Magnetic resonance-guided focused ultrasound (MRgFUS) has been shown to reversibly open the blood-brain barrier (BBB) for targeted drug delivery (1). Research on animal models, including non-human primates (2), has been conducted to investigate the effectiveness and characteristics of BBB openings. Here we describe our initial experience in a pilot clinical study to establish the feasibility, safety and preliminary efficacy of focused ultrasound to temporarily open the BBB to deliver chemotherapy to brain tumors.This phase-one clinical trial of BBB opening by focused ultrasound was approved by Health Canada. A modified clinical MRgFUS brain system (ExAblate 4000, 230 kHz, Insightec, Tirat Carmel, Israel) was used with a 3T MR scanner (Signa MR750, GE Healthcare, Milwaukee, WI, USA). Two hours before the procedure, liposomal doxorubicin Caelyx (Janssen, Toronto, Canada) was intravenously infused over 1 hour at a dose of 30 mg/m2. The patient's head was then shaved and positioned in the FUS array with a stereotactic frame. Two targets close to the posterior margin of the glial tumor were chosen based on T2 images (Fig.1). Each target consisted of a 3x3 grid of 9 spots at 3 mm spacing. For each spot, 2.6 ms on, 30.4 ms off FUS pulses were repeated for 300ms before steering to the next spot. The pattern was repeated periodically resulting in an overall pulse repetition frequency (PRF) for each spot of 0.9%. A bolus injection of 4 ul/kg of Definity microbubbles (Lantheus Medical Imaging, N. Billerica, MA, USA) was applied simultaneously with each sonication (1/5th of the clinical dose for ultrasound imaging). With the first injection of microbubbles, 10s short sonications at 5W, 7W and 9W acoustic power were applied to find the appropriate power level based on feedback of cavitation signals. Cavitation signals were detected by two receivers and sampled at a rate of 2 MHz. Spectrum integration from 75 kHz to 155 kHz was calculated and two threshold levels of the spectrum integration were defined as a safety mechanism based on pre-clinical studies on a trans-human skull pig model (3). 9W was found to be adequate for these targets. 50 s sonications at 9W were then applied at each target, with a separate bolus injection of microbubbles for each. Post sonication, Gd (Gadovist, Bayer)-enhanced 3D FSPGR images were acquired to verify the BBB openings, and T2*-weighted GRE images (TE=15ms) were collected to detect potential hemorrhage. After the treatment, the patient was released from the head frame and MR scans were repeated with an 8-channel head coil for better quality images. The patient underwent routine tumor resection the next day and tissue samples at the two BBB opening targets were collected for quantification of chemotherapy drug concentration.BBB opening was successfully achieved at both locations with clear Gd enhancement of the 3x3 grid patterns (Fig.2). Despite using the same power level, the actual acoustic pressure at the 2nd target was lower than the first due to steering of the FUS beam. Small dark signals within individual sonicated spots in the T2* image (Fig.3) indicated low-level extravasation of red blood cells. The quantification of drug concentration is pending further analysis.

The 3mm spacing of the 9 spots was intentionally designed to form a grid pattern of Gd enhancement for easier confirmation in heterogeneous tumors for the initial cases. We do not expect an impact on other parameters if the spacing needs to be reduced for an more uniform drug distribution within the BBB opening volume. The level of extravastion of RBC was low and not a concern in the tumor environment. Our animal experiments have shown that the cavitation signal can be used during the sonications to control the power level for eliminating the RBC exravasations (4). The current system did not use this method during sonications.

The tumor in this patient was in the right temporal lobe adjacent to the skull. The two targets were ~4 cm lateral from the midline of the brain, and the 2nd target was also ~2.5 cm posterior. Thermal ablations by FUS at these off-centre locations are technically challenging due to excessive skull heating. However, successful BBB openings at these locations were demonstrated at low powers at 230 kHz. If these results can be repeated in other patients without complications, then the method may provide a new way to deliver therapeutic agents into brain for the treatment of tumors and other brain diseases.