O’Reilly et al—Blood-Brain Barrier Closure Time After Ultrasound-Induced Opening

Materials and Methods

Results

Blood-brain barrier opening was achieved at all targeted sites, although 1 multi-point sonication was performed twice after a technical error prevented BBB opening from being initially achieved. T1-weighted imaging revealed that 9 of 10 hemispheres were fully closed by 6 hours after focused US (Figure 2). The remaining hemisphere was fully closed by 24 hours.

The pressure and enhancement data are illustrated in Figure 3. An unpaired t test showed no difference in the peak pressure achieved during the sonications in the single- versus multi-point hemispheres (0.29 ± 0.03 versus 0.30 ± 0.04 MPa; P = .69). In 1 sonication, a software crash at the termination of the sonication prevented the treatment pressures from being recorded, and these data were thus not included in the pressure analysis. Similarly, there was no difference in the mean enhancement observed between hemispheres at time 0 (29.7% ± 18.4% versus 29.7% ± 24.1%; paired t test: P = .9975).

However, the time 0 cross-sectional area of the multi-point opening was found to be approximately 3.5 times larger than that of the single point (14.2 ± 4.7 versus 4.1 ± 3.3 mm²; paired t test: P < .0001). Example spatial profiles of the opened regions are shown in Figure 4. The multi-point profile is slightly asymmetric because the exposures were independently controlled at each of the constituent targets.

At 6 hours, the enhancement in the animal with residual BBB opening was 12% on the slice where opening could be detected and had an opening cross-sectional area of 0.8 mm². Follow-up T2- and T2*-weighted imaging showed no indicators of edema or hemorrhage in any of the animals, including the animal with residual BBB opening at 6 hours after focused US (Figure 5). However, histologic analysis revealed changes in the animal in which the opening persisted past 6 hours (Figure 6). A small ischemic lesion was observed in this animal. This region corresponded with the location of the T1 signal seen at 6 hours. This effect was not observed on adjacent sections (±300 µm). Analysis of the hydrophone data for this sonication showed a small ultraharmonic signal at 29 seconds that was below the trigger threshold of the control algorithm, followed by a larger triggered event at 33 seconds (Figure 6). This target had the highest T1 enhancement of any of the sonications at time 0 (86%) but below-average peak pressure (0.28 MPa). No abnormalities were seen in the surrounding region, where closure occurred within 6 hours, in the contralateral hemisphere, or in the other processed brain. Extravasated red blood cells were not observed in either brain.

Discussion

In this study, we sought to examine the effect of the volume of opening on the time for the BBB to close. Our results suggest that the time for the BBB to close after the sonication is independent of the opening volume when the pressure and volume are decoupled. Previous studies in which a larger opening volume was associated with higher pressures have shown the BBB being restored beginning at the periphery of the disrupted zone.²³ We have overcome this factor by using multiple overlapping foci to flatten the exposure profile. Taken in context with prior studies on BBB closure,^21,23,24,28 this approach supports variations in closing time, for a given tracer or therapeutic agent, as related to the exposure level and degree of effect on the tissue.

In this study, we used ultraharmonic emissions to control the treatment exposures. As the driving pressure is increased, microbubbles become more nonlinear, first emitting harmonic signals, followed by subharmonic and ultraharmonic signals at higher exposures, and finally broadband emissions.³¹ Others have correlated an increase in harmonic emissions, signifying stable cavitation, with successful BBB opening^32,33 and the presence of broadband emissions, indicating inertial cavitation, with damage.³² Although harmonic signals can arise from tissue nonlinearities, subharmonic and ultraharmonic emissions are bubble specific, and as they occur at pressures below the inertial cavitation threshold, we use them to mark an upper safety threshold. In this specific implementation, we have used the first 2 ultraharmonics instead of the subharmonic because of the susceptibility of the subharmonic band to equipment noise in our particular setup. The type of tissue damage observed in the 1 case in this study was not seen in our prior work using the real-time treatment controller,³⁰ although we previously observed some red blood cell extravasation at the 2-hour time point, which was not seen here. The lesion was small (diameter, ≈0.4 mm) and confined to a single histologic slice (slice thickness, 5 µm; distance between slices, 300 µm), and if present, such an effect may have been too small for detection in earlier studies. This effect might also have been missed in the prior study because of the differences in the time points for tissue collection (24 hours versus 2 hours or 7 days). However, small differences also exist in the controller scheme, which could be contributing factors. For example, in the prior work, single-point sonications were used with a 2-Hz pulse repetition frequency. Here we have used multi-point interleaved sonications and thus reduced the per-point pulse repetition frequency to 1 Hz to allow our treatment system to cycle through all of the points within the repetition time. With this approach, the pressure increment between bursts doubled from our prior study, and potentially, this step size is too large. In the hydrophone data, we also observed that during this sonication, ultraharmonic signal components were present but did not satisfy the threshold criteria to register as an event in the control algorithm. To date, we have used a static detection threshold, which was determined empirically based on the goal of keeping a minimum threshold value while avoiding false-positive results due to noise. A dynamic threshold based on the standard deviation of the baseline noise across several reference sonications might provide a more robust approach that could prevent important events from being missed. Furthermore, combining our methods with those proposed by others using the harmonic emissions for control³⁴ may prove the most robust.

As with our previous study with this control method,³⁰ our results show wide variations in the enhancement levels between animals. This technique controls the bubble behavior but may not account for variability due to the bubble concentration or biological response. Future work should investigate these parameters.

Some differences exist between this study and the published works^23,24 that motivated our study. The most important is the use of overlapping foci to create the larger volume, as described in the introduction. Additionally, the species differed (rat versus mouse), as did the US frequency and sonication scheme. In our study, we chose to use 0.5 MHz because of its clinical relevance and larger focal spot size. One limitation of our study was that it was performed at a single frequency, and we cannot do an exact comparison with studies that use substantially higher or lower frequencies. Our slice thickness on the MR images was 3 times that used by Samiotaki et al^23,24; thus, some small effects may have been averaged out. However, in the first study by Samiotaki et al,²³ the delayed opening that was observed could be seen to extend through most of the brain, exceeding our slice thickness.

The most notable difference in the two studies is the delivery of the MRI contrast agent. In this study, we used the recommended clinical dose of 0.1 mmoL/kg, delivered intravenously. Samiotaki et al^23,24 used an intraperitoneal injection of the contrast agent to delay the contrast peak and allow for improved longitudinal imaging. However, the intraperitoneal dose was 60 times the dose in our study, potentially providing greater sensitivity to small changes in the barrier integrity, but that approach might have limited clinical relevance. Finally, in the second study of Samiotaki et al,²⁴ a 50 times higher dose of Definity was used than in our study. It should also be noted that when Definity microbubbles were used,²⁴ the reported time to close for some sonication parameters (<8 hours) was more comparable with that reported here.

Other limitations of this study should be noted. The group size used here was small. Additionally, with the exception of 1 sonication, the barrier was closed by the first follow-up time point. Thus, small differences may exist between the multi- and single-point sonications within the 6-hour window after focused US.

The closure time of the BBB after opening is not the only safety consideration for clinical translation, but our results complement existing behavioral and cognitive studies that have considered the safety of this approach in mice³⁵ and primates.^25,36 The time for the BBB to close after focused US is independent of the opening volume when multiple overlapping foci are used. No differences in the treatment effect were observable by MRI follow-up between larger- and smaller-volume sonications, which suggests that larger-volume focused US BBB opening does not present additional safety risks. These findings have important implications for clinical implementation of this method and the ability to safely treat therapeutically relevant tissue volumes.