Histologic safety of transcranial focused ultrasound neuromodulation and magnetic resonance acoustic radiation force imaging in rhesus macaques and sheep

Abstract

Background: Neuromodulation by transcranial focused ultrasound (FUS) offers the potential to non-invasively treat specific brain regions, with treatment location verified by magnetic resonance acoustic radiation force imaging (MR-ARFI).

Objective: To investigate the safety of these methods prior to widespread clinical use, we report histologic findings in two large animal models following FUS neuromodulation and MR-ARFI.

Methods: Two rhesus macaques and thirteen Dorset sheep were studied. FUS neuromodulation was targeted to the primary visual cortex in rhesus macaques and to subcortical locations, verified by MR-ARFI, in eleven sheep. Both rhesus macaques and five sheep received a single FUS session, whereas six sheep received repeated sessions three to six days apart. The remaining two control sheep did not receive ultrasound but otherwise underwent the same anesthetic and MRI procedures as the eleven experimental sheep. Hematoxylin and eosin-stained sections of brain tissue (harvested zero to eleven days following FUS) were evaluated for tissue damage at FUS and control locations as well as tissue within the path of the FUS beam. TUNEL staining was used to evaluate for the presence of apoptosis in sheep receiving high dose FUS.

Results: No FUS-related pre-mortem histologic findings were observed in the rhesus macaques or in any of the examined sheep. Extravascular red blood cells (RBCs) were present within the meninges of all sheep, regardless of treatment group. Similarly, small aggregates of perivascular RBCs were rarely noted in non-target regions of neural parenchyma of FUS-treated (8/11) and untreated (2/2) sheep. However, no concurrent histologic abnormalities were observed, consistent with RBC extravasation occurring as post-mortem artifact following brain extraction. Sheep within the high dose FUS group were TUNEL-negative at the targeted site of FUS.

Conclusions: The absence of FUS-related histologic findings suggests that the neuromodulation and MR-ARFI protocols evaluated do not cause tissue damage.

Keywords: Focused ultrasound; Magnetic resonance acoustic radiation force imaging; Neuromodulation; Safety.

Figure 1.

Summary of rhesus macaque study parameters. (a) Inclusion characteristics, survival time, and number of histologic samples evaluated for left (L) and right (R) hemispheres. (b) Illustration of rhesus macaque transducer positioning and (c) grid of focused ultrasound sonication in the visual cortex, where each location corresponds to estimated in situ spatial peak-temporal average intensity (I_SPTA values of 0.4, 1.6, 6.4, and 25.8 W/cm², applied in short bursts. Vertical spacing between FUS targets was 10 mm (NHP-1) and 15 mm (NHP-2), and horizontal spacing was 15 mm (NHP-1) and 20 mm (NHP-2). The lower two target locations (1.6 and 25.8 W/cm² I_SPTA) were placed 2 mm above the inion. Three coronal histologic sections were obtained from each hemisphere of the visual cortex (approximate locations shown by red planes). The first histology plane was located near the cortical surface, the second at a depth of approximately 4 mm, and the third at a depth of approximately 20 mm. (d) Illustration of neuromodulation protocol comprising 500 FUS bursts.

Figure 2.

Summary of sheep study parameters. (a) Sheep inclusion characteristics. The two sheep in the control group underwent MRI and anesthesia but no FUS. The eleven sheep that underwent FUS were subdivided into acute (euthanized zero days after FUS study), delayed (euthanized four to seven days after FUS study), repeated (underwent multiple FUS sessions, and euthanized four days after the last FUS study), and high dose groups (underwent prolonged MR-ARFI applications at one location on the last day of study). Days of survival following the first (left-most) and subsequent days of study are reported in split columns where applicable, for MRI without FUS (unshaded cells) and MRI with FUS sessions (shaded cells). The number of evaluated histologic sections is directly related to the number of FUS targets per sheep. (b) Sheep transducer positioning and (c) exemplary focused ultrasound sonication locations (6 locations shown; red circles) shown on axial T2-weighted MRI (cropped to show detail). Histologic sections were obtained from each location targeted with focused ultrasound and additionally from planes approximately 3 mm rostral and caudal to targeted locations (18 sections shown; dashed lines). Illustration of (d) MR-ARFI focal spot localization and (e-g) neuromodulation FUS protocols. Protocols comprised (d) 128, (e) 120 and (f-g) 600 FUS bursts.

Figure 3.

In vivo sheep study parameters. FUS applied acoustic power over time for each animal. Timing spans the total MRI and FUS session. Each cell represents a one minute interval, with color coding to indicate non-zero FUS acoustic powers. Empty cells indicate no FUS.

Figure 4.

In vivo sheep study parameters. (a,d) Range of applied acoustic powers and estimated in situ (b,e) peak pressure and (c,f) spatial peak temporal average intensity for MR-ARFI and neuromodulation, respectively. Total number of FUS bursts applied to each (g) MR-ARFI and (h) neuromodulation location, where animal number is reported below each bar cluster. Individual bars represent unique sonication locations, and bar height indicates number of FUS bursts delivered to that location.

Figure 5.

Distribution of the number of FUS bursts applied to each location with respect to the estimated in situ (a) peak pressure and (b) intensity of each sonication. MR-ARFI sonications (circles) were estimated to have in situ peak pressure between 1.7 and 3.6 MPa, which, due to the short 16 ms sonication times, corresponded to between 5.6 and 26.5 W/cm² I_SPTA. Neuromodulation sonications (triangles) were estimated to have peak in situ pressure between 0.25 and 0.9 MPa, corresponding to 0.6 and 13.8 W/cm² I_SPTA. The color scale indicates the number of locations at which each combination of in situ pressure or intensity and number of FUS bursts was observed. Blue rectangles indicate the range of parameters reported in human neuromodulation studies.

Figure 6.

Focal spot targeting and visualization. (a) Prescribed focal spot is indicated by red cross hairs drawn on T2-weighted MRI. (b) Tissue displacement at the focal spot is shown as an overlay on the MR-ARFI magnitude image. Stray pixels in the displacement map outside the brain are artifact due to slight changes between two MR-ARFI acquisitions.

Figure 7.

Prevalence of histologic findings within in vivo sheep study. The percentage of sections in which histologic findings were observed are reported for each animal by hemisphere (L and R; animal number listed at the top of each column). The number of histologic sections evaluated are reproduced from Fig. 2(a) for convenience. Green boxes indicate hemispheres where focused ultrasound was applied (all other boxes are internal controls or experimental controls). Meningeal and rare perivascular red blood cell extravasation were common histologic findings across all study groups, independent of whether any FUS was applied or which hemisphere was sonicated (in the case of FUS application). Necrosis, macrophage infiltration, red blood cell engulfment (erythrophagocytosis), and intracellular red blood cell breakdown (hemosiderin-laden macrophages), which would be expected to accompany true pre-mortem tissue damage, were not observed.

Figure 8:

Summary of parenchymal red blood cell extravasation foci in H&E-stained sheep brain tissue slides. The number of foci per slide are shown for tissue taken from hemispheres without FUS (blue dots) and hemispheres with FUS (yellow dots) for each study group where applicable. Bars indicate mean and standard error.

Figure 9.

Post-mortem perivascular and meningeal red blood cell extravasation does not differ across sheep treatment groups. Randomly scattered small volumes of extravasated red blood cells (black arrows) were identified adjacent to blood vessels within the neural parenchyma (a-g) and throughout the meninges (h-n) regardless of ultrasound exposure. Black outlines indicate blood vessel walls and delineate intravascular from extravascular red blood cells. No red blood cell extravasation was observed at parenchymal locations targeted with FUS. No associated pre-mortem tissue reactions (i.e., red blood cell engulfment (erythrophagocytosis), red blood cell breakdown (hemosiderosis), necrosis, or edema) were identified in any of the examined sections. Hematoxylin and eosin, scale bar = 50 μm.