1.5-T clinical scanner. Finally, the results of our studies suggest that FeCo/GC nanocrystals are likely to be safe and nontoxic, without causing adverse side effects to the health of animals monitored over several months post-i.v. administration.


FeCo/GC Nanocrystal Synthesis

FeCo/GC Synthesis and Functionalization

FeCo/GC nanocrystal synthesis and concentration determination by calcination and ultraviolet-visible spectroscopy were performed as previously described (10). FeCo/GC nanocrystals were functionalized using a phospholipid (PL)-branched-polyethylene glycol (PEG) surfactant (14). A nanocrystal suspension was prepared by sonicating FeCo/GC nanocrystals in water with 1 mg/mL of surfactant for 1 h. To remove aggregates, the suspension was centrifuged at 15,000 revolutions per minute for 6 h. Excess PL-PEG was removed by filtration and put in phosphate-buffered saline prior to injection. All samples were made fresh and the concentration was determined before use. The difference with the previous sample described in Seo et al. (10) is that PL-branched PEG was used as surfactant instead of the PL-2k-PEG. The centrifugation time was also increased to 6 h instead of 5 min at a higher speed of 15,000 revolutions per minute instead of 10,000 g.

Synthesis of PL-Branched-PEG

This surfactant was prepared by mixing one equivalent of 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-PEG5000- NH2 (Sunbright DSPE-050-PA, NOF Corp., Tokyo, Japan) with 1.5 equivalents of (methyl-poly(ethylene oxide)12)3-poly(ethylene oxide)4-N-hydroxysuccinimide ester (Pierce Biotechnology, Rockford, IL) in methylene dichloride overnight, followed by the addition of two equivalents of N,N-dicyclohexylcarbodiimide (Sigma Aldrich, St. Louis, MO). Dicyclohexylcarbodiimide was used to reactivate any potentially hydrolyzed NHS groups on PEG during storage. The solvent was evaporated after another 24-h reaction and replaced with water. After a 10-min centrifugation at 10,000 g, to remove the insoluble solid (untreated dicyclohexylcarbodiimide), the supernatant was collected, filtered through a 0.2-μm membrane, and lyophilized for storage at –20°C.

Particle Size Measurement

The hydrodynamic size after functionalization with PL-branched-PEG was measured through dynamic light scattering with a Brookhaven 90 Plus Particle Size Analyzer. Light was collected at a scattering angle of 90°.

In Vitro/In Vivo Experiment Setup

All animal experiments were performed according to the protocol approved by the Stanford University Administrative Panel on Laboratory Animal Care.

General Rabbit Protocol

New Zealand white rabbits weighing ∼3.5-5.5 kg were used for all in vivo experiments. Rabbits were first anesthetized with an injection of ketamine/xylazine cocktail with an intramuscular injection (ketamine 35 mg/kg + xylazine 5 mg/kg). Then, contrast agents were injected through the catheterized ear vein. Immediately after the contrast agent injection, 1 mL of saline was injected. The contrast agents were injected in 5 mM concentration solution. The contrast agent injection amount was determined assuming the rabbit blood volume ratio of 60 mL/kg. Given the contrast agent concentration of 5 mM, rabbit blood volume ratio assumption of 60 mL/kg, and 1-mL saline push, to achieve a target blood volume concentration of mM for a kg rabbit, the following formula (Eq. 1) was used to calculate the required injection volume (mL).

equation image(1)

Blood Circulation Time Measurement

To estimate the blood circulation time, normal rabbits were injected with FeCo/GC nanocrystals while the aorta signal was monitored using T1-weighted pulse sequences. The time-dependent aorta signal gave an estimate of the blood circulation time.

Rabbit Aortic Stenosis

Stenosis of the aorta was generated by high-cholesterol feeding, in combination with an amaroid constrictor. The aortic procedure was performed approximately 1 week after initiation of a high-fat diet (1% cholesterol and 4% coconuts oil). All surgical procedures were conducted under general anesthesia and sterile conditions, as previously described (15). The animal was given ketamine (35 mg/kg) and xylazine (5 mg/kg) intramuscularly 30 min before the operation. After endotracheal intubation, anesthesia was maintained with 1.5% halothane and oxygen administered at a rate of 65-80 mL/min through an automatic ventilator (Harvard Respiration Pump). A midline abdominal incision approximately 5 cm long was made and then the abdominal aorta was dissected free. An MR-compatible amaroid constrictor (Research Instruments SW, Escondido, CA) (16) was placed around the abdominal aorta at 1-2 cm below the renal arteries. The constrictor consisted of a casein bar with central lumen housed in an acrylic covering, rather than the standard metallic covering. In the body, the dehydrated casein absorbs fluid and swell, thus inducing a progressive stenosis of the vessel over several months. Finally, the chest was closed and followed by air suctioning to resume physiologic chest pressure. MR images were obtained >6 months after the surgery. After the scan, the rabbit was euthanized by an i.v. injection of pentobarbital (40 mg/kg).

Rabbit Hind Limb Ligation

Unilateral hind limb ischemia (9) was induced in New Zealand white rabbits (3- to 4-kg body weight). Animals were subcutaneously premedicated with ketamine (40 mg/kg) and xylazine (4 mg/kg) and intubated. The endotracheal tube was connected to a ventilator, which maintained ventilation, using 2 L oxygen mixed with 1-3% isoflurane. A longitudinal incision was then performed, extending from a point 4-5 cm proximal to a point 1-2 cm proximal to the patella. The superficial femoral artery was dissected free and ligated at 2 places by 4-0 silk (Ethicon, Sommerville, NJ). Subsequently, the animals were weaned from ventilation and transferred to the animal care facility. MRI images and X-ray angiograms were obtained approximately 6 weeks after the surgery. The rabbits were euthanized after the X-ray imaging.

X-ray Angiography

After the MRI scan, an X-ray angiogram of the rabbit hind limb was acquired using OEC 9600 (40k Vp, 0.2 mA, 30 frames/sec) cardiac C-arm (OEC Medical systems, Inc., Salt Lake City, UT). After induction of anesthesia, as previously described, vascular access was obtained using standard 4-Fr vascular sheath, placed in the right or left carotid artery. A 4-Fr diagnostic catheter (Judkins right type) was advanced into the descending aorta and the aortoangiography was performed. After X-ray angiography, animals were euthanized by KCl infusion. The vessels were excised and were perfusion fixed with phosphate-buffered saline followed by 10% formalin and then placed in 10% formalin for fixation.

Rabbit Femoral Artery Histology

The excised occluded femoral artery was sectioned and stained with hematoxylin-eosin, elastic van Gieson stains. The stained histology confirmed the formation of total occlusion with the rabbit hind limb surgical ligation model.

FeCo/GC Nanocrystal Excretion

Normal rabbits were repeatedly injected (5 mL of 3-mM solution injected five times on day 1 and 9 mL of 3-mM solution injection on day 6) with FeCo/GC nanocrystals. Images were obtained with the same T1-weighted sequence as in the blood circulation time measurement experiment.

In Vitro FeCo/GC Toxicity Study

To test in vitro toxicity of FeCo/GC nanocrystals, 30 μL of a highly concentrated solution was added to macrophage cells in 100 μL of cell medium. Cells were cultured with Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin in a 5% CO2 atmosphere at 37°C. Macrophage cells were incubated in a 96-well plate for 48 h, with varying concentrations of FeCo/GC nanocrystals functionalized with PL-branched-PEG. Prior to measuring cell viability, all medium and nanocrystals were removed from the wells, followed by the addition of 100 μL of fresh medium and 20 μL of CellTiter 96® AQueous One Solution Reagent (Promega Co., Madison, WI). Following incubation for 1.5 h, the absorbance was recorded at 490 nm and the cell viability was determined.

In Vivo FeCo/GC Toxicity Study

To evaluate the toxicity of FeCo/GC nanocrystals in vivo, 16 female Balb/c mice weighing 17-21 g were injected with 4 nm FeCo/GC PL-branched-PEG-carboxylate. The PL-branched-PEG-carboxylate was prepared by mixing four-arm PEG-N-hydroxysuccinimide (Mn ∼10,000; Laysan Bio, Inc., Arab, AL) with 1,2-distearoyl-3-sn-glycero-3-phosphoethanolamine (Lipoid LLC, Ludwigshafen, Germany) in a 1:1 molar ratio overnight and then purifying by dialysis. The injected nanocrystal dose was ∼210 μL of a solution with a metal concentration of 15.5 mM. This dose yields a ∼2.2- to 2.7-mM blood pool concentration, which corresponds to ∼11-14 times the rabbit dose. After 1 or 3 months post-injection, blood was drawn from the submandibular vein or via cardiac puncture into heparinized tubes for analysis.

Histologic evaluations were done on three treated mice and two control mice 3 months after the injection of the nanocrystal solution. All soft tissues were collected and examined, preserved with formalin for 48 h, and then routinely processed for light microscopic examination of hematoxylin-eosin-stained sections.

MRI Data Acquisition

All MRI data were acquired using a 1.5-T GE Excite whole-body MRI system (maximum gradient amplitude: 40 mTm-1; maximum slew rate: 150 mTm-1 ms-1).

Relaxivity (r1 and r2) Measurements

T1 and T2 were measured at six different metal concentrations of FeCo/GC nanocrystal suspensions (0.05, 0.1, 0.2, 0.4, 0.8 mM: Fe and Co concentration combined). The corresponding particle concentration can be obtained by multiplying 6.1854 × 10−8 to the mM concentration (e.g., 0.2 mM metal concentration = 12.38 nM particle concentration). For the T1 measurement, inversion recovery sequences were used (17) with a pulse repetition time (TR) of 6000 ms and inversion time of 50, 100, 150, 200, 250, 300, 350, and 2000 ms. For the measurement of T2, spin-echo sequences were used (17), with a TR of 3000 ms and echo time (TE) of 8, 12, 20, 30, 50, 80, 160, 320 ms. The field of view (FOV) was 22 × 22 cm2 and the image matrix size was 128 × 128.

Blood Circulation Time Measurement

A standard GE head coil was used. Fat was suppressed with a spectral-spatial excitation pulse. Three-dimensional (3D) stack-of-spirals acquisition was designed to have 40 interleaves and 40 kz-phase encodes. The FOV was designed to be 30 × 30 ×8 cm3, with a resolution of 1 × 1 × 2 mm3. TR was 20 ms, with a flip angle of 60°. The resulting volume scan time is 32 sec.

For all the in vivo studies, TR and TE were chosen to be as short as possible under the constraint of obtaining proper FOV and resolution to maximize contrast. High flip angles were used since the Ernst and Anderson (18) angle for the FeCo/Gc nanocrystals at 0.2 mM concentration is around 60°.

In Vivo Demonstration of High Sensitivity

For the comparison of the enhancement effects in rabbit aorta, 3D spoiled-gradient recalled echo (SPGR) images were acquired with a TR of 33.3 ms, TE of 4 ms, flip angle of 60° and fat saturation. The FOV was 18 × 18 × 5.6 cm3, with a matrix size of 192 × 192 × 28.

High Spatial Resolution MRI of Blood Vessels

The high-resolution images comparing ferumoxtran-10 and the FeCo/GC 7-nm particles' performance were obtained using a custom-designed pulse sequence with a spectral-spatial excitation and stack-of-spirals acquisition incorporated into an SPGR sequence. The spectral-spatial excitation pulse (19) was designed to selectively excite water (excluding fat) while simultaneously limiting the slab volume. The pulse duration was 8.536 ms. The FOV was designed to be 4 × 4 × 1 cm3, with a resolution of 78 × 78 × 500 μ m3. TR was 40 ms and TE was 5.128 ms, with a readout duration of 16.224 ms and flip angle of 60°. 3D stack-of-spirals acquisition was designed to have 200 interleaves and 20 kz-phase encodes. The resulting volume scan time is 2 min 40 sec. To improve signal-to-noise ratio (SNR), nine images were averaged.

Imaging Angiogenesis in Rabbit Aorta

The rabbit aorta was imaged with a 3D SPGR sequence with a TR of 35 ms, TE of 4 ms, flip angle of 60°, and fat saturation. The FOV was 18 × 18 × 2.8 cm3, with a matrix size of 256 × 256 × 28.

Monitoring Angiogenic Vessel Growth in Rabbit Hind Limb

Two different protocols were used to image the rabbit hind limb. In both cases, the aforementioned spectral-spatial excitation was used. The large FOV covering both hind limbs was used to show the difference between the two limbs in terms of vessel distribution. The hind limb ischemia model is expected to result in angiogenic vessel growth in a large area of the limb, unlike in the case of local angiogenesis observed in the aortic stenosis. To image a large FOV covering both hind limbs, a 5-inch surface coil was placed on top of a rabbit in supine position. A coronal plane was scanned with an imaging FOV of 20 × 20 × 3 cm3 and resolution of 300 × 300 × 500 μ m3. TR was 40 ms and TE was 5.128 ms, with a spiral readout duration of 7.36 ms and flip angle of 45°. 3D stack-of-spirals acquisition was designed to have 200 interleaves and 60 kz-phase encodes. The resulting volume scan time is 8 min.

Once the region of the stenosis was identified, a small FOV image was obtained using a 1-inch surface coil (Fig. 4c) in higher resolution. The FOV was 6 × 6 × 3 cm3, resolution was 117 × 117 × 500 μ m3 TR was 40 ms and TE was 5.128 ms, with a spiral readout duration of 16.224 ms and flip angle of 45°. 3D stack-of-spirals acquisition for this protocol was also designed to have 200 interleaves and 60 kz-phase encodes. The resulting volume scan time is 8 min. To improve SNR, two images were averaged.

MRI Data Analysis

Relaxivity (r1 and r2) Measurements

The data were analyzed to estimate T1 and T2 values through nonlinear least-squares fits to the inversion recovery and spin-echo decay curves sampled at various TE. The slope of Tinline image and Tinline image vs concentration (r1, r2) was then calculated through linear fitting. The slope, which is the rate of Tinline image and Tinline image change with the contrast concentration, indicates the efficacy of the contrast agent as a T1 or T2 contrast agent (17).

Signal Intensity vs Concentration Calculation

To calculate the expected signal enhancement vs concentration of various contrast agents in Fig. 3b, T1, T2 values expected for each concentration level (C) were calculated from T1, T2 values of blood without any contrast (T10 = 1200 ms, T20 = 327 ms at 1.5 T (20)) and r1, r2 of each contrast agents as follows (Eq. 2).

equation image(2)

Then, the resulting signal intensity from the T1, T2 was calculated for the case of SPGR sequence used for the imaging of the aorta. (TR: 33.3 ms, TE: 4 ms, flip angle: 60°) (Eq. 3).

equation image(3)

equation image was then calculated and plotted to show the enhancement ratio.

Signal Intensity Measurement

For the comparison of the signal intensities, the cross-sectional images were selected along the aorta. Signal intensity was measured by averaging the signal across the ROI selected within the aorta.


For the aortic, circumflex iliac, and femoral artery angiography images, maximum intensity projection (MIP) was calculated from the whole volume or selected slices from the volume (25 slices out of 60 total selected for Fig. 6c). Due to the high contrast provided by FeCo/GC, small vessels are observed in high quality, despite the partial volume effect that can be expected from the lower resolution in the through-plane direction. The smallest visible vessel sizes are down to the in-plane image resolution of 703 μm for the aortic, 78 μm for the circumflex iliac, and 117 μm for the femoral artery angiography.

Signal Intensity Plot

To evaluate the spatial resolution of small vasculature, cross-sectional signal intensity was plotted across small vasculature. A single dot on the plot represents each voxel.

SNR Measurement

Blood pool signal SNR was measured by selecting an ROI within the blood pool. The average signal intensity (Sb) was then divided by the SD (σb) to obtain SNR (Eq. 4).

equation image(4)

Contrast Resolution Measurement

Contrast resolution of the blood pool vs the surrounding tissue was measured by selecting ROIs in the blood pool and the adjacent tissue (usually muscle tissue). Average signal intensity in blood pool ROI (Sb) and the surrounding tissue ROI (Sm) was then measured to calculate contrast resolution as shown in Eq. 5.

equation image(5)


FeCo/GC Nanocrystals

The FeCo-Graphitic shell nanocrystals (Fig. 1a) were synthesized as previously reported (10), with the particle core size distribution of 7 ± 1.2 nm (Fig. 1a). We made an important advancement in the current work by chemical functionalization of the graphitic shells of the nanocrystals using PL with 10 kDa branched-PEG chains (Fig. 1b) (14). Clusters of nanocrystals were removed by increasing the centrifugation time (10). The hydrodynamic size was 23.1 ± 2.6 nm (Fig. 1d). The branched PEG and cluster removal resulted in longer blood circulation time and delayed uptake by the reticuloendothelial system for the improved nanocrystal particles due to branched pegylation (14, 21). A blood circulation half-life of approximately 4 h was measured for nanocrystals i.v. injected into rabbits (Fig. 2). The better declustering reduced the r2/r1 ratio, which is desirable for highly sensitive positive-contrast blood pool imaging. As seen in Fig. 3a, compared to results reported earlier (10), the improved FeCo/GC has higher r1 relaxivity (97 mM-1 s-1) and lower r2 (400 mM-1 s-1) relaxivity and thus lower r2/r1. These relaxivities were among the highest on the per-metal-atom basis among all known MRI contrast agent materials, owing to the superior magnetic properties of FeCo alloy crystals (10, 22–25). Per-particle basis r1 relaxivity is 1.5682 × 1012 mM−1 s−1 and r2 relaxivity is 6.4668 × 1012 mM−1 s−1.

Figure 1.

FeCo/GC nanocrystals with branched-PEG functionalization for in vivo applications. a: Transmit electron microscope (TEM) image showing the FeCo/GC nanocrystals. b: A schematic diagram showing a FeCo/GC nanocrystal with PL-branched-PEG attached. c: Picture of a FeCo/GC solution at 0.2-mM metal molar concentration. d: Hydrodynamic size distribution of 7-nm nanocrystals with PL-branched-PEG measured by dynamic light scattering. The graph shows the number of counts at each diameter out of 313 total counts. The results show that 99% of the sample is between 19.2 and 28.4 nm, but most of it (93%) is at or below 25.8 nm.

Figure 2.

Seven-nanomolar FeCo/GC contrast agent was injected to a normal rabbit and the blood circulation time was measured by monitoring the signal intensity of the aorta. The signal enhancement in the low concentration range (<0.2 mM) is approximately linear. Therefore, the blood circulation half-life can be estimated by estimating the time point where the signal intensity enhancement is half of that obtained immediately after the injection. The blood half-life is estimated to be approximately 4 h.

Figure 3.

Relaxivities of commercial and our FeCo-graphitic carbon shell nanocrystals measured at 1.5 T and highly sensitive MRI enabled by FeCo/GC nanocrystals. a: The newly improved FeCo/GC material with a more uniform size through elimination of clustering resulted in a smaller r2/r1 ratio. These are important properties for in vivo applications where long blood circulation with reduced uptake from the reticuloendothelial system is of interest. b: Calculated signal enhancement vs concentration for commercial and FeCo/GC agents. MRI showing the rabbit aortic signal enhancement by (c) ferumoxtran-10 and (d) FeCo/GC nanocrystals, respectively; 0.2-mM blood pool concentration was targeted for both agents. The difference in resulting signal enhancement factors of 3.8 for ferumoxtran-10 and 9.5 for FeCo/GC was observed, consistent with calculations in (b).

In Vivo Demonstration of High-Sensitivity MRI Using FeCo/GC Contrast Agent

To test the comparative relaxivity and thus sensitivity improvement of FeCo/GC contrast agent, we i.v. injected equal amounts (blood pool target concentration of 0.2 mM of metal) of commercially available ultrasmall particle iron oxide agent ferumoxtran-10 (Combidex®; AMAG Pharmaceuticals, Inc., Cambridge, MA) and our FeCo/GC contrast agents into rabbits and carried out MRI of the rabbit aorta (Fig. 3c,d). At the same targeted 0.2-mM blood pool metal concentration, ferumoxtran-10 exhibited a positive signal enhancement factor of approximately 3.8 (Fig. 3c) with a blood pool SNR (determined by the blood pool signal intensity divided by the SD; see Materials and Methods) of 5.6, and contrast resolution (as defined in Materials and Methods) of 0.60, while FeCo/GC nanocrystals displayed a substantially higher positive enhancement factor of 9.5 (Fig. 3d) and blood pool SNR of 13.9, and contrast resolution of 0.80 consistent with the higher relaxivities of the latter than the former (r1 = 15 mM-1 s-1 vs 97 mM-1 s-1).

Fig. 3b shows the calculated contrast enhancements for various contrast agents (see also Materials and Methods) based on the measured r1, r2 values (Fig. 3a) for a T1-weighted SPGR sequence. Consistent with experimental results, calculations find that FeCo/GC is expected to give a high contrast enhancement factor of approximately 9 (Fig. 3b, red curve) at 0.2 mM per-metal-atom basis concentration. The increased positive enhancement (see the steep upward slope with increasing concentration of the contrast agent) in the low-concentration region is attributed to the high r1 relaxivity. The reversal of the contrast at higher concentration is due to the T2 effects caused by the large r2.

Ferumoxtran-10 has a long circulation time (blood half-life of approximately 24 h), but the r1, r2 relaxation effects are such that even at 1-mM concentration (five times that used for enhancement factor of 9 using FeCo/GC), an enhancement factor of only 7 can be achieved (Fig. 3b, green curve). However, due to contrast reversal, further increase in ferumoxtran-10 concentration will not result in additional contrast enhancement.

To achieve the same contrast enhancement as the FeCo/GC nanocrystals at 0.2-mM concentration, the clinically prescribed Magnevist contrast agent would require a minimum of 10-fold higher concentration in the blood pool (Fig. 3b, dotted blue curve). However, since Gd-based contrast agents, such as Magnevist, have a short blood half-life, it is not viable to equate a 10-fold increase in injection with a 10-fold increase in the actual (and more relevant) blood pool concentration. Therefore, even higher dosage of Gd-based contrast agent would be necessary to achieve the intended concentration within the blood pool.

High-Spatial-Resolution MRI of Blood Vessels

To investigate the potential of FeCo/GC nanocrystals for high-resolution MRI of vessels such as those involved in microvasculature growth, we developed an ultra-high-resolution MR angiography technique by utilizing surface coils (26) (Fig. 4) and highly optimized pulse sequence for high-spatial resolution imaging (27, 28) (see Materials and Methods). Blood vessels in the posterior side of the upper hind limb of healthy rabbits were imaged by MRI following the injection of ferumoxtran-10 and FeCo/GC at a targeted 0.2-mM blood pool concentration. Fig. 4b demonstrates the in vivo visualization of rabbit superficial circumflex iliac artery and branches (29) (blood pool SNR: 11.9; contrast resolution: 0.77) with diameters less than 100 μm, using FeCo/GC. In contrast, ferumoxtran-10, even at five times higher dosage (1-mM targeted blood concentration) than FeCo/GC failed (blood pool SNR: 5.8; contrast resolution: 0.58) to label small vasculature (Fig. 4a).

Figure 4.

In vivo high-resolution MRI. (a) A vasculature MIP image enhanced by 1-mM blood pool concentration of ferumoxtran-10. (b)A vasculature MIP image enhanced by 0.2-mM blood pool concentration of FeCo/GC. (c) Picture of 1-inch-diameter custom-designed surface coil used for this experiment. (d) Cross-sectional signal intensity plot across a small vasculature confirms that it has a sharp single voxel (78 μm) peak, suggesting vessel size on the order of 100 μm. Even at a concentration five times higher than that used for FeCo/GC, ferumoxtran-10 does not provide sufficient contrast to visualize small vasculature (<100 μm).

Imaging Microvessels in Rabbit Aorta

The high sensitivity and long circulation time of FeCo/GC nanocrystals make them ideally suited to label morphologically small vessels as typically displayed by collateral, arteriogenic, and angiogenic vessels. Fig. 5 displays the results of our studies investigating the progressive stenosis of the rabbit aorta through FeCo/GC-labeled MR angiography (Fig. 5a). The results obtained using a 1.5-T clinical MRI with a quadrature head coil and i.v. injection of the FeCo/GC nanocrystals suggest that we were able to label vessel emergence at a submillimeter scale around the aortic occlusion. Despite the limited spatial resolution of the image acquisition, the high vessel contrast (blood pool SNR: 14.0; contrast resolution: 0.81) provided by FeCo/GC nanocrystals allows not only visualization of the larger collateral vessels but also a putative ability to label finer-scale structures such as arteriogenic and potentially angiogenic vasculature (Fig. 5b). This result further provides initial proof of concept of highly sensitive MRI of small microvessel growth using our FeCo/GC nanocrystal contrast enhancement.

Figure 5.

FeCo/GC enhanced MRI at 1.5 T of rabbit microvessels around aorta stenosis. (a) An FeCo/GC-enhanced 3D SPGR MIP image shows small microvessel formation around an aortic stenosis in a rabbit. The spontaneously formed tortuous vessels can be clearly seen in (b).

Monitoring Morphologically Small Microvessels in Angiogenic Rabbit Hind-Limb Model

Next, we explored high-resolution MRI of microvessels using the rabbit hind-limb ischemia model, which has been well established for studying microvessel formation in peripheral arterial disease (9, 30). A total occlusion (Fig. 6) was surgically induced in the right femoral artery, and the spontaneous formation of microvessels was monitored 6 weeks postsurgery by both standard X-ray angiography and our FeCo/GC MRI.

Figure 6.

Normal femoral artery shows large lumen area in the middle (a). The femoral artery taken from the portion identified as the occlusion site (b,c)shows that the artery is completely occluded. The histology was obtained around the ligation(b) and in the proximal portion of the complete total occlusion (CTO) (c). The left images in (b) and (c) show hematoxylin-eosin staining result, and the right images show the elastic van Gieson staining result. These results show that the rabbit hind limb ligation model successfully created CTO.

The presence and location of spontaneously formed microvessels were confirmed through standard X-ray angiography. Here, X-ray angiography revealed a stenosis in the right femoral artery (Fig. 7a). Tortuous small spontaneously formed vessels connecting the proximal and distal parts of the occluded femoral artery were observed (Fig. 7a). An angiogram obtained in a rabbit left hind limb (normal artery without surgery) was also obtained for comparison (Fig. 7b). Note that the X-ray angiogram was obtained by invasively inserting a catheter from the carotid artery down to the femoral artery where the iodine contrast agent was injected during imaging.

Figure 7.

Imaging of microvessel in rabbit hind limb model of peripheral arterial disease. a: X-ray angiogram obtained 6 weeks after the surgery that induced stenosis in the right femoral artery. Tortuous small vessels are seen throughout the right hind limb. b: X-ray angiogram obtained in a normal left hind limb shows normal arterial vessels for comparison. c: MRI (1.5 T, 5-inch surface coil) was performed 6 weeks after the surgery with a FeCo/GC injection targeting a 0.2-mM blood pool concentration. A coronal volume was scanned after the injection, with a 20 × 20 × 3 cm3 FOV and 300 × 300 × 500 μ m3 resolution. MIP image shows the stenosis in the right femoral artery and tortuous microvessel formation throughout the right hind limb (red circle). Note the X-ray and MR images are acquired in different projection angles. d: High-quality fat suppression and image quality are demonstrated in selected cross-sectional images.

The MIP of the FeCo/GC contrast-enhanced MRI image (Fig. 7c) highlighted both the arteries and nearby veins since the imaging was performed in steady state. In the surgically operated right hind limb, narrowing and occlusion of the femoral artery were observed by MRI next to the femoral vein (Fig. 7c, Fig. 8a,b). Interestingly, we also observed a significant number of tortuous small vessels in the right hind limb connecting the proximal and distal part of the femoral artery (see Fig. 7c and Fig. 8), comparable to those seen using the more conventional X-ray angiography (Fig. 7a,b), suggesting that they may indeed represent collateral, arteriogenic, and/or angiogenic microvessels induced by the applied hind limb ischemia (blood pool SNR and contrast resolution in Fig. 7c: 11.4 and 0.71; blood pool SNR and contrast resolution in Fig. 8b,c: 14.2 and 0.70).

Figure 8.

High-resolution MRI of microvessels. a: A zoom-in view of the right hind limb with the stenosis from Fig. 7c. b: A 6 × 6 × 3 cm3 FOV, 117 × 117 × 500 μm3 resolution MIP image (1.5 T, 1-inch surface coil) of the region near the occlusion. c: A MIP of 25 slices selected from the full data set (60 slices). Small microvessels with diameters in the order of 100 μm are observed in the high-resolution images. d: A further zoomed-in view of (c)and its cross-sectional signal intensity plot across a small vasculature confirms that it has a sharp single voxel (117 μm) peak, suggesting vessel size on the order of approximately 100 μm.


In the present study, we investigated the properties of a novel FeCo/GC contrast-enhanced MR angiographic approach that displayed exceptional contrast resolution and sensitivity for MRI using 1.5-T clinical scanners. While significant prior advances in high-resolution imaging of micro-vessels at 1.5 T exist (3–6, 8, 31–34), clear imaging of microvessels on the order of 100 μm has continued to present a challenge. Major challenges include insufficient circulation time for high-resolution imaging, low sensitivity resulting in low contrast in microvessels, unpredictable contrast, high dose, and toxicity (35). An ideal contrast agent for angiography should have a sufficiently long blood half-life (6, 31–33) to allow for image acquisition and should be efficient in generating predictable image contrast with low metal doses (3–5, 31) and without the risk of toxic side effects (33, 34). FeCo/GC nanocrystals are long circulating, are highly sensitive, generate contrast free of complication from complex pharmacokinetics, and are stable and inert.

Here, we demonstrated a significant 20-fold reduction in metal dosage (compared to the clinically used Gd-based contrast dosage) attained by the use of FeCo/GC contrast agent. The dosage of clinical Gd-based contrast agents range from 0.1 to 0.4 mmol/kg of body weight, while 0.2 mmol/kg is a commonly used dose for MR angiography at 1.5 T (36). Targeting a blood pool concentration of 0.2 mM is equivalent to approximately 0.01 mmol/kg dose, which is 1/20 of the Gd-based contrast agent dosage for MR angiography. The high sensitivity of our contrast agent is a direct result of the ultra-high relaxivities of the FeCo/GC nanocrystals originated from the high saturation magnetization of FeCo (approximately three times higher than iron oxide). Although the exact relaxation mechanism is not known, due to the graphitic shell fully surrounding the metal core of the nanocrystal, relaxation must occur by an outer-sphere mechanism. Any nonencapsulated metal that could contribute to the relaxivity is removed by hydrofluoric acid (HF) etching prior to functionalization.

The high sensitivity and long circulation of the FeCo/GC agent, combined with a high-resolution MRI technique, allowed for exceptional SNR and contrast resolution. In MRI, SNR is proportional to the voxel size. Reducing voxel size leads to low SNR. Therefore, compensating for the SNR loss of high-resolution imaging (down to 10–100 μ m) is of great importance. Some of the approaches include using higher field strength scanners since SNR in MRI is approximately proportional to the field strength. However, high-field-strength systems (≥ 7 T) impose many problems over clinical-field-strength systems (1.5 T), including patient safety and comfort, availability, and high cost. Therefore, we chose to utilize the standard clinical field strength (1.5 T) with small surface coils (26, 37) and 3D imaging that allow SNR efficient acquisition. While the use of single small surface coils reduces volume coverage, combining with multiple surface coil phase arrays (38) can easily recover volume coverage, making it possible to translate the current technique into the clinical applications. Likewise, scan time reduction is important since imaging in vivo involves motion. In high-resolution imaging, even a small motion can lead to significant blurring. Therefore, rapid imaging is of great importance. To further accelerate image acquisition, we used 3D spiral encoding (27, 28, 39, 40) that reduced our encoding time by more than a factor of 2. Furthermore, to ensure good background suppression by avoiding fat signal and its blurring, water frequency was selectively excited using spectral-spatial excitation pulses (19).

Like any novel contrast agent, ensuring its biocompatibility and low toxicity is of crucial importance. Here, several lines of evidences point to the biocompatibility of FeCo/GC. First, the metal component is completely enclosed by a graphitic-carbon shell. The metal core remains intact in harsh chemical processes such as soaking in a solution of 10% HF in water (80%) and ethanol (10%). The graphitic-carbon shells are intact and fully encapsulating the metal core in any biologically relevant environment (10). Second, from our rabbit experiments, most of the FeCo/GC material ends up in the reticuloendothelial system, including the liver, after several hours of blood circulation. This leads to dark signal in the liver due to high concentration of the contrast material (Fig. 9a). Importantly, the liver then also starts to excrete the material through the biliary system. This is evidenced by the fact that following the dark liver signal, the intestine starts to light up, suggesting that the FeCo/GC material is being excreted into the intestine (Fig. 9b). The bright signal is also indicative of the continued structural integrity and functionality of the shell material since highly sensitive T1 contrast is only expected to be achieved if the particles stayed intact without aggregation or oxidation. The liver signal also starts to become lighter, which is an indication of reduced contrast concentration in the liver (Fig. 9b). Therefore, the material is believed to be excreted from the body through the biliary system (14). In the future, we expect to perform more complete biodistribution studies to confirm the excretion.

Figure 9.

MIP image of rabbit torso and in vitro toxicity data. a: Before any contrast injection, the liver typically shows bright signal and the intestine shows darker signal than the liver. The rabbit was then injected with contrast agents at multiple time points. b: Six hours after the latest injection, the rabbit liver shows dark signal and the intestine shows bright signal. This suggested that FeCo/GC contrast agent was initially taken up by the liver and then excreted through the biliary system through the intestine and then feces, much like carbon nanotubes, as reported recently. c: Percent viable macrophage cells after a 48-h incubation with varying concentrations of nanocrystals suspended with PL-branched-PEG-(PL-3PEO). The vertical blue line represents the target blood pool concentration (0.2 mM) for imaging, and no toxic effects are observed.

For in vitro toxicity assessment, in addition to the initial toxicity tests (e.g., cell proliferation assay) presented in Seo et al. (10), we incubated macrophage cells in varying concentrations of FeCo/GC solutions for 48 h. We then measured the viability of cells and found no toxicity in vitro, even at concentrations greater than 15 times the blood pool target concentration (Fig. 9c).

In terms of in vivo toxicity, we injected a total of approximately 10 rabbits with the FeCo/GC material in high concentrations (>5 × higher than blood pool target concentration for MRI) and monitored for over 6 months, and no sign of the contrast agent–related adverse health effects was observed. Notably, excretion by the biliary pathway and lack of obvious toxic effects were similar to the observations of carbon nanotubes that have similar graphitic surfaces and chemical functionalization as our nanocrystals (14). Furthermore, we conducted a pilot in vivo toxicity study by monitoring the body weight, behavior, blood chemistry, and hematology (41) in mice for up to 3 months, followed by histologic analysis of soft tissues. Compared to the untreated control group (five mice), there were no abnormalities in body weight (Fig. 10a), behavior, or blood test (Table 1). While accumulation of nanocrystals was present in the cytoplasm of hepatic resident macrophages and splenic macrophages, there was no evidence of tissue damage or loss of organ function of the liver and spleen (Fig. 10b-e) from the histologic evaluation or blood biochemical testing. All other organs examined appeared normal. The presence of nanocrystals in the liver and spleen indicates that the macrophages play a role in the metabolism of this agent and the liver may play an important role in nanocrystal excretion via the hepatobiliary system. The retention of the nanocrystals in the liver and spleen may be due to the long life span of the macrophages in these organs. The lack of organ or tissue damage observed from histologic evaluation, combined with the normal blood biochemical data and lack of obvious behavioral or body weight changes, suggests the overall nontoxic nature of FeCo/GC nanocrystals.

Figure 10.

Body weight and histologic data for in vivo toxicity study. Body weight changes over time (a) in control mice and FeCo/GC injected mice. Mice given the nanocrystal injection at day 0 were given a dose 11-14 times that needed for MRI. The lack of any significant deviations in body weight, as compared to the control mice, supports the argument that there are no obvious toxic effects from the nanocrystals. Hematoxylin-eosin-stained slices of a liver from a control mouse (b), nanocrystal-injected mouse (c), control spleen (d), and nanocrystal-injected spleen(e). The asterisks in (b) and (c) mark the portal triad. The arrows in (c)and (e) mark nanocrystals in macrophage cells. White pulp and red pulp in images (d) and (e)are marked as w.p. and r.p., respectively. The scale bars in all images indicate 50 μm. No damage is observed in any of the tissues collected from the nanocrystal-treated mice.

Table 1. Serum Chemistry and Hematology Data for balb/c Mice Injected With a High Dose of FeCo/GC Nanocrystals at 1 Month and 3 Months Postinjectiona
Liver panel and blood chemistryUnit1 mo3 moControlReference range (literature)
  • a

    The results indicate that most values are either within reference range (41) or only slightly deviate from the given range and do not indicate any toxic side effects due to nanocrystal injection. The blood data indicate that the liver, kidneys, spleen, and immune system were functioning normally, with no signs of damage. Results are based on three to seven mice per group.

  • b

    Reference range given is for CD-1 mice; all other ranges are for Balb/c mice (all mice less than 1 year of age).

  • c

    — = Data not obtained.

Aspartate aminotransferaseIU/L7821.696.935.3109.6143.140–140
Alkaline phosphataseIU/L1376.291.729.4843.547–275
Alanine aminotransferasebIU/L76.34.263.623.354.820.230–100
Total bilirubinmg/dL0.170.16cccc0.4–1.2
Total proteinmg/dL4.804.–6.4
Blood urea nitrogenmg/dL22.32.322.17.8197.210–30
 White blood cell countK/μL5.–5.7
 Red blood cell countM/μL9.–10.6
 Mean corpuscular volumefL46.20.445.–50.3
 Mean corpuscular hemoglobinpg15.–18.4
 Mean corpuscular hemoglobin concentrationg/dL32.–40.6


This work was supported in part by the Stanford Bio-X Interdisciplinary Initiative Grant, CCNT-TR Project 2 at Stanford, NIH 1R01 HL075803, R01 HL078678, and R21 CA133492. The authors thank AMAG Pharmaceuticals, Inc. for providing Ferumoxtran-10 (Combidex®) and GE Healthcare for scanner support. The authors also thank Jennifer Lyons and Juan M. Santos.