Local and reversible blood–brain barrier disruption by noninvasive focused ultrasound at frequencies suitable for trans-skull sonications Kullervo Hynynen, * Nathan McDannold, Nickolai A. Sheikov, Ferenc A. Jolesz, and Natalia Vykhodtseva Department of Radiology, Brigham and Women’s Hospital, and Harvard Medical School, Boston, MA 02115, USA Received 20 November 2003; revised 4 June 2004; accepted 11 June 2004 The purpose of this study was to test the hypothesis that burst ultrasound in the presence of an ultrasound contrast agent can disrupt the blood–brain barrier (BBB) with acoustic parameters suitable for completely noninvasive exposure through the skull. The 10-ms exposures were targeted in the brains of 22 rabbits with a frequency of 690 kHz, a repetition frequency of 1 Hz, and peak rarefactional pressure amplitudes up to 3.1 MPa. The total exposure (sonication) time was 20 s. Prior to each sonication, a bolus of ultrasound contrast agent was injected intravenously. Contrast-enhanced MR images were obtained after the sonications to detect localized BBB disruption via local enhancement in the brain. Brain sections were stained with H&E, TUNEL, and vanadium acid fuchsin (VAF)–toluidine blue staining. In addition, horseradish peroxidase (HRP) was injected into four rabbits prior to sonications and transmission electron microscopy was performed. The MRI contrast enhancement demonstrated BBB disruption at pressure amplitudes starting at 0.4 MPa with approx- imately 50%; at 0.8 MPa, 90%; and at 1.4 MPa, 100% of the sonicated locations showed enhancement. The histology findings following 4 h survival indicated that brain tissue necrosis was induced in approx- imately 70–80% of the sonicated locations at a pressure amplitude level of 2.3 MPa or higher. At lower pressure amplitudes, however, small areas of erythrocyte extravasation were seen. The electron microscopy findings demonstrated HRP passage through vessel walls via both transendothelial and paraendothelial routes. These results demonstrate that completely noninvasive focal disruption of the BBB is possible. D 2004 Elsevier Inc. All rights reserved. Keywords: Ultrasound; Bioeffects; Cavitation; Blood–brain barrier; Apop- tosis; Ischemia Introduction Advances in neuroscience have resulted in the development of new diagnostic and therapeutic agents and genes that may be used to study and treat many central nervous system (CNS) diseases (Pardridge, 2002a). However, the use of these agents is often limited by their access to the CNS via the blood supply because the blood–brain barrier (BBB) protects the brain from foreign molecules (Abbott and Romero, 1996; Kroll and Neuwelt, 1998; Nag, 2003b; Pardridge, 2002a). Several methods have been tested to circumvent the BBB including chemical modification of drugs to make them lipophilic or the use of carriers to aid propagation through the barrier (Pardridge, 2002a,b, 2003). Another option is to utilize an intraarterial infusion of hypertonic solution that initiates a BBB disruption lasting from a few minutes to a few hours (Doolittle et al., 2000). Each of the methods described above requires intra- arterial catheterization and produces diffuse, nonfocal BBB disruption within the entire tissue volume supplied by the injected artery branch (Abbott and Romero, 1996; Kroll and Neuwelt, 1998). The BBB disruption over a large volume, however, may prove a disadvantage for many applications since the agents may have undesired, often dose-limiting side-effects due to their spread within the CNS. To prevent this, a localized and reversible image- guided disruption of the BBB would provide anatomically or functionally targeted drug delivery while preserving an intact BBB to protect the nontargeted regions. Today, such localized drug delivery can only be accomplished by a direct injection of the agent into the targeted brain volume (Kroll and Neuwelt, 1998). Such direct injections are limited by the slow diffusion of molecules in the brain, the need to open the skull and penetrate nontargeted brain tissue, and the attendant risk of neurological damage, bleeding, and infection. There is experimental evidence that focused ultrasound can selectively disrupt the BBB locally in the brain (Mesiwala et al., 2002; Patrick et al., 1990; Vykhodtseva et al., 1995). However, the disruption is often associated with damage to the exposed brain tissue. Recently, a method has been developed wherein the ultrasound exposures are performed in the presence 1053-8119/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2004.06.046 * Corresponding author. Department of Radiology, Brigham and Women’s Hospital, and Harvard Medical School, 75 Francis Street, Boston, MA 02115. Fax: +1 617 278 0610. E-mail address: [email protected] (K. Hynynen). Available online on ScienceDirect (www.sciencedirect.com.) www.elsevier.com/locate/ynimg NeuroImage 24 (2005) 12 – 20
9
Embed
Local and reversible blood–brain barrier disruption by noninvasive focused ultrasound at frequencies suitable for trans-skull sonications
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
www.elsevier.com/locate/ynimg
NeuroImage 24 (2005) 12–20
Local and reversible blood–brain barrier disruption by noninvasive
focused ultrasound at frequencies suitable for trans-skull sonications
Kullervo Hynynen,* Nathan McDannold, Nickolai A. Sheikov,
Ferenc A. Jolesz, and Natalia Vykhodtseva
Department of Radiology, Brigham and Women’s Hospital, and Harvard Medical School, Boston, MA 02115, USA
Received 20 November 2003; revised 4 June 2004; accepted 11 June 2004
The purpose of this study was to test the hypothesis that burst
ultrasound in the presence of an ultrasound contrast agent can disrupt
the blood–brain barrier (BBB) with acoustic parameters suitable for
completely noninvasive exposure through the skull. The 10-ms
exposures were targeted in the brains of 22 rabbits with a frequency
of 690 kHz, a repetition frequency of 1 Hz, and peak rarefactional
pressure amplitudes up to 3.1 MPa. The total exposure (sonication)
time was 20 s. Prior to each sonication, a bolus of ultrasound contrast
agent was injected intravenously. Contrast-enhanced MR images were
obtained after the sonications to detect localized BBB disruption via
local enhancement in the brain. Brain sections were stained with H&E,
TUNEL, and vanadium acid fuchsin (VAF)–toluidine blue staining. In
addition, horseradish peroxidase (HRP) was injected into four rabbits
prior to sonications and transmission electron microscopy was
performed. The MRI contrast enhancement demonstrated BBB
disruption at pressure amplitudes starting at 0.4 MPa with approx-
imately 50%; at 0.8 MPa, 90%; and at 1.4 MPa, 100% of the sonicated
locations showed enhancement. The histology findings following 4 h
survival indicated that brain tissue necrosis was induced in approx-
imately 70–80% of the sonicated locations at a pressure amplitude level
of 2.3 MPa or higher. At lower pressure amplitudes, however, small
areas of erythrocyte extravasation were seen. The electron microscopy
findings demonstrated HRP passage through vessel walls via both
transendothelial and paraendothelial routes. These results demonstrate
that completely noninvasive focal disruption of the BBB is possible.
2% solution in saline) was administered (through the ear vein) to
macroscopically indicate the degree of BBB disruption (Nag,
2003a). HRP type VI (Sigma) dissolved in saline (300 mg/kg) was
injected immediately after the trypan blue through the same vein in
two rabbits. In the other two animals, HRP type II (Sigma) was
administered in the same doses. Two to three locations in each
brain were then sonicated; the animals were sacrificed at
deliberately selected intervals of approximately 1, 5, 20, and 60
min following each sonication. The animals were euthanized
(EuthasolR, Delmara Laboratories Inc., Midlothian, VA, 1 ml iv),
and the brains were fixed either by perfusion through the aorta (one
animal) or by immersion fixation using 2.5% paraformaldehyde +
1.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.2). The two
methods were used to eliminate potential problems associated with
each method. Pieces of about 0.5 mm3 from the sonicated areas
(blue spots) and from nonsonicated areas (controls) were fixed for
2 h in the same fixative. Distance measurements from the MR
Fig. 6. Examples of the histology observations in VAF–toluidine sections. (A) Gra
extravasated erythrocytes were seen within the focal region. (C) Grade 2: microsco
of them associated with evident damage to the brain parenchyma (insert magnifica
lesions (infarct).
images were used to correlate the sample locations and the MR
image analysis. They were then washed in 0.1 M TRIS buffer (pH
7.4) and transferred to an incubation medium containing 3,3V-diaminobenzidine and hydrogen peroxide for 45 min at room
temperature. After postfixation in 2% osmium tetraoxide in PBS
for 2 h, the pieces were dehydrated in ethanol, passed through
propylene oxide, and embedded in Epon-Araldite. Ultrathin
sections unstained or stained with uranyl acetate and lead citrate
were observed with a JEM-1200EX electron microscope at 80 kV.
Results
MRI observations
The BBB opening was observed in the sonicated locations in
the T1-weighted contrast-enhanced scans (Fig. 2). Fig. 3 shows
de 0: no damage was observed in the sonicated location. (B) Grade 1: a few
pic areas of perivascular extravasations over the whole sonicated area, some
tion of an ischemic cell). (D) Grade 3: extensive extravasation; hemorrhagic
Table 3
The histology grading of the sonicated tissues
Grade Description
0 No detected damage
1 One to a few tiny red blood cell extravasations
2 Petechial hemorrhages; mild damage to the brain parenchyma
3 Hemorrhagic or nonhemorrhagic local lesions
Table 4
The number of ischemic and apoptotic cells in the sonicated areas
Pressure
amplitude
(MPa)
Number of
apoptotic cells
Number of
apoptotic
cells in
vessels
Number
of ischemic
cells
Number of
sonicated
locations
0 0.5 F 0.7 0.1 F 0.5 1.9 F 3.3 45
0.4 0.5 F 0.6 0.2 F 0.5 0.7 F 0.5 4
0.5 0.6 F 1.1 0.4 F 0.5 0.6 F 0.8 7
0.8 1.4 F 1.2 1.1 F 1.1 5.3 F 5.2 10
1.1 1.8 F 2.3 1.4 F 1.8 3.6 F 6.5 9
1.4 3.0 F 1.8 2.5 F 1.0 6.7 F 6.4 4
2.3 Many Many 14 F 16 4
3.1 Many Many 12 F 7.2 7
The number of cells was counted in 10 nonoverlapping microscopic fields
chosen in each sonicated area. These counts were averaged for each
pressure amplitude value (mean F SD shown). The pressure amplitude
value 0 MPa represents the average value for all control areas in all of the
brains. When the sonicated areas were compared with the control areas (t
test), the number of apoptotic cells was statistically higher ( P b 0.05) at
pressure amplitude values at or above 0.8 MPa. The number of cells on the
vessel walls was statistically higher than in the control locations in all
sonicated areas at or above 0.5 MPa. The number of ischemic cells was
statistically higher than the control locations at or above 0.8 MPa.
K. Hynynen et al. / NeuroImage 24 (2005) 12–20 17
the signal intensity in T1-weighted imaging in one sonicated
location and likewise in the same anatomical location on the
opposite side of the brain. Following sonication, the signal
intensity at the targeted location increased by 15–20% compared
to the control location. BBB disruption was not detected after
an additional MRI contrast injection 5 h following the
sonication. The signal intensity change was greatest immediately
after the sonications and decreased as a function of time (Fig.
4). Three hours following the sonications, the signal intensity
increase was only approximately 10–20% of that measured
initially.
The signal intensity enhancement due to BBB disruption
increased as a function of the pressure amplitude and was evident
at the lowest value tested (0.4 MPa) (Fig. 5, top). Sixty percent of
the locations had focal contrast enhancement greater than the
standard deviation of the signal in the normal brain at 0.4 MPa.
By 1.4 MPa, all locations showed evidence of BBB disruption
(Fig. 5, bottom).
Histology
Forty-eight locations from among 13 brains were histolog-
ically evaluated in H&E-stained serial sections. In many cases,
the sonicated areas differed little from the nonsonicated regions.
The most noticeable effect observed in other cases was the
areas of extravasation (42/48 locations), ranging from a few
scattered erythrocytes in the mildest cases to extensive
extravasation at the highest pressure amplitude levels (Fig. 6).
These extravasations were limited to the focal volume of the
ultrasound beam. These histological observations were divided
into four grades ranging from no damage (grade 0) to tissue
necrosis (grade 3) (Table 3). The degree of these vascular
effects increased as a function of pressure amplitude when
necrosis was induced at or above 2.3 MPa (Fig. 5B). At
pressure amplitudes below 2.3 MPa, the extravasations were
observed in the absence of obvious parenchymal damage. In the
mildest cases (grade 1), the effects were limited to one or a few
small erythrocyte extravasations, which was the dominant effect
at pressure amplitudes up to approximately 0.5 MPa. In more
severe cases (grade 2), there were multiple areas of erythrocyte
extravasations, light vacuolation of the neuropil adjacent to
some of affected microvessels and alteration in the shape and
stainability of a few neurons and astrocytes. Mild neutrophil
infiltration was seen in the affected regions 48 h following
sonication. Grade 2 damage dominated at exposure levels of
between 0.5 and 1.4 MPa.
In the VAF–toluidine-blue-stained sections with the largest
apparent effect, the entire sonicated area contained on average
five or fewer injured neurons when the pressure amplitude was
below 1.4 MPa. Most of these neurons appeared dark stained
(hyperchromatic) and thus might represent reversibly injured
neurons. At a pressure amplitude of 2.3 MPa, the number of
ischemic neurons increased (Table 4). For tissue effects less than
grade 3, the only cells that were positive for ischemia were
neurons.
TUNEL-stained sections showed a few apoptotic cells in both
in the vessel walls and at the brain parenchyma at pressure
amplitude values lower than 2.3 MPa (Fig. 7). With TUNEL
technique, using counterstaining, some cell types could be clearly
identified according to their size, shape, and specific locations as
endothelial cells, intravascular erythrocytes, or leukocytes. Overall,
the TUNEL technique showed very few positive-stained cells, and
these cells were primarily located in blood vessels and glia.
However, it is possible that some individual cells were small
neurons and could not be easily distinguished from glial cells and
extravascular leukocytes by light microscopy. The average number
of apoptotic cells increased with pressure amplitude. At pressure
amplitudes of 2.3 MPa and higher, the tissue damage was always
associated with multiple apoptotic cells. A few scattered apoptotic
cells were also found outside the sonicated areas (Table 4).
Electron microscopy
In the samples obtained from the nonsonicated areas, HRP
filled the vessel’s lumina in brains fixed by immersion, while in the
cases of the perfusion fixation, the vessels appeared to be empty.
HRP did not penetrate the interendothelial clefts. Vesicles
containing HRP were present in only a few endothelial cells
(ECs). No HRP was seen in the subendothelial space: the basement
membrane, pericytes, or neuropil.
In all of the sonicated areas, passage of HRP through the vessel
walls could be seen. Capillaries, arterioles, and venules were
involved in this process, which showed a dependence on the
circulation time of the tracer. For animals sacrificed approximately
1 min after sonication, the HRP reaction product was seen in small
vacuoles and vesicles located at the luminal front of the ECs. Some
of the vesicles were in the process of formation and taking HRP
from the lumen (Fig. 8). This process was more apparent in the
samples obtained from the immersion-fixed brains. No HRP was
Fig. 7. Examples of apoptotic cells after sonications at 1.4 MPa (A) and 3.1 MPa (B). In (A), only a single small cell, probably a glia cell (arrow), next to
intact neurons was seen to be undergoing apoptosis. In (B), many of the cells within this lesion appeared apoptotic (left: low magnification; right: high
magnification).
K. Hynynen et al. / NeuroImage 24 (2005) 12–2018
present in the interendothelial clefts, basement membrane, peri-
cytes, or neuropil (Fig. 8).
For animals sacrificed approximately 5 min after sonication,
HRP-positive reaction was seen in caveolae, vacuoles, and
lysosome-like profiles in the ECs cytoplasm. The tight junctions
in some of the vessels appeared to be open and permitted HRP to
Fig. 8. Photomicrograph showing a cross section of a capillary 1 min after
sonication in the presence of circulating HRP. HRP fills the lumen (L) but
is not present in the basement membrane (b), pericyte (P), or in the
neuropil (NP). Formation of a caveola (long arrow) incorporating HRP is
seen at the luminal surface of the endothelial cell, while the caveolae at the
abluminal front of the cell are free from HRP (arrowheads). An
interendothelial cleft shows the penetration of HRP into its luminal
opening, but the tracer seems to be stopped at the tight junction (short
arrows). RBC—red blood cell in the lumen; NEC—nucleus of the
endothelial cell (immersion fixation of the brain).
pass through the intercellular cleft, penetrating into the basement
membrane and the perivascular space.
In the samples acquired in animals sacrificed approximately 20
min after sonication, a positive reaction for HRP was observed in
the abluminal zones of the EC, in the basal membrane, and in the
interstitium of the neuropil (Figs. 9A and B).
For a sacrifice time of 1 h following sonication, the
distribution of the HRP was similar. In some sonicated locations,
mild to moderate perivascular and interstitial edema was found.
Small groups of extravasated red blood cells were seen in three of
eleven locations, and a few ruptures of the microvessel walls
were found.
Discussion
The results showed that BBB disruption is possible at a
frequency of 0.69 MHz with minimal damage to the exposed brain
parenchyma cells. The frequency used is suitable for highly
localized sonication through human skull as has been reported
earlier (Clement and Hynynen, 2002). This noninvasive method
may be useful in delivering therapeutic or diagnostic agents in the
central nervous system (Pardridge, 2002a).
The electron microscopy findings confirm the effective
disruption of the BBB by ultrasound under the conditions applied.
The cellular mechanisms of HRP passage through vessel wall
included both transendothelial (via caveolae and cytoplasmic
vacuolar structures) and paraendothelial (via intercellular clefts)
routes. These mechanisms are similar to those observed in other
conditions resulting in BBB disruption (such as hypertension,
injection of bradykinin, hyperosmotic agents, etc.) (Nag, 2003c).
The BBB disruption took place in capillaries, arterioles, and
venules, demonstrating similar mechanisms. The data obtained
show that blood-circulating macromolecules (at least with molec-
ular weight 40,000) invade the brain interstitial space and thus
demonstrate that many therapeutic and diagnostic agents may be