Transcript Depression

How Functional Brain Imaging Can Help
Speed Drug Development and Clinical Trials
Depression
Helen S. Mayberg, MD
Emory University School of Medicine
ASENT meeting 2012
Washington DC
Disclosures
Grant Support:
NIMH, CIHR, NARSAD, Dana Foundation,
Stanley Medical Research Fund, Woodruff Fund
Off-Label Use of Devices: DBS electrodes/pulse generators
1. Medtronics Inc. (U Toronto)
2. St. Jude Medical, Inc (Emory)
Patent:
US2005/0033379A1 (Andres Lozano, co-inventor)
issued March 2008, St. Jude Medical Inc, assignee
Consultant: St Jude Medical Inc / Neuromodulation Division
Emory DBS study: FDA IDE: G060028 (PI: HM)
Clinicaltrials.gov ID#: NCT00367003
devices for research donated by SJM
Imaging Wish-List: Science, Trials, Care, Dev’t
Diagnostic Markers




illness subtypes (heterogeneity for clinical trials)
risk identification (pre-symptomatic intervention?)
response predictors (placebo, responders, nonresp, resistant)
relapse, recurrence potential (Tx continuation, ID hi risk pts?)
Evidence Based Treatment Algorithms
•
•
•
•
Triage pateints for different trials
Identify placebo responders in advance of trials
tailor treatment to what the brain needs
know in advance what treatments won’t work
Needed studies




circuit characterization; variability; genetic, clinical correlates
define treatment specific response pathways (psychotx, drug, somatic)
determine what changes are critical; early surrogates
reliability, practicality of such biomarkers in individual patients
Context: Current State of Treatment Options
 Treatments available but one size does not fit all
• < 40% achieve remission (drug, CBT, other)
• placebo response common in trials
• > 10% become treatment resistant over time
• ECT > 50-70% Remit but > 50% relapse in 6 months
• rTMS 18-24% Resp in 6wks, limited efficacy in pt > 1 failed AD Tx
• VNS 30% Resp at 1yr but <20% long-term Resp
• ketamine (rapid effects, but unsustained)
 Limits to progress, Innovation
•
•
•
•
no pathology, clinical heterogeneity, no clear biomarkers
50 year focus on monoamines, few new leads
animal models: none capture recurrence, relapse, resistance
overinclusive, nonspecific outcome measures,
w/ all symptoms treated equally (COMPARE TO PD)
Hypothesis: Depression and the Brain
gender
family history
temperament
genetics
pre-natal insults
endophenotypes
Subphenotypes
MDD, BP
Melancholic
Atypical
Recurrent
TRD
Biological
Vulnerability
Exogenous
Stressors
homeostasis
Mood Regulatory
Circuits
stress
recovery
Depressive
episode
Phenotypes
post-natal insults
early abuse
life events
medical illness
Regions
Connections
Chemistry
Rx Effects
CBT/PT
Medication
ECT, rTMS, VNS
DBS
P
F
Defining Depression Circuits 1
Identify circuit constituents
Focal Strokes  MRI volume, Glia  MRI volume
PF
Structure
CT, MRI,
pathology
hc
Robinson 1983
Parkinson’s
aCg
Function
PET, fMRI
EEG
Frontal
Cingulate
hippocampus
F9
Drevets 97; Ongur 98
Unipolar
aCg
F9
Mayberg 19990
Bipolar

F9
F9
P40
Sheline, 1999
P40
Mayberg 1994, 1997
F9
P40
F9
P40
Kruger 2003

Frontal
Cingulate
Parietal
Also
Amygdala
Basal
ganglia
Defining Depression Circuits 2
Changes with well characterized treatments
1 week
fluoxetine
pCg
hc
hc
vst
Fr
p
vst
p
Cg25
6 weeks
fluoxetine
Fr
pCg
hc
cg25
vst
p
Fr
hc
ins
p
Mayberg et al.
Biol Psychiatry 2000
Similar time course to neurogenesis, BDNF ∆
Cg25
Subcortical
Brainstem
Limbic
early
Limbic
switch
+
Cortex late
Defining Depression Circuits 2b
responder-nonresponder differences
Cg25
Fluoxetine
Responders
F9
pCg31
hc
hc
Cg25
Cg25
NonResponders
p
F9
F9
hc
hc
Failure to Switch = Non-Response
pCg31
Common Changes Placebo and SSRI
Drug = Placebo Plus
Cg25
Placebo
fluoxetine
Common
Cg25
PCg
Fr9
pCg
Fr9
cg25
Cg25
Cg25
Fr9/46
Active
Fluxotine
pCg
hc
cd
Cg25
cg25
p
hc
p
distinguish Placebo R from Active Drug response with scans?
Am J Psych 159: 728-37, 2002
Also
Hc
BS
Defining Depression Circuits 3
Drug Resp vs Nonresponders
Baseline
Pre-genual Anterior Cingulate 24

F9 pACC24 F9
→
pACC
(r24)
pACC
(r24)

Drug responders
Non-responders
Common Frontal change
Mayberg et al
NeuroReport 1997
Multiple interactive Nodes
More than 1 area of Cg involved
First clue to potential subtypes
rACC
Baseline EEG Theta R>NR to TCA
Pizzagalli AJP 01
Hypothesis
Scan =“insult”+ongoing compensation
baseline heterogeneity defines clinical subtypes
Scan Type
Trigger
overcorrection
A
CBT
illness is failure
to self-correct
network
activity
B partial
meds
recovery
Bad day
symptoms
C
under
failed
Depression
diagnosis
absent
D
adaptive brain
response
Mayberg, J Clin Invest 119:717, 2009
ECT
DBS?
Hypothesis: recovery is
optimized by matching
treatment to state of
network dysregulation
Proof of Principle
Comparison drug to CBT
mF10/9
Change
with
clinical
response
PF9
MCC
PF9

mF9/10
P40
SCC
SSRI (paroxetine)
dPF
dPF
HamD 22+3  6+4
dPF
dPF
Cg24
vPF
vPF
UPD Group 1
Kennedy et al. Am J Psych 2001

Cognitive Behavior Therapy
HamD 20+3  6.7+4
Baseline
Pretreatment
Pts vs Controls
comparable
severity
oF11
thal
vPF
vPF
UPD Group 2
Suggests
Baseline differences
Impacting ultimate
Response to a specific
Treatment
Need to know if it also
Predicts non-response to
The alternative
Goldapple et al. Arch Gen Psych 2004
Evolution of Depression Circuit Model
Template to consider different treatments, common effectts
Cognition
(attention-appraisal-action)
PF
PF9/46
Cg25
PM6
Par40
hc
PCC
MCC
CBT
Emotion Regulation
Self-awareness
insight
mF9/10
thal
amg
mb-sn
pACC24
oF11
Is any one mode
Or clinical feature
Most critical?
Mayberg, Br Med Bul 65:193-207, 2003
Mayberg, J Clin Invest 119:717, 2009
na-vst
Mood
state
Salience
Motivation
sACC25
a-ins
hth
bstem
Interoception
(drive-autonomic-circadian)
MEDS
PF
Meds
PCC
Cg25
P
BS
Isolating Key Components
focus on negative mood
R
Recovery w/SSRI
FDG PET
Transient Sadness
CBF PET
F9
F9
ins
ins
Cg25
Cg25
+
Cg3 4
1
z

Cg31
Cg25
Depressed Patients
4
z
Cg25
Cg25
Cg25
Limbic + Cortex

Reciprocal
Cingulate-Frontal
changes
Healthy Volunteers
Mayberg et al. Am J Psych 156:675-82 1999
Critical Role of the Subcallosal Cingulate
Sad Memory
Tryptophan Deplete
volume; glia
Cortisol Correlate
∆ Spines/Dendrites
SCC
activity
Mayberg
SSRI
SNRI
Drevets, Ongur, Rajkowska
Kalin
Talbot
Placebo
rTMS
ECT
McEwen 1994 etc
VNS
SCC
activity
Mayberg
pre-Cingulotomy
Mayberg
Kennedy
Med NR
Pre-DBS
George
Ketamine
SCC
Dougherty
Greicius
Mayberg
Deakin 2009
Nobler
Hypothesis:
TRD=dysregulated
Cg25 connectivity.
Target the problem
at its origin
Pardo
Direct ‘Circuit’ Modulation using DBS
block aberrant sCg25 activity with 2° effect on connections
mF9
mACC
rACC
mF10
sCg25
4
3
2
1
oF11
PET target
Hth
nAc
Am/hc
Likely remote effects
Cortex
Cognitive control, action
F11
PF9
F10
Cg24 MCC PCC
sCg25
sn
vst
Thal
bs
Striatal-thalamic
drive, motivation
am
hth
ins
hippocampus
Limbic
circadian, stress responses
MRI: target localization
Focus: Treatment Resistant Depression
Toronto: Pilot Proof of principle
Pre-op MRI
Post-op MRI
Toronto Proof of Principle
Pilot: 6 severe TRD, GAF<50
Illness duration avg 5.6 yrs
Failed mult meds, CBT, ECT
6 mo open DBS
4/6 Resp; 3/6 remission
Pre-op PET
∆ 6 months DBS

mF9
dACC
dACC
cc
g
sgCg
vst
ac
SCC25 hth
Electrode
Targeting
Confirm electrode
placement
First patient May 13, 2003
F10
sn
Pts vs Controls
vst
oF11 C25
hth
oF11
 C25
Responders
Funded by NARSAD, Toronto Western hosp
Toronto Long-term Followup
Emory Sham Controlled Trial
3-6 yrs, n=14
Resp
Rem
62.5%
18.8%
46.2% 75%
15.4% 50%
64.3%
42%
IT
OC
avg=42 mo
years after implant
Kennedy S, et al. Am J Psych in Adv Feb 1, 2011
Lozano A, et al. Biol Psych 64:461-67, 2008
HDRS-17 score
24
Remission Response
6 mo
18%
41%
1 yr
36%
36%
2 yr
58%
65%
BP-D/MDD
N=17
18
12
6
No change
in meds for
6 months
0
BL sham 1m
2
3
4
5
6
7
8
9
10
11
12
Holtzheimer et al. Arch Gen Psych Feb 2012
2y
Responder/Nonresponder Differences
surgical precision vs remote effects
Planned Target
Active Contact
Map Remote Effects
cc
g
mF10
mACC
25
mF10
Resp
Hamani
et al localization
J Neurosurg 2009
Simple
uninformative.
Hitting the ‘target’ is not the problem
4
nA
Non-R
ac
sCg
mF10 MCC
oF11
4
3
2
1
Hth
nAc
Am/hc
3
putative tracts
oF
nA
25
Clues from PET changes?
both
mF9
F10
dACC
vst
oF11
C25
hth
Responders
oF11
DTI/DSI
C25
hth
Non-Resp
Local PLUS remote effects
25
nA
Probablistic Tractography
Variable impact on remote ROI
Can this be linked back to patient behavior?
Presurgical Response Predictors
towards optimal patient selection: resting fMRI
Independent Component Analysis (ICA)
Resting State
BOLD fMRI
DBS pts
Difference
Controls
Similar to
PET
Can
potentially
be done in
individuals
mF10
ICA
default mode
component
-
=
SCC25
4
3
ICA - Zscore
Correlation:
baseline
fMRI DFM
with 6 mo
outcome
2
1
SCC FC
R² = 0.6133
 worse IDI-D* better 
0
0.0
0.5
1.0
Alex Franco, 2011
Holtzheimer et al SOBP 2011 abstract
Presurgical Response Predictors
towards optimal patient selection
Baseline resting EEG
1.4
1.2
EEG, 32 sites, Bio-Semi System
4min rest, eyes open
1
0.8
6m Resp
0.6
6m Non-Resp
0.4
  
0.2

0
0 5
10
15
20 25 30 40
45 50 Hz
Similar location to PET and fMRI
Confirms findings, could be a
more practical alternative
6 mo HDRS Change
Baseline
1 mo DBS
6 mo DBS
100
R² = 0.576
80
60
40
20
0
-100
0
-20
Broadway, Hilimire , Corballis. GA Tech unpublished
100
200
300
Theta%Chng at 4 weeks
400
Towards Novel Drug Development
Chemical Specificity within the Cingulate
DBS effects
Ketamine acute

sad induction
Trypt depletion



Deakin AGP 2009
Human
Post
Mortem
Talbot BP 2004
Human
Whole
Brain
Autoradiography
sACC
Hi SERT, 5HT1a
Arango et al Prog Br Res 2002
Hi NMDA
Lo GABA-b
Palomero-Gallagher Human Br Mapping 2009
Future: Imaging Biomarkers
Guide DBS patient selection and parameter optimization
Resting BOLD fMRI
to confirm DBS type
Micro-electrode
Lead localization
DTI tractography
Define optimal contact
mF
mF10
mF
mF10
Cg24
Cg32
BA10
25
sCg
SCC25
sCG
nA
nAc
Amg
Intraoperative LFP
Tune critical 
Voltage Steering:
Volume of tissue activated
Realtime Readouts:
Closed loop adjustments
post
Ipsilat Fr
Bilat Fr Pole
Contral Fr
collaborations at Emory, Yerkes, GA Tech, Cleveland Clinic
Vertex
Depression DBS Collaborators
Emory Clinical DBS 2005Neurosurgery/Neurology
Robert Gross, MD, PhD
Paul Holtzheimer MD
Klaus Mewes, PhD
Steven Garlow, MD PhD
Kevin Gotay, MS
Patricio Riva Posse MD
Donald Bliwise, PhD
Dylan Wint, MD
Kathryn Rahimzadeh, RN
Lori Ritschel PhD (CBT)
Mahlon DeLong, MD
C Ramirez PhD (CBT)
Thomas Wichmann, MD
Sinead Quinn
Psychology/Physiology
Kelsey Hagan
Stephan Hamann PhD
Megan Filkowski
Cory Inman
Andrea
Barrocas
Emory Depression Biomarkers
Otis Smart, PhD
Margaret
Craighead
Ed Craighead
Mike Jutras, PhD
Andrea Crowell MD
Boadie Dunlop
Beth Buffalo, PhD
Tanja Mletzko
Paul Corballis, PhD (GTech)
Imaging Lab
CB Nemeroff
Matt Hilimire BA (GT)
Alex Franco, PhD
Jim Broadway PhD (GT)
Callie McGrath, BS
Yerkes/Animal Models
Amy Alderson, PhD (NPsy)
KiSueng Choi, MS
Donald Rainnie PhD
External Collaborators
Mary Kelley, PhD
Teresa Madsen BS
H Johansenberg PhD (UK)
David Gutman, MD
Leonard Howell PhD
N. P-Gallagher PhD (GR)
C Craddock, PhD
Mar Sanchez PhD
C McIntyre PhD (Ohio)
Jared Moreines, BS
Sue Tye, PhD (AUST)
Clement Hamani MD (TO)
Grants: NARSAD, Woodruff Fund, Emory Healthcare, Stanley Medical
S Pannu PhD (Berkeley)
Research Institute, Dana Foundation, NSF CBN Venture, K23
MH077869,R01MH073719, P50MH077083, RO1MH080880
M Ghovanloo PhD (GTech)
Johns Hopkins 1985-91
UTHSCSA 1991-98
Toronto 1999-2004
Andres Lozano MD PhD
Sidney Kennedy MD
Clement Hamani MD
Zindel Segal PhD (CBT)