Paths to Recovery:Paths to Recovery: Strategies and Insights from Functional

Neuroimaging Studies of Depressiong g p

Helen Mayberg MDHelen Mayberg, MDDepartments of Psychiatry & Neurology

Emory Universityy y

Context: Development of New Treatments for DepressionTreatments for Depression

What we need to know • the “illness” circuit• response pathway(s)• what changes are critical (acute, short, long-term)

anatomical, chemical, electrical• Can these be selectively targetedCan these be selectively targeted

Strategiesffi h i di• efficacy vs mechanisms studies

• time course; R/NR differences• predictors (treatment selection)• predictors (treatment selection)• vulnerability (recurrence, at risk)

Prototype Disorder: Parkinson’s DiseaseSN degeneration with ↓DA

NLNL PDPD

Circuitry changes over time

SND2 D1

18-F-DOPA PET

BG

NLNL PDPD

Ph thPDNL

S ( bl ti /DBS)Pharmacotherapytargets 1° DA Deficit

Surgery (ablation/DBS)Target 2° Motor Circuit changes

On Depression

“It i iti d“It is a positive and active anguish,a sort ofpsychical neuralgia wholly unknown to normal life ”normal life.”William James The Variety ofThe Variety of Religious Experience

What are the critical circuits?

Strategy 1: Anatomical StudiesStrategy 1: Anatomical Studies

BGFr

Post-Stroke LesionsPF9

MRI volume Glial number

hc Cg24d

Cg24/25

Robinson RG et al 1984 Y. Sheline, 1999 OF47

OF47

G. Rajkowska 2002Drevets 1997

cause or effect?

Drevets 1997Ongur 1998

Strategy 2: Functional Studies

vF

Cg

vFBG Disordersw/ depression

FDG PET

HDPD Stroke

mr/wbFDG PETDep vs NonD Pts

F9F9F9

F9+incPrimary

Affective

P40

Cg F9

P40

F9

d

AffectiveDisorders

FDG PETPts vs Controls

Unipolar Bipolar

- decPts vs Controls

Putative Depression Circuit

Cognitive Processingattention-memory-action

pCgaCg24b mCg24c

PF9/46 Par40PM6 hc

rCg24amF9/10 cd-vst thal

Emotion-Cognition Integration

salience

moodstate

oF11rCg24a

amg mb-sn

Covert

salienceself-referencereinforcement

sgCg25

hth bstema-ins

Autonomic Responsesarousal-vegetative-circadian-internal milieu

At Issue: Variability

Unipolar Dep Unipolar DepUnipolar Dep Group 1

aCing

Unipolar DepGroup 2

aCing24

Pre Frontal 9 Prefrontal 9Pre Frontal 9 Prefrontal 9

↑↓M t b li Diff P ti t H lth C t lMetabolic Differences Patients vs Healthy Controls

Clue to Subtypes?Clue to Subtypes?

Depression subtypes may be critical:Differential Treatment Response CBT vs Drug

Dep w/o ELS C<D<BothDep with ELS

D<C=Both**Dep with ELS

Potential Sources of Variability

Biological Vulnerability

ExogenousStressors t t l i ltgender Vulnerability tressors

homeostasis

post-natal insults early abuse life events

medical illness

genderfamily history

gene polymorphismstemperament

Pre natal insults

Mood RegulatoryCircuits

Pre-natal insults

Circuits

Re-stabilized De-stabilized

Depressiveepisodep

phenotype

Variability: Alternative ViewScan “ins lt” + compensation

Brain Response

Scan= “insult” + compensation

Cingulo-Frontal Trigger

Over-correction

Brain Response

therapyFrontalcircuits

real-time partial

remissionpy

activity failed

S t

medication

ECTabsentSymptom onset

ECT

other somatic

DepressionDiagnosis

Treatment selection

Strategy 3: Mapping Core Symptoms

Hypothesis: Mood is THE Primary SymptomHypothesis: Mood is THE Primary Symptom interrupting circuits sustaining this behavior is the GOAL

Mapping Mood Regulatory Circuits: First Studies in Healthy Volunteers

createsad memory

Mood InductionRecollect

First Studies in Healthy Volunteers

yscript

readsad memory

script

Recollect

Re-experience Feel

whereis IT?

CBF PET

Mayberg et al. A J P h

pCgF9Cortex

Am J Psych 156:675-6821999Cg25Cg25aIns HthaInsLimbic

Cg25 Activity and Negative Mood reactivity

Anatomical Variability

Sad Mood inductionWomen volunteersCBF PET

Cg25

Tryptophan Deletion

Cg25

Mayberg, AmJPsych 1999

Cg25

Amg

Tryptophan DeletionMen volunteersCBF SPECT Sad Mood correlates

Cg25

SERT S/S < L/LPezawas et al NMed 2005Talbot, Neuropsychopharm 2005

Cg25

n=51Resting StateHealthy volunteersCBF PET

Cg25

Zald, PNAS 2002

neg affect prev wk

Mapping Mood Regulatory Circuits IIStudies in at-risk populationsStudies in at risk populations

Studies in at Risk PopulationsDepression Stress test?

mF Cg24b

Depression Stress test?

Healthy Volunteers

Cg25 th

mF

Cg25 Cg25

Cg24bVolunteersNEO

risk/resilience

R itt d Pt

Controls

g

HiN LoE

Cg25

LoN HiE

24b

24a

24bmF10

mF9

24amF10 mF10

24b

24a

24b

24a

Remitted Pts Relapse risk

Familial

Rem UP Rem BP BP Sib

24a

oF11 oF11 oF11

24a24aFamilial risk/resilience

on SSRI on Li+ unaffectedKeightley NeuroImage 2003Liotti et al Am J Psych 2002 Krueger et al Am J Psych 2006 Treatment Targets?

Strategy 4: Treatment Effects Drugs vs PsychotherapyDrugs vs Psychotherapy

Brain aerobics

Post vs Pre Treatment Changes:Therapy ≠ MedicationTherapy ≠ Medication

Thalamus

preFrontal9

SubgenualCingulate

Paroxetine ThalHam 20+3

Parietal40

hippocampusCingulate

25Ham 20+3 post 6.7+4

midCing 24c

midCing 24cpreFrontal

9CBT

Ham 22+3 MedialFrontal hippocampusPost

cingulate Orb frontal

post 6+4

↑↓

Goldapple et al. Arch Gen Psych 61:34-41, 2004

CBT ≠ Placebo

CBTpCg PF9aCg24

mF9PF9

oF11

CBTmF9 mF9

Placebofluoxetine

pCg PF9↑

↓fluoxetine Cg25Cg25

PF9Changes

post vs pre

↓

Activefluoxetine

pCg PF9 PF9

Cg25Cg25 insbs

post vs pre treatment

bs

Synergistic Brain Changes: complementary targets

PF9 P40 pCgCBT

attention-cognition-action

PF9 P40

hippocampus -pCg

thalbgmoodstate

mF9/10

aCg24

insightself referencereinforcement

oF11

Cg25

hth bs

a-ins

amdrug-

CBT/SRI inverse

Autonomic-circadianSRI onlyCBT only

Time Course of Medication EffectsEarly subcortical-limbic / Late corticalEarly subcortical limbic / Late cortical

1-wk fluox hchc

pCg

p

Fr

cdp p

pCg6-wks fluox hh

Frcg25 Fr

cg25

p

fluox hcp

hc insCg25

cd

When is switch? Linked to BDNF, neurogenesis?

Time Course of Placebo EffectsExpectation vs Actual Recovery

Expectation1 wk vSt vSt

Cg24F9

No changeOF

vSt

OF

vSt

vSt-oF early

F9Cg24 F9Recovery

Cg25 ins Cg25 St

C 25/F9 l

6 wks Endpoint

Fluox RFluox NR PlaceboCg25/F9 late

Speculative Time Course CBT EffectsEarly Dorsal Cortical / Late limbic-paralimbicEarly Dorsal Cortical / Late limbic-paralimbic

8 weeks:F9 Cg24F9

F9IPTBrody 2003

Final Ham =12.6 (partial R)

↓ Fr Cg only↓ Fr-Cg only

mF10

CBTGoldapple

2004F9

mF9 Cg24

+4

mF9F9

Cg24

pCg pCg

mF102004vF F11hchc

- 4

16 weeks: Full remissionFinal Ham=4.7

↓Frontal + new ↑Hc + reversed ↑Cg

CBT Specificity: OF/mF + Cg24Cognitive System as Primary targetCognitive System as Primary target

+424b

m924b-c

m10

m924b-c

24a 24am10

oF

CBT Changes

- 4m10

Autobiographical Emotional

m10

oF

Emotionalg g pRe-experiencing Re-appraisal

↓ rumination (PF9) ↓ i i ( F9)

Self-Reference

B Levine, Rotman in reviewFossati et al. Am J Psychiatr 2003

Ochsner et al. J Cog Neurosci 2002

↓ re-experiencing (mF9)↑ active reappraisal of meaning (Cg24b) ↑new patterns of learning (hc)* no change in mF10 (self-reference) no change in mF10 (self reference)

Common Cg25 Decreases Across TreatmentsMood System as primary Target

SSRI

Mood System as primary Target

Placebo TMS ECT

↓Cg25

Nobler & SackeimM GeorgeLiotti Mayberg

Anatomy SERT S/S< L/L

AutobiographicalMemory (sad)

TryptophanDepletion (sad)

hc AmCg25Pezawas

Cg25Mayberg

Cg25Talbot

Responder vs Non-responder Differences

FluoxetineResponders

F9hc

Cg25

hcCg25

pCg31

RespondersCg25

Cg25

p

Non- F9 pCg31F9Non-Responders

F9hchc

pCg31F9

Strategy 5: Evidence-based Tx MarkersCoronary Artery Disease as prototype

Acute MIAcute MIAnterolateralinferior

Single Vessel occlusion:Stint / Angioplasty

Multi-Vessel Disease:Bypass graft

Baseline Predictors: Hypothesis

networkcompensation

A CBT neededOver-correction

Triggerremission

Cingulo-Frontalcircuits

partial eitherABReal-timeactivity

remission

meds needed

C

under-correction

y

DepressionDiagnosis

B

absent treatment resistantC

Pretreatment Scan

Diagnosis

Response Predictors: Frontal Cortex (pt vs C)Frontal Cortex (pt vs C)

F9 F9 F9 F9 +4z

C24

- 4z

UPD TorontoUPD Texas UPD TorontoUPD Texas-

Frontal Variability DOES NOT predict brain changesy p gor clinical response to a given treatment

Response Predictors: R t l Ci l tRostral Cingulate

+4z

C24 C24

- 4z

C24a C24a

Non-RespondersResponders

- 4z

Non-Responders Responders

Cingulate 24a variability DOES NOT distinguishResponse to a specific treatmentResponse to a specific treatment

(Drug, Sleep Deprivation)

Prediction of Response Group ROI multinomial logistic model

PC 1 PC 2 PC 3

ROI multinomial logistic model

significance

ba11 0.766 0.211 0.112

ba25 0.794 -0.007 -0.052Variable

CBT-R vs Par-R

ParR v ParNR

PC 1* 0.004 0.039

hc 0.597 -0.453 -0.350

lat9 -0.014 0.749 -0.039

PC 2* 0.063 0.706

PC 3* 0.455 0.368

r24a 0.216 0.611 -0.071

m10 0.361 0.452 0.450

Inte

rval

PC

1

athal 0.327 -0.496 0.656

postcg -0.358 0.022 0.555%

Con

fiden

ce I

n=83; *principle components unrotatedCBT R ParR ParNR

95%

Prediction of Response Group WB lti i l l i ti d lWB multinomial logistic model

PC3 hc

cg25

hc

11

hchc

Variable CBT-R vs P-R

P-R vs P-NR

PC 1* 0.002 0.491

PC1

PF9 M9/C24

hchc

PF46 C24/32mF10 vF47 oF11

PC 3* 0.005 0.591

PC 4* 0.811 0.068

PC1 cd

oF11 25 cg24

bs

oF11Cg25

hchc thcg25

hchc

cg24

PC4

CBT R par R Par NR

drug resistent: Cg25-Subcortical (hc, th, bs)

Direct Testing of a Treatment Response Network

PF9 mF10PF9 mF10

aTh Meta-analysis

Cg24aaTh

CommonCBT

n=119 pts

oF11Cg25 drug

Hc Path Analysis/Structural Equation Modeling Goal: most stable Model that fits all patient groups.Goal: most stable Model that fits all patient groups.

Seminowicz et al, Neuroimage in press 2004

Different Outcomes: Different NetworkConfigurations at Pre-treatment Baseline

Drug ResponsiveCBT Responsive Multiple Drug Failure

Configurations at Pre treatment Baseline

Cg25

Li bi C ti lC ti l C ti l Li bi S b ti lLimbic-CorticalPattern

Cortical-CorticalPattern

Limbic-SubcorticalPattern

dysregulated+ coeff- coeff

dysregulatedCg25 activity +

Fr disconnection

Supporting Evidence of Cg25 Role:Cingulotomy Response Predictor

Surgical

C 25

Surgical Site

Cg25

Baseline overactivity associated with better outcome

DD Dougherty et al J Neurosurgery 2003

Strategy 6: Cg25WM Deep Brain Stimulation (DBS) for Treatment Resistant Depressionp

Treatment Resistance Hypothesis:abnormal Cg25 connectivityabnormal Cg25 connectivity

Compensatory Cortical Response

Sustained↑Cg25activity recovered

OvershootCingulo-Frontalcircuits

Baselineactivity

activity recovered

partialactivity

absent

failed

absent

MDEGoal: Target

the maladaptivecompensatory

NeuroImag Clin NA 13:805-15, 2003

p yresponse at its

hypothesized origin: Cg25

Cg25WM DBS Study: Procedure

X X

Stereotaxic Frame, Local Anesthesia, MRI Targeting Cg25 WMbilateral quadripolar electrodes (Medtronic 3387)Parameters: monopolar, 60ms PW, 130Hz, ~4V (like PD)Criteria: MDE duration>1yr HDRS17>20;Criteria: MDE duration>1yr, HDRS17>20;

Min 4 failed Rx incl ECT; no co-morbidity3M/3W; Age onset: 29+12; past MDE: 4.7+5 MDE dur: 5.6+3yr;3M/3W; Age onset: 29 12; past MDE: 4.7 5 MDE dur: 5.6 3yr; Failed ECT: 5/6 Fam Hx: 5 of 6; Melancholia: 4/6;

Mayberg et al. Neuron, 45:651-660, 2005

Acute Intra-operative Stimulation EffectsContact and voltage specificCo tact a d o tage spec c

Spontaneous Self-Reports Sense of intense calm quiet reliefSense of intense calm, quiet, relief

• Dissipation of visceral symptoms• resolution of the ‘pain,’ dread, void, mental heaviness

Cg25, Insula?

Followed within 10-15 seconds by• ↑ interest, energy, awareness• ↑ attention, motor speed, spont speech F9-Par, NA?

• Δ visual perception; colors, clarity, brightness, details• Δ PANAS: ↑ positive; ↓ negative scores

Adverse EffectsAdverse EffectsNo autonomic, behavioral, mood changesMental slowness at top contacts near cc.

A shift from an all-consuming internal focusto the realization there are other things around to do…

Cg25WM DBS Study: Clinical Outcome

Hamilton-17 ScoreTime Pt 1 Pt 2 Pt 3 Pt 4 Pt 5 Pt 6 Mean

Pre-op Dep baseline 29 22 29 24 26 25 26+31 wk post-op (~2 hr daily) 5 10 12 18 17 12 12+52 wks post-op (off 1 wk) 9 13 23 18 22 n/a 17+62 wks post op (off 1 wk) 9 13 23 18 22 n/a 17 61 month (DBS ON) 10 14 17 20 22 12 16+52 months 13 11 12 18 10 12 13+33 th 2 15 14 25 7 14 13+83 months 2 15 14 25 7 14 13+84 months 4 9 12 24 6 12 11+75 months 5 18 7 22 8 n/a 13+76 months 5 15 9 23 6 12 12+79 months 6 -- 6 -- 5 10 6.8+212 months 1 -- 8 -- 4 4 4.3+212 months 1 8 4 4 4.3 2

as of Feb 1: 13 pts total; 8M/5W; age Range 28-71; 12 UPD; Similar OR effects in most; 9/13 meet Response Criteria (3 month min; most >6 month)

Cg25WM DBS: Putative Mechanisms

Hypothalamus:MOTORDLPFCOMPFC

2410 9

6 4ySleep

Appetitecortisol

SERTS1A

Cd-P2511

S1A S2A D3

GABA

ThalNA

Brainstem:5HT, NE, opiates H b J N i 2000

SNVTAAmygdala

Ongur and Price. JCN 1998

5HT, NE, opiates Haber, J. Neuroscience 2000

Cg25WM DBS CBF PET ChangesParameters: monopolar 60ms PW 130Hz 4VParameters: monopolar, 60ms PW, 130Hz, ~4V

mF10Cg24Cg31Cg24

Which changesAre due toStimulation ofWhite matter

Cg25hthbs

Cg25Target

Which are dueTo transynapticOr plasticity

F9F9F9 F924

gChanges

Need immediateeffects

Recovered

Cg25ins

Dep Baseline

Cg25ins

Post Op Recovered6 mo vs Baseline

Ham17=7.8+3

Dep BaselinePts vs Controls

Ham17=27+2

Post-OpMRI

Proof Of Principle: targeting a critical node has widespread impact

pCgCg24mF9

Cg24mF9

targeting a critical node has widespread impact

Cg25DBS Cg25

mF9

oF11 cg25oF11Hthhth

Cg25WMDTI

tractographybs ?bs Hthhth tractography

fluoxetinepCgCBTpCgCg24mF9

Cg25

bsoF11

Foundation for Studies of R/NR Diff, Unique Chemical Targets

Patient’s Perception of DBS Effects( h / h i hi ?)

Patient 5: 2 months of Stimulation

(where/what is this?)

Patient 5: 2 months of Stimulation.“ the most fundamental change that I can see, is that it isn’t like something has been added—no somethingisn t like something has been added no, something has been taken away. That heavy sinking vortex feelings was always there in some form or another. And

it i ( t i h t t t?)now it is gone. (acute primary change at target?)

It is as if instead of being in the grand canyon, you are now up on a ledge, no longer in a pit. You look around, and you know it is still 800 feet to where I want to be, but you are not in a hole anymore Now it comes downbut you are not in a hole anymore. Now it comes down to you. (new learning, plasticity?)

Putative Depression Circuit in RevisitedCognitive Processing

attention-memory-action

pCgaCg24b mCg24c

PF9/46 Par40PM6CBT hc

C 24mF9/10 cd-vst thalEmotion-Cognition

Integrationmoodstate

oF11rCg24a

amg mb-sn

Covert

Integrationsalience

self-referencereinforcement

state

sgCg25

hth bstema-insDBSMEDShth bstema ins

Autonomic Responsesarousal-vegetative-circadian-internal milieu

MEDS

CollaboratorsJohns Hopkins Bob Robinson Bob Dannals Baltimore 85-91 Sergio Starkstein Carol Peyser

Susan Folstein Petra Jeffries

UTHSCSA Mario Liotti Art SilvaSan Antonio 91-98 Steve Brannan Jan Tekell

Rick Mahurin Scott McGinnisRick Mahurin Scott McGinnis Peter Fox

Rotman Institute Sidney Kennedy Kim Goldapple y y ppU Toronto 99-05 Zindel Segal David Seminowicz

Stephanie Kruger Randy McIntosh Philippe Fossati Robin WestmacottAndres Lozano Mike BagbyAndres Lozano Mike Bagby Heather McNeely Paul Costa Valerie Voon Michelle Keightley Peter Giacobbe Taresa Stefurak

Grants: NIMH, CIHR, NARSAD, Dana Foundation, Stanley Foundation,PD Foundation, S. Rotman Neuropsychiatry Program, Eli Lilly.