10/3/2016

1

Cognitive Reserve: An Evolving Concept

Yaakov Stern, PhD

Columbia University

Financial Disclosure

I have financial relationships to disclose:

Employee of: Columbia University

Consultant for: AXOXANT, Lilly, Takeda

Stockholder in: none

Research support from: NIH, AXOXANT, CA Walnut Commission, Piramal

Honoraria from: occasional academic talks

10/3/2016

2

What is reserve?

Reserve may explain the disjunction between the degree of brain damage and the clinical manifestation of that damage.

Brain Damage Outcome

Reserve

Mechanisms underlying reserve

• Brain reserve:

– More neurons/synapses to lose

– Brain maintenance: Direct effect of lifestyle/  activities on the brain

• Cognitive Reserve: 

– Resilience/plasticity of cognitive networks in the face of disruption 

10/3/2016

3

Passive, threshold model of reserve

Lesion

Lesion

Bra

in R

eser

ve C

apac

ity

FunctionalImpairment

Cutoff

Patient 1 Patient 2

Satz, Neuropsychology 1993

Brain reserve: association between head circumference and Alzheimer’s disease

Schofield, et al, 1997

10/3/2016

4

Brain reserve is not so simpleThe  literature suggests that exercise and environmental stimulation can activate brain plasticity mechanisms and remodel neuronal circuitry in the brain.

They can increase:• Vascularization (exercise)• Neurogenesis in the dentate• Brain volume/Cortical thickness• Neuronal survival and resistance to brain insult • Brain‐derived neurotrophic factor (BDNF) ‐‐benefits brain plasticity processes 

• Relative lack of brain pathology is the biggest contributor to heterogeneity of cognitive aging

• Various lifestyle factors contribute                to resisting the advent of pathology

• Brain maintenance could account for the current level of brain reserve

10/3/2016

5

PLoS ONE 2008

Lesion

Bra

in R

eser

ve C

apac

ity

FunctionalImpairment

Cutoff

Patient 1 Patient 2

Lesion

Active model of reserve

Stern, JCEN 2002

10/3/2016

6

Incident dementia in the Washington Heights study

Group NIncidentCases

RelativeRisk 95% CI

Low Education 264 69 2.02 1.3-3.1

High Education 318 37 1

Low Occupation 327 71 2.25 1.3-3.8

High Occupation 201 17 1

.

Stern et al, JAMA 1994

Valenzuela & Sachdev, Psychological Medicine, 2005

10/3/2016

7

Literacy and memory decline in non‐demented elders

Manly et al, JCEN 2003

VLS WHICAP

N 1023Mean 14.1SD 3.1

N 3443Mean 9.8SD 4.8

10/3/2016

8

Association of education with cognitive decline in the Washington Heights study 

Model‐estimated cognitive trajectories for 76‐year‐old, White, non‐Hispanic Males born 1900‐1909, recruited in 1992, with low (0‐8 years) or high (9‐20 years) education

Zahodne et al, in press

AD Neuropathology

High Reserve (Education)

Low Reserve (Education)

Score at initial visit

10/3/2016

9

Time

6543210-1

SR

T T

otal

Rec

all

20

15

10

5

Education Group

Low Predicted

High Predicted

Low Actual

High Actual

More rapid memory decline in AD patients with higher educational attainment

Stern et al Neurology 1999;53:1942-1957

Hall, C. B. et al. Neurology 2007;69:1657-1664

Blue indicates less than 7 years education (32 Ss), red indicates 8 to 11 years (64 Ss), and green indicates 12 or more years education (21 Ss).

Bronx Aging Study

10/3/2016

10

Father’s occupation

Cognitionat 8 years

Education by 26 years

Own occupationat 43 years

NART at 53

0.380.25

0.50

0.11

0.13

0.36

0.45

0.24

0.20

Richards, JCEN 2003

• Different aspects of life experience contribute independently to CR

• CR is a formative latent variable, not reflective– Important consideration 

for summary measures of reserve

• This observation also provides hope that even experiences later in life might contribute to CR

Low reserve

High reserve

mild

AD

no

rma

lM

CI Diagnostic Threshold

ModerateMild

AD Pathology

Clin

ical

Sev

erity

Reserve, AD pathology, and clinical diagnosis

Stern, JINS 2002

10/3/2016

11

Stern et al, Ann Neurol 1992

Controlling for clinical disease severity, there is an inverse relationship between education and a functional imaging proxy for AD pathology

Education and rCBF

22 years

15 years

18 years

Education * AD path = 0.088, p<.01

Bennett DA et al, Neurology 2003

Interaction of AD pathology and education

Summary Measure of AD Pathology

Glo

bal C

ogni

tive

Fun

ctio

n

10/3/2016

12

Current  study of the  neural implementation of cognitive reserve

Task or NP performance,

Clinical Outcome

VolumeCortical ThicknessWMH BurdenWM Tract integrityResting CBFResting connectivityAmyloid burden

Activation Task PerformanceCognitive DomainsFunction / ADLCognitive decline over time/Incident MCI/AD

Measured IQEducation/LiteracyLeisure ActivityCR Network

Age-or AD-related

pathology

Task-relatednetwork

expression

Measured CR orCR-specific

network

BrainMaintenance

CognitiveReserve

The role of education and verbal abilities in altering the effect of age‐related gray matter differences on cognition

Age Group

Volume

GF

Life experience

Steffener et al, PLoS ONE 2014

10/3/2016

13

Areas with significant mediated moderation 

Age Group

Volume

GF

LE

Steffener et al, PLoS ONE 2014

Mediated moderation sample result

Steffener et al, PLoS ONE 2014

10/3/2016

14

Using Functional Imaging to Study CR• Goal:  To understand how cognitive reserve may be neurally implemented.  – Emphasis on networks mediating CR, not task performance

• Working hypothesis: CR operates through individual differences in how tasks are processed in the brain. 

• Basic approach:  Challenge participants with a demanding task and investigate differences in task‐related activation between individuals with high and low CR. 

• Assumption: Because CR modulates most aspects of cognitive performance in the presence of pathology, this approach should work with most demanding tasks.

Modified Sternberg Task

800

900

1000

1100

1200

1300

1400

1500

1600

1700

1 3 6

set size

RT

(m

s)

Y

E

”Load-related” activation: the change in activation as set size increasesWe focus on load-related activation because CR might be more related to the coping with increases in task demand than to task-specific features.

10/3/2016

15

Load‐dependent Activation During Retention: Neural Reserve and Neural Compensation

• Two patterns were expressed during retention

• The primary network was expressed by both groups, but more efficient in the young

• The second network was expressed only by the elders– Higher expression was 

associated with poorer performance

– Elders with greater volume loss in the primary network were more likely express it

– Thus this is a compensatory pattern

Zarahn et al., Neurobiol Aging 2007Steffener at al., Brain Imaging and Behavior 2009

Elder Young

Compensatory Network

Gre

ater

Net

wor

k E

xpre

ssio

n

Less Efficient Processing (RT slope)

Primary Network

Steffener and Stern, Biochimica et Biophysica Acta 2012

10/3/2016

16

A generalized “task‐invariant” neural representation of  CR • CR allows people to better maintain function in 

multiple activities and cognitive domains in the face of brain pathology.  

• If a particular brain network subserves CR, it should be active across tasks with varying processing demands.  

• Goal: Can we identify a pattern of CR‐related brain activity :– that is common to 12 different tasks– whose expression in other tasks correlates a CR proxy– whose expression moderates the relationship between 

cortical thickness and task performance

Deriving a task‐invariant CR network• 255 subjects from RANN study, age 20‐80, with complete neuroimaging 

for 12 different tasks

• Randomly divide data into training sample of 200 observations and test sample of 41 observations 

• In derivation sample, use scaled Subprofile modeling (SSM) to derive best‐fit NART patterns

• Project derived pattern into test sample and estimate NART in all 12 tasks

• Repeat steps 500 times, each time storing the derived patterns and the test prediction quality

• Compute weighted Z‐map of pattern loadings for the 500 patterns

10/3/2016

17

-1200 -1000 -800 -600 -400 -200 0 200 40090

95

100

105

110

115

120

125

130

135

Pattern expression in ECF task

NA

RT

IQ

Single

Dual

Expression of the task‐invariant CR network in a different fMRI activation task (in different people) correlates with NART

-2 -1.5 -1 -0.5 0 0.50.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Subject scores of structural pattern

VO

CA

B a

ccur

acy

p<0.01

High NARTIQ-pattern

Low NARTIQ-pattern

Expression of the task‐invariant CR network moderates between cortical thickness and task performance

10/3/2016

18

Clinical Implications: Cognitive reserve, aging and AD

• Two individuals who appear the same clinically, whether demented of non‐demented,  can have widely divergent levels of underlying age‐related neural changes or AD pathology.

• Thus, the clinical diagnosis of normal aging, MCI or AD may be accompanied by very minimal pathology or more than enough to meet pathological criteria for AD.

• Measuring CR therefore becomes an important component of diagnosing and characterizing aging and dementia.

Clinical implications: Cognitive reserve, aging and AD

• Optimal clinical evaluation of age‐related cognitive change or AD should include:– A measure of pathology

• age‐related atrophy, amyloid imaging

– A measure of an individual’s CR, that is, the ability the ability to cope with this pathology:  

• Proxies for CR such as education or IQ• fMRI measured expression of “CR networks”

• This type of evaluation is important for – early diagnosis and characterization– prognosis– measuring progression over time– assessing of the effect of interventions

10/3/2016

19

How would reserve‐based interventions work?

Aging/AD Pathology Clinical Disease

BrainReserve

CognitiveReserve

?

Eight out of 11 studies reported that aerobic exercise interventions resulted in … improvements in cognitive capacity.

The largest effects on cognitive function were found on motor function and auditory attention (effect sizes of 1.17 and 0.50 respectively).

Moderate effects were observed for cognitive speed (effect size 0.26) and visual attention (effect size 0.26).

Cochrane Review: Physical activity and enhanced fitness to improve cognitive function in older people without known 

cognitive impairment

10/3/2016

20

Scarmeas et al. JAMA 2009

Physical Activity, Diet and Risk of Alzheimer’s Disease

Problems with cognitive interventions in aging

• Small effect size

• Poor generalization to other cognitive domains (far vs near transfer)

• Poor generalization day‐to‐day functions or IADLs

• Questionable sustainability of effects

• Relation to rate of aging or dementia onset  not established 

10/3/2016

21

Space fortress intervention study• Emphasis Change Training:

– Subjects perform the whole task during training, but are required to systematically change their emphasis on major sub‐components of the task.

– Encourages subjects to explore the response alternative space

– Promotes executive control

– Is associated with improved transfer of training

• Our study had 3 groups: 12 weeks of game play with and without emphasis change training, and a no gameplay control groupThe Space Fortress

Game

.0

.3

.5

.8

1.0

1.3

1.5

1.8

2.0

Week 1 Week 2 Week 3 Week 4 Week 5 Week 6 Week 7 Week 8 Week 9 Week10

Week11

Week12

EC

AC

Fortress Destruction

Space Fortress: Fortress Destruction

Blumen at al, Frontiers in Aging Neuroscience,2010

10/3/2016

22

8

9

9

10

10

11

11

12

12

Baseline Post

EC AC

PC

Time of Test

Space Fortress: Cognitive Outcome

Stern et al, Aging, Neuropsychology and Cognition, 2012

Scr

een

ing

1st

Bas

elin

e vi

sít

2nd

Bas

elin

e vi

sit

RA

ND

OM

IZA

TIO

N

INTENSIVE INTERVENTION

REGULAR HEALTH ADVICE

INT

ER

VE

NT

ION

KIC

K-O

FF

MINI-INTERVENTION

3 6 9 12 15 18 21 24

NUTRITION:7 group sessions,

3 individual sessions

COGNITIVE TRAINING:9 group sessions

Independent training

EXERCISE:1-2x/wk muscle2-4x/wk aerobic

EXERCISE:2x/wk muscle

4-5x/vk aerobic

EXERCISE:2x/wk muscle strength training

5-6x/wk aerobic training

MONITORING AND MANAGEMENT OF METABOLIC AND VASCULAR RISK FACTORS

Nurse: Visit every 3 months, Physician: 3 additional visits

months

INTERVENTION SCHEDULE

COGNITIVE TRAINING:2 group sessios

Independent training

Kivipelto et al., Alzheimer & Dementia 2013

10/3/2016

23

What will the large‐scale project to enhance CR look like?

• Healthy elderly population

• Intensive, extensive, combined interventions

• Long‐term follow‐up

• Outcomes:

– Rapidity of cognitive/functional decline

– Incident dementia

Conclusions

• Epidemiologic and imaging evidence support the concept of reserve

• Reserve is malleable: it is influenced by aspects of experience in every stage of life

• Two forms of reserve:– Brain reserve: passive, supported by brain maintenance– Cognitive reserve: active

• The concept of cognitive reserve is applicable to a wide range of conditions that impact on brain function at all ages

• Imaging studies can help clarify the neural implementation of reserve 

• Influencing reserve may delay or reverse the effects of aging or brain pathology