Transcript Transdisciplinary Imaging Genetics Center

NAMIC Core 3.2
Opportunity & Challenges
Develop methods for combining imaging
and genetic data: imaging genetics links
two distinct forms of data
 Goal: Understand brain function in the
context of an individual’s unique genetic
background
 It is assumed that the integration of these
field will provide new knowledge not
otherwise obtainable: knowledge
discovery

Opportunity & Challenges




Schizophrenia as the exemplar:
Heterogeneous symptoms and course;
Heritable;
Subtle differences in structure and function;
Must involve brain circuitry
Challenges: Behavior and performance, cause
and effect, medication, structure and/or function
Genetic background influences brain
development, function, and structure in both
specific and non specific ways
The challenges
Standard but subjective diagnostic
assessments
 Time course of the disease

– Unclear relationship between clinical profiles,
genotype, and disease progression
– Multiple genes involved
– Multiple internal/external influences

Multiple levels of study, from molecular to
behavioral
A Collaborative Approach to Research
To understand the time course of the disease –
why first episode patients become chronically ill
Premorbid
Poor
15
Prodrome
Function
• First Episode
Good
20
Stable
Relapsing
Progression
30
40
50
Age (Years)
Sheitman BB, Lieberman JA. J Psychiatr Res. 1998(May-Aug);32(3-4):143-150
?
Improving
60
70
Statistical Parametric Map
Mai et al Human Atlas, 2001
??
??
 Fallon’s
PFC’s importance
Implied Circuitry: visual attention and orienting
Clozapine: The First Atypical Antipsychotic

Efficacy
1980s
– Reduction of positive and negative symptoms
– Improvements treatment refractory patient
– Reduction of suicidality in SA & schizo. patients

Side effects
–
–
–
–
–






low EPS,  TD
risk of agranulocytosis
risk of respiratory/cardiac arrest & myopathy
moderate-to-high weight gain
potential for seizures
Receptor binding
– Lowest D2 affinity
– Highest D1 affinity
Potkin et al ,2003
Clozapine Challenges Dogma
 The
EPS associated with
conventional antipsychotics led to
the misconception that EPS were
required for an antipsychotic
 Clozapine’s
lack of EPS
established that EPS are not a
necessary for a therapeutic
response
AIMS Scores for DRD3 Msc I Polymorphism after19
Typical Neuroleptic Treatment
16
14
12
Corrected
10
Mean
8
AIMS
score
6
4
2
1,1
1,2
2,2
0
Ser/Ser
n=34
Ser/Gly
Gly/Gly
n=53
n=25
DRD3 Genotype
F[2,95] = 8.25, p < 0.0005, Power = 0.568, r-square=0.297
Basile et al 2000
UCI Brain Imaging Center
FDG Metabolic Changes With Haloperidol
By D3 Alleles
Gly-Gly
Other Alleles
Negative Symptom Schizophrenia
Failure to activate
frontal cx
Cerebellar attempt to
compensate
Potkin et al A J Psychiatry 2002
The COMT Gene
CHROMOSOME 22
22q11.23
1
22q11.22
27kb
PROMOTER
5´
COMT-MB START CODON
TRANSMEMBRANE SEGMENT
COMT-S START CODON
210 BP
5´-CTCATCACCATCGAGATCAA
NlaIII
NlaIII
5´-GATGACCCTGGTGATAGTGG
NlaIII
G1947  A1947
…CGTG…
..AGVKD..
STOP CODON
PCR

COMT-MB/S:
NlaIII
…CATG…
Val158/108  Met158/108
..AGMKD...
high-activity (3-4X)
low-activity (1X)
thermo-stable
Low Dopamine Available
thermo-labile
More Dopamine Available
SOURCE: NCBI, GEN-BANK, ACCESSION # Z26491
NlaIII
Dopamine terminals in striatum and in
prefrontal cortex are not the same
Striatum
DA
DA transporter
DA receptor
Prefrontal cortex
COMT
NE transporter
modified after: Sesack et al J. Neurosci 199
Weinberger, ICOSR, 2003
COMT Genotype Effects Executive Function
Perseverative Errors (t-scores)
WCST
60
55
50
45
40
35
sibs
30
vv
vm
mm
patients
n = 218
n = 181
controls
n = 58
COMT Genotype
Genotype Effect (F=5.41, df= 2, 449);
p<.004.
Egan et al PNAS 2001
COMT Genotype and Cortical Efficiency During
fMRI Working Memory Task
Val-val>val-met>met-met use more DLPFC to do
same task, SPM 99, p<.005
Egan et al PNAS 2001
Proto-endophenotypes

Combinations of
–
–
–
–
–
Imaging measures (sMRI, FMRI, PET, EEG)
Genotypes
Clinical profiles
Treatment response
Cognitive behavior
Iterative refinements to develop
endophenotypes
 Studies like these represent a wealth of
potential information ---if they can be
combined

Goals
Combine neuroimaging
DNA
With behavioral and
clinical measures
and genetics
0.30
0.26
0.22
0.18
0.14
0.10
0.06
0.02
-0.02
-0.06
-0.10
-0.14
DRD1
5’
-
0.18
0.15
0.14
0.11
ARIP - 20MG ARIP - 30MG RISP - 06MG PLACEBO
Treatment Group
5 6 2 8
1
+
3’
-48
A
To identify useable
endophenotypes &
targeted therapeutics
8
5’
-
3’
-48
G
+
3
Inherited
genotype
Neuroimaging
Clinical and cognitive
measures
How many genes are needed for one disease ?

In complex traits, genes act together and we must
understand “how” if we want to understand the biology of
disease:
modelling gene^gene interactions – the Epistasis effect
Gene A
Gene B
+
+ +
+ + +
+ +
+
+++++++
G72 / 13q
DAAO / 12q
3
MDAAO-5
3
M-22
p value=0.01
p value=0.01
2
2
p value=0.05
p value=0.05
1
1
0
0
106.4 Kb
Risk
Odds
Standard
Model
genotypes
Ratio
error
0.76
G72-DAAO M-22_AA
1.89
M-22_AG
1.82
0.72
MDAAO-5_TT 1.05
0.75
M-22_AA*/
MDAAO-5_TT 5.02
3.95
M-22_AG*/
MDAAO-5_TT 1.73
1.30
C.I. : confident interval ;O.R. : Odds
Odd Ratio
Ratio
120.7 Kb
z
1.59
1.52
0.06
P>|z|
0.11
0.13
0.95
[95% C.I.forO.R]
0.86 - 4.14
0.84 - 3.95
0.26 - 4.24
2.05
0.04
1.08 - 23.45
0.73
0.47
0.40 - 7.52
Strategies for Discovering Novel Candidate
Genes & Drug Targets in Schizophrenia
Candidates From
Replicated Genome Wide
Microsatellite Surveys
Identifying “Hotspots” &
and Genes in ROI
Candidate
Genes
Candidates From
Microarray Screens
(30,000 Genes)
Plus validation with
In situ hybridization
Knowledge of
Pathophysiology of
Neuronal Circuits
Candidates From
Neurotransmitter Systems
Pharmacology of Disease
Candidates From
Microarray Studies in Animals
Drug Models
(e.g., PCP, amphetamine)
Treatment Models
(e.g, neuroleptics)
Computer analysis
Probabilities of
medication response
and development of
side-effects
Efficacy Negative Cognitive DM Weight
Suicide
Clozapine
90 80
25
50 85
2
Asenapine
90 80
50
10 15
?
Olanzapine
80 70
20
70 90
4
Ziprasidone
85 75
30
20 10
?
Neuroarray
WWW:
Analyze Image
Aim 1: Imaging Genetics Conference


The First International Imaging Genetics
Conference was held January 17 and 18,
2005.
To assess the state of the art in the various established
fields of genetics and imaging, and to facilitate the
transdisciplinary fusion needed to optimize the
development of the emerging field of Imaging Genetics.
Legacy Dataset
 fMRI
 PET
 Structural
MRI
 Genetic - SNP
 Clinical measures
 Cognitive measures
 EEG
– 28 subjects, chronic Sz
fMRI: Working Memory

Sternberg task:
5 6 2 8 1
+
8
+
3

Example Results
PET: Continuous Peformance Task

Continuous
Performance Task
(CPT)
– Sustained attention
– Selective attention
– Motor control task
+
0
+
9

PET results:
– Same as fMRI except no
time course data
Structural MRI
 Cortical
thickness measures in mm
 By defined region
Genetics
5HT2A
(T102
C)
DRD2(B DRD2(T DRD2_r
DRD1(D
stNI)
aq1A
s179
deI)
_141
)
9978
DRD2_r
s180
0498
DRD2_r
s464
8317
5058
22
11
22
12
11
11
11
5059
12
11
22
11
11
22
11
5061
12
11
22
12
11
12
11
5064
12
12
22
11
11
11
11
5024
22
11
12
12
11
12
12
5028
22
22
22
11
11
12
11
5030
12
22
22
11
11
12
22
5034
12
12
12
12
11
12
5035
12
11
22
22
11
11
11
5037
12
12
22
11
11
12
11
Clinical Scores
 PANSS
– Thirteen subscales/factors
– Positive, negative, and global summary
scores
– Lindenmayer 5-factors summary
– Marder 5-factors summary
Cognitive Scores
Immediate Word List Recall Total (total words recalled across all 3 trials)
Delayed Word List Recall Total (total words recalled from the 15 presented, after ~25
min delay)
Delayed Word List Recognition Total (total words correctly identified, when presented
visually with 35 distractor words after ~25 min delay)
Visual Recognition Correct (total correct hits; pt is shown 15 geometric shapes, then
those are mixed with 15 similar, distractor, shapes, and pt says 'Yes, I saw that one',
or 'No, I didn't see that one'.
Visual Recognition Correct (total false alarms; pt says 'yes', when he should've said 'no')
Visual Retention Ratio (calculated as Vrcor/Vrfa)
Letter Number Span (total correct; pt hears mixed up numbers and letters, which they
must recite in order--numbers, small to large and then letters--alphabetically.)
Trails A (time to complete a task of connecting numbered circles in order)
Trails A Errors (incorrect numbers connected)
Trails B (time to complete a task of connecting alternating numbered and lettered circles
in order)
Trails B Errors (incorrect numbers or letters connected)
Example Query of Federated Database
How can you predict which prodromal subject will develop
first episode schizophrenia ?
Integrated
View
Mediator
Wrapper
Wrapper
Wrapper
Wrapper
Wrapper
Wrapper
0.30
0.26
0.22
0.18
0.14
0.10
0.06
0.02
-0.02
-0.06
-0.10
-0.14
PET & fMRI
PubMed,
Expasy
0.18
0.15
0.14
0.11
ARIP - 20MG ARIP - 30MG RISP - 06MG PLACEBO
Treatment Group
Structure
Receptor Density
ERP
Web
Clinical
Anatomical Accuracy
Anatomical Accuracy
Anatomical Accuracy

Operational Plan (Fallon led effort)
– Step 1. Core 3-2 will develop operational criteria and
guidelines for differentiation of areas and subareas.
– Step 2. Core 3-2 will develop 10 training sets in which areas
and subareas of BA 9 and 46 have been differentiated as a
rule–based averaged functional anatomical unit
applied to individual subjects.




Needs to be applied to UCI 28 by Tannenbaum
Gliches in Freesurfer, Slicer must be overcome and
features added eg subcortical white matter
segmentation for tractography
Extend to visualiztion (Falco Kuester)
Supplement Slicer with multiple segmentation programs
in addition to Freesurfer
Anatomical Accuracy

Specified Operational Plan
– Step 3. Core 1 will develop algorithms and
methods for defining areas based on the
training dataset.
– Step 4. Iterations of Steps 1 through 3 will
perfect and validate the various methods for
defining areas.
– Step 5. The area identification methods will be
implemented by Core 3.
– Step 6. Validation of the methods by Core 3-2
on new set of subjects.
Identified 80 ROIs Relevant to DBP
of Schizophrenia
LEFT AMYGDALA.txt*
RIGHT AMYGDALA.txt*
LEFT ANGULAR GYRUS.txt*
RIGHT ANGULAR GYRUS.txt*
LEFT ANTERIOR CINGULATE.txt*
RIGHT ANTERIOR CINGULATE.txt*
LEFT ANTERIOR COMMISSURE.txt*
LEFT ANTERIOR NUCLEUS.txt*
RIGHT ANTERIOR COMMISSURE.txt*
RIGHT ANTERIOR NUCLEUS.txt*
LEFT BRODMANN AREA 10.txt*
LEFT BRODMANN AREA 11.txt*
LEFT BRODMANN AREA 13.txt*
RIGHT BRODMANN AREA 10.txt*
RIGHT BRODMANN AREA 11.txt*
RIGHT BRODMANN AREA 13.txt*
LEFT BRODMANN AREA
LEFT BRODMANN AREA
LEFT BRODMANN AREA
LEFT BRODMANN AREA
LEFT BRODMANN AREA
LEFT BRODMANN AREA
RIGHT BRODMANN AREA
RIGHT BRODMANN AREA
RIGHT BRODMANN AREA
RIGHT BRODMANN AREA
RIGHT BRODMANN AREA
RIGHT BRODMANN AREA
17.txt*
18.txt*
19.txt*
1.txt*
20.txt*
21.txt*
LEFT BRODMANN AREA 22.txt*
LEFT BRODMANN AREA 23.txt*
LEFT BRODMANN AREA 24.txt*
LEFT BRODMANN AREA 25.txt*
17.txt*
18.txt*
19.txt*
1.txt*
20.txt*
21.txt*
RIGHT BRODMANN AREA 22.txt*
RIGHT BRODMANN AREA 23.txt*
RIGHT BRODMANN AREA 24.txt*
RIGHT BRODMANN AREA 25.txt*
Circuitry Analysis

Specified Operational Plan
– Step 1. Core 3-2 will collaborate with Core 2 to
implement algorithms for structural equation modeling,
and the canonical variate analysis.

Fallon & Kilpatrick, piloted but as a first step need to
better quantify and automate ROI based on literature,
Knowledge Based Learning as a general tool.
– Step 2. Core 3-2 will use step 1 software to test Core 32 hypotheses.
– Step 3. Core 3-2 in collaboration with Core 2 will extend
the canonical variate analysis methods of Step 1 to
determine images that distinguish among tasks, clinical
symptoms, and cognitive performance.
– Step 4. Core 3-2 and Core 1 will collaborate to integrate
canonical variate analyses with machine learning
approaches for detecting circuitry.
Genetic Analysis in Combination
with Imaging Data

Specified Operational Plan
– Step 1. Core 3 will type multiple genetic
markers at selected genes relevant to
schizophrenia and brain structure.
– Step 2. Core 2 will extend Toronto “in-house”
Phase v2.0 software for measuring two genegene interactions to multiple genes and make
the software more user friendly to
neuroscience and genetic researchers in
general.
– Step 3. Core 3-2 will determine linkage
disequilibrium structure on the genetic data
using specific programs such as Haploview,
GOLD, and 2LD and construct haplotypes.
Genetic Analysis in Combinatin
with Imaging Data

Specified Operational Plan (cont.)
– Step 4. Core 3-2 will complete genetic
analyses on the haplotypes developed,
identified by the Core 3-2 software in Step 3,
and test for gene-gene interaction using
refinement of Toronto Phase v2.0 software
from Step 2.
– Step 5. Core 3-2 will collaborate with Core 1 to
develop methods for combining genetic and
imaging data using machine learning
technologies and Bayesian hierarchical
modeling.
– Step 6. Iterations of Step 5 will develop
predictive models and suggest hypotheses.