Multiple sessions of liposomal doxorubicin delivery via focused ultrasound mediated blood-brain barrier disruption: a safety study

Sonication System

An air-backed, single element, 690 kHz focused piezoelectric transducer (diameter/radius of curvature: 100/80 mm) generated the ultrasound field. It was driven by an arbitrary waveform generator (model 395, Wavetek) and RF amplifier (240L, ENI); electric power was measured with a power meter (E4419B, Agilent,) and dual-directional coupler (C5948-10, Werlatone). Reported exposure levels are absolute peak negative pressure amplitudes measured in water with a membrane hydrophone (Marconi; 0.5 mm diameter). Attenuation by the brain and rat skull is expected to reduce the pressure amplitude by ~30% at this frequency [48] with additional uncertainty arising from standing waves within the skull and increases in skull thickness as the animal ages [48]. The pressure distribution of the transducer was mapped using a 0.2 mm needle hydrophone (Onda, Sunnyvale, CA); its half-maximum diameter and length were 2.3 and 12 mm, respectively. The transducer efficiency was measured using a radiation force-balance.

Acoustic parameters were the same as in our previous study [42]. The sonications consisted of 10 ms bursts applied at a frequency 1 Hz for 60s at a pressure amplitude of 0.55 MPa. This pressure amplitude was initially set based on a prior study in rats with this device [17] and was increased on the basis of the animal age and weight to achieve a consistent level of BBB disruption. This observation was made in our initial treatments and is similar to previous reports [48]. Each sonication was combined with an intravenous injection of a microbubble-based ultrasound contrast agent (Definity; Lantheus) administered at the dose recommended for human ultrasound imaging (10 µl/kg). Each milliliter of Definity contains 1.2×10¹⁰ microbubbles that consist of perfluorocarbon gas-filled lipid shells with a mean diameter of 1.1–3.3µm. To facilitate the injections of such a small volume, the agent was diluted in PBS to 0.1 times its normal concentration. It was injected as a bolus approximately 9 s before each sonication, followed by a 0.2 ml saline flush.

Experimental Setup

The sonication system was operated within a clinical 3T MRI scanner (Signa; GE Healthcare). The transducer was immersed in a small tank of degassed, deionized water and attached to an MRI-compatible, manually-operated positioning system (Figure 1). The animal was laid supine on a tray above this tank, with a water bag providing an acoustic path to the dorsal surface of the head. Images were obtained with a 7.5 cm-diameter transmit/receive MRI surface coil. The animal’s body temperature was maintained with a heated water pad. Before the rat experiments, we visualized heating in a silicone phantom using temperature-sensitive MRI to localize the acoustic focal point in the MRI space. Accurate targeting in vivo was confirmed before the sonications in select animals (typically the first animal sonicated each week) by verifying that the resulting MRI contrast extravasation appeared at the desired target after one sonication.

Animals

All animal experiments were approved by the Institutional Animal Care and Use Committees of Harvard Medical School. Tests were performed in 18 Male Sprague-Dawley rats (Charles River Laboratories; ~250 g) (Table 1). Before each procedure, the animals were anesthetized via intraperitoneal injections of ketamine (80 ml/kg/h) and xylazine (10 ml/kg/h). A catheter was placed in the tail vein, and the hair on the scalp was removed with clippers and depilatory cream. The body temperature was maintained throughout the experiment with a heated water pad. To avoid skin infections that were observed in animals who received Lipo- DOX in a previous study [42], animals receiving three weekly sessions with FUS-BBBD and chemotherapy were treated prophylactically with an antibiotic (Baytril^®, Bayer; 2.5mg/kg).

Optimization of Lipo-DOX delivery

Nine rats were used to verify that clinically-relevant doxorubicin concentrations could be achieved in the brain with FUS-BBBD and the administration of Lipo-DOX. We aimed to achieve a doxorubicin concentration in the brain of at least 0.819 µg/g, which was reported to correlate with a clinical response rate in human tumors [49]. After determining the coordinates of the focal point within the MRI space, treatment planning MRI was acquired, and the focal region was positioned at 2 mm lateral to the midline and 4 mm deep from the dorsal brain surface. The sonicated volumes were centered 2.5 mm anterior to the bregma in the striatum. The striatum was of interest to us because that was the location of the implanted 9L-rat glioma tumors in our previous study [42]. Sonications were performed in a 3×3 grid pattern with 1–1.5 mm spacing. Doxorubicin hydrochloride encapsulated in long-circulating pegylated liposomes (Avanti Polar Lipids, INC), was administered intravenously at a total dosage of 5.67 mg/kg. That dosage was selected on the basis of prior work testing this agent in rats [50].

Lipo-DOX was injected either before (n=5; group A) or 10 min after (n=4; group B) the sonications. In group A, both hemispheres were sonicated. In the right hemisphere, Lipo-DOX was administered in 9 fractions, one before each sonication, mimicking the scheme from the previous rat tumor studies [42,44]. Since Lipo-DOX has a long plasma half-life (2–3 days) [51], we assumed the entire dose was still present in the circulation, and we sonicated the left hemisphere without administering more drug. In group B, the sonications were performed only in one hemisphere and the total dose of Lipo-DOX was injected as a bolus 10 min later. The contralateral hemisphere served as a control. The sonication time was approximately 30 min for the 9 targets. Trypan blue (ICN Biomedical, Aurora, OH; 80 mg/kg) was administered to the animals in Group A and group B through the tail vein after the completion of the sonications and MR imaging. This dye was administered to confirm successful BBB disruption and to mark the sonication targets post mortem for tissue harvesting. Four hours after sonication, the rats were sacrificed, samples of sonicated and control volumes (~30 mg) were collected from the brain, and doxorubicin concentrations were measured with a fluorometer.

Extraction and quantification of doxorubicin

Four hours after the last sonication, each animal was put into a state of deep anesthesia with an overdose of ketamine and xylazine. To flush unabsorbed Lipo-DOX from the cerebral vasculature, the brain was perfused by transcardiac method with normal saline. The brain was cut into 2–3 mm slabs using a brain matrix (ASI-instruments; Warren, MI). Right and left regions of BBB disruption, identifiable by the trypan blue staining, were harvested along with non-sonicated control regions. The doxorubicin concentrations were quantified by linear regression and a standard curve derived from eight serial concentrations. The concentration of doxorubicin in each tissue sample was determined by taking the average of at least three fluorometric readings on a benchtop fluorometer (VersaFluor; Bio-Red Laboratories, Hercules, CA). Measurements were normalized to those obtained from a control animal that did not receive doxorubicin to correct for autofluorescence.

A preparation of acidified alcohol (0.3 N HCl in 50% EtOH) was used to extract doxorubicin from the harvested tissue samples for fluorometric quantification. We assumed a tissue density of 1 g/cm³, so the mass of each sample was assumed to be equal to its volume. Each sample was measured and cut down until its mass was approximately 0.03 g. Each sample was put into a 1.5-mL centrifuge tube with 20 volumes (~600 µL) of acidified alcohol, then lysed with a tissue homogenizer (Bullet Blender; Next Advance, Averill Park, NY) and refrigerate at 4°C for 24 h. Samples were then centrifuged at 4,000 × g for 20 min at 4 deg C. The supernatant was extracted for immediate fluorescence intensity measurement (Excitation/Emission wavelengths: 480/590 nm).

Three weekly sessions

Nine rats were assigned to one of two groups: (1) three weekly sessions with FUS-BBBD and concurrent chemotherapy (FUS+DOX) (N=5), (2) three weekly sessions with FUS-BBBD only (FUS-Only) (N = 4). The sonication parameters and brain regions selected for FUS-BBBD were the same as described above. Sonications were performed in a grid pattern at 5, 9, and 12 targets, respectively, which mimicked the three weekly treatments we administered in our previous study in tumor-bearing rats [42]. The spacing between sonications was 1–1.5 mm. Lipo-DOX was administered intravenously in equal fractions over multiple slow injections just before each sonication. The total Lipo-DOX dose was 5.67 mg/kg in each FUS-BBBD session.

Histology

All of the animals from the FUS-Only and FUS+DOX groups were sacrificed 70 days after the first session to evaluate the histological effects. The animals were deeply anesthetized with ketamine/xylazine, sacrificed, and the brains fixed via transcardial perfusion (0.9% NaCl, 100 mL; 10% buffered formalin phosphate, 250 mL). The brains were then removed, embedded in paraffin, and serially sectioned at 5 µm in the axial plane (perpendicular to the direction of ultrasound beam propagation). Every 50^th section (250 µm apart) was stained with hematoxylin and eosin (H&E). In each animal, the sections with the greatest tissue effects were identified and examined in details under high magnification; the dimensions of any damaged areas were measured.

Magnetic resonance imaging

The experiments were performed under MRI guidance. T2-weighted imaging was used to plan the treatments. BBB disruption was evaluated using T1-weighted imaging acquired before and after administration of the MRI contrast agent, gadopentetate dimeglumine (Magnevist Gd-DTPA; Bayer Healthcare; 0.25 mL/kg). T2*-weighted imaging was used to examine whether petechiae, which are produced by excessive FUS exposures [10], did or did not occur. Detailed imaging parameters are listed in Table 2. For the experiments that examined three weekly sessions of FUS-BBBD and Lipo-DOX, the health of the nine animals was monitored regularly, and MRI was acquired 53 (n=1) or 67 (n=8) days after the first session to evaluate the treatment effects.

Statistical Analysis

The mean doxorubicin concentration in the brain and the standard deviation were calculated for the non-sonicated hemispheres and for the cases where the Lipo-DOX was administered before, during, and 10 minutes after the sonications. The concentrations among the different experimental groups were compared using two-tailed paired student’s t-tests. Values of p < 0.05 were considered statistically significant.

BBB permeabilization

BBB permeabilization and the presence or lack of petechiae were confirmed using contrast-enhanced and T2*-weighted MRI, respectively (Figure 2). A higher pressure amplitude (0.81 vs. 0.55 MPa) was needed for the third session to induce consistent BBB permeabilization, presumably because of an increase in skull or dura thickness as the rats grew [48]. All animals demonstrated hyperintense regions in contrast enhanced T1-weighted images after each session, confirming BBB permeabilization. In several cases, one or more hypointense spots were observed in T2*–weighted MRI in the focal plane or at the brain surface in the cortex, demonstrating the presence of petechiae. Such spots were observed in both hemispheres near the brain surface in 4 of the 5 animals in the animals where doxorubicin concentrations were measured and that received Lipo-DOX before or during after sonication. They were not observed in animals who received Lipo-DOX after sonication. In the animals where histological effects were examined, such spots were observed after the second or third session in 4 of the 5 animals who received both FUS-BBBD and Lipo-DOX and in 3 of the 4 animals who received FUS-BBBD alone. They were not seen in MRI at day 53 or 67. In addition to these hypointense spots in the focal plane, hypointense spots were observed in the cortex at the brain surface in every animal except one who received FUS-BBBD alone. In addition to the examples shown Fig. 2, additional MRI findings are shown below along with histology.

Doxorubicin concentrations

The doxorubicin concentration in the brain tissue was compared between the experimental groups in which Lipo-DOX was injected before, during, or after the sonications (Figure 3). In group A, both hemispheres were sonicated in each animal. In the right hemisphere, Lipo-DOX was administered in 9 fractions, one fraction before each sonication. In the left hemisphere, the sonications were performed without additional Lipo-DOX administration. In group B, the sonications were performed only in one hemisphere, and the total dose of Lipo-DOX was administered in a single, slow injection 10 minutes later. The contralateral hemisphere in these animals served as controls. The total sonication time for 9 targets was approximately 30 min.

In group A, the mean doxorubicin concentrations in the right and left hemispheres were 4.7 ± 0.5 and 4.8 ± 0.5 µg/g, respectively; they were not significantly different (p > 0.05). In group B, the doxorubicin concentration was significantly less (p < 0.001) than in both hemispheres in group A. The mean concentration in these animals was 3.2 ± 0.3 µg/g, a 38% less compared to the volumes in group A. The doxorubicin concentrations were significantly enhanced in all cases (p < 0.0001) compared to the control tissue. In addition, in every case the doxorubicin concentrations in the sonicated areas exceeded our goal of 0.819±0.482 µg/g, a value reported to correlate with a clinical response for human tumors [49].

Histological findings

All animals that received FUS-BBBD alone were found to be unaffected in the targeted plane in the striatum (Figure 4). However, four of the five rats who received FUS-BBBD and Lipo-DOX had small regions with evidence of prior damage in the focal plane (Figure 5). In three of these animals, this was evidenced by the presence of small scars with largest dimensions of 0.5–2.0 mm. These scars consisted of infiltrating macrophages and activated astroglial cells. In the fourth animal, a small (1.0×2.0 mm) cyst formation was observed. No changes were observed in the control hemisphere that received Lipo-DOX only (black boxes in Figure 5).

Evidence of more severe prior damage was observed in the cortex near the brain surface in four of the five animals who received FUS-BBBD and Lipo-DOX and in one of the four animals who received FUS-BBBD alone. The effects were similar in appearance to those in the striatum, but the dimensions of the affected areas, which ranged from 0.5–2.5 mm, were larger. Examples of these effects are shown in Figure 6.

Conclusions

This work demonstrates that FUS-BBBD combined with intravenous administration of Lipo-DOX can deliver clinically-relevant concentrations of doxorubicin to the normal rat brain. Higher concentrations were achieved when we injected the agent either before or during the sonications rather than afterwards. After multiple sessions with FUS-BBBD and Lipo-DOX, the neurotoxicity that was observed in the normal brain in an earlier tumor study with FUS-BBBD and in other works evaluating other doxorubicin delivery methods was not observed here. However, evidence of small areas of prior damage was observed in the targeted area that was not observed with FUS-BBBD or Lipo-DOX administration alone. This damage may have been related to capillary damage produced during the sonications at the exposure levels used here. Overall, this work suggests that FUS-BBBD can be used to deliver Lipo-DOX to the normal brain at the tumor margin with only minor damage. This finding is encouraging for the treatment of the normal brain tissue surrounding an infiltrative tumor such as a glioma. However, due to the presence of small areas of evident prior necrosis that were observed, these results suggest that it would be prudent to use a lower pressure amplitude when sonicating the margins to avoid inertial cavitation.