Focused Ultrasound Blood-Brain Barrier Opening Arrests the Growth and Formation of Cerebral Cavernous Malformations

Abstract

Background: Cerebral cavernous malformations (CCM) are vascular lesions within the central nervous system, consisting of dilated and hemorrhage-prone capillaries. CCMs can cause debilitating neurological symptoms, and surgical excision or stereotactic radiosurgery are the only current treatment options. Meanwhile, transient blood-brain barrier opening (BBBO) with focused ultrasound (FUS) and microbubbles is now understood to exert potentially beneficial bioeffects, such as stimulation of neurogenesis and clearance of amyloid-β. Here, we tested whether FUS BBBO could be deployed therapeutically to control CCM formation and progression in a clinically-representative murine model.

Methods: CCMs were induced in mice by postnatal, endothelial-specific Krit1 ablation. FUS was applied for BBBO with fixed peak-negative pressures (PNPs; 0.2-0.6 MPa) or passive cavitation detection-modulated PNPs. Magnetic resonance imaging (MRI) was used to target FUS treatments, evaluate safety, and measure longitudinal changes in CCM growth after BBBO.

Results: FUS BBBO elicited gadolinium accumulation primarily at the perilesional boundaries of CCMs, rather than lesion cores. Passive cavitation detection and gadolinium contrast enhancement were comparable in CCM and wild-type mice, indicating that Krit1 ablation does not confer differential sensitivity to FUS BBBO. Acutely, CCMs exposed to FUS BBBO remained structurally stable, with no signs of hemorrhage. Longitudinal MRI revealed that FUS BBBO halted the growth of 94% of CCMs treated in the study. At 1 month, FUS BBBO-treated lesions lost, on average, 9% of their pre-sonication volume. In contrast, non-sonicated control lesions grew to 670% of their initial volume. Lesion control with FUS BBBO was accompanied by a marked reduction in the area and mesenchymal appearance of Krit mutant endothelium. Strikingly, in mice receiving multiple BBBO treatments with fixed PNPs, de novo CCM formation was significantly reduced by 81%. Mock treatment plans on MRIs of patients with surgically inaccessible lesions revealed their lesions are amenable to FUS BBBO with current clinical technology.

Conclusions: Our results establish FUS BBBO as a novel, non-invasive modality that can safely arrest murine CCM growth and prevent their de novo formation. As an incisionless, MR image-guided therapy with the ability to target eloquent brain locations, FUS BBBO offers an unparalleled potential to revolutionize the therapeutic experience and enhance the accessibility of treatments for CCM patients.

Figure 1.

FUS effectively opens the BBB within the CCM microenvironment. (A) Confocal image of a CCM (in the absence of FUS) stained with CD31 for endothelial cells. Image depicts the grossly enlarged CCM core (yellow arrow), and moderately dilated perilesional vasculature (white arrows). Scale bar = 100 μm. (B) Top row: Baseline, high-resolution T2-weighted spin echo images used for selecting CCMs for FUS targeting. Arrowheads indicate selected CCMs. Middle row: T1-weighted spin echo images acquired following gadolinium contrast agent injection but immediately prior to FUS application. Circles indicate targeted CCMs, and insets display magnified views of the targeted CCMs. Bottom row: T1-weighted spin echo images acquired following gadolinium contrast agent injection and FUS application. Columns indicate PNPs used for sonication. T1 contrast enhancement is visible following FUS BBBO and localized to perilesional boundaries of the sonicated CCM. (C) Bar graph of T1 contrast enhancement quantified as the fold change in grayscale intensity of sonicated CCMs in the post-image over the pre-image (as seen in A). Gadolinium accumulation following FUS BBBO over the baseline CCM leakiness for PNPs of 0.3 MPa to 0.6 MPa. p=0.0054 for 0.3 MPa and p<0.0001 for 0.4 MPa – 0.6 MPa, one-way ANOVA followed by Dunnett’s multiple comparisons test.

Figure 2.

Acute stability of CCMs exposed to FUS BBBO. (A) High-resolution T2-weighted spin echo images displaying either CCMs prior to sonication (top row) or 24 h following sonication (bottom row). Circles denote targeted CCMs, and insets display magnified views of the targeted CCMs. (B) Targeted CCM volumes prior to sonication and 24 h following sonication on T2-weighted spin echo images with color indicating applied PNP. CCM volume does not significantly demonstrate changes in volume following sonication. p=0.41, Wilcoxon matched-pairs signed rank test. (C) High-resolution susceptibility-weighted images of the same mice in A, displaying either CCMs prior to sonication (top row) or 24 h following sonication (bottom row). (D) Targeted CCM volumes prior to sonication and 24 h following sonication on susceptibility-weighted images with color indicating applied PNP. CCM volume does not significantly demonstrate changes in bleeding following sonication. p=0.34, Wilcoxon matched-pairs signed rank test.

Figure 3.

Comparison of FUS BBBO contrast enhancement and acoustic emission signatures between wild-type and CCM mice. (A) Representative T1-weighted spin echo images acquired following gadolinium contrast agent injection and FUS application in wild-type mice or CCM mice for PNPs of 0.4 MPa – 0.6 MPa. (B) Bar graph of T1 contrast enhancement. Enhancement is comparable in wild-type and CCM mice for PNPs of 0.4 MPa- 0.6 MPa. p=0.92 for 0.4 MPa, p=0.9998 for 0.5 MPa, and p=0.96 for 0.6 MPa; two-way ANOVA with Šidák’s multiple comparison test. (C) Spectrograms of the frequency response for each burst during the FUS application averaged over cohorts of wild-type and CCM mice at PNPs of 0.4 MPa – 0.6 MPa (n=3 mice per group and 2–3 sonication replicates per mouse). (D) Subharmonic, first ultraharmonic, and broadband emissions for wild-type and CCM mice at PNPs of 0.4 MPa – 0.6 MPa. p>0.4 for all PNPs, two-way ANOVA with Šidák’s multiple comparisons test. (E) Second, third, and fourth harmonic emissions for wild-type and CCM mice at PNPs of 0.4 MPa – 0.6 MPa, indicating that stable cavitation-associated signatures between wild-type and CCM mice are comparable at 0.4 MPa and 0.5 MPa, but not significantly increased in CCM mice at 0.6 MPa. P > 0.7 for 0.4 – 0.5 MPa and 2nd – 4th harmonics; p <0.0001, p = 0.0006, p<0.0001 for 0.6 MPa and 2nd, 3rd, and 4th harmonics, respectively; two-way ANOVA with Šidák’s multiple comparisons test.

Figure 4.

CCM mice are not differentially sensitive to adverse effects generated by FUS BBBO at high PNPs. (A) Representative high resolution, T2-weighted spin echo images of wild-type and CCM mice at 1 d, 7 d, and 30 d post-sonication at PNPs of 0.4 MPa – 0.6 MPa in either a single sonication or repeat sonication treatment regimen. Ovals denote focal column. White arrows denote hyperintensities associated with edema. Yellow arrows denote hypointensities associated with hemosiderin deposition. (B) Scatterplot of ipsilateral-to-contralateral grayscale intensity at 1d post-FUS (when edema is visible) of wild-type and CCM mice for PNPs of 0.4 MPa – 0.6 MPa. p=0.047, comparison of fits with F-test for a 2nd order polynomial regression. (C) Scatterplot of ipsilateral-to-contralateral grayscale intensity at 30d post-FUS (when hemosiderin is visible) of wild-type and CCM mice for PNPs of 0.4 MPa – 0.6 MPa. p=0.77, comparison of fits with F-test for a 2nd order polynomial regression. (D) Line graphs of ipsilateral-to-contralateral grayscale intensities over the one-month imaging period for all PNPs within a mouse model and treatment arm, revealing that edema on day 1 is generally followed by hemosiderin on days 7 and 30. (E) Ipsilateral-to-contralateral grayscale intensities over the one-month imaging period for all PNPs within a mouse model and treatment arm, indicating no significant differences when comparing models at individual PNPs within a treatment arm. p = 0.1368 and p = 0.5386 for both PNPs in the single treatment arm for edema and hemosiderin, respectively; p > 0.7 for PNPs of 0.4 MPa and 0.5 MPa and p = 0.0923 for PNP of 0.6 MPa in the repeat treatment arm for edema; p > 0.5 for all PNPs in the repeat treatment arm for hemosiderin; two-way ANOVA with Holm-Šidák’s multiple comparisons test.

Figure 5.

Real-time PCD-modulation of PNP ensures the safety of sonicated brain tissue without compromising gadolinium delivery. (A) Applied PNP versus time during PCD feedback-controlled approach. Each line indicates the average applied PNP across two sonication targets for the same mouse during a single FUS sonication period. (B) Representative T1-weighted contrast images before and after FUS BBBO with PCD-modulated PNPs. (C) Bar graph of T1 contrast enhancement quantified as the fold change in grayscale intensity of sonicated CCMs in the post-image over the pre-image (as seen in B), indicating successful BBBO. p=0.016, Wilcoxon matched-pairs signed rank test. (D) Bar graph of T1 contrast enhancement quantified as the fold change in grayscale intensity of sonicated CCMs in the post-image over the pre-image for CCM mice with fixed PNP and PCD-modulated PNP cohorts. Graphs reveal that T1 contrast enhancement is greater with PCD-modulated PNP compared to fixed PNP in the same range of applied PNP of 0.2 – 0.4 MPa. p < 0.0001 for PCD vs. 0.2 MPa, p = 0.0018 for PCD vs. 0.3 MPa, p = 0.0368 for PCD vs. 0.4 MPa, p = 0.2864 for PCD vs. 0.5 MPa, and p = 0.9918 for PCD vs. 0.6 MPa, one-way ANOVA with Dunnett’s multiple comparison’s test. (E) Spectrogram of the frequency response for each burst during the FUS application averaged over CCM mice with PCD-modulated PNP (n=4 mice and 2 sonication replicates per mouse). Dotted line denotes separation of baseline sonications without microbubbles and sonications with microbubbles. (F) Subharmonic, broadband, and second harmonic emissions for CCM mice at PCD-modulated PNP and fixed PNPs of 0.4 MPa – 0.6 MPa, indicating comparable acoustic signatures for PNPs less than 0.6 MPa. p > 0.8 for the subharmonic, ultraharmonic, and 2^nd-3^rd harmonic emissions for PCD vs. 0.4 or 0.5 MPa; p > 0.3 for the broadband emissions; p = 0.003 for 2^nd harmonic emissions and 0.6 MPa vs. PCD,0.4 MPa, and 0.5 MPa; two-way ANOVA with Šidák’s multiple comparisons test. (G) Representative high resolution, T2-weighted spin echo images of wild-type and CCM mice at 1 d, 7 d, and 30 d post-sonication at PNPs of 0.4 MPa – 0.6 MPa in either a single sonication or repeat sonication treatment regimen. Ovals denote focal column. (H) Line graphs of ipsilateral-to-contralateral grayscale intensities over the one-month imaging period for CCM mice and all PNP regimens. (I) Scatterplot of ipsilateral-to-contralateral grayscale intensity versus time-averaged PNP for CCM with single treatments and fixed PNP, repeat treatments and fixed PNP, or repeat treatments and PCD-modulated PNP mice on day 1 (left) or day 30 post-FUS (right). For edema, ipsilateral-to-contralateral grayscale intensity is not significantly correlated with PNP; however, for hemosiderin, ipsilateral-to-contralateral grayscale intensity is significantly correlated with PNP. p = 0.8382 for edema and p = 0.0163 for hemosiderin, linear regression with F test.

Figure 6.

FUS BBBO arrests the growth of CCMs. (A, C, E) Longitudinal T2-weighted spin echo images for representative mice in the (A) single sonication with fixed PNP arm, (C) repeat sonication with fixed PNP arm, or (E) repeat sonication with PCD-modulated PNP arm. Black circles indicate non-sonicated, control lesions, and colored circles indicate sonicated lesions corresponding to PNP applied. White arrows denote new lesions formed in non-sonicated hemisphere. (B, D, F) Left: Summary plots comparing the natural log transform of CCM volume between sonicated CCMs and non-sonicated CCMs for mice in the (B) single sonication with fixed PNP arm, (D) repeat sonication with fixed PNP arm, or (F) repeat sonication with PCD-modulated PNP arm. Right: Line graphs of CCM volume for individual CCMs for each treatment group. At 30 days, sonicated CCMs are significantly smaller than non-sonicated control CCMs for all treatment arms. p = 0.0002, p < 0.0001, and p = 0.0131 for the single, fixed PNP; repeat, fixed PNP; and repeat, PCD-mod. PNP arms, respectively; linear mixed effect model and pairwise comparison with Tukey’s adjustment. At 7 days, sonicated CCMs are significantly smaller than non-sonicated CCMs in the repeat FUS and fixed PNP arm. p = 0.0021, linear mixed effect model and pairwise comparison with Tukey’s adjustment.

Figure 7.

FUS BBBO with fixed PNP and repeat sonications can prevent de novo lesion formation. (A, C, E) Top row: T1-weighted spin echo images taken immediately following FUS BBBO with hyperintense signal denoting the focal column. Middle and bottom rows: minimum intensity projection images of longitudinal T2-weighted spin echo images to visualize through 1 mm of the focal column for representative mice in the (A) single sonication with fixed PNP arm, (C) repeat sonication with fixed PNP arm, or (E) repeat sonication with PCD-modulated PNP arm. Black ovals denote contralateral, non-sonicated ROIs for de novo quantification, while colored ovals represent sonicated ROIs. (B, D, F) Paired line graphs comparing the change in CCM number one month following FUS BBBO between the sonicated brain region and the contralateral non-sonicated brain region for mice in the (B) single sonication with fixed PNP arm, (D) repeat sonication with fixed PNP arm, or (F) repeat sonication with PCD-modulated PNP arm. Concentric circles indicate multiple mice with the same number of de novo CCMs. Colors indicate applied PNP. For mice receiving the repeat FUS regimen with fixed PNP, the number of new lesions formed in the sonicated brain region is significantly reduced compared to the contralateral brain region. p = 0.0312, Wilcoxon matched-pairs signed rank test.

Figure 8.

FUS BBBO restores endothelial morphology to the mutated CCM vasculature and remodels CCM immune landscape. (A) Immunofluorescent images of non-sonicated and sonicated CCMs at 30 d post-FUS BBBO with staining for mutated vasculature (Krit1KO), microglia/macrophages (Iba1), and erythrocytes (Ter119). The mutated vasculature in sonicated CCMs had reduced mesenchymal appearance compared to non-sonicated CCMs. (B) Graph of average mutated CCM vasculature area at 1 d, 7 d, and 30 d post-FUS BBBO for non-sonicated and sonicated CCMs, indicating reduced area in sonicated CCMs at 30 d. p = 0.0199, linear mixed effect model and pairwise comparison with Tukey’s adjustment. (C) Immunofluorescent images of non-sonicated and sonicated CCMs at 1 d and 7 d post-FUS BBBO with staining for mutated vasculature (Krit1KO), microglia/macrophages (Iba1), and proliferation (Ki67). (D) Graph of density of microglia/macrophages at 1 d, 7 d, and 30 d post-FUS BBBO for non-sonicated and sonicated CCMs, revealing a reduced number in sonicated lesions at 1 d. p = 0.0003, linear mixed effects model and pairwise comparison with Tukey’s adjustment. (E) Graph of the natural log of the average microglia/macrophage area at 1 d, 7 d, and 30 d post-FUS BBBO for non-sonicated and sonicated CCMs, demonstrating an increase in microglia/macrophage size in sonicated lesions at 1 d. p = 0.0106, linear mixed effect model and pairwise comparison with Tukey’s adjustment. (F) Immunofluorescent images of non-sonicated and sonicated CCMs at 1 d post-FUS BBBO with staining for mutated vasculature (Krit1KO), microglia/macrophages (Iba1), lysosomes (CD68), and erythrocytes (Ter119). Insets display 63x maximum intensity projections of the corresponding 20x image. Arrows denote foamy macrophages. (G) Graph of the natural log of phagocyte density at 1 d, 7 d, and 30 d post-FUS BBBO for non-sonicated and sonicated CCMs, revealing a reduced number in sonicated lesions at 1 d. p = 0.0009, linear mixed effects model and pairwise comparison with Tukey’s adjustment. (H) Graph of the natural log of the percent of erythrocytes colocalized in microglia/macrophages at 1 d, 7 d, and 30 d post-FUS BBBO for non-sonicated and sonicated CCMs, indicating a smaller amount in sonicated lesions at 7 d. p = 0.0303, linear mixed effects model and pairwise comparison with Tukey’s adjustment.

Figure 9.

Current clinical FUS systems are equipped to treat CCMs in patients. (A) Stereotactic radiosurgery (SRS) treatment plans for 3 CCM patients with surgically inaccessible lesions. Yellow and green lines are 12.5 Gy and 6.3 Gy isodoses, respectively. (B) Mock FUS BBBO treatment plans using the NaviFUS clinical system software, demonstrating the feasibility of CCM treatment with current clinical FUS systems. Red, grouped focal points denote treatment of CCM with 43 sonication points spanning 2 cm in diameter and 8.65 cm³ in volume.