EARLY TESTING FOR ALZHEIMERS

1
EARLY TESTINGFOR ALZHEIMERS

• Session 32 February 28, 2006
• Session Producer
• Jennifer Vey, Asst Vice-President
• Underwriting Clinical Services
• LifePlans, Inc.

2
PANELISTS

• Robert Pokorski, MD, Vice-President, Worldwide
  Medical Research and Development, Gen Re
  LifeHealth
• Philip Barackman, FSA, MAAA, FLMI,
  Vice-President, Profit Center Head for Long Term
  Care Reinsurance, Gen Re LifeHealth

3
(No Transcript)
4
(No Transcript)
5
Early Testing for Alzheimers Disease

• Genetics / APOE
• REVEAL Study
• Underwriting

6
Introduction

• Alois Alzheimer reported Alzheimers disease (AD)
  in 50-year-old woman (1907)
• AD causes 2/3 of dementia
• 7.1 of all U.S. deaths due to AD in 1995
• Roughly same as cerebrovascular disease (3rd
  cause of death)
• 2 types
• Early-onset (familial)- 2 of AD
• Late-onset (nonfamilial, sporadic, APOE-related)-
  98 of AD

7
Early-onset (Familial)

• Autosomal dominant ? ½ of children affected
• Age at onset
• 40s, 50s, sometimes even 20s
• Never after 65
• Caused by 3 genes that disrupt amyloid brain
  metabolism
• APP (amyloid precursor protein)
• Presenilin 1
• Presenilin 2

8
Late-onset (nonfamilial, sporadic, APOE-related)

• The gene APOE produces the lipoprotein
  apolipoprotein E (APOE)
• APOE primary component of very low density
  lipoproteins (VLDL) which carry cholesterol from
  blood to liver
• APOE important for cholesterol transport within
  brain

9
Apolipoprotein E (APOE) genotypes

• Genotype- Three alleles (versions) of APOE gene
• e2- Lower AD risk
• e3- Average AD risk
• e4- Higher AD risk
• APOE genotyping not routine, but many patients
  request it (299)

10
APOE affects age at onset

• Age at which 50 have AD
• e4/e4- Age 68
• e3e/4- Age 76
• e3/e3- Age 84
• By age 80, percent with AD
• e4/e4- 91
• e3/e4- 48
• No e4- 20
• e2/e2 overly represented among centenarians
• No e4/4 has reached 90 without AD

11
Causes of Alzheimers disease
Nongenetic (possible)
Genetic (definite)
Toxins
APP
Most common pathway Environment genes (APOE,
others) aging
Viruses
Presenilin 1
Prions
Presenilin 2
Low education
Downs syndrome
Head trauma
Other genes
12
Pathophysiology
Gene mutation ? Increased brain production of
amyloid ß peptide and/or defective amyloid ß
peptide
Amyloid ß peptide accumulates as plaques that
disrupt neurotransmission between neurons
Neurons malfunction / die
Dementia
13
Life expectancy after initial clinicaldiagnosis
decreases with older age
14
Treatment not good, will improve

• May slow progression for 6-12 months (some better
  than others)
• Razadyne (galantamine)- Anticholinergic
• Exelon (rivastigmine)- Anticholinergic
• Aricept (donepezil)- Anticholinergic
• Cognex (tacrine)- Anticholinergic
• Namenda (memantine)- Glutamate regulator
• Evista (raloxifene)- Selective estrogen receptor
  blocker

• Ineffective
• Vitamin E (NEJM 20053522379-88)
• Statins (Arch Neurol 2005621047-51)
• Experimental (lab mice)
• Amyloid ß peptide vaccine for prevention /
  treatment of existing AD

15
Early Testing for Alzheimers Disease

• Genetics / APOE
• REVEAL study
• Underwriting

16
REVEAL first study to show that genetic
information changes how people buy insurance

• Genetic Testing For Alzheimers Disease And Its
  Impact On Insurance Purchasing Behavior
• Widespread genetic testing for Alzheimers
  susceptibility could present dilemmas for
  long-term care insurance
• by Cathleen D. Zick, Charles J. Mathews, J. Scott
  Roberts, Robert Cook-Deegan, Robert J. Pokorski,
  and Robert C. Green
• Health Affairs 2005 (March)24483-90

17
REVEAL (Risk Evaluation and Education for
Alzheimers Disease) Study

• Randomized controlled trial to evaluate impact of
  genetic education (learning APOE genotype) on 148
  adult children of AD patients
• Control group (46 subjects)- Told of AD risk
  based on age / gender / family history
• Intervention group (102 subjects)- Told of AD
  risk based on age / gender / family history /
  APOE genotype

18
Subjects with family history of AD morelikely to
have already purchased LTC

• Percentage with insurance at beginning of study
• Health- 97
• Life- 78
• Disability- 40
• LTC- 20
• In U.S., 7 of aged 65 have LTC insurance,
  compared to 20 baseline in REVEAL
• Even before APOE known, subjects with family
  history of AD were 3 times more likely to have
  already purchased LTC
• Other possible reasons for greater baseline LTC
  coverage were higher income / education /
  socioeconomic status

19
Results
Life
e4 positive more likely to think of changing
coverage
e4 positive much more likely to change / think of
changing coverage, e4 negative less likely to
change
LTC
20
AD perfect storm for LTC antiselection

• Perfect storm
• High likelihood of needing care
• APOE gives significant (but incomplete)
  predictive information
• LTC private market in U.S.
• Overall, e4 positive subjects 5.8 times more
  likely to increase LTC coverage
• Authors suggest LTC purchase might increase
  further with older age, i.e., no perceived need
  to buy LTC at age 52 (average age of study
  subjects)

21
Early Testing for Alzheimers Disease

• Genetics / APOE
• REVEAL study
• Underwriting

22
Incidence rate of AD doubles every 5 years
Age 62- 1 new case per 1000 people
Age 79- 1 new case per 100 people
Are underwriting requirements, especially
cognitive testing, in line with AD incidence
rates?
Age 94- 1 new case per 10 people
Age 105- Incidence rate equals mortality rate
23
Family history (FH)

• 25 of late-onset AD have close relative with AD
• Ask older applicants about FH
• At what age was condition diagnosed (e.g., AD,
  heart disease, cancer), not At what age did
  death occur from condition
• FH important if
• Any AD under age 60
• AD in two or more elderly parents, siblings,
  aunts, uncles

24
Likelihood of AD varieswith age, family history,
APOE
Annual incidence of AD between ages 65 and 80
Population 0.3 First degree relative
0.8 APOE e3/e4 1.6
Cupples LA, et al. Genet Med 20046192-6.
25
Attending physicians statement often not helpful
Physician usually does not know about mild AD
Physician may not record AD diagnosis in records
Potential harm of early identification of
dementia relates to repercussions of the label of
cognitive impairment, including inability to
obtain life or health insurance (J Am Geriatr
Soc 2000481430-34)
26
Memory loss AD until proven otherwise

• If memory loss
• Full evaluation by specialist
• Cognitive tests- Borderline low score often /
  usually early AD
• MRI, blood, etc., for other impairments-
  Depression, brain tumor, multi-infarct dementia,
  systemic illness
• Time (12 months, minimum) for physician to follow
  patient, repeat tests (mild cognitive impairment
  ? AD at 10-15 per year)

27
Applicants with known AD

• Suicide risk- Claims within contestable period
  might be accidental death
• Moral risk if beneficiary initiates insurance
  purchase
• Legal risk- Often cannot contest
  misrepresentation (applicant forgot past medical
  history)
• Risk classification difficult because great
  variation in survival data in medical literature

28
AD major concern for LTCI products with LTC
ridereven without APOE issue

• AD already most common / most costly / longest
  LTC claim
• Dementia definition weak
• Easy to misrepresent Significant impairment on
  cognitive test
• Easy to misrepresent Need for continuous
  supervision
• REVEAL
• People with FH of AD 3 times more likely to
  already have LTC
• e4 positive subjects 5.8 times more likely to
  increase LTC coverage

29
AD incidence will increase

• Growing emphasis on screening for prevention /
  early treatment (when better treatment becomes
  available)
• Routine screening in future for people with early
  memory loss, family history of AD, worried
  well, healthy elderly
• www.medafile.com
• Free online memory test for early identification
  of AD
• Suggested frequency- Age 50-65, each year 65,
  every 3 months

30
Tests to screen for AD
Neuropsychological tests
Delayed word recall, other

• MRI- Small hippocampus (memory)
• PET (positron emission tomography) scan- Slower
  glucose metabolism

Brain scans
Electroencephalograph (EEG)
Abnormal pattern predicts AD
Abnormal with AD (strawberry, lemon, smoke,
lilac, leather), online purchase
Scratch and sniff test
Genetic predisposition
Family history, apolipoprotein e4
Mathematical models
Combine data from above to predict AD
31
Claims scenarios when AD incidence increases
No breakthrough in treatment
? number of claims, ? claim duration
What happens if incidence of AD increases in 2010
due to routine screening?
Breakthrough in treatment
? number of claims, ? claim duration
? number of claims, ? claim duration
? number of claims, ? claim duration
? number of claims, ? claim duration
What happens if AD cured, controlled, or
prevented in 2025?
? ? ? in number, duration of claims
32
What insurers can dowhen/before AD incidence
increases

• More cognitive tests based on age, amount at
  risk, cost / benefit analysis
• e.g., age 60- 500,000
• e.g., age 65- 250,000
• Use family history to modify risk
• Physicians statement often wont reveal existing
  AD

Underwriting
Claims
More restrictive benefit criteria
Increase price
Pricing
Product design
Reduce risk per contract provisions
33
Why insurance companies do dumb things

• Denial of problem
• Tension between short- and long term goals
• e.g., need to meet quarterly sales targets
  greater than worry about long term risks
• Crompton, Robert B. Why insurance companies do
  dumb things. Contingencies Nov/Dec 2003, p. 8-12

34
Why insurance companies do dumb things
Antiselection not a problem
Weve priced for this
Dont need more cognitive tests
Well sort this out next quarter
Do any of these beliefs apply to LTC insurance?
Underwriting protects us
We can always raise rates if mistakes in product
design
Claims people protect us
Market share our 1 goal (same as many former LTC
insurers)
Treatment will decrease claims
Ill be long retired by then
35
Early Testing for Alzheimers Disease

• Robert Pokorski, MD, MBA

36
(No Transcript)
37
(No Transcript)
38
Introduction

• How Insurance Works Theory and Reality
• Possible Pricing/Profitability Implications
• Managing The Risk

39
How Insurance Works

• There are things we know that we know
• There are things we know that we don't know 
• But, there are also things we dont know that
  we don't know
•   Donald Rumsfeld 
• PS There are things we thought we didnt know
  well enough to act on, chose to ignore, and now
  we look really, really dumb!

40
How Insurance Works

• Theory
• Same risks / same rates (underwriting)
• Actual versus expected experience is statistical
  (actuarial)
• Experience is predictable (law of large numbers)
• If above not met, that becomes known and is
  correctable

41
How Insurance Works

• Reality
• Underwriting basis, soft/hard data, applicant
  knows more
• Actuarial assumptions can turn out not to fit
  reality
• Predictable experience need large numbers plus
  time
• Threats - slow to anticipate, recognize, address,
  or unable to sufficiently address
• All insurance lines grapple with these realities

42
How Insurance Works

• Reality
• Risk/rate classes are not populated by identical
  risks
• Class or subclass may benefit at the expense of
  others
• Problems can arise, if that subsidy
• is significant
• adversely influences buyers decisions
  (anti-selection)
• causes experience to become adverse,
  unpredictable
• promotes cherry picking by competitors

43
How Insurance Works

• ex until the 1970s, life insurance for smokers
  was highly subsidized by non-smokers
• insurers initially slow to offer S/NS rates
• first lost market share, made agents unhappy, but
  was able to increase profitability by attracting
  non-smokers
• last had little choice, mostly writing smokers at
  inadequate, undifferentiated rates
• follow the leader or follow the bleeder?
• ex LTC, before spouse discounts couples
  subsidized singles

44
How Insurance Works

• Risk classification doesnt need to be perfect,
  BUT
• requires successfully dealing with asymmetric
  information (Nobel Prize in Economics, 2001)
• new information leads to new behavior (e.g.,
  REVEAL)
• underwriting is on the front line
• Need to learn from history or ...
• Need to help educate policymakers, regulators,
  and customers as to adverse consequences of
  requiring subsidies among rate/risk classes in
  private insurance

45
Possible Pricing/Profitability Implications

• Past Present Future
• Role of genetics? APOE e4 risk factor How
  many will test?
• Family history? FH risk factor Degree of
  anti-select?
• Anti-selection demod Use in AD models?
• Low awareness Industry will respond
• Ignored in pricing,u/w New pricing,U/W?
• Regulatory
  restrictions Regulatory response?
• Market response?

46
Possible Pricing/Profitability Implications
47
Possible Pricing/Profitability Implications

• Simplicity is the ultimate sophistication.
  Leonardo Da Vinci

Source Ashford JW. J Mol Neurosc 200523157-65
48
Possible Pricing/Profitability Implications

49
Possible Pricing/Profitability Implications

50
Possible Pricing/Profitability Implications

• Major Caveats Include
• not the worst-case scenario (if all risks were
  e4/e4)
• assumes no anti-selection by benefit amount or
  options
• assumes prevalence-based AD risk relationships
• does not reflect receiving less premium from
  extra claims
• does not assume higher/lower persistency for
  e4-pos/neg
• assumes AD in expected LR consistent with e4 mix
  US Pop gt60
• assumes no break-through in AD treatment or
  prevention
• no rate increases

51
Possible Pricing/Profitability Implications

• All models are wrong. Some models are useful.
  George Box
• A Mathematical Model of Alzheimers Disease And
  The APOE Gene, ASTIN Bulletin, Vol 30, No 1,
  2000, pp 69-110Authors Angus Macdonald and Delme
  Pritchard
• Includes good primer on human genetics
• AD model (Markov type), not an economic insurance
  model
• Projection model for AD, specific to sex, age,
  APOE genotype
• 4 buckets no AD, AD not inst, AD inst, dead
• Uses transition probabilities from
  medical/epidemiological studies
• Case-based studies may yield higher risk factors
  than true pop

52
Possible Pricing/Profitability Implications

• Genetics, Alzheimers Disease,and Long-Term
  Care Insurance North American Actuarial Journal,
  Vol 5, No 2, pp 54-78,Authors Angus Macdonald
  and Delme Pritchard
• builds on authors AD model
• concludes adverse selection increases costs
  significantly only if
• population risk not much smaller than case-based
  studies
• carriers of the e4 allele are very much more
  likely to buy LTC
• but assumed
• low-risk genotypes just as likely to insure after
  testing as before
• benefit amount and options do not depend on
  genetic test result 
• single premium only

53
Possible Pricing/Profitability Implications

•  

• low risk genotypes just as likely to insure
  after testing as before
• assumes no anti-selection by benefit amount or
  options
• case-based study relative risk could overstate
  that of population
• benefits start 1 year before institutionalization

54
Possible Pricing/Profitability Implications

•  

• low risk genotypes just as likely to insure
  after testing as before
• assumes no anti-selection by benefit amount or
  options
• case-based study relative risk could overstate
  that of population
• benefits start 1 year before institutionalization

55
Managing The Risk

• Product
• Benefits
• higher/lower AD risk apps likely to choose
  higher/lower benefit
• continue to offer a wide choice of benefit
  options?
• continue to offer unlimited lifetime benefit?
• offer new benefits for AD preventive and
  treatment therapies?
• Wording
• CI trigger defensible against claims for
  pre-sympto/mild AD?
• is AD covered rather than not excluded?
• Encourage existing insureds to test?

56
Managing The Risk

• Pricing
• anti-selection is certain, size of impact less
  certain, but likely significant
• FH and e4-status are significant AD risk factors
• low market penetration compounds anti-selection
• REVEAL Study
• subjects with family history were 3x more likely
  to have LTCI
• e4 pos 5.8x more likely to purchase/increase LTCI
• Review AD literature, studies, and models
• Lack of data, use common sense, rather than
  ignore key factors

57
Managing The Risk

• Marketing
• will agents and financial planners recommend APOE
  testing?
• will emphasizing AD in marketing attract more FH
  and e4 pos?
• Underwriting
• family history question?
• had any test that indicated a higher likelihood
  of developing AD?
• Claims
• pay only based on objective criteria behind the
  CI trigger wording?
• fear of lawsuits?
• pay on AD diagnosis or early-dementia?
• pay if claimant wants to be on claim, but doesnt
  really qualify?

58
Managing The Risk

• Regulatory
• most regulate use of genetic info, but exempt
  life insurance
• most prohibit use of genetic info in employment
  and health ins
• some states have extended such limitation to LTCI
• ACLI Genetic Testing Laws Regs Affecting
  Life, LTC, and DI
• Customer/Public Relations
• unfair to use genetic info? only a risk, not a
  condition, not my fault
• fair that low risk insureds should subsidize high
  risk insureds?
• private insurance is optional and vulnerable to
  subsidy issue
• public programs that mandate participation are
  less vulnerable

59
Managing The Risk

• Capital/Investors
• APOE AD risk factor has been known for over a
  decade
• little publicity to date that connects the dots
  to LTCI
• when (not if) that occurs, there will be
  questions to answer
• what is the potential impact?
• can it be managed?
• what are you doing about it?
• do you know what youre doing?

60
Early Testing for Alzheimers Disease

• Philip Barackman, FSA, MAAA, FLMI
• February 2006