SCREENING TOOLS FOR MCI (mild cognitive impairment) J. Wesson Ashford, M.D., Ph.D. Clinical Professor (affiliated) Department of Psychiatry & Behavioral Sciences Stanford University Senior Research Scientist Stanford / VA Aging Clinical Research Stanford University and VA Palo Alto Health Care System International Conference on Alzheimer’s disease July 12, 2010 Slides at: www.medafile.com (Dr. Ashford’s lectures)
84
Embed
SCREENING TOOLS FOR MCI (mild cognitive impairment) J. Wesson Ashford, M.D., Ph.D. Clinical Professor (affiliated) Department of Psychiatry & Behavioral.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
SCREENING TOOLS FOR MCI(mild cognitive impairment)
J. Wesson Ashford, M.D., Ph.D.
Clinical Professor (affiliated) Department of Psychiatry & Behavioral Sciences
Stanford University
Senior Research ScientistStanford / VA Aging Clinical Research
Stanford University and VA Palo Alto Health Care System
Factors for Deciding whethera Screening Test is Cost-Effective
1) Benefit of a true positive screen
2) Benefit of a true negative screen
3) Cost of a false positive screen
4) Cost of a false negative screen
5) Incidence of the disease (in population)
6) Test sensitivity (in population)
7) Test specificity (in population)
8) Test cost
$W = Cost–Worthiness Calculation $W > ($B x I x Se) – ($C x (1-I) x (1-Sp)) - $T
• BENEFIT– $B = benefit of a true positive diagnosis
• Earlier diagnosis may mean proportionally greater savings• Estimate: (100 years – age ) x $1000• Save up to $50,000 (e.g., nursing home cost for 1 year)
– (after treatment cost deduction at age 50, none at age 100)– (cost-savings may vary according to your locale)
– True negative = real peace of mind (no money)
• COST– $C = cost of a false positive diagnosis
• $500 for further evaluation– (time, stress of suspecting dementia)
– False negative = false peace of mind (no price)
• I = incidence (new occurrences each year, by age)• Se = sensitivity of test = True positive / I• Sp = specificity of test = True negative / (1-I) = (1-False positive/(1-I)• $T = cost of test, time to take (Subject, Tester)
Kraemer, Evaluating Medical Tests, Sage, 1992
Benefits of Early Alzheimer Diagnosis:Social
• Undiagnosed AD patients face avoidable problems – Social, financial
• Early education of caregivers– How to handle patient (choices, getting started)
• Advance planning while patient is competent– Will, proxy, power of attorney, advance directives
• Reduce family stress and misunderstanding– Caregiver burden, blame, denial
• Promote safety– Driving, compliance, cooking, etc.
• Patient’s and family’s right to know– Especially about genetic risks
• Promote advocacy– For research and treatment development
Benefits of Early Alzheimer Diagnosis
Medical
• Early diagnosis and treatment and appropriate intervention may:– improve overall course substantially
– lessen disease burden on caregivers / society
• Specific treatments now available for dementia (anti-cholinesterases, memantine)
– Improve cognition
– Improve function (ADLs)
– Delay conversion from Mild Cognitive Impairment to AD
– Slow underlying disease process, the sooner the better
– Decreased development of behavior problems
– Delay nursing home placement, possibly over 20 months
– Delay nursing home placement longer if started earlier
Benefits of Early Treatment ofAlzheimer’s Disease
• Neurophysiological pathways in patients with AD are still viable and are a target for treatment
• Opportunity to reduce from a higher level: – Functional decline– Cognitive decline– Caregiver burden
Need to estimate net benefit monetarily
(key factor in determining case for screening)Estimate benefit = (100 years – age ) x $1000
Estimated Age-related Benefitof Early Alzheimer Treatment
0
10000
20000
30000
40000
50000
50 60 70 80 90 100
AGE
Dol
lar
savi
ngs
from
de
laye
d nu
rsin
g ho
me
plac
emen
t
Value of Diagnosis versus Time-Index
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
-10 -8 -6 -4 -2 0 2 4 6 8 10
Estimated years into illness(TimeIndex Scale)
Re
lati
ve
va
lue
of
de
tec
tio
n
Value across continuum
Value at transition
Value early
Cost of False-Positive Screen
• Referral of normal individual for further testing– (more specific testing)
• Value of individual’s time
• Cost of additional testing
• Estimate cost = $500 per false-positive screen
• This does not and should not include the cost of untoward results of misdiagnosis, medication side-effects, or malpractice – quality management should address these
Other Benefits and Costs of Screening
• Benefit of true-positive screen = intangible– Peace of mind – Plan further into future
• Cost of false-negative screen = wash– Delay in diagnosis and treatment– No different from current condition
INCIDENCE OF MCI(Hazard per year)
Based on estimate of 4 million AD patients with dementia in US in 2000, with an incidence that doubles every 5 years, illness duration of 8 years.
Assume average of 5 years from onset of MCI to onset of dementia
U.S. mortality, dementia, MCI rate by age (mortality = 2000 CDC / 2000 census)
0.0001
0.0010
0.0100
0.1000
1.0000
0 10 20 30 40 50 60 70 80 90 100
Age (years)
Haz
ard
/ y
ear
Males, 2t = 8.2yrs
Females, 2t = 7.5 yrs
dementia incidence, 2t = 5 yrs
MCI incidence, 2t = 5yrs
JW Ashford, MD PhD, 2003; See: Raber et al., 2004 (Incidence for a to a + 1 year)
The Gompertz survival curve explains 99.7% of male and female mortalityVariance between 30 and 95 y/o:
U(t) = Ro * exp (alpha * t)
Dementia rate, assume Td = 5 yrs
0.0001
0.001
0.01
0.1
1
10
100
1000
50 60 70 80 90 100
Age (years)
Haz
ard
/ y
ear
mean rate
APOE 4/4 (x7.5)
APOE 3/4 (x2)
APOE 3/3 (x0.6)
Early onset (x200)
Using the Gompertz equationto model rate of dementiaincrease with age:
U(t) = Ro * exp (alpha * t)
MCI rate, assume Td = 5 yrs
0.0001
0.001
0.01
0.1
1
10
100
1000
50 60 70 80 90 100
Age (years)
Ha
zard
/ y
ea
r
mean rateAPOE 4/4APOE 3/4APOE 3/3Early Onset
Probability of Dementia Onset
0
0.01
0.02
0.03
0.04
0.05
0.06
50 60 70 80 90 100
Age
pro
bab
ility
/ yr
mean rate
APOE 4/4
APOE 3/4
APOE 3/3
Using Gompertz equationsto model probability of dementia with age:
D(t) = U(t) * S(t)
Probability of MCI Onset
0
0.01
0.02
0.03
0.04
0.05
0.06
50 60 70 80 90 100
Age
pro
bab
ility
/ yr
mean rate
APOE 4/4
APOE 3/4
APOE 3/3
Probability of Dementia Onset
0
0.01
0.02
0.03
0.04
50 60 70 80 90 100
Age (single mortality correction)
pro
b/
yr
* l
ive p
op
ula
tio
n APOE 4/4-MAPOE 4/4-FAPOE 3/4-MAPOE 3/4-FAPOE 3/3-MAPOE 3/3-F
Critical Factors for Developing Cost-Effective Screening
1) Develop benefit of a true positive screen- Need effective disease slowing treatments
2) Define value of genetic testing- Need to recognize central role of APOE genotype
3) Determine sensitivity and specificity of tests- Parameters must apply to population
4) Need to determine cost-worthiness- This must be determined for each test
5) Specific tests must be optimally sequenced- Frequent cognitive screens triggering biomarker tests
Need to Develop Better Screening and Early Assessment Tools
• Trait factors – determine at 50 y/o to plan screening– Genetic vulnerability testing (or family history)– Vulnerability factors (education, occupation, head injury, blood pressure)
• State factors (begin annually at appropriate age)– Early recognition (10 early warning signs), ADLs– Screening tools (6th vital sign in elderly)
• Brief clinical screens vs. computerized tests• Tests need to assess likely level of function
– Detecting early change over time• Measuring rate, predicting progression
• Positive diagnostic tests– CSF – amyloid levels low (early), tau levels elevated (MCI)– Brain scan – PET – f-DG, f-DDNP, f-amyloid ligands (early)– Dementia severity tested on “time-index” continuum
Alzheimer's Disease Top 10 Warning Signs (not early)
1. Recent memory changes affecting daily life2. Challenges in problem solving and planning3. Difficulty performing familiar tasks4. Disorientation to time and/or place5. Difficulty understanding visual images and/or spatial
relationships6. Problems with spoken and written language (eg,
paraphasia, agraphia)7. Misplacing things8. Poor judgment9. Withdrawal from activities (eg, social, work)10.Changes in personality and/or mood
Alzheimer's Association. 10 Signs of Alzheimer's. Available at: http://www.alz.org/alzheimers_disease_10_signs_of_alzheimers.asp. Accessed April 20, 2009.
Need a Top 10 Early Warning Signs
Challenges With the Mini-Mental State Examination• Mini-Mental State Exam (MMSE)
– Folstein MF, et al. J Psychiatr Res. 1975;12:189-198.
• Several items do not provide adequate information
• Adds noise rather than discrimination between demented and nondemented individuals, particularly in early AD, MCI
• Poor range for measuring change– Large standard error of measurement
• Poor power for assessing medication benefit
• Inadequate screening tool
• Too long– Better, shorter tests are available
• Copyright is being enforced (test is not free)Ashford JW. Aging Health. 2008;4:399-432.
Spearman Correlations Between Neuropsychological and MRI Volumetric Data
Grey Mat. White Mat. Right Hipp. Left Hippo. Right Ento Left Ento
MIS
Controls 0.18 0.112 0.185 0.243 –0.085 –0.205
MCI –0.022 –0.213 0.430a 0.378 0.156 0.21
AD –0.100 0.033 0.192 0.23 –0.012 –0.061
FCSRT learning
Controls 0.25 0.249 0.048 0.252 –0.214 –0.152
MCI –0.044 –0.243 0.469a 0.383 0.374a 0.424a
AD –0.032 –0.224 –0.091 0.211 –0.074 –0.168
FCSRT delayed
Controls 0.161 0.136 0.028 0.233 –0.325 –0.295
MCI –0.010 –0.267 0.554b 0.424a 0.426a 0.407a
AD –0.205 –0.126 0.286 0.451a 0.104 0.081
Abbreviations: AD, Alzheimer Disease; ento, entorhinal; hipp., hippocampus; mat., matterFCSRT, Free and Cued Selective Reminding Test; MIS, Memory Impairment Screen;a Significant correlations are flagged with P < .05.b Significant correlations are flagged with P < .001.
The MCIS For Clinical Practice & Research
Takes 10 Minutes
Accuracy1-4 is:
96-97% for Normal vs. Mild Cognitive Impairment.
99% for Normal vs. Mild Dementia.
Improves Signal:Noise Ratio by 100% over standard scoring methods5.
16 culturally unbiased, equivalent wordlists randomly selected without replacement in each patient to minimize test-retest effects5.
Available in English, Spanish and Japanese.
Adopted in all Medicare regions.
1Shankle et al. PNAS. 20052Trenkle et al. J. Alz. Dis. 2007.3Cho et al. Jap. J. Exp. Med. 2007.4Tabara et al. Hypertension Research. 2009.5Shankle et al. Alz. & Dementia, 2009.
Classification algorithm &Memory Performance Index
(MPI) scaling
1This method explains the maximum possible amount of the raw data’s variance for the class of linear methods.
In contrast to FA & PCA, Correspondence analysis accounts for differences due to heterogeneous samples.
Optimal Scores Vary By: List Position Exposure Frequency Delay Being Recalled or Not
Item Responses Are Usually Scored As 0 or 1: All Items Have Equal Value
Wordlist Memory Task: 4 Trials
Word 1Word 2 Word 3 Word 4 Word 5 Word 6Word 7 Word 8 Word 9Word
10
Wordlist Development1 million common nouns.Frequency, range, and diversity of usage statistics paralleled CERAD and ADAS-Cog Wordlists
600 nouns met these criteria
Constructed 10-word lists that met the following requirements Each word: could be used only once.could only have 1 or 2 syllableshas unique letter or sound.has no homonyms or antonyms in list.has low associability with all other list words.Each target list word can be matched on all above criteria with a word in
its accompanying distracter list.
16 Wordlists Met All Above Criteria(Subjects Must Be Tested 9 Times Before They See The Same Wordlist
Validity compared to Clinical Diagnosis (Kappa statistic)2 0.92 ± 0.09
False Negative Rate Based on Long-Term Care Claims After 3 years exposure: N=250,0004 0.008-0.095%
*The underlying etiologies of the MCI syndrome in the primary care, community and academic samples included Alzheimer’s disease, Lewy Body disease, Parkinson’s disease, Frontal Temporal Lobe dementia, normal pressure hydrocephalus, cerebrovascular disease, alcohol dependence, traumatic brain injury, metabolic disorders, and depressive pseudo-dementia.
1Shankle et al. PNAS: 2005. 2Trenkle et al. J. Alz Dis: 2007. 3Cho et al. Am J. Alz Dis Other Dem. 2008. 4Cohen et al. National Underwriter, 2009.
Psychometric Properties
Japanese MCIS vs. Biomarkers
Cho et al., 2009
Comparing Standard Recall & MCIS Scoring Method (MPI)
Delayed, Immediate or Total Free RecallR2 = 23.4-26.9% of variance explained
MPI ScoreR2 = 55.5% of variance explained
Regression of Recall Scores or MCIS Scoring Method (MPI) Score AgainstAge, Gender, Education, Race, Method of Administration & Wordlist Used1
N=121,481 Applicants for Long-Term Care Insurance: Ages 20-100
1Shankle et al. Alz. And Dementia. 5; 2009: 295-306.
Effect sizes (Cohen’s d) were as follows: Effect of Race, gender, and wordlist on MPI Score were negligible (<0.02) Effect of Education & phone vs. in-person testing on MPI Score were small
(0.02-0.05) Effect of Age on MPI Score was large (0.68) Effect of all covariates on Free Recall scores was negligible or small (<
0.09)
Time to Administer Available Short Screening Tests
Top cognitive tests studied for BRIEF SCREENING for MCI
• Brief Alzheimer Screen 2 – 3 min• Mini-cog + FAQ 5 - 8 min• MIS + Isaacs Set Test 4 – 6 min• MCIS 10 min
A suitably accurate cognitive test for MCI is not available.
Because on variability between individuals, MCI screening requires longitudinal assessment!!
Need to Develop More Sensitive and Specific Tools for MCI
• Genetic vulnerability testing (trait risk)– APOE genotype + age is among the best currently
• Improve awareness of vulnerability factors, ask the “right questions” of the patient or informant (education, occupation, head injury)
• Early recognition “10 warning signs”– Activities of daily living (ADLs), behavior changes, forgetting
• Increase suspicion and use available screening tools (while new and better tools/tests are developed)– "6th vital sign" in elderly
• Utilize current diagnostic tests that can best identify probable AD– Cerebrospinal fluid: tau levels, amyloid levels
– Brain scan, PET scan: f-2DG, f-DDNP, f-amyloid-ligands
• More routine use of mild dementia severity assessments
• Detect early change over time– Measure rate of change, predict progression
Memory / MCI / Dementia Screening Test
• Need test for cognitive screening of patients for early Alzheimer’s disease
• Test needs to be on multiple platforms– Doctor’s offices– Best if computerized for rapid, objective assessment– Internet-based testing – CD-ROM distribution– Kiosk administration (eg, drug stores, shopping malls)
• Test needs to be very brief (~1-minute)• Multiple test-forms needed so it can be repeated often (quarterly)• Annual screening annually after age 50 years
– Repeated every 3 months for individuals over 65 years or with concerns/risk factors
– Variety of versions allow daily testing as an exercise
• Any change over time needs to be detected• The test should be free (or cost very little)
Test to screen patients for dementia, AD: Subjects are asked to respond to images that are repetitions of previously shown images.– Computerized test (computer or web - 3 minutes)– KIOSK administration (clinic check-in)– Group administration (Power-Point – 6 minutes)
• On the paper & pencil version, each slide is shown for 5 seconds. The test-taker is ask to fill in the circle next to the number for a repeated slide. After a practice set, the 50-slide test takes 4 minutes and 10 seconds.
• For the computerized test, each image is shown for 3 seconds, and the subject pushes the space bare to indicate recognition of a repeated picture.
• Estimate level (based on 2,000 patients, caregivers)– >90% very good– 80-90% good– 70-80% consider mild cognitive impairment– <70% dementia
MEMTRAX Memory Test
116 subjects – mostly elderly normals, some young, some dementia patientsFalse positive errors (false recognition) – 33(64);6(58);47(27)—4,18,23,34(1);1,2,8(0)
- mean – 8.3% (sd-14.5%) errors per itemFalse negative errors (failure to recognize) – 35(33);27(20);5(16)—32(4);24(3);45(3)
- second presentation (#15): mean- 10.5% (sd-6.2%) errors per item- third presentation (#10) mean – 5.7% (sd-2.5%) errors per item- second 10 vs. same third 10: 10.5% (sd-3.4%) vs 6.6% (sd-2.5%)
Performance in 116 subjects
0
5
10
15
20
25
0 5 10 15 20 25
Number False negative
Num
ber
Fals
e po
siti
ve
Probable Normal
? fronto-temporaldementia
? MCI
? dementia
RandomPerformance
Regression
True Negative Performance
y = -0.0352x + 25.564
R2 = 0.039
y = -0.0597x + 27.24
R2 = 0.141
1213141516171819202122232425
40.0 50.0 60.0 70.0 80.0 90.0 100.0
Age (years)
Nu
mb
er
Co
rre
ct
Male true-
Female true-
Linear (Male true-)
Linear (Female true-)
True Positive Performance
y = -0.0438x + 27.029
R2 = 0.0617
y = -0.0418x + 26.746
R2 = 0.0605
1213141516171819202122232425
40.0 50.0 60.0 70.0 80.0 90.0 100.0
Age (years)
Nu
mb
er
Co
rre
ct
Male true+
Female true+
Linear (Male true+)
Linear (Female true+)
CONCLUSIONS on MEMTRAX
• A short, computerized test provides a measure of cognitive function, including memory and attention, on a robust continuum, establishing a baseline of cognitive function and potentially predicting the presence of dementia– Computerized version – 2-3 minutes, fun game, provides reaction
time measure– Paper&Pencil, with PowerPoint slide show, can be given to a
large audience
• Testing for reliability and validity are Classical Test Theory concepts– Modern Test Theory examines performance across individual
items on a continuum • (varied by first repeat vs second repeat, number of slides
between first show and first repeat, etc.– Analysis for maximum likelihood level of cognition (both
recognition and attention), provides information about dementia probability
– Information about visuo-spatial and language function is available
MEMTRAX - Memory Test(to detect AD onset)
• New test to screen patients for AD: – World-Wide Web – based testing– CD-distribution– KIOSK administration (grocery stores, drug stores)
• Determine level of ability / impairment• Test takes about 1-minute• Test can be repeated often (e.g., weekly, quarterly)• Any change over time can be detected• Experimental tests at: www.medafile.com• Social network tests at: www.memtrax.net
Comprehensive Screening Plan• At age 50 years: initial screen, review risks
– Review dementia family history – strongly consider APOE genotyping– Review of systems, vital signs– Brief cognitive evaluation – establish baseline for longitudinal assessment– Complete blood count (CBC), B12, cholesterol– Begin yearly assessments if high risk
• At age 55–60 years: follow-up assessments– Review of systems, vital signs– Brief cognitive evaluation using longitudinal measures!!– CBC, B12, cholesterol
• At age 65 years and older: begin annual assessments– Review of systems, vital signs– Brief cognitive evaluation watching longitudinal changes– CBC, B12, cholesterol
Secondary Screen:Specific Testing
• More cognitive testing
• Complete orientation testing
• Test ability to name animals and vegetables in 1 minute
• Ask for recall of 10 items after distraction
• Test praxis
• Draw clock, cube
• Talk with a knowledgeable informant
• Ask questions about activities of daily living
• Ask questions about depression, sleep
Potential AD BiomarkersPotential AD BiomarkersProbably not cost-worthy as screening tests,Probably not cost-worthy as screening tests,but may be useful for secondary screeningbut may be useful for secondary screening
• Blood, urine ABlood, urine Aββ40? A40? Aββ42? Neuritic threads?42? Neuritic threads?– Most studies suggest not helpful, may be wrongMost studies suggest not helpful, may be wrong
• Protein levels in blood – Leptin, ProteomicsProtein levels in blood – Leptin, Proteomics– Lower Leptin predicts MCI progression to dementiaLower Leptin predicts MCI progression to dementia
– Structural (volumetric assessments)Structural (volumetric assessments)– Functional (FDG-PET, SPECT)Functional (FDG-PET, SPECT)– Specific protein imaging (PET)Specific protein imaging (PET)
Serum Leptin levels and cognition in the elderly
78
MiId
Mod
erat
e
Nor
mal
Sev
ere
10
20
Lept
in (
ng/m
l)
Data: Satoris, Inc.
AD
In elderly, higher serum leptin appears to protect against cognitive decline (5 yr prospective study, 2,871 elders, Holden et al., 2009)
Patients with AD have lower serum leptin levels compared to controls, independent of BMI (Power et al., 2001)
In elderly, higher serum leptin appears to protect against cognitive decline (5 yr prospective study, 2,871 elders, Holden et al., 2009)
Patients with AD have lower serum leptin levels compared to controls, independent of BMI (Power et al., 2001)
Correlation networks of Alzheimer disease (AD) signature proteins in plasma of controls without dementia and patients with AD.
PROTEOMICS:Expression patterns of Alzheimer disease (AD) signature proteins discriminate between plasma samples from patients with AD and controls.Britshgi & Wyss-Coray, 2009
0
100
200
300
400
500
600
700
AD Patients Control Patients
CSF in Alzheimer’s Disease, both MCI and Dementia patients:
Low Aβ and High Tau
Aβ Tau
Co
nce
ntr
atio
n (
pg
/mL
)
Sunderland T, et al. JAMA. 2003;289:2094-2103.
ADNI data, 2008
ADNI Data – CSF ABeta, total tau
Comparison p-value
33 vs 34 <.0001
33 vs 44 <.0001
34 vs 44 0.08
Normal vs MCI 0.57
Normal vs Mild AD 0.15
MCI vs Mild AD 0.20
Comparison p-value
33 vs 34 0.07
33 vs 44 0.67
34 vs 44 0.99
Normal vs MCI 0.05
Normal vs Mild AD <.01
MCI vs Mild AD 0.06
ADNI CSF Data – total tauNumber of participants that provided CSF at baseline Ages +std of participants that provided CSF at baseline
CSF tau levels ± std
APOE genotype
Normal MCI Mild AD
33 67 (72%) 82 (44%) 29 (31%)
34 24 (26%) 81 (44%) 42 (45%)
44 2 (2%) 22 (12%) 22 (24%)
APOE genotype
Normal MCI Mild AD
33 75.8 ± 5.0 75.4 ± 8.4 76.3 ± 8.6
34 75.8 ± 6.0 73.9 ± 6.7 75.6 ± 6.6
44 77.0 ± 1.4 72.2 ± 6.0 69.8 ± 7.0
APOE genotype
Normal MCI Mild AD
3367.8 ± 26.9
83.6 ± 40.8 123.8 ± 68.6
3481.8 ± 42.6
122.4 ± 72.7 113.3 ± 42.0
44 71.0 ± 2.8 110.6 ± 45.9 128.9 ± 53.1
APOE genotype
Normal MCI Mild AD
33 212.4 ± 48.4 189.1 ± 59.8 168.8 ± 52.3
34 156.0 ± 47.8 148.4 ± 42.4 139.0 ± 27.2
44 126.0 ± 2.8 119.8 ± 23.5 116.2 ± 22.3
CSF ABeta levels ± std
Future directions for MCI screening
• Successful treatments for MCI
• APOE genotyping – routine at 50 y/o
• Preventive measures based on genetics
• Longitudinal assessment of memory
• Computer games to monitor cognition– quick, fun, inexpensive
• Can beta-amloid deposition be controlled by mental, physical exercises, better sleep?