Drug Delivery to the Brain by Focused Ultrasound and ... · using Two-photon Fluorescent Microscopy...

Drug Delivery to the Brain by

Focused Ultrasound and Microbubble Mediated Blood-brain Barrier Disruption:

Vascular-level Investigation using Two-photon Fluorescent Microscopy

by

Tam Quy Nhan

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Department of Medical Biophysics University of Toronto

© Copyright by Tam Quy Nhan 2015

ii

Abstract

Drug Delivery to the Brain by Focused Ultrasound and Microbubble

Mediated Blood-brain Barrier Disruption: Vascular-level Investigation

using Two-photon Fluorescent Microscopy

Tam Quy Nhan

Doctor of Philosophy

Department of Medical Biophysics

University of Toronto

2015

The use of focused ultrasound (FUS) in combination with microbubbles (MBs) to transiently and

noninvasively disrupt the blood-brain barrier (BBB) has been an active research topic which

could ultimately revolutionize the way drugs are delivered into the brain parenchyma for

treatment of central nervous system (CNS) pathologies. Stemmed from its prospective clinical

application, the overarching goal of this research is to explore the underlying physical

mechanisms and fundamental biological effects via the use of two-photon fluorescence

microscopy (2PFM). Based on the insights gained from these microscopic evaluations, this

research also aims to draw the connection between the kinetics of blood-brain barrier disruption

(BBBD) and the resulting effect of localized drug deposition in the treated brain.

To provide a robust solution for dorsal approach of FUS exposure and in vivo 2PFM

imaging of the cerebral microvasculature, a thin ring-shaped transducer has been designed and

characterized. Two modes of vibration (thickness and height) from the transducer configuration

were investigated for their effectiveness at inducing BBBD in a rat model. With the transducer

operating in the thickness mode at 1.2 MHz frequency, shallow and localized BBBD near the

cortical surface of the animal brain was detected via 2PFM and confirmed by Evans blue (EB)

iii

extravasation. Acoustic pressures ranging from 0.2 to 0.8 MPa, which is the typical threshold for

BBBD, were reliably produced and the evidence of successful BBBD was shown.

Using the aforementioned system design, we conducted a series of 2PFM imaging

sessions while delivering dextran-conjugated fluorescent dyes of various sizes into the rat’s blood

circulatory system and inducing BBBD at different acoustic pressure levels in the 0.2-0.8 MPa

range. Analyses of these time-lapsed microscopic data allowed for quantitative measurements of

the enhanced permeability of blood vessels within the imaging field of view upon incidences of

BBB opening. Derived from these quantitative analyses, the dependency of the vascular

permeability on the test substance size and on the applied acoustic pressured was established. In

addition, we identified two types of leakage kinetics - fast and slow - that exhibit distinctive

permeability constants, temporal disruption onsets, and pertinent vessel diameter. Such direct

assessment of vascular permeability offers insightful and practical knowledge towards treatment

strategies of BBBD-based drug delivery.

To further translate these relevant findings obtained from preclinical studies into the

clinical setting, we developed a mathematical framework that closely depicts the transient and

reversible kinetics of BBBD and rendered the spatio-temporal distribution of the intended drug at

a targeted brain region. In this pilot study, we considered Doxorubicin (Dox) as the therapeutic

agent of choice due to its available preclinical data and promising results of Dox delivery with

high efficiency under FUS treatment. The constructed model predicts Dox concentrations within

three compartments - plasma, extracellular, and intracellular - that are governed by various

transport processes (e.g. diffusion in interstitial space, exchange across vessel wall, clearance by

cerebral spinal fluid, and uptake by brain cells). By examining several clinical treatment

parameters (e.g. sonication scheme, permeability enhancement, and injection mode), our

simulation outputs are in agreement with experimental findings in a rat model by Park et al. In

particular, we identified the optimal time delay between two consecutive sonications to be 10

min. We estimated the intracellular concentration to be 400-1200 ng/g tissue in response to 10

min spacing double sonication and permeability constant range of 0.01-0.03 min-1. Considering

the flexibility of the FUS+MBs assisted BBBD technique in delivering various therapeutic agents

of diverse size and chemical properties to the brain, this simulation study can be adapted for

other drugs in order to assist the treatment planning process for different CNS disease conditions.

iv

Acknowledgements

First and foremost, I would like to express deepest gratitude to my supervisor, Dr. Kullervo

Hynynen. His dedication towards the research avenue of therapeutic focused ultrasound and his

vision of revolutionizing healthcare via FUS applications are awe-inspiring. His “big picture”

outlook has driven my motivation to the day-to-day research activities and his guidance

approach has allowed me to grow as an independent researcher.

Next, I would like to acknowledge the tremendous support from my committee members.

The feedbacks I gained through official meetings or personal exchanges were very helpful in

steering my projects in the right direction. In particular, Dr. Bojana Stefanovic has provided

valuable insights pertaining to in vivo experiments with the two-photon microscopy system.

Despite their offices being distant from Sunnybrook Hospital, I truly appreciated Dr. Lothar Lilge

and Dr. Shirley Wu for their time generosity and willingness to travel uptown for every

committee meeting. Furthermore, Dr. Lothar Lilge has been my “go-to” scientist whenever I was

stuck with a general biophysics concept or a specific technical challenge. Meanwhile, Dr. Shirley

Wu has offered her knowledge in novel drug design as well as collaboratively provided us with

nanoparticle sample for “fluorescent microbubbles” project.

This thesis work would be impossible without the enormous support from “C7-ers”.

First, I am forever indebted to Dr. Alison Burgess’ guidance throughout my entire PhD journey.

Not only passionately listening to my research challenges and genially sharing her ideas every

time I knocked on her office door, she has been extremely patient with editing and reviewing any

piece of my writing whether it is a conference abstract, or a manuscript, or this very thesis.

Beyond that, she has set an inspirational example of a “superwoman” who could juggle work and

life in a caring and fun manner. Second, I would like to express gratitude to other great research

minds: Dr. David Goertz, Dr. Rajiv Chopra Dr. Yuexi Huang, Dr. Meaghan O’Reilly and Dr. Sam

Pichardo for their utter willingness to engage in any scientific exchanges whenever I seek their

advice. Third, I consider myself particularly lucky to be in a “well-equipped” laboratory with an

enthusiastic team of technicians. The electronics-related component of my research (e.g. matching

circuit, power meter, delicate soldering) has greatly benefited from the assistance of Dr. Junho

v

Song, Sam Gunaseelan and Ping Wu. The mechanical tools required for in vitro and in vivo

experimental setup have been promptly and cleverly built by Fedon Orfanidis. As a core of this

thesis work, two-photon imaging on rat brain would not be feasible without the incredible

microsurgical skills from Shawna Rideout-Gros and Alex Garces. I am thankful for all their

dedications and positive attitudes, which has enabled me to push through the trials and

tribulations. Last, I earnestly value the friendship that I had gained throughout these four years

at the C7 lab. I am grateful for all the wonderful memories that I had shared with the current

peers (Alec Hughes, Christopher Acconcia, Dan Pajek, Mathew Carias, Nazanin Hosseinkhan,

Nicholas Ellens, Ryan Jones, Ryan Alkins) as well the past members (Dr. Aki Pulkkinen, Dr.

Arvin Arani, Dr. Brandon Helfield, Leila Shaffaf, Patrick Leonard and Dr. Robert Staruch).

Outside the C7 lab, I was fortunate to learn from and interact with other scientists and

engineers at Sunnybrook Research Institute, including Adrienne Dorr, Ross Williams, Mike Lee,

Dr. Naomi Matsuura and Dr. Minseok Seo. I am sincerely thankful for their kindness in assisting

me with numerous practical aspects of the project. Furthermore, I would like to acknowledge the

funding support the Natural Sciences and Engineering Research Council of Canada (NSERC

CGS-D3), the Canadian Institutes of Health Research (CIHR), the National Institutes of Health

(NIH), the Canada Research Chair program and the Department of Medical Biophysics at

University of Toronto.

Last but not least, I want to accredit the accomplishment to my family. To my parents and

sister, thank you for your unconditional love and heartfelt encouragements. To my in-laws, I

appreciate what you had done during these past years to make my life easier so I could dedicate

all my energy to pursue my studies. Finally, to my husband Lam Phan – a.k.a my awesome

badminton partner, this achievement could never happen without your constant “cheering and

pushing” since day One. You showed me your full support when I decided to leave my job to

pursue my life-long dream of obtaining a Ph.D. Thank you for being the first-hand editor to all of

my writing, as well as being a chauffeur upon the late nights and weekend calls. Most of all, I am

grateful for your positive outlook in life and work hard mentality, which were the driving forces

to help me to cross the finish line.

vi

Table of Contents

1 Background ........................................................................................................................................... 1

1.1 Challenges to drug delivery to the brain ........................................................................................ 1

1.1.1 Current status & treatment for CNS pathology ...................................................................... 1

1.1.1.1 Neurological disorders ........................................................................................................ 1

1.1.1.2 Brain cancer ........................................................................................................................... 2

1.1.2 Structure & function of the blood-brain barrier ...................................................................... 3

1.1.3 Methods to bypass the BBB for drug delivery ......................................................................... 4

1.2 FUS+MBs induced BBBD for drug delivery ................................................................................... 6

1.2.1 Therapeutic ultrasound .............................................................................................................. 6

1.2.1.1 Thermal effects ...................................................................................................................... 7

1.2.1.2 Non-thermal effects ............................................................................................................ 10

1.2.2 Basic components of FUS+MBs mediated BBBD .................................................................. 11

1.2.2.1 Ultrasound induced BBBD ................................................................................................ 12

1.2.2.2 Microbubbles assisted BBBD ............................................................................................ 12

1.2.2.3 Ultrasound parameters for BBBD..................................................................................... 14

1.2.3 Cellular mechanisms ................................................................................................................. 19

1.2.4 Physical mechanisms ................................................................................................................ 20

1.3 Pre-clinical progresses of BBBD-based drug delivery ................................................................. 22

1.3.1 Delivery of macromolecules & therapeutic agents ............................................................... 22

1.3.1.1 Chemotherapy .................................................................................................................... 22

1.3.1.2 Novel agents for targeting brain tumor & metastasis ................................................... 23

1.3.1.3 Immunotherapy for Alzheimer’s disease (AD) .............................................................. 25

1.3.1.4 Gene therapy for Huntington’s disease (HD) ................................................................. 25

1.3.1.5 Stem cell therapy ................................................................................................................ 26

1.3.2 Safety evaluation ....................................................................................................................... 26

1.3.2.1 Reversibility of BBB opening ............................................................................................ 26

1.3.2.2 Short-term & long-term effect on tissue .......................................................................... 27

1.3.2.3 Extravasation of blood-borne material ............................................................................ 28

vii

1.3.2.4 Behavioral tests ................................................................................................................... 28

1.4 Clinical translation of BBBD-based drug delivery ....................................................................... 29

1.4.1 Transcranial ultrasound exposure .......................................................................................... 29

1.4.2 Assessment methods of FUS+MBs induced BBBD ............................................................... 30

1.5 Research objectives ........................................................................................................................... 31

1.5.1 Problem statement ..................................................................................................................... 31

1.5.2 Specific aims ............................................................................................................................... 32

1.5.3 Thesis outline ............................................................................................................................. 33

2 Transducer design and characterization for dorsal-based FUS exposure and 2PFM imaging

of in vivo BBBD in a rat model .............................................................................................................. 35

2.1 Overview on basics of transducer .................................................................................................. 35

2.1.1 Piezoelectric effect ..................................................................................................................... 35

2.1.2 Resonance frequency................................................................................................................. 36

2.1.3 Modes of vibration .................................................................................................................... 38

2.1.4 Transducer structure and backing .......................................................................................... 39

2.2 Research motivation ......................................................................................................................... 40

2.3 Materials & methods ........................................................................................................................ 42

2.3.1 Transducer design ..................................................................................................................... 42

2.3.2 Transducer characterization ..................................................................................................... 42

2.3.3 Experimental setup for BBBD induction and in vivo 2PFM imaging ................................ 44

2.4 Results ................................................................................................................................................ 46

2.4.1 Transducer fabrication .............................................................................................................. 46

2.4.2 US pressure resulting from different mode of vibration ..................................................... 47

2.4.3 Output acoustic pressure .......................................................................................................... 48

2.4.4 2PFM imaging of BBBD in a rat model ................................................................................... 51

2.5 Discussion .......................................................................................................................................... 54

2.6 Conclusions ....................................................................................................................................... 61

3 Quantitative evaluation of enhanced permeability of BBB using 2PFM .............................. 62

3.1 Overview on 2PFM........................................................................................................................... 62

3.1.1 Basic principle of 2PFM ............................................................................................................ 62

3.1.2 Design of 2PFM .......................................................................................................................... 63

3.1.3 Two-photon versus single-photon fluorescence microscopy .............................................. 65

viii

3.2 Research motivation ......................................................................................................................... 66


3.3.1 Animal preparation ................................................................................................................... 67

3.3.2 FUS parameters for BBBD ........................................................................................................ 67

3.3.3 2PFM imaging ............................................................................................................................ 68

3.3.4 Analysis of 2PFM data .............................................................................................................. 69

3.3.5 Statistical analysis ...................................................................................................................... 71

3.4 Results ................................................................................................................................................ 71

3.4.1 Effect of acoustic pressure on enhanced BBB permeability ................................................. 71

3.4.2 Effect of substance size on enhanced BBB permeability ...................................................... 71

3.4.3 Temporal onset of BBBD is correlated with permeability and appears to be controlled by

acoustic pressure................................................................................................................................. 72

3.4.4 Effect of vessel diameter on enhanced BBB permeability .................................................... 74

3.5 Discussion .......................................................................................................................................... 75

3.6 Conclusions ....................................................................................................................................... 78

4 Modelling localized delivery of Doxorubicin to the brain based on FUS-enhanced

permeabilization of BBB ......................................................................................................................... 79

4.1 Introduction ....................................................................................................................................... 79


4.2.1 Model geometry ......................................................................................................................... 81

4.2.2 Model assumption ..................................................................................................................... 81

4.2.3 Mathematical model of drug transport and distribution..................................................... 82

4.2.3.1 Plasma compartment ......................................................................................................... 83

4.2.3.2 Extravascular-extracellular compartment ....................................................................... 83

4.2.3.3 Intracellular compartment................................................................................................. 84

4.2.4 Model parameters ...................................................................................................................... 85

4.2.5 Boundary conditions ................................................................................................................. 86

4.2.6 Numerical methods ................................................................................................................... 86

4.3 Results ................................................................................................................................................ 88

4.3.1 Increase in Dox delivery by FUS induced BBB permeability .............................................. 88

4.3.2 Compare the effect of sonication schemes on Dox delivery ................................................ 90

4.3.3 Effect of BBB permeability enhancement level on Dox delivery ........................................ 93

ix

4.3.4 Effect of injection modes on Dox delivery ............................................................................. 94

4.4 Discussion .......................................................................................................................................... 95

4.5 Conclusions ..................................................................................................................................... 100

5 Conclusions & Future Work .......................................................................................................... 101

5.1 Summary of findings ..................................................................................................................... 101

5.2 Limitations ....................................................................................................................................... 104

5.2.1 Transducer handling ............................................................................................................... 104

5.2.2 Delicate microsurgery of rat brain ........................................................................................ 105

5.2.3 Limitations of Current Simulation Study ............................................................................. 106

5.3 Future directions ............................................................................................................................. 106

5.3.1 Incorporation of passive cavitation detection ..................................................................... 107

5.3.2 Imaging fluorescent MBs during BBBD using 2PFM ......................................................... 107

5.3.3 Extending 2PFM-based BBBD study to other therapeutic agents .................................... 108

5.3.4 Extending the simulation model to other therapeutic agents ........................................... 109

5.4 Clinical perspectives ...................................................................................................................... 110

Appendix A: Imaging nanoparticle-incorporated microbbubles in vivo using two-photon

fluorescence microscopy ........................................................................................................................ 112

Appendix B: Correlation between substance size and its permeability at the BBB ................... 115

x

List of Tables

Table 1.1: Comparison of BBBD pressure threshold data from other studies to the aggregated data

by McDannold et al. [91]. .......................................................................................................................... 16

Table 1.2: Summary of different imaging tracers and monitoring techniques being used for BBBD

preclinical study ......................................................................................................................................... 22

Table 2.1: Summary of resonant frequencies from three vibration modes of .................................... 47

Table 4.1: Pharmacokinetics and pharmacodynamic parameters of Doxorubicin ............................ 87

Table B.1: Summary of measured permeability constants of 1 kDa – 500 kDa test substance upon

BBBD induced at 0.4 MPa and 0.6 MPa FUS acoustic pressure ......................................................... 117

Table B.2: Normalized permeability versus log(MW) ......................................................................... 117

xi

List of Figures Figure 1.1: Percentage of total disability-adjusted life years (DALYs) for various diseases and

neurological conditions. Adapted from [1]............................................................................................... 2

Figure 1.2: Illustrative structure of the blood-brain barrier. This figure is taken from [20] with the

permission of author. ................................................................................................................................... 4

Figure 1.3: The spectrum of ultrasound with respect to audible sound (top) and different

frequency ranges for diagnostic versus therapeutic applications (bottom). Adapted from [34]. ..... 6

Figure 1.4: Diagram depicts different ultrasound parameters including acoustic pressure

amplitude, pulse duration, pulse repetition frequency, total exposure time. ................................... 14

Figure 1.5: (A) The relationship between BBBD pressure threshold and US frequency. (B)

Constant mechanical index over the frequency range of 0.25-2 MHz. Adapted from [91]. ............ 15

Figure 1.6: BBBD enhancement as a function of burst length: (A) Study by McDannold et al. using

MRI technique at a fixed set of parameters: 0.69 MHz frequency, 1 Hz PRF and 0.5 MPa acoustic

pressure (adapted from [97]). (B) Study by Bing et al. using MRI technique at a fixed set of

parameters: 5.7 MHz frequency, 10 Hz PRF and 2.7 MPa acoustic pressure (adapted from [90]).

(C) Study by Choi et al. using optical imaging technique at a fixed set of parameters: 1.5 MHz

frequency, 10 Hz PRF and 0.46 MPa acoustic pressure (adapted from [98]) ..................................... 17

Figure 1.7: BBBD enhancement as a function of burst repetition frequency: (A) Study by

McDannold et al. using MRI technique at a fixed set of parameters: 0.69 MHz frequency, 10 ms

burst length and 0.5 MPa acoustic pressure (adapted from [97]). (B) Study by Choi et al. using

optical imaging technique at a fixed set of parameters: 1.5 MHz frequency, 20 ms burst length

and 0.45 MPa acoustic pressure (adapted from [98]). ........................................................................... 18

Figure 1.8: (A) BBBD enhancement as a function of total exposure duration. (B) Evaluation of

associated tissue damage using histological score (0-No damage; 1-Scattered microhemorrhages

accompanied with selective neuronal injury; 2-Large-sized hemorrhages with selective neuronal

injury and small necrotic areas; 3-Localized lesion). Adapted from [100]. ........................................ 18

Figure 1.9: A proposed model of MB oscillatory phases (i.e. compression and expansion) which

results in sonoporation phenomenon. Adapted from [106]. ................................................................ 20

Figure 1.10: Postulated physical mechanisms of MB cavitation and the associated biological

effects. This figure is taken from [20] with the permission of author. ................................................ 21

xii

Figure 2.1: Electric dipole moments in Weiss domains: (A) Exhibit random orientations before the poling process, (B) Become uniformly aligned during the poling process, (C) Remain well-aligned after the temperature is returned below the Curie point and the external voltage is removed. 36

Figure 2.2: An example of 1D piezoelectric crystal undergoing contraction and expansion phase

and the corresonding displacement at the two end nodes. .................................................................. 38

Figure 2.3: Distinction between thickness mode and lateral mode. .................................................... 39

Figure 2.4: Basic components of an ultrasound transducer [200] ........................................................ 40

Figure 2.5: (A) Dimension of thin cylindrical transducer; (B) Electrical impedance amplitude (top)

and phase measurements (bottom) of two thin cylindrical transducers of identical outer diameter

(do = 10 mm) and thickness (t = 1.5 mm) but different height: Transducer 1 (left, h = 0.85 mm);

Transducer 2 (right, h = 1.10 mm). Resonant peaks associated with 3 vibration modes (R – Radial,

T – Thickness, H – Height) are indicated. ............................................................................................... 43

Figure 2.6: Schematics of setup for optical hydrophone scan. ............................................................. 44

Figure 2.7: The in vivo US+MB assisted BBBD experimental set up with a cylindrical transducer.

(A-B) Side-view and top-view images demonstrate how the transducer is situated within the

cranial window; (C) The actual image of Wistar rat underneath the 2PFM system; (D) The

complete schematic of dorsal attachment of transducer and coverslip .............................................. 46

Figure 2.8: 2D contour and line profiles of the pressure field generated by Transducer 1 as

obtained from optical hydrophone scans. In these scans, z = 0 is set to the coverslip surface. The

first two rows show axial profiles (xz and yz slices), whereas the last row presents lateral profiles

(xy slices) at the focal region. For 2D contour profiles, as indicated, the first column corresponds

to height mode, whereas the second and third columns correspond to the thickness mode at the

fundamental frequency and the third harmonic, respectively. Line profiles at peak pressure are

extracted from the 2D profiles of the same kind (xz, yz, xy) and superimposed to compare the

focal zone location associated with each vibration mode. .................................................................... 49

Figure 2.9: 2D contour and line profiles of the pressure field generated by Transducer 2 as

obtained from optical hydrophone scans. In these scans, z = 0 is set to the coverslip surface. The

first two rows show axial profiles (xz and yz slices), whereas the last row presents lateral profiles

(xy slices) at the focal region. For 2D contour profiles, as indicated, the first column corresponds


fundamental frequency and the third harmonic, respectively. Line profiles at peak pressure are

extracted from the 2D profiles of the same kind (xz, yz, xy) and superimposed to compare the

focal zone location associated with each vibration mode. .................................................................... 50

xiii

Figure 2.10: Comparison of acoustic peak pressure vs. electrical applied power for thickness and

height modes of both transducers measured at the focal region ......................................................... 51

Figure 2.11: Left: A coronal brain section through imaging window shows localized distribution

of BBBD indicated by EB extravasation (arrow). Right: A close-up image of EB extravasation

region (outlined by dotted boundary) with measured dimensions. ................................................... 52

Figure 2.12: (A) An example of fast leakage of dextran-conjugated Texas Red under FUS+MBs

induced BBBD at 1.2 MHz frequency, 10 ms pulse duration, 1 Hz PRF, 120 s exposure duration

and 0.6 MPa pressure. (B) Quantitative analysis of fluorescent signal intensity associated with

intra- and extra-vascular compartments (represented by dashed and solid rectangle, respectively)

for the fast leakage shown in (A). (C) An example of slow leakage of dextra-conjugated Texas Red

under FUS+MBs induced BBBD at 0.4 MPa pressure, whereas other sonication parameters

remained similar to (A). (D) Quantitative analysis of fluorescent signal intensity associated with

intra- and extra-vascular compartments (represented by dashed and solid rectangle, respectively)

for the slow leakage shown in (C). Scale bar: 100 µm ........................................................................... 53

Figure 2.13: A summary of successful BBBD events, as well as the occurrence of two leakage

modes (fast vs. slow) at different acoustic pressure, while other sonication parameters were

maintained at 1.2 MHz frequency, 10 ms pulse duration, 1 Hz PRF and 120 s exposure duration.

....................................................................................................................................................................... 54

Figure 2.14: LDV measurement of the coverslip vibration when the transducer operates in height

mode (A-B) and thickness mode (C-D). In each sonication mode, the images on the left (A,C) and

right (B,D) show the two opposite phases (i.e. 1800 phase difference) of the coverslip vibration. . 58

Figure 3.1: Jablonski diagram to differentiate the single-photon (A) and two-photon (B) excitation

process. Adapted from [221]. .................................................................................................................... 63

Figure 3.2: A basic design of a two-photon fluorescent microscopy system. Adapted from [221]. 64

Figure 3.3: Comparison of excitation and fluorescence focal volume generated by single-photon

(left) and two-photon (right) imaging method [223] ............................................................................. 65

Figure 3.4: In vivo BBBD induced by FUS+MBs and monitored by 2PFM imaging. A)

Experimental timeline. B) 4D XYZT acquisition of 2PFM imaging. .................................................... 68

Figure 3.5: Data analysis of 2PFM data capturing fluorescent dye leakage upon BBBD. A) Depth

projection images illustrate the transient BBBD induced by MBs & FUS at 0.6 MPa (scale bar:

100µm). Sonication and MB injection occurred during the first 2 minutes while the vessels

remained impermeable to dextran conjugated Texas Red TR10kDa. As soon as sonication ceased,

disruption started at multiple vessels within the imaging FOV and extravascular signal increases

over time. B) Quantitative measurement of averaged fluorescent signal intensities associated with

xiv

intravascular and extravascular compartment over time. C) Permeability was evaluated

accordingly. ................................................................................................................................................. 70

Figure 3.6: Effect of acoustic pressure on permeability dextran conjugated Texas Red across the

BBB. Permeabilities were measured for all 20 cases of (A) TR10kDa and (B) TR70kDa delivered

across the BBB. Two-way ANOVA in combination with Bonferroni post-tests were used to

determine the statistical significance in permeabilities between different pressure level. .............. 73

Figure 3.7: Effect of substance size on enhanced BBB permeability. At each pressure, average

permeability constant (A) and average volume fraction (B) was compared between TR10kDa and

TR70kDa. Two-way ANOVA in combination with Bonferroni post-tests were performed as

multiple comparisons. ............................................................................................................................... 73

Figure 3.8: BBBD onset in relation to permeability and acoustic pressure. A) Inverse relationship

between BBBD onset and permeability. B) Inverse relationship between BBBD onset and acoustic

pressure. One-way ANOVA followed by Bonferroni’s Multiple Comparison Test confirms a

statistical significance in BBBD onset between 0.4 MPa and 0.6 MPa. ............................................... 74

Figure 3.9: Effect of vessel diameter on enhanced BBB permeability. A) Vessel size distribution in

correlation with permeability constant: large vessels (40-70 µm) are prone to slow leakage kinetics

and low permeability; whereas smaller vessels (10-40 µm) are subjected to fast leakage kinetics

and high permeability. B) Statistical analysis (two-tailed t test) indicates significant difference (p <

0.0001) in vessel size responsible for fast and slow leakage types. ..................................................... 75

Figure 4.1: (A) Contrast-enhanced axial T1w-MRI of focused ultrasound (FUS)-induced blood-

brain barrier disruption (BBBD). (B) Model geometry for simulation. (C) Model mesh with 706

triangular elements. (D) Permeability kinetics of free Dox (thin red) and bound Dox (thick blue)

across the BBB at FUS treatment region following a single-sonication (SS). ..................................... 82

Figure 4.2: (A) 2D map depicting spatial (x-direction) and temporal (y-direction) distribution of

intracellular Dox concentration at sonicated region and surrounding tissue followed a single-

sonication. Dash line represents the boundary between the sonicated region and the surrounding

tissue. (B) Spatial profiles of intracellular Dox concentration at 6h - 48h. ......................................... 88

Figure 4.3: Time-dependent spatial-mean free Dox (thin red) and bound Dox (thick blue)

concentration in the (A) extravascular-extracellular compartment and (B) intracellular

compartment of the sonicated region (solid) and the surrounding tissue (dashed). Note: The

range of y-axis in (A) is significantly lower than that in (B)................................................................. 90

Figure 4.4: (A)-(C) Effect of double-sonication (DS): Permeability kinetics of free Dox (A) and

bound Dox (B) are contrasted among DS of various intervals (10 min, 30 min, 60 min, 120 min)

against the Control and single-sonication (SS). (C) Time-dependent spatially-averaged profiles of

intracellular Dox concentration within the sonicated region are contrasted among Control, SS,

xv

DS10, DS30, DS60, DS120. (D)-(E) Effect of triple-sonication (TS): Permeability kinetics of free Dox

(D) and bound Dox (E) are contrasted among TS of various intervals (10 min, 30 min, 60 min, 120

min) against the Control (no sonication) and single-sonication (SS). (F) Time-dependent spatially-

averaged profiles of intracellular Dox concentration within the sonicated region are contrasted

among Control, SS, TS10, TS30, TS60, TS120. (G) Comparison of temporally-peaked spatially-

averaged intracellular Dox resulting from different sonication schemes: Single columns represent

the Control and SS case whereas double columns are associated with DS (filled) and TS (striped)

at various delayed intervals. ..................................................................................................................... 92

Figure 4.5: Temporally-peaked spatially-averaged intracellular Dox concentration within the

sonicated region as a function of Ktrans, an indicator of the blood-brain barrier (BBB) permeability

enhancement. Dotted line (red) represents therapeutic level of Dox resulting in a clinical response

for human tumors in vivo. SS, single-sonication; DS10, double-sonication of 10 minute interval;

TS10, triple-sonication of 10 minute interval. ......................................................................................... 94

Figure 4.6: Effect of injection mode (bolus injection and infusion over different durations) on Dox

delivery. (A) Temporally-peaked plasma. (B)-(D) Temporally-peaked spatially-averaged

intracellular Dox concentration within the sonicated region followed: (B) Single-sonication (SS),

(C) Double-sonication of 10 minute interval, (D) Triple-sonication of 10 minute interval. ............. 96

Figure A.1: Nanoparticle-labelled Definity® microbubbles on a glass slide are imaged by the

2PFM system ............................................................................................................................................. 112

Figure A.2: Second example of MB visualization and FITC500 leakage under FUS+MBs induced

BBBD at 0.6 MPa. (A) A vessel (red rectangle) was randomly selected from the 512x521 µm2

imaging FOV. Scale bar: 50 µm. (B) Time-lapsed XY images of the selected vessel over the

duration of 15 minutes. Sonication occurred during the first 2 minutes. At T = 3 min, a mixture of

green and blue signals start emerging in the extravascular space along a 50 µm vessel edge, as

indicated by the red arrow. Over the course of 3-15 minutes, the signal becomes more

pronounced. However, in contrast to the first example, it is interesting to note from this example

that the blue signal appears to dominate the green signal and exhibits a dotted-feature. We

speculated that the MBs might be shedding the NPs coating on their surface. .............................. 114

Figure A.3: Third example of MB visualization and FITC500 leakage under FUS+MBs induced

BBBD at 0.6 MPa. (A) A vessel (red rectangle) was randomly selected from the 512x521 µm2

imaging FOV. Scale bar: 50 µm. (B) Time-lapsed XY images of the selected vessel over the

duration of 15 minutes. Sonication took place during the first 2 minutes. At T = 8 min, a mixture

of green and blue signals start emerging in the extravascular space near the bifurcation point of

the vessel, as indicated by the red arrow. Similar to the second example, these current images also

exhibit a lower level of green signal relative to the blue one. Here, punctated extravasation is also

noticed. Another remarkable feature is the deformation of the vessel wall, as outlined by the

dashed oval................................................................................................................................................ 114

xvi

Figure B.1: Measured permeability of dextran conjugated FITC 500kDa across the compromised

BBB as induced at different acoustic pressure level of FUS exposure .............................................. 115

Figure B.2: Average permeability constants at each acoustic pressure level are compared among

TR10kDa, TR70kDa and FITC500kDa ................................................................................................... 116

Figure B.3: A) Enhanced permeability as a function of molecular weight for imaging tracers of

MW between 1-500 kDa. It is noted that the permeability constants were obtained from two

independent imaging modalities (DCE-MRI and 2PFM) but BBBD was induced at comparable

acoustic pressure and microbubble size. (B) Normalized permeability versus log(MW) displays a

linear relationship ..................................................................................................................................... 117

xvii

Glossary

2PFM Two-photon microscopy

AD Alzheimer’s disease

ATP Adenosine triphosphate

AuNP Gold nanoparticle

Aβ Amyloid-beta

BBB Blood-brain barrier

BBBD Blood-brain barrier disruption

BCNU Carmustine

BNCT Boron neutron capture therapy

BPA-f Boronophenylalanine-fructose

BTB Blood-tumor barrier

C3F8 Octafluoropropane

CNS Central nervous system

CSF Cerebral spinal fluid

CT Computed tomography

DALY Disability-adjusted life year

DEC-MRI Dynamic contrast enhanced MRI

DNA Deoxyribonucleic acid

DoF Depth of field

Dox Doxorubicin

DS Double sonication

DTPA Diethylenetriamine pentaacetate

EB Evans Blue

EC Endothelial cells

FOV Field of view

FUS Focused ultrasound

FWHM Full width at half maximum

Gd Gadolinium

H&E Hematoxylin and eosin

HCT Hematocrit

HD Huntington’s Disease

HIFU High-intensity focused ultrasound

HRP Horseradish peroxidise

Htt Huntingtin

ICA Intra-carotid administration

ISF Interstitial fluid

xviii

IV Intravenous

Ktrans Transfer coefficient

MB Microbubble

MI Mechanical index

MR(I) Magnetic resonance (imaging)

MRIgFUS MRI guided FUS

MTX Methotrexate

MW Molecular weight

NA Numerical aperture

NK Natural killer

PD Parkinson’s disease

P-gp P-glycoprotein

PRF Pulse repetition frequency

PZT Lead Zirconate Titanate

RNA Ribonucleic acid

ROI Region of interest

shRNA Short hairpin RNA

siRNA Small interfering RNA

SPECT Single photon emission computed tomography

SS Single sonication

T1w-MRI T1-weighted MRI

T2w-MRI T2-weighted MRI

TJ Tight junction

TMZ Temozolomide

TR Texas Red

TS Triple sonication

TUNEL Terminal deoxynucleotidyl transferase dUTP nick end labeling

US Ultrasound

WD Working distance

1

1 Background

1.1 Challenges to drug delivery to the brain

1.1.1 Current status & treatment for CNS pathology

1.1.1.1 Neurological disorders

According to the Global Burden of Disease report conducted by the World Health Organization,

neurological disorders represent 6.3% of the total global burden based on disability-adjusted life

years (DALYs) [1]. In fact, the disease burden from neurological disorders is greater than

HIV/AIDS and ischaemic heart disease (Figure 1.1). Prevalent neurological disorders include

epilepsy, Parkinson’s disease (PD), multiple sclerosis, Alzheimer’s disease (AD) and other

dementias. With the growth of the aging population, the financial cost implied by central nervous

system (CNS) diseases necessitates the increased establishment of prevention and treatment

programs.

With advancement in understanding of CNS disease etiology, there has been remarkable

progress in the discovery of novel therapeutic agents to treat CNS disorders. Thus far, several

therapeutics options have been developed: 1) molecules and antibodies for immunotherapy to

target specific regions in the brain [2]; 2) growth factors that either promote regeneration or

inhibit degeneration of cells to improve neurodegenerative conditions [3]; and 3) viral-vectors for

gene therapy to replace or correct mutated genes [4] .

Nevertheless, one of the key questions when treating CNS disorder is how to safely and

effectively deliver therapeutic agents into a specific region of the brain. Since our brains are

encased by the skull, directly accessing the brain tissue requires the use of invasive surgical

procedures such as the creation of burr-holes or removing parts of the skull. Non-invasive drug

delivery via intravenous (IV) administration is infeasible because the route used to access the

brain is blocked by the blood-brain barrier (BBB). While vascular delivery works for other tissues

and organs, more than 90% of existing pharmaceutical agents fail to breach the BBB and enter the

brain in therapeutically relevant quantities [5]. As discussed in Section 1.1.3, tremendous research

2

efforts are underway to develop viable approaches to direct therapeutic drugs into the CNS

safely, effectively and robustly.

Figure 1.1: Percentage of total disability-adjusted life years (DALYs) for various diseases and

neurological conditions. Adapted from [1]

1.1.1.2 Brain cancer

Brain tumors are the major leading cause of solid cancer death in children and young adults (ages

20-39) [6]. In 2013, there were approximately 23,000 new cases and approximately 14,000 deaths

from brain tumors in the United States. [7].

Brain malignancies are classified into primary tumors (originating in the brain) and

metastases (spreading from other cancerous regions such as the lung, breast and colon). Within

the tumor, the BBB is replaced by the blood-tumor barrier (BTB), which is less restricted and more

leaky. Treating primary and metastatic brain malignancies remains a challenge, however, due to

the heterogeneity of the BTB integrity, which implies variability and irregularity of the BBB

permeability across the tumor region. While the tumor core is comprised of leaky vessels,

elevated interstitial pressure drives chemotherapeutic agents outward and prevents these drugs

from reaching cancerous cells at an adequate concentration [8]. Increasing the delivery dose can

enhance the therapeutic efficacy in the tumor region, at the cost of elevating systemic cytoxicity

and potentially causing uncontrolled deleterious side effects in other organs when these drugs

are introduced into the circulatory system [9]. Furthermore, at the tumor periphery where the

3

BBB remains relatively intact, untreated infiltrated cells often lead to tumor re-growth [10]. Thus,

targeting malignant cancerous cells at this peripheral region is crucial for complete eradication of

brain tumors. To address these clinical concerns, one needs to develop techniques that allow for

local and non-invasive delivery of an anti-cancer agent to the CNS such that a sufficiently high

concentration can be achieved in the targeted brain tumor while the overall systemic

concentration is minimized.

1.1.2 Structure & function of the blood-brain barrier

As a specific vascular structure within the CNS, the BBB plays a major role in separating the brain

parenchyma from the circulatory system. Structurally, the BBB is composed of a single layer of

specialized endothelial cells (ECs) that is further protected by pericytes, microglia and astrocytic

endfeet at the basal surface (Figure 1.2). In contrast to systemic ECs, brain ECs exhibit several

distinctive properties such as an abundance of mitochondria [11], an increase in efflux activities

[12], an absence of fenestrations [13], and a reduction of pinocytosis [14]. These adjacent brain

ECs are securely joined by tight junction (TJs) proteins (e.g. claudins, occludins, cadherins,

cingulin) [15]. With their intracellular domains attaching to the EC cytoskeleton and their

extracellular domains forming homodimers with neighbouring proteins, these adhesion

molecules physically create a restrictive paracellular gateway.

The structural and morphological properties of the BBB allow it to effectively control and

maintain homeostasis of the neural environment. For instance, ion concentrations are closely

monitored by the BBB to provide proper synaptic and axonal signalling for neurotransmitters

[16]. Essential nutrients (e.g. oxygen, glucose) required by neurons are selectively delivered

across the BBB via distinct transcellular processes such as passive diffusion and active transport

(e.g. carrier-mediated and receptor-mediated transcytosis). Diffusion across EC membranes is

only applicable for molecules of low molecular weight (MW < 400Da) and high lipophilicity [17].

Small and hydrophilic nutrients are supplied to proximal neurons within the brain parenchyma

by active transport processes [18]. Carrier-mediated transport is initiated by the binding of

molecules to a membrane protein followed by the movement of the molecules from the EC’s

luminal to the basal surface. In contrast, receptor-mediated transport involves vesicle formation

at the cell membrane and encapsulation of targeted proteins that are released once the vesicle

reaches the basal membrane of the EC [19].

4

While the protective nature of the BBB plays a crucial role in maintaining optimal

conditions for neuronal function and prevents access of neurotoxins, it concurrently acts as a

barrier to restrict drug access from the blood compartment to the brain parenchyma. As a result,

most currently available therapeutic agents are blocked by the BBB. This is further exacerbated by

highly expressed efflux transporters (e.g. p-glycoprotein and multidrug resistant proteins) that

expel lipophilic drugs out of the brain parenchyma. Thus, the highly regulated BBB imposes

significant challenges in delivering potentially effective diagnostic and highly potent therapeutic

agents in the treatment of CNS diseases.

Figure 1.2: Illustrative structure of the blood-brain barrier. This figure is taken from [20] with the

permission of author.

1.1.3 Methods to bypass the BBB for drug delivery

Several methods have been proposed to overcome the BBB and deposit drugs into brain tissue.

Firstly, the BBB can be circumvented using convection enhanced delivery. This process is

achieved by either directly injecting the drug into a desired location (e.g. intracerebroventricular

and intracerebral injections) [21] or implanting a pharmaceutical wafer at the treatment site (e.g.

polymer wafer containing nerve growth factor) [22]. The major advantage of these methods is the

spatial localization of drug deposition. However, this route of delivery requires stereotaxically

positioning and neurosurgically tracking a needle into a potentially deep brain region. Such an

invasive procedure can expose patients to craniotomy-associated risks and post-treatment

adverse effects such as abnormal healing, brain edema, intracranial hypertension and infection

[23]. In addition, this convection enhanced delivery approach is constrained by the diffusion of

5

the drug from the injection site, which becomes a challenge for certain treatments that require

large coverage areas.

Secondly, the delivery of therapeutic compounds can be transcellularly assisted by active

transport processes. For example, small molecules such as oligonucleotides can be directed across

the BBB via carrier-mediated transport in a similar manner to glucose [24], whereas the delivery

of large molecules relies on receptor-mediated transcytosis in which the entire ligand-receptor

complex is shuttled through the BBB’s ECs [25]. Drawbacks associated with this transcellular

delivery method include the specificity and selectivity of drugs, low delivery efficiency as well as

potential drug resistance underpinned by efflux transporters [26] [27]. Moreover, this approach is

relatively expensive because it involves precise chemical modification (e.g. lipidization to the

polar end of therapeutic molecules to enhance their passive diffusion into the BBB) or novel drug

design (e.g. increased carrier-affinity or resemblance to endogenous ligands to promote

transcytosis).

Thirdly, the paracellular pathway can be exploited for the purpose of drug delivery by

temporarily disrupting the BBB tight junctions. As the most common method currently being

used in clinical trials (e.g. Phase I clinical trial in treating glioblastoma multiforme), intracarotid

infusion of hyperosmotic or hypertonic solutions (e.g. arabinose or mannitol) into

microvasculature results in the shrinkage of ECs and reversible disintegration of the TJs [28].

However, this “global” BBB opening effect could pose complications of increased vascular

volume, widespread toxicity exposure, heterogeneous opening and potentially significant edema.

Overall, the second and third pathways present a similar limitation; the lack of selective

and local drug deposition into the brain parenchyma. Therefore, a non-invasive and site-directed

delivery mechanism would be preferable. In 2001, Hynynen et al. introduced a novel technique of

disrupting the BBB using focused ultrasound (FUS) and microbubbles (MBs) [29]. By injecting

pre-formed MBs into the systemic circulation and irradiating the treatment region with a FUS

beam, BBB disruption (BBBD) can be induced to facilitate drug delivery to the brain in a non-

invasive, reversible, transient and localized fashion. As the core of this thesis work, the basic

components relevant to FUS+MBs induced BBBD method will be discussed in greater details in

Section 1.2.2.

6

1.2 FUS+MBs induced BBBD for drug delivery

1.2.1 Therapeutic ultrasound

As a form of non-ionizing radiation energy, ultrasound had been explored since the 1940’s for

uses in medical diagnostics and image-guided interventions. Operating in a frequency range of

2-20 MHz (Figure 1.3), ultrasound imaging has been extensively used for abdominal, cardiac,

obstetric, urological, cerebrovascular, ophthalmological and breast examinations [30]. The

therapeutic potential of ultrasound was first recognized via experiments by Wood and Loomis in

1924 and 1927, respectively [31]. In applying ultrasound at frequencies of 0.1-0.7 MHz and a

10W/cm2 output intensity, they observed burns and lesions in tissue samples. The earliest clinical

use of ultrasound was found in the treatment of sciatica in 1930 [32]. Two decades later, low

power ultrasound at 1 MHz frequency was employed in physiotherapy for the treatment of

tendinitis [33]. Ultrasound-assisted lithotripsy was also introduced as an alternative surgical

procedure for dissolving kidney stones.

Figure 1.3: The spectrum of ultrasound with respect to audible sound (top) and different frequency

ranges for diagnostic versus therapeutic applications (bottom). Adapted from [34].

7

The use of focused ultrasound in neurosurgery was first explored by Lynn and Putnam in

1942 [35]. In this investigation, the researchers applied high frequency and short wavelength

ultrasound in 37 animals to cause well-defined lesions on the cortical and subcortical areas in a

prompter manner as compared to radiation therapy. This work serves as the earliest evidence of

ultrasound therapy for brain tissue ablation. In the 1950’s, Francis Fry and William Fry used the

earlier observations and designed an intricate transducer system that enables the targeting of

deep brain structures such as basal ganglia [36] [37]. The established instrumentation had made it

possible to conduct cranial surgery procedures to treat patients with neurological disorders,

including Parkinson’s disease [38] and brain cancer [39]. These early investigations had laid the

foundation for the subsequent development of modern high-intensity focused ultrasound (HIFU)

surgical tools, from which BBBD (discussed in Section 1.2.2) has greatly benefited from.

In addition to those afore-mentioned clinical successes, other therapeutic applications of

ultrasound were developed for uterine fibroid ablation, cataract removal, hemostasis treatment,

transdermal drug delivery and bone fracture healing [33]. At typical operating frequencies of 0.5-

5 MHz, therapeutic ultrasound relies on the interaction between the acoustic energy and

biological tissues and the resulting thermal and non-thermal effects. Properties and

characteristics of each effect will be discussed in the following sections.

1.2.1.1 Thermal effects

Ultrasound induces heating in biological tissues by depositing energy along its propagation path.

Overall, the rate of temperature increase in the targeted tissue is dependent on both tissue

properties (e.g. the tissue ultrasound absorption coefficient, blood perfusion rate, thermal

diffusion and conduction) and the ultrasound exposure parameters (e.g. frequency, exposure

duration, intensity). Mathematically, the change in temperature due to ultrasonic energy

deposition can be described by the Pennes bioheat transfer equation below [40]:

(1.1)

where

is the rate of temperature change at a target location, is the tissue density, and is

the specific heat capacity of tissue. The first term on the right hand side of the above equation

accounts for the thermal conductivity of tissue, where is the thermal diffusivity constant. The

8

second term serves as a sink function driven by blood perfusion, where is the perfusion rate,

is the specific heat capacity of blood, and measures the temperature difference from

ambient. Lastly, is the rate of generated heat per unit volume and this source term is directly

related to ultrasonic exposure condition in Equation 1.2 below [41], [42]:

(1.2)

where is the absorption coefficient, is the ultrasound temporal-average intensity, is the

applied acoustic pressure amplitude squared, is the ultrasound absorption coefficient, and is

the speed of sound in target tissue.

Due to the temperature dependence of biochemical reaction and enzymatic activity, a

temperature rise at a specific tissue location will affect these reaction rates and introduce changes

in cellular structure. Study by Dickson and Calderwood have shown that the cellular and

physiological adverse effects are insignificant at temperatures below 400C [43]. However, when

the temperature was maintained at 400C over an extended period of time, irreversible

conformational changes were observed. Therefore, by raising the tissue temperature above 400C,

ultrasound-assisted hyperthermia can be leveraged for various therapeutic applications.

At moderate temperatures of 400C - 450C, mild hyperthermia is applied over timescales of

minutes to hours to enhance perfusion and modify immune response [33]. The concept of

ultrasound-assisted mild hyperthermia was also explored for cancer therapy by overcoming the

typical thermotolerance exhibited by tumor cells [44], [45]. Given an irregularly-vascularised

tumor where its microenvironment is hypoxic and acidic, the tumor cells could endure an

elevated heat-induced stress better than healthy normal cells. Under such circumstance, mild

heating was considered beneficial for radio- and chemo-sensitization by promoting oxygen

delivery within the tumor regions as well as deactivating proteins that are responsible for

restoring damaged DNA. In addition, mild hyperthermia also found its application in enhancing

and localizing drug release [46]–[48]. For instance, thermal-sensitive liposomes can be

administered intravenously yet the encapsulated drug would be restrictively released at the

heated tumor site.

9

When the high temperature above 500C is maintained over the duration of seconds to

minutes, rapid protein denaturation and subsequent cellular apoptosis will occur. This extreme

heating regime, also regarded as thermal ablation, is achieved via the use of HIFU (e.g. 1000

W/cm2 intensity and 0.5-7 MHz frequency). Such irreversible and permanent tissue coagulation

with a lesion size of a few millimeters has application for the minimally-invasive surgery of

uterine fibroid and other tumor masses (e.g. brain, breast, liver, bone and prostate) [49]. Via

concurrent magnetic resonance imaging (MRI) or ultrasound (US)-guidance, HIFU ablation of

uterine fibroids is an approved clinical procedure in several countries within North America,

Europe and Asia. Meanwhile, its potential use for thermal ablation in other organs is currently

being evaluated in several clinical trials [50], [51].

Following a large number of experiments on the thermotolerance of healthy and

tumorous cells, the breakpoint for cell killing effect has been determined to be around 430C. The

theoretical basis of this phenomenon is underlined by the Arrhenius activation energy. To

quantify the thermal-induced impact (e.g. amount of cell killing) on treated tissue upon different

heating regimens, the concept of thermal isoeffective dose with respect to a reference temperature

of 430C is considered. For a particular hyperthermia treatment with ultrasound-induced

temperature and exposure duration , the cumulative equivalent minutes at 430C ( ) can

be calculated via the following formula:

(1.3)

where for and for [52], [53]. The two distinct values of

indicate different rates of cell killing in response to the temperature deviation from the 430C

benchmark. In particular, this rate is doubled for every 10C temperature rise above 430C and

reduced by a factor of 4 for 10C temperature drop below 430C. Consequently, represents

the readjustment in thermal exposure, measured in minutes, in order to compensate for any

temperature difference from the 430C level. Using the convention, thermal doses have

been determined for various tissues. For instance, based on thermal ablation experiments on pig’s

colon, esophagus and muscle, values were found to be 30 min, 120 min and 240 min,

respective [54], [55]. In general, 240 has been used as a conservative threshold to achieve

thermal coagulation [56], [57].

10

1.2.1.2 Non-thermal effects

Interaction between ultrasound and biological systems can also occur via non-thermal

mechanisms such as radiation force and cavitation [34], [58], [59]. Upon the exertion of radiation

force, the acoustic wave can displace an obstacle along its path. The magnitude of radiation force

is dependent on the absorption coefficient of the medium, the temporal average intensity of the

acoustic wave at a specific location, and the speed of sound [60]. The resulting tissue motion

elicited by acoustic radiation force has been exploited for ultrasound diagnostic techniques such

as impulse and shear imaging [61], [62].

Another class of non-thermal effects arising from ultrasound is the acoustic cavitation

event and its secondary mechanical activities [63]. Cavitation can occur in the absence or presence

of injected microbubbles. Without introducing exogenous microbubbles, cavitation process is

ultrasonically-stimulated in a liquid material, where nucleation sites are rapidly initiated and

micron-sized pockets of gas are subsequently formed. In a complex medium such as tissue, the

exact nature of the cavitation phenomenon is still not fully understood. Nevertheless, under the

exposure of ultrasound, these microbubbles (either derived from the endogenous cavitation

nuclei or presented from the exogenously preformed solution) will respond accordingly to the

compressional and rarefactional acoustic pressure cycles. Depending on the pressure amplitude,

two regimes of cavitation exist: “stable cavitation” and “inertial cavitation”. In the former case,

when the pressure amplitude of the external acoustic field is not too high, these microbubbles can

undergo volume oscillations with moderate increases and decreases their radiuses. Such

oscillation of microbubbles results in micro-streaming of surrounding fluid, which exerts shear

stress on nearby tissue boundaries [31]. In the latter case, when the acoustic pressure amplitude is

sufficiently high, microbubbles would grow in volume and subsequently implode. Such transient

and violent collapse leads to various destructive mechanical impacts such as shock-waves,

microjetting, and the release of free radicals.

Overall, these acoustic cavitation events and secondary activities triggered by ultrasound

could have a biological impact on the cellular membrane and tissue integrity. As a characteristic

example of a therapeutic application based on ultrasonic non-thermal effects, lithotripsy involves

the generation of shock waves to break up kidney and bladder stones [64]. Typical lithotripter

devices operate at 150 kHz central frequency and generate 3000-5000 shock waves. These

11

waveforms typically consist of 1 μs long 50 MPa positive pressure impulse followed by a 4 μs

long 10 MPa negative pressure tail [33]. Using a low pulse repetition frequency of 1 Hz and a

peak pressure up to 80 MPa, the spatially-peaked temporally average intensity ( ) generated

by lithotripter device is below 0.1 W/cm2 [65]. When the temperature rise is less than 20C,

biological effects of thermal origin in exposed tissue are considered to be negligible. More recent

applications derived from acoustic cavitation include: sonothrombolysis to recanalize acute

intracranial arterial occlusion [66][67], sonoporation to induce cell membrane porosity for gene

transfection [68]–[71], and BBBD to facilitate drug delivery to the brain [29], [72]–[74]. Using an

occluded rabbit femoral arteries model, in vivo blood clot lysis has been conducted at 1.51 MHz

frequency, 1 ms burst length, 0.1% duty cycle and 20 MPa acoustic peak pressure [75]. Cavitation

thresholds considered for sonothrombolysis appear to increase linearly with frequency at a rate of

5.3 MPa/MHz [76]. Acoustic pressure threshold involved in sonoporation process has been

thoroughly explored by Deng and colleagues [70]. Using voltage clamp technique to measure the

transmembrane current at a single cell level, repeatability of sonoporation has been

demonstrated. For instance, by applying 1.5 MHz frequency, 13-40 µs pulse duration and 1.5-1.7

MPa pressure amplitude, pore formation (of 100-170 nm radius) can be induced on the membrane

of Zenopus oocyte [71]. For smaller sized cells such as HEK-293, the pressure threshold (at 1.25

MHz frequency and 4 µs pulse) was identified to be 0.17 MPa and the pore size was measured to

be 15-35 nm [69]. Such ability to control and quantify the degree of cell membrane disruption has

great implications for nonviral gene transfection and targeted drug delivery to cell [68]. Lastly,

microbubble-assisted BBBD was typically achieved at a pressure threshold between 0.2-1 MPa

over the frequency range of 0.25-2 MHz, as discussed in Section 1.2.2.3.

1.2.2 Basic components of FUS+MBs mediated BBBD

As mentioned earlier in Section 1.1.3, BBBD using FUS and MBs is a novel technique that allows

for the non-invasive opening of the BBB. Compared to other available methods (as reviewed in

Section 1.1.3), the promptness, localization and reversibility of this BBBD technique has positive

implications for drug delivery to a specific region of the brain. Since FUS+MBs mediated BBBD is

the central component of my thesis, the fundamental elements (e.g. microbubble properties,

ultrasound parameters) behind the technique will be reviewed in Sections 1.2.2.1 – 1.2.2.3.

Subsequently, the underlying cellular and physical mechanisms will be discussed in Sections

12

1.2.3 and 1.2.4. Preclinical results contributed by several research groups towards this field will be

summarized in Section 1.3.. Lastly, essential aspects that are pertinent to the implementation of

BBBD in a clinical setting will be highlighted in Section 1.4.

1.2.2.1 Ultrasound induced BBBD

During the early 1950s, Bakay et al. reported the first evidence of BBBD within the margins of

HIFU-induced lesions [77]. By exposing animal brains to ultrasound alone, later studies by Shealy

et at. (in 1965), Ballantine et al. (in 1960) and Salahuddin et al. (in 1988) also observed selective

alteration of the BBB [78]–[80]. In 1990, Patrick and colleagues were the first to propose the

utilization of this phenomenon for the delivery of chemotherapeutic agents into brain tumors

[81]. Nevertheless, these initial attempts at thermally-induced BBBD were accompanied by tissue

damage. The first experiment to demonstrate BBBD without associated damage was conducted

by Vykhodtseva et al. in 1995 [82]. In this study, the authors also reported the capture of

subharmonic emissions, suggesting the involvement of cavitation. Further investigations by

Mesiwala et al. and McDannold et al. aimed to minimize brain tissue damage and explored the

threshold for thermally-induced BBBD [83], [84]. Overall, these initial research efforts suggested

that cavitation- and thermal-based FUS can result in BBBD with potential applications for drug

delivery to the brain. However, its inconsistency in inducing BBBD raised the concerns regarding

safety and efficacy. In 2001, Hynynen et al. introduced for the first time the use of preformed

microbubbles to assist the FUS-induced BBBD process and lead to more consistent result.

1.2.2.2 Microbubbles assisted BBBD

Microbubbles were first invented in 1968 for myocardial contrast echocardiography applications

[85]. Since then, they have been extensively used as ultrasound contrast agents owing to their

strong backscattering properties (e.g. 20-30 dB enhancement), which arise from the acoustic

impedance mismatch between the gas core and the surrounding tissue [86], [87]. Early

generations of MBs, however, exhibited shortcomings such as unstable shell structures, rapid size

reduction, and limited half-life [88]. These technical limitations were later overcome by the use of

higher-molecular-weight and low solubility gases (such as octafluoropropane C3F8). Overall,

commercially-available MBs are biologically inert, physiologically-compatible and non-

aggregated. With similar rheologicical properties to red blood cells, MBs can be sufficiently

13

retained within the blood pool. MB shells are eliminated from the body via the

reticuloendothelial system, and the gases can escape via the lung.

Two commercial MBs being used in the United States and Canada are DefinityTM

(Lantheus Medical Imaging, MA, USA) and OptisonTM (GE Heathcare, Milwaukee, WI).

Definity® has a lipid shell, whereas Optison® has an albumin-coated shell [89]. Their mean

diameters are 1-3 µm and 2-5 µm, respectively. The recommended clinical dose for IV bolus

injection is 10 µl/kg for Definity and 0.5 ml (cumulative) for Optison [90].

Beyond their conventional contrast-enhanced imaging applications (e.g. diagnosis of

cardiovascular and renal diseases , characterization of highly-vascularised tumor [91]–[94]), MBs

have also been employed for therapeutic purposes, such as aiding the BBBD process at a reduced

acoustic energy (i.e. by two-orders of magnitude compared to HIFU alone) [29]. In contrary to the

earlier HIFU approaches that resort solely to ultrasound to induce BBBD (see Section 1.2.2.1), MB-

aided methods relies on the mechanical effects introduced by the preformed MBs onto the

cerebral vasculature. These physical mechanisms that MBs induce on the BBB will be discussed

in greater detail in Section 1.2.4.

When exposed to an ultrasound field, MBs undergo volume oscillation. In the linear

regime, the undamped natural resonant frequency of an encapsulated MB is given by [95]:

(1.4)

where is the radius, is the polytropic exponent ( = 1.06 for Definity TM), is the hydrostatic

pressure in the surrounding liquid (100kPa), is the liquid density (1000 kg/m) and is the

shell stiffness. Based on the ringdown signals of Definity microbubbles, experimental study by

Sun et al. had validated the relationship between the resonant frequency and bubble initial

diameter [96]. Overall, their research findings suggested that the effective frequency range for US

and MB interaction is between 2-10 MHz. As a result, past experiments involving MBs-assisted

BBBD in animal models have utilized ultrasound frequencies within the range of 28 kHz [97] - 8

MHz [98].

14

1.2.2.3 Ultrasound parameters for BBBD

Using small animal models (e.g. rodents, rabbits), various ultrasound parameters have been

investigated for BBBD. In optimizing the exposure condition for successful BBB opening,

different ultrasound parameters were closely examined including the acoustic frequency, in situ

pressure, burst duration, burst repetition frequency (PRF), and total exposure time (see Figure

1.4). The influence of each parameter on the BBBD outcome will be summarized in the following

sections.

Figure 1.4: Diagram depicts different ultrasound parameters including acoustic pressure amplitude, pulse duration, pulse repetition frequency, total exposure time.

Frequency: BBBD has been demonstrated over the frequency range of 28 kHz to 8 MHz

[97], [98]. For any transcranial application in a clinical setting, it is important to note the intrinsic

trade-off between frequency and quality of the focused beam. On one hand, a lower frequency is

desired owing to minimal attenuation. At the same time, precise targeting requires a tight focus,

which is achievable at a higher frequency. Thus, the ultrasound frequency is selected so as to

maintain a balance between low attenuation and a sharp focal volume.

Pressure amplitude: The relationship between the BBBD pressure threshold and US

frequency was originally explored by McDannold et al. [99]. Figure 1.5(A) displays the data

points obtained from BBB permeability enhancement experiments on rats and rabbits. Over the

frequency range of 0.25-2 MHz, the pressure threshold for BBBD was established to be linearly

correlated with the square root of frequency as given by the following equation:

(1.5)

15

Furthermore, is referred to as a “mechanical index (MI)”, which was found to be 0.46

MPa/MHz1/2 (Figure 1.5(B)). The authors suggested that this constant can be used as a

meaningful index to evaluate MBs and FUS-induced bioeffects on vasculature. In fact, studies by

other groups in conducting BBBD at various frequencies ranging between 0.3-8 MHz have shown

comparable MI values to that deduced from McDannold’s study. Table 1.1 summarizes the

pressure threshold and other sonication parameters being employed to achieve BBBD at a specific

ultrasound frequency. However, at a much lower frequency of 28 kHz, Liu at al. had reported an

MI constant as high as 4.78 MPa/MHz1/2 to induce BBBD [97]. One potential explanation for the

observed high MI value is the significant offset between the ultrasound frequency and the

optimal resonance frequency of microbubbles. In addition, since gross observation of EB

extravasation was used in this study to identify BBBD region, the detection technique might be

less sensitive than the MRI method used in other studies.

Figure 1.5: (A) The relationship between BBBD pressure threshold and US frequency. (B) Constant

mechanical index over the frequency range of 0.25-2 MHz. Adapted from [99].

16

Table 1.1: Comparison of BBBD pressure threshold data from other studies to the aggregated data by McDannold et al. [99].

Frequency (MHz)

BBBD Pressure Threshold (MPa)

Other Sonication Parameters

Reference Burst Length

(ms)

Pulse Repetition Frequency

(Hz)

Exposure Duration

(s)

0.3 0.49 10 1 30 [100]

1.0 0.39 0.25 400 20 [101]

1.5 0.55 10 1 30 [102]

1.5 0.67 20 10 30 [103]

2.2 0.52 14 10 180 [104]

5.7 1.2

20 10 30 [98] 6.7 1.4

8.0 1.6

Burst duration: By fixing the other parameters (e.g. frequency, pressure, PRF) and varying

the burst duration, several research groups have observed that a longer burst is associated with a

greater extent of BBBD [55], [57], [58]. For example, the study by McDannold and colleagues was

conducted at 0.69 MHz frequency, 1Hz PRF and 0.5 MPa pressure while the burst length was set

to 0.1, 1 and 10 ms. Such an adjustment in the burst length led to a 3-fold increase in the mean

MRI signal intensity enhancement (Figure 1.6(A)). Using a different set of acoustic parameters

(5.7 MHz frequency, 10 Hz PRF and 2.7 MPa applied pressure), Bing et al. also observed an

elevation in the MRI contrast-to-noise ratio when the pulse duration was stretched from 2 µs to 20

ms (Figure 1.6(B)). Using an optical technique to quantify the trans-BBB delivery of 3-kDa dextran

at 1.5MHz frequency, 10 Hz PRF and 0.46 MPa pressure, Choi et al. reported a subsequent

increase in the optical density upon extending the burst duration over a range of 0.033-30 ms

(Figure 1.6(C)). It is postulated that such an increase in pulse length would further the stimulation

period of non-thermal activities such as MB radial oscillation, displacement by radiation force or

streaming-derived velocity of the fluid [99]. Nonetheless, extending the burst length beyond 10

ms showed no further improvement in BBBD outcome [29]. One probable cause for this observed

17

saturation is the complete destruction of MBs during the first 10 ms, leaving no MBs for the

remaining burst to act on.

Figure 1.6: BBBD enhancement as a function of burst length: (A) Study by McDannold et al. using MRI

technique at 0.69 MHz frequency, 1 Hz PRF and 0.5 MPa acoustic pressure (adapted from [105]). (B) Study by Bing et al. using MRI technique at 5.7 MHz frequency, 10 Hz PRF and 2.7 MPa acoustic

pressure (adapted from [98]). (C) Study by Choi et al. using optical imaging technique at 1.5 MHz

frequency, 10 Hz PRF and 0.46 MPa acoustic pressure (adapted from [106])

Pulse repetition frequency (PRF): An initial study by McDannold and colleagues,

conducted at 0.69 MHz, with a 10 ms burst length and 0.5 MPa acoustic pressure, has suggested

that raising the PRF from 0.5 Hz to 5 Hz would produce little change on the level of BBBD (Figure

1.7(A)) [105]. For a different array of ultrasound parameters (e.g. 1.5 MHz frequency, 20 ms burst

length and 0.45 MPa acoustic pressure), Choi et al. demonstrated that BBBD enhancement occurs

when the PRF is raised from 0.1 Hz to 1 Hz and remains unchanged for PRFs above 1 Hz (Figure

1.7(B)) [106]. Goertz et al. attributed this ineffectiveness at high PRFs to a MB destruction-

reperfusion phenomenon [107]. In particular, the authors noted that applying a high pressure and

more bursts would effectively deplete more MBs in circulation. For instance, raising the acoustic

pressure from 0.44 MPa to 0.88 MPa has been found to double the depletion ratio as well as

lengthen the recovery time for MB reperfusion (e.g. from 0.6 s to 2.7 s). On the same note,

increasing the PRF from 0.2 Hz to 5 Hz also dramatically eliminates MB population from the

blood. Therefore, if the pulses are too closely-spaced, these MBs would not have sufficient time

for complete reperfusion.

18

Figure 1.7: BBBD enhancement as a function of burst repetition frequency: (A) Study by McDannold et al. using MRI technique at 0.69 MHz frequency, 10 ms burst length and 0.5 MPa acoustic pressure

(adapted from [105]). (B) Study by Choi et al. using optical imaging technique at 1.5 MHz frequency, 20

ms burst length and 0.45 MPa acoustic pressure (adapted from [106]).

Total exposure time: The effect of exposure time on the degree of BBBD has been

comprehensively investigated by Chopra and colleagues over the total duration of 30-1200 s

while fixing other acoustic parameters (e.g. 1.08 MHz frequency, 10ms burst length, 1 Hz PRF

and 0.38 MPa pressure amplitude) [108]. Figure 1.8(A) illustrates a greater extent of BBBD in

relation to longer exposures. However, as shown in Figure 1.8(B), exposure time extending

beyond 300 s leads to irreversible damage such as large-sized hemorrhages and neuronal injury.

Figure 1.8: (A) BBBD enhancement as a function of total exposure duration. (B) Evaluation of associated tissue damage using histological score (0-No damage; 1-Scattered microhemorrhages accompanied with selective neuronal injury; 2-Large-sized hemorrhages with selective neuronal injury and small necrotic

areas; 3-Localized lesion). Adapted from [108].

19

1.2.3 Cellular mechanisms

To gain insight into the distinctive cellular mechanisms underlined by FUS-induced BBBD,

researchers have extensively investigated the modulation in ultrastructures of the cerebral

vasculature using electron microscopy, histology (staining) and immunohistochemistry (labelled

antibodies). Thus far, two major routes of transport have been observed: transcellular and

paracellular passage.

Transcellular pathway: In a rabbit model treated with FUS, Sheikov et al. reported

evidence of fenestrations and the formation of cytoplasmic channel along with an increased

number of vesicles and vacuoles in the endothelial cells of the BBB in regions treated with FUS

[109], [110]. In particular, photomicrographs of sonicated domains indicated caveolae-assisted

transport of labeled IgG molecules. In another study, using horseradish peroxidise (HRP) as a

tracer, the authors compared the endothelial “pinocytotic densities” (i.e. the number of HRP-

positive vesicles per m2 of the cell cytoplasm) and discovered that arterioles are more active in

vesicular transport as compared to capillaries and venules [110]. Supporting studies by other

groups further suggested that the endocytosis and transcytosis processes be activated by the

upregulation of caveolin and clathrin [111]–[113]. Using independent characterization methods

(e.g. immunohistochemistry, western blot and transmission electron microscopy) to examine rat

brains exposed to FUS, Deng et al. established the optimal time point of caveolin-1 expression to

be 1 hour post treatment [112]. In another study on ultrasound-promoted gene transfection, Paula

and colleagues further noted the co-localization of labeled plasmid DNA with clathrin following

ultrasound exposure. The evidence indicates that DNA is taken up by cells via clathrin-facilitated

endocytosis [113]. Lastly, sonoporation has also been considered as another transcellular

transport mechanism that might take place during BBBD [68]–[71]. As hypothesized in Figure 1.9,

such disintegration and rupture of the cellular membrane is triggered by the pushing and pulling

forces during MB oscillatory phases (i.e. compression and expansion phases) [111], [114]. Using

fluorescent microscopy and a high-speed camera, these researchers were able to visualize

FUS+MB-evoked pore formation and the subsequent influx of tracer molecules such as dextrans

and propidium iodide across an in vitro endothelial cell membrane.

20

Figure 1.9: A proposed model of MB oscillatory phases (i.e. compression and expansion) which results

in sonoporation phenomenon. Adapted from [114].

Paracellular pathway: Using immunoelectron microscopy to trace the distribution of IV-

delivered gold nanoparticles on a sonicated area of a rabbit brain, Sheikov et al. offered initial

evidence of macromolecule transport from the blood to brain compartment via interendothelial

clefts [109]. In a follow-up study, using lanthanum chloride (139Da) and HRP (40kDa) as imaging

tracers, the authors consistently observed the excursion of these particles along the

interendothelial gaps [115]. Furthermore, by monitoring the expression level of several TJ

proteins (e.g. occludin, claudin-5 and zonula occluden-1), the research group detected reduced

expression of these TJ proteins within 1-2 hours post-sonication followed by a gradual recovery

within the 4-24 hour window. Applying FUS+MB treatment to tumor-bearing rats, Zhang et al.

similarly observed a noticeable decrease in both the mRNA and protein levels [116]. Such

characteristics substantiate the temporary disassembling of the TJs, which implies their

prevalence in transport regulation at the BBB. Given its role in supporting tissue homeostasis and

neuronal network integrity, one expects that gap junction breakdown would trigger a cascade

effect on the cellular signaling response. In fact, a study by Jalali et al. validated this hypothesis by

demonstrating that Akt signaling pathway activation is exhibited by neuronal cells within the

compromised BBB region [117]. Another study by Alonso et al. also reported a high expression

level of phosphorylated Connexin 36 and 43 that were attributable to neurons and astrocytes in

response to the imbalanced extracellular homeostasis [118].

1.2.4 Physical mechanisms

Although the precise physical mechanisms of BBBD remain undetermined, plausible

explanations of MB behaviour inside vessels have been proposed. Depending on the acoustic

energy, two cavitation regimes could exist: stable cavitation (driven by low acoustic pressure)

21

and inertial cavitation (triggered by high acoustic pressure). Under a stable cavitation condition,

bioeffects that MBs exert on the proximal microvasculature can originate from various physical

phenomena, as illustrated in Figure 1.10. For instance, the acoustic radiation force, which is

generated by an extended pulse, could push the MBs towards the vessel wall [98], [119]. Periodic

contraction and expansion of MBs could perturb the surrounding fluid and give rise to a

secondary acoustic microstreaming effect, which in turn stimulates the cellular membrane and

modulates the ion channel [120], [121]. In addition, MB oscillations induce shear and

circumferential stresses which could transiently disrupt the TJs [29], [122]. On the other hand,

inertial cavitation is associated with MB implosion which leads to shockwaves, microjet and the

release of free radicals as illustrated in Figure 1.10. As a consequence, such violent events could

cause damage to the ECs lining of the BBB [123]–[125].

Using two-photon microscopy for real-time analysis of rodents’ cortical tissue under FUS

treatment, distinct characteristics of BBB opening have been identified based on the patterns of

IV-administered fluorescent tracers leaking out of the cerebral vasculature [126], [127]. Upon the

exposure to low acoustic pressure, a slowly diffusive outflow of the dye was noted along the

entire segment of the affected vessel, suggesting transcellular passage as a potential mechanism.

Conversely, increasing in situ acoustic pressure yielded a temporally-fast and spatially-localized

disruption. Such focal leakage of fluorescent agent from a single point of the vessel could be

attributed to the paracellular route. Lastly, two-photon microscopic images of sonicated rodent

brain also revealed the vasoconstriction of arterioles and transient deferral of blood flow for as

much as five minutes [128], suggesting that vasospasm may be involved during the BBBD

process.

Figure 1.10: Postulated physical mechanisms of MB cavitation and the associated biological effects. This

figure is taken from [20] with the permission of author.

22

1.3 Pre-clinical progresses of BBBD-based drug delivery

1.3.1 Delivery of macromolecules & therapeutic agents

Since the introduction of the FUS+MBs mediated BBBD method, numerous preclinical

experiments have been conducted to verify the entry of a particular macromolecule into the brain

parenchyma. For instance, via independent monitoring techniques (as highlighted in Table 1.2),

the passage of different imaging tracers of various sizes across the BBB has been examined.

Table 1.2: Summary of different imaging tracers and monitoring techniques being used for BBBD preclinical study

Monitoring Technique Imaging Tracer References

MRI Omniscan (573 Da)

Magnevist (938 Da)

[129]

[29]

Histology Trypan Blue (960 Da) bound to albumin (70 kDa)

Evans Blue (960 Da) bound to albumin (70 kDa)

[130]

[131]

Immunology Horseradish Peroxidase (40 kDa) [115]

Optical imaging Dextran-conjugated Fluorophore (3kDa to 2 MDa) [132]

In moving FUS-based BBBD towards clinical translation, researchers have recently

demonstrated the effectiveness in delivering numerous therapeutic agents. These studies were

conducted using various disease models which replicate neural-deficit conditions. In the

following sections, different clinical applications that can be realized with FUS treatment will be

reviewed.

1.3.1.1 Chemotherapy

The treatment of newly-diagnosed and recurrent brain tumors has been examined using different

chemotherapeutic agents such as Doxorubincin (Dox, 544 Da), Carmustine (BCNU, 214 Da),

Temozolomide (TMZ, 194 Da) and Methotrexate (MTX, 545 Da).

During their initial attempt at FUS-enhanced delivery of liposomal Dox (e.g. Doxil, Ben

Venue Laboratories, OH, USA), Treat et al. achieved therapeutic efficacy on 9L glioma tumor

bearing rats by demonstrating delayed tumor progression and an enhanced population survival

rate [133], [134]. A recent study has further suggested that repeated treatments of

23

FUS+MBs+Doxil (e.g. 3 weekly sessions) would further improve the treatment outcome [135]. In

particular, histological analyses have shown a complete eradication of tumor cells in 4 out of 8

animals that underwent multiple FUS+MBs+Doxil sessions.

As another well-established chemotherapeutic agent, BCNU has been used in randomized

trials for the treatment of glioblastoma multiforme [136], [137]. Aiming to restrain its toxicity

within the tumor regions, Liu and colleagues have employed the FUS+MBs technique as a

targeting strategy and demonstrated increased access of the chemo-drug across both normal BBB

and leaky BTB by 340% and 202%, respectively, with respect to the non-FUS-assisted approach

[73]. The authors showed that combining FUS+MBs+BCNU yields the greatest outcome in

prolonging the survival of tumor-implanted rats.

Acting as an alkylating adjuvant, TMZ has been augmented to radiotherapy to control

tumor progression [138]. Given its current application in phase-III clinical trials, minimizing

systemic toxicity is desirable for patient recovery. This objective has been examined by Wei et al.

via the use of FUS+MBs as an adaptive procedure. In lowering the TMZ injected dose, these

researchers were able to enhance TMZ deposition with detectable CSF/plasma ratio

improvement, from 22% to 39% [139]. Additionally, the authors reported a noticeable

suppression effect on tumor volume as well as prominent survival rate.

Lastly, the efficacy of MTZ in treating medulloblastoma has been explored by Mei et al.

[131]. Despite its anti-metabolic nature, which had previously shown effectiveness in inducing

apoptosis of medulloblastoma cell lines, MTZ exhibits low permeability across the BBB due to its

hydrophilicity. As a conventional delivery route, intra-carotid administration (ICA) has seen an

increased risk of hemorrhage, micro-embolism and cerebrospinal fluid outflow [17], [140].

Therefore, to circumvent these ICA-associated side effects, Mei and colleagues have proposed the

use of FUS+MBs while delivering MTZ intravenously. A comparative analysis has confirmed a

3.7 fold increase in MTX concentration by FUS+MBs+MTZ.

1.3.1.2 Novel agents for targeting brain tumor & metastasis

Among potential antibody-based anticancer agents, Herceptin (i.e. trastuzumab, 148 kDa) has

been recognized for its therapeutic effectiveness on breast cancer patients [141]. To expand its

utilization for treatment of brain metastasis, Kinoshita et al. employed MRI-guided FUS to breach

24

the BBB and facilitate a local deposition of this epidermal growth factor receptor-2 monoclonal

antibody at a specific region of the rodent brain [142]. As a result, successful opening of the BBB

enabled the delivery of Herceptin to the sonicated area at a significant dose (e.g. 3,257 ng/g of

tissue at 0.8 MPa acoustic pressure).

With the ability to target tumor-associated antigens to exert their cytolytic activity on

tumor cells, natural killer (NK) cells have been proposed for immune therapy. To increase the

access of these NK cells across the BBB, Alkins et al. induced BBBD on tumor-bearing rats by

applying typical FUS parameters (551.5 kHz frequency, 0.33 MPa average peak pressure) [143]

[144]. In tracking the distribution of these iron-loaded NK cells via high-resolution MRI and

histological techniques, the authors noted a 10-fold increase in the NK cell population that

successfully breached the BBB and entered the brain parenchyma. This novel demonstration

highlights the potential application of FUS+MBs in the treatment of brain metastasis and solid

malignancies.

Given their biocompatibility, nanoparticles (NPs, 10–50 nm) exhibit promising clinical

utility in diagnostic and therapeutic applications, such as gene targeting, cancer imaging and

thermal-based tumor ablation [145]–[154]. To gain access to the brain tissue, gold NPs (AuNPs)

were initially introduced into circulation via IV injection and subsequently delivered to a specific

brain region using FUS+MBs mediated BBBD [155]. With an effective deposition of AuNPs into

the brain tissue, this preclinical proof-of-concept study implies huge potential of AuNPs-

facilitated therapy for CNS disorders. In a different study, the concept of drug-bearing magnetic

NPs (MNPs) has been explored by Deng and Huang [156]. By attaching epirubicin to MNPs and

applying magnetic targeting along with ultrasound-induced BBBD, the authors demonstrated

that such combined strategy can further enhance the drug concentration inside the tumor cells

and thus improve the overall outcome of targeted delivery of chemotherapeutic agents to the

brain.

Recently, FUS-induced BBBD has also been proposed as an alternative to the conventional

use of osmotic agents to deliver Boronophenylalanine (e.g. 10B-enriched L-4-

boronophenylalanine-fructose, or BPA-f) across the BBB. As a crucial catalyst for boron neutron

capture therapy (BNCT) [157], BPA-f are designed for selective binding to malignant cells and

thereby increased sensitivity to neutron radiation. Considering the effectiveness of BNCT relying

25

on a full-range penetration of BPA-f into both tumor core and infiltrating cells, enhancing the

permeability of this catalyst across the BBB is clinically advantageous. Using FUS+MBs for BBBD

and mass spectroscopy for BPA-f concentration quantification, Alkins and colleagues have

demonstrated the potential role of sonication in improving 10B accumulation in 9L gliosarcoma

tumor bearing rats. These findings highlight clinical implications of FUS-induced BBBD for BNCT

in particular, and the treatment of glioblastoma in general [157] .

1.3.1.3 Immunotherapy for Alzheimer’s disease (AD)

The accumulation of amyloid-beta (Aβ) peptide has been identified as the primary pathology of

AD; therefore, reducing this plaque burden is a potential alleviation and treatment strategy.

However, efforts in delivering diagnostic substances to label the affected regions and therapeutic

materials to disassemble these toxic amyloid plaques have been challenged by the intact BBB. To

increase the penetration of these immunotherapeutic agents for specific targeting of AD,

Raymond et al. originally proposed the use of FUS+MBs [158] . Jordao and colleagues further

demonstrated this concept by treating plaque-bearing transgenic mice with BAM-10, an anti-Aβ

antibody, in conjunction with MRI-guided and transcranial FUS therapy [159].

Immunofluorescence analysis of cortical tissue at 4 hours, 2 days and 4 days confirmed the

binding of BAM-10 to Aβ-plaques. Most importantly, in comparing the plaque counts between

the control and sonicated hemisphere 4-days post treatment, the authors noticed a significant

plaque clearance in the latter case. This promising evidence suggested that MRIgFUS

immunotherapy could potentially offer an effective treatment for AD patients.

1.3.1.4 Gene therapy for Huntington’s disease (HD)

HD is a genetic disorder that is associated with cell death in specific regions of the brain

including the caudate, putamen and cerebral cortex. HD is typified by advancing symptoms

including emotional turmoil, cognitive loss and physical deterioration [160]. Despite the current

lack of clinical treatments, suppressing the mutant Huntingtin (Htt) gene via short hairpin RNA

(shRNA) or small interfering RNA (siRNA) has been considered as a prospective approach [161]

[162] [163]. To facilitate a transient and localized delivery of siRNA to the affected brain region,

Burgess et al. have employed FUS+MBs to deliver the therapeutic agent across the BBB while

targeting the BBB opening via MRI [164]. Based on a dose-dependent analysis, this pioneering

study not only showed evidence of effective Htt knockdown, but furthermore identified the

26

optimal siRNA dose that could sufficiently reduce the Htt expression level. These findings are

instrumental for non-invasive and directly-targeted gene therapy treatment of HD.

1.3.1.5 Stem cell therapy

Neural stem cell therapy has been explored for the treatment of various neurological conditions

such as Parkinson’s disease, amyotrophic lateral sclerosis, traumatic brain injury, spinal cord

injury and ischemic stroke [165]–[167]. The clinical benefit of this therapeutic approach relies on

the migration of neural stem cells to the brain regions that exhibit cell loss and successive

differentiation of the progenitors into specialized cells (e.g. neurons, astrocytes, and

oligodendrocytes) for replacement and restoration [168],[169]. While these rescuing cells have

been shown to improve disease symptoms (e.g. enhanced cognition, recovered and functionality),

the conventional transplantation methodology is deemed invasive and complicated due to the

required craniotomy. To avoid this surgical procedure, Burgess and colleagues have suggested

that stem cells can be presented into the blood stream and safely transferred from circulation into

the brain parenchyma via the precise application of transcranial FUS to a targeted brain structure

(e.g. striatum, hippocampus) [170]. In this investigation, successful passage of iron-labeled neural

stem cells across the BBB was established via MR imaging. More importantly, their survival at the

transplanted site and progressive differentiation into neurons was corroborated by

immunohistochemical analyses at 4 hours and 24 hours post treatment, respectively. These

important findings serve as another constructive proof-of-concept for the clinical translation of

FUS.

1.3.2 Safety evaluation

As a dictating factor for its clinical adoption, safety evaluation on FUS+MBs induced BBBD

technique must be conducted. In this section, short- and long-term risks associated with FUS

treatment will be reviewed.

1.3.2.1 Reversibility of BBB opening

The time window that the BBB remains open post-sonication is a key parameter for treatment

planning. Given the instrumental role of TJ complexes in protecting the BBB, their expression

levels at discrete time points post-FUS treatment have been monitored via immuno-electron

27

microscopy [115]. From their analyses, Sheikov et al. noticed the reduction of TJ proteins at 1-2

hours and subsequent restoration at 6-24 hours. Beyond the protein expression level, work by

Shang et al. examined the presence of claudin-5, occludin and zonula occluden-1 at mRNA level

[171]. These authors also observed the absence and recovery level of these proteins at 3 and 12

hours, respectively. At a macroscopic level, the permeability of the BBB was monitored via

contrast-enhanced T1 weighted (T1w)-MRI and Evans Blue (EB) detection [112], [172]. By

repeating the scans at different time points between 0 and 24 hours after treatment, as well as 1

week follow-up, these studies confirmed that the permeability peaked around 1-2 hours post

sonication and the BBB gradually resealed itself within 4-12 hours. Aside from the conventional

MR contrast agents Gadolinium (Gd) and tissue labelling with EB to indicate BBB opening, Marty

and colleagues further mapped the closure half-time as a function of macromolecule size by

experimenting with a wide range of hydrodynamic diameters (1-65 nm) [173]. In doing so, the

authors established an inverse relationship between the size of these contrast agents and their

passage duration across the BBB. For instance, under a similar FUS+MBs exposure condition, the

BBB gateway would pass small molecules (e.g. 1 nm) for up to 10 hours, while permitting the

passage of large iron oxide particles (e.g. 25 nm) for only a few minutes.

In general, these experimental evidences commonly showed the gradual reversibility of

BBB with a complete closure within 24 hours provided that commercial MBs (e.g. Definity,

SonoVue, Optison) and optimized acoustic parameters (e.g. mechanical index < 0.46, < 5 % duty

cycle, PRF ~ 1 Hz) are being employed. However, using custom-made monodispersed MBs with

large diameter (e.g. 4-5 µm and 6-8 µm), Samiotaki et al. observed a drastically delayed BBB

closure of up to 5 days along with noticeable cell loss and tissue damage [174].

1.3.2.2 Short-term & long-term effect on tissue

Using TUNEL assay and H&E (Hematoxylin and eosin) staining to evaluate the bio-effects

underlined by microvasculature damage at the sonication zone, McDannold et al. confirmed that

neuronal damage and associated detrimental effects can be avoided [175]. Nevertheless, they

noted small extravasations and mild inflammation that lasted for 3 days post treatment but

diminished after 4 weeks. In another study by Hynynen and colleagues, histology analyses were

performed on the brain tissue of treated animals to determine the short-term (e.g. 7-9 days after

BBBD) and long-term impacts (e.g. 4-5 weeks survival post FUS treatment) [176]. Aside from a

28

few scattering erythrocytes in the vicinity of disrupted blood vessels, short-term analyses

exhibited no obvious lesions, while histological findings regarding long-term investigation also

confirmed no adverse effects on neurons.

1.3.2.3 Extravasation of blood-borne material

There have been concerns that, other than the drug itself, the widening of the TJs would

unintentionally lead to leakage of blood-borne materials such as chemokines and erythrocytes

into the targeted region [102]. However, it has been shown that such unfavourable phenomena

could be minimized by using optimal sonication parameters. In particular, at a frequency of 1.5

MHz, using a 10 ms burst length, 1 Hz PRF and 30 s total duration, Liu et al. assessed the

inflammatory response at sonication sites via histological fluorescent antibody staining [177].

When the brain region was sonicated at 2.45 MPa, a noticeable macrophage infiltration was

observed at the treatment zone, whereas sonications at 1.1 MPa did not result in any apparent

monocyte accumulation.

Albumin is another blood-borne substance that could inadvertently enter the brain upon

BBBD. This blood-borne entity could pose a neurotoxic threat to the surrounding parenchyma.

Nevertheless, by labelling albumin with EB, Alonso et al. demonstrated the prompt clearance of

albumin primarily by microglial and astroglial cells at 30 minutes post FUS treatment. At a fast

uptake rate over the course of 24 hours, these specialized cells conjecturally phagocytise

albumins, thereby protecting neurons from apoptosis [178].

1.3.2.4 Behavioral tests

Behavioral assessments of preclinical BBBD-based treatment were originally carried out by

Howles and colleagues [104]. In their study, BBB opening was induced in mice with Definity MBs

and transcranial unfocused ultrasound (2.15 MHz frequency, 0.8 MPa pressure). The mice were

tested and scored for the level of activity, arousal and responsiveness. In particular, the

behavioural test parameters included: body position, spatial locomotion tail elevation, touch-

escape, grip strength, righting reflex, etc. According to the total behaviour score, these mice

displayed a 13% decline in their response at 3 hours post sonication but were restored to the

original state within 1 day. Recently, McDannold et al. extended the safety evaluation from

rodents to primates [179]. Upon undergoing repeated FUS+MBs induced BBBD at the central

29

visual regions, rhesus macaques were assessed for their behavioral and visual deficits. A

longitudinal study over several weeks demonstrated that these primates, which had been

previously trained to conduct complex visual acuity tasks, continued to perform well in cognitive

tests and display functional recovery following each FUS treatment. Lastly, behavioral

evaluations have been directed in transgenic AD mice [180]. Based on a 1 month follow-up study

after exposing the hippocampus to multiple MRIgFUS treatments, the novel arm Y-maze test was

performed to assess the changes in spatial memory of AD mice. Along with an evident reduction

in the amyloid-beta plaque load, a significant increase in exploration time at the novel arm

strongly indicated the effectiveness and safety of the FUS treatment in reversing the

abnormalities inflicted by AD.

1.4 Clinical translation of BBBD-based drug delivery

In order to shift BBBD-based drug delivery from the pre-clinical phase into clinical

implementation, it is necessary to consider two of the following technical components:

transcranial exposure of ultrasound and real-time monitoring of BBBD.

1.4.1 Transcranial ultrasound exposure

While depositing ultrasound energy into small animals (e.g. rodents and rabbits) for preclinical

investigation has been feasible, translating FUS treatments to humans and large animals (e.g. pigs

and primates) has been deemed challenging due to the more intricate skull structures [29], [181],

[182]. Given the considerable thickness and varying density of the skull, transmitted energy

experiences substantial loss at the interface between skull and soft tissue [183]. While skull

removal would circumvent this technical challenge, such a procedure is clinically invasive. As an

alternative, focusing the acoustic energy across the skull has been proposed [184]. In theory, a

sharp focus can be achieved using a single-element hemispherical transducer. However, the non-

uniformities in the skull thickness and density lead to wave-front distortion and aberration,

which in turn causes defocusing and enlargement of the beam at the target region. To circumvent

this undesired effect, three approaches have been considered: (1) shear wave transmission [185];

(2) frequency lowering [186]; and (3) utilization of phased array transducers [187], [188].

30

The first solution relies on the reduced distortion of shear wave in comparison to its

longitudinal counterpart [185]. By deliberately raising the incident angle of the incoming waves

above Snell’s critical angle, the arriving wave can be partially transformed into a shear mode

within the bone layers and be subsequently switched back to a longitudinal mode in the soft

tissue. Owing to its better impedance match between the shear wave speed and the speed of

sound in water, shear mode suffers less distortion and refraction. The second scheme requires the

use of sub-megahertz frequencies to suppress the frequency-dependence scattering and

absorption processes [176], [189], [190]. However, low frequency ultrasound leads to a larger focal

spot at the treatment site as well as increases the potential of standing waves, which has been

suggested to be the cause of hemorrhaging in the brain [191]. The last method offers improved

functionality by driving multiple elements of the transducer array at individually controllable

phase and amplitude such that constructive interference of ultrasound waves is attained at a

desired location. However, deemed as tedious, complex and costly, such a procedure requires

patient-specific CT scans for skull thickness calculations, computational implementation of wave

propagation simulations and multi-channel electronic design for individual element control.

Incorporating those aforementioned concepts (i.e. lowered frequency and phase-array),

the ExAblate 4000 (Insightec, Israel) system prototype contains 1024 elements and operates at 220

kHz and 650 kHz. At the former frequency, the resulting full width at half maximum (FWHM) of

the focal volume is 3.0 mm and 5.8 mm in lateral and axial direction, respectively. Originally

designed for thermal ablation procedures, this MRI-compatible system has been adapted for

preclinical investigations of BBBD applications in which significantly lower transmitted power is

required and therefore the associated skull heating effect is diminished. Currently, the system has

been employed in several clinical transcranial FUS studies conducted on large animals such as

pigs and primates [192], [179]. The findings from these studies are encouraging for clinical trials

on humans.

1.4.2 Assessment methods of FUS+MBs induced BBBD

In vivo monitoring techniques to detect BBBD and evaluate its efficacy have been an on-going

research topic. MRI is the most commonly used non-invasive imaging modality. Unlike thermal

ablation treatment that relies on MRI thermometry to directly monitor temperature elevation at

the lesion, BBBD is undetectable via temperature change. Instead, three contrast-enhanced MRI

31

methods are prevalently engaged during BBBD investigation: T1 weighted (T1w), T2 weighted

(T2w) and dynamic contrast enhanced (DCE) MRI. Contrast enhanced T1w-MRI corroborates

successful BBB opening by detecting the leakage of MR contrast agent extravasating out of the

compromised vessels and entering the brain parenchyma [29], [175], [193]. Conversely, T2w-MRI

sequence is applied to identify any hemorrhage resulting from FUS treatment [102], [194]. As a

quantitative method to assess the BBB permeability, DCE-MRI facilitates the measurement of

transfer coefficient, Ktrans. By measuring Ktrans at discrete time points (e.g. every 1.5 hour), Park et

al. have characterized the kinetics of BBB opening and estimated the closure half-life from 1 to 3.4

hours [195]. In addition, using DCE-MRI, Vlachos and colleagues have established the

relationship between permeability and externally controlled parameters (e.g. acoustic pressure

and microbubble diameter) [196]. Lastly, DCE-MRI was employed by Marty et al. to examine the

dependence of BBB closure time on substance size via the delivery of MR contrast agent with

different hydrodynamic diameters [173].

At a superior sensitivity as compared to MRI, micro-single photon emission computed

tomography/computed tomography (micro-SPECT/CT) has been used in combination with

99mTc diethylenetriamine pentaacetate (99mTc-DTPA) for a quantitative evaluation of BBBD [197],

[198]. As a common non-diffusible radio-tracer in nuclear medicine, 99mTc-DTPA has been

clinically adopted to assess BBB permeability breakdown and predict the neurologic outcome of

patients with acute stroke [199]. Therefore, this brain imaging modality can be potentially applied

to spatially resolve the BBB disrupted area. However, the major limitation exhibited by both MRI

and micro-SPECT/CT techniques is the reliance on a specific contrast agent which possesses

particular pharmacokinetics parameters rather than those relevant to the drug of interest. This

drawback can be overcome with the use of two-photon fluorescent microscopy, as further

discussed in Section 1.5.

1.5 Research objectives

1.5.1 Problem statement

Despite tremendous progress being made thus far to characterize FUS+MBs induced BBBD,

understanding of vascular and cellular mechanisms associated with the therapy is still

incomplete, as is our knowledge of the temporal-spatial distribution of therapeutic agents upon

32

their delivery to the brain parenchyma. The overarching goal of this thesis is, therefore, to bridge

this gap with the use of two-photon fluorescent microscopy (2PFM). This imaging modality, in

comparison to MRI, offers sufficient temporal and spatial resolution to track real-time, transient

biophysical behaviours of BBB opening at a microscopic level. Additionally, 2PFM exhibits many

other benefits over conventional fluorescent microscopy (e.g. confocal , total internal reflection)

such as a low level of tissue phototoxicity, a reduced effect of photobleaching, and an ability to

access deep into the tissue (e.g. beyond 500 µm in vivo) [200]. Lastly, with the design

advancement in tuning the physical and chemical properties of novel drugs (e.g. MW,

lipophilicity) and labelling them with fluorophores, these agents can be readily visualized for

their in vivo interaction with the physiological system. In fact, 2PFM has been used as a principal

method in many pre-clinical studies to measure drug transport parameters in tumors [201],

observe structural and functional changes in live animal kidney [202] and track the development

of thrombus in mouse mesenteric vessels [203]. Considering all of these prominent features,

2PFM is a suitable tool for investigating the microscopic mechanisms of FUS+MBs mediated

BBBD.

1.5.2 Specific aims

Structurally, the thesis work consists of three components:

1. Transducer design and characterization for dorsal-based focused ultrasound exposure and

two-photon fluorescent microscopic imaging of BBBD

The first aim was to design a transducer system that can be integrated into the existing two-

photon fluorescent microscope and concurrently allows for effective sonication on a murine

brain. In recognizing the advantage of dorsal-based over ventral-sonication, which will be further

explained in Chapter 2, a robust design was proposed and evaluated for its compatibility with

both two-photon imaging and FUS treatment. Given its cylindrical configuration, two modes of

vibration (thickness and height) generated by the transducer were analyzed and contrasted to

select the most suitable mode for BBBD application. Following these characterization studies, the

intended system was tested on a rat model and histology analysis was performed to validate the

transducer performance in vivo.

33

2. Quantitative assessment of BBB permeability based on two-photon fluorescent microscopic

imaging

Using the robust transducer design from above, we conducted in vivo BBBD experiments on rat

brain while acquiring time-lapsed microscopic evidence of induced BBB opening exhibited by the

microvasculature volume. BBBD was induced at various acoustic pressure levels (ranging from

0.2 - 0.8 MPa) and leakage kinetics were examined for fluorescent dyes of different molecular

weights (e.g. 10 kDa and 70 kDa dextran-conjugated Texas Red). A data processing pipeline was

established to expediently visualize these 4D-XYZT images and automatically segment the

vessels within the imaging field of view (FOV). Lastly, a quantification method was developed to

systemically assess the permeability of these fluorescent markers exiting the compromised vessels

and entering the parenchyma at the event of BBBD.

3. Model the delivery of Doxorubicin o the brain in the context of FUS-induced BBBD

By applying the quantitative measures of BBB permeability as input parameters, a

pharmacokinetics model was constructed to closely reflect pertinent transport mechanisms

involved in the BBBD-based drug delivery process and subsequently predict the spatio-temporal

profile of drug distribution. With Doxorubicin being a well-established chemotherapeutic agent

whose permeability enhancement across the BBB was recently realized via the use of FUS+MBs,

we are motivated to tailor the pharmacokinetics model towards this drug. By adopting pre-

determined pharmacokinetic parameters of Doxorubicin (e.g. plasma half-life, diffusion constant,

and cellular transport rate) along with their extrapolated permeability kinetics constant, we

mathematically obtained the drug concentration in the extravascular compartment and validated

against the experimentally available data [195]. Beyond the validation between the simulation

and experimentation results, we further examined several clinical treatment factors such as

sonication scheme, the change in BBB permeability and injection mode. Essentially, we envision

that this mathematical framework can potentially be used to guide future treatment planning in

the context of FUS+MBs induced BBB opening.

1.5.3 Thesis outline

Chapter 1 provides a brief literature review on the current status of drug delivery to the brain for

treatment CNS pathology. Subsequent components of the chapter are dedicated to describing the

34

groundwork behind FUS+MBs methodology to safely breach the well-protected BBB for drug

delivery purpose. Thereafter, overall objectives and specific aims for my thesis work are

rationalized and established. Details on implementation and results obtained from these three

aims will then be addressed in Chapters 2, 3, and 4, respectively. In addition, the basic principles

of transducer will be covered in Chapter 2 whereas the fundamental background on two-photon

microscopy will be included in Chapter 3. Resting on the findings in the preceding chapters,

Chapter 5 presents a summary, clinical perspective, and future directions. Lastly, additional

results from two side projects exploring “fluorescent microbubbles” and “substance size

dependence of permeability” will be covered in the Appendix A and B, respectively.

35

2 Transducer design and characterization

for dorsal-based FUS exposure and 2PFM

imaging of in vivo BBBD in a rat model 1

Constituting Aim 1 of the thesis, this chapter lays out the design of a transducer system that

enables concurrent 2PFM imaging and FUS application on a dorsal surface of a rat brain. Prior to

in-depth discussion of the transducer design, an overview on basics of transducer will be

provided. Thereafter, the system specification and complete characterization results in both in

vitro and in vivo studies will be the main topic of discussion.

2.1 Overview on basics of transducer

2.1.1 Piezoelectric effect

The generation of ultrasound is underlined by the piezoelectric effect. This phenomenon was first

noted by Pierre and Jacques Curie in 1880 when an electric charge was produced upon the

application of pressure to Rochelle salt (i.e. quartz crystal) [204]. In particular, the generated

voltage is proportional to the applied mechanical pressure. Reversely, when an electric field is

applied across its surface, the piezoelectric crystal would undergo mechanical expansion and

contraction due to the induced stress.

Two most commonly seen piezoelectric materials are Barium Titanate (BaTiO3) and Lead

Zirconate Titatnate (PbZrTiO3, or PZT). At a microscopic level, the crystalline grains are consisted

of internally aligned domains called Weiss domains. Below a temperature point referred to as

“Curie point”, these Weiss domains exhibit random polarization as illustrated in Figure 2.1(A).

For instance, the “Curie point” for BaTiO3 and PZT are 1200C and 3200C, respectively [205], [206].

During the poling process, the crystal’s temperature is raised to the “Curie point” and a strong

voltage is applied across its electrodes. As a result, these Weiss domains are uniformly aligned

1 Adapted from the article: Nhan T, Burgess A, Hynynen K. Transducer design and characterization of dorsal-based ultrasound

exposure and two-photon imaging of in vivo blood-brain barrier disruption in a rat model. IEEE Trans Ultrason Ferroelectr Freq

Control 2013; 60: 1376-85.

36

along the poling axis (refer to Figure 2.1(B)). Once the temperature is lowered below the Curie

point and externally applied voltage is removed, the electric dipole moments of these Weiss

domains remain polarized as shown in Figure 2.1(C).

Figure 2.1: Electric dipole moments in Weiss domains: (A) Exhibit random orientations before the poling process, (B) Become uniformly aligned during the poling process, (C) Remain well-aligned after the temperature is returned below the Curie point and the external voltage is removed.

2.1.2 Resonance frequency

When a piezoelectric crystal is exposed to an external alternating voltage, the electrical energy

will be converted to the mechanical energy, resulting in the vibration at the driving frequency.

This motion reaches its maximum at a resonant frequency that is directly related to the dimension

of the crystal. To derive this natural resonance frequency, let’s consider a 1D example as depicted

in Figure 2.2. Particle displacement at any point along the material is governed by the wave

equation:

(2.1)

where is the maximum oscillation amplitude; whereas is the wave number, which can be

expressed in term of wavelength , frequency and speed of sound as following:

(2.2)

37

and is the angular frequency, which is directly related to the frequency as following:

(2.3)

When the entire 1D crystal line undergoes vibration (i.e. expansion and contraction), its two ends

always oscillate in the opposite direction. For instance, let’s consider time when the crystal

expands, the particle displacement at and x can be written as:

(2.4)

(2.5)

Combining Equation 2.1 and 2.4, we obtain:

, where n=1, 3, 5... (2.6)

Similarly, combining Equation 2.1 and 2.5, we obtain:

, where n=1, 3, 5... (2.7)

Given the conditions in (2.6) and (2.7), we deduce:

, where n=1, 3, 5...

, where n=1, 3, 5... (2.8)

Hence, the natural frequency at which the crystal vibrates depends on its length L and the speed

of sound inside the material.

38

Figure 2.2: An example of 1D piezoelectric crystal undergoing contraction and expansion phase and the corresonding displacement at the two end nodes.

2.1.3 Modes of vibration

Given the degree of freedom in a 3D crystal, the vibration can occur in three directions and the

resonant frequency associated with each mode will depend on its dimension as shown in

Equation 2.7. However, with respect to the poling direction, modes of vibration can be

categorized as thickness and lateral. Let’s assume the electrodes are located on the top and

bottom surface and the poling direction is parallel to the z axis, as illustrated in Figure 2.3. If the

vibration occurs along the z-axis and the poling direction, it is classified as the thickness mode.

Meanwhile, if the vibration axis is along the x-axis or y-axis (i.e. perpendicular to the poling

direction), it is regarded as lateral mode. Due to its matching alignment with the poling direction,

the former mode exhibits a greater piezoelectric constant as compared to the latter. In other

words, under the same applied voltage, the thickness mode would generate a stronger response.

For instance, for PZT material, the piezoelectric constant associated with the thickness mode is

about twice that of the lateral mode (e.g. 271x10-12 m/V versus 131x10-12 m/V for PZT-5E, and

603x10-12 m/V versus 303x10-12 m/V for PZT-7B [207]).

39

Figure 2.3: Distinction between thickness mode and lateral mode.

2.1.4 Transducer structure and backing

The basic components of an ultrasound transducer are depicted in Figure 2.4. The PZT crystal

might be consisted of a single or multiple active elements. A cable wire allows to transmit and

receive electrical signal to the PZT crystal. An insulation case is used to isolate the transducer

from the electrical signal interference. For imaging transducers, a backing material is required for

two main reasons. First, it serves as a damping material to suppress the ringing effect of the PZT

crystal after an applied electrical impulse. The shorten pulse length helps improve the axial

resolution. Second, the backing material increases the bandwidth (i.e. the range of frequencies

around the central frequency). While it is desirable to embed a backing material for imaging

transducer, therapeutic transducers are typically air-backed as they require a narrow frequency

bandwidth [208]. A matching layer in front of the PZT element helps reduces the impedance

mismatch and enables the maximal transmission of ultrasound from the emitter to the receiver

medium (e.g. biological tissues). For therapeutic applications, either aqueous gel or water can be

used to couple the ultrasound into the tissue. When the transducer surface is flat, aqueous gel can

be used as a coupling medium. For a curved transducer (e.g. concave, spherical) and complex

tissue geometries, water is used.

40

Figure 2.4: Basic components of an ultrasound transducer [208]

2.2 Research motivation

FUS in combination with MB contrast agents has been established as an effective method for

BBBD and targeted drug delivery to the brain [29], [209]. As mentioned in Section 1.4.2, MRI has

been the primary imaging modality used to guide and evaluate FUS treatments. In particular,

gadolinium-based MR contrast agents are delivered intravenously to show macroscopic signal

enhancement in T1-weighted MR images in the area of BBBD [210]. These images have been

analyzed to compare the extent of BBBD as a function of MB parameters (e.g. concentration [105],

MB size [211], MB shell type [89], injected method [212]) and ultrasound (US) parameters (e.g. US

frequency [176], [99], pulse repetition frequency (PRF) [105], [106], burst length [105], and

exposure time [108]). However, the trade-off between temporal and spatial resolution inherent to

MRI technologies impedes the exploration of microscopic and dynamic mechanisms associated

with BBBD. In contrast, 2PFM imaging modality, as previously outlined in Section 1.5.1,

addresses the present limitations exhibited by MRI and allows for visualization of BBBD at a

cerebral vascular level.

The combination of simultaneous optical imaging and application of FUS, however,

encounters several challenges. First, in contrast to the MR setting, where both the transducer and

positioning system can reside in the MR magnet bore without interfering with the imaging

process [194], in 2PFM, the lack of complete transparency will limit the depth over which

experiments are made. Second, the spatial constraints of the microscope limit the size and

geometry of the transducer.

41

Raymond and colleagues integrated an ultrasound transducer into a commercial two-

photon microscope for in vivo optical imaging and ultrasound treatment in the murine brain

[126]. In the experimental setup, a dorsal cranial window was exposed to the microscope

objective lens for imaging, while a spherical transducer located on the ventral side emitted

ultrasound toward the imaging zone. Using this design, the authors showed proof-of-concept and

identifiable evidence of different BBB opening response from microvessels (i.e. microdisruption

vs. slow disruption) [128]. However, this setup is prone to variability and distortion of the FUS

field at the imaging zone due to scattering and absorption of ultrasound along its propagation

through various anatomical structures (e.g. trachea, skull, and/or air cavities).

Here, we aim to design and characterize a transducer which attaches to the dorsal surface

of the animal skull and is conducive to simultaneous 2PFM imaging. In brief, the integrated

design is comprised of a thin cylindrical transducer coupled to a coverslip that is attached to a

dorsal cranial window of the animal. Aside from a spherical configuration, a cylindrical

transducer also facilitates geometric focusing at the far-field. Theoretical calculation of the far

field response from a continuous-wave excitation of a cylindrical transducer inferred that its

lateral pressure response in the Fraunhofer zone can be described as a Bessel function [213]. With

its improvements in both lateral resolution and depth of field, cylindrical transducers were used

extensively in ultrasound imaging [214][215]. In our work, however, cylindrical transducer was

utilized as a therapeutic device.

In this study, we have characterized and confirmed that a uniform FUS profile with

adequate acoustic pressure for BBBD can be achieved while maintaining optical transparency

necessary for in vivo imaging of the animal cerebral microvasculature. The transducer dimensions

were optimized and the two vibration modes (thickness and height) of the transducer

configuration were characterized by quantitatively mapping the acoustic pressure field profile

using an optical fiber hydrophone. These analyses provide a baseline to evaluate the suitability

for using each of these vibration modes for BBBD in vivo. Lastly, we obtained in vivo real-time

2PFM images of successful BBBD with FUS+MB using this transducer design. The localized

extravasation of Evan’s Blue (EB) following BBBD demonstrates the transducer’s usability and

robustness for in vivo investigations.

42

2.3 Materials & methods

2.3.1 Transducer design

Thin ring-shaped transducer of specific height was cut from a lead zirconate titanate (PZT-4) tube

that was custom-ordered from EDO, Salt Lake City, UT, USA (Figure 2.5(A)). The tube was

radially-poled and electrodes were affixed to the inner and outer wall surfaces. The inner

diameter of 8.5 mm allowed the transducer to fit around the objective lens, whereas the outer

diameter of 10 mm was small enough to fit onto one hemisphere of the rat’s cranial window. Two

transducers with different heights (0.85 mm for Transducer 1 and 1.10 mm for Transducer 2) were

created and tested.

After fabrication, transducers were matched to a 50 Ω impedance and 0o phase load at the

desired frequency to maximize the electrical driving power. Each mode of each transducer

resonated at a different frequency and thus required its own custom matching circuit. The

transducer was driven by a function generator (33210A, 10 MHz Function/Arbitrary Waveform

Generator, Agilent, Palo Alto, CA, USA) and a 53 dB RF power amplifier (NP Technologies Inc.,

Newbury Park, CA, USA). The applied forward and reflected RF-power during sonication was

recorded using an in-house manufactured power meter. The electrical forward and reflected

powers, used as an indicator of any change in the loading condition of transducer, were

monitored during any characterization studies as well as during the in vivo experiment.

2.3.2 Transducer characterization

For quantitative mapping of the US field pressure profile generated by a transducer in different

vibration modes (height vs. thickness mode), fiber-optic hydrophone scans were performed. In

the case of the thickness mode, both the fundamental frequency and the third harmonic were

examined. At each mode of vibration, the transducer was excited by a sinusoidal burst signal (25

cycles/pulse, PRF = 100 Hz) at 5 electrical input power levels (ranging from 0.10 to 2.12 W). The

ultrasound pressure field radiating from the transducer was measured with a fiber-optic

hydrophone with an active element diameter of 10 μm (Precision Acoustics, Dorchester, UK).

43

Figure 2.5: (A) Dimension of thin cylindrical transducer; (B) Electrical impedance amplitude (top) and phase measurements (bottom) of two thin cylindrical transducers of identical outer diameter (do = 10 mm) and thickness (t = 1.5 mm) but different height: Transducer 1 (left, h = 0.85 mm); Transducer 2 (right, h = 1.10 mm). Resonant peaks associated with 3 vibration modes (R – Radial, T – Thickness, H – Height) are indicated.

As depicted in Figure 2.6, the transducer was attached to a coverslip with cyanoacrylate

glue. To avoid direct contact with the transducer during the measurements, we let the transducer

float on the water surface and gently clamped the electrical cables that were soldered onto the

transducer electrodes to hold it in place. A 45 × 50 × 120 cm3 water tank was lined with 13 mm

thick anechoic rubber (Global Rubber Products, Scarborough, Ontario, Canada) to minimize any

acoustic reflections from the tank walls. The tank was filled with degassed, deionized water

(Resistivity > 16 MΩ-cm), with dissolved oxygen level below 1 ppm. The hydrophone was affixed

vertically in the water tank underneath the coverslip by a Parker/Velmax three-dimensional

scanning system (Parker, Hannifin, PA, USA; Velmax Inc., Broomfield, NY, USA). Using a

44

Cartesian coordinate system, the lateral origin was set at the center of the ring-shaped transducer,

whereas zero depth (z = 0) was set at the coverslip surface. The scanned area and step size were

controlled by a software interface written in LabView (National Instrument, Austin, TX, USA),

and communication with the positioning system occurred via a parallel port. For lateral profiles

(in xy-plane) of the pressure field at focal depth, a 4 x 4 mm2 area centered at the origin of ring-

shaped transducer was scanned with the spatial resolution of 0.2 mm in both the x and y

directions. For depth field profiles, a 4 x 10 mm2 area cross-sectioned at y = 0 (for xz-plane) and x

= 0 (for yz-plane) was scanned with the lateral spatial resolution of 0.2 mm and the axial spatial

resolution of 0.5 mm. The measurements were captured on a digital oscilloscope (TDS 3012B,

Tektronix, Richardson, TX, USA) at a temporal sampling rate of 12.5 MHz and averaged over 16

traces at each spatial location. The averaged voltage signal and associated (x, y, z) location were

saved on the computer via a General Purpose Interface Bus (GPIB).

Figure 2.6: Schematics of setup for optical hydrophone scan.

2.3.3 Experimental setup for BBBD induction and in vivo 2PFM imaging

Male Wistar rats with a weight range of 120-200 g were used. All the procedures were approved

by the institutional Animal Care and Use Committee and were in accordance with the Canadian

Council on Animal Care. Anesthesia was induced in the animal using 5% isoflurane which was

45

reduced to 2% for the duration of the experiment and monitored using a pulse oximeter. The

animal was positioned in a stereotaxic frame and the tail vein was cannulated for injection of the

fluorophore and MB contrast agents. A 5 mm diameter cranial window was created on the right

side of the skull about 3 mm lateral from the midline and 3 mm posterolateral to bregma to allow

for 2PFM imaging (Figure 2.7(A)-(B)). The cranial window was covered with 1% agarose and a

coverslip attached to the thin ring-shaped transducer was secured on top with cyanoacrylate

glue. The overall design is compatible with 2PFM imaging since the ring-shaped transducer

serves as a well for the water immersion objective lens (Olympus XLPLN, 25x magnification,

numerical aperture (NA) = 1.05, working distance (WD) = 2 mm) while the optical transparency

required for imaging is maintained by using a glass coverslip. Upon the completion of

craniotomy and transducer attachment, the animal was transferred onto the microscope stage

(FV1000MPE, Olympus, Tokyo, Japan) (Figure 2.7(C)). Dextran-conjugated Texas Red

fluorophore (10 kDa MW, Invitrogen, Burlington, ON, Canada) was injected through the tail vein,

allowing for visualization of the cerebral vasculature. Two-photon excitation of the fluorophore

was achieved with a mode-lock Ti:Sapphire laser unit (Mai-Tai, Spectra-Physics, Mountain View,

CA, USA) at an 810 nm center wavelength, 100 fs pulse length and 80 MHz pulse repetition rate.

The complete schematic of the dorsal attachment of the transducer and coverslip is depicted in

Figure 2.7(D).

Prior to sonication, Definity MBs (Lantheus Medical Imaging, Billerica, MA, USA) (0.02

ml/kg) were injected through the tail vein at the onset of sonication. Typical sonication

parameters were fixed at 1.2 MHz frequency, 10 ms pulse duration, 1 Hz PRF, 120 s exposure

duration. Meanwhile, the acoustic pressure was varied at four different levels: 0.2, 0.4, 0.6 and 0.8

MPa.

Lateral images of 512 x 512 pixels (1 µm spatial resolution and 8 µs/pixel temporal

resolution) were captured below the cortical surface in a stacking-mode to a cortical depth of up

to 300 µm in 10 µm increments. The 300 µm stacking distance is sufficient to capture a good

representation of vessel sizes for a proper investigation of BBBD. 2PFM imaging took place

continuously throughout the course of the experiment. 2PFM imaging sessions typically lasted

for 15-30 minutes. After that, the animal was injected with EB at the concentration of 100 mg/kg

and was survived for 15 minutes before being euthanized. The brain was removed and

46

submerged in formalin solution for 24 hours. The brain was then cut in the coronal plane through

the centre of the cranial window to evaluate EB extravasation.

Figure 2.7: The in vivo US+MB assisted BBBD experimental set up with a cylindrical transducer. (A-B) Side-view and top-view images demonstrate how the transducer is situated within the cranial window; (C) The actual image of Wistar rat underneath the 2PFM system; (D) The complete schematic of dorsal attachment of transducer and coverslip

2.4 Results

2.4.1 Transducer fabrication

We created 2 transducers from cylindrical pieces of PZT-4 with different heights that would be

suitable for use with 2PFM. The cylindrical configuration yields three modes of vibration: radial,

thickness, and height, of which resonant frequency (f) is dictated by the physical size of each

dimension (s) and the speed (co) at which an US wave traverses via the relation f = co/2s

[216][204] . For a typical PZT material, co = 4000 m/s [213]. The outer diameter and thickness

were preset by the original material, so the transducers had fixed resonant frequency in radial

and thickness mode. We varied the frequency in height mode by cutting and lapping the

transducers to a desired height. In order to accommodate the bulky objective lens, the height was

limited to 1.10 mm, therefore we chose to set the height for Transducer 1 at 0.85 mm and

Transducer 2 at 1.10 mm.

47

The expected resonant frequency was calculated and summarized in Table 2.1.

Empirically, resonant frequencies were further verified by measuring the amplitude of the

electrical impedance and phase angle of each transducer using a network analyzer (E5061B, ENA

Series, Agilent Technologies, Santa Clara, CA, USA) as shown in Figure 2.5(B). Since the radial

mode resonant frequency of 200 kHz is less prevalent for BBBD application due to the potential

standing wave effect [217], it was excluded from further analysis. The resonant frequencies for

the first transducer were fT = 1.2 MHz and fH = 2.35 MHz for thickness and height mode,

respectively. For the second transducer, these frequencies were fT = 1.2 MHz and fH = 1.82 MHz.

Table 2.1: Summary of resonant frequencies from three vibration modes of the two fabricated transducers

Dimension

Transducer 1 Transducer 2

Size (mm) Expected Resonant

Frequency (MHz) Size (mm)

Expected Resonant

Frequency (MHz)

Outer Diameter (do) 10.00 ± 0.05 0.20 ± 0.01 10.00 ± 0.05 0.20 ± 0.01

Thickness (t) 1.50 ± 0.05 1.33 ± 0.04 1.50 ± 0.05 1.33 ± 0.04

Height (h) 0.85 ± 0.05 2.35 ± 0.14 1.10 ± 0.05 1.82 ± 0.08

2.4.2 US pressure resulting from different mode of vibration

2D pressure profiles obtained from optical hydrophone scans for Transducer 1 were normalized

and are shown in Figure 2.8. The first two rows are results from axial scans (xz and yz slices), and

the last row shows results from lateral scans (xy slice). As indicated, the first column corresponds


fundamental frequency and the third harmonic, respectively. To compare the difference in focal

zone location associated with each mode of vibration further, line profiles passing through the

peak pressure were extracted from the 2D profiles of the same kind (xz, yz and xy) and

superimposed onto the same graph as shown in the fourth column in Figure 2.8. Similar results

obtained from optical hydrophone scan for Transducer 2 are presented in Figure 2.9.

48

In all cases, axial US pressure profiles reveal central main lobes (containing 70% of the

acoustic energy) accompanied by symmetrical side lobes. However, for both transducers, the

axial line profiles from the two modes of vibration display a difference in the location and extent

of the focal spot. In particular, both the fundamental frequency and the third harmonic in

thickness mode result in a significantly shallower depth of field (DoF) as compared to height

mode. In thickness mode, the acoustic peak pressure resides at 1-1.5 mm below the coverslip

whereas the acoustic pressure yielded from the height mode focuses at a depth of 4-4.5 mm. In

addition, both the fundamental frequency and the third harmonic in thickness mode yield a

tighter axial focus when comparing the axial extension (along z-direction) of each contour profile.

For example, the blue contour indicates that 70% of the acoustic energy is deposited within 4 mm

depth in the case of thickness mode. In contrast, this same amount of acoustic energy is spread

over a 10 mm depth range when the transducer is driven in height mode. Considering the 2PFM

imaging field extends only 1 mm in depth from the coverslip surface, such differences in the

location and extent of the focal spot are important.

Lateral US pressure profiles at the focal zone (e.g. scanned at z = 1-1.5 mm for thickness

mode, and z = 4-4.5 mm for height mode) reveal a circularly symmetrical and uniform focal spot

in both modes for both transducers. Most importantly for these cases, lateral FWHMs are found

to be 500 µm in diameter, which sufficiently overlaps with the typical 512 x 512 μm2 lateral FOV

of the 2PFM imaging field, ensuring the majority of the scanned area is exposed to US.

2.4.3 Output acoustic pressure

In addition to a qualitative comparison of the US field profiles resulting from two different

vibration modes of the same transducer, absolute peak acoustic pressures were also measured at

the focal region (i.e. (x, y, z) = (0, 0, 4-4.5 mm) for height mode and (x, y, z) = (0, 0, 1-1.5 mm) for

thickness mode as a function of applied electrical power. Results presented in Figure 2.10 exhibit

a square-root increase of pressure with power, as expected. It is noted that data shown in Figure

2.10 only include height and thickness modes at the fundamental frequencies, whereas the third

harmonic frequencies have been excluded due to its relatively low signal to noise ratio.

49

Figure 2.8: 2D contour and line profiles of the pressure field generated by Transducer 1 as obtained from optical hydrophone scans. In these scans, z = 0 is set to the coverslip surface. The first two rows show axial profiles (xz and yz slices), whereas the last row presents lateral profiles (xy slices) at the focal region. For 2D contour profiles, as indicated, the first column corresponds to height mode, whereas the second and third columns correspond to the thickness mode at the fundamental frequency and the third harmonic, respectively. Line profiles at peak pressure are extracted from the 2D profiles of the same kind (xz, yz, xy) and superimposed to compare the focal zone location associated with each vibration mode.

50

Figure 2.9: 2D contour and line profiles of the pressure field generated by Transducer 2 as obtained from optical hydrophone scans. In these scans, z = 0 is set to the coverslip surface. The first two rows show axial profiles (xz and yz slices), whereas the last row presents lateral profiles (xy slices) at the focal region. For 2D contour profiles, as indicated, the first column corresponds to height mode, whereas the second and third columns correspond to the thickness mode at the fundamental frequency and the third harmonic, respectively. Line profiles at peak pressure are extracted from the 2D profiles of the same kind (xz, yz, xy) and superimposed to compare the focal zone location associated with each vibration mode.

Despite the height difference between Transducer 1 and 2, similarity in acoustic output

pressure is observed in each mode of vibration. It is also worth noting that at the same electrical

input power the acoustic pressure generated from the thickness mode is nearly doubled that of

the height mode. Lastly, within the range of applied electrical power under investigation (up to

2.12 W), acoustic peak pressures up to 1 MPa are achievable by driving these thin ring-shaped

51

transducers in thickness mode, and this range of acoustic pressures is adequate for inducing BBB

disruption [99], [128], [218].

Figure 2.10: Comparison of acoustic peak pressure vs. electrical applied power for thickness and height modes of both transducers measured at the focal region

2.4.4 2PFM imaging of BBBD in a rat model

Using thickness mode of Transducer 2 (fT = 1.2 MHz), we obtained evidence of BBBD in vivo at 4

acoustic pressures (0.2, 0.4, 0.6 and 0.8 MPa) with n = 12 per group. For the control rats (i.e. MBs

were administered but not sonicated), no evidence of fluorescent dye leakage was seen. EB

extravasation following sonication at 0.8 MPa, shown in Figure 2.11, demonstrates that the BBBD

is localized to the brain surface area in the sonicated hemisphere. As illustrated in a high

magnification image, EB extravasation region was outlined and measured for its depth and

lateral dimension. Their respective values of 0.75 mm and 0.6 mm are comparable to 75%

pressure profiles generated by thickness mode of Transducer 2 as shown in Figure 2.9.

0 0.5 1 1.5 2 2.50

0.2

0.4

0.6

0.8

1

Electrical Forward Power (W)

Acoustic P

eak P

ressure

(M

Pa)

Transducer 1 - Thickness Mode

Transducer 1 - Height Mode

Transducer 2 - Thickness Mode

Transducer 2 - Height Mode

52

Figure 2.11: Left: A coronal brain section through imaging window shows localized distribution of BBBD indicated by EB extravasation (arrow). Right: A close-up image of EB extravasation region (outlined by dotted boundary) with measured dimensions.

Figure 2.12(A) and Figure 2.12(C) provide two examples of 2PFM monitoring of Texas

Red leakage from the cerebral vasculature upon BBBD induced at the acoustic pressure of 0.6

MPa and 0.4 MPa, respectively. In both cases, MB injection and sonication occurred during the

first 0 – 2 minutes (i.e. the first two frames) when the vessels were still intact. In Figure 2.12(A),

disruption occurred promptly 2 minutes later, with dye leakage originating focally from the

bifurcation point (indicated by the white arrow) of the vessel branch and quickly flooding the

entire FOV. In contrast, the leakage of Texas Red in Figure 2.12(C) was delayed 10 minutes post-

sonication and exhibited much slower kinetics. We also noted for this type of leakage, the

disruption occurred over an extended segment of the vessel wall in the FOV (indicated by the

white arrow) rather than focal origin.

To quantitatively analyze the disruption kinetics, average intra- and extra-vascular

fluorescent signal over time were calculated. Representative regions of the intra- and extra-

vascular compartments are defined as dashed and solid white rectangles, respectively, at t = 0 on

the first frame (Figure 2.12(A) and Figure 2.12(C)). As shown in Figure 2.11(B) and 2.11(D), intra-

vascular signal curves (dashed line) exhibit a reduction over time due to plasma clearance and

diffusion into the interstitium. Despite their overall increasing profile, the extra-vascular signal

curve in Figure 2.12(B) presents a steep slope over 3-minute duration and reaches plateau at 5

minutes. On the other hand, Figure 2.12(D) demonstrates a slower kinetics with 10-minute delay

prior to intensity increase.

53

Figure 2.12: (A) An example of fast leakage of dextran-conjugated Texas Red under FUS+MBs induced BBBD at 1.2 MHz frequency, 10 ms pulse duration, 1 Hz PRF, 120 s exposure duration and 0.6 MPa pressure. (B) Quantitative analysis of fluorescent signal intensity associated with intra- and extra-vascular compartments (represented by dashed and solid rectangle, respectively) for the fast leakage shown in (A). (C) An example of slow leakage of dextran-conjugated Texas Red under FUS+MBs induced BBBD at 0.4 MPa pressure, whereas other sonication parameters remained similar to (A). (D) Quantitative analysis of fluorescent signal intensity associated with intra- and extra-vascular compartments (represented by dashed and solid rectangle, respectively) for the slow leakage shown in (C). Scale bar: 100 µm

Based on their distinct temporal kinetics and spatial profile of extravasation, we classified

the first and second example shown in Figure 2.12 as fast and slow disruption, respectively. By

applying this classification for the entire data set, we predicted the probability of achieving fast

and slow leakage at each pressure level. Figure 2.13 summarizes the measurements, including the

unsuccessful BBBD trials. At the acoustic pressure of 0.2 MPa, BBBD was only achieved for 50%

of the trials, from which most cases are slow leakage. As the pressure increases, the probability of

successful BBBD increases, with higher likelihood of inducing fast leakage than slow leakage.

54

Figure 2.13: A summary of successful BBBD events, as well as the occurrence of two leakage modes (fast

vs. slow) at different acoustic pressure, while other sonication parameters were maintained at 1.2 MHz

frequency, 10 ms pulse duration, 1 Hz PRF and 120 s exposure duration.

2.5 Discussion

Our design of thin ring-shaped transducers enables US sonication from a dorsal approach to be

combined with 2PFM imaging for monitoring and evaluating BBBD. The dorsal-based design

resolves the inherent limitations of the previously published ventral design [126], [128] including

imperfect transmission due to strong attenuation (90%), possible in vivo reverberations due to

reflection at the cranial window on the dorsal skull, and decreased capability for precise

targeting.

Complete characterization and comparison of US pressure profiles between height mode

and thickness mode strongly suggested that the transducers should be driven in thickness mode

for BBBD. Using height mode leads to inadvertent sonication of a brain region deeper than the

imaging zone of the 2PFM, thereby making optical detection of the BBBD impossible. Mismatches

between the imaging zone and the sonication zone due to operation in height mode manifested in

low success rate during the initial experimentation. It is hypothesized that peak pressure

localized at the ventral side of the rat’s skull where MBs enter the cranial circulation, MBs are

prone to be destructed by US; thus it is likely that new MBs do not replenish the imaging FOV.

Instead, the region of disruption might occur somewhere deeper in the tissue explaining the

55

observation where the FOV was flooded with the fluorescent dye without any visible site of

BBBD.

The effect of potential cross-coupling due to dimension similarity between thickness and

height of the transducer was also explored in this study by comparing the US profiles from two

transducers of similar thickness but with different heights. At the greatest allowable height of 1.1

mm, which is close to the preset thickness of 1.5 mm, cross-coupling did not occur. Instead,

despite their differences in height, both transducers yield similar and consistent US pressure

profile associated with each mode of vibration.

When the ring transducer operates in height mode, it can be treated as a plane annulus.

As discussed in Section 3.6.2 of Cobbold’s textbook, the plane annulus can be constructed by

removing a disk transducer of a small radius Ri from another disk transducer of a larger radius Ro

[213]. In the case of our particular ring transducer, Ri = 3.5 mm and Ro = 5 mm. Hence, the

impulse response of the annulus can be written as , where is the

vector distance from any field point to the transducer surface. This vector distance can further

be decomposed into radial distance r and axial distance z. Furthermore, the impulse response for

a disk or piston transducer of radius R has been thoroughly examined and discussed in Section

3.3.1 of Cobbold’s textbook [213]. In particular, for the entire range of possible radial position r,

the impulse response is given by:

(2.9)

where , , and is the speed of sound in

the propagating medium.

In the same manner as the impulse response, the CW pressure response of a plane

annulus ring can be derived by using the superposition principle and thus can be expressed as

[213]. Under the assumption of a “rigid baffle” boundary

condition, the square of the pressure amplitude for on-axis observation points ( ) is given by

Equation (3.22) in Cobbold’s textbook and is re-written as below:

56

(2.10)

where is the medium density, and is the particle velocity (i.e. particle movement from the

transducer surface). As a result, maximum pressure takes place where

. For

and , it can be deduced that . This is the distance between the transducer and

the last maximum of the intensity. Based on these analyses of pressure maxima and minima

generated by a piston transducer of a specific radius R, the pressure field of an annulus ring can

be deduced. As an illustration of the resulting on-axis CW pressure magnitude response from an

annulus with and , Figure 3.32 in Cobbold’s textbook reveals that the first and

last peak of the annulus transducer locates at and , respectively. Given

that our ring transducer operates at 1.8 MHz frequency in height mode, the corresponding

wavelength is

. Hence, and . With a

comparable size as the annulus considered in Figure 3.32 of Cobbold’s textbook, we could

roughly estimate the axial location of the first and last pressure peak of our ring transducer, when

operating in height mode, to be and

, respectively. Based on experimental data shown in Figure 2.8 and Figure 2.9, we note

that the axial peak pressure generated by the ring transducers in height mode falls within the

range between [ ].

In contrast to the well-understood far-field pressure profile generated by the ring-shaped

transducer operating in height mode as described above, the operation principle of its thickness

mode remains undetermined. To gain insights into the role of the coverslip during the transducer

vibration, we coated the front surface of the coverslip with a thin gold layer and performed laser

Doppler vibrometer (LDV) experiments. The transducer was excited by a sinusoidal pulse signal

(10 cycles/pulse, PRF = 100 Hz, Vinput = 200mV). During the transducer sonication (either in the

thickness or height mode), the laser beam from the sensor head (PSV-400, 1 mW helium neon,

wavelength λ = 633 nm) was directed to the gold-coated coverslip surface. At a 00 incident angle

and a spot size spot size of approximately 10 μm, the laser beam scanned the entire coverslip

surface. The number of sampling points is approximately 2500. The reflected optical signal at

each sampling point was collected and the velocity decoder measured the displacement at that

particular location.

57

Figure 2.14 demonstrates LDV measurement of the coverslip vibration when the

transducer operates in height mode (top images) and thickness mode (bottom images). In each

sonication mode, the images on the left and right show the two opposite phases (i.e. 1800 phase

difference) of the coverslip vibration. For both height and thickness mode, we observed the wave

formation and propagation, which presumably originates from the ring area and exhibits

constructive interference in the center of the coverslip. However, we noticed the wavelength and

amplitude difference of the coverslip vibration in these two scenarios. In the former case, shorter

wavelength and lower undulation were noted. In the latter case, longer wavelength and much

significant oscillation amplitude of coverslip surface movement were seen. In fact, the difference

in wavelength can be explained by the frequency difference between the height and thickness

mode (refer to Table 2.1). Meanwhile, the greater vibrational amplitude of the coverslip in the

thickness mode can be attributed to its higher piezoelectric constant due to the matching

alignment of the dipole moment with the poling direction (as discussed in Section 2.1.3). In other

words, the transducer operating in the thickness mode would generate a stronger response than

that in the height mode under the same externally applied voltage.

Due to the strong coupling of the ring transducer to the coverslip during the thickness

mode vibration, we anticipate the coverslip might act as a secondary ultrasound emitter.

However, unlike the typical piston transducer in which the entire surface exhibits a uniform

particle movement (e.g. plane wave with constant phase, ), the particle velocity on

the coverslip surface has different phase depending on its radial distance from the center. In other

words, the particle velocity on the coverslip surface can be described as ),

where accounts for the radially-dependent phase offset. Analyses of such phenomenon have

been described in the technical notes by Aarts and Janssen [219], [220]. In their study, they

estimated the acoustical quantities such as the sound pressure on-axis, directivity and total

radiated power of a harmonically vibrating membrane. Figure 2 in the cited paper indicates two

pressure peaks: one immediately at z = 0 and one at z = R. With R = 5 mm in our case, the on-axis

pressure profile generated by harmonically vibrating coverslip is predicted to peak within the

axial range 0-5 mm and significantly drops at further distance. This might explain the near-field

focal zone as observed in the thickness mode of our ring-shaped transducers (refer to Figure 2.8

and Figure 2.9).

58

Figure 2.14: LDV measurement of the coverslip vibration when the transducer operates in height mode

(A-B) and thickness mode (C-D). In each sonication mode, the images on the left (A,C) and right (B,D)

show the two opposite phases (i.e. 1800 phase difference) of the coverslip vibration.

Study of water loading effect on the pressure profile output has been emphasized in

Shaffaf’s thesis work for multi-element ring-shaped transducer [221]. The author observed that by

varying the loading water condition from “fully rounded” to “flat” and “low”, the location of the

axial peak pressure and the axial profile can be affected by the water volume. In her thesis, the

author noted that multi-element ring transducer experienced focal pressure variation greater than

0.1 MPa when different loading conditions were applied. However, the profiles from “fully

rounded” and “flat” water loading are acceptable when the axial peak location remains within 2

mm below the transducer base. Meanwhile, the “low” water loading condition was found to

cause unwanted changes on the axial pressure profile. Another issue being raised in Shaffaf’s

thesis is the sensitivity of the pressure profile to the objective height. The author noted that,

during in vivo experiments, sonications are performed with the two-photon objective situating in

the ring center and being lifted up and down within a depth range of 0.5 mm. The author

repeated this movement of the objective in an in vitro setup, where the multi-element ring

59

transducer was loaded with water and the objective lense was first put into contact with the

water surface and subsequently lowered at 0.1 mm increment. In response to the objective

movement from zero to 0.5 mm below the water surface, the axial location of peak pressure was

found to have shifted from 0.1 to 1.6 mm. However, the peak focal pressure does not vary

significantly. Since the objective movement is inherent in the 2PFM experiments, these findings

suggest that we should pay more attention to the objective position during experiments and be

aware of the potential variability of in situ pressure profile when inducing BBBD.

Due to its small size, contact with water would have more prominent effect on the

performance of each element of the phased-array transducer. However, for the single-element

ring-shaped transducer used in our study, we expected lower sensitivity to the water loading

condition and the objective height as compared to the multi-element ring transducer being

investigated in Shaffaf’s thesis. Despite the lack of direct sensitivity study for our single-element

ring-shaped transducers, we noted that the impedance curve of these transducers remained

stable whenever a new coverslip was replaced and a session of matching circuit was conducted.

As a result, the matching circuit required minimal tuning, as long as the water loading was

maintained at the “full” condition. In other words, we could easily achieve 50 Ω and 00 phase

without significantly altering the capacity and inductance of the matching circuit. Furthermore,

during in vitro hydrophone scans of the same single-element ring-shaped transducer on different

days, we were able to reproduce the pressure profile and the focal pressure level for a particular

externally applied voltage. With the potential of different water loading conditions on different

experiment days, the ability to repeat the experimental results suggested that these single-

element ring-shaped transducers might not be as sensitive to water loading condition as the

multi-element transducer. Lastly, during sonication session of any in vivo experiment, we

recorded the forwarded and reflected RF-power using the in-house built power meter. From

these recorded data, we noted that both of the forward and reflected powers remained stable for

the same applied voltage over different experiment days. Therefore, we anticipated the objective

height might not significantly alter the sonication condition (i.e. focal pressure level) during in

vivo experiments.

Robust and effective induction of BBBD in vivo monitored by real time imaging with

2PFM system has stringent requirements: (1) suitable frequencies and acoustic pressure (or

60

mechanical index) for FUS+MB mediated BBBD [99], [128], [218], [127]; (2) laterally uniform US

pressure field within the 512 x 512 μm2 imaging FOV; (3) axially shallow and confined US focal

zone due to the limited depth penetration of the optical system. The in vitro results demonstrate

that the thin ring-shaped transducer driven in thickness mode has met all of these conditions. At

the resonant frequency of 1.2 MHz, the transducer driven in thickness mode can effectively cause

in vivo BBBD at the dorsal surface of the brain as observed with the 2PFM. Detection of EB

extravasation further confirms BBBD was localized to the focal region near the cranial surface of

the sonicated hemisphere.

Using this dorsal-based ring-shaped design, Cho et al had demonstrated the initial proof-

of-concept dorsal sonication on 20 rats [127]; however, different operating modes and

corresponding pressure profiles were not well understood and fully explored. In this study, with

insights gained from a complete transducer characterization, we continued the investigation into

BBBD using two-photon microscopy guidance in a larger group of animals (n = 48) to identify the

probability of successful BBBD induction at each pressure level as well as examine associated

leakage kinetics. Similar to Cho et al, we determined the leakage type to be fast or slow by

observing the kinetics at origin of disruption. Fast disruption was determined by leakage kinetics

originating from a focal point on a vessel. Slow disruption was characterized by the leakage of

small amounts of dye along the length of a vessel. This latter disruption type is more prone at

low pressure range of 0.2 – 0.6 MPa at the frequency of 1.2 MHz. At 0.8 MPa, 90% of BBBD events

exhibit fast kinetics and focal origin. This transitional pressure agrees well with the expected

pressure of 0.5 MPa when considering the operating frequency of 1.2 MHz used in our 2PFM-

guided experiment and the mechanical index for BBBD threshold of 0.46 observed by McDannold

et al. [99].

It has been postulated that slow leakage is facilitated by transcellular transport of material

outwards from the lumen. As evident from electron microscopy, an increase in vesicles and

vacuoles as well as fenestration and cytoplasmic channel formation was observed in the BBB

endothelial cells [110]. A recent study by Deng et al [222] also verified the upregulation of

caveolin-1 protein, which is essential for the caveolar invaginations. In contrast, fast leakage is

postulated to be due to paracellular transport. This mechanism is probably of a physical nature,

61

in which MB oscillations in the microvessel exert circumferential and shear stress onto the vessel

wall, leading to opening of the TJ [115].

Results established in this study indicated that varying the acoustic pressure amplitude

may allow us to control the mechanism of BBB disruption. Beyond differentiating and selecting

various leakage types, our transducer system will also be used to investigate the closure kinetics

of the BBB [195], the integral role of support cells in maintaining and repairing the BBB following

disruption [16], [118], [223], or to evaluate BBBD-based drug delivery methods by tracing

fluorescently-labeled drugs in the brain tissue [13], [224].

2.6 Conclusions

A compact 2PFM compatible ultrasound transducer design has been fabricated and

characterized. In order to avoid the complications exhibited in ventral sonication, the transducer

is placed dorsally. The cylindrical geometry of the transducer overcomes the constraints imposed

by the 2PFM imaging objective. In addition, the height and thickness mode for the dorsal-based

transducer has been characterized and indicates that the latter mode is better-suited for optically

observed BBBD application due to the constrained US pressure profiles. Lastly, our integrated

system has demonstrated the capability to induce various BBBD modes by controlling the

acoustic pressure. This system will be employed in the latter studies to investigate the

biophysical effects of BBBD at the microscopic level and to measure the permeability constant of

various substance across the compromised BBB in murine model.

62

3 Quantitative evaluation of enhanced

permeability of BBB using 2PFM 2

In leveraging the 2PFM-integrated transducer system, FUS+MBs induced BBBD can be robustly

conducted on rat brain in vivo while the cerebral vasculature of the studied subject is closely

monitored. Prior to a delving into the details of experimental implementation, an overview on the

basic principle of two-photon fluorescent microscopy will be presented. Furthermore, this

chapter will be dedicated to introducing a quantification method that has been established to

measure BBB permeability from 2PFM data. Findings on the BBB permeability kinetics and its

dependence on extrinsic parameters (e.g. acoustic pressure, substance size) are highlighted

therein.

3.1 Overview on 2PFM

The theory of two-photon absorption by atoms was first described by Maria Goeppert Mayer in

her doctoral thesis in 1930 [225]. Based on this concept, the first two-photon fluorescence

microscopy system was designed by Denk, Webb and colleagues in 1990 [226]. Since then, it had

been extensively used as a 3D imaging tool for biomedical research in conducting detailed

examination of biological samples in vivo [227], [228]. Its major advantages over other

conventional optical microscopy techniques include the deeper penetration into scattering tissue

as well as reduced photodamage and photobleaching of fluorophore [229].

3.1.1 Basic principle of 2PFM

Fluorescence is an optical phenomenon in which a fluorophore is elevated from the electronic

ground state to the first electronic excited state after absorbing a photon of an appropriate energy.

Such process is depicted via the Jablonski diagram as shown in Figure 3.1(A). At room

temperature, the absorption of this single photon (as described by the blue arrow) occurs within

2 Adapted from the article: Nhan T, Burgess A, Cho EE, Stefanovic B, Lilge L, Hynynen K. Drug delivery to the brain by focused

ultrasound induced blood-brain barrier disruption: quantitative evalution of enhanced permeability of cerebral vasculature using two-

photon microscopy. J Control Release 2013; 172: 274-80.

63

an attosecond time window (i.e. ~10-18 s). After one to ten nanoseconds (i.e. ~10-8 – 10-9 s) delay,

the fluorophore returns to the ground state while a fluorescent photon of slightly lower energy is

emitted (as presented by the steel arrow). This linear excitation is typically achieved with the

light of green-ultraviolet wavelength. Meanwhile, a similar fluorescence process can also be

achieved with multiple excitation photons whose energies fall into the infrared spectral regime.

For instance, a simultaneous absorption of two infrared photons whose energies are

approximately half of that in the former scenario could also lift the fluorophore to the first

electronic excited state. Its subsequent relaxation to the ground state would result in a same

emission photon, as demonstrated in Figure 3.1(B). However, such multi-photon excitation is

classified as non-linear optical phenomenon, which relies on the quantum transitions of virtual

states and requires extremely high photon flux (e.g. 1020-1030 photons/(cm2s). Given the short-

lived nature of these virtual states, the probability of multi-photon absorption is a quadratic

function of the light intensity [225].

Figure 3.1: Jablonski diagram to differentiate the single-photon (A) and two-photon (B) excitation

process. Adapted from [230].

3.1.2 Design of 2PFM

To facilitate the stringent condition of high photon density to achieve previously-mentioned non-

linear effect, a femtosecond laser is required. Owing to the advanced design of stable mode-

locked infrared laser [231], generation of pulsed light (rather than continuous-wave light) at high

flux level is conveniently attainable. In particular, a Ti:sapphire laser is employed to create 100 fs

64

ultrafast pulses of appreciable peak power. Furthermore, with the aid of high numerical aperture

microscope objective, the laser beam can be focused to a diffraction-limited volume. In such

circumstance, generated photons are spatially and temporally compacted, thus enabling the non-

linear process.

With the Ti:shappire laser system being the core element, a typical 2PFM design is

illustrated in Figure 3.2. The excitation beam exiting the laser is steered along the

epiluminescence light path, reflected by a dichroic mirror and eventually focused into the

specimen via the microscope objective. Using a galvanometer for lateral positioning and a piezo-

objective for axial control, the diffracton-limited focal volume can be guided at the sample over

3D. Upon two-photon excitation of either endogenous or exogenous fluorophores in the tissue

sample, the emission fluorescence signal is captured by the same objective and directed through

the dichroic mirror. Subsequently, different filters can be used to separate the emission signals by

their spectral range. Lastly, analog-to-digital signal conversion takes place via a photomultiplier

tube (PMT). Owing to its significant sensitivity to light in the ultraviolet, visible and near-

infrared spectrum, the PMT detector surface generates a current in response to an incident

photon. The current is then amplified via a series of dynode stages inside the vacuum

phototubes. Along with a large active sensor for photon collection, the current multiplication

process results in a high gain, low noise and ultra-fast response, which are advantageous for the

detection of low photon flux. Thus, PMT can lead to a 160dB or 108 fold signal amplification [232].

Figure 3.2: A basic design of a two-photon fluorescent microscopy system. Adapted from [230].

65

3.1.3 Two-photon versus single-photon fluorescence microscopy

Due to its nonlinear nature of the two-photon excitation, the fluorescence is confined to the focal

center of the laser beam, where the excitation intensity is greatest. In other words, the quadratic

dependence of two-photon imaging technique allows for a pinpoint excitation and detection

volume at a deep location within thick samples [233]. Figure 3.3 demonstrates the confinement of

excitation volume offered by two-photon imaging in comparison to that generated by single-

photon method. As further shown in Figure 3.3, the fluorescence power of 2PFM decay can be

expressed with 1/z2, where z is the axial distance away from the focus [233]. Thus, 2PFM enables

rejection of background signal outside of the focal region. This leads to an appreciable

improvement in image resolution. Lastly, given that Rayleigh scattering effect is highly frequency

dependent, the use of infrared laser beam for two-photon excitation process implies a significant

reduction in light scattering [234]. With the combination of these advantageous features, 2PFM

has been proven to achieve 2-3 fold improvement in depth penetration as compared to single-

photon confocal microscopy. For instance, experimental study by Kobat et al. has achieved the

fundamental depth limit in scattering tissue by demonstrating the ability to record imaging depth

of 1.6 mm in an in vivo mouse cortex using 2PFM [235].

Figure 3.3: Comparison of excitation and fluorescence focal volume generated by single-photon (left)

and two-photon (right) imaging method [233]

66

3.2 Research motivation

Intensive efforts in drug development have led to the invention of numerous therapeutic agents

with the potential to treat CNS diseases and disorders including chemotherapeutic agents for the

treatment of brain tumors and metastases [236] as well as chemokines, growth factors, and viral

vectors for the treatment of neurodegenerative diseases [2]–[4]. At the emergence of FUS+MBs

triggered BBBD as a non-invasive technique for targeted drug delivery to the brain, these novel

therapeutic agents will be more likely to reach the local treatment site at a sufficient concentration

whereas their accumulation elsewhere in the brain and body can be limited.

Guidance and evaluation of FUS-induced BBBD has been primarily performed by MRI.

Typically, T1w-MRI images are acquired to confirm successful delivery of Gadolinium based MR

contrast agents across the BBB [29], [129], [175], whereas T2w-MRI images serve to verify the

absence of edema and tissue damage [84], [212]. In addition, quantitative measurement of the

permeability of the BBB in the targeted region (e.g. hippocampus or striatum) can be done via

DCE-MRI [195], [196], [237], [238]. However, spatial resolution (e.g. lateral: 86x86 µm2; slice

thickness: 500 µm) offered by 9.4T MRI limits permeability measurement to a macroscopic brain

volume of 2-35 mm3 [196], [238].

In exploiting the 2PFM imaging modality, we aim to go beyond the volumetric average as

achieved in DCE-MRI and resolve the gradient of drug concentration to differentiate therapeutic

range from toxicity level. Furthermore, a dorsal approach for FUS exposure makes it possible to

induce BBBD in a reliable manner and better control the in situ acoustic pressure [127], [239]–

[241]. These studies, however, characterized the microscopic leakage patterns qualitatively but

did not attempt to quantify the rate of agent delivery.

Here, we propose a quantitative methodology to analyze the 2PFM images post BBBD. By

extracting and correlating intravascular and extravascular signals from the time-lapse 2PFM

images, a permeability constant of the cerebral vasculature network within the FOV can be

determined. Measured vascular permeability is then correlated with the applied acoustic

pressure, disruption onset and vessel diameter to shed light on the potential mechanisms which

control BBBD. These insights are crucial for guiding future treatments utilizing BBBD-based drug

delivery to the brain.

67


3.3.1 Animal preparation

Male Wistar rats of 120-200 g weight range were used in this study (n = 40). All the procedures

were approved by the institutional Animal Care and Use Committee and were in accordance

with the guideline by Canadian Council on Animal Care. The details on animal preparation and

microsurgical procedures have been discussed in Section 2.3.3. Once the surgery was completed,

the rat stabilized on the stereotactic stage was transferred to the microscope for BBBD induction

and 2PFM imaging.

3.3.2 FUS parameters for BBBD

The experimental timeline is shown in Figure 3.4(A). Prior to sonication, Definity MBs (Lantheus

Medical Imaging, Billerica, MA, USA) of 1.1-3.3 µm mean diameter were diluted with saline (1:10

v/v) and injected through the tail vein at a final concentration of 0.02 ml/kg. A PZT-4 cylindrical

transducer (diameter = 10 mm, thickness = 1.5 mm, height = 1.1 mm) was used for sonication. A

complete characterization study of the transducer design, which facilitates dorsal application of

FUS and simultaneous 2PFM imaging, has been described previously in Chapter 2. Briefly, the

transducer was operated in the thickness mode at a frequency of 1.2 MHz to produce a circularly

uniform focal spot that coincides with the microscope’s lateral imaging FOV (512x512 µm2). The

ultrasound depth of field generated by the transducer is shallow (1mm immediately beneath the

coverslip), ensuring that it overlapped with the light depth penetration of the 2PFM. The

transducer was driven by a function generator (Agilent, Palo Alto, CA, USA) and a 53 dB RF

power amplifier (NP Technologies Inc., Newbury Park, CA, USA) with typical BBBD sonication

parameters (10 ms pulse duration, 1 Hz pulse repetition frequency, 120 s total sonication

duration). The applied forward and reflected RF-power during sonication was recorded using an

in-house constructed power meter.

68

Figure 3.4: In vivo BBBD induced by FUS+MBs and monitored by 2PFM imaging. A) Experimental timeline. B) 4D XYZT acquisition of 2PFM imaging.

3.3.3 2PFM imaging

To visualize the cerebral vasculature, 10 kDa or 70 kDa dextran-conjugated Texas Red

(Invitrogen, Burlington, ON, Canada) was injected through the tail vein. The two molecular

weights (MWs) were chosen to represent equivalent therapeutic substances with sizes ranging

from small proteins and siRNAs up to albumin-bound drugs that can be delivered across the

BBB. If we assume the protein has a spherical shape, the minimal radius of a sphere that could

contain a certain mass of protein is given by: , where is in Dalton and is

in nm [242]. Hence, for 10-70 kDa, the minimal radius is estimated to be 1-3 nm.

The animal was positioned below the microscope stage (FV1000MPE, Olympus, Tokyo,

Japan) and the cranial window was aligned underneath a water-immersion objective (Olympus

69

XLPLN, Tokyo, Japan) with 25x magnification power, 1.05 numerical aperture and 2 mm working

distance. Two-photon excitation of Texas Red fluorescent dye was achieved with a mode-lock

Ti:Sapphire laser unit (Mai-Tai, Spectra-Physics, Mountain View, CA, USA) emitting at 810 nm

wavelength, 100 fs pulse width and 80 MHz repetition rate. Scanning was performed in an XYZT

order (Figure 3.4(B)), in which lateral images of 512 x 512 pixels (0.99 µm resolution, 8 µs/pixel)

were captured below the cortical surface up to 300 µm depth (i.e. cortical layers I and II) in a

stacking mode at 10 µm increments. This scanning depth allows imaging vessels within the pial

layer. As depicted in Figure 3.4(B), 2PFM imaging was continuous over the course of the

experiment; from the injection of MBs, through the 120 s sonication, and following leakage of the

fluorescent dye upon BBBD. Typically, each data set consists of 40-50 stacks with the acquisition

time of 15 to 30 seconds per stack.

3.3.4 Analysis of 2PFM data

4D XYZT microscopic data of a superficial cortical tissue volume was visualized in Matlab (The

Mathworks, Natick, MA, USA) as a maximum intensity projection map along z direction (Figure

3.5(A)). To separate the intravascular and extravascular compartment, automatic vessel

segmentation was performed on each individual Z-slice at the initial time point (when BBB was

impermeable to either dextran). Once the intravascular regions of interest (ROIs) were masked

based on the segmented vessels, extravascular ROIs were identified by subtracting the

intravascular ROIs from the imaging FOV. Fluorescent intensity associated with each

compartment, Ii(t) and Ie(t), was then calculated by averaging over all pixels within the

compartment ROIs over the entire depth (Figure 3.5(B)).

To measure permeability from fluorescent intensity change in the intravascular and

extravascular space, we applied the formulation developed by Dreher et al. [243]. In their model,

the rate of solute transport across a blood vessel wall is given by the Kedem-Katchalsky equation

to account for both convection and diffusion processes [243]–[245]. However, due to its unknown

direction and magnitude, the convective term was ignored and its influence was lumped into the

latter term [246]. Given the linearity between the dye concentration and the fluorescent signal

intensity in the plasma and the extravascular space (which has been established via an

independent in vitro fluorometry study), an apparent permeability α(t) measuring exchange

capacity between the two compartments can be determined via the following equation:

70

ie

ei

e

VV

tI

HCT

tI

dtdIt

/

)(

1

)(

/)(

(3.1)

where Ve/Vi is the volume fraction between extravascular and intravascular compartments,

which was simply obtained from the vessel segmentation. Similar to DCE-MRI studies by Park et

al. [195] and Vlachos et al. [196], [238], hematocrit (HCT) of 45% was assigned to account for the

average HCT level of all blood vessels within the imaging FOV [247], [248]. Apparent

permeability αapp for each studied subject is the average value of α(t) from 3 to 30 minutes,

corresponding to the peaked duration of BBBD (Figure 3.5(C)).

Figure 3.5: Data analysis of 2PFM data capturing fluorescent dye leakage upon BBBD. A) Depth projection images illustrate the transient BBBD induced by MBs & FUS at 0.6 MPa (scale bar: 100µm). Sonication and MB injection occurred during the first 2 minutes while the vessels remained impermeable to dextran conjugated Texas Red TR10kDa. As soon as sonication ceased, disruption started at multiple vessels within the imaging FOV and extravascular signal increases over time. B) Quantitative measurement of averaged fluorescent signal intensities associated with intravascular and extravascular compartment over time. C) Permeability was evaluated accordingly.

71

3.3.5 Statistical analysis

At each acoustic pressure, the apparent permeability αapp of TR10kDa and TR70kDa, as well as

the volume fraction Vi/Ve, were reported as mean (±standard deviation) over 5 animal subjects.

For 40 cases of permeability measurements, comparison among 4 pressure groups and 2

molecular weights (MW = 10kDa and 70kDa) was performed using two-way ANOVA followed

by Bonferroni post-tests in GraphPad Prism (GraphPad Software Inc., CA, USA). Correlation

between the acoustic pressure and BBBD temporal onset was evaluated by one-way ANOVA

followed by Bonferroni’s Multiple Comparison Test. The correlation between vessel size and

leakage type (fast vs. slow kinetics) was assessed using two-tailed unpaired Student`s t test. For

all of these analyses, p < 0.05 were considered statistically significant.

3.4 Results

3.4.1 Effect of acoustic pressure on enhanced BBB permeability

For TR10kDa and TR70kDa, we analyzed 5 data sets per pressure and the evaluated permeability

values are presented in Figure 3.6(A) and Figure 3.6(B), respectively. Both scattering plots reveal

an increasing trend for the permeability at higher pressure. To quantitatively explore the

relationship of permeability and acoustic pressure, linear regression was applied for each MW.

Best-fit values of slope for TR10kDa and TR70kDa were found to be 0.039±0.005 min-1/MPa and

0.018±0.005 min-1/MPa, respectively. Best-fit values of X-intercept for TR10kDa and TR70kDa

were found to be 0.16 MPa and 0.11 MPa, respectively. In addition, two-way ANOVA in

combination with Bonferroni post-tests confirmed statistical significance in permeability of

TR10kDa induced at low acoustic pressure of 0.4 MPa with respect to higher acoustic pressure of

0.6 MPa and 0.8 MPa. For TR70kDa, no significant difference of permeability between 0.4 MPa

and 0.6 MPa was found, whereas statistical significance was observed for resulting permeabilities

at 0.4 MPa and 0.8 MPa.

3.4.2 Effect of substance size on enhanced BBB permeability

To determine the impact of substance size (or MW) on BBB permeability, average values of two

MWs at similar acoustic pressure are plotted side by side as shown in Figure 3.7(A). Two-way

ANOVA followed by Bonferroni post-tests showed that, at high pressure, a significantly greater

72

permeability of TR10kDa compared to TR70kDa was demonstrated (e.g. p < 0.05 at 0.6 MPa and

0.8 MPa). This observation is consistent with the expected inverse relationship between molecular

size and its permeation across the BBB. To confirm that these permeability measurements were

not confounded by differences in cerebral vasculature volume, two-way ANOVA followed by

Bonferroni post-tests were performed on the volume fraction Vi/Ve between TR10kDa and

TR70kDa at 4 acoustic pressure levels (n = 5 per group) and found no statistically significant

differences for all 4 pairs.

3.4.3 Temporal onset of BBBD is correlated with permeability and appears

to be controlled by acoustic pressure

To investigate the kinetics of BBBD, temporal onset (i.e. time point when leakage was initiated)

was recorded for all 40 data sets and plotted against the permeability. As evident in Figure 3.8(A),

an inverse relationship between these two entities was observed. 10 minutes was chosen as the

temporal benchmark to separate fast leakage from slow leakage as this duration is the averaged

time required for receptor-mediated transcytosis [248], [249]. Fast leakage exhibits short BBBD

onset and substantially higher permeability constants, whereas slow leakage presents delayed

BBBD onset and very small permeability.

To examine the connection between BBBD onset and the applied acoustic pressure,

averaged temporal onset was plotted at each pressure (n = 10 per pressure), as shown in Figure

3.8(B). Due to its gradual increase in leakage, it is difficult to define BBBD onset time based on the

extravascular signal curve (e.g. Figure 3.5(B)). Therefore, we used the peak time of the

permeability curve (e.g. Figure 3.5(C)) as the benchmark for BBBD onset time. In doing so, we

further noted that the peak time of approximately 4-5 minutes agrees with visible leakage

evidence of Texas Red extravasating out of initially intact vasculature as shown in Figure 3.5(A).

Overall, higher pressure appears to yield prompt onset. One-way ANOVA followed by

Bonferroni’s Multiple Comparison Test reported no statistical significance between 0.2-0.4 MPa

and 0.6-0.8 MPa, whereas significant difference between average BBBD onset at 0.4 MPa (13.5±7.6

minutes) and at 0.6 MPa (6.3±4.3 minutes) was noted. This strongly suggests that different

leakage kinetics (slow or fast) can be controlled via the applied acoustic pressure.

73

Figure 3.6: Effect of acoustic pressure on permeability dextran conjugated Texas Red across the BBB. Permeabilities were measured for all 20 cases of (A) TR10kDa and (B) TR70kDa delivered across the BBB. Two-way ANOVA in combination with Bonferroni post-tests were used to determine the statistical significance in permeabilities between different pressure level.

Figure 3.7: Effect of substance size on enhanced BBB permeability. At each pressure, average permeability constant (A) and average volume fraction (B) was compared between TR10kDa and TR70kDa. Two-way ANOVA in combination with Bonferroni post-tests were performed as multiple comparisons.

74

Figure 3.8: BBBD onset in relation to permeability and acoustic pressure. A) Inverse relationship between BBBD onset and permeability. B) Inverse relationship between BBBD onset and acoustic pressure. One-way ANOVA followed by Bonferroni’s Multiple Comparison Test confirms a statistical significance in BBBD onset between 0.4 MPa and 0.6 MPa.

3.4.4 Effect of vessel diameter on enhanced BBB permeability

We further looked into the effect of vessel diameter on the enhancement of BBB permeability by

measuring the average diameter of vessels undergoing disruption within the imaging FOV and

plotting the value against the corresponding permeability constant as illustrated in Figure 3.9(A).

Overall, permeability appears to be inversely related to vessel diameter. By applying the

predefined criteria (10 minutes benchmark of BBBD onset and negligible permeability constant),

data points associated with fast and slow leakage are separated by the dotted line in Figure

3.9(A). Here, we noted that fast leakage is prevalent in small vessels (10-40 µm diameter),

whereas slow leakage occurs more commonly in larger vessels (40-70 µm diameter). As revealed

in Figure 3.9(B), p value from two-tailed unpaired Student`s t test confirms the correlation

between the vessel size and the resulting leakage type.

75

Figure 3.9: Effect of vessel diameter on enhanced BBB permeability. A) Vessel size distribution in correlation with permeability constant: large vessels (40-70 µm) are prone to slow leakage kinetics and low permeability; whereas smaller vessels (10-40 µm) are subjected to fast leakage kinetics and high permeability. B) Statistical analysis (two-tailed t test) indicates significant difference (p < 0.0001) in vessel size responsible for fast and slow leakage types.

3.5 Discussion

Past investigations into the kinetics of BBBD permeability had been carried out using DCE-MRI.

For instance, Vlachos et al. demonstrated the reconstructed permeability map of the murine

hippocampus superimposed onto coronal and transverse T1w images of the brain [196], [238].

The same group also confirmed the dependence of permeability on acoustic pressures and

microbubble sizes. Park et al. further compared the permeability enhancement of double

sonication to single sonication. These analyses supported the use of DCE-MRI as an in vivo tool

for quantifying the efficacy of FUS induced BBB opening. However, detecting disruption and

measuring permeability at vascular level is of fundamental importance for resolving

concentration gradients of the delivered drugs, from which therapeutic range and excessive

toxicity range can both be identified. Furthermore, permeability provides estimation of drug

concentration in the interstitial space. To perform these analyses, 2PFM is required.

Our study builds on the initial work of Raymond et al. who demonstrated the use of 2PFM

for a comprehensive investigation of BBBD [128]. Our previously published study has advanced

this field by introducing dorsally applied FUS which enhances the robustness of BBBD at a

reliable in situ pressure [127], [239]. In the present work, we presented a quantitative analysis

technique which allows for characterization of permeability from 2PFM time-lapsed images. High

76

spatial resolution from 2PFM imaging enabled discerning investigation into vessel diameter and

the temporal onset of BBBD in correlation with applied acoustic pressure and resulting disruption

kinetics.

Permeability constants had been previously reported for 1 kDa Gd-DTPA via DCE-MRI

measurement as 1.1e-2 min-1 and 3.9e-2 min-1 when BBBD was achieved at frequency of 1.5MHz

and acoustic pressure of 0.45 MPa and 0.6 MPa, respectively [196]. At comparable FUS

parameters, these values are two-fold and five-fold higher than 2PFM monitored BBB

permeability enhancement of TR10kDa and TR70kDa, respectively. Although differences in

permeability are expected due to its ten-fold smaller size, the distinctions between the two

imaging modalities should be noted. In 2PFM, a small cortical tissue volume (512x512x100 μm3) is

directly measured and vessels of narrow size distribution undergoing BBBD are readily detected.

In contrast, value reported from DCE-MRI is corresponding to a much larger brain tissue volume

(2-35 mm3) [195], [196], [238] that contains only a fraction of vessels at broader size distribution

being disrupted while the rest of vascular tree remain intact. Nevertheless, these data fit the

overall expected trend of an inverse correlation between MW and permeability constant and

eliminates the need to conduct further 2PFM on small MW compounds. Furthermore, we

speculate that the permeability of large molecule therapeutics would follow the same trend as

seen here and thus exhibit lower permeability than the 70 kDa compound. However, one

limitation of this study is that we cannot predict how drugs with charges or other modifications

will extravasate upon BBBD. Similarly, although stem cells and immune cells have been shown to

cross the BBB upon the application of FUS+MBs [144], [170], this study cannot elicit the rate of

their extravasation. Due to the ability of the cells to interact with the BBB, understanding the

kinetics may require further analysis.

From this study, we also noted that the permeability constant is linearly related to the

applied acoustic pressure, as well as inversely related to the onset of BBBD. At higher acoustic

pressure, one would expect the greater oscillatory amplitude of MBs during stable cavitation, or

the likelihood of inertial cavitation which triggers the collapse of MBs [250]. These MB activities

could readily prompt the opening of BBB. We also observed that transition from low to high

permeability (Figure 3.6), as well as from long to short BBBD onset (Figure 3.8(B)) takes place at

the pressure range of 0.4-0.6 MPa. This transitional pressure agrees well with the expected

pressure of 0.5 MPa when considering the operating frequency of 1.2 MHz used in our 2PFM-

77

guided experiment and the mechanical index for BBBD threshold of 0.46 observed by McDannold

et al. [99].

By classifying the leakage types into fast and slow kinetics based on BBBD onset (Figure

3.8(A)), we found that 20 of 22 data points in the first group exhibiting high permeability (from

0.005 min-1 up to 0.036 min-1). On the other hand, in the latter group, 16 of 18 data points possess

permeability constants below 0.005 min-1. Mechanistically, we speculate that the fast leakage is

caused by the opening TJs, leading to fluorescent dye leakage out of the blood vessel. In some

cases, Ie(t) curve exhibits saturation after the initial ramp-up (15-20 minutes post sonication),

indicating quick repair and closure of the BBB of a few vessels in the imaging FOV. One study

suggested that a possible mechanism for quick repair involves the recruitment of astrocytes and

microglia to the disruption site [251]. Although opening of TJs might be responsible for the

typical focal pattern of disruption, another possible mechanism of fast leakage is cellular

sonoporation [68]–[71], [114]. This might explain why some fast leakage occurred extensively

along a segment of blood vessel rather than from a single focal point. Notably for this

mechanism, the pores also reseal quickly (4-10 seconds) [70]. In contrast, we predict that slow

leakage is facilitated by transcytosis, which is limited under normal condition of the BBB.

However, FUS-induced oscillation of MBs may activate endothelial cell receptors to promote

transcellular transport of molecules from the lumen to interstitial space [109], [110], [222]. It was

previously suggested that transcytosis of low-density lipoproteins across the endothelial cells

lining the vessel wall takes at least 15 minutes [248]. This lag may account for the slow onset of

leakage observed in our 2PFM experiments.

As reflected in the permeability constant, the extent of BBB opening is inversely related to

the vessel diameter (Figure 3.9(A)). In a simulation study, Hosseinkhah et al. had shown similar

dependency of shear stress on the ratio between vessel and initial bubble radii (i.e., largest shear

stresses were obtained at lowest rv/r0 values) [252]. Based on these data, we speculate that when

MBs with a narrow size distribution are administered, smaller vessels will experience higher

shear stress and be more prone to BBB opening. This conjecture also describes the observed link

between vessel size and leakage type (shown in Figure 3.9(B)), where smaller vessels are more

inclined to undergo fast leakage due to the applied shear stress, while slow leakage is

predominately seen in larger vessels under trigger by minor perturbations and activation of

cellular receptors. Furthermore, this notion is in agreement with electron microscopy

78

observations where higher level of active vesicular transport of blood-borne tracer molecules was

found in the arterioles as compared to the capillaries [21].

Lastly, the enhanced permeability, time of BBBD onset, and leakage kinetics (fast vs. slow)

are affected by the applied acoustic pressure. This suggests that it is possible to control the

leakage type and tailor drug delivery for specific treatment procedures by altering the FUS

parameters. For instance, the high permeability and prompt opening associated with fast leakage

may benefit delivery of small MW drugs with short plasma half-lives. In contrast, slow leakage

may be more suitable for delivery of large MW substances, with increased plasma half-lives that

allows for extended availability for transcytocis across the BBB. In fact, by considering the

difference in plasma half-lives between 10 kDa and 70 kDa agent (e.g. 10 minutes vs. 25 minutes

[243]) and their averaged permeability constant associated with fast leakage (e.g. 0.0205 min-1 vs.

0.0110 min-1 , respectively (Figure 3.9(A)), we calculated the accumulation of each agent delivered

to the extravascular compartment over 1 hour. As a result, the fraction of concentration-time area

under the curve (AUC) between extravascular space and plasma was estimated for 10 kDa and 70

kDa agent to be 0.61 and 0.35, respectively. Meanwhile, given a low averaged permeability

constant of 0.0025 min-1 for both MWs (Figure 3.9(A)), the fraction of concentration-time AUC

between the two compartments was found to be relatively comparable (e.g. 0.11 for 10 kDa and

0.09 for 70 kDa). Therefore, this quantitative approximation supports the aforementioned

postulations on suitable delivery approaches for therapeutic agent of different MWs (e.g. fast

leakage for small MWs and slow leakage for large MWs).

3.6 Conclusions

This study demonstrates the capability to measure permeability of the compromised BBB at a

vascular level. In particular, the leakage of 10 kDa and 70 kDa dextran conjugated Texas Red (TR)

induced at the acoustic pressure range of 0.2-0.8 MPa were quantified. For both substances, a

linear regression was applied on the permeability constant against the acoustic pressure and the

slope from best-fit was found to be 0.039 ± 0.005 min-1/MPa and 0.018 ± 0.005 min-1/MPa,

respectively. In addition, the pressure threshold for successfully induced BBBD was confirmed to

be 0.4-0.6MPa. Facilitated by 2PFM imaging technique, the direct assessment of vascular

permeability and insights on its dependency on acoustic pressure, vessel size and leakage kinetics

brings us one step closer to clinical implementation of BBBD-based drug delivery.

79

4 Modelling localized delivery of

Doxorubicin to the brain based on FUS-

enhanced permeabilization of BBB 3

4.1 Introduction

The focus of this chapter is the development of a mathematical framework that potentially can be

used to guide treatment planning for FUS+MBs induced BBBD. The simulation study discussed

within this chapter aims to incorporate those experimentally-measured BBB permeability

constants from Chapter 2 as the input parameters of the underlying pharmacokinetics model. In

addition, with Doxorubicin (Dox) being an effective anticancer drug that has been used to target a

wide range of malignant cancers, including neuroblastomas [253], [254], we are motivated to

tailor the simulation study towards this drug by adopting its pre-determined pharmacokinetic

parameters such as plasma half-life, diffusion constant and cellular transport rate.

With molecular weight (MW) of 544 Da, Dox is among the chemotherapeutic substances

that are inhibited by the BBB [255]. To overcome the challenge in delivering drugs through the

BBB, it is advantageous to temporarily open these barriers and uniformly enhance their

permeability to therapeutic agents. There are several available options, including implantation of

drug polymer substrate into the brain [256], direct injection to promote convectively-assisted

transport [21], global opening the BBB via osmotic agent [257], and transcranial focused

ultrasound (FUS) [29]. Among all, FUS+MBs strategy allows for transient, reversible, and “local”

BBB permeability enhancement. At the sonicated region where local BBB has been compromised,

intravenously injected chemotherapeutic agents can exit the circulation to enter the brain

parenchyma. With respect to brain tumors, the aim is that the sonicated regions receive

therapeutic level of chemotherapy agents while the rest of the brain is protected from the

cytotoxicity effect due to the remaining functional BBB [133]. Recently, several research groups

3 Adapted from the article: Nhan T, Burgess A, Lilge L, Hynynen K. Modeling localized delivery of Doxorubicin to the brain following

focused ultrasound enhanced blood-brain barrier permeability. Phys Med Biol 2014; 59: 5987-6004.

80

have demonstrated preclinical success in distributing Dox into a confined brain volume with

intact BBB using FUS-induced permeability enhancement thereof [133], [195]. Achieving

therapeutic level at the target site resulted in a reduction in tumor size and an increase in survival

time [134], [135].

These promising preclinical findings necessitate the development of a mathematical

model which could anticipate the effect of enhanced BBB permeability on Dox delivery to the

brain within areas of intact BBB. Several simulation studies in the past had extensively

investigated the delivery of free, stealth and thermosensitive Dox to a peripheral tumor site [258]–

[262]. However, these existing models address the transport of Dox to the hepatoma rather than

the brain, and therefore do not account for permeability restrictions and increased clearance by

cerebral spinal fluid (CSF) turnover.

Tailored towards drug delivery to the CNS, a pharmacokinetic model by Patlak and

Fenstermacher was the first development that addresses the limited exchange of solute across the

BBB following systemic administration [263]. The authors also first introduced a model of the

brain as a tissue unit whose surface is bathed by CSF. The role of CSF turnover in washing the

drug out of brain tissue was further explored by Collins and Dedrick [264]. However, these initial

models did not account for the uptake and metabolism of drug by the intracellular compartment.

In addition, drug binding in both plasma and interstitial space was neglected in these previous

models, whereas high affinity to protein is prevalent phenomenon for Dox. Lastly, in the advent

of temporally and spatially increasing the BBB permeability level within the treatment region, the

exchange kinetics between plasma and interstitium requires modification to incorporate such

effect.

The pharmacokinetic model described in this paper, therefore, is a pioneering study that

considers the BBBD condition with FUS treatment while Dox is being administered

intravenously. In particular, we employ prior knowledge acquired from experimental data such

as BBB transfer constant (Ktrans) and the closure characteristics of the BBB [195], [265], as well as

other pre-determined pharmacokinetic constants for Dox to predict its temporal and spatial

concentration profiles at the target region [262], [266]. By exploring the parameters of BBB

kinetics following FUS exposure, this original work can serve as an initial step towards future

treatment planning of FUS-based drug delivery for brain pathologies.

81


4.2.1 Model geometry

For a typical half-MHz frequency and 0.8 f-number spherically curved transducer employed in

transcranial FUS-induced BBB opening, the focal volume at a target brain region exhibits an

elongated ellipsoid with half-maximum pressure amplitude diameter and length of 2.3 mm and

14 mm, respectively [170], [209]. The dimensions of focal pressure profile generated by the

transducer were measured in a water tank using hydrophone scan. Using Gd-DTPA contrast-

enhanced T1w-MRI, the sonicated volume with resulting BBBD can be identified via

hyperintense signal. Figure 4.1(A) demonstrates an axial T1w image with a lateral cross-section of

the sonication ellipsoid indicated by the white arrow. Since the axial dimension of the ellipsoidal

sonication volume is relatively greater than its lateral dimension (i.e. 14 mm versus 2.3 mm), the

gradient effect of drug concentration is more prominent along radial direction as compared to the

axial direction. Therefore, in this simulation study, we only examine the lateral cross-section of

the treatment volume by considering 2D circular model geometry with a closely defined

sonication condition. As illustrated in Figure 4.1(B), the dimension imposed on the circular core

represents the “sonicated area” that will experience the greatest changes in vascular permeability

after FUS treatment, whereas the surrounding tissue of 1.85 mm thickness indicates minimal

permeability enhancement effect. In particular, we employed a Gaussian function with a FWHM

of 2.3 mm to closely describe the spatial-variation of the permeability resulted from the focal

pressure gradient across the treatment area. The geometry and computation mesh are generated

in COMSOL Multiphysics ® (COMSOL, Inc., Burlington, MA, USA). The final mesh consists of

706 triangular elements (Figure 4.1(C)). This is deduced from a convergence test in which a 4-time

increase in mesh elements affects the simulated Dox concentration only by 1%.

4.2.2 Model assumption

With an aim to macroscopically determine the concentration profile of Dox within the sonication

region, we exclude any microscopic features such as blood vessels, cells, and the interstitial

matrix from the model. It is noted that the permeability enhancement level upon BBB disruption

depends on the vascularisation of the targeted brain region. At the macroscopic level, however,

the vascularisation effect has been included in the measured transfer constant Ktrans, which

82

represents the overall leakiness of the entire treatment region. Over a 3 mm distance (i.e. radius of

the sonicated region and surrounding periphery), variations in transport parameters on

microscopic length scale up to the order of inter-capillary distance will average out. As a result,

continuous, radially-dependent and distributed sources are used in the model equations.

Figure 4.1: (A) Contrast-enhanced axial T1w-MRI of focused ultrasound (FUS)-induced blood-brain barrier disruption (BBBD). (B) Model geometry for simulation. (C) Model mesh with 706 triangular elements. (D) Permeability kinetics of free Dox (thin red) and bound Dox (thick blue) across the BBB at FUS treatment region following a single-sonication (SS).

4.2.3 Mathematical model of drug transport and distribution

To describe Dox transport and distribution within the treatment region of the brain, three

compartments are considered: Plasma (or Intravascular), Extravascular-Extracellular, and

83

Extravascular-Intracellular, where Dox concentration associated with each compartment is

denoted as Cp(t), Ce(r,t), and Ci(r,t), respectively (Table 4.1).

4.2.3.1 Plasma compartment

Following an IV administration, Dox plasma concentration Cp(t) decreases over time due to the

clearance effect. For a bolus injection, plasma pharmacokinetics of Dox can be described as a

triexponential decay function:

α β (4.1)

where D is the total injection dose; A, B,C are compartmental distribution parameters and α, β,

are the elimination rate constants. These values are extracted from past experimental

investigation where plasma of rat was collected after the injection of Dox and sampled at various

time points [266]. For continuous infusion, the time-dependent plasma profile is given by:

α α

β β

(4.2a)

α α α

β β β

(4.2b)

where T is the infusion duration [258].

To account for high affinity of Dox to plasma proteins (e.g. albumin), Dox plasma

concentration is partitioned into free Dox, Cpf, and bound Dox, Cpb, with distinct molecular

weights (MW of 544 Da and 70 kDa, respectively) [262]. Their concentration profiles, Cpf and Cpb,

are directly related to the total plasma concentration:

(4.3)

(4.4)

where s=0.75 is the binding fraction adopted in these simulations [262].

4.2.3.2 Extravascular-extracellular compartment

Dox concentration within the extravascular-extracellular compartment is governed by the

subsequent transport processes: interaction with the plasma proteins, exchange across the vessel

84

walls, diffusion in the interstitial fluid (ISF), clearance by the cerebral spinal fluid (CSF) and

uptake by the brain cells [262]. As a result, free Dox and bound Dox concentration in this

compartment, Cef(r,t) and Ceb(r,t), can be described by the following rate equations [262], [264]:

(4.5)

(4.6)

The first term in right-hand-side of both equations accounts for the diffusion process, with

Def and Deb representing the diffusion coefficient of free Dox and bound Dox, respectively. The

second term describes the gain of Dox from blood-brain exchange that is underlined by the BBB

permeability enhancement, Pf and Pb, respectively. The third term depicts the loss of Dox due to

CSF clearance at a replenish rate of Kcsf. The conversion (Sb) of free Dox from bound Dox can be

expressed as [262]:

(4.7)

where kd and ka are the dissociation and association rate constants.

Since only free Dox can cross the cell membrane to enter the intracellular space, the source

term for cellular uptake (Su) only appears in the rate equation of Cef (Equation 4.5) [262].

Considering both efflux and influx of Dox into and out of the intracellular compartment, the

cellular uptake rate is given by [262]:

(4.8)

where ρ is the cell density, and Rm is the transmembrane transport rate. Michaelis-Menten

kinetics formula are applied for the influx and efflux functions, where ki and ke are the reaction

rates deduced from experimental data [261], and φ=0.4 is the volume fraction of extracellular

space.

4.2.3.3 Intracellular compartment

In direct relation to the cellular uptake rate Su, intracellular concentration of free Dox is given by

[262]:

(4.9)

As the key variable to be evaluated from the simulation, Ci(r,t) has a unit of ng/105 cells.

However, to validate against experimental measurement, we convert Ci(r,t) into a final unit of

ng/g by assuming that 1 g wet tissue contains 109 cells [267].

85

4.2.4 Model parameters

Table 1 summarizes all the constants relevant to Dox pharmacokinetics and tissue properties that

are applied throughout Equations 4.1 - 4.9. To conform to a clinical dose of 5 mg/kg, D=1.25 mg

is used by assuming a rat weight of 250 g. To incorporate the transient amplification and gradual

closure of vascular permeability upon FUS-induced BBB opening at the sonication zone, we

express Pf and Pb in Equation 4.5 - 4.6 in form of an exponential decay function:

(4.10)

where the subscript i represents either f (free Dox) or b (bound Dox). Ei is the initial permeability

enhanced by FUS application, and Ri is the decay rate due to closure of the BBB. Past

experimental studies had confirmed that the extent of permeability as well as the closure time of

the BBB are strongly dependent on molecular size [173], [265]. To integrate this effect for free Dox

and bound Dox, we correlate Ef and Eb to the Ktrans value of Gd-DTPA. Since Gd-DTPA is a

typical MRI contrast agent that has been safely used as a surrogate tracer for monitoring and

guiding FUS-induced BBB opening procedure, its Ktrans constant, readily obtained from the DCE-

MRI, is a good indicator of BBB leakiness. Considering Gd-DTPA’s MW of ~1kDa, we extrapolate

Ef and Eb for free Dox (0.5 kDa) and bound Dox (70 kDa), based on the inverse relationship with

their sizes (i.e. Ei/E1kDa = 1 – 0.5 log(MWi)) [196], [211], [265].

(4.11)

(4.12)

Given Gd-DTPA’s Ktrans of 0.01 min-1, which can be readily measured via DCE-MRI, Ef and

Eb will take value of 1.15e-2 min-1 and 0.8e-3 min-1, respectively. On the other hand, size-

dependent reversibility kinetics of the BBB for free Dox and bound Dox are extracted from a

study by Marty et al. [173], where the time window of BBB passage for a given nanoparticle size

was measured and a theoretical model was proposed to predict the half closure time as a function

of the hydrodynamic diameter of the nanoparticle. Based on their theoretical model, we

concluded that Rf =8.6 h and Rb=3.6 h for free Dox and bound Dox, respectively.

Applying these above-mentioned values for Ef, Eb, Rf and Rb to Equation 4.10, we plot the

permeability kinetics profiles, Pf(t) and Pb(t), at the center of the sonicated region (Figure 4.1(D)).

86

In essence, due to its smaller size, Pf-sonicated(t) exhibits higher amplitude and longer opening than

Pb-sonicated (t). As revealed in the semilog plot, once the BBB has closed the permeability of free Dox

and bound Dox at sonicated region return to the minimal permeability level of the normal intact

tissue, which is 100-fold lower than Pf-sonicated and Pb-sonicated at t=0 [195], [265].

In the simulation study, we explore three key treatment aspects that have influence on the

outcome of Dox delivery. First, we examine the different sonication schemes which are based on

the number of sonications (e.g. single, double, and triple) and the duration between two

consecutive treatments (e.g. 10 min, 30 min, 60 min, 120 min). Second, we investigate the BBB

permeability enhancement level indicated via the Ktrans constant. Last, various modes of Dox

intravenous administrations (e.g. bolus injection, continuous infusion ranging from 10 min to 360

min) were studied.

4.2.5 Boundary conditions

Continuity is imposed at the boundary between the sonication core and the surrounding area. On

the outer edge of the normal tissue layer, Neumann boundary condition is applied to ensure

insulation and symmetry [262].

4.2.6 Numerical methods

To obtain spatial and temporal solution for Cef(r,t), Ceb(r,t) and Ci(r,t), a system of 3 coupled

differential equations (Equation 4.5, 4.6, 4.9) are solved simultaneously using the Convection and

Diffusion Module in COMSOL Multiphysics 3.5a. The time-dependent solver is selected with a

relative tolerance of 0.01 and an absolute tolerance of 0.001. The direct linear system solver

(UMFPACK) is used. The generalized alpha method is set for time stepping with free time step

taken by the solver at maximum interval of 50 seconds (s). This value is chosen based on our

time-step sensitivity tests. The simulation is run to cover a total of 48 hours (h).

87

Table 4.1: Pharmacokinetics and pharmacodynamic parameters of Doxorubicin

Parameter Description Value Unit Reference

S Binding ratio of Dox-protein 0.75 − [262]

A Plasma compartment distribution 2.14e-3 ml-1 [266]

B Plasma compartment distribution 2.37e-4 ml-1 [266]

C Plasma compartment distribution 6.64e-5 ml-1 [266]

α Elimination rate of Dox 2.16e-3 s-1 [266]

β Elimination rate of Dox 6.36e-5 s-1 [266]

Elimination rate of Dox 1.11e-5 s-1 [266]

D Injection dose 1.25 mg [195]

T Infusion duration 10 – 360 min [258]

Def Diffusion constant of free Dox 1.58e-10 m2/s [261], [268]

Deb Diffusion constant of bound Dox 4.17e-12 m2/s [269], [270]

Kcsf Replenish rate of CSF 1.4e-4 s-1 [271]

kd Dissociation rate of bound Dox 0.278 s-1 [261]

ka Association rate of free Dox 0.833 s-1 [261]

ρ Cell density 1e10 105cells/m3 [261]

Rm Transmembrane transport rate 4.67e-15 kg/(105cells s) [258]

ki Cellular efflux rate 1.37e-12 kg/m3 [272]

ke Cellular influx rate 2.19e-4 kg/m3 [272]

φ Volume fraction of extracellular space 0.4 − [258]

88

4.3 Results

4.3.1 Increase in Dox delivery by FUS induced BBB permeability

To elucidate the effect of FUS on improving Dox delivery at a target brain region, we consider the

BBB permeability kinetics of free Dox and bound Dox in the sonicated area and surrounding

tissue given Gd-DTPA’s Ktrans of 0.01 min-1 (Figure 4.1(D)). In effect, the spatial-temporal

distribution map of intracellular Dox concentration is shown in Figure 4.2(A). Figure 4.2(B)

displays distinct spatial profiles of the intracellular concentration at various time points, ranging

from 6 h to 48 h. As evident from both the 2D map and 1D line profiles, following a bolus

injection of Dox and FUS application, intracellular concentration Ci peaks at t=6-12 h and tapers

off thereafter. More importantly, high intracellular Dox concentration is localized within the

sonicated domain (r < 1.25 mm), whereas the surrounding edge (1.25 mm < r < 3 mm) is spared

from high toxicity level by maintaining below 150 ng/g throughout 48 h post treatment.

Figure 4.2: (A) 2D map depicting spatial (x-direction) and temporal (y-direction) distribution of intracellular Dox concentration at sonicated region and surrounding tissue followed a single-sonication. Dash line represents the boundary between the sonicated region and the surrounding tissue. (B) Spatial profiles of intracellular Dox concentration at 6h - 48h.

89

To quantitatively emphasize the ability of FUS-induced BBB opening to target a specific

treatment region without affecting the surrounding tissue environment, we evaluate the

spatially-averaged concentration associated with each domain over time. Spatial-mean profiles of

extravascular-extracellular concentration for free Dox and bound Dox, Cef(t) and Ceb(t) (Figure

4.3(A)) as well as intracellular concentration, Ci(t) (Figure 4.3(B)) within the sonicated zone (solid

curves) substantiate their counterparts associated with the surrounding periphery (dashed

curves). In comparing the range of y-axis in Figure 4.3(A) and Figure 4.3(B), we also note that

Cef(t) and Ceb(t) curves in Figure 4.3(A) are significantly lower than Ci(t) level. This indicates that

Dox from the extravascular space are exhaustively utilized throughout various physiological

processes (vascular exchange, CSF clearance processes and cellular uptake) with free Dox being

prominently taken up by brain cells. Considering an IV dose of 5 mg/kg and assuming a murine

blood density of 77 g blood per kg of body weight, we expect Cp to be 65 µg/g. In contrast, Figure

4.3(A) indicates the peak values of Cef and Ceb in the sonicated region to be 2 ng/g and 5 ng/g,

respectively; whereas the peak value of Ci in the sonicated domain is 280 ng/g as revealed in

Figure 4.3(B).

Therefore, we can deduce the overall relationship among these concentration entities: Cp

>> Ci >> Ceb > Cef. Overall, the extravascular Dox concentration is Ce = Ceb + Cef = 6 ng/g, which

is 40-fold lower than Ci. This ratio between Ci/Ce is similar to that observed in other studies [258],

[262]. In Zhan et al.’s work, for 2 hour infusion of Dox, the simulation results show that Ci peaks

at 1.5 ng/105 cells or 1.5 x 104 ng/g tissue (by assuming 1g tissue contains 109 cells). On the other

hand, their findings suggest Ce = Ceb + Cef = 4 x 10-4 kg/m3 = 400 ng/g (by assuming tissue

density of 1000 kg/m3). This indicates the ratio between Ci and Ce to be 37.5. In the study by El-

Kareh and Secomb, the authors also found that for the total amount of drug initially in the

circulation of 100 mg, the peak extracellular drug concentration is 2 µg/ml whereas a typical peak

intracellular concentration is 60 µg/ml. The implication for such dominant ratio of Ci versus Ce is

the extremely efficient uptake of Dox by the cell.

90

Figure 4.3: Time-dependent spatial-mean free Dox (thin red) and bound Dox (thick blue) concentration in the (A) extravascular-extracellular compartment and (B) intracellular compartment of the sonicated region (solid) and the surrounding tissue (dashed). Note: The range of y-axis in (A) is significantly lower than that in (B).

4.3.2 Compare the effect of sonication schemes on Dox delivery

In fixing the injection dose of Dox and applying different sonication schemes (e.g.. single

sonication (SS), double sonication with 10 min interval (DS10), double sonication with 120 min

interval (DS120)), Park et al. had demonstrated that DS10 yields the optimal delivery of Dox

across the BBB [195]. From their DCE-MRI study, the authors concluded that a second sonication

would raise the Ktrans level twice as high as the first sonication and prolong the time window of

BBB opening.

Here, we modelled these multiple sonication conditions by adopting the kinetic

characteristics of BBB permeability in correspondence to each sonication scheme for the

simulation input. Aside from exploring the separation interval of 10 min and 120 min between

two consecutive sonications, we further investigate two other intermediate durations (e.g. 30 min

and 60 min) as well as expanding the simulation study to examine a triple sonication (TS)

condition to elicit the level of improvement in drug delivery.

91

The simulation results for DS are summarized in Figure 4.4(A)-(C), whereas data

corresponding to TS are presented in Figure 4.4(D)-(F). In the first four panels, permeability

kinetics curves of free Dox (top row) and bound Dox (second row) are compared among multiple

sonication at a fixed delay interval (e.g. 10 min, 30 min, 60 min, 120 min) against the control case

(i.e. no sonication) and SS. Given an initial Ktrans of 0.01 min-1, depending on when the next

sonication occurs, an instantaneous increase in the permeability amplitude of 0.01 min-1 will have

an additive effect on the current permeability level that is remaining from the previous sonication

session. After each FUS-triggered enhancement, the permeability curve undergoes the decay due

to restoration of the BBB.

In response to these distinct permeability kinetic profiles, spatially-averaged intracellular

Dox concentrations Ci(t) in the treatment domain are simulated and plotted as a function of time.

In Figure 4.4(C), we note that all four Ci(t) profiles corresponding to DS scheme with four

different delay intervals are superior to the SS-based profile, while SS offers two-order of

magnitude improvement in Dox deposition as compared to the control case. Similarly, relative to

SS and the control case, TS schemes yield significant enhancement of Dox delivery to the

intracellular compartment (Figure 4.4(F)).

To assemble the therapeutic effectiveness of all sonication schemes explored in the

simulation study, their corresponding temporal peak values of Ci(t) are contrasted in Figure

4.4(G). Overall, TS schemes result in the greatest Dox delivery, followed by DS and finally SS.

However, while DS exhibits low sensitivity to the delay interval between two consecutive

sonications, a much wider spread in temporal peak of intracellular concentrations (ranging from

435 ng/g to 565 ng/g) is observed for TS10, TS30, TS60 and TS120. Lastly, we note that both DS

and TS schemes achieve optimal Dox deposition at the targeted region with 10 min delay interval

in comparison to other delay time frame. This notion corroborates Park et al.’s experimental

results, showing that DS10 provides greatest drug delivery among three sonication scenarios

under investigation (SS, DS10 and DS120).

92

Figure 4.4: (A)-(C) Effect of double-sonication (DS): Permeability kinetics of free Dox (A) and bound Dox (B) are contrasted among DS of various intervals (10 min, 30 min, 60 min, 120 min) against the Control and single-sonication (SS). (C) Time-dependent spatially-averaged profiles of intracellular Dox concentration within the sonicated region are contrasted among Control, SS, DS10, DS30, DS60, DS120. (D)-(E) Effect of triple-sonication (TS): Permeability kinetics of free Dox (D) and bound Dox (E) are contrasted among TS of various intervals (10 min, 30 min, 60 min, 120 min) against the Control (no sonication) and single-sonication (SS). (F) Time-dependent spatially-averaged profiles of intracellular Dox concentration within the sonicated region are contrasted among Control, SS, TS10, TS30, TS60, TS120. (G) Comparison of temporally-peaked spatially-averaged intracellular Dox resulting from different sonication schemes: Single columns represent the Control and SS case whereas double columns are associated with DS (filled) and TS (striped) at various delayed intervals.

93

4.3.3 Effect of BBB permeability enhancement level on Dox delivery

Previous studies using MRI and two-photon microscopy had demonstrated that the degree of

BBB permeability enhancement can be controlled by adjusting the applied acoustic pressure [196],

[265]. For instance, in response to MB injection at clinical dose and FUS application at a frequency

range of 1-1.5 MHz, BBB permeability of 0.01-0.04 min-1 is linearly achievable by tuning the

acoustic pressure from 0.2-0.8 MPa [195], [196], [265]. To translate the effect of BBB permeability

enhancement to therapeutic implication of Dox delivery to the brain, we compute the temporal

peak intracellular concentration from the spatial-mean profile as a function of Ktrans.

Based on their optimal sonication spacing interval, we select DS10 and TS10, along with

SS, to study their dependence on Ktrans. Temporally-peaked spatially-averaged intracellular Dox

resulting from these 3 sonication scenarios are summarized in Figure 4.5 over the clinically-

relevant Ktrans range of 0.01-0.04 min-1. While Dox concentration measurement in Park’s study was

done at 16 hours post sonication, temporal peaks of spatially-averaged intracellular Dox

concentration profiles from our simulation results occur between 6 and 12 hours (Figure 4.4(C)

and 4F). Nevertheless, it is reasonable to compare the temporally-peaked spatially-averaged

value from the simulation to the concentration measurement in Park’s experiment due to the slow

fall-off at t = 16 hours with respect to peaked values. Firstly, we note that our simulation predicts

the DS10-induced Ci concentration of Dox over the Ktrans range of 0.01-0.03 min-1 to be 400-1200

ng/g tissue, which agrees well with the experimental findings from Park et al. For all 3 sonication

schemes, increasing Ktrans will lead to the improved Dox accumulation in brain cells. Displaying

the steepest slope, a linear regression curve associated with TS10 suggests this sonication scheme

result in an improvement of Dox delivery by 1.3-fold and 2.2-fold relative to DS10 and SS,

respectively. Given that the therapeutic dose of 819 ng/g tumor is correlated with a clinical

response in breast cancer patients [273], we use this level of Dox concentration as benchmark

(shown by the red dotted line in Figure 4.5) to demonstrate that it can be exceeded by FUS

induced permeability enhancement of 0.01-0.04 min-1.

94

Figure 4.5: Temporally-peaked spatially-averaged intracellular Dox concentration within the sonicated region as a function of Ktrans, an indicator of the blood-brain barrier (BBB) permeability enhancement. Dotted line (red) represents therapeutic level of Dox resulting in a clinical response for human tumors in vivo. SS, single-sonication; DS10, double-sonication of 10 minute interval; TS10, triple-sonication of 10 minute interval.

4.3.4 Effect of injection modes on Dox delivery

Despite the promising evidence that therapeutic level of Dox can be reached for effective cell-

killing activity at the FUS-treated region, the cardiotoxicity effect of Dox remains a major concern

for clinical practices. Therefore, we further utilize the model framework to determine the optimal

mode of injection in this context of transient BBB opening and closure. In particular, we compare

bolus injection against continuous infusion of various durations ranging from 10 min to 360 min,

of which their associating peaked plasma concentrations are presented in Figure 4.6(A). Overall,

the bolus injection leads to the highest peak plasma concentration whereas longer infusion

substantially reduces the peak value. For instance, 360 min infusion lowers the peak plasma

concentration by 10-fold relative to the bolus injection.

In response to these distinct injection modes, intracellular Dox concentrations are

simulated for each sonication scheme (SS, DS10, TS10) and their temporally-peaked spatially-

95

averaged values are shown in Figure 4.6(B)-(D), respectively. For SS, highest intracellular Dox

concentration is attainable with the bolus injection. However, the intracellular Dox concentration

level remains relatively high and plateau over the 10-60 min infusion. In contrast, DS10 and TS10

exhibit the best outcome with infusion over a time window of 20-30 min and 30-60 min,

respectively. One similar trend shared among all three sonication scenarios (Figure 4.6(B)-(D)) is

the drop-off in Dox delivery with respect to their associated peaked concentration when the

infusion duration is extended beyond 60 min. This feature suggests that, for each sonication

scheme, a long period infusion will not warrant the greatest deposition of Dox to the targeted

brain region despite its subdued peak plasma concentration.

4.4 Discussion

We have established a mathematical framework to simulate the efficacy of delivering Dox into

brain cells within the FUS-induced BBB opening region. We adopted pre-determined

pharmacokinetic parameters for Dox (e.g. plasma half-life, albumin binding ratio, diffusion

constant, cellular influx and efflux rate, etc.) in conjunction with permeability kinetics of Dox

across a compromised BBB under FUS treatment (e.g. initial amplitude of Ktrans, half closure time).

The results from Figure 4.2 demonstrate a relatively low drug concentration beyond the

sonication zone. This attribute supports the appropriateness of the 2D circular assumption for an

insonation ellipsoid of 2.3 mm by 14 mm. While keeping the simulation inputs consistent with

experimental conditions (e.g. injection dose, Ktrans value, sonication dimension), the simulation

outputs agree well with experimental data reported by Park et al [195]. Firstly, the Dox

concentrations achieved with three different sonication schemes (SS, DS10, DS120) are in

consensus with Park’s observation that DS10 yields the optimal delivery. Secondly, our

simulation shows Dox concentration delivered to the intracellular compartment is linearly

correlated to the Ktrans constant [195]. Finally, our predicted Dox concentration in response to the

Ktrans range of 0.01-0.03 min-1 varies between 400-1200 ng/g, which is congruent with the scope

reported by Park et al.

96

Figure 4.6: Effect of injection mode (bolus injection and infusion over different durations) on Dox delivery. (A) Temporally-peaked plasma. (B)-(D) Temporally-peaked spatially-averaged intracellular Dox concentration within the sonicated region followed: (B) Single-sonication (SS), (C) Double-sonication of 10 minute interval, (D) Triple-sonication of 10 minute interval.

Despite these agreements between our model prediction and their experimental data, two

major discrepancies were noted. By enforcing minimal permeability for the control side of the

brain by setting its Ktrans level two orders of magnitude below the FUS-mediated permeability,

our simulation reveals an extremely low Dox concentration for the control case (e.g. 6 ng/g

tissue) whereas the experimental data showed a substantially higher concentration of 300 ng/g

tissue. Given that an intact BBB is impermeable to lipid-soluble substances larger than 400 Da

[274] due to the presence of TJs and highly-expressed P-glycoprotein (P-gp) [12], it is unexpected

to observe a significant level of Dox concentration in an untreated brain. Such difference between

our low concentration forecast and the high reported measurement remains unexplained.

97

However, we note that another group has reported a value of 5 ng/g for the control case when

using mass spectrometer to quantify Dox concentration in their tissue sample [275].

The second discrepancy between our model and published experimental data is the

concentration level of DS120 with respect to SS. Our simulation suggests DS120 is less effective

than DS10 while remaining better than SS. On the contrary, Park et al. showed no difference

between DS120 and SS when Dox concentration was averaged for different animals over a large

range of Ktrans. Such discrepancy warrants further investigation in future study, when more

relevant experimental data are available. Nonetheless, it is certain that if multi-sonication (e.g.

double or triple sonication) is considered for the treatment planning, a narrow time window for

the subsequent sonication following the previous session would be recommended (Figure 4.4(G)).

In general, as corroborated from both empirical evidence and simulation result, it is

promising that FUS allows for sufficient delivery of high and localized distribution of Dox in the

treatment region of the brain. With the sonication area dictated by the ultrasound frequency in

the range of 0.5-1.5 MHz, a target diameter below 2.5 mm can be achieved (Figure 4.1(A)). As a

result, highest concentration is contained within the center of sonicated core, while the

surrounding tissue is spared from elevated level of toxicity (Figure 4.2). At a threshold dose of

819±482 ng/g that was considered as the therapeutic benchmark based on clinical response in

breast cancer patients, Figure 4.5 confirms the Ktrans level of 0.02 min-1 is adequate to reach the

required threshold when DS10 or TS10 is employed. Moreover, in the context of brain glioma

cells, the therapeutic dose might even be lower. For instance, in vitro study by Muldoon et al.

revealed that half maximal cytotoxic dose (EC50) of Dox varies from 0.03-0.07 µg/ml, which is

equivalent to 30-70 ng/g tissue [276]. A lower level of Ktrans translates to reduced acoustic

pressure and MB concentrations required to induce BBB opening, which ultimately implies lesser

risk of red blood cell extravasations, petechia, or tissue damage [196], [277].

The last parameter explored in this study is the optimal injection mode to be employed

during FUS-induced BBB opening treatment. It is noted that the majority of preclinical

experiments on Dox delivery had been conducted with a bolus administration. Despite its

convenience for experimental procedure, there is a major concern in clinical practice regarding its

associated cardiotoxicity which is strongly tied to the peak plasma concentration [258]. In this

study, by maintaining the same injection dose yet applying continuous infusion, we notice the

98

peak plasma concentration is significantly suppressed by stretching the infusion period (Figure

4.6(A)). However, for all three sonication schemes (SS, DS10, TS10), simulation results reveal that

peaked intracellular Dox concentration begins to taper off when infusion is extended beyond 60

min, whereas shorter infusion time (between 20-60 min) allows for the balance between the

minimal cardiotoxicity effect and the sufficient Dox delivery. This simulation finding also

suggests future experimental studies should investigate and evaluate the benefit of continuous

infusion of different intervals.

Generalization of the model: Despite the specific examination of Dox delivery in this

simulation study, the established mathematical model can be adapted to other therapeutic agents.

In that circumstance, cellular uptake constants (e.g. efflux/influx rate and transmembrane

transport rate) would be adjusted accordingly for the drug of choice. Difference in drug size also

affects the diffusion, plasma clearance rate as well as the extent and closure time of the BBB.

Lastly, in addition to sonication scheme and injection method investigated throughout this work,

sensitivity study should also be conducted on the delivery dose to determine the combination of

optimal parameters for guiding the treatment planning procedure [139], [164].

An example of model adaptation for Methotrexate: In the following section, we demonstrate

the adjustment of the current mathematical model towards Methotrexate (MTX). Furthermore,

we could compare the simulation results to the experimental data reported by Mei et al. [172]. In

that study, the authors applied FUS and MBs for targeted delivery of MTX to the rabbit brain. In

particular, 20 mg/kg of MTX was injected into the ear vein of rabbits (weighing between 2.5 – 3.5

kg). FUS was applied at 1.1 MHz and the respective lateral and longitudinal beam width were

measured to be 3 mm and 8 mm, respectively. The right hemispheres of the brains received FUS

exposure and 0.03 ml/kg SonoVue ultrasound contrast agent was administered. T1w fast spin

echo contrast-enhanced MRI was performed to confirm successful BBB disruption and the

animals were sacrificed 1 hour after FUS treatment. MTX concentrations in the targeted sites were

determined by high-performance liquid chromatographic analytical procedure. As shown in

Table 2 of their paper, the averaged MTX concentration of this treated group (n=5) was reported

to be 7.412 ± 1.471 µg/g tissue. (1)

With a MW of 454 Da, we expect MTX to exhibit similar size-dependent pharmacokinetic

properties with Doxorubicin including permeability across the BBB and the diffusion constant in

99

the ISF. If we assume sonication conditions in Mei’s paper resulted in a similar Ktrans level of 0.01 -

0.04 min-1, we can approximately estimate the MTX concentration delivered to the intracellular

space based on our simulation. For a single sonication (which was implemented in Mei’s study),

Figure 4.4C in our simulation study predicts a drug concentration of 100 ng/g tissue at t = 1 hour

(which is the time point that the rabbits were sacrificed in Mei’s work). Based on the relationship

between the temporally-peaked spatially-averaged drug concentration and Ktrans depicted in

Figure 4.5 of this thesis work, we further anticipated the spatially-averaged drug concentration in

the intracellular space at t = 1 hour to be approximately 100 – 500 ng/g tissue when Ktrans ranges

from 0.01 – 0.04 min-1. (2)

However, we noted that Mei’s experiment used a dose of 20 mg/kg for IV-administered

MTX, which is 4-fold higher than Dox (i.e. 5 mg/kg). Another scalable difference is the reduced

binding of MTX to plasma protein in blood [278]. While the plasma concentration of free Dox is

Cpf = 0.25Cp (where Cp is the total plasma concentration), the plasma concentration of free MTX

becomes Cpf = 0.5Cp , two-fold higher than the plasma concentration of free Dox (see Equation 4.3

in Chapter 4). Therefore, these two distinct aspects lead to a MTX concentration that is 8 times

higher than Dox concentration. (3)

Combining arguments (2) and (3) above, we estimated the spatially-averaged MTX

concentration in the intracellular space at t = 1 hour to be approximately 800 – 4000 ng/g tissue

(or 0.8 – 4.0 µg/g tissue) when Ktrans ranges from 0.01 – 0.04 min-1. Overal, this quick set of

calculations allows us to roughly estimate the expected MTX concentration in order to compare

against the value reported in Mei’s study (7.412 ± 1.471 µg/g tissue), as summarized in argument

(1). While the upper limit of simulation-derived MTX concentration (4.0 µg/g tissue) is relatively

comparable to experimental value (7.412 ± 1.471 µg/g tissue), it is worthy to note that this

estimation assumes MTX and Dox exhibit similar triphasic plasma half-life constants as well as

influx and efflux cellular uptake rates. Overall, the presented estimation of MTX concentration in

the intracellular compartment serves as an example of how the established mathematical

framework could be redesigned for other drugs that are different from DOX.

100

4.5 Conclusions

This simulation study presents the unprecedented mathematical framework to closely depict the

delivery of chemotherapeutic agent Dox into a local volume of the brain using FUS-induced

BBBD. This is accomplished by considering Dox concentrations within three compartments

(plasma, extracellular, intracellular) and accounting for various transport processes (e.g. diffusion

in interstitial space, exchange across vessel wall, clearance by cerebral spinal fluid, uptake by

brain cells). In addition to adopting pre-determined pharmacokinetic parameters for Dox (e.g.

plasma half-life, albumin binding ratio, diffusion constant, cellular influx and efflux rate), we

mathematically depict permeability kinetics of free and bound Dox across a compromised BBB

under FUS treatment based on the initial amplitudes of Ktrans constant and their half closure

times. In examining several clinical treatment factors (e.g. sonication scheme, permeability

enhancement, injection mode), the simulation outputs agree well with experimental data

reported by Park et al [195]. Firstly, the Dox concentrations achieved with three different

sonication schemes (SS, DS10, DS120) are in consensus with their observation that DS10 yields the

optimal delivery. Secondly, our simulation shows Dox concentration delivered to the intracellular

compartment is linearly correlated to the Ktrans constant and its range of 400-1200 ng/g in

response to the Ktrans range of 0.01-0.03 min-1 is congruent with their fluorometrically-measured

Dox concentration [195]. Finally, the model suggests that infusion over a short duration (20-60

min) should be employed along with single-sonication or multiple-sonication at 10 min interval

to ensure maximum delivery to the intracellular compartment while attaining minimal

cardiotoxicity via suppressing peak plasma concentration. While the current model is pertinent to

normal brain tissue in order to ensure a comparable condition as performed by Park et al [195] for

validation purposes, future work in this area will extend to tumor condition of malignant brain

tissue as well as explore the potential drawback of efflux transporters at the BBB and optimal

dose of intravenously-administered Dox.

101

5 Conclusions & Future Work

5.1 Summary of findings

The motivation for this thesis work stemmed from the need to investigate the biophysical

behavior of BBBD induced by FUS+MBs in vivo at a vascular level. Such high spatial and

temporal resolution requirements called for the adoption of 2PFM as the imaging modality over

the commonly-used MRI technique.

In order to incorporate FUS into the existing 2PFM imaging system at Sunnybrook

Research Institute, a suitable transducer design was fundamental. In contrast to the MRI setting,

where both the transducer and positioning system can reside in the MR magnet bore without

interfering with the imaging process [194], the 2PFM setup imposes constraints on the allowable

size and shape of the transducer. In addition, optical transparency along the path between the

microscope objective and the targeted sonication region must be maintained. To accommodate

these technical necessities, a comprehensive list of design specifications was laid out in Chapter 2,

which ultimately confirmed a ring-shaped transducer as the most appropriate design. The inner

diameter of the transducer was sufficiently large (8.5 mm) to allow the transducer to fit around

the objective lens. The outer diameter (10 mm) of the transducer was small enough to sit over one

hemisphere of a rat’s brain. For optimized transducer dimensions, its two vibration modes

(thickness and height) were characterized by mapping the acoustic pressure field profile using an

optical fiber hydrophone. From these measurements, we found that both vibration modes

generate a circularly symmetrical pressure profiles in the lateral plane with a FWHM of 500 µm.

This lateral dimension of the focal zone provided sufficient overlap with the typical 512 x 512 μm2

field of view (FOV) of the 2PFM imaging plane. In the axial direction, however, the thickness

mode results in a significantly shallower depth of field (DoF) compared to the height mode. The

former exhibits a pressure peak 1-1.5 mm below the coverslip, whereas the latter results in a

deeper focal zone 4-4.5 mm away from the coverslip. Furthermore, the thickness mode yields a

tighter axial focus with 70% of its acoustic energy contained within 4 mm depth. In contrast, this

same amount of acoustic energy is spread over a 10 mm range when the transducer is driven in

the height mode. These analyses helped determine the most appropriate vibration mode for the

102

transducer in order to induce BBBD at a brain region accessible to 2PFM. Considering the 2PFM

imaging field extends only 1 mm in depth from the coverslip surface, such differences in the

location and extent of the focal spot generated by these two modes of vibration should be noted.

To ensure the transducer could generate sufficient pressure for BBBD induction, an electrical

input power up to 2.12 W was applied and the acoustic output from the US pressure field profile

was measured. In doing so, an adequate pressure level of 1 MPa was confirmed without causing

thermal damage to the transducer. Lastly, by driving the transducer in the thickness mode and

varying the applied pressure from 0.2 MPa to 0.8 MPa, real-time 2PFM evidence of successful

BBBD was obtained. Prior to sacrificing the rat, EB was injected into the subject intravenously and

allowed for EB extravasation into the sonicated brain region. The brain was subsequently

harvested for histology examination. From coronal sections of the sonicated hemisphere, the EB-

stained region displayed a superficial profile similar to those found for the thickness mode from

the in vitro characterization. All together, these analyses substantiate the suitability and

robustness of the proposed transducer design for concurrent FUS exposure and 2PFM detection

of BBBD induced on the dorsal surface of a rat brain.

As outlined in Chapter 3, the transducer and imaging system were deployed to achieve

statistically-significant data sets that captured the BBBD process. Using 10 kDa and 70 kDa

dextran conjugated Texas Red as vascular markers, the leakage of these fluorescent dyes was

captured once BBBD was successfully triggered. A quantitative algorithm was developed to

analyse these 4D data sets (XYZT) and measure the corresponding permeability of the

compromised BBB at a vascular level. From this study, we were able to map the permeability as a

function of drug size (i.e. MW) and applied acoustic pressure. For instance, a linear regression

was applied on the permeability constant against the acoustic pressure, and the slope from the

best-fit was found to be 0.039 ± 0.005 min-1/MPa and 0.018 ± 0.005 min-1/MPa for MW of 10 kDa

and 70 kDa, respectively. By closely monitoring the time point when leakage was initiated, we

also confirmed an inverse relationship between the temporal onset of BBBD and the permeability

level. Moreover, the temporal onset of BBBD was found to be dictated by the applied acoustic

pressure. Specifically, higher pressure input appears to yield an immediate BBBD onset, whereas

delayed leakage is prevalent at a lower acoustic pressure. Lastly, from these microscopic images,

we were able to measure the diameter of vessels undergoing disruption and correlate this

parameter to the enhanced permeability. In general, an inverse relationship was noted between

103

these two entities. In other words, high permeation was more commonly seen in vessels of 10-40

µm in diameter, whereas low permeation occurred more often in vessels of 40-70 µm in diameter.

Overall, the direct assessment of vascular permeability and observations of its dependency on

acoustic pressure, vessel size and leakage kinetics provides valuable information as this

technique moves towards clinical implementation.

To integrate these experimental findings into a guided treatment planning procedure, the

last component of my thesis work focused on the development of a mathematical framework that

would closely depict the spatio-temporal distribution of a therapeutic agent into a local volume of

the brain in the context of FUS+MBs induced BBBD. The simulation study was performed using

the properties for Doxorubicin (Dox), which is an effective anti-cancer drug that can be used to

target a wide range of malignant cancers and has shown positive outcome on tumor-bearing rats

when BBBD was incorporated [134], [135]. Additionally, Dox was chosen for the investigation

due to the availability of existing pharmacokinetic parameters relevant to the drug (e.g. plasma

half-life, albumin binding ratio, diffusion constant, cellular influx and efflux rate), which can be

conveniently applied as the model inputs. As further required for the model inputs, the

permeability properties of the compromised BBB (as shown in Chapter 3) were incorporated into

the mathematical framework. In particular, we expressed the permeability kinetics of free and

bound Dox at the disrupted BBB in the form of exponential decay functions with specific initial

amplitudes and half-closure times. Lastly, the model also considered other relevant transport

processes (e.g. diffusion in the interstitial space, exchange across the vessel wall, clearance by

cerebral spinal fluid, uptake by brain cells). Given that Dox concentrations in three separate

compartments (i.e. plasma, extracellular, intracellular) were tracked as simulation outputs, we

were able to evaluate the delivery efficacy in response to different clinical treatment conditions

(e.g. sonication scheme, permeability enhancement, and injection mode). Overall, we found that

our simulation outputs were in agreement with the experimental data reported by Park et al [195].

Firstly, the Dox concentrations achieved with three different sonication schemes (SS, DS10,

DS120) consistently agreed on DS10 as the optimal regime. Secondly, our simulation shows that

the Dox concentration delivered to the intracellular compartment is linearly correlated to the

Ktrans constant. As well, its range of 400-1200 ng/g in response to the Ktrans range of 0.01-0.03 min-1

align with their fluorometrically-measured Dox concentration [195]. Finally, an optimal injection

mode was explored in the simulation by comparing intracellular concentrations of the drug

104

corresponding to bolus and infusion over different periods (e.g. between 10 minutes and 360

minutes). Such analyses allow clinicians to select the optimal mode of injection during the

treatment planning process, so that maximum delivery to the intracellular compartment can be

achieved while ensuring minimal cardiotoxicity. In its infancy, the current model is limited to

normal brain. Nevertheless, the framework outlined in Chapter 4 is promising for further

expansion and adaptation towards brain tumor conditions.

5.2 Limitations

Despite the new findings established from this thesis work, there remain limitations inherent to

each research aim. This section will outline these identified shortcomings related to the

transducer usage, the experimental bottleneck and the simulation constraints.

5.2.1 Transducer handling

With its current design, the handling and recalibration of the ring-shaped transducer post in vivo

experiments is inconvenient and cumbersome. In particular, with the transducer-plus-coverslip

ensemble being sealed to the animal skull around the cranial window, it is necessary to pry the

coverslip-transducer set out of the skull to retrieve the system once the animal has been

sacrificed. After each experiment, the transducer, along with the damaged coverslip, is

submerged in an acetone solution to dissolve the glue and lift the transducer from the

contaminated glass. A new coverslip is later attached using superglue. The entire ensemble is

then left for to dry for several hours to ensure the complete curing of the superglue. Finally, the

combined transducer-plus-coverslip set is matched at 50 Ω impedance and 00 phase by fine-

tuning the matching circuit. This complete process of replacing the coverslip and re-matching the

transducer following every in vivo experiment is not only time-consuming, but also likely to

degrade the transducer with each use. While the transducer preparation stage is unavoidable, the

current solution is to increase the overall process throughput and efficiency by making 8 ring-

shaped transducers available. Being cut at a similar height from the same PZT cylinder, these

transducers possess almost identical dimension and generate similar frequency output. Each

transducer also has its own matching circuit. With the availability of eight spare transducers, we

could carry out the coverslip attachment and transducer matching procedure for the entire batch

in one set-up. As well, these newly-matched transducers could be used for up to eight animal

105

experiments, before we need to repeat the transducer preparation process. Another inherent

drawback of these transducers is the fragility due to its small size and the vulnerability at the

soldering sites (i.e. on the transducer electrode). In previous experiments, we encountered

instances where the transducers got damaged and the wires were detached while removing them

from the animal skull. However, this issue can be minimized by avoiding pulling on the wires

and handling the transducer removal process with extra care.

5.2.2 Delicate microsurgery of rat brain

To enable a transparent and accessible optical path from the objective lens to the cortical surface

of the animal brain, a cranial window needs to be created by removing both skull and dura. This

step needs to be successful before the transducer-coverlip system can be installed on top of the

cranial window. In early experiments, we faced numerous setbacks in failing to remove the dura

perfectly without inadvertently damaging the blood vessels in the surgical region. It is critical

that the blood vessels remain intact so that the proceeding BBBD evidence can be captured from

the exposure of FUS and MBs. Another complication with the microsurgical process that our

team encountered was brain swelling due to extended contact with air, and increased intracranial

pressure. The issue can only be resolved by speeding up the dura removal process and promptly

sealing the cranial window with the coverslip. With practice and experience, our veterinary

technicians have been able to improve and fine-tune the entire microsurgical procedure to result

in a good cranial window so that the BBBD experiment and 2PMF imaging can be achieved. Thus,

we would like to emphasize the importance and relevance of strong microsurgical skills required

for this research component. In addition, the other crucial aspects of animal care (e.g. tail vein

canalization, anesthesia stabilization, physiological condition maintenance) should be ensured for

a successful BBBD experiment. Lastly, it is noteworthy that these delicate yet critical

microsurgical steps limit the experiment throughput, in terms of the number of animals and

BBBD studies that can be conducted per day. As a result, it may require a longer time and effort

to complete enough trials to achieve statistical significance for this type of experiment.

Another issue related to the microsurgical and experimental component is the current use

of isoflurane for anesthesia. There have been in vitro evidences of the isoflurane interference on

the ability of astrocytes to support neuronal growth, as well as reduction in levels of brain-

derived neurotrophic factor. As a result, neurons cocultured with astrocytes exposed to

106

isoluorane have been shown to exhibit a 30% decrease in axon outgrowth [290]. Overall, the

potential effect of anesthetics on the neurovascular unit should be considered when interpreting

the mechanism of BBBD induced during animal experiments. As well, other anesthetics should be

explored and compared for their influences on the neurovascular unit.

5.2.3 Limitations of Current Simulation Study

As an initial paradigm that adapts relevant conditions associated with FUS+MBs mediated Dox

delivery, the mathematical model presented in Chapter 4 exhibits several limitations. Firstly, its

current setup is applicable for normal brain conditions, as opposed to tumor pathology, with the

underlying motive of keeping the simulation parameters as close to the experimental conditions

in Park et al.’s study for validation purposes. Thus, for future studies, it will be of great

importance to expand and adapt the model for brain tumors. In those circumstances, an

additional convection process driven by the outward pressure gradient of ISF needs to be

incorporated into the model [262]. Furthermore, Dox pharmacokinetic constants and tissue

properties (e.g. diffusion constant, leakiness of the BTB) must be modified accordingly for the

tumor conditions. Another limitation of the current model is that the efflux mechanisms at the

BBB are neglected. Identified as one of the major obstacles in chemotherapy and treatment of

malignant cancers, the development of multidrug resistance (MDR) is triggered by the

overexpression of P-gp [12]. P-gp is an adenosine triphosphate (ATP)-dependent transporter

which actively pumps a wide range of chemotherapeutic agents back out across the BBB. While

this phenomenon was not accounted for in the present model, we estimate that the

implementation of subsequent efflux mechanisms may ultimately reduce the intracellular Dox

concentrations. Nevertheless, in practice, efflux activities of P-gp could potentially be suppressed

by the administration of P-gp inhibitors (e.g. valspodar, elacridar, zosuquidar) along with

chemotherapeutic agents [279]. This aspect should be explored and verified in future studies.

5.3 Future directions

In addition to addressing the inherent limitations to the thesis work as presented in the preceding

section, future research should also consider the following issues.

107

5.3.1 Incorporation of passive cavitation detection

To ensure safety during clinical BBBD procedures, it is advantageous to closely monitor in real-

time the acoustic activities of MBs inside the cerebral vasculature. This concept has been

demonstrated by several research groups using passive cavitation detection technique to capture

the acoustic emissions from MBs (including harmonic, sub- and ultra-harmonic as well as

wideband signals) [250], [280], [281]. From the original investigation by McDannold et al., the

occurrence of BBBD was accompanied by the amplification of the second and third harmonic

signals. On the other hand, a study by Tung and colleagues suggested the fourth and fifth

harmonic emissions might be a more reliable indicator of in vivo BBB opening. Finally, O’Reilly et

al. demonstrated that safe opening of the BBB via FUS-treatment can be based on the detection of

ultra-harmonic signal level relative to its fundamental frequency. Using a control-feedback

algorithm, the authors showed that sonications can be applied in a controlled manner by

constantly monitoring MB emissions and increasing the power input until sub- or ultraharmonic

signals are detected and then reducing the power to a predetermined level depending on the

desired magnitude of the BBBD.

Given the importance of monitoring MB activities, it is worthwhile to incorporate passive

cavitation detection to the current 2PFM imaging procedure. Essentially, this augmentation

would provide us with additional and useful information on the BBBD signature aside from the

captured optical images of fluorescent dye leaking out of the compromised vessel. Correlating

these acoustic emission signals from MBs to the ensuing BBBD evidences (e.g. fast versus slow

leakage) could shed light on the precise physical mechanisms of FUS+MBs induced BBBD.

5.3.2 Imaging fluorescent MBs during BBBD using 2PFM

Another research avenue that is worthwhile exploring with the 2PFM imaging system is to label

the MBs with fluorophores in order to visualize them inside of blood vessels in vivo. In

collaboration with Dr. Shirley Wu’s team from Leslie Dan Faculty of Pharmacy, we had been able

to incorporate nanoparticles (NPs) onto the surface of Definity® MBs. The details in methodology

and preliminary results are laid out in Appendix A. From the pilot study, we were able to capture

several interesting evidences of MBs “traffic” inside a pre-selected vessel as well as ensuing

leakage of FITC 500kDa that was used as the marker for vasculature. Given the encouraging

108

results, future studies should aim to acquire more data in order to gain statistical significance and

substantiate these early observations.

5.3.3 Extending 2PFM-based BBBD study to other therapeutic agents

Thus far, 2PFM imaging has only been exploited to observe and measure permeability of

fluorescent dyes with different molecular weight. Such investigation for 10 kDa and 70 kDa

dextran-conjugated Texas Red leaking out of the cerebral vasculature of normal rats is discussed

in Chapter 3. In addition, we had extended our experimental study to 500 kDa dextran-

conjugated FITC. The results from this study can be found in Appendix B.

Beyond using fluorophores to gain knowledge about the size-dependence of permeability

across the disrupted BBB, one could also look into actual therapeutic agents. Previously, stem

cells and immune cells have been shown to cross the BBB upon the application of FUS+MBs [144],

[170]. In regards to delivering stem cells or natural killer cells across the BBB, their mechanisms of

crossing the BBB still remain undetermined. Hence, it would be of great benefit to closely monitor

the behavior of these cells crossing the BBB to completely understand their mechanisms. Such

knowledge will allow for optimization of the acoustic parameters to increase their population or

concentration in the extravascular compartment.

Another avenue to be explored is the study of immunotherapeutic drug BAM-10 in the

Alzheimer disease model. In a recent study, Burgess et al used two-photon microscopy to study

changes in FUS-mediated BBB permeability in transgenic (TgCRND8) mice and their non-

transgenic littermates [282]. Using regular fluorescent dye, the authors have identified leakage

from the vasculature after the application of FUS. They found that dye leakage occurred in both

transgenic and non-transgenic mice at similar acoustic pressures but exhibited different leakage

kinetics. Calculations of the permeability constant demonstrated that the vasculature in the

transgenic mice was much less permeable after FUS than the non-transgenic littermates. Further

analysis demonstrated that the change in vessel diameter following FUS was lessened in amyloid

coated vessels. This data suggests that changes in vessel diameter may be directly related to

permeability and the presence of amyloid plaque may reduce the permeability of a vessel after

FUS. Future work could build on this pilot study by employing BAM-10 in order to have better

quantitative permeability measurements of the drug itself in the vasculature of AD.

109

In addition to quantifying the permeability for other therapeutic agents using the

methodology described in Chapter 3, future studies should aim to differentiate the type of vessels

undergoing BBBD. Currently, we could only measure the vessel diameter to infer its vessel type.

Using this approach, any vessel with diameter less than 10 µm was assigned as capillary.

Meanwhile, venules range from 7 - 50 µm in diameter and arterioles range from 10 - 100 µm in

diameter. Therefore, it is imprecise to identify the vessel type based on the diameter

measurement from 2PFM images. A better method would be to directly visualize fluorescent dye

entering the microvasculature system upon the injection and distinguish arterioles from venules

based on the arrival time of the dye. Lastly, future work should also consider histological

examination on the rat brains that have been studied under the 2PFM. Similar to the

investigations by Sheikov et al [109], [110], details on microstructural changes resolved by

transmission electron microscopy can provide critical information on the mechanisms of the fast

and slow BBB disruption, when these analyses are linked to the leakage types observed from the

2PFM.

5.3.4 Extending the simulation model to other therapeutic agents

The mathematical framework presented in Chapter 4 is the first model that depicts BBBD

conditions under FUS+MBs treatment. As well, it is specifically constructed for Doxorubicin due

to the relevance of this chemotherapeutic agent in treating primary brain tumors and metastases.

In other words, all the parameters have been adapted towards this drug. However, considering

how flexible the FUS+MBs assisted BBBD technique can be used for other therapeutic agent of

different size and chemical property, the simulation study can be modified to closely describe the

drug of interest. For instance, the simulation study can be conducted for the delivery of BAM-10

in alleviating AD [158], [283], siRNA in treating HD [164], or stem cells in handling traumatic

brain injury [170]. In such cases, the model inputs have to be modified accordingly to these types

of therapeutic agents (e.g. pharmacokinetic parameters) as well as the physiological conditions of

these particular pathologies (e.g. different permeability in AD vessels [282]).

Since the property of the BBB might be different on various locations within the brain (e.g.

the ventricles versus the cortical layer), the level of BBB opening under a similar FUS+MBs

exposure might vary on each treatment region. Such effect should be considered during both

experimental permeability quantification and simulation for treatment planning. As well, since

the current model discussed in Chapter 4 relies on several pharmacokinetic and

110

pharmacodynamic parameters that had been previously reported by other groups, a sensitivity

study should be conducted to examine the impact of each parameter on the end result (i.e.

intracellular drug concentration). Once developed and validated, these simulation studies will be

beneficial for future treatment strategy of FUS+MBs induced BBBD technique for different CNS

disease conditions.

5.4 Clinical perspectives

The BBBD technique described in this thesis work is based on two components: FUS and MBs,

each of which has made significant progress towards clinical setting. First, referred as a “scalpel-

free” surgery, FUS has been employed under the guidance of MRI to treat patient with severe

essential tremor, which is the most common movement disorder. Presently, Sunnybrook Hospital

in Canada is one of the eight centers around the world that are engaging in this randomized

Phase 3 clinical trial. Between 2012 and now, the institution has transitioned from Phase 1 to

Phase 3 and the treatment outcomes from patients are positive [284]. If the FUS technique

receives the regulatory approval from Health Canada and the U.S. Food and Drug

Administration, this innovative approach will revolutionize the field of neurosurgery. This

development will benefit the progress of FUS-aided BBBD and potentially allow the technique to

be fast-tracked to a clinical practice in the near future.

However, to completely realize the clinical viability of BBBD, further development of a

real-time monitoring procedure needs to be put in place to ensure patient safety during

treatment. Beyond the temporal control methodology (mentioned in Section 5.3.1), the capability

of spatially-mapping cavitation activity within the brain is desirable and ideal for clinical usage.

Recently, by sparsely and pseudo-randomly integrating 128 piezo-ceramic receiver elements into

an existing hemispherical phase-array transducer, O’Reilly et al. demonstrated a dual-purposed

transducer assemble that offers both FUS therapy and US imaging of the MB cloud [285], [286].

Using passive beam-forming techniques, the authors were able to achieve a lateral resolution of

1.25-2 mm and axial resolution of 2-3.5 mm. Given its capability of imaging MB activity down to

single bubble events at pressure below the BBBD threshold, this 3D transcranial ultrasound

imaging technique has great implication for real-time monitoring and intervention of cavitation-

based BBBD in the brain.

111

As an equally important element of BBBD strategy, the novel design of multifunctional

MBs could further add value to clinical advancement. While the current BBBD procedure relies

on commercial MBs for its synergistic effect towards cavitation-mediated FUS treatment,

additional therapeutic potential of MBs can be realized by encapsulating or conjugating

endogenous drugs with MBs [287]–[289]. In such circumstances, aside from their use for

permeability enhancement of the BBB, MBs also serve as a vehicle that shields the unstable

therapeutic agent from exposure to systemic circulation and precisely release them at the targeted

site upon FUS-triggered MB destruction. Ongoing research is focusing on different strategies to

incorporate various therapeutic agents into MB carriers as well as increase the drug payload.

Regardless, it is exciting to witness the progression of MBs reaching beyond the conventional

diagnostic contrast agent for US imaging and offering more possibilities for BBBD-mediated drug

delivery to the brain.

112

Appendix A: Imaging nanoparticle-

incorporated microbbubles in vivo using

two-photon fluorescence microscopy

In collaboration with Dr. Shirley Wu from Leslie Dan Faculty of Pharmacy, we were able to

incorporate nanoparticles (NPs) onto the surface of Definity® MBs. NP solution was prepared by

Preethy Prasad, a graduate student from Dr. Wu’s lab. The solution was custom-made such that

NPs can be excited by the 810 nm pulse laser via the two-photon excitation process. Their

emission wavelength is within 420 - 460 nm, hence can be readily filtered and captured into the

blue channel of the 2PFM imaging system.

From a characterization study, we determined the optimal dose for injecting NP solution

into a new Definity® vial to be 0.25 ml. Once NP solution had been added, the Definity® vial was

then activated using the VialMix®. Figure A.1 shows two examples of “static” fluorescent MBs on

a glass slide as imaged by the 2PFM system.

Figure A.1: Nanoparticle-labelled Definity® microbubbles on a glass slide are imaged by the 2PFM system

After the promising in vitro tests, we moved to in vivo experiments where NPs-labelled

MBs were injected into a rat tail vein. Prior to injection, NP-MB mixture was withdrawn from the

activated Definity® vial and diluted with saline. During these initial attempts, to increase the

likelihood of capturing MBs inside the circulation, we decided to set the dilution ratio to 1:2

113

rather than 1:10 as seen in the clinical standard of normal Definity® MBs. In addition to injecting

the NPs-labelled MBs, we also used 500 kDa dextran-conjugated FITC (FITC500) as the marker

for vasculature. To catch the fast movement of MBs inside a blood vessel, the imaging FOV was

reduced to 5-40 times below the typical XY scanning area (i.e. 512x512 µm2) and the dwelling

speed was double the previous setting (i.e. 8µs/pixel). As well, we repeatedly imaged a single

slice over time (i.e. XYT) rather than operating in a stacking mode (i.e. XYZT). Using the new

setting that allowed for a temporal resolution of 200-400 ms/XY slice, we aimed to capture signal

emissions from both NPs-incorporated MBs (by the blue channel) and FITC 500kDa fluorescent

dye (by the green channel).

Overall, we performed this set of experiments on 12 animals and at total of 24 sonications.

However, only 8 data sets are considered usable for analysis. The rest of the data were discarded

due to either imperfect microsurgical procedure (hence poor-quality cranial window), drifting

imaging FOV, or failing to detect any MBs due to their absence in the selected vessel. It is noted

that this type of experiment is a randomized process in which we hope to catch these rare events

of MBs passing through a randomly pre-selected vessel. Figure A.2 – Figure A.3 display two

interesting data sets in which both MB presence in a selected vessel and extravasation of FITC500

upon BBBD were captured over a series of time-lapsed XY images. In these examples, sonication

was applied at 0.6 MPa acoustic pressure and lasted for 2 minutes. From these time-lapsed

images, we noted the emergence of the blue punctuated-pattern signal along a 50 µm vessel

segment and its prominent presence in the extravascular domain. Based on the dotted-feature, we

speculated that MBs might be shedding the NPs from their surfaces. In the second example, we

also noticed a remarkable deformation of the vessel wall, which might be caused by transcellular

passage of some blood-borne agents.

Based on the pilot study, we demonstrated that imaging these NP-labeled microbubbles in

the in vivo blood vessel is possible. However, these observations are preliminary and future

studies should aim to acquire more data to better understand MB behaviors during BBBD events.

114

Figure A.2: Second example of MB visualization and FITC500 leakage under FUS+MBs induced BBBD at 0.6 MPa. (A) A vessel (red rectangle) was randomly selected from the 512x521 µm2 imaging FOV. Scale bar: 50 µm. (B) Time-lapsed XY images of the selected vessel over the duration of 15 minutes. Sonication occurred during the first 2 minutes. At T = 3 min, a mixture of green and blue signals start emerging in the extravascular space along a 50 µm vessel edge, as indicated by the red arrow. Over the course of 3-15 minutes, the signal becomes more pronounced. However, in contrast to the first example, it is interesting to note from this example that the blue signal appears to dominate the green signal and exhibits a dotted-feature. We speculated that the MBs might be shedding the NPs coating on their surface.

Figure A.3: Third example of MB visualization and FITC500 leakage under FUS+MBs induced BBBD at 0.6 MPa. (A) A vessel (red rectangle) was randomly selected from the 512x521 µm2 imaging FOV. Scale bar: 50 µm. (B) Time-lapsed XY images of the selected vessel over the duration of 15 minutes. Sonication took place during the first 2 minutes. At T = 8 min, a mixture of green and blue signals start emerging in the extravascular space near the bifurcation point of the vessel, as indicated by the red arrow. Similar to the second example, these current images also exhibit a lower level of green signal relative to the blue one. Here, punctated extravasation is also noticed. Another remarkable feature is the deformation of the vessel wall, as outlined by the dashed oval.

115

Appendix B: Correlation between

substance size and its permeability at the

BBB

As a continuation from the successful permeability assessment of 10 kDa and 70 kDa dextran-

conjugated Texas Red via the use of 2PFM imaging, we further extended the study to other

fluorescent dyes of different molecular weights. While such an investigation for small size

substances (e.g. 3 kDa dextran-conjugated Texas Red) was infeasible due to its rapid clearance

from circulation, we were able to conduct 2PFM imaging on large size agents such as 500 kDa

dextran-conjugated FITC.

In brief, the permeability of FITC 500 kDa was evaluated a similar manner as for TR10 and

TR70. In this set of experiment, we induced BBBD at 3 pressure levels: 0.4 MPa (n=3), 0.6 MPa

(n=4) and 0.8 MPa (n=4). The permeability associated with each disruption incident was shown

as a scattered plot in Figure B.1. To compare the results of FITC500 to TR10 and TR7, their

respective average and standard deviation of permeability values at each acoustic pressure are

presented in Figure B.2. At 0.6 MPa and 0.8 MPa, it is evident that permeability is inversely

correlated with the MW.

Figure B.1: Measured permeability of dextran conjugated FITC 500kDa across the compromised BBB as induced at different acoustic pressure level of FUS exposure

0.2

0.4

0.6

0.8

0.000

0.005

0.010

0.015

Acoustic Pressure (MPa)

Perm

eab

ilit

y (

min

-1)

116

Figure B.2: Average permeability constants at each acoustic pressure level are compared among TR10kDa, TR70kDa and FITC500kDa

To further explore the relationship between MW and its permeation across the BBB, we

further include the Ktrans constant of 1 kDa Gd which was obtained from the DCE-MRI method

[196]. ). This study particularly reported the permeability for Gd-DTPA of 1 kDa molecular

weight (MW) leaking out of the vasculature when BBBD was induced using similar acoustic

pressure (e.g. 0.4 - 0.6 MPa) and similar microbubble size (e.g. 1-2 µm). All together, the

aggregated data of 1-500 kDa at 0.4 MPa and 0.6 MPa concluded from two independent imaging

methodologies (DCE-MRI and 2PFM) are summarized in Table B.1. As further shown in Figure

B.3(A), we observed a convergence of permeability toward zero when MW increases from 70 kDa

to 500 kDa. In Table B.2, the logarithm of MW and normalized permeability with respect to 1 kDa

(i.e. Ei/E1kDa, where i represents other larger MW such as 10 kDa, 70 kDa or 500 kDa) are

tabulated. As demonstrated in Figure B.3(B), normalized permeability exhibits a linear

relationship with log(MW). By applying the best fit, we observed:

Ei/E1kDa = 1 – 0.35 log(MWi)

For future study, it would be interesting to even extend towards larger size agents to verify

whether such relationship still holds.

0.2

0.4

0.6

0.8

0.00

0.01

0.02

0.03

0.0410kDa

70kDa

500kDa

Acoustic Pressure (MPa)

Avera

ge

Perm

eab

ility

C

on

sta

nt

(min

-1)

117

Table B.1: Summary of measured permeability constants of 1 kDa – 500 kDa test substance upon BBBD induced at 0.4 MPa and 0.6 MPa FUS acoustic pressure

MW (kDa)

Permeability (min-1) Reference 0.4 MPa 0.6 MPa

1 0.0105 0.0390 Vlachos et al., Mag Res Med 2011, 66: 821-830

10 0.0068 0.0193 Nhan et al., J Controlled Release 2013, 172: 274-280

70 0.0032 0.0073

500 0.0031 0.0028 Unpublished data

Table B.2: Normalized permeability versus log(MW)

MW (kDa)

log(MW)

Normalized Permeability

0.4 MPa 0.6 MPa

1 0 1 1

10 1 0.65 0.50

70 1.85 0.31 0.19

500 2.70 0.29 0.07

Figure B.3: A) Enhanced permeability as a function of molecular weight for imaging tracers of MW between 1-500 kDa. It is noted that the permeability constants were obtained from two independent imaging modalities (DCE-MRI and 2PFM) but BBBD was induced at comparable acoustic pressure and microbubble size. (B) Normalized permeability versus log(MW) displays a linear relationship

118

References

[1] WHO, “Neurological Disorders: public health challenges,” in Neurological Disorders: Public Health Challenges, 2014, pp. 26–39.

[2] H. L. Weiner and D. Frenkel, “Immunology and immunotherapy of Alzheimer’s disease.,” Nat. Rev. Immunol., vol. 6, no. 5, pp. 404–16, May 2006.

[3] M. H. Tuszynski, “Growth-factor gene therapy for neurodegenerative disorders,” Lancet Rev., vol. 1, no. May, pp. 51–57, 2002.

[4] S. T. Lim, M. Airavaara, and B. K. Harvey, “Viral vectors for neurotrophic factor delivery: a gene therapy approach for neurodegenerative diseases of the CNS.,” Pharmacol. Res., vol. 61, no. 1, pp. 14–26, Jan. 2010.

[5] G. W. Goldstein and A. L. Betz, “The blood-brain barrier.,” Sci. Am., vol. 255, no. 3, pp. 74–83, Sep. 1986.

[6] Brain Tumor Foundation Of Canada, “Canadian Facts about Brain Tumours.” .

[7] NIH, “Brain Tumor,” 2014. [Online]. Available: http://www.cancer.gov/cancertopics/types/brain.

[8] M. Wu, H. B. Frieboes, M. A. J. Chaplain, S. R. McDougall, V. Cristini, and J. S. Lowengrub, “The effect of interstitial pressure on therapeutic agent transport: coupling with the tumor blood and lymphatic vascular systems.,” J. Theor. Biol., vol. 355, pp. 194–207, Aug. 2014.

[9] C. A. Graham and T. F. Cloughesy, “Brain tumor treatment: Chemotherapy and other new developments,” in Seminars in Oncology Nursing, 2004, vol. 20, no. 4, pp. 260–272.

[10] A. Pirzkall, C. McGue, S. Saraswathy, S. Cha, R. Liu, S. Vandenberg, K. R. Lamborn, M. S. Berger, S. M. Chang, and S. J. Nelson, “Tumor regrowth between surgery and initiation of adjuvant therapy in patients with newly diagnosed glioblastoma.,” Neuro. Oncol., vol. 11, no. 6, pp. 842–52, Dec. 2009.

[11] W. H. Oldendorf, M. E. Cornford, and W. J. Brown, “The large apparent work capability of the blood-brain barrier: a study of the mitochondrial content of capillary endothelial cells in brain and other tissues of the rat.,” Ann. Neurol., vol. 1, no. 5, pp. 409–17, May 1977.

[12] W. Löscher and H. Potschka, “Blood-brain barrier active efflux transporters: ATP-binding cassette gene family.,” NeuroRx, vol. 2, no. 1, pp. 86–98, Jan. 2005.

[13] J. Fenstermacher and T. Kaye, “Drug ‘diffusion’ within the brain.,” Ann. N. Y. Acad. Sci., vol. 531, pp. 29–39, Jan. 1988.

119

[14] R. Sedlakova, R. R. Shivers, and R. F. Del Maestro, “Ultrastructure of the blood-brain barrier in the rabbit,” J Submicrosc Cytol Pathol., vol. 31, no. 1, pp. 149–61, 1999.

[15] U. Kniesel and H. Wolburg, “Tight junctions of the blood–brain barrier,” Cell. Mol. Neurobiol., vol. 20, no. 1, pp. 57–76, Feb. 2000.

[16] N. J. Abbott, L. Rönnbäck, and E. Hansson, “Astrocyte-endothelial interactions at the blood-brain barrier.,” Nat. Rev. Neurosci., vol. 7, no. 1, pp. 41–53, Jan. 2006.

[17] A. Misra, G. S., and A. Shahiwala, “Drug delivery to the central nervous system: a review.,” J Pharm Pharm. Sci, vol. 6, no. 2, pp. 252–273, 2003.

[18] W. M. Pardridge, “The blood-brain barrier: bottleneck in brain drug development.,” NeuroRx, vol. 2, no. 1, pp. 3–14, Jan. 2005.

[19] A. Tsuji and I. Tamai, “Carrier-mediated or specialized transport of drugs across the blood–brain barrier,” Adv. Drug Deliv. Rev., vol. 36, no. 2–3, pp. 277–290, Apr. 1999.

[20] H.-L. Liu, C.-H. Fan, C.-Y. Ting, and C.-K. Yeh, “Combining microbubbles and ultrasound for drug delivery to brain tumors: current progress and overview.,” Theranostics, vol. 4, no. 4, pp. 432–44, Jan. 2014.

[21] R. J. Mairs, C. L. Wideman, W. J. Angerson, T. L. Whateley, M. S. Reza, J. R. Reeves, L. M. Robertson, A. Neshasteh-Riz, R. Rampling, J. Owens, D. Allan, and D. I. Graham, “Comparison of different methods of intracerebral administration of radioiododeoxyuridine for glioma therapy using a rat model.,” Br. J. Cancer, vol. 82, no. 1, pp. 74–80, Jan. 2000.

[22] C. E. Krewson, M. L. Klarman, and W. M. Saltzman, “Distribution of nerve growth factor following direct delivery to brain interstitium.,” Brain Res., vol. 680, no. 1–2, pp. 196–206, May 1995.

[23] P. C. McGovern, E. Lautenbach, P. J. Brennan, R. A. Lustig, and N. O. Fishman, “Risk factors for postcraniotomy surgical site infection after 1,3-bis (2-chloroethyl)-1-nitrosourea (Gliadel) wafer placement.,” Clin. Infect. Dis., vol. 36, no. 6, pp. 759–65, Mar. 2003.

[24] A. Tsuji, “Small molecular drug transfer across the blood-brain barrier via carrier-mediated transport systems.,” NeuroRx, vol. 2, no. 1, pp. 54–62, Jan. 2005.

[25] A. T. Nies, “The role of membrane transporters in drug delivery to brain tumors.,” Cancer Lett., vol. 254, no. 1, pp. 11–29, Aug. 2007.

[26] L. Juillerat-Jeanneret, “The targeted delivery of cancer drugs across the blood-brain barrier: chemical modifications of drugs or drug-nanoparticles?,” Drug Discov. Today, vol. 13, no. 23–24, pp. 1099–106, Dec. 2008.

120

[27] S. Dallas, D. S. Miller, and R. Bendayan, “Multidrug Resistance-Associated Proteins: Expression and Function in the Central Nervous System,” Pharmacol. Rev., vol. 58, no. 2, pp. 140–161, 2006.

[28] D. F. Kraemer, D. Fortin, and E. A. Neuwelt, “Chemotherapeutic dose intensification for treatment of malignant brain tumors: recent developments and future directions.,” Curr. Neurol. Neurosci. Rep., vol. 2, no. 3, pp. 216–24, May 2002.

[29] K. Hynynen, N. McDannold, N. Vykhodtseva, and F. A. Jolesz, “Noninvasive MR Imaging – guided Focal Opening of the Blood-Brain Barrier in Rabbits,” Radiology, vol. 220, no. 4, pp. 640–646, 2001.

[30] A. Carovac, F. Smajlovic, and D. Junuzovic, “Application of ultrasound in medicine.,” Acta Inform. Med., vol. 19, no. 3, pp. 168–71, Sep. 2011.

[31] M. C. Fyfe and M. I. Bullock, “Therapeutic Ultrasound: Some Historical Background and Development in Knowledge of its Effect on Healing,” Aust. J. Physiother., vol. 31, no. 6, pp. 220–224, 1985.

[32] H. F. Stewart, M. H. Repacholi, and D. A. Benwell, “Ultrasound Therapy,” in Essentials of Medical Ultrasound, Ottawa, Canada: Humana Press, 1982, pp. 181–213.

[33] D. L. Miller, N. B. Smith, M. R. Bailey, G. J. Czarnota, K. Hynynen, and I. R. S. Makin, “Overview of therapeutic ultrasound applications and safety considerations.,” J. Ultrasound Med., vol. 31, no. 4, pp. 623–34, Apr. 2012.

[34] W. D. O’Brien, “Ultrasound-biophysics mechanisms.,” Prog. Biophys. Mol. Biol., vol. 93, no. 1–3, pp. 212–55, Jan. 2007.

[35] J. G. Lynn, “A new method for the generation and use of focused ultrasound in experimental biology,” J. Gen. Physiol., vol. 26, no. 2, pp. 179–193, Nov. 1942.

[36] F. F. J., “Precision high intensity focusing ultrasonic machines for surgery.,” Am. J. Phys. Med., vol. 37, no. 3, pp. 152–6, Jun. 1958.

[37] F. WJ, M. W. H., B. J. W., and F. F. J., “Production of focal destructive lesions in the central nervous system with ultrasound.,” J. Neurosurg., vol. 11, no. 5, pp. 471–8, Sep. 1954.

[38] W. J. Fry and R. Meyers, “Ultrasonic Method of Modifying Brain Structures,” Confin. Neurol, vol. 22, pp. 315–327, 1962.

[39] R. F. Heimburger, “Ultrasound augmentation of central nervous system tumor therapy.,” Indiana Med., vol. 78, no. 6, pp. 469–76, Jun. 1985.

[40] H. H. Pennes, “Analysis of Tissue and Arterial Blood Temperatures in the Resting Human Forearm,” J Appl Physiol, vol. 1, no. 2, pp. 93–122, Aug. 1948.

121

[41] W. L. Nyborg, “Heat generation by ultrasound in a relaxing medium,” J. Acoust. Soc. Am., vol. 70, no. 2, p. 310, Aug. 1981.

[42] T. J. Cavicchi, “Heat generated by ultrasound in an absorbing medium,” J. Acoust. Soc. Am., vol. 76, no. 4, p. 1244, Oct. 1984.

[43] J. A. Dickson and S. K. Calderwood, “TEMPERATURE RANGE AND SELECTIVE SENSITIVITY OF TUMORS TO HYPERTHERMIA: A CRITICAL REVIEW,” Ann. N. Y. Acad. Sci., vol. 335, no. 1 Thermal Chara, pp. 180–205, Mar. 1980.

[44] E. L. Jones, J. R. Oleson, L. R. Prosnitz, T. V Samulski, Z. Vujaskovic, D. Yu, L. L. Sanders, and M. W. Dewhirst, “Randomized trial of hyperthermia and radiation for superficial tumors.,” J. Clin. Oncol., vol. 23, no. 13, pp. 3079–85, May 2005.

[45] E. G. Moros, J. Peñagaricano, P. Novàk, W. L. Straube, and R. J. Myerson, “Present and future technology for simultaneous superficial thermoradiotherapy of breast cancer.,” Int. J. Hyperthermia, vol. 26, no. 7, pp. 699–709, Jan. 2010.

[46] S. Dromi, V. Frenkel, A. Luk, B. Traughber, M. Angstadt, M. Bur, J. Poff, J. Xie, S. K. Libutti, K. C. P. Li, and B. J. Wood, “Pulsed-high intensity focused ultrasound and low temperature-sensitive liposomes for enhanced targeted drug delivery and antitumor effect.,” Clin. Cancer Res., vol. 13, no. 9, pp. 2722–7, May 2007.

[47] M. de Smet, E. Heijman, S. Langereis, N. M. Hijnen, and H. Grüll, “Magnetic resonance imaging of high intensity focused ultrasound mediated drug delivery from temperature-sensitive liposomes: an in vivo proof-of-concept study.,” J. Control. Release, vol. 150, no. 1, pp. 102–10, Feb. 2011.

[48] R. Staruch, R. Chopra, and K. Hynynen, “Localised drug release using MRI-controlled focused ultrasound hyperthermia.,” Int. J. Hyperthermia, vol. 27, no. 2, pp. 156–71, Jan. 2011.

[49] K. Hynynen, “MRI-guided focused ultrasound treatments.,” Ultrasonics, vol. 50, no. 2, pp. 221–9, Feb. 2010.

[50] F. Wu, Z.-B. Wang, Y.-D. Cao, W.-Z. Chen, J. Bai, J.-Z. Zou, and H. Zhu, “A randomised clinical trial of high-intensity focused ultrasound ablation for the treatment of patients with localised breast cancer.,” Br. J. Cancer, vol. 89, no. 12, pp. 2227–33, Dec. 2003.

[51] F. Wu, Z.-B. Wang, H. Zhu, W.-Z. Chen, J.-Z. Zou, J. Bai, K.-Q. Li, C.-B. Jin, F.-L. Xie, and H.-B. Su, “Feasibility of US-guided high-intensity focused ultrasound treatment in patients with advanced pancreatic cancer: initial experience.,” Radiology, vol. 236, no. 3, pp. 1034–40, Sep. 2005.

[52] S. A. Sapareto and W. C. Dewey, “Thermal dose determination in cancer therapy,” Int. J. Radiat. Oncol., vol. 10, no. 6, pp. 787–800, Apr. 1984.

122

[53] M. W. Dewhirst, B. L. Viglianti, M. Lora-Michiels, M. Hanson, and P. J. Hoopes, “Basic principles of thermal dosimetry and thermal thresholds for tissue damage from hyperthermia,” Jul. 2009.

[54] H. Lyngt, O. R. Monge, P. J. Bøhler, and E. K. Rofstad, “Relationships between thermal dose and heat-induced tissue and vascular damage after thermoradiotherapy of locally advanced breast carcinoma,” Jul. 2009.

[55] W. C. Dewey, “Arrhenius relationships from the molecule and cell to the clinic.,” Int. J. Hyperthermia, vol. 25, no. 1, pp. 3–20, Feb. 2009.

[56] N. J. Mcdannold, R. L. King, F. A. Jolesz, and K. H. Hynynen, “Usefulness of MR Imaging – Derived Thermometry and Dosimetry in Determining the Threshold for Tissue Damage Induced by Thermal Surgery in Rabbits,” Radiology, vol. 216, no. 11, pp. 517–523, 2000.

[57] C. Damianou and K. Hynynen, “The effect of various physical parameters on the size and shape of necrosed tissue volume during ultrasound surgery.,” J. Acoust. Soc. Am., vol. 95, no. 3, pp. 1641–9, Mar. 1994.

[58] D. Dalecki, “Mechanical bioeffects of ultrasound.,” Annu. Rev. Biomed. Eng., vol. 6, pp. 229–48, Jan. 2004.

[59] G. ter Haar, “Therapeutic applications of ultrasound.,” Prog. Biophys. Mol. Biol., vol. 93, no. 1–3, pp. 111–29, Jan. 2007.

[60] M. L. Palmeri, A. C. Sharma, R. R. Bouchard, R. W. Nightingale, and K. R. Nightingale, “A finite-element method model of soft tissue response to impulsive acoustic radiation force.,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 52, no. 10, pp. 1699–712, Oct. 2005.

[61] K. Nightingale, M. S. Soo, R. Nightingale, and G. Trahey, “Acoustic radiation force impulse imaging: in vivo demonstration of clinical feasibility,” Ultrasound Med. Biol., vol. 28, no. 2, pp. 227–235, Feb. 2002.

[62] J. Bercoff, M. Tanter, and M. Fink, “Supersonic shear imaging: a new technique for soft tissue elasticity mapping,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 51, no. 4, pp. 396–409, Apr. 2004.

[63] K. G. Baker, V. J. Robertson, and F. A. Duck, “A review of therapeutic ultrasound: biophysical effects.,” Phys. Ther., vol. 81, no. 7, pp. 1351–8, Jul. 2001.

[64] H. W. Herr, “‘Crushing the stone’: a brief history of lithotripsy, the first minimally invasive surgery.,” BJU Int., vol. 102, no. 4, pp. 432–5, Aug. 2008.

[65] A. J. Coleman and J. E. Saunders, “A review of the physical properties and biological effects of the high amplitude acoustic fields used in extracorporeal lithotripsy,” Ultrasonics, vol. 31, no. 2, pp. 75–89, Jan. 1993.

123

[66] J. Eggers, “Sonothrombolysis for treatment of acute ischemic stroke: Current evidence and new developments,” Perspect. Med., vol. 1, no. 1–12, pp. 14–20, Sep. 2012.

[67] M. Rubiera and A. V Alexandrov, “Sonothrombolysis in the management of acute ischemic stroke.,” Am. J. Cardiovasc. Drugs, vol. 10, no. 1, pp. 5–10, Jan. 2010.

[68] Z. Fan, D. Chen, and C. X. Deng, “Improving ultrasound gene transfection efficiency by controlling ultrasound excitation of microbubbles.,” J. Control. Release, vol. 170, no. 3, pp. 401–13, Sep. 2013.

[69] Z. Fan, H. Liu, M. Mayer, and C. X. Deng, “Spatiotemporally controlled single cell sonoporation.,” Proc. Natl. Acad. Sci. U. S. A., vol. 109, no. 41, pp. 16486–91, Oct. 2012.

[70] C. X. Deng, F. Sieling, H. Pan, and J. Cui, “Ultrasound-induced cell membrane porosity.,” Ultrasound Med. Biol., vol. 30, no. 4, pp. 519–26, Apr. 2004.

[71] Y. Zhou, K. Yang, J. Cui, J. Y. Ye, and C. X. Deng, “Controlled permeation of cell membrane by single bubble acoustic cavitation.,” J. Control. Release, vol. 157, no. 1, pp. 103–11, Jan. 2012.

[72] C. X. Deng, “Targeted drug delivery across the blood-brain barrier using ultrasound technique.,” Ther. Deliv., vol. 1, no. 6, pp. 819–48, Dec. 2010.

[73] H.-L. Liu and M.-Y. Hua, “Blood-Brain Barrier Disruption with Focused Ultrasound Enhances Delivery of Chemotherapeutic Drugs for Glioblastoma Treatment,” Radiology, vol. 255, no. 2, 2010.

[74] A. Alonso, E. Reinz, B. Leuchs, J. Kleinschmidt, M. Fatar, B. Geers, I. Lentacker, M. G. Hennerici, S. C. de Smedt, and S. Meairs, “Focal Delivery of AAV2/1-transgenes Into the Rat Brain by Localized Ultrasound-induced BBB Opening.,” Mol. Ther. Nucleic Acids, vol. 2, no. February, p. e73, Jan. 2013.

[75] C. Wright, K. Hynynen, and D. Goertz, “In vitro and in vivo high-intensity focused ultrasound thrombolysis.,” Invest. Radiol., vol. 47, no. 4, pp. 217–25, Apr. 2012.

[76] K. Hynynen, “The threshold for thermally significant cavitation in dog’s thigh muscle in vivo,” Ultrasound Med. Biol., vol. 17, no. 2, pp. 157–169, Jan. 1991.

[77] L. BAKAY, “Ultrasonically Produced Changes in the Blood-Brain Barrier,” Arch. Neurol. Psychiatry, vol. 76, no. 5, p. 457, Nov. 1956.

[78] C. N. Shealy and D. Crafts, “Selective alteration of the blood-brain barrier.,” J. Neurosurg., vol. 23, no. 5, pp. 484–7, Nov. 1965.

[79] H. T. Ballantine, E. Bell, and J. Manlapaz, “Progress and problems in the neurological applications of focused ultrasound.,” J. Neurosurg., vol. 17, pp. 858–76, Sep. 1960.

124

[80] T. S. Salahuddin, B. B. Johansson, H. Kalimo, and Y. Olsson, “Structural changes in the rat brain after carotid infusions of hyperosmolar solutions: A light microscopic and immunohistochemical study,” Neuropathol. Appl. Neurobiol., vol. 14, no. 6, pp. 467–482, Dec. 1988.

[81] J. T. Patrick, M. N. Nolting, S. A. Goss, K. A. Dines, J. L. Clendenon, M. A. Rea, and R. F. Heimburger, “Ultrasound and the blood-brain barrier,” in Consensus on Hyperthermia for the 1990s, Springer, 1990, pp. 369–381.

[82] N. I. Vykhodtseva, K. Hynynen, and C. Damianou, “Histologic effects of high intensity pulsed ultrasound exposure with subharmonic emission in rabit brain in vivo,” Ultrasound Med. Biol., vol. 21, no. 7, pp. 969–979, 1995.

[83] A. Mesiwala, “High-intensity focused ultrasound selectively disrupts the blood-brain barrier in vivo,” Ultrasound Med. Biol., vol. 28, no. 3, pp. 389–400, Mar. 2002.

[84] N. McDannold, N. Vykhodtseva, F. Jolesz, and K. Hynynen, “MRI investigation of the threshold for thermally induced blood-brain barrier disruption and brain tissue damage in the rabbit brain.,” Magn. Reson. Med., vol. 51, no. 5, pp. 913–23, May 2004.

[85] S. Kaul, “Myocardial contrast echocardiography: A 25-year retrospective.,” Circulation, vol. 118, no. 3, pp. 291–308, Jul. 2008.

[86] P. J. . Frinking, A. Bouakaz, J. Kirkhorn, F. J. Ten Cate, and N. de Jong, “Ultrasound contrast imaging: current and new potential methods,” Ultrasound Med. Biol., vol. 26, no. 6, pp. 965–975, Jul. 2000.

[87] P. A. Dayton and K. W. Ferrara, “Targeted imaging using ultrasound.,” J. Magn. Reson. Imaging, vol. 16, no. 4, pp. 362–77, Oct. 2002.

[88] K. Ferrara, R. Pollard, and M. Borden, “Ultrasound microbubble contrast agents: fundamentals and application to gene and drug delivery.,” Annu. Rev. Biomed. Eng., vol. 9, pp. 415–47, Jan. 2007.

[89] N. McDannold, N. Vykhodtseva, and K. Hynynen, “Use of ultrasound pulses combined with Definity for targeted blood-brain barrier disruption: a feasibility study.,” Ultrasound Med. Biol., vol. 33, no. 4, pp. 584–90, May 2007.

[90] Medscape, “Definity, Optison (perflutren) dosing, indications, interactions, adverse effects, and more.” [Online]. Available: http://reference.medscape.com/drug/definity-optison-perflutren-343761. [Accessed: 03-Jun-2014].

[91] J. R. Arnold, T. D. Karamitsos, T. J. Pegg, J. M. Francis, R. Olszewski, N. Searle, R. Senior, S. Neubauer, H. Becher, and J. B. Selvanayagam, “Adenosine stress myocardial contrast echocardiography for the detection of coronary artery disease: a comparison with coronary angiography and cardiac magnetic resonance.,” JACC. Cardiovasc. Imaging, vol. 3, no. 9, pp. 934–43, Sep. 2010.

125

[92] J.-M. Correas, O. Hélénon, and J.-F. Moreau, “Contrast-enhanced ultrasonography of native and transplanted kidney diseases,” Eur. Radiol., vol. 9, no. S3, pp. S394–S400, Nov. 1999.

[93] E. Quaia, “Microbubble ultrasound contrast agents: an update.,” Eur. Radiol., vol. 17, no. 8, pp. 1995–2008, Aug. 2007.

[94] C. J. Harvey, M. J. Blomley, R. J. Eckersley, D. O. Cosgrove, N. Patel, R. A. Heckemann, and J. Butler-Barnes, “Hepatic malignancies: improved detection with pulse-inversion US in late phase of enhancement with SH U 508A-early experience.,” Radiology, vol. 216, no. 3, pp. 903–8, Sep. 2000.

[95] D. E. Goertz, N. de Jong, and A. F. W. van der Steen, “Attenuation and size distribution measurements of Definity and manipulated Definity populations.,” Ultrasound Med. Biol., vol. 33, no. 9, pp. 1376–88, Sep. 2007.

[96] Y. Sun, D. E. Kruse, P. A. Dayton, and K. W. Ferrara, “High-frequency dynamics of ultrasound contrast agents,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 52, no. 11, pp. 1981–1991, Jul. 2005.

[97] H.-L. Liu, C.-H. Pan, C.-Y. Ting, and M.-J. Hsiao, “Opening of the blood-brain barrier by low-frequency (28-kHz) ultrasound: a novel pinhole-assisted mechanical scanning device.,” Ultrasound Med. Biol., vol. 36, no. 2, pp. 325–35, Feb. 2010.

[98] K. F. Bing, G. P. Howles, Y. Qi, M. L. Palmeri, and K. R. Nightingale, “Blood-brain barrier (BBB) disruption using a diagnostic ultrasound scanner and Definity in Mice.,” Ultrasound Med. Biol., vol. 35, no. 8, pp. 1298–308, Aug. 2009.

[99] N. McDannold, N. Vykhodtseva, and K. Hynynen, “Blood-brain barrier disruption induced by focused ultrasound and circulating preformed microbubbles appears to be characterized by the mechanical index.,” Ultrasound Med. Biol., vol. 34, no. 5, pp. 834–40, May 2008.

[100] H.-L. Liu, H.-W. Chen, Z.-H. Kuo, and W.-C. Huang, “Design and experimental evaluations of a low-frequency hemispherical ultrasound phased-array system for transcranial blood-brain barrier disruption.,” IEEE Trans. Biomed. Eng., vol. 55, no. 10, pp. 2407–16, Oct. 2008.

[101] Z. Zhang, C. Xia, Y. Xue, and Y. Liu, “Synergistic effect of low-frequency ultrasound and low-dose bradykinin on increasing permeability of the blood-tumor barrier by opening tight junction.,” J. Neurosci. Res., vol. 87, no. 10, pp. 2282–9, Aug. 2009.

[102] H.-L. Liu, Y.-Y. Wai, W.-S. Chen, J.-C. Chen, P.-H. Hsu, X.-Y. Wu, W.-C. Huang, T.-C. Yen, and J.-J. Wang, “Hemorrhage detection during focused-ultrasound induced blood-brain-barrier opening by using susceptibility-weighted magnetic resonance imaging.,” Ultrasound Med. Biol., vol. 34, no. 4, pp. 598–606, Apr. 2008.

126

[103] J. J. Choi, M. Pernot, T. R. Brown, S. a Small, and E. E. Konofagou, “Spatio-temporal analysis of molecular delivery through the blood-brain barrier using focused ultrasound.,” Phys. Med. Biol., vol. 52, no. 18, pp. 5509–30, Sep. 2007.

[104] G. P. Howles, K. F. Bing, Y. Qi, S. J. Rosenzweig, K. R. Nightingale, and G. A. Johnson, “Contrast-enhanced in vivo magnetic resonance microscopy of the mouse brain enabled by noninvasive opening of the blood-brain barrier with ultrasound.,” Magn. Reson. Med., vol. 64, no. 4, pp. 995–1004, Oct. 2010.

[105] N. McDannold, N. Vykhodtseva, and K. Hynynen, “Effects of acoustic parameters and ultrasound contrast agent dose on focused-ultrasound induced blood-brain barrier disruption.,” Ultrasound Med. Biol., vol. 34, no. 6, pp. 930–7, Jun. 2008.

[106] J. J. Choi, K. Selert, Z. Gao, G. Samiotaki, B. Baseri, and E. E. Konofagou, “Noninvasive and localized blood-brain barrier disruption using focused ultrasound can be achieved at short pulse lengths and low pulse repetition frequencies.,” J. Cereb. blood flow Metab., vol. 31, no. 2, pp. 725–37, Feb. 2011.

[107] D. E. Goertz, C. Wright, and K. Hynynen, “Contrast agent kinetics in the rabbit brain during exposure to therapeutic ultrasound.,” Ultrasound Med. Biol., vol. 36, no. 6, pp. 916–24, Jun. 2010.

[108] R. Chopra, N. Vykhodtseva, and K. Hynynen, “Influence of exposure time and pressure amplitude on blood-brain-barrier opening using transcranial ultrasound exposures.,” ACS Chem. Neurosci., vol. 1, no. 5, pp. 391–398, May 2010.

[109] N. Sheikov, N. McDannold, N. Vykhodtseva, F. Jolesz, and K. Hynynen, “Cellular mechanisms of the blood-brain barrier opening induced by ultrasound in presence of microbubbles.,” Ultrasound Med. Biol., vol. 30, no. 7, pp. 979–89, Jul. 2004.

[110] N. Sheikov, N. McDannold, F. Jolesz, Y.-Z. Zhang, K. Tam, and K. Hynynen, “Brain arterioles show more active vesicular transport of blood-borne tracer molecules than capillaries and venules after focused ultrasound-evoked opening of the blood-brain barrier.,” Ultrasound Med. Biol., vol. 32, no. 9, pp. 1399–409, Sep. 2006.

[111] B. D. M. Meijering, L. J. M. Juffermans, A. van Wamel, R. H. Henning, I. S. Zuhorn, M. Emmer, A. M. G. Versteilen, W. J. Paulus, W. H. van Gilst, K. Kooiman, N. de Jong, R. J. P. Musters, L. E. Deelman, and O. Kamp, “Ultrasound and microbubble-targeted delivery of macromolecules is regulated by induction of endocytosis and pore formation.,” Circ. Res., vol. 104, no. 5, pp. 679–87, Mar. 2009.

[112] J. Deng, Q. Huang, F. Wang, Y. Liu, Z. Wang, Z. Wang, Q. Zhang, B. Lei, and Y. Cheng, “The role of caveolin-1 in blood-brain barrier disruption induced by focused ultrasound combined with microbubbles.,” J. Mol. Neurosci., vol. 46, no. 3, pp. 677–87, Mar. 2012.

127

[113] B. Paula, V. B. Valero-Lapchik, E. J. Paredes-gamero, and S. W. Han, “Therapeutic ultrasound promotes plasmid DNA uptake by clathrin-mediated endocytosis,” J. Gene Med., no. June, pp. 392–401, 2011.

[114] A. van Wamel, K. Kooiman, M. Harteveld, M. Emmer, F. J. ten Cate, M. Versluis, and N. de Jong, “Vibrating microbubbles poking individual cells: drug transfer into cells via sonoporation.,” J. Control. release, vol. 112, no. 2, pp. 149–55, May 2006.

[115] N. Sheikov, N. McDannold, S. Sharma, and K. Hynynen, “Effect of focused ultrasound applied with an ultrasound contrast agent on the tight junctional integrity of the brain microvascular endothelium.,” Ultrasound Med. Biol., vol. 34, no. 7, pp. 1093–104, Jul. 2008.

[116] Z. Zhang, Y. Xue, Y. Liu, and X. Shang, “Additive effect of low-frequency ultrasound and endothelial monocyte-activating polypeptide II on blood-tumor barrier in rats with brain glioma.,” Neurosci. Lett., vol. 481, no. 1, pp. 21–5, Aug. 2010.

[117] S. Jalali, Y. Huang, D. J. Dumont, and K. Hynynen, “Focused ultrasound-mediated bbb disruption is associated with an increase in activation of AKT: experimental study in rats.,” BMC Neurol., vol. 10, no. 1, p. 114, Jan. 2010.

[118] A. Alonso, E. Reinz, J. W. Jenne, M. Fatar, H. Schmidt-Glenewinkel, M. G. Hennerici, and S. Meairs, “Reorganization of gap junctions after focused ultrasound blood-brain barrier opening in the rat brain.,” J. Cereb. blood flow Metab., vol. 30, no. 7, pp. 1394–402, Jul. 2010.

[119] F. E. Borgnis, “Acoustic Radiation Pressure of Plane-Compressional Waves at Oblique Incidence,” J. Acoust. Soc. Am., vol. 24, no. 5, p. 468, Jun. 1952.

[120] V. Sboros, “Response of contrast agents to ultrasound.,” Adv. Drug Deliv. Rev., vol. 60, no. 10, pp. 1117–36, Jul. 2008.

[121] B. Krasovitski and E. Kimmel, “Shear stress induced by a gas bubble pulsating in an ultrasonic field near a wall.,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 51, no. 8, pp. 973–9, Aug. 2004.

[122] C. F. Caskey, S. M. Stieger, S. Qin, P. A. Dayton, and K. W. Ferrara, “Direct observations of ultrasound microbubble contrast agent interaction with the microvessel wall.,” J. Acoust. Soc. Am., vol. 122, no. 2, pp. 1191–200, Aug. 2007.

[123] E. A. Neppiras, “Acoustic cavitation,” Phys. Rep., vol. 61, no. 3, pp. 159–251, May 1980.

[124] G. A. Husseini, M. A. Diaz de la Rosa, E. S. Richardson, D. A. Christensen, and W. G. Pitt, “The role of cavitation in acoustically activated drug delivery.,” J. Control. Release, vol. 107, no. 2, pp. 253–61, Oct. 2005.

[125] G. Basta, L. Venneri, G. Lazzerini, E. Pasanisi, M. Pianelli, N. Vesentini, S. Delturco, C. Kusmic, and E. Picano, “In vitro modulation of intracellular oxidative stress of endothelial

128

cells by diagnostic cardiac ultrasound,” Cardiovasc. Res., vol. 58, no. 1, pp. 156–161, Apr. 2003.

[126] S. B. Raymond, J. Skoch, B. J. Bacskai, and K. Hynynen, “Modular Design for In Vivo Optical Imaging and Ultrasound Treatment in the Murine Brain,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 54, no. 2, pp. 431–434, 2007.

[127] E. E. Cho, J. Drazic, M. Ganguly, B. Stefanovic, and K. Hynynen, “Two-photon fluorescence microscopy study of cerebrovascular dynamics in ultrasound-induced blood-brain barrier opening.,” J. Cereb. blood flow Metab., pp. 1–11, Apr. 2011.

[128] S. B. Raymond, J. Skoch, K. Hynynen, and B. J. Bacskai, “Multiphoton imaging of ultrasound/Optison mediated cerebrovascular effects in vivo.,” J. Cereb. blood flow Metab., vol. 27, no. 2, pp. 393–403, Feb. 2007.

[129] J. J. Choi, M. Pernot, S. A. Small, and E. E. Konofagou, “Noninvasive, transcranial and localized opening of the blood-brain barrier using focused ultrasound in mice.,” Ultrasound Med. Biol., vol. 33, no. 1, pp. 95–104, Jan. 2007.

[130] M. Kinoshita, N. McDannold, F. Jolesz, and K. Hynynen, “Targeted delivery of antibodies through the blood-brain barrier by MRI-guided focused ultrasound.,” Biochem. Biophys. Res. Commun., vol. 340, no. 4, pp. 1085–90, Feb. 2006.

[131] J. Mei, Y. Cheng, Y. Song, Y. Yang, F. Wang, Y. Liu, and Z. Wang, “Experimental study on targeted methotrexate delivery to the rabbit brain via magnetic resonance imaging-guided focused ultrasound.,” J. Ultrasound Med., vol. 28, no. 7, pp. 871–80, Jul. 2009.

[132] J. J. Choi, S. Wang, Y.-S. Tung, B. Morrison, and E. E. Konofagou, “Molecules of various pharmacologically-relevant sizes can cross the ultrasound-induced blood-brain barrier opening in vivo.,” Ultrasound Med. Biol., vol. 36, no. 1, pp. 58–67, Jan. 2010.

[133] L. H. Treat, N. McDannold, N. Vykhodtseva, Y. Zhang, K. Tam, and K. Hynynen, “Targeted delivery of doxorubicin to the rat brain at therapeutic levels using MRI-guided focused ultrasound.,” Int. J. cancer, vol. 121, no. 4, pp. 901–7, Aug. 2007.

[134] L. H. Treat, N. McDannold, Y. Zhang, N. Vykhodtseva, and K. Hynynen, “Improved anti-tumor effect of liposomal doxorubicin after targeted blood-brain barrier disruption by MRI-guided focused ultrasound in rat glioma.,” Ultrasound Med. Biol., vol. 38, no. 10, pp. 1716–25, Oct. 2012.

[135] M. Aryal, N. Vykhodtseva, Y.-Z. Zhang, J. Park, and N. McDannold, “Multiple treatments with liposomal doxorubicin and ultrasound-induced disruption of blood-tumor and blood-brain barriers improve outcomes in a rat glioma model.,” J. Control. release, vol. 169, no. 1–2, pp. 103–11, Jul. 2013.

[136] M. Deutsch, S. B. Green, T. A. Strike, P. C. Burger, J. T. Robertson, R. G. Selker, W. R. Shapiro, J. Mealey, J. Ransohoff, P. Paoletti, K. R. Smith, G. L. Odom, W. E. Hunt, B. Young,

129

E. Alexander, M. D. Walker, and D. A. Pistenmaa, “Results of a randomized trial comparing BCNU plus radiotherapy, streptozotocin plus radiotherapy, BCNU plus hyperfractionated radiotherapy, and BCNU following misonidazole plus radiotherapy in the postoperative treatment of malignant glioma,” Int. J. Radiat. Oncol., vol. 16, no. 6, pp. 1389–1396, Jun. 1989.

[137] Michael D. Walker, J. Eben Alexander, William E. Hunt, Collin S. MacCarty, J. M. Stephen Mahaley, J. John Mealey, Horace A. Norrell, Guy Owens, Joseph Ransohoff, Charles B. Wilson, Edmund A. Gehan, and Thomas A. Strike, “Evaluation of BCNU and/or radiotherapy in the treatment of anaplastic gliomas,” May 2009.

[138] R. Stupp, W. P. Mason, M. J. van den Bent, M. Weller, B. Fisher, M. J. B. Taphoorn, K. Belanger, A. A. Brandes, C. Marosi, U. Bogdahn, J. Curschmann, R. C. Janzer, S. K. Ludwin, T. Gorlia, A. Allgeier, D. Lacombe, J. G. Cairncross, E. Eisenhauer, and R. O. Mirimanoff, “Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma.,” N. Engl. J. Med., vol. 352, no. 10, pp. 987–96, Mar. 2005.

[139] K.-C. Wei, P.-C. Chu, H.-Y. J. Wang, C.-Y. Huang, P.-Y. Chen, H.-C. Tsai, Y.-J. Lu, P.-Y. Lee, I.-C. Tseng, L.-Y. Feng, P.-W. Hsu, T.-C. Yen, and H.-L. Liu, “Focused ultrasound-induced blood-brain barrier opening to enhance temozolomide delivery for glioblastoma treatment: a preclinical study.,” PLoS One, vol. 8, no. 3, p. e58995, Jan. 2013.

[140] W. M. Pardridge, “Drug and Gene Delivery to the BrainThe Vascular Route,” Neuron, vol. 36, no. 4, pp. 555–558, Nov. 2002.

[141] R. Bell, “What can we learn from Herceptin trials in metastatic breast cancer?,” Oncology, vol. 63 Suppl 1, no. Suppl. 1, pp. 39–46, Jan. 2002.

[142] M. Kinoshita, N. McDannold, F. A. Jolesz, and K. Hynynen, “Noninvasive localized delivery of Herceptin to the mouse brain by MRI-guided focused ultrasound-induced blood-brain barrier disruption.,” Proc. Natl. Acad. Sci. U. S. A., vol. 103, no. 31, pp. 11719–23, Aug. 2006.

[143] M. J. Smyth, Y. Hayakawa, K. Takeda, and H. Yagita, “New aspects of natural-killer-cell surveillance and therapy of cancer.,” Nat. Rev. Cancer, vol. 2, no. 11, pp. 850–61, Nov. 2002.

[144] R. Alkins, A. Burgess, M. Ganguly, G. Francia, R. Kerbel, W. S. Wels, and K. Hynynen, “Focused ultrasound delivers targeted immune cells to metastatic brain tumors.,” Cancer Res., vol. 73, no. 6, pp. 1892–9, Mar. 2013.

[145] E. E. Connor, J. Mwamuka, A. Gole, C. J. Murphy, and M. D. Wyatt, “Gold nanoparticles are taken up by human cells but do not cause acute cytotoxicity.,” Small, vol. 1, no. 3, pp. 325–7, Mar. 2005.

[146] K. B. Male, B. Lachance, S. Hrapovic, G. Sunahara, and J. H. T. Luong, “Assessment of cytotoxicity of quantum dots and gold nanoparticles using cell-based impedance spectroscopy.,” Anal. Chem., vol. 80, no. 14, pp. 5487–93, Jul. 2008.

130

[147] E. S. Glazer, C. Zhu, A. N. Hamir, A. Borne, C. S. Thompson, and S. A. Curley, “Biodistribution and acute toxicity of naked gold nanoparticles in a rabbit hepatic tumor model.,” Nanotoxicology, vol. 5, no. 4, pp. 459–68, Dec. 2011.

[148] C. H. J. Choi, C. A. Alabi, P. Webster, and M. E. Davis, “Mechanism of active targeting in solid tumors with transferrin-containing gold nanoparticles.,” Proc. Natl. Acad. Sci. U. S. A., vol. 107, no. 3, pp. 1235–40, Jan. 2010.

[149] J. F. Hainfeld, D. N. Slatkin, and H. M. Smilowitz, “The use of gold nanoparticles to enhance radiotherapy in mice,” Phys. Med. Biol., vol. 49, no. 18, pp. N309–N315, Sep. 2004.

[150] H. Wang, T. B. Huff, D. A. Zweifel, W. He, P. S. Low, A. Wei, and J.-X. Cheng, “In vitro and in vivo two-photon luminescence imaging of single gold nanorods.,” Proc. Natl. Acad. Sci. U. S. A., vol. 102, no. 44, pp. 15752–6, Nov. 2005.

[151] A. Wijaya, S. B. Schaffer, I. G. Pallares, and K. Hamad-Schifferli, “Selective release of multiple DNA oligonucleotides from gold nanorods.,” ACS Nano, vol. 3, no. 1, pp. 80–6, Jan. 2009.

[152] L. R. Hirsch, R. J. Stafford, J. A. Bankson, S. R. Sershen, B. Rivera, R. E. Price, J. D. Hazle, N. J. Halas, and J. L. West, “Nanoshell-mediated near-infrared thermal therapy of tumors under magnetic resonance guidance.,” Proc. Natl. Acad. Sci. U. S. A., vol. 100, no. 23, pp. 13549–54, Nov. 2003.

[153] J. A. Schwartz, A. M. Shetty, R. E. Price, R. J. Stafford, J. C. Wang, R. K. Uthamanthil, K. Pham, R. J. McNichols, C. L. Coleman, and J. D. Payne, “Feasibility study of particle-assisted laser ablation of brain tumors in orthotopic canine model.,” Cancer Res., vol. 69, no. 4, pp. 1659–67, Feb. 2009.

[154] E. S. Day, P. A. Thompson, L. Zhang, N. A. Lewinski, N. Ahmed, R. A. Drezek, S. M. Blaney, and J. L. West, “Nanoshell-mediated photothermal therapy improves survival in a murine glioma model.,” J. Neurooncol., vol. 104, no. 1, pp. 55–63, Aug. 2011.

[155] A. B. Etame, R. J. Diaz, M. A. O’Reilly, C. A. Smith, T. G. Mainprize, K. Hynynen, and J. T. Rutka, “Enhanced delivery of gold nanoparticles with therapeutic potential into the brain using MRI-guided focused ultrasound.,” Nanomedicine, vol. 8, no. 7, pp. 1133–42, Oct. 2012.

[156] C. X. Deng and X. Huang, “Improved outcome of targeted delivery of chemotherapy drugs to the brain using a combined strategy of ultrasound, magnetic targeting and drug-loaded nanoparticles,” Ther. Deliv., vol. 2, no. 2, pp. 137–141, Feb. 2011.

[157] R. D. Alkins, P. M. Brodersen, R. N. S. Sodhi, and K. Hynynen, “Enhancing drug delivery for boron neutron capture therapy of brain tumors with focused ultrasound.,” Neuro. Oncol., vol. 15, no. 9, pp. 1225–35, Sep. 2013.

131

[158] S. B. Raymond, L. H. Treat, J. D. Dewey, N. J. McDannold, K. Hynynen, and B. J. Bacskai, “Ultrasound enhanced delivery of molecular imaging and therapeutic agents in Alzheimer’s disease mouse models.,” PLoS One, vol. 3, no. 5, p. e2175, Jan. 2008.

[159] J. F. Jordão, C. A. Ayala-Grosso, K. Markham, Y. Huang, R. Chopra, J. McLaurin, K. Hynynen, and I. Aubert, “Antibodies targeted to the brain with image-guided focused ultrasound reduces amyloid-beta plaque load in the TgCRND8 mouse model of Alzheimer’s disease.,” PLoS One, vol. 5, no. 5, p. e10549, Jan. 2010.

[160] M. E. MacDonald, C. M. Ambrose, M. P. Duyao, R. H. Myer, C. Lin, L. Srinidhi, G. Barnes, S. A. Taylor, M. James, N. Groot, H. MacFarlane, B. Jenkins, M. A. Anderson, N. S. Wexler, and J. F. Gusella, “A novel gene containing a trinucleotide repeat that is expanded and unstable on Huntington’s disease chromosomes,” Cell, vol. 72, no. 6, pp. 971–983, Mar. 1993.

[161] C. Zuccato, M. Valenza, and E. Cattaneo, “Molecular mechanisms and potential therapeutical targets in Huntington’s disease.,” Physiol. Rev., vol. 90, no. 3, pp. 905–81, Jul. 2010.

[162] S. Q. Harper, P. D. Staber, X. He, S. L. Eliason, I. H. Martins, Q. Mao, L. Yang, R. M. Kotin, H. L. Paulson, and B. L. Davidson, “RNA interference improves motor and neuropathological abnormalities in a Huntington’s disease mouse model.,” Proc. Natl. Acad. Sci. U. S. A., vol. 102, no. 16, pp. 5820–5, Apr. 2005.

[163] M. DiFiglia, M. Sena-Esteves, K. Chase, E. Sapp, E. Pfister, M. Sass, J. Yoder, P. Reeves, R. K. Pandey, K. G. Rajeev, M. Manoharan, D. W. Y. Sah, P. D. Zamore, and N. Aronin, “Therapeutic silencing of mutant huntingtin with siRNA attenuates striatal and cortical neuropathology and behavioral deficits.,” Proc. Natl. Acad. Sci. U. S. A., vol. 104, no. 43, pp. 17204–9, Oct. 2007.

[164] A. Burgess, Y. Huang, W. Querbes, D. W. Sah, and K. Hynynen, “Focused ultrasound for targeted delivery of siRNA and efficient knockdown of Htt expression.,” J. Control. release, vol. 163, no. 2, pp. 125–9, Oct. 2012.

[165] O. Lindvall, G. Sawle, H. Widner, J. C. Rothwell, A. Björklund, D. Brooks, P. Brundin, R. Frackowiak, C. D. Marsden, and P. Odin, “Evidence for long-term survival and function of dopaminergic grafts in progressive Parkinson’s disease.,” Ann. Neurol., vol. 35, no. 2, pp. 172–80, Feb. 1994.

[166] O. Lindvall, P. Brundin, H. Widner, S. Rehncrona, B. Gustavii, R. Frackowiak, K. Leenders, G. Sawle, J. Rothwell, C. Marsden, and al. et, “Grafts of fetal dopamine neurons survive and improve motor function in Parkinson’s disease,” Science (80-. )., vol. 247, no. 4942, pp. 574–577, Feb. 1990.

[167] G. K. Wenning, P. Odin, P. Morrish, S. Rehncrona, H. Widner, P. Brundin, J. C. Rothwell, R. Brown, B. Gustavii, P. Hagell, M. Jahanshahi, G. Sawle, A. Björklund, D. J. Brooks, C. D. Marsden, N. P. Quinn, and O. Lindvall, “Short- and long-term survival and function of

132

unilateral intrastriatal dopaminergic grafts in Parkinson’s disease.,” Ann. Neurol., vol. 42, no. 1, pp. 95–107, Jul. 1997.

[168] S. Gögel, M. Gubernator, and S. L. Minger, “Progress and prospects: stem cells and neurological diseases.,” Gene Ther., vol. 18, no. 1, pp. 1–6, Jan. 2011.

[169] B. Reynolds and S. Weiss, “Generation of neurons and astrocytes from isolated cells of the adult mammalian central nervous system,” Science (80-. )., vol. 255, no. 5052, pp. 1707–1710, Mar. 1992.

[170] A. Burgess, C. A. Ayala-Grosso, M. Ganguly, J. F. Jordão, I. Aubert, and K. Hynynen, “Targeted delivery of neural stem cells to the brain using MRI-guided focused ultrasound to disrupt the blood-brain barrier.,” PLoS One, vol. 6, no. 11, p. e27877, Jan. 2011.

[171] X. Shang, P. Wang, Y. Liu, Z. Zhang, and Y. Xue, “Mechanism of low-frequency ultrasound in opening blood-tumor barrier by tight junction.,” J. Mol. Neurosci., vol. 43, no. 3, pp. 364–9, Mar. 2011.

[172] F. Wang, Y. Cheng, J. Mei, Y. Song, Y. Yang, Y. Liu, and Z. Wang, “Focused Ultrasound Microbubble Destruction-Mediated Changes in Blood-Brain Barrier Permeability Assessed by Contrast-Enhanced Magnetic Resonance Imaging,” J. Utrasound Med., pp. 1501–1509, 2009.

[173] B. Marty, B. Larrat, M. Van Landeghem, C. Robic, P. Robert, M. Port, D. Le Bihan, M. Pernot, M. Tanter, F. Lethimonnier, and S. Mériaux, “Dynamic study of blood-brain barrier closure after its disruption using ultrasound: a quantitative analysis.,” J. Cereb. blood flow Metab., vol. 32, no. 10, pp. 1948–58, Oct. 2012.

[174] G. Samiotaki, F. Vlachos, Y.-S. Tung, and E. E. Konofagou, “A quantitative pressure and microbubble-size dependence study of focused ultrasound-induced blood-brain barrier opening reversibility in vivo using MRI.,” Magn. Reson. Med., vol. 000, pp. 1–9, Aug. 2011.

[175] N. McDannold, N. Vykhodtseva, S. Raymond, F. A. Jolesz, and K. Hynynen, “MRI-guided targeted blood-brain barrier disruption with focused ultrasound: histological findings in rabbits.,” Ultrasound Med. Biol., vol. 31, no. 11, pp. 1527–37, Dec. 2005.

[176] K. Hynynen, N. McDannold, N. Vykhodtseva, S. Raymond, R. Weissleder, F. A. Jolesz, and N. Sheikov, “Focal disruption of the blood-brain barrier due to 260-kHz ultrasound bursts: a method for molecular imaging and targeted drug delivery.,” J. Neurosurg., vol. 105, no. 3, pp. 445–54, Sep. 2006.

[177] H.-L. Liu, Y.-Y. Wai, P.-H. Hsu, L.-A. Lyu, J.-S. Wu, C.-R. Shen, J.-C. Chen, T.-C. Yen, and J.-J. Wang, “In vivo assessment of macrophage CNS infiltration during disruption of the blood-brain barrier with focused ultrasound: a magnetic resonance imaging study.,” J. Cereb. Blood Flow Metab., vol. 30, no. 1, pp. 177–86, Jan. 2010.

133

[178] A. Alonso, E. Reinz, M. Fatar, M. G. Hennerici, and S. Meairs, “Clearance of albumin following ultrasound-induced blood-brain barrier opening is mediated by glial but not neuronal cells.,” Brain Res., vol. 1411, pp. 9–16, Jul. 2011.

[179] N. McDannold, C. D. Arvanitis, N. Vykhodtseva, and M. S. Livingstone, “Temporary disruption of the blood-brain barrier by use of ultrasound and microbubbles: safety and efficacy evaluation in rhesus macaques.,” Cancer Res., vol. 72, no. 14, pp. 3652–63, Jul. 2012.

[180] A. Burgess, S. Dubey, S. Yeung, O. Hough, N. Eterman, I. Aubert, and K. Hynynen, “Alzheimer Disease in a Mouse Model: MR Imaging-guided Focused Ultrasound Targeted to the Hippocampus Opens the Blood-Brain Barrier and Improves Pathologic Abnormalities and Behavior.,” Radiology, p. 140245, Sep. 2014.

[181] P. J. White, G. T. Clement, and K. Hynynen, “Local frequency dependence in transcranial ultrasound transmission.,” Phys. Med. Biol., vol. 51, no. 9, pp. 2293–305, May 2006.

[182] S. Pichardo, V. W. Sin, and K. Hynynen, “Multi-frequency characterization of the speed of sound and attenuation coefficient for longitudinal transmission of freshly excised human skulls.,” Phys. Med. Biol., vol. 56, no. 1, pp. 219–50, Jan. 2011.

[183] K. Hynynen and F. A. Jolesz, “Demonstration of Potential Noninvasive Ultrasound Brain Therapy Through an Intact Skull,” Ultrasound Med. Biol., vol. 24, no. 2, pp. 275–283, Feb. 1998.

[184] a N. Guthkelch, L. P. Carter, J. R. Cassady, K. H. Hynynen, R. P. Iacono, P. C. Johnson, E. a Obbens, R. B. Roemer, J. F. Seeger, and D. S. Shimm, “Treatment of malignant brain tumors with focused ultrasound hyperthermia and radiation: results of a phase I trial.,” J. Neurooncol., vol. 10, no. 3, pp. 271–84, Jun. 1991.

[185] G. T. Clement, P. J. White, and K. Hynynen, “Enhanced ultrasound transmission through the human skull using shear mode conversion,” J. Acoust. Soc. Am., vol. 115, no. 3, p. 1356, 2004.

[186] K. J. Parker, R. M. Lerner, and R. C. Waag, “Attenuation of ultrasound: magnitude and frequency dependence for tissue characterization.,” Radiology, vol. 153, no. 3, pp. 785–8, Dec. 1984.

[187] G. T. Clement and K. Hynynen, “A non-invasive method for focusing ultrasound through the human skull.,” Phys. Med. Biol., vol. 47, no. 8, pp. 1219–36, Apr. 2002.

[188] J.-F. Aubry, M. Tanter, M. Pernot, J.-L. Thomas, and M. Fink, “Experimental demonstration of noninvasive transskull adaptive focusing based on prior computed tomography scans,” J. Acoust. Soc. Am., vol. 113, no. 1, p. 84, Jan. 2003.

[189] A. Y. Ammi, T. D. Mast, I.-H. Huang, T. A. Abruzzo, C.-C. Coussios, G. J. Shaw, and C. K. Holland, “Characterization of ultrasound propagation through ex-vivo human temporal bone.,” Ultrasound Med. Biol., vol. 34, no. 10, pp. 1578–89, Oct. 2008.

134

[190] F. J. Fry, S. A. Goss, and J. T. Patrick, “Transkull focal lesions in cat brain produced by ultrasound.,” J. Neurosurg., vol. 54, no. 5, pp. 659–63, May 1981.

[191] S. C. Tang and G. T. Clement, “Standing-wave suppression for transcranial ultrasound by random modulation.,” IEEE Trans. Biomed. Eng., vol. 57, no. 1, pp. 203–5, Jan. 2010.

[192] R. Alkins, Y. Huang, D. Pajek, and K. Hynynen, “Cavitation-based third ventriculostomy using MRI-guided focused ultrasound.,” J. Neurosurg., vol. 119, no. 6, pp. 1520–9, Dec. 2013.

[193] K. Hynynen, “The threshold for brain damage in rabbits induced by bursts of ultrasound in the presence of an ultrasound contrast agent (Optison®),” Ultrasound Med. Biol., vol. 29, no. 3, pp. 473–481, Mar. 2003.

[194] R. Chopra, L. Curiel, R. Staruch, L. Morrison, and K. Hynynen, “An MRI-compatible system for focused ultrasound experiments in small animal models,” Med. Phys., vol. 36, no. 5, p. 1867, 2009.

[195] J. Park, Y. Zhang, N. Vykhodtseva, F. A. Jolesz, and N. J. McDannold, “The kinetics of blood brain barrier permeability and targeted doxorubicin delivery into brain induced by focused ultrasound.,” J. Control. release, vol. 162, no. 1, pp. 134–42, Aug. 2012.

[196] F. Vlachos, Y.-S. Tung, and E. Konofagou, “Permeability dependence study of the focused ultrasound-induced blood-brain barrier opening at distinct pressures and microbubble diameters using DCE-MRI.,” Magn. Reson. Med., vol. 830, pp. 821–830, Apr. 2011.

[197] K.-J. Lin, H.-L. Liu, P.-H. Hsu, Y.-H. Chung, W.-C. Huang, J.-C. Chen, S.-P. Wey, T.-C. Yen, and I.-T. Hsiao, “Quantitative micro-SPECT/CT for detecting focused ultrasound-induced blood-brain barrier opening in the rat.,” Nucl. Med. Biol., vol. 36, no. 7, pp. 853–61, Oct. 2009.

[198] F.-Y. Yang, H.-E. Wang, G.-L. Lin, M.-C. Teng, H.-H. Lin, T.-T. Wong, and R.-S. Liu, “Micro-SPECT/CT-based pharmacokinetic analysis of 99mTc-diethylenetriaminepentaacetic acid in rats with blood-brain barrier disruption induced by focused ultrasound.,” J. Nucl. Med., vol. 52, no. 3, pp. 478–84, Mar. 2011.

[199] M. Lorberboym, Y. Lampl, and M. Sadeh, “Correlation of 99mTc-DTPA SPECT of the blood-brain barrier with neurologic outcome after acute stroke.,” J. Nucl. Med., pp. 1898–904, 2003.

[200] M. Oheim, D. J. Michael, M. Geisbauer, D. Madsen, and R. H. Chow, “Principles of two-photon excitation fluorescence microscopy and other nonlinear imaging approaches.,” Adv. Drug Deliv. Rev., vol. 58, no. 7, pp. 788–808, Sep. 2006.

[201] G. Alexandrakis, E. B. Brown, R. T. Tong, T. D. McKee, R. B. Campbell, Y. Boucher, and R. K. Jain, “Two-photon fluorescence correlation microscopy reveals the two-phase nature of transport in tumors.,” Nat. Med., vol. 10, no. 2, pp. 203–7, Feb. 2004.

135

[202] S. L. Ashworth, R. M. Sandoval, G. A. Tanner, and B. A. Molitoris, “Two-photon microscopy: visualization of kidney dynamics.,” Kidney Int., vol. 72, no. 4, pp. 416–21, Aug. 2007.

[203] M. M. Kamocka, J. Mu, X. Liu, N. Chen, A. Zollman, B. Sturonas-Brown, K. Dunn, Z. Xu, D. Z. Chen, M. S. Alber, and E. D. Rosen, “Two-photon intravital imaging of thrombus development.,” J. Biomed. Opt., vol. 15, no. 1, p. 016020, 2011.

[204] A. Arnau, Piezoelectric Transducers and Applications. Springer Publishing. Berlin, 2008.

[205] D. F. R. M. G. Harwood, P. Popper, “Curie Point of Barium Titanate,” Nature, vol. 160, pp. 58–59, 1947.

[206] “Morgan Advanced Materials.” [Online]. Available: http://www.morganelectroceramics.com/resources/piezo-ceramic-tutorials/typical-properties/.

[207] Murata Manufacturing Co., “Characteristics of Piezoelectric Ceramics,” 2008.

[208] “Echocardiographer.org.” [Online]. Available: http://echocardiographer.org/Echo Physics/BasicTransducers.html.

[209] K. Hynynen, N. McDannold, N. A. Sheikov, F. A. Jolesz, and N. Vykhodtseva, “Local and reversible blood-brain barrier disruption by noninvasive focused ultrasound at frequencies suitable for trans-skull sonications.,” Neuroimage, vol. 24, no. 1, pp. 12–20, Jan. 2005.

[210] F.-Y. Yang, S.-C. Horng, Y.-S. Lin, and Y.-H. Kao, “Association between contrast-enhanced MR images and blood-brain barrier disruption following transcranial focused ultrasound.,” J. Magn. Reson. imaging, vol. 32, no. 3, pp. 593–9, Sep. 2010.

[211] J. J. Choi, J. a Feshitan, B. Baseri, S. Wang, Y.-S. Tung, M. A. Borden, and E. E. Konofagou, “Microbubble-size dependence of focused ultrasound-induced blood-brain barrier opening in mice in vivo.,” IEEE Trans. Biomed. Eng., vol. 57, no. 1, pp. 145–54, Jan. 2010.

[212] M. O’Reilly, A. C. Waspe, M. Ganguly, and K. Hynynen, “Focused-ultrasound disruption of the blood-brain barrier using closely-timed short pulses: influence of sonication parameters and injection rate.,” Ultrasound Med. Biol., vol. 37, no. 4, pp. 587–94, Apr. 2011.

[213] R. Cobbold, Foundations of Biomedical Ultrasound. Oxford University Press. New York, 2007.

[214] D. Vilkomerson, Acoustic imaging with thin annular apertures. Plenum Press. New York, 1973.

[215] S. F. Foster, M. S. Patterson, M. Arditi, and J. W. Hunt, “The conical scanner: a two transducer ultrasound scatter imaging technique,” Ultrason. Imaging, vol. 3, pp. 62–82, 1981.

136

[216] T. Szabo, Diagnostic Ultrasound Imaging: Inside and Out. Elsevier Academic Press. San Diego, 2004.

[217] M. O’Reilly, Y. Huang, and K. Hynynen, “The impact of standing wave effects on transcranial focused ultrasound disruption of the blood-brain barrier in a rat model.,” Phys. Med. Biol., vol. 55, no. 18, pp. 5251–67, Sep. 2010.

[218] G. P. Howles, K. F. Bing, Y. Qi, S. J. Rosenzweig, K. R. Nightingale, and G. A. Johnson, “Contrast-enhanced in vivo magnetic resonance microscopy of the mouse brain enabled by noninvasive opening of the blood-brain barrier with ultrasound.,” Magn. Reson. Med., vol. 64, no. 4, pp. 995–1004, Oct. 2010.

[219] R. M. Aarts, “Acoustic holography for piston sound radiation with non-uniform velocity profiles,” 2010.

[220] R. M. Aarts and A. J. E. M. Janssen, “Estimating the velocity profile and acoustical quantities of a harmonically vibrating membrane from on-axis pressure data.”

[221] L. Shaffaf, “Development of a Phased Array Focused Ultrasound Transducer for Two - -Photon Microscopy Guided Neural Studies,” 2013.

[222] J. Deng, Q. Huang, F. Wang, Y. Liu, Z. Wang, Z. Wang, Q. Zhang, B. Lei, and Y. Cheng, “The Role of Caveolin-1 in Blood-Brain Barrier Disruption Induced by Focused Ultrasound Combined with Microbubbles.,” J. Mol. Neurosci., Aug. 2011.

[223] A. Armulik, G. Genové, M. Mäe, M. H. Nisancioglu, E. Wallgard, C. Niaudet, L. He, J. Norlin, P. Lindblom, K. Strittmatter, B. R. Johansson, and C. Betsholtz, “Pericytes regulate the blood-brain barrier.,” Nature, vol. 468, no. 7323, pp. 557–61, Nov. 2010.

[224] S. M. Stieger, C. F. Caskey, R. H. Adamson, F.-R. E. Curry, E. R. Wisner, and K. W. Ferrara, “Enhancement of Vascular Permeability with Low- Frequency Contrast-enhanced Ultrasound in the Chorioallantoic Membrane Model,” Radiology, vol. 243, no. 1, pp. 112–121, 2007.

[225] M. Göppert-Mayer, “Elementary processes with two quantum transitions,” Ann. Phys., vol. 18, no. 7–8, pp. 466–479, Aug. 2009.

[226] W. Denk, J. H. Strickler, W. W. Webb, N. Series, and N. Apr, “Two-Photon Laser Scanning Fluorescence Microscopy,” Science (80-. )., vol. 248, no. 4951, pp. 73–76, 1990.

[227] B. R. Masters, P. T. So, and E. Gratton, “Multiphoton excitation fluorescence microscopy and spectroscopy of in vivo human skin.,” Biophys. J., vol. 72, no. 6, pp. 2405–12, Jun. 1997.

[228] D. W. Piston, R. G. Summers, S. M. Knobel, and J. B. Morrill, “Characterization of Involution during Sea Urchin Gastrulation Using Two-Photon Excited Photorelease and Confocal Microscopy,” Microsc. Microanal., vol. 4, no. 04, pp. 404–414, Aug. 1998.

137

[229] V. E. Centonze and J. G. White, “Multiphoton excitation provides optical sections from deeper within scattering specimens than confocal imaging.,” Biophys. J., vol. 75, no. 4, pp. 2015–24, Oct. 1998.

[230] P. T. C. So, “Two-photon Fluorescence Light Microscopy,” pp. 1–5, 2002.

[231] P. F. Curley, A. I. Ferguson, J. G. White, and W. B. Amos, “Application of a femtosecond self-sustaining mode-locked Ti: sapphire laser to the field of laser scanning confocal microscopy,” Opt. quantum Electron., vol. 24, no. 8, pp. 851–859, 1992.

[232] C. Marmonier, “Photomultiplier Tube: Principles & Applications,” no. September, 2002.

[233] “Biomicroscopy Lab.” [Online]. Available: http://biomicroscopy.bu.edu/research/nonlinear-microscopy.

[234] M. Gu, X. Gan, A. Kisteman, and M. G. Xu, “Comparison of penetration depth between two-photon excitation and single-photon excitation in imaging through turbid tissue media,” Appl. Phys. Lett., vol. 77, no. 10, p. 1551, 2000.

[235] D. Kobat, N. G. Horton, and C. Xu, “In vivo two-photon microscopy to 1.6-mm depth in mouse cortex.,” J. Biomed. Opt., vol. 16, no. 10, p. 106014, Oct. 2011.

[236] J. T. Huse and E. C. Holland, “Targeting brain cancer: advances in the molecular pathology of malignant glioma and medulloblastoma.,” Nat. Rev. Cancer, vol. 10, no. 5, pp. 319–31, May 2010.

[237] P. S. Tofts, G. Brix, D. L. Buckley, J. L. Evelhoch, E. Henderson, M. V Knopp, H. B. W. Larsson, T. Lee, N. A. Mayr, G. J. M. Parker, R. E. Port, J. Taylor, and R. M. Weisskoff, “Estimating Kinetic Parameters from Dynamic Contrast-Enhanced T1-Weighted MRI of a Diffusable Tracer: Standardized Quantities and Symbols,” Magn. Reson. Imaging, vol. 232, pp. 223–232, 1999.

[238] F. Vlachos, Y.-S. Tung, and E. E. Konofagou, “Permeability assessment of the focused ultrasound-induced blood-brain barrier opening using dynamic contrast-enhanced MRI.,” Phys. Med. Biol., vol. 55, no. 18, pp. 5451–66, Sep. 2010.

[239] T. Nhan, A. Burgess, and K. Hynynen, “Transducer Design and Characterization for Dorsal Based Ultrasound Exposure and Two Photon Imaging of in vivo Blood-Brain Barrier Disruption in a Rat Model,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 2013.

[240] A. Burgess, E. E. Cho, L. Shaffaf, T. Nhan, C. Poon, and K. Hynynen, “The use of two-photon microscopy to study the biological effects of focused ultrsound on the brain,” SPIE Proc., vol. 8226, pp. 822642–7, Feb. 2012.

[241] T. Nhan, A. Burgess, and K. Hynynen, “Transducer Design and Characterization for Dorsal-Based Ultrasound Exposure and Two-Photon Imaging of In Vivo Blood-Brain

138

Barrier Disruption in a Rat Model,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 60, no. 7, pp. 1376–1385, Jul. 2013.

[242] H. P. Erickson, “Size and shape of protein molecules at the nanometer level determined by sedimentation, gel filtration, and electron microscopy.,” Biol. Proced. Online, vol. 11, pp. 32–51, Jan. 2009.

[243] M. R. Dreher, W. Liu, C. R. Michelich, M. W. Dewhirst, F. Yuan, and A. Chilkoti, “Tumor vascular permeability, accumulation, and penetration of macromolecular drug carriers.,” J. Natl. Cancer Inst., vol. 98, no. 5, pp. 335–44, Mar. 2006.

[244] F. Yuan, M. Leunig, S. K. Huang, D. A. Berk, D. Papahadjopoulos, and R. K. Jain, “Mirovascular Permeability and Interstitial Penetration of Sterically Stabilized ( Stealth ) Liposomes in a Human Tumor Xenograft,” Cancer Res., pp. 3352–3356, 1994.

[245] M. W. Dewhirst, S. Shan, Y. Cao, B. Moeller, F. Yuan, and C.-Y. Li, “Intravital fluorescence facilitates measurement of multiple physiologic functions and gene expression in tumors of live animals.,” Dis. Markers, vol. 18, no. 5–6, pp. 293–311, Jan. 2002.

[246] C. C. Michel and F. E. Curry, “Microvascular permeability.,” Physiol. Rev., vol. 79, no. 3, pp. 703–61, Jul. 1999.

[247] A. R. Pries, D. Neuhaus, and P. Gaehtgens, “Blood viscosity in tube flow : dependence on diameter and hematocrit,” Am. J. Physiol. Heart Circ. Physiol., pp. 1770–1778, 1992.

[248] B. Dehouck, L. Fenart, M.-P. Dehouck, A. Pierce, G. Torpier, and R. Cecchelli, “A new function for the LDL receptor: transcytosis of LDL across the blood–brain barrier,” J. Cell Biol., vol. 138, no. 4, pp. 877–889, 1997.

[249] R. L. Roberts, R. E. Fine, and A. Sandra, “Receptor-mediated endocytosis of transferrin at the blood-brain barrier.,” J. Cell Sci., vol. 104 ( Pt 2, pp. 521–32, Feb. 1993.

[250] N. McDannold, N. Vykhodtseva, and K. Hynynen, “Targeted disruption of the blood-brain barrier with focused ultrasound: association with cavitation activity.,” Phys. Med. Biol., vol. 51, no. 4, pp. 793–807, Feb. 2006.

[251] A. Burgess, E. E. Cho, L. Shaffaf, T. Nhan, C. Poon, and K. Hynynen, “The use of two-photon microscopy to study the biological effects of focused ultrasound on the brain,” in Proc. SPIE, 2012, p. 822642−822647.

[252] N. Hosseinkhah and K. Hynynen, “A three-dimensional model of an ultrasound contrast agent gas bubble and its mechanical effects on microvessels.,” Phys. Med. Biol., vol. 57, no. 3, pp. 785–808, Feb. 2012.

[253] A. Garaventa, R. Luksch, S. Biasotti, G. Severi, M. R. Pizzitola, E. Viscardi, A. Prete, S. Mastrangelo, M. Podda, R. Haupt, and B. De Bernardi, “A phase II study of topotecan with

139

vincristine and doxorubicin in children with recurrent/refractory neuroblastoma.,” Cancer, vol. 98, no. 11, pp. 2488–94, Dec. 2003.

[254] M. S. Lesniak, U. Upadhyay, R. Goodwin, B. Tyler, and H. Brem, “Local delivery of doxorubicin for the treatment of malignant brain tumors in rats.,” Anticancer Res., vol. 25, no. 6B, pp. 3825–31, 2005.

[255] C. Rousselle, P. Clair, J. M. Lefauconnier, M. Kaczorek, J. M. Scherrmann, and J. Temsamani, “New advances in the transport of doxorubicin through the blood-brain barrier by a peptide vector-mediated strategy.,” Mol. Pharmacol., vol. 57, no. 4, pp. 679–86, Apr. 2000.

[256] H. Brem, M. S. Mahaley, N. A. Vick, K. L. Black, S. C. Schold, P. C. Burger, A. H. Friedman, I. S. Ciric, T. W. Eller, and J. W. Cozzens, “Interstitial chemotherapy with drug polymer implants for the treatment of recurrent gliomas.,” J. Neurosurg., vol. 74, no. 3, pp. 441–6, Mar. 1991.

[257] S. I. Rapoport, “Advances in osmotic opening of the blood-brain barrier to enhance CNS chemotherapy.,” Expert Opin. Investig. Drugs, vol. 10, no. 10, pp. 1809–18, Oct. 2001.

[258] A. W. El-Kareh and T. W. Secomb, “A mathematical model for comparison of bolus injection, continuous infusion, and liposomal delivery of doxorubicin to tumor cells.,” Neoplasia, vol. 2, no. 4, pp. 325–38, 2000.

[259] Y. M. Goh, H. L. Kong, and C. H. Wang, “Simulation of the delivery of doxorubicin to hepatoma.,” Pharm. Res., vol. 18, no. 6, pp. 761–70, Jun. 2001.

[260] A. W. El-Kareh and T. W. Secomb, “Two-Mechanism Peak Concentration Model for Cellular Pharmacodynamics of Doxorubicin,” Neoplasia, vol. 7, no. 7, pp. 705–713, Jul. 2005.

[261] S. Eikenberry, “A tumor cord model for doxorubicin delivery and dose optimization in solid tumors.,” Theor. Biol. Med. Model., vol. 6, pp. 16–36, Jan. 2009.

[262] W. Zhan and X. Y. Xu, “A mathematical model for thermosensitive liposomal delivery of Doxorubicin to solid tumour.,” J. Drug Deliv., vol. 2013, pp. 1–13, Jan. 2013.

[263] C. S. Patlak and J. D. Fenstermacher, “Measurements of dog blood-brain transfer constants by ventriculocisternal perfusion.,” The American journal of physiology, vol. 229, no. 4. pp. 877–84, Oct-1975.

[264] J. M. Collins and R. L. Dedrick, “Distributed model for drug delivery to CSF and brain tissue,” Am. J. Physiol. Regul. Integr. Comp. Physiol., vol. 245, pp. 303–310, 1983.

[265] T. Nhan, A. Burgess, E. E. Cho, B. Stefanovic, L. Lilge, and K. Hynynen, “Drug delivery to the brain by focused ultrasound induced blood-brain barrier disruption: quantitative evaluation of enhanced permeability of cerebral vasculature using two-photon microscopy.,” J. Control. release, vol. 172, no. 1, pp. 274–80, Nov. 2013.

140

[266] A. Rahman, D. Carmichael, M. Harris, and J. K. Roh, “Comparative Pharmacokinetics of Free Doxorubicin and Doxorubicin Entrapped in Cardiolipin Liposomes,” Cancer Res., vol. 46, no. 5, pp. 2295–2299, May 1986.

[267] V. T. Devita, R. C. Young, and G. P. Canellos, “Combination versus single agent chemotherapy: A review of the basis for selection of drug treatment of cancer,” Cancer, vol. 35, no. 1, pp. 98–110, Jan. 1975.

[268] C. Nicholson, “Diffusion and related transport mechanisms in brain tissue,” Meas. Tech., vol. 64, pp. 815–884, 2001.

[269] J. Nugent and R. Jain, “Extravascular Diffusion in Normal and Neoplastic Tissues,” Cancer Res., vol. 44, pp. 238–244, 1984.

[270] C. Nicholson, K. C. Chen, S. Hrabĕtová, and L. Tao, “Diffusion of molecules in brain extracellular space: theory and experiment.,” Prog. Brain Res., vol. 125, no. 1993, pp. 129–54, Jan. 2000.

[271] M. Hammarlund-Udenaes, M. Fridén, S. Syvänen, and A. Gupta, “On the rate and extent of drug delivery to the brain.,” Pharm. Res., vol. 25, no. 8, pp. 1737–50, Aug. 2008.

[272] D. J. Kerr, A. M. Kerr, R. I. Freshney, and S. B. Kaye, “Comparative intracellular uptake of adriamycin and 4’-deoxydoxorubicin by non-small cell lung tumor cells in culture and its relationship to cell survival.,” Biochem. Pharmacol., vol. 35, no. 16, pp. 2817–23, Aug. 1986.

[273] J. Cummings and C. S. McArdle, “Studies on the in vivo disposition of adriamycin in human tumours which exhibit different responses to the drug.,” Br. J. Cancer, vol. 53, no. 6, pp. 835–8, Jun. 1986.

[274] W. M. Pardridge, “The blood-brain barrier and neurotherapeutics.,” NeuroRx, vol. 2, no. 1, pp. 1–2, Jan. 2005.

[275] I. Sardi, G. la Marca, M. G. Giovannini, S. Malvagia, R. Guerrini, L. Genitori, M. Massimino, and M. Aricò, “Detection of doxorubicin hydrochloride accumulation in the rat brain after morphine treatment by mass spectrometry.,” Cancer Chemother. Pharmacol., vol. 67, no. 6, pp. 1333–40, Jun. 2011.

[276] L. L. Muldoon and E. A. Neuwelt, “BR96–DOX Immunoconjugate Targeting of Chemotherapy in Brain Tumor Models,” J. Neurooncol., vol. 65, no. 1, pp. 49–62, Oct. 2003.

[277] F.-Y. Yang, W.-M. Fu, R.-S. Yang, H.-C. Liou, K.-H. Kang, and W.-L. Lin, “Quantitative evaluation of focused ultrasound with a contrast agent on blood-brain barrier disruption.,” Ultrasound Med. Biol., vol. 33, no. 9, pp. 1421–7, Sep. 2007.

[278] “Chemical structure for METHOTREXATE,” Toxicology Data Network. [Online]. Available: http://toxnet.nlm.nih.gov/cgi-bin/sis/search/a?dbs+hsdb:@term+@DOCNO+3123.

141

[279] V. Laquintana, A. Trapani, N. Denora, F. Wang, J. M. Gallo, and G. Trapani, “New strategies to deliver anticancer drugs to brain tumors.,” Expert Opin. Drug Deliv., vol. 6, no. 10, pp. 1017–32, Oct. 2009.

[280] Y.-S. Tung, F. Vlachos, J. J. Choi, T. Deffieux, K. Selert, and E. E. Konofagou, “In vivo transcranial cavitation threshold detection during ultrasound-induced blood-brain barrier opening in mice.,” Phys. Med. Biol., vol. 55, no. 20, pp. 6141–55, Oct. 2010.

[281] M. A. O’Reilly and K. Hynynen, “Feedback-controlled Focused Ultrasound Disruption by Using an Acoustic Emissions – based Controller,” Radiology, vol. 263, no. 1, 2012.

[282] A. Burgess, T. Nhan, C. Moffatt, A. L. Klibanov, and K. Hynynen, “Analysis of focused ultrasound-induced blood-brain barrier permeability in a mouse model of Alzheimer’s disease using two-photon microscopy.,” J. Control. Release, vol. 192, pp. 243–8, Oct. 2014.

[283] J. F. Jordão, C. a Ayala-Grosso, K. Markham, Y. Huang, R. Chopra, J. McLaurin, K. Hynynen, and I. Aubert, “Antibodies targeted to the brain with image-guided focused ultrasound reduces amyloid-beta plaque load in the TgCRND8 mouse model of Alzheimer’s disease.,” PLoS One, vol. 5, no. 5, p. e10549, Jan. 2010.

[284] “Next step in scalpel-free surgery,” Sunnybrook Hospital. [Online]. Available: http://sunnybrook.ca/media/item.asp?i=1098.

[285] M. A. O’Reilly and K. Hynynen, “A super-resolution ultrasound method for brain vascular mapping.,” Med. Phys., vol. 40, no. 11, p. 110701, Nov. 2013.

[286] M. A. O’Reilly, R. M. Jones, and K. Hynynen, “Three-dimensional transcranial ultrasound imaging of microbubble clouds using a sparse hemispherical array.,” IEEE Trans. Biomed. Eng., vol. 61, no. 4, pp. 1285–94, Apr. 2014.

[287] E. C. Unger, T. Porter, W. Culp, R. Labell, T. Matsunaga, and R. Zutshi, “Therapeutic applications of lipid-coated microbubbles.,” Adv. Drug Deliv. Rev., vol. 56, no. 9, pp. 1291–314, May 2004.

[288] S. Hernot and A. L. Klibanov, “Microbubbles in ultrasound-triggered drug and gene delivery.,” Adv. Drug Deliv. Rev., vol. 60, no. 10, pp. 1153–66, Jun. 2008.

[289] M. A. Borden, C. F. Caskey, E. Little, R. J. Gillies, and K. W. Ferrara, “DNA and polylysine adsorption and multilayer construction onto cationic lipid-coated microbubbles.,” Langmuir, vol. 23, no. 18, pp. 9401–8, Aug. 2007.

[290] Y. K. Ryu, S. Khan, S. C. Smith, and C.D. Mintz, “Isoflurane impairs the capacity of astrocytes to support neuronal development in a mouse dissociated coculture model.,” J. Neurosurg Anesthesiol., vol. 26, no. 4, pp. 363–8, Oct. 2014.

Drug Delivery to the Brain by Focused Ultrasound and ... · using Two-photon Fluorescent Microscopy...

Documents

Transcript of Drug Delivery to the Brain by Focused Ultrasound and ... · using Two-photon Fluorescent Microscopy...