Simultaneous PET/fMRI
for Imaging Neuroreceptor Dynamics
by
MASSAO 8USETT INSTmUTE
TECHNOLOGY
Christin Y. Sander
S EP 2 5 2014
M.Sc., University College London (2009)
M.Eng., Imperial College London (2008)
L BRARIES
Submitted to the Department of
Electrical Engineering and Computer Science
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy in Electrical Engineering and Computer Science
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
September 2014
@ Massachusetts Institute of Technology 2014. All rights reserved.
Signature redacted
Author ...........
.
.................
Department of
Electrical Engineering and Computer Science
August 20, 2014
Certified by....
S ~ignature
redacted
.
...........
Bruce R. Rosen
Director, Athinoula A. Martinos Center
Profes sor of Radiology and Healt} Sciences & Technology
esas"up ervisor
Certified by.
Sig nature redacted
Joseph B. Mandeville
Instructor in Radiology, Assistant in Physics
j,-9/
Accepted by ......
,
Thesis Supervisor
e
Signature redacted
6 /Q
...........
eslie Kolodziejski
Professor of Electrical Engineering
Chair, Department Committee on Graduate Theses
2
Simultaneous PET/fMRI
for Imaging Neuroreceptor Dynamics
by
Christin Y. Sander
Submitted to the Department of
Electrical Engineering and Computer Science
on August 20, 2014, in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy in Electrical Engineering and Computer Science
Abstract
Whole-brain neuroimaging is a key technique for studying brain function and connectivity.
Recent advances in combining two imaging modalities - magnetic resonance imaging (MRI)
and positron emission tomography (PET) - into one integrated scanner, have created the
opportunity to explore the underlying neurochemistry of brain function in more detail.
Imaging these dynamics plays an important role for understanding drug action and function
of neurochemical pathways in the brain and is crucial, yet largely unexplored, for creating
and evaluating treatment of neurological and psychiatric disorders.
In this thesis, we first address technological challenges in simultaneous PET/MRI by
designing, building and evaluating PET compatible MR probes for brain imaging, which
enable highly sensitive dual modality imaging. We then develop simultaneous imaging
methods with PET and functional MRI to assess and validate relationships between receptor
occupancy and changes in brain activity due to pharmacological challenges targeting the
dopamine system. Our results indicate that dopamine receptor occupancies and vascular
responses are correlated in anatomical space and with pharmacological dose. Moreover,
the temporal dynamics of the signals show that a direct neurovascular coupling between
receptor occupancy and hemodynamics exists and that a temporal divergence between PET
and fMRI can be used to investigate previously unexplored neurochemical parameters and
adaptation mechanisms in vivo. Overall, our findings provide insight into dopaminergic
receptor dynamics and their effects on high-level brain function, paving a way to address
receptor-specific brain dysfunction effectively.
Thesis Supervisor: Bruce R. Rosen
Title: Director, Athinoula A. Martinos Center
Professor of Radiology and Health Sciences & Technology
Thesis Supervisor: Joseph B. Mandeville
Title: Instructor in Radiology, Assistant in Physics
3
4
Acknowledgments
This thesis would not have been possible without the support, help and love of many
individuals.
My advisor Bruce Rosen has been a source of inspiration to me all these years:
I am grateful for all the insightful discussions and thoughtful advice. From him, I
learned not only how to evaluate scientific topics critically but also about a passion
for science. I feel fortunate to have such a brilliant, kind and supportive mentor.
My sincerest thanks go to Joe Mandeville: His invaluable expertise and intellect
have shaped many aspects of this thesis. He has taught me pivotal skills, starting
from experimental imaging procedures to exploring novel methods and approaches
to solving problems. I truly value all his input and unwavering support throughout
these years.
I thank Larry Wald for giving me the opportunity to pursue the hardware side
of PET/MRI, which really set a foundation for my research direction. I really enjoy
being part of the Wald lab and am thankful for his guidance.
I extend my deep gratitude to my thesis committee members: Elfar Adalsteinsson
has always encouraged me to shape my own academic pathway and to take initiative while staying focused on important milestones. Martha Gray provided me with
invaluable feedback on my thesis work and advice that I have appreciated very much.
I am very grateful to all the mentors who have shaped many aspects of my thesis
work and who have supported me with their enthusiasm: I thank Jacob Hooker for
his support and excitement towards my projects and for encouraging me to think
outside the box during many of our inspiring discussions. I thank Ciprian Catana
for supporting my work with great enthusiasm and for guiding my attention to many
important technical aspects. I am grateful to Boris Keil for sharing his knowledge on
the art of RF coil building. Finally, many thanks also go to Nathaniel Alpert, Giorgio
Bonmassar, Gitte Knudsen, Marc Normandin, Julie Price and Wim Vanduffel for their
contributions and stimulating discussions.
5
It has been a real pleasure to interact with everyone at the Martinos Center and
I am truly grateful for the efforts from many people that were vital in shaping my
work and establishing the PET/MR research over the years at the Martinos Center:
Spencer Bowen, Kevin Chen, Dan Chonde and David Izquierdo for their contributions
to the BrainPET tools and PET reconstructions. Bastien Guerin, Hanne Hansen,
Jon Polimeni, Marjorie Villien and Hsiao-Ying Wey for great discussions and their
collaborations in PET/MRI. Grae Arabasz and Shirley Hsu for making experiments
run smoothly. Helen Deng for her help with animal work and Avilash Cramer for
his coil contributions. Ehimen Aisaborhale, Steve Carlin, Chris Moseley, Nathan
Schauer, Judit Sore and Colin Wilson for all their radiotracer productions. And I
would like to thank all of them and more for their daily inspiration, friendships and
support: Berkin Bilgic, Steven Cauley, Itthi Chatnuntawech, Clarissa Cooley, JP
Coutu, Cornelius Eichner, Audrey Fan, Trina Kok, Cris LaPierre, Azma Mareyam,
Mark Schuppert, Kawin Setsompop, Jason Stockmann, Jeff Stout, Thomas Witzel,
Filiz Yetizir, Wei Zhao and Bo Zhu.
I would like to acknowledge the funding support I received from the "Studienstiftung des Deutschen Volkes" and its European Recovery Program, the MIT-endowed
Cronin fellowship, the CIMIT/MIT foundation and NIH training grants. All these
have enabled me to explore my research topics and to obtain a multi-faceted experience during graduate school.
Many thanks to the EECS graduate office, Leslie Kolodziejski, Janet Fischer and
Terry Orlando, whose kind support I have valued very much. Thanks also go to
Donna Crowe, Arlene Wint and the admin support at the Martinos for their friendly
help and organization.
My warmest thanks go to all my friends both at MIT and beyond, who made my
experience in graduate school and in Boston extraordinary. I will cherish the memories
of precious times spent together. I thank Flavio for his cheer and support throughout
this journey. Last but not least, I am deeply grateful to my sister Michelle and my
parents for their unwavering love, support and encouragement in all my endeavors.
6
Contents
1 Introduction and Motivation
26
27
Background and Context
Imaging Brain Function ................
. . . . . . . . .
27
2.2
Integrated PET/MR Scanners . . . . . . . . . .
. . . . . . . . .
29
Radiofrequency Coils for PET/MRI . . .
. . . . . . . . .
30
2.3
Dynamics of Brain Function . . . . . . . . . . .
. . . . . . . . .
32
2.4
The Dopamine Receptor System . . . . . . . . .
. . . . . . . . .
33
PET Imaging of Dopamine Receptors . .
. . . . . . . . .
35
. . . . . . . . . . . . . . . . .
. . . . . . . . .
38
2.5.1
The Classical Receptor Occupancy Model
. . . . . . . . .
38
2.5.2
The Internalization Model . . . . . . . .
. . . . . . . . .
40
2.5
Receptor Theory
.
2.4.1
.
.
.
2.2.1
.
.
2.1
.
2
Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
1.1
23
3 A PET Compatible 31-channel MR Brain A rray Coil
3.1
Introduction .....................
3.2
M ethods ....................
43
43
. . . . . . . . . . . . . .
45
Existing Hardware Components ...
. . . . . . . . . . . . . .
45
3.2.2
PET Compatibility of Coil Design .
. . . . . . . . . . . . . .
45
3.2.3
Array Coil Design ............
. . . . . . . . . . . . . .
47
3.2.4
Coil Construction ............
. . . . . . . . . . . . . .
48
3.2.5' Coil Bench Tests . . . . . . . . . .
. . . . . . . . . . . . . .
50
MRI Data Acquisition and Analysis
. . . . . . . . . . . . . .
52
3.2.6
.
.
.
3.2.1
7
3.2.7
CT Data Acquisition and Processing
52
3.2.8
Derivation of Attenuation Map and Attenuation Correction. .
53
3.2.9
PET Data Acquisition and Reconstruction . . . . . . . . . . .
53
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
54
3.3.1
Coil Component Evaluation . . . . . . . . . . . . . . . . . . .
54
3.3.2
MR Performance of Array Coil
. . . . . . . . . . . . . . . . .
55
3.3.3
PET Performance/Attenuation of Array Coil . . . . . . . . . .
57
3.4
Discussion . . . . . . . . . . . . . . . . . . . . .-. . . . . . . . . . . .
61
3.5
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
67
3.6
Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
68
3.3
4 Neurovascular Coupling to D2/D3 Receptor Occupancy using simultaneous PET/fMRI
. . . . . . . . .
69
4.2 Materials and Methods . . . . . . . . . . . . . . . .
. . . . . . . . .
71
.
Introduction . . . . . . . . . . . . . . . . . . . . . .
.
4.1
69
Animal Studies . . . . . . . . . . . . . . . .
. . . . . . . . .
71
4.2.2
Study Design . . . . . . . . . . . . . . . . .
. . . . . . . . .
71
4.2.3
PET/fMRI Image Acquisition . . . . . . . .
. . . . . . . . .
72
4.2.4
fMRI Data Analysis
. . . . . . . . . . . . .
. . . . . . . . .
73
4.2.5
PET Data Analysis . . . . . . . . . . . . . .
. . . . . . . . .
74
4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . .
75
.
.
.
.
.
.
4.2.1
Kinetic Modeling Results . . . . . . . . . . .
. . . . . . . . .
76
4.3.2
Temporal Correlation . . . . . . . . . . . . .
. . . . . . . . .
77
4.3.3
Spatial Correlation . . . . . . . . . . . . . .
. . . . . . . . .
78
4.3.4
Occupancies and Relationship to fMRI . . .
. . . . . . . . .
79
4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . .
82
.
.
.
.
.
4.3.1
Regional Dose Response . . . . . . . . . . .
. . . . . . . . .
82
4.4.2
Temporal Comparisons . . . . . . . . . . . .
. . . . . . . . .
83
4.4.3
Neurovascular Coupling mediated by D2/D3R Antagonism
4.4.4
Basal Dopamine Occupancy . . . . . . . . .
.
.
.
4.4.1
8
.
83
. . . . . . . . .
84
.
4.5
4.4.5
Study Limitations . . . . . . . . . . . . . . . . . . . . . . . . .
85
4.4.6
Potential Clinical Applications . . . . . . . . . . . . . . . . . .
86
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
87
5 Imaging Agonist-induced D2/D3 Receptor Internalization
89
5.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
89
5.2
T heory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
91
5.2.1
PET and fMRI Signal Models . . . . . . . . . . . . . . . . . .
91
5.2.2
Classical Occupancy Model: Efficacy of Drugs . . . . . . . . .
93
5.2.3
Internalization Model . . . . . . . . . . . . . . . . . . . . . . .
94
M ethods . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . .
95
5.3.1
Animal Model . . . . . . . . . . . . . . . . . . . . . . . . . . .
95
5.3.2
Study Design . . . . . . . . . . . . . . . . . . . . . . . . . . .
95
5.3.3
PET/MR Image Acquisition and Reconstruction . . . . . . . .
96
5.3.4
PET and fMRI Data Analysis . . . . . . . . . . . . . . . . . .
97
5.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
97
5.3
5.4.1
Simulations of the Classical Occupancy Model . . . . . . . . .
97
5.4.2
Simulations of the Internalization Model . . . . . . . . . . . .
98
5.4.3
Spatial Correlation of Occupancy and CBV
5.4.4
Dose Response from Occupancy and CBV . . . . . . . . . . . 102
5.4.5
Temporal Dissociation between PET/MR Timecourses
5.4.6
Estimation of Internalization Constants . . . . . . . . . . . . . 105
. . . . . . . . . . 101
. . . .
103
5.5
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.6
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
6 Effects of Flow Changes on D2/D3 Radiotracer Dynamics
109
6.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.2
M ethods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
6.2.1
Animal Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
6.2.2
Study Design with Hypercapnia . . . . . . . . . . . . . . . . . 111
6.2.3
PET/MR Image Acquisition . . . . . . . . . . . . . . . . . . . 112
9
6.3
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
6.3.1
Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . 112
6.3.2
Experimental Results: ["C]raclopride . . . . . . . . . . . . . . 113
6.3.3
Experimental Results: [18F]fallypride . . . . . . . . . . . . . . 115
6.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
6.5
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
7 Conclusions and Future Work
121
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
7.1
Sum mary
7.2
Imaging Neuroreceptor Dynamics . . . . . . . . . . . . . . . . . . . . 124
7.3
7.2.1
Imaging other Neuroreceptor Systems . . . . . . . . . . . . . . 124
7.2.2
Imaging Neurochemical Connectivity . . . . . . . . . . . . . . 125
7.2.3
Development of Biological Models . . . . . . . . . . . . . . . . 125
Clinical Applications of PET/fMRI . . . . . . . . . . . . . . . . . . . 126
7.3.1
Evaluation of Psychiatric Drugs . . . . . . . . . . . . . . . . . 126
7.3.2
The Dopamine Hypothesis in Schizophrenia . . . . . . . . . . 127
7.3.3
Deep Brain Stimulation in Parkinson's Disease and Depression 128
7.4 Beyond Imaging with PET/MRI. . . . . . . . . . . . . . . . . . . . . 128
7.4.1
Simultaneous Imaging with EEG-PET/MRI . . . . . . . . . . 128
A Appendix: PET Compatible Brain Array Coils for Non-Human Primate Imaging
131
A .1 M otivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 131
A.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 132
A.3 Performance of NHP coils . . . . . . . . . . . . . . . . . . . . . . . . 133
A.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
B Appendix: Details of Biological Models
137
B.1 Relationship between Raclopride and Dopamine Occupancies . . . . . 137
B.2 Occupancy and Dynamic Binding Potential ..................
140
B.3 Forward-model Simulation and Analysis ................
143
10
List of Figures
2-1
Illustration of the chain of biochemical processes in functional brain
activation and the associated imaging modalities that can measure
events at various stages. . . . . . . . . . . . . . . . . . . . . . . . . .
29
2-2 Available RF coils for the detection of MR signals. While commercially
available coils with 32 channels exist for standard MRI systems, an 82-3
channel coil exists for PET/MR imaging with the BrainPET scanner.
31
An overview of the correlations of fMRI with other measurements. . .
33
2-4 Anatomy of the basal ganglia and their subdivisions in the human brain. 34
2-5
Compartmental models for PET kinetic modeling. Left: General 2tissue compartmental model. Right: Simplified reference tissue model
applicable for the radiotracer ["C]raclopride.
. . . . . . . . . . . . .
36
2-6 The classical occupancy model for the case of a D2 antagonist exposure. 39
2-7 The internalization model for the case of a D2 agonist exposure.
3-1
40
511 keV attenuation map of coil components and its evaluation for the
final design choices of the 31-channel PET/MR array coil.
3-2
. .
. . . . . .
46
View of the 31-channel PET/MR array coil with sparse elements in
the PET FOV, its division into an anterior and posterior part and its
fit into a local transmit coil. . . . . . . . . . . . . . . . . . . . . . . .
3-3
47
Circuit schematic for coil elements with a long coaxial cable that allows placement of the preamplifier outside the PET FOV. In order to
maintain a A/2 phase shift, a pi phase shifter is located before the
preamplifier chain.
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
49
3-4 The 31-channel coil design allowed bundled output cables to be eliminated from the PET FOV. (a) 31-channel coil with output cables. (b)
Superior end of the 31-channel coil inside the local transmit coil with
output cables leaving the coil. . . . . . . . . . . . . . . . . . . . . . .
50
3-5 Noise correlation matrices for the 8-channel PET/MR, the 31-channel
PET/MR and the 32-channel MR-only array. The average noise correlation is 11.8%, 21.4% and 12.4% respectively. . . . . . . . . . . . .
3-6
56
Signal to noise maps of a human head-shaped water phantom (sagittal
and axial slice) for the 8-channel PET/MR, the 31-channel PET/MR
and a standard 32-channel MR-only array. . . . . . . . . . . . . . . .
3-7
57
1/g factor maps for the 8-channel PET/MR, the 31-channel PET/MR
and the 32-channel MR-only array for acceleration factors R in 1D and
2D, with corresponding histograms of 1/g factor maps for the three
coils in comparison. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3-8
58
Axial CT scans for z-positions -7.5 cm, 0 cm and 7.5 cm of the 8channel PET/MR, the 31-channel PET/MR and a standard MR-only
32-channel coil. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3-9
59
Sinograms calculated from a forward projection of CT scans, with mean
attenuation correction factors (ACF) of the 31-channel PET/MR coil,
the 8-channel PET/MR and a standard MR-only 32-channel coil. .
60
3-10 PET images of a cylindrical phantom filled with [18F]-solution, acquired
using the setup with the 8-channel, 31-channel or no coil as reference;
and corresponding images of relative change in percent. . . . . . . . .
62
3-11 Plots of the mean value of relative change in photon counts between a
phantom with and without a coil in the PET FOV for five ROIs over
the length of the cylinder phantom. . . . . . . . . . . . . . . . . . . .
63
3-12 Simultaneously acquired [ 1 8F]FDG PET and MRI in a human subject.
Artifact-free PET images demonstrate an accurate implementation of
the 31-channel coil attenuation correction. MR images show the superior g-factor and SNR of the 31-channel coil. . . . . . . . . . . . . . .
12
65
4-1
Schematic illustrating the basic PET/fMRI paradigm for antagonism
at D2/D3R . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
76
4-2 PET time activity curves and modeling fits for three raclopride injections. 77
4-3 PET specific binding overlaid with CBV timecourses for two doses or
raclopride. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4-4 Parameter maps showing voxelwise DBP'PD
maps from PET and
%CBVP** change from fMRI for four raclopride injections. . . . . . .
4-5 %CBVpek vs. raclopride occupancy.
5-1
. . . . . . . . . . . . . . . . . .
80
81
Schematic illustrating a compartmental model that describes receptor
internalization at dopaminergic neurons.
5-2
78
. . . . . . . . . . . . . . . .
92
Simulated CBV timecourses (without internalization) for the same occupancies but varying efficacies of a ligand. Antagonists (e = 0) show
a positive CBV signal, whereas full agonists (e = 1) show a negative
CBV signal. For partial agonists, the response depends on the basal
DA occupancy.
5-3
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
98
Simulation results from the proposed model of receptor internalization
that show how PET and fMRI signal timecourses are affected for different rates of internalization due to a D2/D3 agonist injection at time
t=O. With very short time constants for internalization (5 min), the
fMRI timecourse is shortened, whereas PET occupancy stays elevated
for much longer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
99
5-4 Simulated PET occupancy measures for the duration of a dynamic
PET scan for different affinities a of internalized receptors. . . . . . . 100
5-5
Voxelwise parametric maps showing PET and fMRI parameters due to
three different dose injections of the D2/D3 agonist quinpirole. . . . . 101
5-6 Plot of CBVP*k measures for putamen and caudate ROIs against peak
occupancy due to three dose injections of quinpirole injections (0.1, 0.2
and 0.3 mg/kg). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
13
5-7 Top row: PET time activity curves for the caudate and cerebellum
ROIs for three different doses of quinpirole injection at 35- min, with
kinetic modeling fits from SRTM2 with cerebellum as the reference.
Bottom row: Corresponding CBV timecourses that show the negative
response due to the challenge in the caudate ROI. . . . . . . . . . . . 103
5-8 CBV timecourses in the caudate ROI due to a repeated injection of
quinpirole at 0.2 mg/kg. The first dose, administered at 30 min, elicits
a negative CBV response. But a second dose given 100 min later at
130 min does not elicit any CBV response.
. . . . . . . . . . . . . . 104
5-9 Voxelwise maps showing a direct comparison of the PET and CBV
response in putamen and caudate for pharmacological challenges with
a D2/D3 antagonist (raclopride, left) and the D2/D3 agonist (quinpirole, left). The differences in the signals can give an estimate of basal
dopamine occupancies in vivo.
6-1
Simulations of a reference tissue model with kinetic parameters for
["C]raclopride.
6-2
. . . . . . . . . . . . . . . . . . . . . 106
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Experimental results from a [ 11C]raclopride bolus injection and hypercapnic challenges of 7% CO 2 applied three times during the experiment. 114
6-3
Experimental results from a [18F]fallypride bolus injection and hypercapnic challenges of 7% CO2 applied five times during the experiment. 116
6-4
Experimental results from a [18F]fallypride bolus injection and hypercapnic challenges of 7% CO 2 applied before the start of the experiment
to observe flow-induced uptake differences. . . . . . . . . . . . . . . . 117
A-1 Small and large 8-channel array coils for simultaneous PET/MR imaging of non-human primates.
. . . . . . . . . . . . . . . . . . . . . . . 132
A-2 Top: 51.1 keV attenuation map of cross-section of small and large monkey coil and mMR human coil. Bottom: Sinograms of attenuation
correction factors for each coil. . . . . . . . . . . . . . . . . . . . . . . 133
14
A-3 SNR maps (top) and 1/g factor maps (bottom) of the small and large
monkey coil compared to the human head-neck coil. . . . . . . . . . . 134
A-4 PET/MR images acquired with the small monkey coil on the BrainPET
PET/MR scanner with the radiotracer [ 1 C]raclopride (top) and on the
whole-body PET/MR with the radiotracer ["C]CW4 (bottom). . ...
135
B-1 Simulations of a multi-compartmental model based upon first-order
kinetics: Specific binding of raclopride vs. time and for increasing
dose, normalized to the maximum value in order to observe differences
in shape due to receptor saturation. . . . . . . . . . . . . . . . . . . . 138
B-2 The relationship between the fractional occupancies of raclopride (ORAc)
and the change in fractional occupancy of DA
(A9 DA),
which drives
the fMRI signal, using the linear approximation of Eq. (B.3), (dashed
black) or estimating effects of DA displacement (blue) and displacement plus release (red). Calculations assume a 20% basal DA occupancy. 140
B-3 Comparison of an SRTM fit (black) with a dynamic fit (red) that allows the use of DBPND (for the 16 pg/kg RAC mass dose in M2).
The dynamic method shows an improved fit visually and by X 2 /DOF
measures.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
B-4 Analysis of simulated data to determine the accuracy of occupancy
estimates relative to true peak change in occupancy. Simulations were
analyzed within the SRTM framework by GLM. . . . . . . . . . . . . 144
B-5 Estimate of specific binding computed from the data, compared with
specific binding, modeled with a full reference tissue model based on
fitted parameters for the two highest RAC mass doses.
15
. . . . . . . . 145
16
List of Tables
3.1
Relative change values in percent (average
based analysis.
4.1
std. dev.) from an ROI-
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
64
Summary of parameters and outcomes of the PET/fMRI paradigm for
putamen and two animals (M1, M2). . . . . . . . . . . . . . . . . . .
72
B.1 Parameters used in forward-model simulations. . . . . . . . . . . . . .
146
17
18
List of Abbreviations and Symbols
a
Ratio of affinity of the radiotracer to internalized and external receptors
Bavaji
(a = 1/KD,
1/KD,ett)
Receptors available for binding
BDA
D2 receptors bound by DA
Bmax
Total density of D2 receptors
BP
Binding potential (BP = Bmax/KD)
BPND
Binding potential (non-displaceable) (BPND = Bavai/KD)
BRAC
D2 receptors bound by RAC
BW
Bandwidth
CBV
Cerebral blood volume
CT
Computed Tomography
D2R/D3R
D2/D3 (dopamine) receptor
DA
Dopamine
DBPND
"Dynamic" binding potential (Eq. 4.1)
etCO 2
End-tidal CO 2
eL
Efficacy of the ligand (drug) L
FDA,
FL
fDA, fL
Free dopamine, ligand (drug) concentration
Free dopamine, ligand (drug) concentration normalized to
baseline DA occupancy
OA0
fMRI
Functional magnetic resonance imaging
FOV
Field of view
GRAPPA
Generalized autocalibrating partially parallel acquisitions
19
Hounsfield units
K1
Delivery rate constant (for transfer from arterial plasma to tissue) (in min-')
Kj'
Delivery rate constant to reference region (cerebellum) (in min- 1
k2
Rate constant for transfer from tissue to arterial plasma (in min-1)
k3
Rate constant for transfer from free to bound compartment
)
HU
(k 3 = konBma)
Rate constant for transfer from bound to free compartment
k4
)
(k 4 = koff), (in min- 1
KD,DA
KD,
L
Dissociation constant for dopamine at D2 receptors
Dissociation constant for a ligand (drug) at D2 receptors
Dissociation constant for radiotracer (raclopride) at D2/D3 receptors
Kint
Internalization time constant (in min- 1
L
Ligand (challenge, or drug)
Ml, M2
Animals (monkeys) in experiments
MRI
Magnetic Resonance Imaging
NHP
Non-human primate
pcASL
Pseudo-continuous arterial spin labeling
PET
Positron Emission Tomography
R,
Ratio of delivery to region of interest and reference tissue
)
KD
(R1 = K 1 /K()
RAC
Raclopride
RF
Radiofrequency
Rx
Receive
ROI
Region of interest
SENSE
Sensitivity encoding (for fast MRI)
SRTM
Simplified reference tissue model [1]
SRTM2
Simplified reference tissue model reduced to 2 parameters
TAC
Time-activity curve
TE
Echo time
TR
Repetition time
20
Tx
Transmit
0avail
Fraction of available receptors
0(O)
DA
Fractional dopamine occupancy at baseline
ODA, ORAC
Fractional occupancies of dopamine and raclopride
OPET
Occupancy as measured by PET
X 2 /DOF
Chi-square per degree of freedom
21
22
Chapter 1
Introduction and Motivation
Whole-brain in vivo neuroimaging studies to date have made an impact by advancing high-resolution anatomy and investigating brain function through structural and
functional connections. Yet, our understanding of neurochemical connections, their
dynamic interactions and its effect on high-level brain function is limited, even though
these aspects are of utmost importance for any drug acting on neurochemical pathways in the brain. While we know a tremendous amount about neurotransmitter
and receptor systems, relatively little is understood about the chemical underpinning
of neurotransmission and its modulation in vivo. Methods that can simultaneously
link molecular level changes to broader functional signaling have been largely unexplored until now, partly due to the lack of measurement techniques. In this thesis,
we address this knowledge gap by developing new technology and methodology for
simultaneous imaging with positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) that enables the mapping of dynamic neuroreceptor
function.
A central piece of this thesis is the cutting-edge technology that integrated MRI
with PET imaging. At the start of this thesisl the first human prototype simultaneous
PET/MR scanner - the first of a new generation of a number of multimodal PET/MR
scanners - had just recently been installed. While there were technical challenges to
overcome, the question of why PET and MR imaging should be done simultaneously
was still a topic of debate. By advancing the detection technology for PET/MRI and
23
then making combined measurements with PET and fMRI as proposed in this thesis,
we were able to show a strong example of the importance of an integrated PET/MRI
technology.
As a first step, we aimed to advance PET/MR detector imaging technology to be
able to produce highly sensitive images for both modalities, which is particularly important for functional imaging applications. MR detectors that have a high number
of array channels have proven valuable for increased image sensitivity and acquisition
speed [2]. But simultaneous acquisition of MR and PET images requires the placement of the MR detection coil inside the PET detector ring where it absorbs and
scatters photons. This constraint is the principal barrier to achieving optimum sensitivity on each modality. Although MR detectors for PET/MR scanners exist, they
have a limited number of channels and are not necessarily optimized for PET compatibility. We addressed this issue by designing an MR detector for human brain imaging
with a high channel count (31 channels) that improved imaging for PET/MR. In addition, we built two PET compatible MR coils for dedicated large animal imaging
that enabled the functional imaging studies presented in this thesis.
Following the construction of specialized coils that are highly sensitive for PET/MR
imaging, the natural next step was to extract complementary information from two
measurements: High spatial and temporal resolution from fMRI and the specific
underlying neurochemistry from PET. Being able to characterize the relationship between receptor occupancy measures and its downstream functional effects marks a
first step for understanding receptor-specific functional activation. In the context
of the dopamine receptor system, we thus devised imaging methods and biological
models that enabled us to demonstrate the relationships between dopaminergic occupancy measures from PET and changes in brain activity inferred by fMRI. Overall,
our results demonstrate the utility of simultaneous PET/fMRI for investigations of
neurovascular coupling that correlate neurochemistry with hemodynamic changes in
vivo for any receptor system with an available PET tracer.
The brain is a complex system that can modulate its neuronal response patterns
and adapt to extreme environments. Certain pharmacological stimuli are known to
24
induce receptor trafficking mechanisms in order to modulate their downstream functional response. Although such adaptation mechanisms have been explored with in
vitro measurements, they are not well understood in the context of in vivo synaptic
activity - one of the reasons being that no techniques exist to measure these phenomena in vivo. In light of this, we extended our initial functional imaging methodologies
with PET/fMRI to provide a measurement of receptor trafficking in vivo. Together
with the proposition of a relevant biological model, we are able to extract information from both modalities to learn more about the dynamics of receptor adaptation
mechanisms in vivo.
Finally, we addressed some known controversies in the quantification of PET imaging. The question that probably every PET modeling scientist has asked him- or herself at some point, is whether changes in blood flow can influence the quantification
of dynamic PET imaging results in the case of receptor-targeted studies. While this
question has been pursued mainly through simulations, the answers strongly depend
on the assumptions being used and have thus been met with controversy. By proposing imaging methods and carrying out experiments with PET/MRI, we demonstrated
how simultaneous acquisition of PET and flow imaging techniques from MRI can be
used to evaluate the question of flow changes on radiotracer dynamics.
Overall, this thesis' contributions to the field of simultaneous PET/MR imaging
were to develop both unique hardware components and advanced physiological modeling, and devise experiments that enabled us to better understand the in vivo connections between neuroreceptor dynamics and the distributed function of the brain.
The methods and results presented in this thesis can put forward new questions and
inspire solutions by exploring the dynamics of how function of the brain is driven neurochemically. In addition, they have the potential to provide a platform that could
guide us about when and where to intervene with neurochemistry-modulating drugs
to restore healthy brain function.
25
1.1
Thesis Outline
This thesis covers the following topics:
" Chapter 2: Background and Context
An introduction to combined imaging with PET and MRI and its value in imaging
brain function is given. Relevant background to concepts described in this thesis are
covered, and the work of the thesis is placed in context.
" Chapter 3: A PET'Compatible 31-channel MR Brain Array Coil
The design, construction and performance evaluation of a multichannel array coil for
highly parallel imaging of the brain is described. The coil is designed to have low
attenuation of 511 keV (PET) photons for the use in a combined PET/MR scanner.
" Chapter 4: Neurovascular Coupling to D2/D3 Receptor Occupancy using
simultaneous PET/fMRI
Functional imaging studies of the dopamine D2/D3 system with simultaneous PET/fMRI
are presented. Using a pharmacological stimulus that targets D2/D3 receptors, coupling mechanisms between receptor occupancy and hemodynamics are described.
" Chapter 5: Imaging Agonist-induced D2/D3 Receptor Internalization
This extends the concepts of Chapter 4: Dopamine receptor adaptation mechanisms
are presented by (i) proposing a model for receptor internalization and (ii) carrying
out imaging experiments with simultaneous PET/fMRI with a D2/D3 agonist.
" Chapter 6: Effects of Flow Changes on D2/D3 Radiotracer Dynamics
The results from the studies from Chapter 4 and 5 are validated and experimental
results provide evidence that changes in blood flow do not affect radiotracer kinetics.
" Chapter 7: Conclusions and Future Work
A summary of the main chapters in this thesis is given and topics for future work are
proposed and discussed.
26
Chapter 2
Background and Context
The development of functional magnetic resonance imaging (fMRI) techniques has
advanced behavioral and translational neuroscience by extending traditional anatomical brain imaging to include maps of human brain function. Concurrently, positron
emission tomography (PET) has become the gold standard for the imaging of specific
in vivo neurochemistry.
The combination of these two modalities into one scanner
opens up a promising new research area with the potential to provide insight into
neurochemistry of the brain and to understand unexplored questions on the -underlying biochemical nature of the functional signal in MRI. As a background for the
relevant topics of this thesis, this chapter provides an overview of the current scanner
technology of integrated PET/MRI and its challenges. Following that, we summarize
the state of the art for imaging brain function and how PET/MRI can be applied
to answer open questions in brain function. Since this thesis focuses on imaging the
dopamine system, we then give an overview the dopamine receptor system and how it
can be quantified in vivo with PET, together with current issues and open questions
in receptor models.
2.1
Imaging Brain Function
Over the last three decades, there has been an expansion of imaging methods to understand brain anatomy and function. Imaging modalities like Computer Tomography
27
(CT), MRI and PET are routinely used in medical diagnosis, prognosis and therapy
planning. Further development of MR imaging techniques and the discovery of novel
PET radiotracers have extended traditional imaging of morphology to include maps
of brain function and metabolism. In the neurosciences, functional brain imaging has
hence become a tool to explore how the brain processes information. Together with
advances in our understanding of brain anatomy, fMRI methods provide the ability
to record changes in brain activity with a spatial resolution at the submillimeter level
and a temporal resolution of seconds or less. PET has added specificity by delineating maps of neuroreceptors with receptor-specific radiotracers, albeit with lesser
spatial resolution and usually static images. Therefore, neuroimaging studies to date
have expanded our understanding of. the location, propagation and connections of
human brain activity. Still other tools like electroencephalography (EEG) and magnetoencephalography (MEG) have focused on the exact timing of neurotransmission,
although there is a lack of tomographic information.
Functional MRI measures a hemodynamic response that is secondary to underlying neuronal activation. It is understood that the metabolism involved in neuronal
activity causes a local vascular response that can be detected by fMRI. The notion of
"neurovascular coupling" describes the link between a neuronal response and subsequent changes in vascular parameters. The major elements that are generally known
to play a role in neurovascular coupling are illustrated in Figure 2-1. At the most
basic level, neurotransmitter concentrations and receptor occupancy at the synapse
provide the conversion of a chemical signaling link to electrical activity in the neuron.
Neuronal activity is then linked to a vascular response through energy consumption
and metabolism, in which astrocytes are known to play a significant role. Each of
these stages can be probed for through in vivo whole-brain imaging or recording techniques as depicted at the bottom. While some correlations, e.g. between EEG and
MEG recordings and fMRI have provided some insight in vivo, the exact biochemical
changes and links of the chain involved in this coupling process are still not fully
understood yet.
28
&
Energy
consumption
metabolism
MRL:
CMRQ2
P F,-T:
Figure 2-1: Illustration of the chain of biochemical processes in functional brain
activation and the associated imaging modalities that can measure events at various
stages.
2.2
Integrated PET/MR Scanners
Magnetic resonance imaging (MRI) and positron emission tomography (PET) have
become established imaging modalities with a wide range of applications in the clinic
and for biomedical research. Especially for the field of neuroimaging, both technologies have advanced diagnosis and treatment through in vivo whole-brain imaging
techniques.
The strengths of MRI lie not only in its high resolution and excellent
soft tissue contrast for morphological imaging, but also in imaging physiological parameters such as hemodynamics, diffusion or metabolites. However, MRI is limited
in its sensitivity and specificity for biochemical changes. Those features are exactly
what PET excels at: Its sensitivity in the picomolar range (10-" to
is extraordinary compared to MR sensitivity of around 10-
3
10-12
mol/L)
to 10' mol/L [3]. Fur-
thermore, PET has very high biochemical specificity, such that quantification of in
29
vivo biochemistry is possible. This is one of the reasons why it has become the gold
standard for imaging glucose metabolism, with its main clinical application in oncology. Nevertheless, it usually lacks anatomical information and is inferior in its
spatial and temporal resolution compared to MRI. The combination of both MRI
and PET thus integrates complementary imaging techniques, allowing for the simultaneous measurement and mapping of specific neurochemistry, functional physiology
and its underlying anatomy.
Dual-modality scanner technology has been made possible through a number of
technical developments. But the integration of MRI and PET into one scanner poses
many challenges, such as the interaction of the high magnetic field from MRI and the
radiofrequency interference between PET and MRI [4]. A crucial step forward was
thus to replace the photomultiplier tubes, which are sensitive to the magnetic fields
encountered in MRI and are thus not an option, with avalanche photodiodes in the
PET detectors [5, 6, 7]. The development of the first preclinical systems advanced
the technology and showed the feasibility of integrated PET/MRI [8]. Although a
more general development, the design of wide-bore MR scanners allowed to make
room inside the bore for PET detectors and PET camera components for humansized scanners. The first human prototype integrated PET/MR scanner was then
introduced in 2007 and installed at a handful of sites for the first time in the following
few years [9]. This first human prototype PET/MR system is the central technology
that served as a basis for this thesis. It consists of a BrainPET camera that can
be inserted into the standard bore of an MR scanner. In 2011, a commercial fullyintegrated version whole-body PET/MRI system was introduced [10, 11].
2.2.1
Radiofrequency Coils for PET/MRI
While the proof-of-principle of simultaneous imaging with MRI and PET has been
demonstrated through preclinical systems, hardware improvements and developments
are still needed and are active areas of research. Especially on the MR hardware
development side, key improvements are necessary to enable the detection of dualmodality images with high sensitivity. State-of-the-art receive coils for imaging the
30
8-channel coil
32-channel coil
for the BrainPET
for 3T MRI
Figure 2-2: Available RF coils for the detection of MR signals. While commercially
available coils with 32 channels exist for standard MRI systems, an 8-channel coil
exists for PET/MR imaging with the BrainPET scanner.
brain in a PET/MR system include an 8-channel coil (for the BrainPET insert) or a
12-channel coil (for the whole-body PET/MRI) for the detection of the MR signal.
However, for MR-only systems, multichannel array coils with up to 64 channels have
been shown to improve the sensitivity by factors of 2 to 4 and the acquisition speed
of MR images by factors up to 10-fold acceleration in 2D [12, 13].
An example of
currently available coils for the detection of the MR signal is shown in Figure 2-2.
High sensitivity and the use of parallel imaging techniques for faster imaging are
especially important for functional imaging applications. The development of highly
sensitive MR coils capable of parallel imaging for a PET/MR system would thus
enable a number of advanced imaging applications. Because RF coils are located as
close as possible to the object of interest to maximize sensitivity, they are located in
the space before the PET detectors, i.e. inside the field of view of the PET camera.
PET relies on the detection of 511 keV photon pairs that are emitted from an injected
radioactive isotope inside the organism. Hence, any material that is placed in front of
the PET detectors can attenuate or scatter the 511 keV photons, and thus reduce the
sensitivity of PET. This poses a new challenge for designing multichannel array coils
for combined PET and MR imaging: While MR sensitivity is maximized with a high
channel count and thus more materials and highly attenuating metals in the FOV,
PET sensitivity is best when not disturbed by additional objects in the PET FOV. In
31
order to optimize image sensitivity for both PET and MRI at the same time, it is thus
necessary to rethink the design of multichannel array coils for PET compatibility.
In this thesis, we thus address challenges for the design of PET compatible multichannel array coils by proposing and evaluating novel design criteria for PET compatibility. We then design, build and evaluate a 31-channel PET compatible MR coil
for human brain imaging that enables highly sensitive measurements of fMRI data.
The coil is designed to be used in the prototype BrainPET PET/MR system. The
development integrates RF circuit theory, mechanical and materials engineering and
neuroanatomy to develop the array for the needed specifications. In addition, we also
construct two 8-channel PET compatible coils for non-human primate imaging, which
can be used with both the prototype BrainPET as well as the whole-body PET/MR
scanner.
With this unique technology, we are able to achieve extremely fast, high resolution
MRI data together with highly sensitive PET measurements, enabling highly sensitive
functional imaging applications that are described in Chapter 3 of this thesis.
2.3
Dynamics of Brain Function
A relatively unexplored question to date is how neurochemical messaging supports
the aforementioned functional signaling chain (see Fig. 2-1). fMRI signal as a measure of brain function has been previously correlated with other physiological parameters. Advances in our understanding of neurovascular coupling mechanisms have been
brought about by comparing fMRI signal with other techniques that either directly
measure neuronal activity or quantify related metabolic or hemodynamic parameters.
An overview of the correlations of fMRI with other measurements is given in Figure
2-3. The aim in this thesis was thus to use simultaneous PET and fMRI to investigate the link between neuroreceptor-specific activation and its downstream effect on
hemodynamics - a measure of brain function.
While receptor-specific function and occupancy have not been compared for any
neuroreceptor system, the work presented in this thesis focuses on the dopamine
32
Metabolism:
'
CMR02
CmRGiuMR
- -- -----Vasculature:
CBV, CBF
Blood vessel size
Blood oxygenation
Neuronal activity
(non-invasive):
EEG, MEG
as a
ctionNeuronal
'
activity
(invasive):
Local field potentials
Multi-unit activity
Novel approach:
Receptor Occupancy
Figure 2-3: An overview of the correlations of fMRI with other measurements.
system as an example. The dopamine system is described in the following section.
2.4
The Dopamine Receptor System
Dopamine (DA) is a neurotransmitter (a monoamine derived from tyrosine and belongs to the group of catecholaminergic neurotransmitters) that plays an important
role in synaptic transmission and is known to be affected in a number of neurological
and psychiatric brain disorders. Although it has been shown since the 1960's that
dopamine (and other compounds like serotonin or noradrenaline) is a transmitter of
the central nervous system, an understanding of its mechanism of action was still
lacking [14]. Subsequent discoveries of how dopamine is involved in the signal transduction in the nervous system, its action at dopaminergic synapses and involvement
in diseases of motor behavior were awarded with the Nobel Prize in Physiology or
Medicine in 2000. To date, an understanding of dopaminergic transmission has advanced our knowledge of the normal function of the brain and behaviors related to
natural rewards, learning and motivation [15], mood [16] or drug abuse [17]. More-
33
subauttnuft
thalarmus
n--
Figure 2-4: Anatomy of the basal ganglia and their subdivisions in the human brain.
Modified from [20, 21].
over, we know that disturbances in the dopamine signal transduction underlies a
number of neurodegenerative or psychiatric diseases. Hence, the dopamine system
has become a primary target for drug action in a number of neurodegenerative or
psychiatric diseases, including Parkinson's disease [18] or schizophrenia [19].
The distribution of dopaminergic receptors in the brain is highly concentrated to
the basal ganglia, and is thus confined to an anatomically well delineated region at
the center of the brain. The basal ganglia are a collection of gray matter structures
that include the striatum (which comprises the putamen, the caudate nucleus and the
nucleus accumbens), the globus pallidus, the subthalamic nucleus and the substantia
nigra (Figure 2-4). The largest component is the striatum, which receives inputs from
the cerebral cortex and relays information to other nuclei of the basal ganglia (the
globus pallidus). Together, the basal ganglia are associated with a variety of functions
including motor control and reward behavior.
The dopamine receptor system includes five subreceptors, named Dl to D5. All
receptors function as G protein-coupled receptors, i.e. the effect of a ligand binding a
receptors leads to activation via a complex second messenger system. The dopamine
34
receptor system can be divided into two families: (i) Di-like receptors include D1
and D5 and their activation leads to excitation of the target neuron; (ii) D2-like receptors include D2, D3 and D4 and their effects are inhibitory. The neurotransmitter
dopamine can bind to all receptors and thus exhibit excitation or inhibition, but other
ligands, especially drug targets, can more specific to subtypes of receptors such as
D2/D3. In mammals, the D1 receptor is the most abundant in the brain, followed by
the D2 receptor. The concentrations of D3, D4 and D5 receptors are much less.
Due to these characteristics of the dopamine system - its clinical and neuroscientific importance, its anatomical confinement, its well characterized subreceptor system
and the availability of good radiotracers (as explained in the next section) - we chose
the dopamine system as a target to take a first look at characterizing the relationship
between receptor occupancy and function.
2.4.1
PET Imaging of Dopamine Receptors
A number of PET radiotracers that target the dopamine system have been developed
[3]. One of the most successful radiotracers include ["C]raclopride [22] due to its
favorable kinetics, which include fast uptake and relatively fast wash-out. It specifically targets D2/D3 receptors with high affinity and can thus produce specific maps of
D2/D3 receptors in the brain. ["C]raclopride has been well-characterized in previous
imaging studies, so its quantification of PET data can be achieved through kinetic
modeling approaches with a so-called reference tissue model [23].
Quantification of PET Data
Quantification of PET data can be achieved through kinetic modeling. Traditionally,
compartmental models that describe the pharmacokinetic behavior of the radioligand
are applied to fit the acquired PET data to regions of interest in the brain. The
number of parameters that can be reliably estimated is inherently limited by the
amount of information acquired from the PET data. For many radiotracers, a 2-tissue
compartmental model (one compartment that describes the free tracer in tissue, and
35
2-tissue compartmental model
Simplified reference tissue model (SRTM)
K1,
Tissue concentrationB
g
---- --- ------- -----
CT
R
Tssue
ka
K,
C
k2
CP
k4
Free ligand
K
IO
Specific
~
L Plaa
Figure 2-5:
concentration
IW2
Compartmental models for PET kinetic modeling.
Reference
tissue
Left:
General 2-
tissue compartmental model. Right: Simplified reference tissue model applicable for
the radiotracer ["C]raclopride.
another that describes specific binding of the tracer to its target) provides a good fit to
the data (see Figure 2-5). In these cases, it is necessary to take arterial blood samples,
a non-trivial and invasive method, during the imaging session in order to extract the
timecourse of the radiotracer in blood plasma (arterial input function) and use this
as an input to the data in order to estimate the tissue parameters. The kinetics of
["C]raclopride (the exchange between non-specific and specific binding is assumed to
be fast) make it possible to simplify this model to a 1-tissue compartmental model and
use an image-derived reference region as the input to the model [1]. For raclopride,
the cerebellum has been validated as an appropriate reference region since it has only
very small concentrations of dopamine receptors.
The main outcome parameter from kinetic modeling is called "binding potential"
BP. Binding potential is defined as the ratio of target or receptor density Bmax
to radiotracer affinity KD towards the target and is thus a combined measure of
the number of available neuroreceptors and the affinity of the radiotracer to that
receptor.
Practially, in PET imaging studies, the total number of receptors that
is generally measured is not Bm,
but the available receptor pool Bavaji. Receptor
occupancies are calculated as the relative change in BP. Traditionally, one measure
of BP is derived from one entire dynamic PET imaging session, lasting around 1-
36
2 hours. Dynamic PET in this setting refers to the timecourse of dynamic tracer
kinetic uptake, not the dynamic response of the receptor systems themselves. All
kinetic rates that are estimated through PET are thus static quantities. This is in
contrast to fMRI, which looks at hemodynamics due to neuronal activity. Hence,
in the context of drawing comparisons between PET and fMRI, there have been no
standard measures developed for comparing signals. However, we address this issue
by introducing dynamic measures of occupancy in Chapter 4.
Binding potential is a weighted average of receptor density and tracer affinity over
time. With PET alone, it is impossible to tell the difference between a change in the
occupancy of receptors, a change in density or a change in affinity. In order to deal
with this issue, it is often assumed that the total receptor density stays constant,
or that KD stays constant - an assumption that is not always accurate. In the next
section, we discuss how this inherent limitation in PET may be circumvented by using
fMRI and how this may help in decoupling the occupancy/density of receptors from
the affinity of the ligand.
Open Questions in PET Quantification
One question that has persisted through decades of PET literature is whether changes
in blood flow could influence the delivery and washout of an injected receptor-specific
radiotracer, and thus affect quantification results in PET. Kinetic modeling in PET
assumes steady-state conditions, which includes a constant value for blood flow. But
physiology can change over the course of ther PET scan, which usually lasts at least
1-2 hours for a full scan session. Although the topic has been investigated, mainly
through simulations, the reported results have been controversial as they depend on
varying assumptions [24, 251.
Pharmacological challenges (i.e. the dose at which a particular drug is given causes
a pharmacological effect) are known to cause changes in physiology. Some drugs that
target the dopamine system, such as amphetamine, can produce flow changes up
to 50% or more. In this context, it has thus been unclear whether specific binding
measures (as determined by an observed change in binding potential from PET) could
37
be confounded by a change in blood flow.
MR techniques such as arterial spin labeling can measure absolute values of perfusion and thus monitor cerebral blood flow measures over the timecourse of an experiment. Being able to acquire PET and MRI at the same time provide the unprecedented ability to experimentally determine the influence of changes in blood flow on
PET radiotracer kinetics, with the results shown in Chapter 6. Our studies focus
on radiotracers that target the dopamine system, but we present an imaging method
that can be generally applied to evaluate the effect of flow on PET time activity
curves for any radiotracer by inducing controlled changes in blood flow.
2.5
Receptor Theory
In combining PET and fMRI measures, our goal is to learn more about the dynamic
neurochemical events occurring at the synapse and how it relates to downstream
function. In this section, we introduce some general notions and models of receptor
theory that formed an integral part in devising the PET/fMRI studies carried out in
Chapters 4 and 5 of this thesis.
2.5.1
The Classical Receptor Occupancy Model
PET imaging has been used to measure acute fluctuations in the synaptic concentration of neurotransmitters. The principle underlying such studies is the competition
between neurotransmitter and radiotracer for the binding to neuroreceptors.
It is
assumed that changes in the concentration of neurotransmitters or exogenous ligands
translate to changes in receptor occupancy that can be detected as a change in binding
potential with PET.
The classical receptor occupancy model describes the action of agonists and competitive antagonists at receptors and the resulting observed effect. Studying a correlation between function and occupancy has been proposed as early as 1937 by Clark
et al. [26]. The basis of the description is mass-action kinetics, such that drug action is proportional to the receptor occupied at equilibrium.
38
The main modeling
Baseline
0
4
Stimulated
+PET
PET specific
binding large
in D2-rich
signal
regions
Baseline
dopamine
~20%
Small inhibitory signal
due to baseline DA
Inhibition blocked due to antagonist
(b)
(a)
ITjD2 receptor
N
[' 1C]raclopride
Dopamine (DA)
Antagonist
Figure 2-6: The classical occupancy model for the case of a D2 antagonist exposure.
(a) At baseline, approximately 20% of D2 receptors are occupied by endogenous
dopamine. The corresponding PET specific binding is large in regions where D2
receptor concentrations are high. (b) With an antagonist challenge that targets D2
receptors, the antagonist is now competing with the basal dopamine levels and the
PET radiotracer. At high doses, the majority of all receptors is bound by antagonist.
The latter does not elicit a functional response on its own but causes a blockage of the
initial inhibition due to endogenous dopamine. The PET signal in D2-rich regions is
decreased in this case.
assumptions in this approach are:
i. Receptors can have two states: They can either be bound by a ligand or available
for binding.
ii. The magnitude of the response due to binding is directly proportional to the
concentration of the drug bound.
iii. The maximum response is elicited with all receptors occupied at steady state by
a full agonist.
Figure 2-6 illustrates the main ideas behind the classical occupancy model for the
case of a D2 antagonist.
Even though the relationship between drug occupancy and its functional effects
39
Baseline
Stimulated
t:
Internalized
ta
40PET
signal
PET specific
binding large
+PET
signal
in D2-rich
regions
Baseline
dopamine
Downstream
inhibitory
functional
response
(b)
(a)
-
'j?
2 receptor
N7 Dopamine
[f"Cilraclopride
(c)
TinternalizedT
receptors
*'Agonist
Figure 2-7: The internalization model for the case of a D2 agonist exposure. (a) At
baseline, 20 % of receptors are occupied by endogenous dopamine and thus the PET
radiotracer gives a high signal in D2-rich regions. (b) With a D2 agonist challenge,
both endogenous dopamine and the PET radiotracer binding is decreased. The receptors occupied by agonist elicit a downstream functional response that is inhibitory. (c)
After a certain amount of time or exposure of the agonist drug, receptors can become
internalized. This serves as a desensitization mechanism that limits the downstream
functional response.
has been suggested a long time ago, there have been no investigations to validate
or confirm the relationship of specific drug exposure to functional measures until
now. Hence, our application of PET/fMRI in Chapter 4 sought to characterize this
relationship.
For this purpose, we initially consider a scenario that leads to the
physiologically simplest results and choose to investigate the effects of an antagonist
that is specific to D2/D3 receptors and characterize its consequences of binding in
space, time and with dose.
2.5.2
The Internalization Model
More recently, it has become clear that dopamine receptors are not static entities
but rather are dynamically regulated through internalization mechanisms.
Hence,
extensions and alternatives to the classical occupancy model have been proposed
that include this feature and its parameters.
40
The internalization model makes the
following assumptions:
i. Receptors can not only be bound, or available for binding, but can also internalize
by endocytosis to intracellular space.
ii. If internalized, receptors can change their affinity for binding and are not functionally active.
Figure 2-7 shows a schematic of the internalization model and the notion of
agonist-induced receptor internalization at the postsynaptic membrane.
Measurements of synaptic adaptation mechanisms, such as receptor internalization
or externalization, can be made in cell culture, and a few, very recent studies have
used post-mortem tissue from brain slices [27] to examine alterations in dopamine
transporter endocytosis. However, the overall story is complicated by the lack of
techniques to measure internalization in vivo. It is believed that receptors can be
modulated during strong reward behaviors or exposure to pharmacological challenges
with agonists. However, there are no non-invasive ways to estimate these dynamic
receptor behaviors in vivo.
Our hypothesis for the experiments of Chapter 5 is that both function and measures of occupancy by PET can be modulated by receptor internalization, and that
simultaneous PET/fMRI can provide the first in vivo measurement of receptor adaptation mechanisms. Having characterized the relationship between PET and fMRI
with a D2/D3 antagonist that does not cause internalization, we investigate the same
target D2/D3 with an agent that is known to cause internalization in vitro. The
details of this study are described in Chapter 5.
41
42
Chapter 3
A PET Compatible 31-channel MR
Brain Array Coil
3.1
Introduction
Imaging technology for simultaneous Magnetic Resonance Imaging (MRI) and Positron
Emission Tomography (PET) has been developed in the last decade for small-animal
imaging [5, 7] and more recently for human imaging applications [9]. Integrated simultaneous PET and MRI allow a unique combination of morphology and function
by combining high resolution anatomical MR, hemodynamic changes from functional
MRI (fMRI), and specific functional molecular imaging from PET [28]. The potential and the growing interest for studies with simultaneous positron emission tomography/magnetic resonance (PET/MR) imaging [29 calls for advanced detection
capabilities with integrated PET/MR scanners, including new radiofrequency (RF)
coils that improve sensitivity and parallel imaging.
From the MR perspective, large coil arrays for highly parallel imaging with MRonly scanners have demonstrated increased image sensitivity and acquisition speed
compared to single-channel coils [2, 13, 30]. The development of parallel imaging has
allowed accelerated image acquisition and advanced reconstruction techniques with
array coils [31, 32]. For imaging of the brain, multi-channel array coils with up to
32 channels (or 64 channels for head/neck) are commercially available and arrays
43
with higher numbers of channels have been demonstrated [12]. However, the molar
sensitivity in the functional/molecular domains of MRI is still orders of magnitude
lower compared to PET [5] and hence there is a continuing drive for higher sensitivity.
In PET/MR imaging, the RF coils for the MR signal detection are located within
the PET field-of-view (FOV) and thus degrade PET image quality through additional
attenuation and scatter. These phenomena result in decreased sensitivity in PET
images (i.e. increased image noise or longer PET acquisition times to make up for the
counts absorbed by the MR array) as well as in the introduction of artifacts if the coil
attenuation is not accounted for [33]. Since standard RF coils are not PET-optimized
and contain highly attenuating materials at non-uniformly distributed points in the
PET FOV, artifacts on PET images due to RF coils can be a severe problem, especially
for multi-channel arrays. While it is necessary to correct for the attenuation of the
RF coil sitting inside the PET FOV, such corrections can be challenging and artifacts
cannot always be fully removed [34, 35, 36, 37]. Thus, it is crucial to minimize the
attenuation of photons for best PET image quality.
The design and development of multi-channel array coils for the use in an integrated PET/MR system poses new challenges and requires a different design approach
compared to conventional MR arrays. For brain imaging with simultaneous PET/MR
systems, MR arrays with a maximum of 12 channels are currently available. Higher
number of channels would increase MR image quality, but would also be expected to
increase the attenuation of 511 keV PET photons. Hence, it is especially important
for the construction of highly parallel arrays to adopt a targeted design approach that
takes into account PET compatibility while maximizing the MR detection capability.
Other considerations are important, such as highly constrained space due to the small
geometry of some current generation PET/MR brain scanners.
In this chapter, we present the design and performance of a 31-channel PETcompatible MR brain array coil for simultaneous PET/MR. Detailed design considerations for the challenges with integrated PET/MR systems are addressed. These
include choices for electrical components and their arrangement within the PET scanner bore. According to these criteria, the array coil is designed in a sparse config44
uration of 511 keV absorptive materials and aims to optimize the arrangement of
components, such as preamplifiers and cables. The performance of the coil is evaluated according to MR SNR and parallel imaging measures as well as the principal
factors for sensitive and artifact-free PET images: attenuation at 511 keV and the
performance of coil attenuation corrections.
3.2
3.2.1
Methods
Existing Hardware Components
The PET-compatible array coil was designed for a prototype simultaneous PET/MR
scanner, composed of a 3 Tesla MRI scanner (MAGNETOM Trio, Tim system,
Siemens AG, Healthcare Sector, Erlangen Germany) and a PET camera insert (BrainPET, Siemens AG, Healthcare Sector, Erlangen Germany) with magnetic field insensitive avalanche photodiodes as scintillation detectors. The standard coil provided
for this scanner includes an 8-channel receive coil with a local circularly polarized
birdcage transmit coil. The PET camera has inner and outer physical diameters of
35 and 60 cm respectively, with an axial PET FOV of 19.25 cm.
3.2.2
PET Compatibility of Coil Design
The 31-channel coil was designed to minimize the attenuation and scatter of 511 keV
photons and thus be PET-compatible. To assess the attenuation properties of coil
components inside the PET FOV, CT scans of components were taken with a clinical
CT scanner (Discovery STE, 'GE Healthcare, Chalfont St. Giles, UK). Hounsfield
units (HU) from the CT scans were converted into linear attenuation coefficients at
511 keV PET energies via a bilinear transformation [38]. Additionally, estimates of
attenuation at 511 keV were calculated with the Beer-Lambert law, using documented
values of linear attenuation coefficients [39].
Coil components were evaluated for their tradeoff between PET compatibility
(attenuation properties), MR SNR properties and necessary practical considerations
45
0.
02
-
250
-18aWg
F
wg
re
wir2
250
-1.2mm
-2.2mm
dia. coax
dia. coax
-200
i10%
-thin 40mm coax
250
-thin
100mm coax
200200 - -thick 100mm
0- ---thlckl00m
ia.150
---Al foil
150
$100
100
C cc100
0 50
'.
50
U 50
0
50
100 I
0
50
100
0
50
100
Depth into phantom (mm)
Depth into phantom (mm)
Depth into phantom (mm)
........ ...... .............. ~~~~~~~~~~~~~~~~~~~~~~~~.......
... ..... ............. .......................................-....-.......................... ...... .....................
18 awg Cu wire
1.2 mm diameter
silver coax cable
Remove from the
PET FOV
Thickness < 3 mm
Figure 3-1: 511 keV attenuation map of coil components evaluated for the design
of the 31-channel PET/MR array coil (1st row), showing high attenuation of the
preamplifiers and thick metal or plastic components. Each electrical component was
tested for its MR SNR in a single-loop configuration on a water phantom (2nd row).
The final design choices reflect the goal to reduce attenuation but preserve SNR for
MRI (3rd row).
(e.g. mechanical stability). In Figure 3-1, a range of coil components and their properties are displayed. Various wire thicknesses (16 to 24 awg), and different materials
(copper and aluminum) for conducting loops were investigated. Also, two types of
coaxial cables with a diameter of 2.2 and 1.2 mm were compared. The impact of using longer coaxial cables to connect conducting loops and preamplifiers was assessed
in order to consider the removal of the preamplifiers from the PET FOV. Finally,
the attenuation of preamplifiers on circuit boards (Figure 3-1, column 3) and plastic
housing samples with various thicknesses (2 mm to 12 mm) was measured.
To test the performance of components for MR imaging, we built single test loop
coils for each coil component option and compared SNR performance. Additionally,
silver-based conductive epoxy was evaluated as an alternative to solder for connecting
components because of its lower density that would limit attenuation.
The final
design choice was based on maximizing sensitivity for the two imaging modalities,
while minimizing loss of signal.
46
Figure 3-2: The 31-channel PET/MR array coil is divided into an anterior and posterior part, with 15 and 16 channels respectively. (a) Side view of the array, with
overlapping loop design and preamplifiers outside the FOV of the PET camera. (b)
Fit of the receive coil with cover into the local transmit coil. (c) CAD design of the
coil with an anterior and posterior part and a magnification of the mechanical rim
system that allows for an overlapped loop design between the two coil parts.
3.2.3
Array Coil Design
The 31-channel receive (Rx) array coil was designed to be placed inside a commercially
available local head transmit coil of the BrainPET camera (Figure 3-2).
The Rx
helmet former was designed with a 3D CAD program (Sketchup, Google, Mountain
View, CA, USA). As the transmit (Tx) coil accommodates space for a cylinder with
a diameter of 28.5 cm, the Rx was designed with a maximum diameter of 28.3 cm
to achieve a tight fit into the Tx. Since the Tx coil consists of a splitable housing,
we incorporated a split coil former into the design of the Rx for easy patient access
(Figure 3-2-c).
The overall shape and size of the helmet curvature was based on an aligned head
MRI and scaled to the 95th percentile of male head circumferences.
In order not
to obstruct the subjects view and to be able to utilize visual stimulus equipment,
cutouts for the eyes were incorporated. The shape of the posterior housing follows
natural head contours to support the patients neck and improve sensitivity towards
the occipital lobe and cerebellum.
The entire coil was closed with a thin plastic
housing to provide coverage of electronics.
The final design of the array coil was
printed from polycarbonate plastic on a three-dimensional printer (Fortus 400mc,
Stratasys, Eden Prairie, MN, USA).
47
3.2.4
Coil Construction
The layout of the array with overlapped circular coil elements was established with
a hexagonal and pentagonal tiling pattern (6). The pattern was engraved onto the
helmet with the CAD program, in addition to standoffs for preamplifiers and cable
trap holders. The coil loop diameters were derived from the size of the pentagon or
hexagon tiles, resulting in a diameter of 75 mm and 90 mm respectively. The array was
designed to consist of 31 loop elements, plus 1 volume receive channel associated with
the Tx. Since the scanner has 32 available receive channels in total, this design used
the maximum number of channels possible. The anterior coil segment comprised 15
loop coils, where 13 loops correspond to hexagons and 2 loops to pentagon positions.
In the posterior segment, we incorporated 16 loops (13 at hexagon and 3 at pentagons
locations).
Based on our component evaluation, we used an 18-awg thick tin-plated copper
wire (Figure 3-1). Bridges were bent into the wires to allow conductor crossings without contact. Each loop was split into three segments of equal lengths, interleaved with
components mounted on small FR4 circuit boards (Figure 3-3). A split into three wire
segments (instead of six segments that may be used for coils of this size) reduces the
total number of components on the coil, further minimizing attenuation. The circuit
boards were produced with a rapid prototyping circuit router (T-Tech-7000, T-Tech,
Inc., Norcross, GA, USA) and their size was held as small as possible. A variable
capacitor C1 was used for fine-tuning the loop resonance to the resonant frequency
(123.25 MHz) of the scanner. A fuse F (570 mA rating) served as passive protection
against potentially large currents induced by the transmit system. The array coil was
geometrically decoupled using an overlapped coil element layout. Adjacent loops between the anterior/posterior housing were overlapped with a mechanical rim system,
with loops in the anterior segment bent over the housings rim structure to achieve
critical overlap with its nearest neighbors from the posterior housing segment (Figure
3-1-c). In order to utilize a sparse housing structure around the eyes, the two eye
loops were decoupled by a shared impedance.
48
C,
F
Phase shifter
L,
Coil
90 MM
coax
C4
C3
C
L2
PyC6 MPreamp
XD 0_
C2
Figure 3-3: Circuit schematic for coil elements with a long coaxial cable that allows
placement of the preamplifier outside the PET FOV. In order to maintain a A/2 phase
shift, a pi phase shifter is located before the preamplifier chain.
Each coil elements output circuitry comprised a capacitive voltage divider (C3,
C4) to match the loop impedance to the preamplifier as shown in Figure 3-3. An active
detuning circuit across the matching capacitor C4 was composed of a variable inductor
Li (hand-wound) and a PIN diode D (Macom, MA4P4002B-402, Lowell, MA, USA).
During transmit, the PIN diode is forward biased and the activated parallel LC4
circuit provides a high impedance within the coil loop. Prior to mounting the output
circuitry to the coil helmet, these traps were pre-tuned to the Larmor frequency using
a small sniffer probe for S11 monitoring. After populating the array coil, the active
detuning traps were fine-tuned using a double-probe slightly coupled into the loop
under test and adjusted via a S21 measurement.
Preamplifier decoupling was achieved by transforming the low impedance of the
preamplifier to a short across the PIN diode using a A/2 phase shift. This activates the
detuning trap to transform a serial high impedance into the coil loop. Thus, current
flow is minimized and inductive coupling to neighboring coils is reduced. Since all
preamplifiers are placed outside the PET FOV, the distance of the coil element and
preamplifiers varies from 10 mm to 100 mm. We compensated for these distances
with a pi phase shifter (Figure 3-3) located between coaxial cable and preamplifier,
while maintaining the desired A/2 phase shift needed to transform the preamplifier
input impedance to a short across the diode.
49
Figure 3-4: The 31-channel coil design allowed bundled output cables to be eliminated
from the PET FOV. (a) 31-channel coil with output cables. (b) Superior end of the
31-channel coil inside the local transmit coil with output cables leaving the coil. Due
to the arrangement of preamplifiers and cable plugs, three Rx and one Tx output
cable leave the coil at the head end, and one Rx output cable leaves the coil at the
neck end.
Due to space constraints and preamplifier placement outside the PET FOV, we
placed 23 preamplifiers at the superior end of the coil, with the remaining eight
placed at the inferior end of the helmet. We aligned all preamplifiers in z-direction to
the magnetic field BO to prevent Hall effect issues on the drain current of the Field
Effect Transistor [40]. Output cables from the preamplifiers were bundled together
with eight outputs into one cable trap each to prevent common mode currents and
interference during RF transmit.
During the design layout, special attention was paid to prevent output cables from
passing through the PET FOV. Three cable outputs with eight channels each were
positioned to leave the coil towards the back of the scanner. One of these plugs also
included the transmission lines for receive and transmit paths for the local birdcage
Tx. The eight channels with preamplifiers positioned at the inferior end of the coil
were bundled into one plug exiting at the neck end of the housing.
These plug
arrangements allowed a complete elimination of output cables from the PET FOV
(see Figure 3-4).
3.2.5
Coil Bench Tests
Bench testing during construction was carried out with a custom-made coil-plug simulator that provides 10 V for the preamplifiers and has the ability to manually apply
50
a bias current (100 mA) to each PIN diode on the array elements.
Bench mea-
surements verified element tuning, active detuning, nearest-neighbor coupling and
preamplifier decoupling for each coil element. The ratio of unloaded-to-loaded quality factor (Qu/QL) was measured for each coil element size. The Qu/QL-ratios were
obtained for a loop with no coaxial cable or preamplifier and were tested with the
loops external to the array assembly. Initially, all loops but the element under test
were detuned with a DC bias. The S21 measure of a double-probe was used to adjust active detuning by maximizing the difference between S21 measures in the tuned
and unturned state. Geometrical decoupling of adjacent loop pairs was carried out
by a direct S21 measurement, in which coaxial cables were connected from the each
preamplifier input socket of two adjacent loops to the network analyzer while all other
elements were kept actively detuned. The overlap percentage of each coil pair was
then adjusted empirically by bending the overlapped part of the wire of each loop to
achieve isolation between -13 dB and -18 dB.
We measured preamplifier decoupling of a given loop with all other loops detuned.
Preamplifier decoupling was measured as the change in S21, while the coil output
coaxial cable was terminated with two different match conditions [41]: first, with a
powered preamplifier (low impedance termination) and second, with a power matched
source impedance (i.e. 50 Q). For each loop in turn, the bias voltage was turned off
to measure preamplifier decoupling.
For the scanner setup with the BrainPET insert, a commercially available PEToptimized local transmit coil (Siemens AG, Healthcare Sector, Erlangen, Germany)
encloses the 31-channel Rx. Due to the close proximity of the two coils, we observed
both a shift and a split in resonant frequency for the transmit coil when the detuned
array coil was placed inside the birdcage. This was attributed to residual coupling
between the two coils and we had to carefully retune and restore circularly polarized
symmetry of the transmit coil. This would be a lesser issue in whole body PET/MR
applications, where the body Tx is located further away from the Rx array.
51
3.2.6
MRI Data Acquisition and Analysis
Images were acquired with a head-shaped phantom filled with physiological saline and
gadolinium. SNR measurements were taken with proton density weighted gradient
echo images (repetition time TR = 30 ms, echo time TE = 6 ms, flip angle FA
= 30, slice thickness = 7 mm, matrix = 192 x 192 pixels, FOVMR = 235 x 235
mm2 , bandwidth BW
=
200 Hz/pixel). SNR maps were computed with the noise
covariance weighted root sum-of-squares from individual channel images [42]. In
order to assess noise amplification in SENSE reconstructions, g-factor maps [31] were
calculated from the complex coil sensitivity maps and the noise correlation matrix
using the same imaging sequence but with an adjusted FOV (210 x 210 mm 2 ). SNR
and g-factor measures were compared to those from the 8-channel head coil dedicated
for the PET/MR system. Additionally, comparisons were made to a commercially
available 32-channel head coil in the same MR scanner but with the PET insert taken
out and a body coil used for transmit. Human PET/MR images were acquired for
comparisons between the 8-channel and 31-channel coil. Ti-weighted MR images
were acquired with the multiecho MPRAGE sequence (MEMPRAGE [43]) at 1mm
isotropic resolution and with TR = 2530 ms, TE = 1.64 ms, TI = 1200 ms, FA = 7
and BW = 651 Hz/pixel.
3.2.7
CT Data Acquisition and Processing
CT scans of coil components and the final coil design were acquired with a clinical CT
scanner (GE Discovery STE). Images were acquired in helical scanning mode with
x-ray tube current-time-product = 580 mAs, maximum voltage kVp = 140 kV, spiral
pitch factor = 0.98, slice thickness = 0.625 mm, matrix = 512 x 512 pixels. The FOV
was adjusted to the size of the coil in each case.
To evaluate the attenuation of each coil, CT scans were processed for an estimate
of the coil attenuation at 511 keV in sinogram space: First, CT images in Hounsfield
Units (HU) were converted to an attenuation map in units of linear attenuation coefficients at 511 keV (in cm') by a bilinear transformation [38]. The Radon transform
52
was then applied to bring the data into sinogram space of angle versus radius. A
sinogram of attenuation correction factors (ACF) was then obtained by evaluating
the exponential of the Radon transform result.
3.2.8
Derivation of Attenuation Map and Attenuation Correction
An attenuation map of the coil was created from the CT images of the 31-channel
array. Streak artifacts on the CT images due to highly attenuating metal components
were removed manually before extrapolating to 511 keV energies. In order to obtain
an accurate positioning of the attenuation map in the PET FOV, four fillable fiducial
markers were placed at fixed locations on the coil. The markers were filled with
contrast agent during the CT scan and ['IF]-solution for a PET scan. The fiducial
markers from the CT were then registered to the PET to place the coil attenuation
map in an accurate position in the PET FOV. Attenuation maps were blurred to
match PET resolution using a 3 mm FWHM Gaussian kernel and noise reduction.
3.2.9
PET Data Acquisition and Reconstruction
To test the CT-derived coil attenuation map, PET data were acquired with an [' 8 F]solution filled phantom ( 0.4 mCi in 1.9 L). The phantom was placed in a fixed position
on the patient bed and scanned in three configurations in the following order: with
(i) the 31-channel or (ii) the 8-channel coils in place, or (iii) without any coil in the
PET FOV. To compensate for the reduced activity over time (for the 2nd and 3rd
scans), scan durations were adjusted to match the same activity levels in each scan:
(i) 12.6 min, (ii) 17.3 min and (iii) 20 min for the three configurations above. An
attenuation map of the phantom was derived from an initial PET image that was first
reconstructed without attenuation correction and then thresholded to create a binary
mask. The linear attenuation coefficient corresponding to water (0.096 cm- 1) was
assigned to all the voxels inside the mask. The same attenuation map was used for all
reconstructions, co-registered to each initial PET image to account for small changes
53
in position.
PET data were reconstructed with a standard 3D ordinary Poisson
expectation maximization algorithm (OP-EM, 32 iterations), using both prompt and
variance-reduced random coincidence events [44] as well as normalization, scatter and
attenuation sinograms [45]. The reconstructed volume consisted of a 256 x 256 x 153
matrix with isotropic voxel size of 1.25 mm. Final images were smoothed with a 3 mm
FWHM Gaussian kernel (8 mm for the computation of difference images). Images
were reconstructed either including or ignoring the coil attenuation maps. Normalized
difference images were computed as the difference of phantom images with coil and
the reference image (no coil in PET FOV) and were divided by the reference image
and masked to compute relative change inside the phantom.
Human PET (and MR) data was acquired at 73 and 116 min after injection of
[1 8 F]FDG (5 mCi injected activity), first with the 31-channel and then with the 8channel coil in place. To match the total activity (and thus SNR) for the two scans,
the duration of the PET scan was adjusted to 15 min and 20 min for the 31-channel
and 8-channel coil acquisitions, respectively. A pseudo-CT attenuation map of the
head was derived from an anatomical MEMPRAGE scan (see MRI data acquisition
and analysis) and used for attenuation correction of the head [46]. Otherwise, PET
data were reconstructed as described above. All studies were carried out in compliance
with the Institutional Review Board at Massachusetts General Hospital.
3.3
3.3.1
Results
Coil Component Evaluation
The material PET/MR performance trade-offs and final design choices for the 31channel PET/MR coil are summarized in Figure 3-1. For the loop wire choices (Figure
3-1, column 1), SNR values decreased with wire diameter. Loops built with conductive
epoxy showed a lower quality factor and connections were not strong enough for a
stable array, so that standard solder was used in all cases. Relative to a standard
16 awg wire, the 18 awg and 24 awg wire showed a 6.2% and 22% reduction in SNR
54
respectively. The aluminum foil loop showed a 25% decrease in SNR. Calculations
from the Beer-Lambert law showed an estimated maximum beam attenuation of 9%
from a 16 awg wire, 7% from an 18 awg wire and 3% from a 24 awg wire. Based on
these measurements, the 18 awg wire was chosen for the final design as a compromise
to preserve SNR and limit attenuation of 511 keV photons. The second column
in Figure 3-1 compares the performance of two coaxial cables of different thickness
that connect the coil and preamplifier. The graph illustrates the SNR performance
as a function of depth into the phantom. Both coaxial cables showed similar SNR
performance, such that the thinner 1.2 mm silver coaxial cable was chosen for the
final design. With the CT scan and attenuation map (Figure 3-1, column 3), we
showed that the preamplifiers are the most attenuating components of the coil. If the
preamplifier is moved away from the PET FOV, i.e. the distance between loop and
preamplifier is increased from 40 mm to 100 mm, an average loss of 8% in SNR is
shown. This small loss was deemed acceptable in light of the reduced attenuation from
preamplifiers. A distance of 100 mm was the largest length of coaxial cable needed
for all preamplifiers to be placed outside the PET FOV. The plastic housing (Figure
3-1, column 4) was chosen to have minimal thickness wherever possible, taking into
account mechanical stability of the housing. The shell was thus chosen to have a
thickness of less than 2.7 mm on a flat surface, which corresponds to an attenuation
of 3% according to the Beer-Lambert law.
3.3.2
MR Performance of Array Coil
The single loop unloaded/loaded quality factor for a 90 mm loop was measured as
230/45, ( 5.1 ratio) for a single coil loop. In Figure 3-5, the noise correlation matrix
of the 31-channel PET/MR coil is shown in comparison with the 8-channel PET/MR
and the 31-channel MR-only phased array coil. The noise correlation ranges from
1% to 65% and averages with 21.4%, excluding the diagonal elements. Geometric
decoupling on the bench achieved an average attenuation of -15.2 dB, with minimum
attenuation of -13.0 dB and maximum of -22.2 dB. Preamplifier decoupling allowed
for an additional -25 dB attenuation. To compare, the 8-channel PET/MR and 3255
8ch
8ch PET/MR
PET/MR
PET/MR
31ch
3lch PET/MR
32ch MR-only
32ch
MR-only
1
0.8
0.6
0.4
0.2
COF = 11.8%
COF = 21.4%
COF = 12.4%
0
Figure 3-5: Noise correlation matrices for the 8-channel PET/MR, the 31-channel
PET/MR and the 32-channel MR-only array. The average noise correlation is 11.8%,
21.4% and 12.4% respectively.
channel MR-only array achieve an average noise correlation of 11.8% (max: 20%,
min: 1%) and 12.4% (max: 44%, min: 0.1%) respectively.
The MR SNR maps of a representative sagittal and axial slice for each of the three
coils are shown in Figure 3-6. The gain in SNR, relative to the 8-channel array, was
calculated for regions of interests with an area of 18 x 18 mm 2 placed at locations
(a) to (f) (Figure 3-6). The highest SNR values of the 31-channel array are found at
the periphery, close to the coil elements. In the sagittal slice, SNR gains were 2.3 (a,
frontal region), 1.6 (b, center) and 1.7 (c, posterior region). In the axial slice, SNR
gains were 2.2 (d, anterior region), 2.1 (f, posterior region) and 1.8 (e, center).
Figure 3-7 (top) shows inverted g-factor maps for an axial slice (same slice position
as for the SNR maps) for the three coils in comparison. Four one-dimensional accelerations in the anterior-posterior direction with acceleration factor R = 2 to R = 5 and
four two-dimensional accelerations with additional right-left directions are compared
for the three coils. For the 8-channel array, the g-factor increases rapidly for R >
2, translating into increased image noise amplification, whereas the 31-channel array
consistently shows an overall lower g-factor. Comparison to the 32-channel MR-only
array demonstrates that the 31-channel PET-optimized array coil provides similar
parallel imaging capabilities.
Figure 3-7 (bottom) shows corresponding histograms formed from the 1/g-factor
56
31ch PET/MR
8ch PET/MR
32ch MR-only
SNR
800
700
600
500
400
300
200
100
0
Figure 3-6: Signal to noise maps of a human head-shaped water phantom (sagittal
and axial slice) for the 8-channel PET/MR, the 31-channel PET/MR and a standard
32-channel MR-only array. The gain in SNR compared to the 8ch PET/MR coil for
the ROIs shown is: (a) 2.3, (b) 1.6, (c) 1.7, (d) 2.2, (e) 1.8, (f) 2.1. The SNR of the
31-channel PET/MR coil shows a large improvement compared to the 8-channel coil,
and shows no overall SNR loss compared to the 32-channel MR-only array.
maps above, with histogram entries binned to 20 entries. For accelerations of R
2 and R = 3, the 31-channel outperforms the 8-channel array with the majority of
voxel distributions near 1/g
=
1. Higher accelerations of R
=
4 and R = 5 results
in a complete loss of high 1/g factors for the 8-channel array, whereas the 31-channel
array shows an improved performance with a similar 1/g-factor distribution to that
of a conventional 32-channel array.
3.3.3
PET Performance/Attenuation of Array Coil
Figure 3-8 shows transverse slices of the CT images for each of the three coils at the
center z = 0 and 7.5 cm from the center in each direction (z =
7.5 cm). All slices
lie within the PET FOV. The CT images show that there are more components on a
57
R=2x2
R=2x3
R=3x3
-
R=5
---
R=3x4
a:1/g
1-
--
0.
0.8
0.6
cW
0.
04
0.75
2 0.25
2
1
0.75
0.5 0.25 7 --
0
-
0.5
1/g factor
-
110
--
0.5
1/g factor
1
0
0.5
1/g factor
10
0.5
1/g factor
1
0
0.5
1/g factor
10
0.5
1/g factor
10
0.5
1/gfactor
1
0
0.5
1/gfactor
1
Figure 3-7: Top: 1/g factor maps for the 8-channel PET/MR, the 31-channel
PET/MR and the 32-channel MR-only array for acceleration factors R in 1D and
2D. Bottom: Corresponding histograms of 1/g factor maps for the three coils in
comparison. Compared to the 8-channel PET/MR coil, the 31-channel PET/MR
coil shows much higher 1/g factors values at all acceleration stages. Despite the design constraints for PET compatibility, the 31-channel PET-optimized array performs
similarly to an MR-only array with an equivalent number of channels.
standard 32-channel MR-only coil that would attenuate photons. The main absorbing components are densely packed electrical components from the preamplifiers and
highly attenuating cables, cable traps and cable plugs.
Figure 3-9 shows sinograms of attenuation correction factors (ACF) for the corresponding slices of the CT images shown in Figure 3-8, with the mean ACF displayed
for the each slice (z = 0 and z = 7.5 cm). At z
=
0, the 31-channel PET/MR at-
tenuates 12% more with respect to the 8-channel PET/MR, whereas the 32-channel
MR-only attenuates 77% more. Similar results are obtained in both directions away
from the center. The overall mean ACF for a total axial length of 20 cm, with 40
58
H.U.
2000
E
0
LO
N
1000
E
N
0
E
N
-1000
Figure 3-8: Axial CT scans for z-positions -7.5 cm, 0 cm and 7.5 cm of the 8-channel
PET/MR, the 31-channel PET/MR and a standard MR-only 32-channel coil. While
the 8-channel coil shows the least material in the PET FOV, the 32-channel MR-only
coil shows attenuating coil components, such as the plugs, preamplifiers and cable
traps. Compared to these two coils, the 31-channel PET/MR coil shows more metal
due to the higher number of coils but does not have large attenuating coil components
in the PET FOV.
slices taken into account, spaced at 0.5 cm each is 1.13 for the 8-channel PET/MR
coil, 1.33 for the 31-channel PET/MR coil and 2.40 for the 32-channel MR-only coil.
Hence, we estimate that the 31-channel PET/MR coil attenuates 18% more on average and the 32-channel MR-only coil attenuates 112% times more, compared to the
8-channel PET/MR coil.
The artifact load of the 31-channel coil and the loss in photon counts due to
the coil are demonstrated in Figure 3-10 with PET phantom data.
Images from
an [18F]-solution-filled cylindrical phantom with the 8-channel or 31-channel coil in
place are compared to a reference dataset without any coils present in the PET FOV.
59
31ch PET/MR
8ch PET/MR
32ch MR-only
0
E0
ACF
-4
~50
LO
Q)100
N
150
0
E
3
050
100
N
<150
2
0
E0
50
LC)
0)
N
<150
-10
10
0
radial position
10
0
-10
radial position
-10
10
0
radial position
1
Figure 3-9: Sinograms calculated from a forward projection of the CT scans (shown
in Figure 3-8) for z-positions -7.5 cm, 0 cm and 7.5 cm. The mean attenuation
correction factors (ACF) of the 31-channel PET/MR coil, the 8-channel PET/MR
and a standard MR-only 32-channel coil are shown for each slice. This estimation
shows that while the 31-channel PET/MR would only attenuate between 18-54% of
the 511 keV photons (similar to the 8-channel PET/MR coil), an MR-only standard
32-channel coil would attenuate much more photons (up to 218%).
Normalized difference images relative to the reference image are shown next to the
phantom images for each case.
At the top in Figure 3-10, a maximum intensity
projection of the attenuation map including the coils is shown and visualizes the
amount of attenuation inside the PET FOV. Without accounting for attenuation
correction of the coil (Figure 3-10, middle rows), artifacts and an overall loss in
photon counts are visible.
On average, the 8-channel and 31-channel coil show a
60
14% and 17% loss in photon counts, respectively. Based on this phantom data, the
31-channel coil attenuates 21% more on average than the 8-channel coil. In addition,
an analysis based on 5 regions of interest (ROIs, shown in Figure 3-10, right) was
carried out. Table 3.1 shows the loss in photon counts in terms of relative change for
ROIs 1 to 5 in a plane at the center of the PET FOV. A plot of these ROIs over all
planes along the length of the phantom is shown in Figure 3-11.
Attenuation correction was implemented for the 31-channel coil with an attenuation map created from CT scans. As a result, image artifacts were removed, so that
the PET images are comparable to the reference image without any coil in the PET
FOV (Figure 3-10, bottom rows). Moreover, attenuation due to the coil was largely
accounted for, such that measured losses compared to the reference image reduced to
2% and 4% on average for the 8-channel and 31-channel coils, respectively. From an
ROI-based analysis with ROIs 1 to 5, we found that ROIs at the center of the coil
had the least relative change compared to the reference image and all ROIs showed a
maximum of -5% change on average. The values for each ROI are listed in Table 3.1,
and the values for all planes along the length of the phantom are shown in Figure
3-11.
Simultaneously acquired images from a healthy volunteer demonstrate the arrays
optimization towards obtaining highly sensitive PET/MR imaging data (Figure 3-12).
The [18F]FDG-PET data shows artifact-free images, reconstructed with respective coil
attenuation maps. For MR images taken with the 31-channel coil, the higher SNR
and lower g-factor translate into improved image quality, as shown in the Ti-weighted
anatomical images (MEMPRAGE) with acceleration factors R=2 and R=4.
3.4
Discussion
Overall, the design of the PET-compatible array coil took into consideration three
main design criteria: First, sensitivity improvement of the MR image compared to
available coil designs for PET/MR systems; second, removal of all possible components (i.e. the preamplifiers) outside the PET FOV and third, the choice of coil
61
31ch PET/MR
8ch PET/MR
No coil
(reference)
Is-
IL
NE C
0
I
4
0
kBq/mL
0
0
.C
V
E*FW
20
%Relative
-20%
0
chane
20%
Figure 3-10: PET images (in gray-scale) of a cylindrical phantom filled with [ 18 F]solution acquired using the setup with the 8-channel (left), 31-channel (center) or
no coil as reference (right); and corresponding images of relative change (blue-red
scale) in percent (with respect to the reference image with no coils). Top row: A
maximum intensity projection (MIP) of the attenuation map shows the location of
the phantom in the coil and the materials inside the PET FOV. Middle rows: Coronal
and transaxial slices of the phantom for each setup without coil attenuation correction
(AC) applied. Images show artifacts and loss in photon counts, compared to the
reference image without coils in the PET FOV. Bottom rows: Coronal and transaxial
slices of the phantom with attenuation correction of the coils implemented, in which
images do not display any artifacts.
components within the PET FOV for minimal attenuation of 511 keV photons while
maintaining sensitivity from the MR signal.
62
8-channel
31-channel
10
0---------------------------
0--------------------------
O
.
o
o
10
-10
-10
- -20
-20
*
60
140
80
100
120
-- ROI 1 -RO
2 -RO
10
60
80
100
3 -ROI 4 -- ROI 5
120
140
10
r0
-10
0-10
S
-20 --
2060
80
100
120
Plane number (z-direction)
140
60
80
100
120
Plane number (z-direction)
140
Figure 3-11: Plots of the mean value of relative change in photon counts between a
phantom with and without a coil in the PET FOV for five ROIs (shown in Figure
3-10) over the length of the cylinder phantom. Without coil attenuation corrections
(AC), all ROIs show a loss in photons over the entire length of the phantom, with
the 31-channel coil showing a slightly higher but comparable loss in photon counts
to the 8-channel coil. With coil AC implemented, this loss is corrected for, showing
a reduced loss compared to a phantom with no coil in the PET FOV.
The choice of electrical components within the PET FOV was a compromise between achieving the highest possible MR SNR and low attenuation/artifacts for the
PET acquisition. Generally, electrical components that allow for high SNR are larger
or contain more highly attenuating materials on circuit boards. But if the similar
performance can be achieved with smaller components, those should be the preferred
choice. It was thus necessary to evaluate all components for their properties before
designing the coil. For example, thicker wires correspond to a higher quality factor in
measurements and would be desirable, except for thicker wires causing more photon
attenuation. As one tradeoff, we chose a thinner 18 awg copper wire for coil loops,
contrary to previous coil designs, in which 26% thicker (16 awg) wires were used [47].
Since the 18 awg wire loops only showed 6% less SNR, this was an acceptable tradeoff.
Smaller wire loops showed much larger decreases in SNR and thus were not chosen.
63
8-channel coil
31-channel coil
without coil attenuation correction
8-channel coil
31-channel coil
with coil attenuation correction
ROi 1
-12%
3%
-18%
2%
-3%
3%
-5%
2%
ROI 2
-14%
3%
-18%
2%
-2%
3%
-5%
2%
ROI 3
-14%
2%
-19%
3%
-2%
3%
-4%
3%
RO 4
-18%
2%
-20%
3%
0%
3%
0% 3%
ROI 5
-12%
2%
-16%
3%
0% 3%
3%
3%
Table 3.1: Relative change values in percent (average
std. dev.) from an ROIbased analysis for the five regions from PET phantom data shown in Figure 3-10.
Data without coil attenuation correction quantify the loss in photon counts due to the
8-channel or 31-channel coil inside the PET FOV. With coil attenuation corrections
implemented, most of the attenuation due to the coils was corrected for.
In coils with fewer receive channels, this may not have a large impact, but in the
case of 31 or 32 channels, artifacts can become more pronounced. Hence, component
choices and their arrangement should be taken into account during the coil design
stage.
We evaluated the MR performance of the constructed 31-channel array with MR
noise correlations, SNR maps and g-factor maps. We demonstrated superior MR
performance of the 31-channel array over current existing arrays (an 8-channel receive
array dedicated to the BrainPET scanner). At the same time, SNR and acceleration
properties were not diminished in the 31-channel PET/MR coil compared to a 32channel MR-only coil, even though we made some compromises on the electrical
components to reduce attenuation.
To assess the PET performance, we analyzed CT scans in sinogram space and
determined attenuation correction factors at 511 keV. Moreover, we analyzed PET
phantom data to compare the attenuation of the 31-channel coil to the existing 8channel coil and a reference phantom without coils in the PET FOV. We showed that
the constructed 31-channel PET/MR array has a similar attenuation to the 8-channel
64
Figure 3-12: Simultaneously acquired PET ([ 1 8F]FDG) and MRI in a human subject
shown as a fused image (left). Artifact-free PET images (left center) demonstrate an
accurate implementation of the 31-channel coil attenuation correction. MR images
obtained with the 8-channel and 31-channel coil show TI anatomy (MEMPRAGE at
acceleration R=2 (center right) and R=4 (right)) and the superior g-factor and SNR
of the 31-channel coil.
PET/MR array, even though it has nearly four times the number of channels. Additionally, the 31-channel array shows an improvement over conventional 32-channel
coils in terms of attenuation properties. Thus the coil design can be seen as either
improving MR performance at similar PET performance compared to the 8-channel
PET-optimized array, or improving the PET performance at similar MR performance
when compared to a 32-channel MR array.
Attenuation correction of PET images is a necessary processing step for accurate
quantification of PET datasets. However, the implementation of attenuation corrections for MR coils is not straightforward: The geometry of the MR coil cannot be
imaged since there are no transmission sources or CT available at PET/MR scanners
that account for a reliable attenuation map before each scan.
Prior knowledge of
the attenuation map and its specific location is thus required.
Therefore, it is de-
65
sirable that components are fixed in a stable position to avoid relative movement of
components that could partially invalidate attenuation maps acquired previously. To
address this issue, the 31-channel coil wasesigped to fit tightly into the table bed
and local transmit array because small displacements in space can lead to artifacts
that can be challenging to correct for [34]. Minimizing attenuation from MR coils
during the design stage avoids unnecessary artifacts and increases true coincidences
in PET images, leading to higher SNR. While the conversion of Hounsfield units into
linear attenuation coefficients via a bilinear transform is an established method for
human tissues [38, 46], applying the same conversions on highly attenuating metals
can introduce errors. Attenuation properties of coil components are usually not exactly known as extrapolation to 511 keV energies from CT scans is approximate. It
is thus challenging to correct for all components accurately [34]. On the CT images
used for deriving attenuation maps, beam hardening artifacts often cause an overestimation of the Hounsfield units of highly attenuating electrical components [37]. The
standard conversion to linear attenuation coefficients we applied may thus overestimate the attenuation properties of the RF coil and it may result in overcorrection of
the attenuation of the coil in PET images. While 511 keV transmission scans may
provide a more accurate alternative to derive attenuation maps [35], the resolution
of standard transmission images is too low for the high-resolution BrainPET camera
employed here. In order to further improve the accuracy of attenuation evaluations
and corrections, targeted conversion methods for non-tissue components and highly
attenuating electronics need to be developed.
The design criteria for PET-compatible coils presented here are generally valid for
multi-channel coils in PET/MR imaging. A similar, though slightly modified, design
of the 31-channel coil presented here can be employed for the construction of 32channel coils for newer-generation PET/MR scanners. New generation commercially
available PET/MR scanners are targeted towards whole-body imaging, in which a
longer PET FOV is present and automatic table movements allow for the acquisition
of several bed positions in PET. In this case, preamplifiers would need to be moved an
additional 5 cm further away from the coil to adjust to the PET FOV. Moreover, all
66
preamplifiers should be moved to the superior end of the coil in order not to interfere
with subsequent whole-body scans. Overall, the majority of design criteria can be
employed as proposed in this work though.
The design of PET-compatible arrays is equally important for sequential PET/MR
systems. In an integrated sequential scan setup, the subject and coils are generally
not moved as the scan table switches position from one scanner to the other [48] to
preserve the position of the patient. Hence, sequential PET/MR scanners face the
same challenges from MR coil attenuation and specially designed coils are needed for
such applications as well.
3.5
Conclusions
In this study, we presented design criteria for PET-compatible multichannel MR
arrays and showed the design, construction and performance of a 31-channel receive
array for human brain imaging that is optimized for simultaneous PET/MR imaging.
The array is characterized by a sparse configuration of 511 keV absorptive materials
in the PET FOV, placement of preamplifiers outside the PET FOV and a tight fit
into a local transmit coil within a confined space of a BrainPET camera.
The 31-channel PET-compatible array coil enables higher sensitivity and better
image quality for both PET and MRL The superior performance of the 31-channel
array was demonstrated with SNR and 1/g-factor maps in the MR domain, and with
attenuation measurements of 511 keV absorptive materials in the PET domain. We
successfully implemented accurate attenuation correction of the coil and demonstrated
this array coil enables new applications of high quality simultaneous PET and MR
brain imaging in the clinic or research domain.
67
,
the coil performance with simultaneously acquired PET/MR brain images. Hence,
3.6
Extensions
Based on the concepts and design criteria presented for the construction of the 31channel coil in this chapter, we developed two PET compatible 8-channel coils that are
specialized for simultaneous PET/MR imaging of non-human primates. These coils
enabled highly sensitive functional imaging applications presented in Chapters 4, 5
and 6. The details of the construction and performance of these coils are presented
in Appendix A.
68
Chapter 4
Neurovascular Coupling to D2/D3
Receptor Occupancy using
simultaneous PET/fMRI
4.1
Introduction
The neural basis of functional magnetic resonance imaging (fMRI) signals has been a
topic of extensive investigation. Experimental correlations of fMRI signals or cerebral
blood flow with relative changes in glucose or oxygen utilization [49, 50, 51] have been
performed in order to, determine whether hemodynamic responses reflect a coupling
with metabolism - a hypothesis with a long history [52]. To relate hemodynamic
responses more directly to neural circuitry, fMRI or optical imaging signals have
been correlated with electrophysiology using graded levels of stimulation in preclinical
models [53, 54, 55, 56]. These results generally demonstrate a monotonic coupling
between electrical activity and evoked cortical hemodynamic responses using sensory
stimuli. However, the methodology is not readily extensible to human subjects, and
the implications of such studies are difficult to generalize to pharmacotherapies or
other drug stimuli that target specific neurotransmitter systems.
Recent technological advances in multimodal imaging have enabled the simulta69
neous acquisition of MRI and positron emission tomography (PET) data [7]. One of
the motivations for conjoining these modalities is the potential for new insights into
neural function based upon the complementary natures of PET and fMRI. In activation paradigms, fMRI provides excellent spatio-temporal resolution for localizing
changes in brain activity but offers little insight into the underlying neurotransmission. Conversely, while PET has more limited temporal and spatial resolution for
functional measurements, it offers high sensitivity and neurochemical specificity. Together, PET and MRI measures have the potential to help clarify the neurochemical
basis of changes in fMRI signal induced by selective exogenous ligands or endogenous neurotransmitter. Perhaps because combined PET/MRI systems have become
available only recently, there have been no reports to date that have dynamically
compared the functional output reported by fMRI with in vivo occupancy for any
neuroreceptor system targeted by a selective ligand.
The dopamine (DA) system has been a primary target of PET studies due to its
role in numerous diseases, including Parkinsons disease, schizophrenia, and behaviors
related to natural rewards and drug abuse. The basal ganglia present an ideal target
system for initial PET/fMRI studies of neurovascular coupling, because this system
has been studied widely in humans and non-human primates (NHP) using PET radiotracers that have been thoroughly characterized [57]. fMRI signals strongly correlate
with evoked DA in rodents under some circumstances [58], suggesting that DA is
driving the fMRI response. Stimulation of D2/D3R produces inhibition of basal ganglia in fMRI studies in rodents [59, 60]. This finding is consistent with G-protein
coupled pathways leading to D2/D3R-mediated inhibition of adenylate cyclase and
cyclic AMP, a second messenger in the ATP pathway leading to energy production
[61].
In this study, responses of the dopaminergic system in NHP were assessed with
PET/fMRI in order to characterize the dynamic relationship between hemodynamic
responses and changes in neurochemistry. The PET ligand ["C]raclopride (RAC), a
selective D2/D3R antagonist, was administered at decreasing specific activities (i.e.,
increasing total mass doses) in order to evoke fMRI responses. The latter were com70
pared to receptor occupancy measures from PET in the domains of time, space, and
dose. Thereby, we test the hypothesis that fMRI signal changes are correlated with
DA displacement in each of these domains. This study initiates a new methodology
for investigating neurovascular/neurochemical coupling in health and disease using
simultaneous PET/fMRI.
4.2
Materials and Methods
4.2.1
Animal Studies
Two male rhesus macaques M1 (7 years) and M2 (5 years) with mean weight of 11.7
kg and 7.5 kg underwent imaging. Anesthesia was induced with ketamine (20 mg/kg
ketamine with 0.4 mg/kg diazepam, or 10 mg/kg ketamine with 0.5 mg/kg xylazene),
and a catheter for injections was placed in the saphenous vein. During scans, anesthesia was maintained by isoflurane (0.8 - 1.5 %, mixed with pure oxygen) through
an intubation tube without ventilation. Physiological changes (blood pressure, pulse,
end-tidal CO 2 and breathing rate) were monitored continuously throughout the experiment. The procedures complied with the regulations of the Subcommittee on
Research Animal Care.
4.2.2
Study Design
Four different RAC mass doses were administered by bolus injection in two animals
(Table 4.1). Injections of high mass doses of RAC were separated by at least two
weeks. Additionally, five low-mass dose studies (three with M1, two with M2) were
acquired in order to compute average low-dose fMRI signals. An additional mass dose
of RAC (1.4 pg/kg) was administered in M1 during a PET-only scan and used in the
KD computation for M1. For each scan, 4.6 - 6.6 mCi of [ 1 C]RAC was administered
intravenously over approximately 30 seconds. ["C]RAC was synthesized from the
0-desmethyl RAC precursor and ["C]methyl iodide. The synthesis and subsequent
purification by high-performance liquid chromatography was performed according to
71
Study parameters*
No. 1
No. 2
No. 3
No. 4
0.3
4.5
16
40
1350
53
8.9
6
Inj. RAC mass (pg/kg)
Spec. activity (pCi/nmol)
Study outcomes
M1
M2
M1
M2
M1
M2
M1
M2
BPND
3.0
5.6
1.8
3.2
1.1
1.6
0.3
0.9
DBPNea
3.3
5.1
1.6
2.4
0.8
1.1
0.2
0.6
0
0
40
43
63
71
90
84
0
0
50
52
74
77
94
88
- 1*
2.2
5.8
5.6
10
7.3
12
RAC occ.t
Peak RAC occ.t 0e
Peak ACBV (%)
(%)
5*
Table 4.1: Summary of parameters and outcomes of the PET/fMRI paradigm for
putamen and two animals (M1, M2). *Average values from studies in two animals.
tOccupancies are computed relative to dose 1. $Average CBV from repeated studies.
[22] with minor modifications. In order to vary the specific activity and total mass for
the five studies, unlabeled RAC was added after synthesis to the saline formulation
to obtain the specific activities in Table 4.1.
4.2.3
PET/fMRI Image Acquisition
Simultaneous dynamic PET/fMRI scans were acquired on a research-dedicated human PET/MR scanner, which consists of a 3 T MRI scanner (MAGNETOM Trio,
Tim system, Siemens AG, Healthcare Sector, Erlangen Germany) and an MR compatible PET insert (BrainPET, Siemens AG, Healthcare Sector, Erlangen Germany
[9]). A vendor-supplied circularly polarized local transmit birdcage coil with an 8channel receive array was used for MR imaging in animal M1.
A smaller, more
sensitive custom-built 8-channel NHP receive array was used for imaging animal M2.
The phased-array receivers enabled a two-fold acceleration with GRAPPA [32] in the
anterior-posterior direction in order to reduce the echo time (TE = 23 ms) and im-
72
age distortions during fMRI. We employed multi-slice echo-planar imaging (EPI) with
=
110 x 72.8mm 2
,
whole-brain coverage and an isotropic resolution of 1.3 mm (FOVMR
BW = 1350 Hz/pixel) with a temporal resolution of 3 sec (TR). Prior to fMRI, ferumoxytol (Feraheme, AMAG Pharmaceuticals, Cambridge MA) (45) was administered
intravenously at 10 mg/kg to improve fMRI detection power [62]. Dynamic acquisition of fMRI images started approximately 20 min before RAC injection and lasted
90 min in total.
PET emission data were acquired in list-mode format for 90 min starting with
ligand injection and sorted in the line-of-response space. Images were reconstructed
with the ordinary Poisson expectation maximization algorithm with 32 iterations.
With the PET camera used in this study, the highest image resolution was on the
order of 2 to 3 mm [4]. Corrections for scatter and attenuation of the head and the
radiofrequency coil were applied during reconstruction as described in [34]. The final
reconstructed volume consisted of 76 slices with 128 x 128 pixels (2.5 mm isotropic
voxels), with framing intervals of 10 x 30 sec, followed by 85 x 1 min.
4.2.4
fMRI Data Analysis
fMRI (and PET) data were registered to the Saleem-Logothetis stereotaxic space
[63] using a population-averaged MRI-based template for rhesus monkey brain [64].
Prior to statistical analysis using the general linear model (GLM), fMRI data were
spatially smoothed with a 2.5 mm Gaussian kernel to correspond approximately to
the PET resolution. A gamma-variate function modeled the fMRI temporal response
to RAC infusion. The time-to-peak response of the gamma function was adjusted to
minimize the X 2 /DOF of the GLM fit to data for mass doses 2-4 (Table 4.1). The
specific binding signal from PET defined the time-to-peak for the lowest mass dose,
which did not produce observable changes in fMRI signal. Maximum changes in fMRI
signal were defined as peak magnitudes of the gamma-variate regressor scaled by the
GLM. These values were converted to changes in %CBV using standard methods [65].
73
4.2.5
PET Data Analysis
By convention, BPND is defined at steady-state concentration [66] under the assumption that specific binding has no time dependence. However, binding is not constant
in time in this (Figure 4-1) or other challenge protocols, so that it is unclear how
BPND
relates to dynamic fMRI measures. Nevertheless, we report BPND values using
a conventional analysis to facilitate literature comparisons. Additionally, we employed
an alternative analysis that explicitly incorporates time dependence into occupancy
estimates and allows the comparison of peak occupancy with peak CBV measures.
We thus defined a related binding-dependent quantity, the dynamic binding potential
DBPND(t):
DBPND (t)
R 1 xk2'
(4.1)
k 2a(t)
DBPND(t)
was determined using a time-dependent parameter from the SRTM
(53, 54). This enabled computation of peak occupancies:
A
(t) ADBPND(t)
max
PFTC
JRDBP
ak(4.2)
(t)
IA
The term (e.g. Table 4.1) is the DBPND value that corresponds to the peak
occupancy as defined in Eq. 4.2.
PET data were analyzed using the cerebellum ROI as a reference region, where
D2/D3R concentration has been reported to be 103 times smaller than in putamen
[67]. Although non-negligible specific binding to D2/D3R may be detectable with
other ultra-high affinity ligands, [ 11C]raclopride has negligible specific binding in the
cerebellum [68]. BPND values were obtained with the multilinear reference tissue
model (MRTM [69]) using the standard three independent parameters R1, k 2a(t), ks.
Alternatively, we employed the two-parameter MRTM2 [69] and included the timevarying parameter k2a(t) following existing methods [70, 71]. This dynamic analysis
was applied to determine peak RAC occupancy OIP92(t) from Eq.
maximum value of the k 2a(t) curve.
4.2 using the
Simulations demonstrated that MRTM2 with
a time-dependent k2a(t) provides a more accurate determination of peak occupancy
74
than the conventional three-parameter implementation of SRTM for large doses of
RAC (Fig. B-1). Specific binding estimates were derived by subtracting cerebellum
TACs from other ROIs after inclusion of the R1 term from SRTM: (S =
CT-RCREF)-
Parametric images were generated using voxel-wise modeling based upon publicly
available software (www.nitrc.org/projects/jip).
4.3
Results
Figure 4-1 demonstrates the basic paradigm and hypothesis employed in this study.
With a very low mass of [ 11C]RAC, the PET signal is sensitive to available D2/D3R in
the baseline state (Fig. 4-la). Due to much higher density of D2R relative to D3R in
caudate and putamen [72], ["C]RAC specific binding is highly weighted by D2R density in these regions. We hypothesize that increasing masses of non-radioactive RAC
(decreasing specific activities) displace progressively more DA, reducing PET binding
at D2/D3R, while invoking an fMRI signal (Fig. 4-1-b). This is illustrated in the
temporal domain by simulations of a 2-tissue compartmental model (Fig. 4-1-c): The
wash-in/wash-out of RAC causes a time-dependent displacement of basal DA from
D2/D3R and a reduction of available D2/D3R. However, pharmacokinetics largely
dictates the shape of time-dependent changes in specific binding, which differs only
subtly versus mass until high occupancies in simulations (Fig. B-1 in the Appendix).
Experimental PET time-activity curves (TAC) from 5 injections of ["C]RAC with
different specific activities in one NHP (Ml) are shown in Figure 4-1d for whole
putamen and normalized to peak activity, together with the low-mass cerebellum TAC
as a reference. As the total injected mass of RAC increases, unlabeled RAC occupies
progressively more binding sites and thus reduces specific binding by labeled RAC,
so that putamen TACs start to resemble the shape of the non-specific (cerebellum)
TAC.
75
Pharmacologic dose
(b)
Low RAC mass dose
(a)
4fMRI
Large PET signal
in D2-rich regions
Basal DA
-20%
signal
r
80
Dopa9ine
[C]raclopride
Racopride
1
(d)
Available
_
(C)
signal
Postsynaptic membrane
Postsynaptic membrane
D2 receptor
4PET
with RAC
Occupied with DA
-Occupied
--
0.'.6
C60a
00
0-'.4
r40-
20
20
0
-
C0.
0.2
20
40
60
0
80
time (min)
20
40
60
80
time (min)
Figure 4-1: Schematic illustrating the basic PET/fMRI paradigm: (a) ["C]RAC binds
specifically to available D2/D3R in the baseline state. (b) Antagonism at D2/D3R
causes a positive fMRI signal, while unlabeled RAC competes with ["C]RAC, decreasing the PET signal. (c) Simulations show the dynamics of RAC and DA occupancy
at D2/D3R, together with available D2/D3R, all expressed as a percentage of total
D2/D3R. (d) Experimental results show changes in the shape of PET TACs for whole
putamen, normalized to peak, in order of increasing RAC mass dose (black to light
gray) and a low-mass cerebellum TAC (dashed).
4.3.1
Kinetic Modeling Results
We hypothesized that changes in RAC specific binding are related to fMRI responses
under the assumption that DA displacement drives the fMRI signal.
Hence, we
employed two analyses within the simplified reference tissue model framework (SRTM
[1]): 1) a conventional steady-state binding potential (BPND) analysis; 2) an analysis
based upon dynamic binding potentials, DBPND(t), as defined in Eq. 4.1 in Methods,
in order to compare peak changes in PET binding with peak changes in fMRI signal.
Kinetic model fits are illustrated in Figure 4-2 for three different RAC mass doses
76
(c) 40 pg/kg
TAO
(b) 16 jig/kg
(a) 0.3 pg/kg
U
30,-N
3
Fit
--
20
S
z
-.S20-NS
-
-.
-
DBPND
10
01
0
20
40
60
time (min)
1
80 0
20
60
40
time (min)
80 0
20
40
60
time (min)
80
0
Figure 4-2: PET TAC for whole putamen (dots) for the lowest mass dose (a) and
the two highest masses (b-c) with fit to the data (black). DBPND (Eq. 4.1) (gray),
together with SRTM estimates of non-specific (NS = RiCREF) or specific binding
(S = CT - RCREF) are displayed over time. Dynamic binding analyses demonstrated
an improved fit to data.
in MI. Dynamic analyses (Fig. 4-2(b)-(c)) reduced the X 2/DOF and Akaike information criterion measures for all fits. Representative curve fits comparing the SRTM
and dynamic analysis are shown in Figure B-3 in the Appendix. Moreover, dynamic
analyses of forward-model simulations determined that consistently produced good
agreement with true peak occupancies (Fig. B-3 in the Appendix), whereas analyses
based upon BPND did not consistently reflect peak or average occupancies across all
mass doses. For the low-mass RAC dose, values from both animals averaged 4.2
in putamen and 3.8+ 1.3 in caudate, while BPND values averaged 4.3
and 4.1
were 0.93
4.3.2
1.3
1.8 in putamen
1.9 in caudate. Average R1 and k' parameters from the DBPND analysis
0.08 and 0.27
0.30, respectively.
Temporal Correlation
Temporal responses from fMRI for whole putamen are shown in Fig.
4-3(a)-(b)
for the two highest injected masses of RAC in one animal and are overlaid with
approximations of specific binding (S
=
CT - R1CREF), derived directly from PET
77
X10'3
X 10~3
16 g/kg
-(a) PET : 8
0
ft%
(b) 40 g/kg :8
fMRI :6'
4
*..
*-
< 2.
-2-.*
0
20
40
60
time (min)
-
-2
-2L
80
0
20
40
60
time (min)
80
-
Figure 4-3: Temporal responses from PET specific binding estimate (S
CT
RiCREF, black line), derived from experimental TACs, overlaid with CBV changes
(binned to 1 min intervals, gray dots) for putamen for the two highest RAC masses.
Timecourses from CBV and S resemble each other.
TACs. Using this specific binding estimate allowed a direct temporal comparison of
responses derived from PET and fMRI data. S provides a reasonably accurate index
of specific binding as modeled with a 2-tissue compartmental model (Fig. B-5 in the
Appendix). The timecourses of PET specific binding estimates and fMRI responses
resembled each other, as time-to-peak measures and durations were similar.
Fits to the fMRI temporal response and DBPND(t) were determined by varying
the time-to-peak
(T)
of gamma-variate regressors (t/re(/7) in order to optimize the
goodness of fit to the data within the GLM. For both fMRI and PET data, X 2 /DOF
were slowly varying functions of the time-to-peak. All X 2 /DOF values increased by
less than 10% across a 4 min window centered upon the optimal value.
Optimal
time-to-peak values from fMRI and PET analyses differed by less than 4 min for each
animal at the two highest doses, for which the fMRI signal was robust.
4.3.3
Spatial Correlation
The upper row of Figure 4-4 shows parametric maps from PET for 4 mass doses, and
the lower row of Figure 4-4 shows the corresponding fMRI maps in units of %CBV.
A p-value threshold of p < 0.03, computed with a mixed effects model [73] across two
animals was applied to all CBV maps. For the lowest RAC mass, the maps exhibited
78
high specific binding and a high binding potential in the striatum, but hardly any
detectable fMRI signal. In the voxelwise maps, 49% of voxels within the putamen
were above a value of and 23% within the putamen exceeded a value of . Results were
reversed for the highest RAC mass dose, for which binding potential was very small
while fMRI reported a large positive change in CBV. The peak CBV response relative
to baseline CBV averaged 9.8% 3.5% across whole putamen and 4.3% 1.7% across
caudate. Voxelwise analysis showed that 45% of voxels within the putamen show
CBV > 8% and 19% of voxels in putamen recorded CBV > 12%.
The high-binding regions of the striatum and the neurovascular responses were
remarkably similar on the voxel-wise maps (Fig. 4-4). As mass dose increased, the
specific binding signal in the striatum decreased visibly in the parametric images,
while CBV increased from no detectable signal to a prominent signal within the
striatum.
Occupancies and Relationship to fMRI
4.3.4
Table 4.1 lists values for BPND, steady-state RAC occupancies and peak occupancies
(Eq. 4.2) for whole putamen from the SRTM and the dynamic binding analysis,
together with peak changes in CBV for each RAC mass dose for both animals. Binding
potentials (BPND) extrapolated to true zero occupancy for whole putamen were 3.3
and 5.0 for NHP M1 and M2, respectively. These were computed from five (Ml) or
four (M2) RAC mass doses with the Hanes-Woolf plot, in which 1/BPND is plotted
versus free ligand concentration in tissue. The latter was estimated from specific
activity values and the first 60 min of data from the cerebellum TAC. Based upon
these extrapolated baseline binding potentials, occupancies for whole putamen ranged
from
-
7% to 91% for M1 and < 1% to 82% for M2. The average RAC dissociation
constant KD across animals was 1.9
0.6 nM (assuming a RAC free fraction
IND
=
0.12 [74]), which is within the range of literature values [75].
Figure 4-5-a-b shows the relationship between peak RAC occupancies and peak
%CBV changes for whole putamen and caudate in each animal. Data points were
described by regression with a power law with two parameters: For M1, the exponent
79
V.-tJAUfW0 J / K
40
JA11%
/
Increasing raclopride mass dose
16 /k
JA
'g
DBPNP"k
4
0
%CBV
15
0
Figure 4-4: Top row:
DBPak parameter maps from PET data, overlaid on an
anatomical MR atlas. Bottom row: Maps of %CBVPeak change from fMRI data,
windowed by a p-value map with p < 0.03. All maps are created from data from
two animals with a mixed effects model. Similarities in the spatial distribution of
PET/fMRI signals and the dose dependencies support the idea that antagonism of
RAC at D2/D3R is elicited the CBV changes.
b was 1.7 and 1.4 in putamen and caudate, respectively, with R 2 values of 0.97 and
0.39. For M2, the corresponding exponents were 1.4 and 1.6, with R 2 values of 0.99
and 0.88. The regression with a power fit is concordant with a model describing the
relationship between DA and RAC occupancy (Eq. B.1 and B.2 in the Appendix).
If we assume a 2-fold increase in extracellular DA due to RAC injection, model
calculations and a corresponding power law fit predict an exponent of b = 1.6 (Fig.
B-2 in the Appendix), which agrees with our experimental data.
From a regional analysis, we found that CBVpeak responses from the putamen
ROI were much larger compared to caudate in both animals (Fig. 4-5). The putamen/caudate ratio of the CBVpeak response at 80% occupancy, estimated from the
experimental data fit, was 2.2
0.2. Including the fitted CBVpek values for all oc80
14
(a) M1
10
(b) M2
> Putamen
12 -Q
Caudate
putamen=0.97, b=1.7
8) R203,
8
caudate
0
39
=putamen=0.99, b=1.4
=.
, b14
Rcaudate
0
2
0
20
40
=0.88, b=1.6
--
-0
60 peak
%Occupancy (0*Ac)
0
100
0
20
40
60
80
%Occupancy ( pAC
100
Figure 4-5:
%CBVpeak vs. occupancy can be approximated by a power-law fit
(with exponent b) for putamen and caudate ROI for M1 and M2. Data show a
monotonically increasing relationship, with putamen exhibiting approximately twice
the CBV magnitude compared to caudate. Error bars show within-session uncertainty
from the GLM analysis and SEM for repeated fMRI studies with the lowest RAC mass
dose.
cupancies between 50% and 100%, the average putamen/caudate ratio was 2.2
1.7.
Regional differences in receptor densities were 9% 2% less in caudate, as determined
by baseline
BPND
values. However, this alone does not provide an explanation for
the much smaller functional response in caudate.
We conclude that there is a monotonically increasing relationship between RAC
occupancy and changes in %CBV that showed a good fit to a superlinear function.
Moreover, the magnitude of the CBV response for all RAC masses exhibited a regional
dependence within the striatum.
81
4.4
Discussion
In this study, we investigated the relationship between neurovascular responses and
receptor occupancies using simultaneous PET/fMRI and variable mass doses of the
D2/D3R antagonist RAC. As injected RAC mass dose was increased, the reduction
in [ 11C]RAC specific binding correlated with CBV increases in the striatum spatially,
temporally and with RAC mass dose. These data suggest that vascular responses to
D2R-like antagonists are coupled to changes in neuroreceptor occupancy.
By directly comparing changes in neuroreceptor occupancy with a simultaneously
induced functional response, we demonstrated an approach that can be employed
widely using both antagonists and agonists. This opens up the possibility to probe
mechanisms of neurovascular coupling or to investigate brain circuitry by relating
selective neuroreceptor binding to local and distant functional responses.
Studies
of this design can be used to develop multi-receptor models that describe the fMRI
response to endogenous neurotransmitter [76] in order to clarify the neurochemical
basis of behaviors like reward [77, 78]. Finally, simultaneous PET/fMRI studies using
targeted PET ligands may provide information about basal receptor occupancy, a
quantity that is not routinely measured by PET alone.
4.4.1
Regional Dose Response
The maps of DBPND and %CBV in Figure 4-4 each showed localized signals in striatal
regions that exhibited a strong dependence on injected RAC mass. At low injected
masses, DBPND maps showed high specific binding indicative of a large number of
available receptors in basal ganglia, as expected. At high RAC masses, specific binding of ["C]RAC was decreased by competition with unlabeled RAC. However, fMRI
showed progressively larger increases in CBV at higher mass doses that presumably
reflect transient decreases of DA basal occupancy at D2/D3R. Since DA produces
an inhibitory effect at D2/D3R, an explanation for the increase in CBV is that RAC
blocks the inhibitory effect of DA upon D2/D3R. Thus, fMRI responses were observed
predominantly in regions with high D2/D3R density.
82
4.4.2
Temporal Comparisons
The temporal response of the fMRI signal closely matched the kinetics of ["C]RAC
specific binding estimates (Fig. 4-3). The temporal correlation is intriguing given
the numerous physiological mechanisms that could cause temporal responses from
the two modalities to diverge. For instance, a large dose of an agonist might promote receptor internalization. This mechanism has been suggested to explain the
prolonged decrease in raclopride displacement following infusion of a large dose of
amphetamine [79]. With internalization, fMRI signal should be shortened in time,
so that divergent temporal responses could offer insight into dynamic physiological
adaptations like agonist-induced receptor internalization. However, our data show
that a D2/D3 antagonist produces changes in binding potentials and fMRI signals
that are well matched in time, which is consistent with a classical competition model
[79] between RAC and DA for synaptic binding sites.
4.4.3
Neurovascular Coupling mediated by D2/D3R Antagonism
In this study, data showed a monotonically increasing relationship between changes
in CBV and RAC occupancies in subregions of basal ganglia. A postulate of linearity between tissue function and receptor occupancy was proposed as early as 1937
[26]. For pharmacological stimuli, receptor occupancy is an important synaptic quantity that underlies the genesis of a functional response. Moreover, correlations between hemodynamics and electrophysiology have shown that vascular responses are
monotonically related to integrative synaptic quantities like local field potentials and
multi-unit activity (6-8, 25). As a basis for discussion of the observed CBV-occupancy
relationship, we propose a model for the fMRI response: CBV changes (ACBV) are
driven by relative changes in receptor occupancies (AO) and are scaled by ligand efficacies (e), the local density of receptors (Bm.) and excitatory/inhibitory neurovascular
coupling constants (N). In a general framework, CBV changes due to exogenous
or endogenous agonists reflect the sum of influences mediated by multiple receptor
83
subtypes:
#receptors #ligands
ACBV =
Nj6jjBmax,iAdij
(4.3)
i=1j=
Antagonists like RAC have affinity but no efficacy at target receptors. However,
RAC produces a physiological response by displacing DA. As a first step, we adopt
a first-order (linear) approximation for our results that ignores any additional DA
release (AGDA = -ORA9(c9
, see Eq. B.3, Fig. B-2 in the Appendix) and Eq. 4.3
simplifies to:
ACBV
(4.4)
= -N2Bmax,299RAC
This description relates a positive ACBV to RAC occupancy, basal DA occupancy,
.
and Bmax via an inhibitory N 2
The precise nature of the relationship between
ORAC
and the change in DA occu-
pancy should be dependent on the magnitude of additional DA release induced by
the binding of RAC to presynaptic autoreceptors [80]. A classical competition model
predicts that changes in fMRI signal are insensitive to DA release at very high RAC
occupancies. At moderate occupancies, increased synaptic DA will more effectively
compete with RAC, causing the relationship between DA and RAC occupancies to become superlinear (Fig. B-2 in the Appendix). This effect may explain the superlinear
correlation between RAC occupancy and CBV in Fig. 4-5.
4.4.4
Basal Dopamine Occupancy
As observed from the CBV-occupancy plot (Fig. 4-5), caudate exhibited a remarkably
smaller CBV increase than putamen with pharmacologic doses of RAC. The proposed
coupling model in the previous section provides a potential explanation for this difference by showing that both receptor density and basal D2/D3R occupancy affect
the CBV response.
We can apply Eq. 4.4 to calculate the ratio of basal D2/D3R
occupancy between putamen and caudate based on our study data: Taking into account a small difference in receptor densities from our PET measurements (9% less
receptors in caudate), this ratio is 2.0. Several studies have measured higher levels
84
of basal DA in putamen compared to caudate using microdialysis in NHP in vivo
(27, 28), with basal DA levels up to 2-fold higher in the putamen. Further evidence
of higher basal D2/D3R occupancy in putamen is provided by human post-mortem
[81] and PET studies with basal DA depletion [82]; both methods showed a 1.3 to
1.5-fold difference. Conversely, some studies indicate similar levels of basal DA or
basal D2/D3R occupancy in putamen/caudate [83, 84, 85]. Our results suggest that
PET/fMRI measurements of this type offer the potential to non-invasively assess relative differences in basal neurotransmitter occupancy across regions, or ultimately
across subject groups.
4.4.5
Study Limitations
There are a number of limitations with this animal model. Studies were performed in
isoflurane-anesthetized NHP to facilitate the use of pharmacological challenges that
achieved very high occupancies. Hence, changes in basal DA induced by anesthesia
could potentially affect the magnitude of fMRI signal changes in this paradigm. Very
high levels of isoflurane (3%) increase basal DA in rats, but effects are not significant
at 1% isoflurane [86], as used in this study. In some experiments, we observed large,
transient and repetitive changes in fMRI signal that were consistent with isofluraneinduced burst suppression, a noise source that would not be present in human studies.
Additionally, we employed an MRI contrast agent that increases the magnitude of
fMRI signal changes. With endogenous contrast techniques, it may be difficult to
detect small signal changes due to pharmacological stimuli. The MR contrast agent
now has been used for fMRI in human subjects [87J, although repeated within-subject
measurements presumably will be restricted.
There are also limitations to our interpretation of D2R-mediated neurovascular
coupling. Our results and proposed model are consistent with a neurovascular coupling mechanism that is driven by occupancy as one aspect of synaptic processes.
However, we cannot exclude direct actions of DA on the vasculature that are uncoupled with neural activity. Our model excludes other factors (e.g., neurotransmitter
release) that also could contribute to neural activity and changes in fMRI signal.
85
Moreover, neural pathways that project to a region of interest may produce functional contributions that are not directly related to local occupancy, although such
situations can provide useful information about neural connectivity.
Within the framework of an occupancy-based model like Eq. 4.3, there are potential contributions to the fMRI response other than changes in D2R occupancy.
We assumed a linear relationship between DA occupancy and fMRI signal, together
with a superlinear relationship between RAC and DA occupancies.
Although this
model produces good agreement with data, there is no routine way to measure DA
occupancy directly in order to verify the linear assumption of Eq. 4.3. Additionally,
RAC-mediated DA release can cause a positive fMRI signal contribution due to stimulation of Di-like excitatory receptors. In order to address this issue quantitatively,
further studies will be needed in order to clarify the magnitude of fMRI signal changes
associated with changes in D1 occupancy.
4.4.6
Potential Clinical Applications
Clinically, simultaneous PET/fMRI appears to have broad roles for diagnosis and
monitoring of therapy [28]. Based upon this study, one potential application is the
estimation of basal receptor occupancies, which are coupled to basal neurotransmitter
levels. Basal occupancies are not routinely measurable by PET alone unless neurotransmitter is fully depleted from the brain. DA depletion studies have employed PET
to assess differences in basal D2/D3R occupAncy in patient and control populations
[88]. While such studies have helped clarify important aspects of basal DA function,
depletion can produce motor dysfunction and a whole spectrum of psychiatric symptoms [89]. Other methodological concerns include incomplete depletion and receptor
externalization or upregulation [90]. More tolerable methods for assessing basal occupancy might facilitate the identification of therapeutic correlates of basal occupancy,
following paths established for D2/D3R occupancy by antipsychotics [91].
In this PET/fMRI study, the DA-bound receptor population was unmasked transiently by antagonist displacement rather than by depletion. If studies demonstrate
sufficient reproducibility/specificity, a single injection of an antagonist can in prin86
ciple provide a relative index of basal occupancy for use in cross-sectional studies.
D2/D3R antagonist doses will need to produce measurable fMRI signals in humans
while avoiding the noxious side effects that accompany occupancies above the thera-
peutic window of 70-80% for antipsychotics [92].
4.5
Conclusions
We demonstrated an imaging methodology in which a PET antagonist can be administered at pharmacologic doses to measure neurovascular responses simultaneously with
receptor occupancy. By measuring cerebral responses using different specific activities
of ["C]raclopride with PET/fMRI, we observed that the signals from both modalities correlated in anatomical and temporal domains. Our results suggest a monotonic coupling relationship between neurovascular responses and dopamine D2/D3R
occupancy in basal ganglia across a wide dynamic range. We observed distinct relative CBV magnitudes between putamen and caudate, consistent with higher basal
dopamine levels in putamen. These results demonstrate that the concurrent assessment of hemodynamics and receptor-specific neurotransmission with simultaneous
PET/fMRI offers new possibilities for performing dynamic neurotransmitter mapping and understanding distributed functions of the brain in preclinical and clinical
studies.
87
88
Chapter 5
Imaging Agonist-induced D2/D3
Receptor Internalization
5.1
Introduction
Simultaneous functional imaging with positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) allows the concurrent measurement of
receptor occupancies and hemodynamic parameters.
More importantly, not only
steady-state neurochemistry can be measured but rather, dynamic changes in kinetic
and biochemical parameters over time can be assessed simultaneously. Using pharmacological challenges that target the dopamine system with a D2/D3 antagonist,
we have shown that PET and fMRI signal changes can be matched in time and thus
provide evidence that hemodynamics are directly linked to receptor occupancy by
neurovascular coupling mechanisms (Chapter 4). Combining information from PET
and fMRI while perturbing neuroreceptor states has the potential to enhance our understanding of neurobiological parameters as it enables us to extract novel information
about synaptic events and their downstream effects in vivo.
In the context of drug-receptor interactions, the classical occupancy theory postulates that receptors can either be in a bound or unbound state at the postsynaptic
membrane and that binding to or unbinding from receptors causes a functional response. In PET blocking or competition studies, the occupancy model carries the
89
assumption that changes in radiotracer binding potential directly reflect receptor
availability. The classical occupancy model provides a valid explanation for some experimental data: For certain PET ligands, decreases in binding potential accompany
increases in dopamine concentration, as measured by microdialysis [93, 94].
The
tight relationship between endogenous dopamine measured from microdialysis and
cerebral blood volume changes is also concordant with the classical model [60]. In
addition, similar temporal responses from fMRI and PET receptor occupancy due to
D2 antagonism shown in Chapter 4 support the classical occupancy theory.
However, there are a number of reported experiments with features that cannot be
explained by the classical occupancy theory. One validated observation is that the decrease in binding potential of D2 receptor antagonist radiotracers due to amphetamine
lasts much longer than the timecourse of the drug or extracellular dopamine concentrations. Moreover, radiotracers such as spiperone and pimozide (non-benzamides)
seem to show binding properties that oppose those predicted from the classical occupancy theory. Agonist-induced receptor internalization is a concept that attempts
to explain these contradictions to the classical theory. The central idea behind this
internalization model it that receptor availability is reduced by the entrapment of
receptors intracellularly.
Receptor internalization is a synaptic receptor adaptation mechanism that modulates the cellular and thus downstream response. In vitro, it has been shown to
occur at D2 receptors in response to D2 agonists [95]. Moreover, it has been shown
that internalization can be induced relatively fast (within minutes of D2 agonist exposure), but that receptors can stay internalized for a very long time (several hours
or days) [96]. These receptor adaptation mechanisms can directly influence imaging
data and their outcomes, especially in agonist-induced competition studies. Despite
these indications, there are currently no methods to non-invasively measure receptor
internalization or to validate rates of receptor internalization in vivo.
The purpose of the work presented in this chapter thus is to investigate how receptor adaptation mechanisms can affect simultaneous measurements with PET and
fMRI and to deduce indices of receptor adaptation mechanisms non-invasively in vivo.
90
First, a model of agonist-induced receptor internalization is proposed, together with
a neurovascular coupling model that shows how PET and fMRI data are affected by
various rates of internalization. Based on the model, we hypothesize that internalization prolongs the decrease in binding potential from [ 11C]raclopride but shortens
the fMRI temporal response. Second, imaging experiments with ["C]raclopride-PET
and fMRI in non-human primates are performed using injections of graded doses of
the D2 agonist quinpirole. The temporal and dose responses from these simultaneous
data allow a first assessment of an in vivo measurement of agonist-induced receptor
adaptation mechanisms with PET/fMRI.
5.2
Theory
We previously proposed a neurovascular coupling model that relates receptor occupancy directly to changes in CBV as measured by fMRI (Chaper 4), in agreement with
the classical occupancy theory. Here, we expand this neurovascular coupling model
to cases in which the classical occupancy model does not hold and we investigate the
results for varying ligand properties. The compartmental model and its relationship
to PET and fMRI signals is illustrated in Figure 5-1. Applying this model allows us
to draw more in-depth connections between the PET and fMRI signals.
5.2.1
PET and fMRI Signal Models
In this section, we present the signal models that describe signal changes in fMRI
and PET. Through a neurovascular coupling model, changes in hemodynamics (such
as ACBV) are directly related to changes in receptor occupancy. If a ligand L that
is specific to D2R (e.g. a D2 agonist as in this work) is administered, the agents
contributing to the CBV response are ligand L and endogenous dopamine DA:
ACBV =
NDeLBmaOL - ND2BmxA9DA.
(5.1)
In this equation, ND2 denotes the neurovascular coupling constant for D2R, e the
91
PET
kin
fMRI
Figure 5-1: Schematic illustrating a compartmental model that describes receptor
internalization at dopaminergic neurons. The total number of receptors (Bmax) are
composed of the available receptors at the postsynaptic membrane, those bound by
an injected agonist, by endogenous dopamine and internalized receptors. Occupied
receptors are in exchange with free ligands in the synaptic space. Receptors that are
occupied due to agonist exposure can trigger internalization. Since externalization
mechanisms are known to be very slow, we assume that ket = 0 for the duration
of modeled timecourses. The parameters that determine the PET and fMRI signal
changes are depicted in blue and green, respectively. This shows that PET and fMRI
timecourses contain complementary information about receptor adaptation mechanisms.
efficacy of the ligand L (see following section), Bmax the total density of receptors and
0 the occupancy of the ligand (subscript L) or endogenous dopamine (subscript DA).
Receptor occupancy as measured by PET
(OPET),
probes for the number of avail-
able receptors and can thus be described as the summation over compartments of
92
receptor occupancies 0:
OPET = OL + (ODA
-
-
9
int
-
(5.2)
The affinity of the radiotracer to internalized receptors is described by a. Specifically, a is the ratio of affinities (a = .
D~
t
The relation to known measures, such as (dynamic) binding potential DBP is
then:
9
PET
-
(ADBP
BP(0 )
DBP = (1
5.2.2
-
(
DBP\
BP()),
OPET)BP( 0 ).
(5.3)
(5.4)
Classical Occupancy Model: Efficacy of Drugs
The classical occupancy model can be described with a compartmental model (see
Figure 5-1).
It describes how antagonists and full or partial agonist occupancies
are related to changes in CBV, under the assumption that no receptor adaptation
mechanisms occur. The following assumptions are made in this model:
i. The total number of receptors Bm, stays constant on the postsynaptic membrane.
ii. Receptor occupancy causes a functional response (as described by CBV) that
is dependent on the efficacy of the ligand. This response is described by the
neurovascular coupling model in Figure 5.1.
iii. Efficacy e of an antagonist is zero and that of a full agonist is 1. Partial agonists
have an efficacy between 0 and 1.
To investigate the effect of drug efficacy on the CBV response, we simulate a
compartmental model that describes the occupancy dynamics of an injected ligand
(which can be an agonist, or an antagonist) in the presence of 20% basal dopamine
occupancy at the D2R receptors.
93
The model can be described with first order differential equations:
dOL
d
(5.5)
= kon,L0avaiFL - koff,LOL,
9
d DA
= kon,DAdavai1FDA -
Oavail =
1
-
OL -
(5.6)
koff,DAODA,
ODA
-
dt
The concentrations of the free ligand and free DA can both be normalized to the
baseline DA occupancy at time zero, such that the above equations can be modified
to:
dOL = koff,L OavaifL (KD)
dt
dt
5.2.3
/(1
((O)
_
koff,DA
-
dt
Oavail =
1
-
L-
javailfDA
DA
1 _ OD(A
OL),
(5.7)
- ODO)
_
DA )
(5.8)
ODA
-
dODA
-
KD,L
Internalization Model
High-affinity agonists are known to cause receptor internalization [95], and thus we
expect the classical occupancy model not to hold true in these cases. Instead, we
proposed an internalization model that again describes the related CBV response
(and the PET response) through the neurovascular coupling model. The following
assumptions are made for this model:
i. Receptors can be internalized due to high-affinity agonist binding to receptors.
ii. At t = 0, no receptors are internalized.
iii. The externalization constant ke, is very long, and is thus negligible for timescales
< 2 hours.
iv. Once internalized, receptors can be bound with an affinity a by the PET radiotracer. If a = 0, internalized receptors do not bind the radiotracer; if 0 < a < 1,
94
the affinity is reduced compared to receptors on the postsynaptic membrane and
if a > 1, the affinity is increased (e.g. due to the radiotracer being entrapped
inside the cell membrane).
v. Internalized receptors do not contribute to the functional response (CBV).
Figure 5-1 shows a schematic of the internalization model as described with a compartmental model. The corresponding equations can be expressed as follows:
dOL = off,L
dt
dDA
dt
Oava/L
off,DA
K
DA
availfDA
DA
9
-ODA ),
OL
-
intOL,
(5.9)
(5.10)
DA
-kfA1-0
(5.11)
do'n' = kint(OL + ODA),
dt
1 - OL - ODA - Oint
5.3
Methods
5.3.1
Animal Model
-
Oavail =
DA
KD,L
Two animals (male rhesus macaques) underwent PET/MR imaging in 8 sessions. All
animals were anesthetized, initially with 10 mg/kg ketamine and 0.5 mg/kg xylazene,
and maintained with isoflurane (1%, mixed with oxygen) after intubation. All procedures complied with the regulations of the Subcommittee for Research Animal Care
at Massachusetts General Hospital.
5.3.2
Study Design
[1 C]raclopride was injected as a bolus + infusion paradigm. A kwr, = 52 min and 81
min was used for animal M1 and M2 respectively. The bolus was administered by
hand over a duration of 30 seconds, after which infusion at a rate of 0.01 ml/s was
95
started with an automatic pump (Medrad Spectra Solaris). After 35 min, the highaffinity D2 agonist quinpirole was injected at three different doses in both animals. To
test whether the fMRI signal is sensitive to multiple quinpirole injections after some
delay (for the 0.1 and 0.2 mg/kg dose in animal M2), a second quinpirole injection
with the same dose, respectively, was administered 2 hours after the first dose. In
two separate experimental sessions, the medium-affinity D2 agonist ropinirole was
injected at 0.1 mg/kg and 0.3 mg/kg in animal M2.
5.3.3
PET/MR Image Acquisition and Reconstruction
Simultaneous PET and MR data were acquired on a prototype scanner that consists of a BrainPET insert and a Tim Trio 3T MR scanner (Siemens AG, Healthcare
Sector, Erlangen Germany). A custom-built tight-fitting PET compatible 8-channel
NHP receive array together with a vendor-supplied local circularly polarized transmit coil was used for MR imaging. The phased-array enabled 2-fold acceleration
with GRAPPA [32] in the anterior-posterior direction. Whole-brain fMRI data were
acquired for the duration of PET imaging with multi-slice echo-planar imaging that
had an isotropic resolution of 1.3 mm and a temporal resolution of 3 sec (TR). Other
parameters included FOVMR
=
110 x 72.8 mm 2, BW=1350 Hz/pixel. To improve
fMRI detection power, ferumoxytol (Feraheme, AMAG Pharmaceuticals, Cambridge,
MA) was injected at 10 mg/kg prior to fMRI [62].
PET emission data were acquired in list-mode format for 100 min starting with
radiotracer injection. Images were reconstructed with a standard 3D Poisson orderedsubset expectation maximization algorithm using prompt and variance-reduced random coincidence events. Normalization, scatter and attenuation sinograms (including
attenuation of the radiofrequency coil) were included in the reconstruction. The reconstructed volume consisted of 2.5 x 2.5 x 2.5 mm voxels in a 128 x 128 x 76 matrix,
with framing intervals of 10 x 30 sec, followed by 1 min frames.
96
5.3.4
PET and fMRI Data Analysis
PET and MR data were registered to the Saleem-Logothetis stereotaxic space [63], in
which regions of interest were defined in a standard way. After spatially smoothing
fMRI data with a 2.5 mm Gaussian kernel, statistical analysis was carried out using
the general linear model. The temporal response to the drug injection was modeled
with a gamma-variate function, in which the time-to-peak was adjusted to minimize
the X 2 /DOF of the GLM fit to the data. The resulting signal changes were then
converted to changes in CBV with known methods [65].
PET kinetic modeling was carried out with the reference tissue model SRTM2,
with cerebellum as the reference ROI. Since binding does not stay constant but is
dynamically altered as a result of the D2 agonist drug challenge, our kinetic analysis
included the time-dependent parameter k 2a(t) and a previously defined dynamic binding potential (DBP, Eq. 4.1). Parametric images from voxel-wise kinetic modeling
were generated with open-access software (www.nitrc.org/projects/jip).
5.4
Results
5.4.1
Simulations of the Classical Occupancy Model
Simulations of the neurovascular coupling model in Eq. 5.1, together with the classical
occupancy model described in Equations 5.7 and 5.8, predict CBV timecourses for
a range of ligands with varying efficacies (Figure 5-2). For a ligand with efficacy
zero (i.e. an antagonist), the CBV response is purely positive, whereas for a ligand
with efficacy 1 (i.e. a full agonist), the CBV response is purely negative, with both
following a similar timecourse. For ligands that have efficacy 0 < e < 1, we found
two cases:
1. If the efficacy of the partial agonist is close to 1, and larger than the basal
occupancy of the endogenous neurotransmitter, the CBV response is similar to
that of a full agonist, but with diminished amplitude.
97
10
-CBV:e=
-CBV:
E=
-CBV:E=
-CBV: E =
-CBV: e =
-CBV: E =
-CBV: e =
-CBV:E=1
5
0
-5
10
E
-10
0
0.05
0.1
0.15
0.2
0.3
0.6
-15
10 -20
-25
-30
-35
-40
0
10
20
40
30
time (in min)
50
60
Figure 5-2: Simulated CBV timecourses (without internalization) for the same occupancies but varying efficacies of a ligand. Baseline occupancies of endogenous
dopamine are assumed to be 20%. Antagonists (e = 0) show a positive CBV signal,
whereas full agonists (E = 1) show a negative CBV signal. For partial agonists, the
response depends on the basal DA occupancy: If efficacy is high enough, the partial
agonist response is similar to a full agonist response. But if efficacy is low, the CBV
response of a low-efficacy partial agonist (0 < E ; 0.2) can become biphasic.
2. If the efficacy of the partial agonist is close to zero, and smaller than the basal
occupancy of the endogenous neurotransmitter, the CBV response becomes
biphasic, with an initial negative peak, followed by a later positive peak.
5.4.2
Simulations of the Internalization Model
We investigated how internalization affects the CBV and PET timecourses with the
internalization model proposed in Equations 5.9 - 5.11. CBV timecourses were modeled using Eq. 5.1 and occupancies measured with PET were modeled using Eq. 5.1
and 5.2. For the results in Figure 5-3, we assumed a full agonist injection with efficacy
of 1, and that the radiotracer does not bind to internalized receptors. With no inter-
98
0.8-
-60
.6
.. CBV: No internalization
*..CBV: 1/k =30 min
-50
40.6
-
-40
.CBV:
1/k
=5mn
'"5
0 PET No internalization
-PET
+30
.
-20-0
0
PET
Int
1/k
tnt
30 min
5 min
0.2
-_10
0
*
.
+,
0
10
0'
,ow
20
-*M- ,,-
30
40
time (in min)
50
60
Figure 5-3: Simulation results from the proposed model of receptor internalization
that show how PET and fMRI signal timecourses are affected for different rates of
internalization due to a D2/D3 agonist injection at time t=0. If no internalization
occurs, PET and fMRI signals are matched in time. If internalization occurs with a
moderate time constant of 30 min, PET and fMRI signals start to diverge. With very
short time constants for internalization (5 min), the fMRI timecourse is shortened,
whereas PET occupancy stays elevated for much longer.
nalization (red curves), CBV changes follow the timecourse of receptor occupancies.
However, if agonist-induced internalization occurs, the PET and CBV timecourses
start to diverge, resulting in shorter fMRI signals and a prolonged elevated PET
occupancy.
The shortening of CBV timecourses is dependent on how fast internalization occurs in vivo. Whereas the CBV timecourse with no internalization peaks at 8 min.
and lasts about 50 min until it returns to baseline, a moderate internalization time
constant of 30 min shows a peak at 4.6 min (blue curves) and a fast internalization
constant of 5 min (green curves) causes the CBV peak to be shifted to 2.4 min. The
duration of the CBV timecourse for internalization constants of 30 min and 5 min is
abbreviated to approximately 30 min and 10 min, respectively.
The prolonged decrease in the PET signal, which is equivalent to a long-lasting
observed increase in receptor occupancy, depends both on the internalization constant
99
1
I
I
-
aw 0.5
-
= 0.2
a = 0.5
-
a=1.5
S--a
M
0
0
UJ -0.5-
-1'
0
20
time (in min)
40
60
Figure 5-4: Simulated PET occupancy measures for the duration of a dynamic PET
scan for different affinities a of internalized receptors. If internalized receptors are not
accessible to the radiotracer (a = 0), occupancy is very high and stays elevated for
the duration of the experiment. If the internalized receptors are low-affinity receptors
(0 < a < 1), occupancy is still increased for a prolonged time. However, if affinity
for internalized receptors does not changes or is even increased (a > 1), occupancy
may appear as reduced to negative values, in which case the time activity curve will
show an increase due to the agonist exposure.
and whether the radiotracer can bind to internalized receptors. Our simulation results
show that fast internalization constants (5 min) cause a larger number of receptors
to be internalized.
This results in a higher occupancy level compared to a slow
internalization constant, if internalized receptors are not accessible to radiotracer
binding (a = 0). If the radiotracer binds to internalized receptors with a lower affinity
(0 > a > 1), PET occupancy still stays artificially increased for a prolonged time
(Figure 5-4) but at a lower occupancy level. However, if the radiotracer can bind with
equal affinity to internalized receptors (a = 1), the timecourse of occupancy matches
that of CBV and returns to an increased occupancy quickly, depending on the initial
endogenous basal occupancy. If affinity for the radiotracer is increased (a
=
1.5), e.g.
if it cannot diffuse out of the cell as easily, receptor occupancy may even seem reduced
and the measured PET binding potential will increase from its baseline value.
100
Tracer:
["C]raclopride
Increasing quinpirole mass dose
-0
0.1 mg/kg
0.2 mg/kg
0.3 mg/kg
DBPpak
ND
4
0
%CBVpeak
10
-15
Figure 5-5: Voxelwise parametric maps showing PET and fMRI parameters due to
three different dose injections of the D2/D3 agonist quinpirole. Top: DBPNk maps
show the dose-dependent decrease in binding potential in the striatal regions. Bottom:
CBVp'a maps show negative CBV signals that are localized to the striatum and
increase in magnitude with dose.
5.4.3
Spatial Correlation of Occupancy and CBV
Parametric maps from PET kinetic modeling results (DBPN"Da maps) and fMRI
statistical analysis (CBVP** maps) are shown in Figure 5-5. Specific binding from
PET and CBV changes from fMRI both show a localized response in putamen and
caudate only. Moreover, the measurements from both modalities are dose-dependent:
Specific binding decreases with increasing dose of quinpirole, whereas the negative
CBV signal shows a larger magnitude as dose increases. Contrary to the antagonist
dataset from Chapter 4, the CBV response is negative, following injection of the
agonist quinpirole.
The voxelwise maps show that the CBV signal change in the caudate is noticeably
larger compared to putamen for all doses.
101
> Putamen
0 Caudate
-2
cc
-4
-1
M
0
-6-8
-12
0
20
40
60
80
100
%Occupancy (peak)
Quinp
Figure 5-6: Plot of CBVPea' measures for putamen and caudate ROIs against peak
occupancy due to three dose injections of quinpirole injections (0.1, 0.2 and 0.3 mg/kg)
in animal M2. Data show a monotonically decreasing relationship that is fitted with
a power-law. Error bars show within-session uncertainty from the GLM analysis.
5.4.4
Dose Response from Occupancy and CBV
Figure 5-6 shows a direct comparison of the CBVpeak results to peak occupancy measures in whole putamen and caudate. The datapoints can be described with a power
law fit (a(6pea)b), such that CBVPeak changes exhibit a monotonically decreasing
function as occupancy increases in the case of the D2/D3 agonist challenge.
BPND values at baseline (before injection of the challenge) were 5.5 t 0.3 for putamen and 4.7 t 0.6 for caudate in animal M2. For the three dose injections of 0.1, 0.2
and 0.3 mg/kg quinpirole, binding potential measures decreased to DBPNPeDa
of 4.0, 2.5 and 1.8 in the putamen and 3.3, 2.2 and 1.3 in the caudate, respectively.
Peak occupancy values were thus 29%, 56% and 65% in putamen and 32%, 58%, 68%
in caudate for the three doses.
CBV peak values for the three doses (0.1, 0.2, 0.3 mg/kg quinpirole) were -5.1%,
-7.1% and -8.0% in putamen and 6.5%, -8.7% and -9.1% in caudate.
As already
visible on the voxelwise CBV maps, the values from caudate were larger in all cases
compared to putamen.
102
xin
0.3 mg/kg
0.2 mg/kg
0.1 mg/kg
4
4
x 10
o Caudate TAC
Cerebellum TAC
-- Dynamic SRTM fit
x 104
CIV
5
5
0
20
5
40
10
60 80 OC10
nd)0 20 40 60 80 100
tie
time(mm)time
(min)
)
-5
20
40 60 80 100
time (min)
5
0
0
ati
0
-
50.
-10
-1
0
20
40 60 80 100
time (mi)
0
20
40 60 80 100
time (min)
0
20
40 60 80 100
time (min)
Figure 5-7: Top row: PET time activity curves for the caudate and cerebellum ROIs
for three different doses of quinpirole injection at 35 min (for animal M2), with kinetic
modeling fits from SRTM2 with cerebellum as the reference (black line). The dotted
line at 35 min. indicates the time at which the quinpirole challenge was administered.
Bottom row: Corresponding CBV timecourses that show the negative response due
to the challenge in the caudate ROL.
5.4.5
Temporal Dissociation between PET/MR Timecourses
Temporal Response from Quinpirole Dose Injections
Figure 5-7 shows the PET time activity curves (TACs) for caudate and the reference
region cerebellum for the three doses of quinpirole injection for animal M2 (marked
as the black dashed line), together with the corresponding kinetic modeling fits from
dynamic MRTM2.
In all cases, a sigmoid function provided better fits to the data
compared to a gamma function, according to Akaike criteria. In all cases, the DBP
measures stayed decreased for the duration of the experiment.
For each of the three dose injections, the corresponding/simultaneously acquired
%CBV timecourse are shown in the bottom row. The %CBV signal change peaks at
2.2, 1.0 and 2.0 min for the three doses of 0.1, 0.2 and 0.3 mg/kg quinpirole injections,
and returns to baseline (< 0.1%) within 14.0, 5.8 and 15.2 minutes. In contrast to this
103
n
-10
0
20
40
2
60 80 100 120 140 160
time (min)
Figure 5-8: CBV timecourses in the caudate ROI due to a repeated injection of
quinpirole at 0.2 mg/kg. The first dose, administered at 30 min, elicits a negative
CBV response. But a second dose given 100 min later at 130 min does not elicit any
CBV response.
very fast CBV response, the PET signal shows a signal reduction in specific binding
regions that remains suppressed for the duration of the experiment.
Repeated Quinpirole Injections
A second quinpirole injection (for the 0.1 and 0.2 mg/kg doses) was administered
2 hours after the first dose to test if the first response could be replicated or had
been altered. In both experimental sessions with the two doses, the second injection
showed no detectable change in the fMRI signal. The timecourse of the measured
fMRI signal with the second injection of quinpirole for the 0.2 mg/kg dose is shown
in Figure 5-8.
Temporal Response from Ropinirole Injections
The temporal response from a 0.1 and 0.3 mg/kg ropinirole injection showed robust
negative responses in the striatum. While the signal also peaked quickly, at 3.4 and
2.6 min after the injection for each dose, the return to baseline was slower compared
to the quinpirole fMRI data: The CBV response due to the 0.1 mg/kg ropinirole dose
lasted 26.3 min and that due to the 0.3 mg/kg dose lasted 19.6 min. The duration of
the CBV signal was defined as starting with the onset of the GLM regressor until it
returned to < 0.1% CBV signal change.
104
5.4.6
Estimation of Internalization Constants
Fitting our proposed internalization model to the data, we estimate in vivo internalization constants kit to be between 3 to 9 min for the D2/D3 agonist quinpirole.
Estimates of kit for the less potent I)2/D3 #gonist ropinirole are 12 to 19 min.
5.5
Discussion
In summary, we have shown that pharmacological doses of the D2/D3 agonist quinpirole elicited dose-dependent decreases in D2/D3 receptor occupancy, together with
decreases in CBV in the basal ganglia of anesthetized non-human primates. Despite
the correlation of the signals in space, we observed a temporal dissociation of the
signals measured from PET and fMRI: While receptor availability measured with
[ 1 C]raclopride stayed decreased for a prolonged amount of time, CBV signals were
very short and returned to baseline much faster. Simulations of the proposed model
that includes receptor internalization are in agreement with the experimental data,
suggesting that receptor adaptation mechanisms can be observed with PET/fMRI
measurements.
The spatial correlation between changes in D2/D3 receptor occupancy and changes
in CBV support a neurovascular coupling mechanism during receptor-specific activation. This relationship holds both for D2/D3 agonists, as shown in this study, as well
as for D2/D3 antagonists, as demonstrated in Chapter 4, and can be described by
the neurovascular coupling model in Equation 5.1. The dose-dependent relationship
observed with agonist and antagonist exposure provides further evidence of a coupling
mechanism.
By combining measurements from both agonists and antagonists, an estimate of
basal DA occupancy may be calculated. The CBV-occupancy plot in Figure 5-6 shows
that putamen consistently showed a smaller CBV response compared to caudate,
whereas the signal magnitudes in these two ROIs are reversed for antagonist data
(see Figure 5-9). If the assumption is made that peak CBV changes are directly
related to occupancy as described in the neurovascular coupling model, the change
105
Antagonist
IR I nln
f
Agonist
o mn&r
Figure 5-9: Voxelwise maps showing a direct comparison of the PET and CBV response in putamen and caudate for pharmacological challenges with a D2/D3 antagonist (raclopride, left) and the D2/D3 agonist (quinpirole, left). The differences in
the signals can give an estimate of basal dopamine occupancies in vivo.
in CBV due to a D2/D3 agonist is directly proportional to (1 - 0(0)). Together with
data from antagonist studies, one can calculate the basal dopamine occupancies in
the putamen and caudate ROIs. Based on our data, the basal dopamine occupancy
is 20% in putamen and 10% in caudate. This is consistent with measurements in the
literature [83, 84, 85].
PET and fMRI signal changes due to the D2/D3 agonist injection showed diverging
temporal responses. On the one hand, the CBV timecourse seemed to be much shorter
compared to the kinetics of the agonist quinpirole [97]. On the other hand, decreases
in occupancy measured with ["C]raclopride seemed to last much longer. Since we
have previously shown that PET and fMRI temporal responses are matched for the
case of a D2/D3 antagonist, the temporal divergence in the case of an agonist suggests
that additional physiological changes occur. The D2/D3 agonist quinpirole has been
shown to induce receptor internalization in vitro [95]. Moreover, this desensitization
mechanism occurs rapidly, with an internalization constant of 5 min. Several PET
studies have shown evidence that the measured PET signal is affected by receptor
trafficking and remains suppressed for several hours [96]. This is consistent with our
PET data. If receptors desensitize, our model predicts that the fMRI response will be
abbreviated as desensitized receptors become functionally inactive. Our fMRI data
106
are consistent with such a model of receptor desensitization. We propose that receptor
adaptation mechanisms have opposite effects on the observed duration of PET and
fMRI signals. Integrating complementary information from both modalities in this
way allows us to learn more about the underlying biochemical processes of brain
function.
We investigated a number of features within our proposed model of receptor internalization. For example, internalized receptors can display altered affinity (e.g. a
lower affinity). Such low-affinity receptors are usually still accessible to antagonists
like raclopride but are generally assumed not to bind endogenous agonists. It is, however, unknown whether quinpirole can cross the cell membrane and still bind to these
receptors. This should not affect the CBV signal as internalized receptors are functionally inactive, whether bound or not. But the PET signal can be influenced with a
change in affinity for raclopride (Figure 5-4): If affinity of internalized receptors does
not change, then the PET signal is a measure of total receptor availability. If the
affinity is reduced, occupancy may seem lower, and if affinity is increased, occupancy
may even be observed to increase. The latter scenario could occur if the radiotracer
enters the cell membrane but then accumulates inside the membrane, thus effectively
increasing the affinity.
The rates of receptor desensitization are challenging to measure in vivo, but we
showed that we can probe for this mechanism with dynamic PET/fMRI measurements. Comparing our model to the experimental data, we found plausible rates of
internalization between 3 - 9 min. This is consistent with in vitro measurements of
internalization rate constants for quinpirole on the order of 5 min [95].
Once receptors are internalized, the desensitization state may be preserved for
hours or more. This is supported by the observed prolonged decrease in availability
of D2/D3 receptors as observed by PET, together with the fact that receptor function did not recover with a second injection of the D2/D3 agonist quinpirole after 2
hours. These findings are in agreement with electrophysiological recordings, in which
reapplication of quinpirole fails to initiate a second response in dopaminergic neurons of the ventral tegmental area [98]. Contrary to D1 receptors, D2 receptors have
107
been suggested to undergo degradation post-endocytosis, which may explain the long
recovery period needed for resensitization of D2 receptors.
To look at effects that can alter internalization rates, we used the D2/D3 agonist
ropinirole, which is seven times less potent compared to quinpirole [99]. We hypothesized that ropinirole reduces the rate at which receptors desensitize. Our CBV results
show a somewhat slower return to baseline for ropinirole, suggesting that receptor desensitization is not as pronounced compared to the highly potent quinpirole, which is
often used a reference for evaluating the efficacy of other D2/D3 agonists [100].
5.6
Conclusions
We investigated the effects of pharmacological doses of D2/D3 agonist injections in
anesthetized non-human primates and proposed a model that describes receptor internalization and its effect on PET/fMRI data. Consistent with the proposed model,
we found that PET and CBV timecourses diverge, with PET specific binding staying
suppressed for a prolonged amount of time and with CBV signals being shortened.
Experimental data with a D2/D3 agonist showed that PET and fMRI signals match in
anatomical space and with injected dose, as predicted through the proposed neurovascular coupling model. Overall, we demonstrated a model together with experimental
data that provides first measurements of receptor trafficking dynamics in vivo with
simultaneous PET/fMRI. Characterizing receptor adaptation mechanisms in vivo has
the potential to inform treatment plans, especially for the use of agonists, and expand
our understanding of the long-term effects of drug exposure.
108
Chapter 6
Effects of Flow Changes on D2/D3
Radiotracer Dynamics
6.1
Introduction
The influence of changes in radiotracer delivery on the dynamics in receptor-specific
PET studies has been widely discussed in the literature [70, 24, 25]. Pharmacological challenges or activation paradigms are known to cause large and rapid changes
in blood flow, as shown with BOLD or CBV measurements in functional magnetic
resonance imaging studies [101, 76]. In PET, such pharmacological challenges are
applied to measure the reduction in PET signal that is attributed to radiotracer displacement. However, if the delivery and washout of the tracer is altered during the
timeframe of a PET study, it is not clear to date whether and how this may affect
measures of receptor binding.
Several varying assumptions have been made to simulate the precise mechanism
of how blood flow affects the kinetic parameters in a compartmental model. Some
studies have argued that blood flow changes affect both the plasma-to-tissue transfer
rate K1 and the tissue-to-plasma transfer rate k 2 in the same manner [70, 24]. In this
scenario, small changes of up to 25% were simulated as having a negligible effect [25].
A different assumption is to decouple K1 and k2 , which results in more pronounced
changes in simulated time activity curves (TAC) due to a change in K1 or k 2. But
109
it remains questionable whether the latter scenario is physiologically relevant [24].
Hence, the conclusions drawn from simulations of changes in blood flow have been
controversial.
One of the main questions is whether specific binding measures are affected by
varying values of blood flow. One experimental investigation is reported [102], in
which hyperventilation was used to decrease blood flow artificially while evaluating
receptor-specific PET data from [ 1 1C]raclopride. While changes in the overall shape
of TACs were observed, an overall decrease in blood flow for the duration of the
experiment did not seem to affect binding potentials.
Radiotracers that have a large extraction fraction are often assumed to be more
flow-sensitive, one example being [1 8F]fallypride.
Cumming et al. postulated that
differences in binding potentials are due to differences in flow values in healthy volunteers [103]. [18F]fallypride has also been used for dopamine displacement studies
and characterization of extra-striatal receptors.
PET displacement studies using a within-scan challenge have been adopted in
several study designs, together with respective methods for analysis [74]. In this context, the dopamine D2/D3 receptor system has been a popular target, partly due to
kinetically favorable and well-characterized radiotracers. It is usually assumed that
the kinetic parameters of the radiotracer are constants throughout a study, except
for the binding term k3. Yet, we know that pharmacologic challenges, such as amphetamine, can produce relatively large transient changes in flow [101], so that the
assumption on constant rates may not hold true.
Large flow changes of up to an
increase by 2-fold from baseline have not been explicitly investigated and could have
a more pronounced effect on radiotracer dynamics compared to more moderate flow
changes as previously investigated.
The goal of this study was to experimentally measure controlled changes in blood
flow and their influence on PET time activity curves (TACs) with simultaneous
PET/MRI. Flow was increased by inducing hypercapnia for specific periods of time
during the dynamic PET acquisition and was simultaneously measured with arterial spin labeling.
Two tracers specific to the dopamine receptors D2/D3 were
110
investigated:
["C]raclopride, a well-validated tracer for displacement studies, and
[' 8 F]fallypride, a tracer previously suggested to be susceptible to flow changes. We
carried out simulations of flow changes and compared this to experimental results,
thus providing insight into the assumptions used for the kinetic parameters affected
by flow.
6.2
6.2.1
Methods
Animal Model
Imaging studies were carried out in two baboons, both with a weight of approximately 10 kg. Initial anesthesia was induced with 10 mg/kg ketamine and 0.5 mg/kg
xylazene, and maintained during the scan by isoflurane (1%, mixed with air). Baboons were ventilated throughout the entire study and physiological changes in blood
pressure, pulse, end-tidal CO 2 and breathing rate were monitored during the experiment. All procedures complied with the regulations of the Subcommittee for Research
Animal Care at Massachusetts General Hospital.
6.2.2
Study Design with Hypercapnia
To induce hypercapnia (and thus increases in flow), animals were ventilated with 7%
CO 2 mixed with air for a period of 10-15 min at a time. This exposure to 7% CO2
was repeated in a block design fashion, interleaved with a period of 10-15 min of
ventilation with air. The onset of the first hypercapnia challenge was timed to occur
either 5 min before or 20-30 min after injection of the radiotracer. The first scenario
allows measurements of the effect of flow changes during the uptake of the tracer,
whereas the latter mimics the timing normally used for within-scan pharmacological
challenges.
111
6.2.3
PET/MR Image Acquisition
Simultaneous PET and MR data were acquired on a whole-body human PET/MR
scanner at 3 T (mMR, Siemens AG, Healthcare Sector, Erlangen Germany).
A
custom-built PET compatible 8-channel NHP receive array together with the body
coil for transmit was used for MR imaging. For quantitative flow measurements,
pseudo-continuous arterial spin labeling (pcASL) with whole-brain coverage was employed for the entire duration of the experiment (FOVead = 128 mm, resolution =
2x 2 x 4 mm, TE = 15 ms, TR = 3.5 s, post-labelling delay = 800 ms, BW = 1474 Hz).
The phased-array receiver enabled a two-fold acceleration with GRAPPA [32] in the
anterior-posterior direction. PET emission data were acquired in list-mode format
for 90 min ([' 1 C]raclopride) or 180 min ([ 18 F]fallypride) starting with ligand injection. Images were reconstructed with the ordinary Poisson expectation maximization
algorithm with 3 iterations and 21 subsets. Corrections for scatter and attenuation
of the head and the radiofrequency coil were applied during reconstruction. The final
reconstructed volume consisted of 4 x 4 x 2 mm voxels in a 172 x 172 x 127 matrix,
with framing intervals of 10 x 30 sec, followed by 1 min frames.
6.3
6.3.1
Results
Simulation- Results
We simulated the effect of flow changes on radiotracer for a full reference tissue
compartmental model [23]. Flow was assumed to be related to K1 by K1 = extraction fraction x flow and k 2 was assumed to change proportionally with K 1 , so that
the volume of distribution remained constant. The graph in Figure 6-1 shows the
simulated TACs for ["C]raclopride for specifically bound tissue tracer and the free
fraction in tissue. The corresponding change in the rate constants that were simulated are shown in the bottom graph and compared to a PET baseline TAC without
flow changes. While small deviations from the PET baseline curve were observed due
to the simulated flow changes, a kinetic modeling analysis of the simulated curves
112
Free ligand
-Specifically
bound
-PET
-PET
baseline
30E20-
10
cc
0
0
10
20
30
40
50
60
70
80
90
K
~Kref
0.
.---
1
time (in min.)
Simulations of a reference tissue model with kinetic parameters for
["C]raclopride. Upper graph: Time activity curves for a specific binding region,
the free ligand and the computed PET signal show very small deviations compared
to a baseline PET curve. Lower graph: Timecourse of simulated rate constants K 1
k2 , and Klref (for the reference region, in min- 1 ) for an doubling of flow starting at
,
Figure 6-1:
20, 40 and 60 min.
with a simplified reference tissue model [1] did not show differences in BPND due to
simulated flow changes.
6.3.2
Experimental Results: ["C]raclopride
Figure 6-2 shows the experimental results for the entire duration of 90 min from a
representative scanning session with ["C]raclopride. The first graph shows the time
activity curves (TACs) for the putamen and cerebellum for an experiment, in which
hypercapnia was applied in an ON-OFF block design, with each condition lasting 10
min. Hypercapnia was applied three times for a duration of 10 min, starting at 20,
40 and 60 min after tracer injection, as indicated by the blue-shaded areas on the
graph. The TACs do not show any prominent changes due to a change in flow.
The kinetic modeling fit for the putamen ROI from SRTM2 is shown in purple.
The second graph in Figure 6-2 shows the residuals between the fit and the data, but
113
-
::-60
e
.E
WSI
w
40%CO
S20
%CO
PET utamen
PET Drebellumn
%CO
0 or
5
+
+
+~0#+
4F++
+
+
+
+
-5
_250 --o-20--100
-
030
ASL Putamen
ASL Cerebellum
150
-
+44
0
0
U
W200
20
40
time (min)
60
80
Figure 6-2: Experimental results from a [ 1 C]raclopride bolus injection and hypercapnic challenges of 7% CO 2 applied three times during the experiment. Upper graph:
Time activity curves for putamen and cerebellum show no effects due to hypercapnia
(timing indicated by blue-shaded areas). Upper middle: Residuals (in kBq/ml) between the kinetic modeling fit and data. Lower middle: Measured flow changes with
pcASL (MRI) show increases in flow in the putamen and cerebellum ROIs. Lower
graph: End-tidal CO 2 values for the duration of the experiment.
no periodic changes that match the flow pattern are detectable.
Increases in cerebral blood flow, as measured by pcASL, in response to the hypercapnic challenges were measured in all regions, albeit with different magnitudes as
shown for the putamen and cerebellum ROIs in the third graph of Figure 6-2. Flow
114
increased up to 110% in the putamen and up to 245% in the cerebellum, relative
to baseline values. Quantitative analysis of the pcASL data in the putamen showed
that flow increased from a baseline value of 59 ml/100g/min to 114, 116 and 119
ml/100g/min for the first, second and third hypercapnic challenge, respectively. In
the cerebellum, baseline values of flow were lower (37 ml/100g/min) but increased to
96, 111 and 126 ml/1OOg/min for the three hypercapnic challenges.
End tidal CO 2 (etCO 2) measures (fourth graph in Figure 6-2) correlate well in
time with the measured changes in blood flow by pcASL. Baseline values started
at 35 mmHg and increased by a maximum of 20 mmHg to 55 mmHg during the
administration of hypercapnia.
Quantification of the ["C]raclopride PET data resulted in binding potentials
BPND =
4.5 and BPND = 5.7 for two experimental runs, in which hypercapnia
was applied starting at 20 min after injection of the radiotracer. In order to compare
values to a no-flow-change condition, a kinetic analysis was performed that took into
account only the first 20 min of the data. The corresponding estimates for binding
potentials are BPND = 4.6 and BPND = 5.7, which are very similar to the BPND
values that take into account the full 90 min of the PET data for quantification.
Moreover, a time-dependent flow or k2. term did not improve the model fit.
6.3.3
Experimental Results: [ 18F]fallypride
In all [18F]fallypride experiments, TACs did not show any noticeable changes that
match to the pattern of flow applied.
Figure 6-3 (first graph) shows the experimental results for applying five hypercapnia challenges in an ON-OFF design over the timecourse of the experiment of 180
min. Each hypercapnia challenge lasted 12 min, with onsets at 30, 60, 90, 120 and
150 min. of the experiment (after injection of the radiotracer), as indicated by the
green-shaded areas of 7% CO2 exposure.
Relative blood flow values measured with pcASL are shown in the second graph in
Figure 6-3 for the putamen and cerebellum ROIs. In the putamen, flow increased by
up to 178% and in the cerebellum flow increased by up to 246% during hypercapnia.
115
E
PET Putamen
PETmm
40 -
930
7%
20i
.2
s
C
ocerebelu
7%
7%
7%
1
ASL Putamen
ASL Cerebellum
-
.2 150
-
a100
50
60-
W20-
0
20
40
60
80
100
time (min)
120
140
160
180
Figure 6-3: Experimental results from a ["Ffallypride bolus injection and hypercapnic challenges of 7% C2 applied fiv eti
r
ig
experiment. Upper graph:
Time activity curves for putamen and cerebellum are not noticeably affected with
hypercapnia (timing of challenge indicated by shaded areas). Middle graph: Measured flow changes with pcASL (MRI) show increases in flow in the putamnen and
cerebellum ROls during hypercapnia. Lower graph: End-tidal C02 values for the
duration of the experiment.
Initial etC02 values were 32 mmHg and increased to a maximum of 52 mmHg
during hypercapnia, as shown in the third graph of Figure 6-3.
The timing of the hypercapnia challenge was also altered to start before and during
injection of the radiotracer, such that flow effect during radiotracer update could be
assessed. The TACs for [ 18F]fallypride from an experiment, in which hypercapnia was
applied starting 5 min before radiotracer injection, are shown in Figure 6-4. Despite
the change in the timing of the challenge, no changes in the shape of the TACs were
observed.
Changes in flow, measured by pcASL, and etCO 2 values measured throughout
the timecourse of the experiment are shown in the second and third graph in Figure
6-4. The flow measurements show that flow was increased by up to 145% in the
116
40-
7% C02
7% CC2
SPET Putamer
30k~
AW
PET Cerebell m
-
20
-
10
2000
-+-
ASL Putamen
- 100-
CM
0
0
60-
40-
20
Lu
20-
-40
20
0
-20
60
40
time (min)
80
100
120
140
Figure 6-4: Experimental results from a [' 5F]fallypride bolus injection and hypercapnic challenges of 7% CO 2 applied before the start of the experiment to observe
flow-induced uptake differences. Upper graph: The shapes of the time activity curves
for putamen and cerebellum are not changed with hypercapnia (timing of challenge
indicated by shaded areas). Middle graph: Measured flow changes with pcASL (MRI)
show increases in flow in the putamen ROI during hypercapnia. Lower graph: Endtidal CO 2 values for the duration of the experiment.
putamen during hypercapnia. These values corresponded well to the changes observed
in etCO 2 , as shown in the graphs.
The kinetic modeling results for [ 18F]fallypride with the SRTM2 model did not
show changes outside a normal test-retest range. For the experiment in Figure 6-3,
parameter fits yielded BPND= 16.0 for the putamen. For the experiment in Figure
6-4, we found that BPND
=
14.7. A baseline study with no hypercapnia applied in
the same animal gave a BPND of 15.6. In addition to modeling the full curve, we
also applied an analysis that only took into account the first 30 min of the data.
Although all BPND values were higher compared to taking into account the full curve
for modeling (a statistical bias occurs in these cases), the values were all within the
same range: BPND
=
16.1 at baseline and BPND = 16.6 for the experiment in Figure
6-3 and BPND =15.2 for the experiment in Figure 6-4.
117
6.4
Discussion
In this study, we investigated the effect of increasing blood flow on radiotracer kinetics. By applying hypercapnia challenges, cerebral blood flow increased more than
2-fold in anesthetized NHPs. Simultaneous acquisition of dynamic PET data and
pseudo-continuous arterial spin labeling with MRI enabled the measurement of radiotracer kinetics together with induced flow changes. For the investigated radiotracers [ 1 C]raclopride and [ 18F]fallypride, flow changes did not change the shape of the
time activity curves, independent of the timing of the hypercapnic challenge. This
suggests that flow does not affect the kinetics of these radiotracers and that changes
in radiotracers may have less of an impact than sometimes assumed.
Hypercapnic challenges were employed since hypercapnia produces large flow changes
that generally exceed flow changes occurring in the brain under normal conditions.
Hence, this design allowed for an assessment of flow effects in a worst-case scenario.
We thus expect that our results are valid for most conditions that increase flow, including pharmacological challenges, which generally do not increase flow more than
50%.
The primary goal of this study was to investigate the effect of flow changes that
are known to occur during pharmacological challenges.
From fMRI studies (e.g.
Chapters 4 and 5) and correlations between CBF and CBV [104], we know that flow
changes induced by injected drugs can rapidly increase flow (within minutes) and the
timecourse may depend on the half-life of the drug or other adaptation mechanisms
(see Chapter 5). In PET competition studies, injection of pharmacological challenges
can be given before, with or during the radiotracer administration. The timing of a
flow change may be important because varying parts of the radiotracer curve may be
affected differently: During uptake of the radiotracer, i.e. at the beginning, kinetics
are dominated by K1 and k 2 ; but during the washout phase, kinetics are dominated
by k3 and k4 . In our experiments, we thus varied the timing of the challenge in order
to address these potential differences. Yet, we did not detect any effects of flow on
the time activity curve or with radiotracer quantification.
118
We used bolus injections to test for the effects of flow on radiotracer quantification.
The results for infusion, or bolus plus infusion paradigms are expected to be the
same. Both with infusion and bolus plus infusion, steady-state is reached after an
initial equilibration period. During steady-state, flow changes are not expected to
affect tracer delivery or washout because the blood and tissue concentration remain
at steady state.
Our simulation results showed that large flow changes had very little effect on the
PET TACs, especially when noise was included in the model. Moreover, doubling flow
did not have effects on measures of binding potential. This is in general agreement
with simulation results in the literature that assume that flow changes affect both
K1 and k 2 to the same proportion [70, 24]. Simulated flow in previous literature was
generally smaller (10% to 50% increases in K1 and k 2), which may be the reason
why no changes due to flow were detected in TACs, contrary to small changes we
observed. Nevertheless, our experimental results suggest that an accurate model for
flow changes includes a proportional change in both K1 and k2 . This is in contrast
to previous findings where only one of these parameters was changed [25].
Pharmacological challenges usually produce flow patterns that differ in regions of
the brain. Especially when using reference region models for a kinetic analysis in
PET, differences in flow between the two regions can potentially affect the estimation
of parameters. In the data presented here, we observed differences in flow between
the striatal regions (specific binding region) and the cerebellum (reference region),
which suggests that the flow changes we induced are a good model for addressing
the question of pharmacological challenges that might act more locally. In addition,
since we did not observe any alterations in the TACs due to a change in flow, our
results can be applied even for competition studies, in which only specific regions of
the brain are targeted.
119
6.5
Conclusions
In summary, we have experimentally investigated the effect of flow changes on PET
radiotracer kinetics using simultaneous measurements of PET and MRI. Even with
very large flow changes induced with a hypercapnic challenge, ["C]raclopride and
[18 F]fallypride TACs did not show any changes, indicating that flow does not influence
these radiotracers. The flow-insensitivity of ["C]raclopride validates our findings
from Chapters 4 and 5. Our findings are important for kinetic modeling approaches
and their assumptions on flow. Furthermore, our results show, for the first time,
experimental validation of the effect of flow for within-scan challenges and provide a
framework for testing the flow effects of other existing or novel radiotracers.
120
Chapter 7
Conclusions and Future Work
7.1
Summary
In this thesis, we presented technical developments, methods and models for imaging
neuroreceptor function with simultaneous PET/fMRI. The techniques were applied to
measure the dynamics of dopamine receptor function in the brain, demonstrating that
we can establish and quantify relationships between D2/D3 receptor occupancy and
hemodynamics. Our findings show, for the first time, the relationship between functional dynamics and receptor-specific binding and provide insights into neurochemical
parameters and neuronal adaptation mechanisms, such as receptor internalization.
As a first step, we successfully designed, built and evaluated a 31-channel parallel
phased array for a prototype simultaneous PET/MR scanner to enable highly sensitive dual-modality imaging. We developed general design criteria for PET compatible
coils that can be applied for coil designs in newer generation combined scanners. Additionally, we developed two PET compatible coils for the purpose of PET/MR imaging
of non-human primates. With this technology, we are able to achieve extremely stable
and fast, high resolution MRI data together with highly sensitive PET measurements,
enabling the functional imaging applications presented in this thesis.
Second, we employed simultaneous neuroimaging with PET/fMRI to demonstrate the relationship between changes in receptor occupancy measured by PET and
changes in brain activity inferred by fMRI. By administering the D2/D3 dopamine
121
receptor antagonist ["C]raclopride at varying specific activities to anesthetized nonhuman primates, we mapped associations between changes in receptor occupancy and
hemodynamics (cerebral blood volume) in the domains of space, time, and dose. We
showed a monotonic coupling between vascular responses and receptor occupancies,
and proposed a neurovascular coupling model that relates dopaminergic occupancies
to hemodynamic changes in the basal ganglia. Overall, we demonstrated the utility
of simultaneous PET/fMRI for investigations of neurovascular coupling that correlate
neurochemistry with hemodynamic changes in vivo for any receptor system with an
available PET tracer.
Third, we expanded on the above results to compare PET/fMRI signals with a
dopamine D2/D3 receptor agonist, thereby devising new methods to image neurobiological changes at receptors. Mechanisms such as receptor internalizations have been
shown in vitro but, due to lack of in vivo techniques, had not been investigated in
the context of in vivo synaptic activity. We showed that receptor trafficking causes
PET and fMRI signals to diverge in the temporal domain, both with a model and
experimentally, using graded doses of the D2/D3 agonist quinpirole. These results
axe valuable as they provide a way to quantify rates of receptor internalization in
vivo, a measurement that was only possible in vitro before.
Fourth, we validated the findings from our simultaneous PET/fMRI studies above
by identifying how blood flow changes affect radiotracer quantification. By inducing
controlled flow changes through hypercapnia and measuring these flow changes and
dopaminergic radiotracer kinetics at the same time, we showed that flow does not affect PET quantification. With these results, we were able to experimentally provide
insight into the long-debated question of flow effects in the PET literature, and propose methodologies that can be used to experimentally verify any kinetic modeling
predictions for other existing and novel radiotracers.
Overall, the results of this thesis advance the field of simultaneous PET/fMRI
through hardware developments, imaging methodology, biological models and experimental results. By characterizing the dynamic relationship between the functional
output reported by fMRI and dopaminergic occupancy due to selective ligands in
122
vivo, we demonstrated how imaging the dynamic neurochemistry of the brain and its
connections and communication pathways will enable us to understand the action of
any neurochemistry-modulating drug.
While this is the closing chapter of this thesis, we are excited that the findings
discussed here may only be the beginning of a new research area that can have an
impact on how we understand the distributed function of the brain and its underlying
neurochemical dynamics. The concepts originating from this thesis can be translated
both to other neuroreceptor systems and human brain imaging applications, including
the study of receptor-specific functional activation of the brain in health and disease.
The following section outline some future potentials and applications of the topics
presented in this thesis.
123
7.2
Imaging Neuroreceptor Dynamics
7.2.1
Imaging other Neuroreceptor Systems
In this thesis, we have demonstrated a relationship between receptor occupancy and
function at the D2/D3 dopamine receptor system. However, we would expect other
receptor systems to show similar neurovascular coupling relationships.
Interplay between the D1 and D2 Dopamine Receptors
While the D2-like dopamine receptors are inhibitory and thus produced a negative
cerebral blood volume response in our studies, the Di-like receptors are excitatory.
PET radiotracers for D1 receptors are available and well-characterized, with the main
examples being the Dl antagonists ["C]NNC112 or ["C]SCH23390. Pharmacological
challenges with antagonists or agonists (such as the drug SKF81297) at the D1 receptor can expand our understanding of the overall function of the dopamine receptor
system. Initial results of studies targeting the D1 receptor system are in line with
the findings we presented in the thesis here: At D1, the agonist SKF81297 produced
a positive CBV response [105].
Endogenous dopamine and other dopaminergic drugs have affinities for both the
D1 and the D2 receptor.
The interplay between the excitatory and inhibitory re-
sponses in the dopamine system are not well understood to date. However, a homeostatic balance of these responses may be crucial for healthy brain function. Literature
suggests that dopamine has a higher affinity for the D2 receptor than the D1 receptor [106], such that fMRI responses may be weighted by the ratio of the affinities.
We proposed a model for this interaction, together with some experimental evidence
through amphetamine challenges, which releases endogenous dopamine [76].
By characterizing the dynamics and functional effects of the dopamine receptor
subtypes in a bottom-up approach, we will gain insights into complex drug action
at the dopamine system. In this way, drug treatments can be optimized to target a
number of receptors in a balanced way to maintain a healthy homeostasis of excitatory
and inhibitory signals in the brain.
124
7.2.2
Imaging Neurochemical Connectivity
Imaging whole-brain molecular dynamics through functional and structural connectivity is a promising way to understand both neurotransmitter and drug pathways.
Imaging functional connectivity with MRI (fcMRI) is mainly performed while the
brain is at "rest" (hence also its name resting-state fMRI). Because brain activity is
present even in the absence of a specific task or stimulation, brain regions show spontaneous fluctuations. Methodologies in fcMRI have revealed a number of networks
that exists, such as the default mode network.
The most widely used techniques for analyzing fMRI data for functional connectivity rely either on independent component analysis or a seed-based analysis. The
latter usually provides a way to look at specific connectivity in brain areas of interest.
However, choosing a seed is not always trivial. Here, PET data can give guidance on
how to choose a seed and one can use priors from PET to target a neurochemicallyspecific analysis. One example is to use the raphe nucleus as the seed region, a small
structure in the brain stem known as the site of serotonin release to the rest of the
brain. Connectivity analyses can also be carried out on dynamic PET datasets, especially in datasets where evolving PET technology has increased the sensitivity of
detectors and shorter timeframes for PET data can be used. Integrating fluctuations
in receptor-specific networks with functional connectivity fMRI data can yield insight
into the role of receptor dynamics for networks of the brain.
7.2.3
Development of Biological Models
Non-invasive imaging with PET and fMRI provides crucial insights into the brain
function in vivo, but transferring knowledge from other techniques and building physiologically relevant models is equally important.
Quantification of PET data relies on building kinetic models. In order to understand the true dynamics of receptor-specific activation, it is necessary to expand
these models to include a dynamic description of data. However, attention needs to
be paid not to over-estimate the data by trying to fit too many unknown variables.
125
Nevertheless, there are experimental methods (e.g. constant infusions, double tracer
injections) and interventions (such as dynamic ON-OFF stimuli) that can help in the
evaluation of dynamics [107].
Invasive techniques, such as microdialysis, and in vitro measurements can measure biological parameters that are not accessible with in vivo imaging techniques.
For example, we rely on in vitro data to measure the affinity of drugs to a receptor.
Building on this knowledge, we can devise biological models that predict the functional responses we expect to see in imaging data. Moreover, features and outcomes
of models that integrate information from other research fields can be validated from
many angles, and can break down various aspects that contribute to the complexity
of brain function.
7.3
Clinical Applications of PET/fMRI
Clinically, simultaneous PET/fMRI can have a significant role in the diagnosis and
monitoring of therapy for psychiatric or neurologic disease. While the studies presented in this thesis were mainly carried out in a non-human primate model and
served as basis for developing the imaging methodologies, many methods are readily
translatable to humans.
7.3.1
Evaluation of Psychiatric Drugs
Drug development for treatment of psychiatric disease is a long and costly process.
Often, a large pool of potential ligands is evaluated first in vitro, then in animal models
in vivo and finally in human clinical trials. Parameters that are used for screening of
initial ligands include specificity and affinity to the target. However, efficacy of the
drug can often only be evaluated at a much later stage when ligands have undergone
a long chain of validation procedures. Using PET/fMRI as a screening technique,
we can start to predict functional outcomes in animal models. This can provide an
early understanding of agonist, partial agonist or antagonist properties of the ligand
in question and help to make the screening process more efficient.
126
The exact mechanism of action of existing psychiatric drugs is often not known.
Although targets and affinity are characterized, there is little knowledge about the
links between drug targets and symptomatic relief. D2/D3 antagonists have been
used for treatment in schizophrenia or migraines. The administration of drug doses is
largely determined by the therapeutic window nowadays, which is solely dependent on
occupancy measures that target an occupancy between 70 -80%. However, with some
drugs, it may be possible that the downstream functional effects are a useful measure
of symptom relief and lower occupancies are enough to achieve the desired outcome.
An assessment with PET/fMRI for optimal dosing strategies could be crucial when
trying to avoid side effects that often appear at high occupancies. Being able to
characterize the nuances in functional effects due to exposure of different drugs with
PET/fMRI may lead to more targeted and personalized treatment plans.
7.3.2
The Dopamine Hypothesis in Schizophrenia
In Chapters 4 and 5, we proposed methods for determining basal dopamine occupancies in vivo. Basal dopamine occupancies seem to play a crucial role in schizophrenia:
With (relatively invasive) studies in patients suffering from schizophrenia, it was found
that basal dopamine is elevated [107]. Perhaps even more important, a decrease in
basal dopamine occupancies seems to be correlated with a decrease in the positive
symptoms in schizophrenia through drug treatment [107]. Current measurements of
basal dopamine occupancies in vivo include the depletion of dopamine from its receptors, i.e. dopamine resources are depleted, such that occupancy of the total receptors
goes to zero. However, this measurement is associated with severe motor dysfunction
and psychiatric symptoms due to the low levels of dopamine, with study participants
being hospitalized for weeks after the measurement. Our proposed model with measurements from PET/fMRI may allow for the evaluation of basal dopamine levels for
larger groups of schizophrenic patients and can thus expand our knowledge on the
role of dopamine in schizophrenia.
127
7.3.3
Deep Brain Stimulation in Parkinson's Disease and Depression
Deep brain stimulation (DBS) through implantable neurostimulation electrodes is an
important symptomatic therapy in Parkinson's disease (PD). Despite its success, the
neurophysiological mechanisms underlying the efficacy of DBS remain a subject of
debate. Using methods for dopaminergic imaging with simultaneous PET/MRI may
be able to address some of these issues through dynamic whole-brain molecular and
functional imaging. Performed concurrently with DBS, PET/fMRI has the potential
to provide a new platform for enhancing treatment strategies in PD, by visualizing
both local neurochemical modulation and distant network circuit function simultaneously. An understanding of the interplay between synaptic events and function is
crucial for advancing the therapeutic use of DBS.
Furthermore, DBS is not only used to relieve symptoms of PD but more recently
has also found application in severe long-term depression, if patients are resistant to
treatment by drugs. The target for stimulation in these cases is usually towards the
ventral striatum. However, little is known about how the treatment works, the exact
location for optimal stimulation targets or the duration of treatment.
Dopamine
seems to play a role in long-term depression, but little is known about the specific
neurochemistry or connections involved in depression. Imaging studies with PET and
fMRI during DBS can provide measures of the neurochemical dynamics and can yield
insight into these unanswered questions.
7.4
7.4.1
Beyond Imaging with PET/MRI
Simultaneous Imaging with EEG-PET/MRI
To understand the full picture of functional brain activation, starting with synaptic
transmission and occupancy, we need to explore techniques beyond PET and fMRI.
To date, other non-invasive techniques exist that offer unique insights into neuronal
activation, one of these being electroencephalography (EEG). Combined EEG and
128
fMRI studies have already provided insights into the neural basis of fMRI signal. But
the combined information obtained from dense-array EEG, fMRI and PET together
can lead to advances in basic science (e.g., neuronal and neurovascular coupling with
exploration of specific receptor occupancies), and better assessment of neuropathologies (e.g., a more robust and reliable detection of epileptogenic foci compared to
EEG [108], EEG/fMRI [109] or PET [110] alone). Technologically, we have shown
that novel technology for invisible dense-array EEG caps can facilitate simultaneous
imaging with PET/MRI [111]. Using this technology for studying receptor-specific
activation while monitoring synchronized electrical activity will allow us to further
understand links in the neurovascular coupling chain.
129
130
Appendix A
PET Compatible Brain Array Coils
for Non-Human Primate Imaging
A.1
Motivation
Simultaneous imaging with MRI and PET offers the opportunity to obtain complementary functional data not only in human but also in animal models, such as
non-human primates (NHP). However, current PET/MR scanners are limited in the
selection of PET-optimized RF coils and provide only a few coils for human imaging studies. Smaller brain structures in NHP brain imaging demand higher spatial
resolution with a corresponding reduction in sensitivity from the smaller voxel size.
Thus, attempting to image small monkey head sizes in human arrays is suboptimal.
Specifically, functional imaging of NHP can show substantial degradation in SNR and
parallel imaging performance compared to coils with the appropriate size and shape
for a given monkey head. Hence, we developed two PET compatible 8-channel array
coils for simultaneous PET/MR imaging of anesthetized NHP, based on the concepts
presented in Chapter 3. Our goal is enhanced MRI sensitivity and acceleration improvements compared to commercially available human coils, with minimal impact
on the PET image quality.
131
Small monkey coil
Large monkey coil
Figure A-1: Small and large 8-channel array coils for simultaneous PET/MR imaging
of non-human primates.
A.2
Materials and Methods
Two PET compatible 8-channel coil arrays for NHP brain imaging were constructed
for use with two simultaneous PET/MR scanners (whole-body and brain-only). Both
coils were optimized to reduce 511 keV -y-ray attenuation by locating dense electronic
components (e.g. preamplifiers) outside the field-of-view (FOV) of the PET camera
and by utilizing lightweight housing materials. The monkey coils were built on two
tight fitting cylindrical helmets, with diameters of 11.5 cm and 15 cm, respectively
(Figure A-1). The small coil was targeted to closely fit macaques and small baboons
in a helmet shape, whereas the large cylindrical coil was designed to accommodate a
larger MR FOV and growth towards large baboons. The coil housing was designed in
a 3D CAD program and 3D printed using polycarbonate plastics. The monkeys lower
face remains unobstructed to facilitate anesthesia ventilation.
Coil elements were
built in an overlapped loop design using rectangular coils distributed around the
132
Small NHP coil
Large NHPcoIl
Human coil
SNR
150
100
s0
/g
80
60
40
'20
Figure A-2: Top: 511 keV attenuation map of cross-section of small and large monkey
coil and mMR human coil. Bottom: Sinograms of attenuation correction factors for
each coil.
cylindrical helmets. Low impedance preamplifiers were used to decouple next-nearest
neighbor elements. The loops were kept sparse in conductive material (18-awg thin
cooper wire, according to the design criteria presented in Chapter 3, Fig. 3-1) to
reduce attenuation properties. CT scans of the final coils were carried out to evaluate
the final coil attenuation and to create maps for attenuation correction. A bilinear
transformation was applied to calculate 511 keV attenuation maps [38] and sinograms
of attenuation correction factors were computed by applying the Radon transform.
We compared the coil performances to a commercially available 16-channel "PETfriendly" human head-neck coil available on a whole-body PET/MR scanner.
A.3
Performance of NHP coils
Figure A-2 shows the attenuation maps of a representative cross-sectional slice derived
from a CT scan of the small and large monkey coils compared to the human head-neck
coil, with the latter coil being much larger and containing more material. The bottom
133
Small NHP coil
Human coil
Large NHPcoll
0.5
0
3
0
173
50
100
150
-10
0
10
radius
-10
0
10
radius
-20
0
radius
20
Figure A-3: SNR maps (top) and 1/g factor maps (bottom) of the small and large
monkey coil compared to the human head-neck coil.
row of Figure A-2 shows the sinograms of attenuation correction factors that provide
an estimate of the total attenuation of the coil. The average attenuation coefficient
values for the small 8-channel, the large 8-channel and the head-neck coil are 1.06,
1.06 and 1.65, with maximum values of 2.8, 3.2 and 27.3.
Figure A-3 shows the SNR and g-factor maps (R=2) for both 8-channel monkey
coils in comparison to the head-neck coil. The small and large coil showed an average
SNR increase of 3.6-fold and 2.2-fold, respectively, when compared to the commercial
coil. 1/g-factor maps in Figure A-3 show substantial lower noise application using
the dedicated monkey coil with moderate acceleration (R=2).
Figure A-4 depicts the images of two simultaneous PET/MR studies that were
acquired in a rhesus macaque and a baboon with the small and large 8-channel coil,
respectively.
On the top, a summed PET image of the radiotracer ["C]raclopride
(specific to D2/D3 dopamine receptors) with the striatum clearly visible and on the
bottom, the radiotracer ["C]CW4 (specific to orexin OX2 receptors) is displayed.
134
Figure A-4: PET/MR images acquired with the small monkey coil on the BrainPET
PET/MR scanner with the radiotracer ["C]raclopride (top) and on the whole-body
PET/MR with the radiotracer ["C]CW4 (bottom).
A.4
Conclusions
We developed two array coils for PET/MR imaging in NHP studies, which provide
high sensitivity and moderate parallel imaging capabilities, with minimal -- ray attenuation. These coils are well suited for simultaneous functional PET/MRI applications
in NHPs and enabled the imaging studies presented in Chapter 4, 5 and 6 in this thesis.
135
136
Appendix B
Details of Biological Models
B.1
Relationship between Raclopride and Dopamine
Occupancies
Raclopride (RAC) competes with dopamine (DA) for binding to the D2 receptors
(D2R), such that increasing doses of RAC displace progressively more DA. To establish an approximate relationship between these occupancies, we assume transient
equilibrium conditions, which are applicable at the peak response of RAC occupancy.
Moreover, we assume that the non-displaceable tissue free fraction of [' 1 C]RAC stays
constant. Since the onset rate constant is a function of the available receptor pool
(k 3 = konBma), the equilibrium condition can be written as [112]:
(Bmx
-
BDA
-
BRAC) FDA = KD,DABDA,
where
KD,DA =
koff,DA
kon,DA
or
ODA =
(1
-
ORAC)
137
FDA
FDA + KD,DA'
,
(B.1)
.0.8
0(0.6
Co
1%
-0
---
D0.4
30%
E
1-1-60%.
z
--
SV.2rRaclopride
0
0
90%
20
occupancy
40
60
time (min)
80
Figure B-1: Simulations of a multi-compartmental model based upon first-order kinetics: Specific binding of raclopride vs. time and for increasing dose, normalized to
the maximum value in order to observe differences in shape due to receptor saturation.
with
Bmax, BDA
and
BRAC
Bmax
and
BRAC
B
ORAC
-
BDA
ODA =
Bmax
refer to the total number of D2R receptors bound by DA,
and receptors bound by RAC, respectively.
FDA
is the free DA concentration,
KD,DA
is the dissociation constant for DA at D2R, and fractional occupancies are denoted
by ODA and ORAC- A solution for the free DA concentration should account both for
displacement of bound DA into the free pool and release of additional DA due to
RAC binding presynaptic D2 autoreceptors. We assume the latter process occurs in
proportion to RAC occupancy:
+B(0
FDA =FDA
-
BDA
+60RAC
0(0)
KD,DA
Bmax(O
(DA
(-
-
ODA)
+6ORAC.
(B.2)
O(0)
We can employ Eqs. (B.1) and (B.2) to relate ODA and ORAC under several scenarios:
138
1. If we assume the free pool is large in comparison to the bound pool and we
ignore RAC-induced DA release, this is equivalent to ignoring the second and
third terms in Eq. (B.2). Under these conditions, the change in DA occupancy
(AODA =
9
DA -
91) is proportional to RAC occupancy and DA occupancy, and
we refer to this model as the linear approximation in the manuscript.
AODA = -RACnDA
(B.3)
2. If we ignore RAC-induced DA release but conserve free plus bound dopamine by
including the second term in Eq. (B.2), then the relationship in Eq. (B.3) deviates only subtly from linearity using literature parameters (Fig. B-2). Reported
values for Bm, in striatum for humans and NHP are about 20 nM [113, 114]
and the
KD,DA
for D2R has been estimated to be about 100 nM [115]. While
this model produces a relationship between RAC and DA occupancies that is
not strictly linear, there is no meaningful difference between this model and the
linear approximation.
3. To include the effect of DA release due to antagonism of autoreceptors, we
must rely upon data from rats, where IP injection of a dose slightly larger
than the maximal dose used in this study produced an approximate doubling
of extracellular DA [80]. Under these conditions, the relationship between RAC
and DA occupancies is described by a quadratic equation that deviates from
linearity at middle occupancies (Fig. B-2). This model produces a superlinear
relationship between DA and RAC occupancies.
Negative changes in DA occupancy are plotted on the y-axis in Fig. B-2 to provide
plots comparable to Figure 4-5. The three curves correspond to the three assumptions
above. Fitting a power function (a (eRAC)b) to the superlinear relation that. assumes
additional DA release (Fig. B-2, red curve), gives an exponent of b = 1.6. This
value is very similar to experimental values we observed in Figure 4-5. Thus, one
interpretation of Figure 4-5 is that a superlinear relationship between RAC occupancy
139
1
0.2
-Linear approx.
-- DA conserved
0
DA released
m 0.15 -
0 0.1-
0-0.05-
0
0
0.2
04
0.6
0.8
Raclopride occ. (ORAd
1
Figure B-2: The relationship between the fractional occupancies of raclopride (ORAC)
and the change in fractional occupancy of DA (AODA), which drives the fMRI signal,
using the linear approximation of Eq. (B.3), (dashed black) or estimating effects of
DA displacement (blue) and displacement plus release (red). Calculations assume a
20% basal DA occupancy.
and fMRI signal results from a superlinear relationship between RAC occupancy and
changes in DA occupancy, together with a linear relationship between DA occupancy
and the fMRI response.
B.2
Occupancy and Dynamic Binding Potential
Binding potential relative to a non-displaceable tissue compartment (BPND) is defined
in terms of the available concentration of receptors for binding (Bavail) relative to the
dissociation constant (KD), scaled by the fraction of ligand that is freely dissolved
in tissue water [66]. For this study, Bavan is equal to the total density of receptors
(Bm,, ) reduced by the fraction occupied by DA (ODA = BDA/Bmax) and RAC (ORAC)Consequently, the absolute change ABPND, relative to a tracer dose BP0(, can be
expressed in terms of occupancies:
140
ABPND
_
(1 -
9(0)
(1
-
1_
ORAC
1-
-
ORAC)
ODA -
01
+ AODA
(0)
(B.4)
DA
$
In all notations, 9 denotes the theoretical occupancy of DA or RAC, whereas
denotes the measured occupancy value derived from PET imaging data. The superscript (0) denotes basal DA conditions. The change in DA occupancy can be related
to RAC occupancy from Eq. (B.3), so that the fractional change in
BPND
becomes
an index of RAC occupancy:
A
ORAC =
ABPND
B
ORAC
(B.5)
ND
All computations of BPND or the related 'dynamic' binding potential (DBPND,
Eq. 4.1) were performed within the framework of the simplified reference tissue model
(SRTM [1]). In differential form, the time derivative of a tissue TAC (OT) can be
written in terms of a reference region concentration
(CREF),
an index (R1 ) of the
extraction-flow product relative to the reference region, a time constant for outflow
from the reference region into plasma (k'), and the binding potential (BPND) as:
UT =
R1CREF
+ k2CREF - k2aCT
= R1OREF + R1k'CREF
2 -
Rik'
2
1+ BPNDC
T
(B.6)
Although analytical solutions [1] or GLM techniques [70, 116] often solve for three
local parameters, the reference region rate constant (k2) is a global parameter, and
thus Eq. (B.4) contains an extra degree of freedom for each voxel or ROI [117]. Two
questions arise regarding the determination of an appropriate index of occupancy for
comparison with a dynamic fMRI response like CBV(t):
1. Should we employ a steady-state analysis based upon BPND, even though
physiology clearly is changing?
BPND
represents a weighted average across the 90 min
141
'o 40
Q30
--
Putamen TAC
SRTM fit
Dynamic fit
10
0
0
0
20
40
60
time (min)
80
Figure B-3: Comparison of an SRTM fit (black) with a dynamic fit (red) that allows
the use of DBPND (for the 16 jg/kg RAC mass dose in M2). The dynamic method
shows an improved fit visually and by X 2 /DOF measures.
measurement interval, and the weighting function changes versus the injected mass
dose. A dynamic index of receptor availability might offer a less biased alternative.
A dynamic binding potential (DBPND, Eq. 4.1) is a temporal analog of BPND that is
.
determined in analysis by including a temporal dependence in parameter k 2a(t) [70]
2. Should we fix parameter k' or employ the original 3-parameter SRTM? From
inspection, it is clear that k' will be biased toward zero to reduce noise from the last
two terms when BP is small, and this bias should affect other parameters as well.
Because specific binding varies with mass dose, a 3-parameter model may exhibit
more bias versus mass.
Based on this reasoning, we employed a standard steady-state BPND analysis
with a 2-parameter SRTM and a dynamic analysis by adding a time-dependent
k 2 a(t)
term. A comparison of the two model fits is shown in Fig. B-3 for the 16 jg/kg
dataset in animal M2. The dynamic analysis provides a better fit visually and based
on X2
/DOF values, which are 2.8 x 106 and 0.7 x 106 for the SRTM and dynamic
analysis respectively.
142
B.3
Forward-model Simulation and Analysis
This section describes simulations that 1) investigated the accuracy of SRTM estimates of occupancy using the co-infusion paradigm employed in this study, and
2) compared simplified estimates of specific binding based upon SRTM with actual
specific binding in a simulation model with separate free and specifically bound compartments.
Since our experimental paradigm used progressively larger doses of RAC to reduce BPND, we investigated the accuracy of linear SRTM measurements of occupancy
as a function of RAC dosage using forward-model simulations, which included compartments for plasma, free and specifically bound RAC, and free and specifically
bound DA [114]. Literature values were employed for all rate constants, dissociation
constants, and the concentration of D2R (Table B.1). Plasma kinetics were approximated by a gamma-variate profile for RAC injection followed by bi-exponential decay
of RAC from plasma; these kinetics were adjusted to match the reference region TAC
from experimental data. Synthetic noise [118] was applied to TACs to approximate
experimental data from basal ganglia ROIs. Since simulations separated free and
specifically bound compartments, true values for specific binding were known as a
function of time. Hence, we could assess the accuracy of SRTM analyses based upon
BPND
values versus time-dependent binding models.
We analyzed simulated data by the GLM using the same software that was employed to analyze real data. Analyses used the integral form of Eq. (B.6) and either
three local constant parameters [70], as in Eq. (B.6), or two local parameters [69].
Additionally, we evaluated the effect of a time-dependent k2 , (following [70]). The
time dependence was implemented as a gamma-variate function specified by a single parameter defining the time to peak (-r). Thus, SRTM included either three or
four local parameters, and the reduced model (MRTM2) included a global value for
k (kG) and either two or three local parameters, with the latter version explicitly
143
C-)
-,,C
0
100
O 3-param model
0 2-param,y(t)
80
El
E
60
E
40
0
CL
(m
20
IL
(14
20
0
40
60
80
True ABP/BP,
100
Figure B-4: Analysis of simulated data to determine the accuracy of occupancy estimates relative to true peak change in occupancy. Simulations were analyzed within
the SRTM framework by GLM: Conventional 3-parameter SRTM analyses with a
fixed BPND (black squares) systematically underestimate peak occupancy, whereas
the 2-parameter SRTM (red dots) accurately determines peak occupancy when a
gamma-variate function is embedded in parameter k2a-
formulated as follows:
CT(t) =
R1
CREF + k2,G
- k2 a
J
CREF(t)dt)
CT(t)dt - k2 ay
J
-y(t, T)CT(t)dt
(B.7)
where
'y(t, T) =
(t/)
e(1-t/r)
When employing a time-dependent k2a term, the time-to-peak parameter was
determined in two different ways to determine the sensitivity of results to the choice
of the time constant -r: (ii) by minimizing the X 2 /DOF of the GLM fit to data for
each dose, or (ii) by minimizing the X 2 /DOF for the lowest dose, and maintaining
that value of r for all doses.
The main simulation results are summarized in Fig. B-4. The conventional SRTM
144
X 10-3
16 g/kg
4 --
0
40 sg/kg
4-
0
T-R, *CR
- --Spec. bind.
20
40
60
time (min)
80
Figure B-5: Estimate of specific binding computed from the data (S=CT-RCREF,
black line) compared with specific binding (dashed), modeled with a full reference
tissue model based on fitted parameters for the two highest RAC mass doses in M1
(16 pg/kg and 40 pg/kg).
model produced estimates of RAC occupancies that underestimated true peak occupancies by 10-15% at high doses (black). Best results were obtained by adding a
gamma time-dependence to MRTM2 (red) in which the variable r was optimized for
each dose. Errors in occupancy were <2% at all doses. Intermediate results with up
to 7% error were obtained by using fixed values of r, based upon fits within MRTM2
for data at the lowest mass dose. Fitting up to 4 parameters (SRTM with a gamma
time-dependence on k2a) did not provide stable estimates of r across doses within
simulations.
These results suggest that the most accurate method for determining peak changes
in RAC occupancies in this co-infusion paradigm is to (i) fit parameter k' in a highbinding region (e.g., putamen) at low mass dose, (ii) eliminate the extra degree of
freedom incorporated into the three-parameter SRTM model by fixing k2, and (iii)
add a time-dependent binding term that is optimized at each dose by minimizing the
X 2 /DOF in a high-binding region (e.g., putamen).
145
Table B.1: Parameters used in forward-model simulations.
VD
Value
17 nM
3.8 nM
25%
3.36
0.17 min- 1
0.7
Species
NHP
Human
Human
NHP
NHP
NHP
Reference
Endres et al. (1997) [114]
Farde et al. (1986) [120]
Verhoeff et al. (2002) [121]
from Bm., KD, and 9 DA
Endres et al. (1997) [114]
Endres et al. (1997) [114]
(distribution vol. in REF region)
R 1 = K 1 /K 1 = k2/k2
KD,DA
k4,DA = koff,DA
k4= koff,RAC
0.8
100 nM
1 min- 1
0.07 min-'
NHP
Human
NHP
NHP
Measured from data
Fisher et al. (1995) [115]
Logan et al. (1991) [118]
Endres et al. (1997) [114]
Parameter
Bmax
KD
ODA
in basal state
BPND
Kj
In a full reference tissue model (FRTM [23]), specific binding is known, whereas
SRTM analysis determines only the ratio of the rate constants (k 3 , k4 ) between the free
and specifically bound compartments, as well as an effective rate constant (k2a) for
washout. In measurements and fits, the offset rate constant (k4 ) has been determined
to be about three to ten times smaller than the washout rate constant [114, 119], so
that the free and non-specifically bound compartment changes much more rapidly in
time than the specifically bound compartment. Thus, an approximate index of specific
binding can be obtained through SRTM by correcting the tissue concentration TAC
for ligand delivery and washout by subtracting the reference region and scaling for
the relative flow-extraction product (R1 ) as Cs ~ S = CT - R1CREF. Fig.
B-5
compares specific binding Cs as determined from an analysis with the FRTM to the
specific binding estimate S. In both cases, the peak of S is shifted by no more than
a few minutes compared to Cs, which is comparable to the uncertainty in the timeto-peak response of fMRI signal and PET specific binding in this study based upon
optimization of the X2 /DOF, as described in the Results of Chapter 4.
146
Bibliography
[1] A. A. Lammertsma and S. P. Hume, "Simplified reference tissue model for
PET receptor studies," Neurolmage, vol. 4, no. 3, pp. 153-158, Dec. 1996.
[2] G. C. Wiggins, C. Triantafyllou, A. Potthast, A. Reykowski, M. Nittka,
and L. L. Wald, "32-channel 3 tesla receive-only phasedarray head coil with
soccerball element geometry," Magnetic Resonance in Medicine, vol. 56, no. 1,
pp. 216-223, Jul. 2006.
[3] L.-H. Shen, M.-H. Liao, and Y.-C. Tseng, "Recent advances in imaging of
dopaminergic neurons for evaluation of neuropsychiatric disorders," Journalof
Biomedicine and Biotechnology, vol. 2012, pp. 1-14, 2012.
[4] B. J. Pichler, A. Kolb, T. N.gele, and H.-P. Schlemmer, "PET/MRI: paving
the way for the next generation of clinical multimodality imaging applications,"
J Nucl Med, vol. 51, no. 3, pp. 333-336, Mar. 2010.
[5] C. Catana, Y. Wu, M. S. Judenhofer, J. Qi, B. J. Pichler, and S. R. Cherry,
"Simultaneous acquisition of multislice PET and MR images: Initial results
with a MR-Compatible PET scanner," J Nucl Med, vol. 47, no. 12, pp.
1968-1976, Dec. 2006.
[6] R. Grazioso, N. Zhang, J. Corbeil, M. Schmand, R. Ladebeck, M. Vester,
G. Schnur, W. Renz, and H. Fischer, "APD-based PET detector for
simultaneous PET/MR imaging," Nuclear Instruments and Methods in Physics
Research Section A: Accelerators, Spectrometers, Detectors and Associated
Equipment, vol. 569, no. 2, pp. 301-305, Dec. 2006.
[7] M. S. Judenhofer, H. F. Wehrl, D. F. Newport, C. Catana, S. B. Siegel,
M. Becker, A. Thielscher, M. Kneilling, M. P. Lichy, M. Eichner, K. Klingel,
G. Reischl, S. Widmaier, M. R6cken, R. E. Nutt, H.-J.. Machulla, K. Uludag,
S. R. Cherry, C. D. Claussen, and B. J. Pichler, "Simultaneous PET-MRI: a
new approach for functional and morphological imaging," Nature Medicine,
vol. 14, no. 4, pp. 459-465, Mar. 2008.
[8] H. F. Wehrl, M. S. Judenhofer, S. Wiehr, and B. J. Pichler, "Preclinical PET/MR: technological advances and new perspectives in biomedical
research," Eur J Nucl Med Mol Imaging, vol. 36, no. 1, pp. 56-68, Mar. 2009.
147
[9] M. Schmand, Z. Burbar, J. Corbeil, N. Zhang, C. Michael, L. Byars,
L. Eriksson, R. Grazioso, M. Martin, A. Moor, J. Camp, V. Matschl,
R. Ladebeck, W. Renz, H. Fischer, K. Jattke, G. Schnur, N. Rietsch,
B. Bendriem, and W.-D. Heiss, "BrainPET: first human tomograph for
simultaneous (functional) PET and MR imaging," Society of Nuclear Medicine
Annual Meeting Abstracts, vol. 48, no. Supplement 2, p. 45P, May 2007.
[10] G. Delso, S. Ffirst, B. Jakoby, R. Ladebeck, C. Ganter, S. G. Nekolla,
M. Schwaiger, and S. I. Ziegler, "Performance measurements of the siemens
mMR integrated whole-body PET/MR scanner," J Nucl Med, vol. 52, no. 12,
pp. 1914-1922, Dec. 2011.
[11] H. H. Quick, R. Ladebeck, and J. C. Georgi, "Whole-body MR/PET hybrid
imaging: Technical considerations, clinical workflow, and initial results,"
MA GNETOM Flash, vol. 1, pp. 88-100, 2011.
[12] B. Keil, J. N. Blau, S. Biber, P. Hoecht, V. Tountcheva, K. Setsompop,
C. Triantafyllou, and L. L. Wald, "A 64-channel 3T array coil for accelerated
brain MRI," Magn Reson Med, vol. 70, no. 1, pp. 248-258, Jul. 2013.
[13] B. Keil and L. L. Wald, "Massively parallel MRI detector arrays," Journal of
Magnetic Resonance, vol. 229, pp. 75-89, Apr. 2013.
[14] A. Bj6rklund and S. B. Dunnett, "Dopamine neuron systems in the brain: an
update," Trends in Neurosciences, vol. 30, no. 5, pp. 194-202, May 2007.
[15] R. A. Wise, "Dopamine, learning and motivation," Nat Rev Neurosci, vol. 5,
no. 6, pp. 483-494, Jun. 2004.
[16] W. Schultz, "Behavioral dopamine signals," Trends in Neurosciences, vol. 30,
no. 5, pp. 203-210, May 2007.
[17] N. D. Volkow, G.-J. Wang, J. S. Fowler, D. Tomasi, and F. Telang, "Addiction:
Beyond dopamine reward circuitry," PNAS, vol. 108, no. 37, pp. 15037-15042,
Sep. 2011.
[18] J. Schulz, M. Gerlach, G. Gille, W. Kuhn, M. Miingersdorf, P. Riederer,
M. Sfidmeyer, and A. Ludolph, "Basic science in parkinson's disease: its
impact on clinical practice," Journal of Neurology, vol. 258, no. 0, pp. 299-306,
2011.
[19] 0. D. Howes and S. Kapur, "The dopamine hypothesis of schizophrenia:
Version III-The final common pathway," Schizophr Bull, vol. 35, no. 3, pp.
549-562, May 2009.
[20] Wikipedia. (2014) Putamen - wikipedia, the free encyclopedia.
[21] Wikipedia. (2014) Basal ganglia - wikipedia, the free encyclopedia.
148
[22] L. Farde, E. Ehrin, L. Eriksson, T. Greitz, H. Hall, C. G. Hedstrdm, J. E.
Litton, and G. Sedvall, "Substituted benzamides as ligands for visualization
of dopamine receptor binding in the human brain by positron emission
tomography," Proceedings of the National Academy of Sciences, vol. 82, no. 11,
p. 3863, 1985.
[23] A. A. Lammertsma, C. J. Bench, S. P. Hume, S. Osman, K. Gunn, D. J.
Brooks, and R. S. J. Frackowiak, "Comparison of methods for analysis of
clinical [11C]Raclopride studies," Journalof CerebralBlood Flow & Metabolism,
vol. 16, no. 1, pp. 42-52, 1996.
[24] M. Normandin and E. Morris, "Estimating neurotransmitter kinetics with ntPET: a simulation study of temporal precision and effects of biased data,"
Neuroimage, vol. 39, no. 3, pp. 1162-1179, 2008.
[25] S. Pappata, "In vivo detection of striatal dopamine release during reward:
A PET study with [11C]Raclopride and a single dynamic scan approach,"
Neurolmage, vol. 16, no. 4, pp. 1015-1027, Aug. 2002.
[26] A. J. Clark, Generalpharmacology. Berlin; New York: Springer-Verlag, 1937.
[27] L. R. Gabriel, S. Wu, P. Kearney, K. D. Bellv6, C. Standley, K. E. Fogarty,
and H. E. Melikian, "Dopamine transporter endocytic trafficking in striatal
dopaminergic neurons: Differential dependence on dynamin and the actin
cytoskeleton," J. Neurosci., vol. 33, no. 45, pp. 17836-17846, Nov. 2013.
[28] C. Catana, A. Drzezga, W.-D. Heiss, and B. R. Rosen, "PET/MRI for
neurologic applications," J Nucl Med, vol. 53, no. 12, pp. 1916-1925, Dec.
2012.
[29] C. Catana, D. Procissi, Y. Wu, M. S. Judenhofer, J. Qi, B. J. Pichler, R. E.
Jacobs, and S. R. Cherry, "Simultaneous in vivo positron emission tomography and magnetic resonance imaging," Proceedings of the NationalAcademy of
Sciences, vol. 105, no. 10, p. 3705, 2008.
[30] G. C. Wiggins, J. R. Polimeni, A. Potthast, M. Schmitt, V. Alagappan, and
L. L. Wald, "96channel receiveonly head coil for 3 tesla: Design optimization
and evaluation," Magnetic Resonance in Medicine, vol. 62, no. 3, pp. 754-762,
Sep. 2009.
[31] K. P. Pruessmann, M. Weiger, M. B. Scheidegger, and P. Boesiger, "SENSE:
sensitivity encoding for fast MRI," Magn Reson Med, vol. 42, no. 5, pp.
952-962, Nov. 1999.
[32] M. A. Griswold, P. M. Jakob, R. M. Heidemann, M. Nittka, V. Jellus,
J. Wang, B. Kiefer, and A. Haase, "Generalized autocalibrating partially
parallel acquisitions (GRAPPA)," Magnetic Resonance in Medicine, vol. 47,
no. 6, pp. 1202-1210, Jun. 2002.
149
[33] L. Tellmann, H. H. Quick, A. Bockisch, H. Herzog, and T. Beyer, "The effect
of MR surface coils on PET quantification in whole-body PET/MR: results
from a pseudo-PET/MR phantom study," Medical Physics, vol. 38, no. 5, pp.
2795-2805, 2011.
[34] C. Catana, A. van der Kouwe, T. Benner, C. J. Michel, M. Hamm, M. Fenchel,
B. Fischl, B. Rosen, M. Schmand, and A. G. Sorensen, "Toward implementing
an MRI-Based PET attenuation-correction method for neurologic studies on
the MR-PET brain prototype," Journal of Nuclear Medicine, vol. 51, no. 9,
pp. 1431-1438, Sep. 2010.
[35] L. R. MacDonald, S. Kohlmyer, C. Liu, T. K. Lewellen, and P. E. Kinahan,
"Effects of MR surface coils on PET quantification," Med Phys, vol. 38, no. 6,
pp. 2948-2956, Jun. 2011.
[36] D. H. Paulus, H. Braun, B. Aklan, and H. H. Quick, "Simultaneous PET/MR
imaging: MR-based attenuation correction of local radiofrequency surface
coils," Medical Physics, vol. 39, no. 7, pp. 4306-4315, Jun. 2012.
[37] D. H. Paulus, L. Tellmann, and H. H. Quick, "Towards improved hardware
component attenuation correction in PET/MR hybrid imaging," Phys. Med.
Biol., vol. 58, no. 22, p. 8021, Nov. 2013.
[38] C. Burger, G. Goerres, S. Schoenes, A. Buck, A. Lonn, and G. Von Schulthess,
"PET attenuation coefficients from CT images: experimental evaluation of the
transformation of CT into PET 511-keV attenuation coefficients," European
journal of nuclear medicine and molecular imaging, vol. 29, no. 7, pp. 922-927,
2002.
[39] J. Hubbell, "Photon mass attenuation and energy-absorption coefficients," The
International Journal of Applied Radiation and Isotopes, vol. 33, no. 11, pp.
1269-1290, 1982.
[40] C. Possanzini and M. Boutelje, "Influence of magnetic field on preamplifiers
using GaAs FET technology," in Proc. Intl. Soc. Mag. Reson. Med. 16, 2008,
p. 1123.
[41] A. Reykowski, S. M. Wright, and J. R. Porter, "Design of matching networks
for low noise preamplifiers," Magnetic Resonance in Medicine, vol. 33, no. 6,
pp. 848-852, Jun. 1995.
[42] P. Kellman and E. R. McVeigh, "Image reconstruction in SNR units: A
general method for SNR measurement," Magnetic Resonance in Medicine,
vol. 54, no. 6, pp. 1439-1447, Dec. 2005.
[43] A. J. van der Kouwe, T. Benner, D. H. Salat, - and B. Fischl, "Brain
morphometry with multiecho MPRAGE," Neurolmage, vol. 40, no. 2, pp.
559-569, Apr. 2008.
150
[44] L. Byars, M. Sibomana, Z. Burbar, J. Jones, V. Panin, W. Barker, J.-S. Liow,
R. Carson, and C. Michel, "Variance reduction on randoms from coincidence
histograms for the HRRT," in 2005 IEEE Nuclear Science Symposium Conference Record, vol. 5, Oct. 2005, pp. 2622-2626.
[45] D. B. Chonde, N. Abolmaali, G. Arabasz, A. R. Guimaraes, and C. Catana,
"Effect of MRI acoustic noise on cerebral fludeoxyglucose uptake in
simultaneous MR-PET imaging:," Investigative Radiology, vol. 48, no. 5, pp.
302-312, May 2013.
[46] D. Izquierdo-Gacia, "New SPM8-based MRAC method for simultaneous
PET/MR brain images: comparison with state-of-the-art non-rigid registration
methods," in PSMR, vol. 72, 2014.
[47] B. Keil, V. Alagappan, A. Mareyam, J. A. McNab, K. Fujimoto, V. Tountcheva,
C. Triantafyllou, D. D. Dilks, N. Kanwisher, W. Lin, P. E. Grant, and L. L.
Wald, "Size-optimized 32-channel brain arrays for. 3T pediatric imaging,"
Magnetic Resonance in Medicine, vol. 66, no. 6, pp. 1777-1787, Dec. 2011.
[48] B. Zhang, D. Pal, Z. Hu, N. Ojha, T. Guo, G. Muswick, C.-h. Tung,
and J. Kaste, "Attenuation correction for MR table and coils for a sequential PET/MR system," in Nuclear Science Symposium Conference Record
(NSS/MIC), 2009 IEEE, Nov. 2009, pp. 3303 -3306.
[49] P. T. Fox, M. E. Raichle, M. A. Mintun, and C. Dence, "Nonoxidative glucose
consumption during focal physiologic neural activity," Science, vol. 241, no.
4864, pp. 462-464, 1988.
[50] T. L. Davis, K. K. Kwong, R. M. Weisskoff, and B. R. Rosen, "Calibrated
functional MRI: mapping the dynamics of oxidative metabolism," Proceedings
of the National Academy of Sciences, vol. 95, no. 4, p. 1834, 1998.
[51] R. D. Hoge, J. Atkinson, B. Gill, G. R. Crelier, S. Marrett, and G. B.
Pike, "Linear coupling between cerebral blood flow and oxygen consumption
in activated human cortex," Proceedings of the National Academy of Sciences,
vol. 96, no. 16, p. 9403, 1999.
[52] C. S. Roy and C. S. Sherrington, "On the regulation of the blood-supply of
the brain," The Journal of physiology, vol. 11, no. 1-2, p. 85, 1890.
[53] A. Devor, A. K. Dunn, M. L. Andermann, I. Ulbert, D. A. Boas, and A. M.
Dale, "Coupling of total hemoglobin concentration, oxygenation, and neural
activity in rat somatosensory cortex," Neuron, vol. 39, no. 2, pp. 353-359, Jul.
2003.
[54] N. K. Logothetis, J. Pauls, M. Augath, T. Trinath, and A. Oeltermann,
"Neurophysiological investigation of the basis of the fMRI signal," Nature, vol.
412, no. 6843, pp. 150-157, Jul. 2001.
151
[55] S. A. Sheth, M. Nemoto, M. Guiou, M. Walker, N. Pouratian, and A. W.
Toga, "Linear and nonlinear relationships between neuronal activity, oxygen
metabolism, and hemodynamic responses," Neuron, vol. 42, no. 2, pp. 347-355,
Apr. 2004.
[56] P. J. Magistretti and L. Pellerin, "Cellular mechanisms of brain energy
metabolism and their relevance to functional brain imaging." Philos Trans R
Soc Lond B Biol Sci, vol. 354, no. 1387, pp. 1155-1163, Jul. 1999.
[57] P. H. Elsinga, K. Hatano, and K. Ishiwata, "PET tracers for imaging of the
dopaminergic system," Curr. Med. Chem., vol. 13, no. 18, pp. 2139-2153, 2006.
[58] Y. I. Chen, A. L. Brownell, W. Galpern, 0. Isacson, M. Bogdanov, M. F. Beal,
E. Livni, B. R. Rosen, and B. G. Jenkins, "Detection of dopaminergic cell loss
and neural transplantation using pharmacological MRI, PET and behavioral
assessment," Neuroreport, vol. 10, no. 14, pp. 2881-2886, 1999.
[59] Y.-C. Chen, J.-K. Choi, S. Andersen, B. Rosen, and B. Jenkins, "Mapping
dopamine D2/D3 receptor function using pharmacological magnetic resonance
imaging," Psychopharmacology, vol. 180, no. 4, pp. 705-715, 2005.
[60] J.-K. Choi, Y. I. Chen, E. Hamel, and B. G. Jenkins, "Brain hemodynamic
changes mediated by dopamine receptors: Role of the cerebral microvasculature
in dopamine-mediated neurovascular coupling," Neurolmage, vol. 30, no. 3,
pp. 700-712, Apr. 2006.
[61] J. R. Traynor and R. R. Neubig, "Regulators of g protein signaling & drugs of
abuse," Mol. Interv., vol. 5, no. 1, pp. 30-41, Feb. 2005.
[62] J. B. Mandeville, "IRON fMRI measurements of CBV and implications for
BOLD signal," Neurolmage, vol. 62, no. 2, pp. 1000-1008, Aug. 2012.
[63] K. S. Saleem and N. K. Logothetis, A Combined MRI and Histology Atlas of
the Rhesus Monkey Brain. Burlington: Elsevier, 2006.
[64] D. McLaren, K. Kosmatka, T. Oakes, C. Kroenke, S. Kohama, J. Matochik,
D. Ingram, and S. Johnson, "A population-average MRI-based atlas collection
of the rhesus macaque," Neuroimage, vol. 45, no. 1, pp. 52-59, 2009.
[65] J. B. Mandeville, J. J. A. Marota, B. E. Kosofsky, J. R. Keltner, R. Weissleder,
B. R. Rosen, and R. M. Weisskoff, "Dynamic functional imaging of relative
cerebral blood volume during rat forepaw stimulation," Magnetic Resonance
in Medicine, vol. 39, no. 4, pp. 615-624, Apr. 1998.
[66] R. Innis, V. Cunningham, J. Delforge, M. Fujita, A. Gjedde, R. Gunn,
J. Holden, S. Houle, S. Huang, M. Ichise et al., "Consensus nomenclature for
in vivo imaging of reversibly binding radioligands," Journal of Cerebral Blood
Flow & Metabolism, vol. 27, no. 9, pp. 1533-1539, 2007.
152
[67] H. Hall, L. Farde, C. Halldin, Y. L. Hurd, S. Pauli, and G. Sedvall,
"Autoradiographic localization of extrastriatal D2-dopamine receptors in the
human brain using [1251]epidepride," Synapse, vol. 23, no. 2, pp. 115-123,
1996.
[68] L. H. Pinborg, C. Videbaek, M. Ziebell, T. Mackeprang, L. Friberg,
H. Rasmussen, G. M. Knudsen, and B. Y. Glenthoj, "[1231]Epidepride binding
to cerebellar dopamine D2/D3 receptors is displaceable: Implications for the
use of cerebellum as a reference region," Neurolmage, vol. 34, no. 4, pp.
1450-1453, Feb. 2007.
[69] M. Ichise, J.-S. Liow, J.-Q. Lu, A. Takano, K. Model, H. Toyama, T. Suhara,
K. Suzuki, R. B. Innis, and R. E. Carson, "Linearized reference tissue
parametric imaging methods: Application to [11C]DASB positron emission
tomography studies of the serotonin transporter in human brain," Journal of
CerebralBlood Flow & Metabolism, vol. 23, no. 9, pp. 1096-1112, 2003.
[70] N. M. Alpert, R. D. Badgaiyan, E. Livni, and A. J. Fischman, "A novel
method for noninvasive detection of neuromodulatory changes in specific
neurotransmitter systems," Neurolmage, vol. 19, no. 3, pp. 1049-1060, Jul.
2003.
[71] M. D. Normandin, W. K. Schiffer, and E. D. Morris, "A linear model for
estimation of neurotransmitter response profiles from dynamic PET data,"
Neurolmage, vol. 59, no. 3, pp. 2689-2699, Feb. 2012.
[72] E. A. Rabiner, M. Slifstein, J. Nobrega, C. Plisson, M. Huiban, R. Raymond,
M. Diwan, A. A. Wilson, P. McCormick, G. Gentile, R. N. Gunn, and M. A.
Laruelle, "In vivo quantification of regional dopamine-d3 receptor binding
potential of (+)-PHNO: studies in non-human primates and transgenic mice,"
Synapse, vol. 63, no. 9, pp. 782-793, 2009.
[73] K. Worsley, C. Liao, J. Aston, V. Petre, G. Duncan, F. Morales, and A. Evans,
"A general statistical analysis for fMRI data," Neurolmage, vol. 15, no. 1, pp.
1-15, Jan. 2002.
[74] R. E. Carson, A. Breier, A. de Bartolomeis, R. C. Saunders, T. P. Su,
B. Schmall, M. G. Der, D. Pickar, and W. C. Eckelman, "Quantification
of amphetamine-induced changes in [11C] raclopride binding with continuous
infusion," J Cereb Blood Flow Metab, vol. 17, no. 4, pp. 437-447, Apr. 1997.
[75] R. Narendran, D.-R. Hwang, M. Slifstein, Y. Hwang, Y. Huang, J. Ekelund,
0. Guillin, E. Scher, D. Martinez, and M. Laruelle, "Measurement of
the proportion of D2 receptors configured in state of high affinity for
agonists in vivo: A positron emission tomography study using [11C]N-Propylnorapomorphine and [11C]Raclopride in baboons," J Pharmacol Exp Ther,
vol. 315, no. 1, pp. 80-90, Oct. 2005.
153
[76] J. B. Mandeville, C. Y. Sander, B. G. Jenkins, J. M. Hooker, C. Catana,
W. Vanduffel, N. M. Alpert, B. R. Rosen, and M. D. Normandin, "A
receptor-based model for dopamine-induced fMRI signal," Neurolmage, vol. 75,
pp. 46-57, Jul. 2013.
[77] B. H. Schott, L. Minuzzi, R. M. Krebs, D. Elmenhorst, M. Lang, 0. H.
Winz, C. I. Seidenbecher, H. H. Coenen, H.-J. Heinze, K. Zilles, E. Diizel,
and A. Bauer, "Mesolimbic functional magnetic resonance imaging activations
during reward anticipation correlate with reward-related ventral striatal
dopamine release," J. Neurosci., vol. 28, no. 52, pp. 14311-14319, Dec. 2008.
[78] N. B. L. Urban, M. Slifstein, S. Meda, X. Xu, R. Ayoub, 0. Medina,
G. D. Pearlson, J. H. Krystal, and A. Abi-Dargham, "Imaging human
reward processing with positron emission tomography and functional magnetic
resonance imaging," Psychopharmacology,Nov. 2011.
&
[79] M. Laruelle, "Imaging synaptic neurotransmission with in vivo binding
competition techniques: A critical review," Journal of Cerebral Blood Flow
Metabolism, vol. 20, no. 3, pp. 423-451, 2000.
[80] P. Hertel, "Enhanced cortical dopamine output and antipsychotic-like effects
of raclopride by 2 adrenoceptor blockade," Science, vol. 286, no. 5437, pp.
105-107, Oct. 1999.
[81] D. S. Goldstein, P. Sullivan, C. Holmes, I. J. Kopin, M. J. Basile, and D. C.
Mash, "Catechols in post-mortem brain of patients with parkinson disease,"
European Journalof Neurology, vol. 18, no. 5, pp. 703-710, 2011.
[82] P. Riccardi, R. Baldwin, R. Salomon, S. Anderson, M. S. Ansari, R. Li,
B. Dawant, A. Bauernfeind, D. Schmidt, and R. Kessler, "Estimation of
baseline dopamine d2 receptor occupancy in striatum and extrastriatal regions
in humans with positron emission tomography with [18F] fallypride," Biological
Psychiatry, vol. 63, no. 2, pp. 241-244, Jan. 2008.
[83] L. S. Kegeles, A. Abi-Dargham, W. G. Frankle, R. Gil, T. B. Cooper,
M. Slifstein, D.-R. Hwang, Y. Huang, S. N. Haber, and M. Laruelle,
"Increased synaptic dopamine function in associative regions of the striatum
in schizophrenia," Archives of generalpsychiatry, vol. 67, no. 3, p. 231, 2010.
[84] D. Martinez, K. Greene, A. Broft, D. Kumar, F. Liu, R. Narendran,
M. Slifstein, R. Van Heertum, and H. D. Kleber, "Lower level of endogenous
dopamine in patients with cocaine dependence: Findings from PET imaging
of D2/D3 receptors following acute dopamine depletion," American Journal of
Psychiatry, vol. 166, no. 10, pp. 1170-1177, Sep. 2009.
[85] M. Morissette and T. Di Paolo, "Effect of estradiol on striatal dopamine
activity of female hemiparkinsonian monkeys," Journal of Neuroscience
Research, vol. 87, no. 7, pp. 1634-1644, 2009.
154
[86] Y. U. Adachi, S. Yamada, M. Satomoto, H. Higuchi, K. Watanabe, and
T. Kazama, "Isoflurane anesthesia induces biphasic effect on dopamine release
in the rat striatum," Brain Research Bulletin, vol. 67, no. 3, pp. 176-181, Oct.
2005.
[87] D. Qiu, G. Zaharchuk, T. Christen, W. W. Ni, and M. E. Moseley, "Contrastenhanced functional blood volume imaging (CE-fBVI): enhanced sensitivity for
brain activation in humans using the ultrasmall superparamagnetic iron oxide
agent ferumoxytol," Neurolmage, vol. 62, no. 3, pp. 1726-1731, Sep. 2012.
[88] L. Booij, A. J. W. V. d. Does, and W. J. Riedel, "Monoamine depletion in
psychiatric and healthy populations: review," Molecular Psychiatry, vol. 8,
no. 12, pp. 951-973, 2003.
[89] L. de Haan, "Subjective experiences during dopamine depletion," American
Journal of Psychiatry, vol. 162, no. 9, pp. 1755-1755, Sep. 2005.
[90] N. Guo, D.-R. Hwang, E.-S. Lo, Y.-Y. Huang, M. Laruelle, and
A. Abi-Dargham, "Dopamine depletion and in vivo binding of PET dl
receptor radioligands: implications for imaging studies in schizophrenia,"
Neuropsychopharmacology,vol. 28, no. 9, pp. 1703-1711, Sep. 2003.
[91] H. Uchida, H. Takeuchi, A. Graff-Guerrero, T. Suzuki, K. Watanabe, and
D. C. Mamo, "Dopamine D2 receptor occupancy and clinical effects: a
systematic review and pooled analysis," J Clin Psychopharmacol, vol. 31,
no. 4, pp. 497-502, Aug. 2011.
[92] M. Nord and L. Farde, "Antipsychotic occupancy of dopamine receptors in
schizophrenia," CNS Neuroscience & Therapeutics, vol. 17, no. 2, pp. 97-103,
2011.
[93] A. Breier, T.-P. Su, R. Saunders, R. E. Carson, B. S. Kolachana,
A. d. Bartolomeis, D. R. Weinberger, N. Weisenfeld, A. K. Malhotra,
W. C. Eckelman, and D. Pickar, "Schizophrenia is associated with elevated
amphetamine-induced synaptic dopamine concentrations: Evidence from a
novel positron emission tomography method," PNAS, vol. 94, no. 6, pp.
2569-2574, Mar. 1997.
[94] M. Laruelle, C. D. D'Souza, R. M. Baldwin, A. Abi-Dargham, S. J. Kanes,
C. L. Fingado, J. P. Seibyl, S. S. Zoghbi, M. B. Bowers, P. Jatlow, D. S.
Charney, and R. B. Innis, "Imaging D2 receptor occupancy by endogenous
dopamine in humans," Neuropsychopharmacology, vol. 17, no. 3, pp. 162-174,
1997.
[95] N. Guo, W. Guo, M. Kralikova, M. Jiang, I. Schieren, R. Narendran,
M. Slifstein, A. Abi-Dargham, M. Laruelle, J. A. Javitch, and S. Rayport,
"Impact of D2 receptor internalization on binding affinity of neuroimaging
155
radiotracers,"
Neuropsychopharmacology, vol. 35, no. 3, pp. 806-817, Dec.
2009.
[96] M. Skinbjerg, J.-S. Liow, N. Seneca, J. Hong, S. Lu, A. Thorsell, M. Heilig,
V. W. Pike, C. Halldin, D. R. Sibley, and R. B. Innis, "D2 dopamine
receptor internalization prolongs the decrease of radioligand binding after
amphetamine: A PET study in a receptor internalization-deficient mouse
model," Neurolmage, vol. 50, no. 4, pp. 1402-1407, May 2010.
[97] B. Levant, D. E. Grigoriadis, and E. B. DeSouza, "Characterization of
[3H]quinpirole binding to D2-like dopamine receptors in rat brain." J
Pharmacol Exp Ther, vol. 262, no. 3, pp. 929-935, Sep. 1992.
[98] S. E. Bartlett, J. Enquist, F. W. Hopf, J. H. Lee, F. Gladher, V. Kharazia,
M. Waldhoer, W. S. Mailliard, R. Armstrong, A. Bonci, and J. L. Whistler,
"Dopamine responsiveness is regulated by targeted sorting of d2 receptors,"
PNAS, vol. 102, no. 32, pp. 11521-11526, Aug. 2005.
[99] A. K. Berger, T. Green, S. J. Siegel, E. J. Nestler, and R. P. Hammer, "cAMP
response element binding protein phosphorylation in nucleus accumbens
underlies sustained recovery of sensorimotor gating following repeated D2-like
receptor agonist treatment in rats," Biological Psychiatry, vol. 69, no. 3, pp.
288-294, Feb. 2011.
[100] S. Perachon, J.-C. Schwartz, and P. Sokoloff, "Functional potencies of new
antiparkinsonian drugs at recombinant human dopamine dl, d2 and d3
receptors," European Journalof Pharmacology, vol. 366, no. 2-3, pp. 293-300,
Feb. 1999.
[101] J. C. Price, W. C. Drevets, J. Ruszkiewicz, P. J. Greer, V. L. Villemagne,
L. Xu, S. Mazumdar, M. N. Cantwell, and C. A. Mathis, "Sequential H2150
PET studies in baboons: Before and after amphetamine," Journal of Nuclear
Medicine, vol. 43, no. 8, pp. 1090 -1100, 2002.
[102] J. Logan, N. D. Volkow, J. S. Fowler, G.-J. Wang, S. L. Dewey, R. MacGregor,
D. Schlyer, S. J. Gatley, N. Pappas, P. King, R. Hitzemann, and S. Vitkun,
"Effects of blood flow on [nC]Raclopride binding in the brain:
Model
simulations and kinetic analysis of PET data," J Cereb Blood Flow Metab,
vol. 14, no. 6, pp. 995-1010, Nov. 1994.
[103] P. Cumming, G. Xiong, C. L. Fougere, A. Rominger, P. Bartenstein, H.-G.
Buchholz, M. Piel, F. R6sch, G. Griinder, and I. Vernaleken, "Surrogate
markers for cerebral blood flow correlate with [18F]-fallypride binding potential
at dopamine D2/3 receptors in human striatum," Synapse, vol. 67, no. 4, pp.
199-203, Apr. 2013.
156
[104] R. L. Grubb, M. E. Raichle, J. 0. Eichling, and M. M. Ter-Pogossian, "The
effects of changes in PaCO2 cerebral blood volume, blood flow, and vascular
mean transit time," Stroke, vol. 5, no. 5, pp. 630-639, Sep. 1974.
[105] D. W. Wooten, J. B. Mandeville, C. Y. Sander, J. M. Hooker, G. El Fakhri,
N. M. Alpert, and M. D. Normandin, "Simultaneous PET/fMRI mapping of
dopamine dl receptor occupancy and stimulation," in 10th InternationalSymposium on FunctionalNeuroReceptor Mapping of the Living Brain, 2014.
[106] D. Marcellino, J. Kehr, L. F. Agnati, and K. Fuxe, "Increased affinity
of dopamine for D2-like versus Di-like receptors. relevance for volume
transmission in interpreting PET findings," Synapse, vol. 66, no. 3, pp.
196-203, 2012.
[107] M. Villien, H.-Y. Wey, J. B. Mandeville, C. Catana, J. R. Polimeni, C. Y.
Sander, N. R. Ziircher, D. B. Chonde, J. S. Fowler, B. R. Rosen, and
J. M. Hooker, "Dynamic functional imaging of brain glucose utilization using
fPET-FDG," Neurolmage, vol. 100, pp. 192-199, Oct. 2014.
[108] A. Abi-Dargham, J. Rodenhiser, D. Printz, Y. Zea-Ponce, R. Gil, L. S.
Kegeles, R. Weiss, T. B. Cooper, J. J. Mann, R. L. V. Heertum, J. M.
Gorman, and M. Laxuelle, "Increased baseline occupancy of d2 receptors by
dopamine in schizophrenia," PNAS, vol. 97, no. 14, pp. 8104-8109, Jul. 2000.
[109] M. Yamazaki, M. Terrill, A. Fujimoto, T. Yamamoto, and D. M. Tucker,
"Integrating dense array EEG in the presurgical evaluation of temporal lobe
epilepsy," ISRN Neurol, vol. 2012, Nov. 2012.
[110] R. Thornton, S. Vulliemoz, R. Rodionov, D. W. Carmichael, U. J. Chaudhary,
B. Diehl, H. Laufs, C. Vollmar, A. W. McEvoy, M. C. Walker, F. Bartolomei,
M. Guye, P. Chauvel, J. S. Duncan, and L. Lemieux, "Epileptic networks
in focal cortical dysplasia revealed using electroencephalography-functional
magnetic resonance imaging," Annals of Neurology, vol. 70, no. 5, pp. 822-837,
2011.
[111] R. Casse, C. C. Rowe, M. Newton, S. U. Berlangieri, and A. M. Scott, "Positron
emission tomography and epilepsy," Mol Imaging Biol, vol. 4, no. 5, pp. 338351, Oct. 2002.
[112] C. Y. Sander, C. Catana, A. Zhu, C. Poulsen, B. R. Rosen, and G. Bonmassar,
"Invisible dense-array EEG net for simultaneous EEG-PET/MR imaging," in
Proc. Intl. Soc. Mag. Reson. Med. 22, ser. 0782, 2014.
[113] M. Slifstein and M. Laruelle, "Models and methods for derivation of in
vivo neuroreceptor parameters with PET and SPECT reversible radiotracers,"
Nuclear Medicine and Biology, vol. 28, no. 5, pp. 595-608, Jul. 2001.
157
[114] P. Cumming, "Absolute abundances and affinity states of dopamine receptors
in mammalian brain: A review," Synapse, vol. 65, no. 9, pp. 892-909, 2011.
[115] C. J. Endres, B. S. Kolachana, R. C. Saunders, T. Su, D. Weinberger,
A. Breier, W. C. Eckelman, and R. E. Carson, "Kinetic modeling of
[11C]Raclopride: combined PET-Microdialysis studies," Journal of Cerebral
Blood Flow & Metabolism, vol. 17, no. 9, pp. 932-942, Sep. 1997.
[116] R. E. Fisher, E. D. Morris, N. M. Alpert, and A. J. Fischman, "In vivo imaging
of neuromodulatory synaptic transmission using PET: a review of relevant
neurophysiology," Human brain mapping, vol. 3, no. 1, pp. 24-34, 1995.
[117] Y. Zhou, C. J. Endres, J. R. Bra~i, S.-C. Huang, and D. F. Wong,
"Linear regression with spatial constraint to generate parametric images of
ligand-receptor dynamic PET studies with a simplified reference tissue model,"
Neurolmage, vol. 18, no. 4, pp. 975-989, Apr. 2003.
[118] Y. Wu and R. E. Carson, "Noise reduction in the simplified reference tissue
model for neuroreceptor functional imaging," Journal of Cerebral Blood Flow
& Metabolism, vol. 22, no. 12, pp. 1440-1452, Dec. 2002.
[119] J. Logan, J. S. Fowler, N. D. Volkow, Y. S. Ding, G.-J. Wang, and D. L.
Alexoff, "A strategy for removing the bias in the graphical analysis method,"
Journal of Cerebral Blood Flow & Metabolism, vol. 21, no. 3, pp. 307-320,
2001.
[120] S. Kapur and P. Seeman, "Antipsychotic agents differ in how fast they come
off the dopamine d2 receptors. implications for atypical antipsychotic action."
Journal of Psychiatry and Neuroscience, vol. 25, no. 2, pp. 161-166, Mar.
2000, PMID: 10740989 PMCID: PMC1408069.
[121] Farde L, Hall H, Ehrin E, and Sedvall G, "Quantitative analysis of D2 dopamine
receptor binding in the living human brain by PET." Science (New York, N. Y.),
vol. 231, no. 4735, pp. 258-61, 1986.
[122] N. P. L. G. Verhoeff, D. Hussey, M. Lee, J. Tauscher, G. Papatheodorou, A. A.
Wilson, S. Houle, and S. Kapur, "Dopamine depletion results in increased
neostriatal D2, but not D1, receptor binding in humans," Molecularpsychiatry,
vol. 7, no. 3, pp. 233, 322-328, 2002, PMID: 11920160.
158