|Home | About | Journals | Submit | Contact Us | Français|
Focused ultrasound has been discovered to locally and reversibly increase permeability of the blood–brain barrier (BBB). However, inappropriate sonication of the BBB may cause complications, such as hemorrhage and brain tissue damage. Tissue damage may be controlled by selecting optimal sonication parameters. In this study, we sought to investigate the feasibility of labeling cells with superparamagnetic iron oxide particles to assess the inflammatory response during focused-ultrasound-induced BBB opening. We show that infiltration of phagocytes does not occur using optimal parameters of sonication. Taken together, the results of our study support the usefulness and safety of focused-ultrasound-induced BBB opening for enhancing drug delivery to the brain. These findings may have implications for the optimization of sonication parameters.
The potential of focused ultrasound to increase the permeability of the blood–brain barrier (BBB) was initially reported more than a decade ago (Mesiwala et al, 2002; Vykhodtseva et al, 1995). If microbubbles are combined with ultrasound exposure, the effects of ultrasound can be focused to reduce the acoustic intensity required to produce BBB opening (Hynynen et al, 2003, 2005, 2006; Kinoshita et al, 2006; McDannold et al, 2005). This can diminish the risk of tissue damage and make the technique more easily applied through the intact skull (Hynynen et al, 2006). Therefore, focused-ultrasound-induced BBB opening is an advantageous technique for drug delivery to the central nervous system (CNS), because the opening is both localized and temporary.
In most previous studies, confirmation of the opening of the BBB was obtained using contrast-enhanced T1-weighted magnetic resonance imaging (MRI). The leakage of gadolinium-based MR contrast agents can be used as an indicator of BBB opening in T1-weighted scans (Hynynen et al, 2001, 2003; McDannold et al, 2005; Runge et al, 1985). In addition, T2*-weighted MRI sequences may be useful for the detection of hemorrhage during focused-ultrasound-induced BBB opening (Liu et al, 2008).
Inappropriate sonication is the main cause of complications encountered while using focused-ultrasound-induced BBB opening. Possible mechanisms of brain damage after application of inappropriate sonication include transcytosis, passage through endothelial cell cytoplasmic openings, opening of tight junctions, and free passage through the injured endothelium. (Sheikov et al, 2004, 2008). The widespread opening of tight junctions after a high-intensity-focused ultrasound may favor the passage of proteins, chemokines, or blood cells between the vessel lumen and the surrounding brain tissue, resulting in brain hemorrhage and tissue damage (Liu et al, 2008). The leakage of proinflammatory molecules and chemokines into the brain milieu may in turn promote macrophage infiltration and homing. McDannold and coworkers initially showed that focused ultrasound exposure for BBB opening may result in erythrocyte extravasations. Occasionally, a mild inflammatory reaction with macrophage infiltration has also been observed (McDannold et al, 2005). However, it is unclear whether activated macrophages originate from circulation or from in situ microglia. Activated macrophages from circulation, in concert with other immune competent cells, infiltrate the CNS in different neurologic disorders (Nilupul Perera et al, 2006; Price et al, 2003; Stoll et al, 1998), and macrophage tracking by MRI has been applied in several diseases afflicting the CNS.
Superparamagnetic iron oxide (SPIO) nanoparticles are used for intracellular magnetic labeling of cells to monitor cell trafficking by MR imaging. Superparamagnetic iron oxide particles exhibit magnetic moments that align in an applied magnetic field, creating extremely large microscopic field gradients around the particles that de-phase the neighboring proton magnetic moments, thereby reducing the T2* relaxation time. As SPIO particles are phagocytosed by activated macrophages (Dousset et al, 1999a), brain SPIO-enhanced MRI might be useful for in vivo monitoring of immune cell migration into the CNS. This technique may thus allow the identification of diagnostic and prognostic markers in vivo. For example, the study of different experimental autoimmune encephalomyelitis models by MRI with SPIO has suggested that disruption of the BBB is the initial key step in the development of inflammatory lesions (Dousset et al, 1999a1999b; Linker et al, 2006; Rausch et al, 2003; Xu et al, 1998). Similarly, MRI studies in animal models of tumor, nerve trauma, brain ischemia, and autoimmune neuritis have shown that macrophages are capable of infiltrating the CNS parenchyma (Bendszus and Stoll, 2003; Kleinschnitz et al, 2003; Nolte et al, 2005; Stoll et al, 2004).
The assessment of macrophage CNS infiltration and activity using MRI with SPIO has been previously reported (Dousset et al, 2006; Rausch et al, 2001; Weissleder et al, 1989). In this study, we sought to investigate (1) the feasibility of labeling cells with SPIO particles for assessing the occurrence of an inflammatory response during focused-ultrasound-induced BBB opening, and (2) whether the optimization of sonication parameters can reduce the occurrence of brain macrophage infiltration during BBB disruption (Figure 1).
All animal experiments were approved by our Institutional Animal Experiment Committee. A total of 16 adult male Sprague–Dawley rats were used in this study. The animals were anesthetized using a mixture of ketamine (40mg/kg) and xylazine (10mg/kg). A craniotomy was performed 3 days before ultrasound sonications (i.e., 2 days before SPIO injections) to provide a path for the ultrasound beam, and a cranial window (~1 × 1cm2) was made using a high-speed drill. Skull defects were covered with saline-soaked gauze to prevent dehydration until focused ultrasound was applied. The dura was kept intact during craniotomy. As cutting was performed using a bone saw along the edge of the opened region, no scab was left in the ultrasound beam passage to avoid wave interference.
Monocytes were labeled using commercial SPIO nanoparticles (Resovist, Schering AG, Berlin, Germany, carboxydextran-coated, hydrodynamic size: 60nm). A total of 11 animals received a dose of 15μmol/kg through the tail vein ~24h before sonication and MRI scans. The remaining five animals did not receive SPIO and served as controls. A summary of experiments is listed in Table 1.
The experimental protocol is depicted in Figure 2A. A focused ultrasound transducer (Imasonics, Besancon, France; diameter: 60mm, radius of curvature: 80mm, frequency: 1.5MHz) was used to generate a beam of ultrasound energy. An arbitrary function generator (33120A, Agilent, Palo Alto, CA, USA; DS345, Stanford Research Systems, Sunnyvale, CA, USA) was used to generate the driving signal, which was then fed into a radio frequency power amplifier (150A100B, Amplifier Research, Souderton, PA, USA). The ultrasound pressure field and peak ultrasonic pressure were measured and calibrated in an acrylic water tank using a polyvinylidene difluoride-type hydrophone (Onda, Sunnyvale, CA, USA; calibration range: 50kHz to 20MHz, diameter: 0.5mm) mounted on an in-house designed three-dimensional positioning system. The half-maximum pressure amplitude diameter and length of the produced focal spot were 2 and 10mm, respectively. Measurements of calibrated pressure magnitudes and focal beam dimensions were both conducted in a degassed water-filled tank.
A different acrylic water tank with a 4 × 4cm2 opening at the bottom was used for animal experiments. To allow the entry of ultrasound, a thin film device was used to seal the opening. Degassed water was filled into the cranial openings to serve as an acoustic coupling device. Animals were placed directly under the water tank with the head tightly attached to the thin-film window.
Ultrasound energy was then delivered into the brain. In the ultrasound exposure experiments, all the brain samples (the only exception being the control tumor group) were sonicated in the presence of an ultrasound microbubble-based contrast agent (sulfur hexafluoride microbubbles 2.0 to 5.0μm in diameter; SonoVue, Bracco, Milan, Italy) injected intravenously before sonications. Each bolus injection contained 0.025mL/kg microbubbles (~0.1mL/kg, normal adult dose) mixed with 0.2mL of saline solution and then flushed by 0.2mL heparin. The center of the focal zone was placed at a 2 to 3mm penetration depth for each hemisphere. Burst-mode ultrasound was used, with a burst length of 10msecs, a pulse-repetition frequency of 1Hz, and a duration of 30secs. To open the BBB either with or without hemorrhagic damage, we used acoustic negative-peak pressure values (spatial peak and temporal peak) of 2.45 and 1.1MPa, respectively (the peak values reported in this study have the possibility of being an underestimated value because of the limited resolution of the hydrophone). To achieve a complete opening of the BBB at a low acoustic pressure (1.1MPa), we performed six different sonications at different points (patterns shown in Figure 2). Microbubbles were injected twice. The first three sonications were conducted consecutively after the first microbubbles injection. The fourth to sixth sonications were conducted after the second microbubbles injection, with at least a 5-min interval between two distinct microbubbles injections. Sonications were specifically delivered consecutively and unilaterally in the frontal brain, with 1-mm gaps between sonication points. The delay between each sonication was <5secs. At high pressure values (2.45MPa), BBB disruption was easily achieved using a single sonication point. Sonications were performed outside the MRI scanning room.
Images were acquired using a 3-T scanner (Trio with Tim, Magnetom, Siemens, Erlangen, Germany) equipped with a homemade surface coil. After sonication, animals placed in an acrylic plastic holder were transferred to the scanning room and then placed inside the bore of the magnet. Contrast-enhanced T1-weighted turbo spin-echo images were obtained as a reference after BBB disruption with the following imaging parameters: TR (repetition time)/TE (echo time)=534/11msec, slice thickness=1.5mm, matrix size=134 × 192, field of view=57 × 83mm2, echo train length=5, and equivalent voxel size=0.4 × 0.4 × 1.5mm3. In all, 10 measurements were acquired in ~5mins. A dose of 0.25mL/kg of gadolinium (Gd-diethylene triamine pentaacetic acid (DTPA), Magnevist, Berlex Laboratories, Wayne, NJ, USA) was administered before the image scan, followed by a saline flush solution (0.2mL/kg) containing 0.2mL/kg heparin.
Heavy T2*-weighted three-dimensional fast low-angle shot sequences were subsequently acquired for detecting longitudinal changes in the brain parenchyma. Imaging parameters were as follows: TR/TE/flip angle=28msec/20msec/15°, slice thickness=0.7mm, number of slices=16, matrix size=256 × 384, and field of view=80 × 130mm2, producing a resultant voxel size of 0.3 × 0.3 × 0.7mm3, without leaving spaces between adjacent image slices (distance factor=0%). The image region of interest covered the entire animal brain to achieve a three-dimensional brain reconstruction, as well as for tracking the SPIO-laden macrophages. A total of 5 measurements were acquired in ~7mins. Animals were continuously imaged within 3h (at 5-min intervals), as well as at 4 and 24h. A series of T2* images were acquired at the following time points: (1) 10mins before sonication, (2) continuously starting from 10mins after sonication to 3h after sonication at 10-min intervals, (3) 4h after sonication, and (4) 24h after sonication.
To longitudinally monitor slice-based signal intensity (SI) changes from T2* images, the images were first realigned in Matlab (Mathworks, Natric, MA, USA) using a homemade registration algorithm (Dhawan, 2003). Thereafter, regions of interest of 3 × 3 voxels from a single image slide were selected and mean SI values were calculated. Moreover, the volume of hypointense regions in T2* images was calculated using a simple thresholding algorithm (Dhawan, 2003). First, images were linearly interpolated from consecutive slices to a voxel size of 0.3 × 0.3 × 0.35mm3 to reduce anisotropy bias in slice direction (i.e., analyzed slices: 31). Images were normalized to the maximum signal intensity of the tissue parenchyma. A binary mask was applied with a threshold of 0.5. Volume was calculated as the sum of pixels in the region of interest multiplied by the known voxel size. Increases in volume were compared in 2.45MPa sonications either with or without the injection of SPIO particles. The differences in volume increase (observed between 0.5 and 24h after sonication) were compared using unpaired two-tailed Student's t-tests.
Tissue preparation was performed after in vivo MR analysis. Histopathology was performed on 10-μm sections from paraformaldehyde-fixed, paraffin-embedded brain samples. Macrophage infiltration in the brain parenchyma was visualized using the FITC (fluorescein isothiocyanate)-conjugated CD11b mAb (M1/70.15) as described previously (Maruyama et al, 2005). For immunocytochemistry, CD11b+peritoneal exudates cells (PECs) in tissue culture dishes were rinsed with phosphate-buffered saline, fixed in 4% paraformaldehyde for 7mins, rinsed with phosphate-buffered saline, blocked in 2% bovine serum albumin containing 0.1% Triton-X-phosphate-buffered saline, and stained with lymphatic vessel endothelial HA receptor (LYVE)-1 (1:500), podoplanin (1:500), or Prox-1 (1:500). After rinsing and blocking, the samples were subsequently stained with FITC-conjugated CD11b (1:250) secondary antibodies, washed with phosphate-buffered saline, and covered with a Vectorshield with DAPI (4'-6-Diamidino-2-phenylindole) applied for fluorescence. Microscopy was performed using a Zeiss Axiophot microscope (Carl Zeiss Vision, Oberkochen, Germany).
Prussian blue staining (ferric hexacyanoferrate and hydrochloric acid, Sigma, St Louis, MO, USA) was performed to detect focal iron accumulation. Slides were placed in a staining jar containing a hydrochloric acid–potassium ferrocyanide solution for 30mins at room temperature. The slides were rinsed in distilled water and were counterstained by nuclear fast red for 5mins. Microscopy was performed using a Zeiss Axioplan 2 imaging microscope with AxioVision 4.1 imaging software, AxioCam HRc camera, and Zeiss objectives Fluar 10 × /0.50, Plan-Apochrome 20 × /0.75, and Plan-Neofluar 100 × /1.30 oil (Carl Zeiss Vision).
As shown in Figure 2B, BBB disruption was obtained using either multiple sonications at an acoustic pressure of 1.1MPa (upper panel) or a single sonication at the high acoustic pressure of 2.45MPa (lower panel). Repeated sonications at 1.1MPa resulted in a larger area of BBB opening compared with that obtained using a single high-pressure sonication (2mm in diameter at the focal depth). Only few areas of erythrocyte extravasations were evident in regions of BBB opening after sonications at 1.1MPa. Conversely, a single 2.45-MPa sonication produced a severe intracerebral hemorrhage (red arrow) within the area of BBB disruption (white arrow).
Figure 3 shows MRI images of control animals and of those receiving SPIO. Gadolinium-enhanced T1 images showed an increase in hyperintense signal at two sites of sonication (left panel: six sonications at 1.1MPa; right panel: single sonication at 2.45MPa). Disruption of the BBB was confirmed in both cases. Imaging was continued until 24h after sonication (i.e., 48h after SPIO injection in SPIO-treated rats). The T2* image obtained immediately after the T1 image showed hypointensity at the right sonication site. This reflects the presence of an intracerebral hemorrhage due to the excessive acoustic pressure applied during sonication. Notably, the application of six sonications at a pressure of 1.1MPa did not result in the occurrence of intracerebral hemorrhage. It indicates that this acoustic pressure was relatively safe and below the threshold for hemorrhage.
Subsequent T2* images obtained at 4 and 24h showed different patterns. The area of hypointensity was specifically relatively constant in the control group, whereas it was significantly expanded in SPIO-treated rats receiving 2.45MPa sonications. Notably, the signal loss on T2* images caused by intracerebral hemorrhage in animals sonicated at 2.45MPa showed a consistent hypointense pattern through 24h.
Figure 4 shows the reconstructed hypointense volume in the same animals used for the experiments reported in Figure 2. The hypointense volume in SPIO-negative rats was found to be rather unchanged (ranging from 0.61 to 0.76mm3), whereas an apparent increase was evident in SPIO-treated animals (from 0.19mm3 at 0.5h to 1.87mm3 at 24h, a nearly nine-fold increase in this case). Figure 5 shows the comparisons of the increases in hypointense volume observed in consecutive T2* images in the control (2.45MPa) and SPIO-laden (2.45MPa) groups. The control group showed a homogenous and slight decrease (−10.8%±7.3%), whereas the SPIO-laden group showed a heterogeneously increased volume (320.9%±246.6%). The differences in volume increase between the two groups were statistically significant (P=0.001).
The occurrence of an inflammatory response in areas of BBB opening is also evident in T2* images obtained at 3h. Figure 4 shows the analysis of MRI SI time course in SPIO-treated animals. Five selected regions are zoomed and depicted (Figure 6A), including blood vessels neighboring the brain (regions 1 and 2), brain areas sonicated at 1.1 and 2.45MPa (regions 3 and 5), and midbrain artery (region 4). Ultrasonic energy was delivered at 10mins after the initiation of T2* scans (arrow). Five selected time points are shown in Figure 6B, and the corresponding time changes in SI were recorded accordingly (Figure 6C). As shown in Figure 6, multiple sonications at 1.1MPa (region 3) did not result in the occurrence of brain hemorrhagic damage, and no apparent signal change was evident. In addition, there was no apparent SPIO-laden monocyte infiltration in these areas. Conversely, a single sonication at 2.45MPa induced the formation of intracerebral hemorrhage that, in turn, resulted in an immediate 10% signal decrease, followed by another gradual decrease. The total decrease in SI at the end of imaging was 15% at the end of 3h. These findings suggest a continuous infiltration of SPIO-labeled macrophages at sites sonicated at 2.45MPa (Figure 3). It is interesting that an ~30% SI decrease was evident in selected neighboring blood vessels. This decrease peaked at ~60mins and subsequently increased to baseline values. A similar pattern of SI decrease was detected in the midbrain artery (region 4), although to a lesser extent (~12% from baseline values). The pattern in SI changes in circulatory areas (regions 1, 2, and 4) was characterized by a quick SI rebound. This suggests that SPIO-laden macrophages did not stall in these areas. This pattern was therefore different from that observed in areas of suspected inflammation (region 5). Accordingly, inflammatory areas were characterized by a gradual decrease in SI because of infiltration and accumulation of SPIO-laden macrophages.
Superparamagnetic iron oxide-laden macrophages were also confirmed histologically by FITC-conjugated CD-11b immunostaining, as well as by Prussian blue iron staining. Figure 7 compares different focused ultrasound sonication protocols. Specifically, multiple sonications at 1.1MPa (upper panel), and a single sonication at 2.45Mpa either without (middle panel) or with (lower panel) an SPIO injection, are depicted. As expected, repeated sonications at 1.1MPa did not elicit an inflammatory response at irradiated sites. Accordingly, no CD-11b-positive cells were evident. Although some fluorescent-positive cells were evident, they were likely to belong to local microglia. In contrast, areas sonicated at 2.45MPa showed the presence of CD11b-positive cells (panels E and H), suggesting that sonication with high acoustic pressures was associated with significant macrophage infiltration in the brain parenchyma. Iron deposits as assessed by Prussian blue staining were evident in rats treated with SPIO particles only (panel I versus panel G).
In this study, we showed that the brain inflammatory response that accompanies ultrasound-induced BBB disruption can be detected using the cell-labeling MRI technique. Our findings may have implications for the optimization of ultrasound parameters.
The phagocytic capacity of activated macrophages and their ability to migrate to inflammation sites have provided the rationale for in vivo labeling and monitoring of these cells. The endocytotic loading of circulating macrophages with SPIO has been used in this study to monitor inflammatory response after ultrasound-induced BBB disruption in vivo using MRI. The results of our study indicate that a proper selection of sonication parameters can reduce the occurrence of brain macrophage infiltration produced by microbubble-enhanced focused ultrasound stimulation. Our data suggest that hemorrhagic brain damage occurs with the use of high-pressure sonications, resulting in phagocyte infiltration into the brain parenchyma. Intriguingly, the occurrence of an inflammatory response may be avoided using proper sonication parameters.
In this study, we showed that superparamagnetic iron oxide MRI can be used successfully to monitor macrophage infiltration into the brain parenchyma after BBB opening by focused ultrasound. The SPIO-laden activity of infiltrating macrophages was characterized by a temporal change in the hypointense pattern detectable both in the damaged brain and in blood vessels. This activity, however, did not originate from signal changes caused by nontargeted SPIO nanoparticles.
Previous studies have consistently shown that the blood pool half-life of SPIO lasts from several minutes (Weissleder et al, 1989) to a maximum of 30mins (Kato et al, 1999). After injection of SPIO, the pattern of SI change is characterized by a sharp decrease immediately after SPIO injection, followed by a monotonic increase, and a return to baseline values. In our experiments, SPIO particles were injected into animals 24h before sonication when no brain damage had occurred. Therefore, the majority of SPIO that was not phagocytosed by macrophages should have been removed from systemic circulation when our sonication experiments were carried out. In addition, we consistently showed that changes in the hypointense pattern were always detectable after 2.45-MPa sonication, but not after 1.1-MPa sonication (data not shown). It is thus unlikely that the temporal signal decrease detected in cerebral blood vessels (Figure 6) could be because of the presence of free SPIO particles in this circulatory territory or because of the effect of gadolinium injection on T2*-weighted image. Moreover, as SPIO was injected before BBB opening, the possibility that changes in hypointensity may be due to the leakage of free-SPIO particles through the disrupted BBB should be excluded. Therefore, we hypothesize that the SI change observed in circulation could be attributed to a transient response of SPIO-laden macrophages recruited into the intracerebral hemorrhage because of high-dose sonication (see regions 1 and 2 in Figure 6).
The time lapse between intravenous injection and scanning is an important parameter that should be carefully established. This time frame should exceed the blood phase of free contrast agent molecules and should be sufficiently long to allow the accumulation of iron particles within macrophage phagosomes. In this regard, it has been previously shown that all particles had accumulated in macrophages 24h after intravenous injection, thereby ensuring an optimal use of the macrophage delivery system (Kleinschnitz et al, 2003; Weissleder et al, 1990).
Inappropriate delivery of ultrasonic energy may have undesired side effects. In this regard, several potential complications associated with focused ultrasound-induced BBB opening have been described, including erythrocyte extravasation, microhemorrhage, physical damage of endothelial cells, neuronal damage (hyperchromatic cells, vacuolation), necrotic neuronal death due to excessive acoustic pressure, glial cell neuron pyknosis, or neuron cell apoptosis (Hynynen et al, 2005, 2006; Treat et al, 2007). Thus far, however, it is still unclear whether such complications are accompanied by an inflammatory response. In this study, we showed for the first time that a significant cellular inflammatory reaction is associated with the occurrence of intracerebral hemorrhage. It is noteworthy that the acoustic pressures used to induce intact BBB disruption or the BBB disruption associated with intracerebral hemorrhage used in this study is in keeping with those previously reported when focused ultrasound was used at a similar frequency (Hynynen et al, 2001). Such a response, however, may be the consequence of cerebral hemorrhage and is unlikely to be directly triggered by specific sonication conditions.
In recent years, macrophage tracking by MRI with SPIO has been repeatedly used in numerous animal models of CNS diseases. In a rat model of permanent middle cerebral artery occlusion, SPIO was injected intravenously into animals at 5h, and at days 1, 2, 4, and 7. On T2-weighted images, patch areas of signal loss were evident in infarcted areas until day 4, and were found to be decreased thereafter (Rausch et al, 2001). In another model of ischemia (photothrombosis model), the authors reported that in vivo labeling of monocytes with SPIO can be confirmed by spatial agreement with CD11b+ areas, as well as by the incorporation of SPIO nanoparticles into infiltrated monocytes (Saleh et al, 2004; Schroeter et al, 2004). It is noteworthy that the time course of contrast enhancement has been described to correspond to the time window of monocyte influx (Kleinschnitz et al, 2003). In addition, the major wave of macrophage activity was delayed by 5 to 6 days after the induction of cerebral infarcts. Altogether, these findings suggest that the recruitment of hematogenous macrophages occurs similarly in a narrow time interval between days 3 and 6 after the induction of cerebral ischemia.
In this study, we provided evidence that SPIO-laden monocyte activity was evident at the hemorrhagic brain site as early as 4h after sonication and can be detected early (within 1h) in the bloodstream. It subsequently increased gradually until 24h, thereby suggesting that monocyte activity was instantly activated by sonication. This phenomenon could be explained by the fact that focused ultrasound-induced damage is capable of inducing hemorrhagic stroke, but not the ischemic subtype. It is noteworthy that the time course of the inflammatory response in our study was similar to that reported in a previous investigation on intracerebral hemorrhage (Del Bigio et al, 1999). Accordingly, these authors showed that intracerebral hemorrhage is accompanied by a significant inflammatory response characterized by immune cell infiltration within 12h of hematoma onset, with a more marked infiltration of macrophages 1 to 2 days thereafter (Del Bigio et al, 1999).
The CNS is characterized by two major monocyte-related populations, i.e., highly ramified resident microglia and hematopoietic perivascular macrophages It is now acknowledged that the CNS does exhibit intense macrophage activity in the early phases of numerous pathologic conditions, including inflammation, demyelination, trauma, or infections (Flaris et al, 1993). Resting microglia in the CNS parenchyma are ramified and easily discernible from macrophages. Upon activation, however, microglia undergo morphologic changes and become indistinguishable from hematogenous macrophages on morphologic grounds. The SPIO labeling of resident microglia cells should be ruled out in our study, as the BBB was still intact when SPIO was administered intravenously. We conclude that the pattern of hypointensity observed in longitudinal T2* MRI studies was clearly because of SPIO-laden monocytes and monocyte-derived macrophages.
Professor ZH Cho (Neuroscience Research Institute, South Korea) is deeply acknowledged for his support on MRI coil design.
Conflict of interest
The authors declare no conflict of interest.