Dumb Side of Intelligence

1
Dumb Side of Intelligence Research Association
Studies
Tony Payton CIGMR
2

• Association studies powerful methods for
• identifying genes that influence behavioural
• or disease phenotype

• Simple principle compare allele frequencies
• in cases and controls

• Inconsistency in literature
• Examples
• Reducing spurious findings
• Manchester cognitive study

3
Publication trends in field of psychiatric
genetics
4

• Multiple positive associations without failure
• to replicate is rare

• 166 disease/gene associations investigated
• ?3 groups
• 6 consistently replicated
• Hirschhorn et al, (2002) Genet Med 4, 45-61

5
Serotonin Transporter Gene (5-HTT)

• Short allele ? transcriptional efficiency

• Possession of short allele associated with
  neuroticism
• 8 of genetic variance (505 subjects, P0.0031)
• Lesch et al. 1996. Science, 274, 1527-1531

6
A total of 22 studies some positive but majority
negative
7
Adapted from Moyzis et al, 2002
Association between 7 48bp repeat in exon 3 of
DRD4 and novelty seeking (Benjamin et al, 1996,
Nature Genetics)
To date 15 studies performed reporting both
positive and negative associations
By 2001 23 publications for DRD4 VNTR and ADHD
8

• Alcoholism
• Schizophrenia
• Novelty seeking
• Attention deficit hyperactivity disorder
• Depression

9
Late-onset Alzheimers disease

• In contrast to neuropsychiatric conditions
• defined histological features
• diagnostic criterea

• 2003 gt10 publications/month
• 50 genes associated

• 1 consistently replicated by
• multiple groups
• ?4 allele of APOE

10
Intelligence

• General intelligence is a diverse range of
• mental abilities
• Processing speed
• Vocabulary ability
• Memory

• Cognitive ability is highly heritable
• gt110, 000 twin/sib pairs
• heritability approx. 50
• Bouchard and McGue, 1981

11
Genes associated with intelligence

• Insulin-like growth factor 2 receptor (Chorney
  et al, 1998)
• Muscle segment homeobox 1 (Fischer et al, 1999)
• Catechol-O-methyltransferase (Egan et al, 2001)
• Dopamine receptor D2 (Bartres-Faz et al, 2002)
• Cholinergic muscarinic 2 receptor (Comings et
  al, 2003)
• Brain derived neurotrophic factor (Egan et al,
  2003)
• Cathepsin D (Payton et al, 2003)
• Serotonin receptor 2A (Quevain et al, 2003)

12
Has intelligence research learned from
previous behavioural studies?
13
Failure to replicate

• IGF2R Hill et al, 2002
• DRD2 Moises et al, 2001
• Ball et al, 1998
• Petrill et al, 1997

Whats going on?
14
Complexity

• 1.5 million SNPs in 33,000 genes

• Multiple diseases associated with a single gene
• APOE AD, cancer, heart disease, diabetes, MS
• 30 other diseases/traits

• Different outcomes
• Dementia Alzheimers disease, vascular disease,
• frontotemporal, Lewy body
• combinations of above

• Multiple alleles
• APOE 3 common alleles, 6 genotypes

• Subgroup splits
• men vs women, young vs old etc

15
Fishing for genes without getting bitten 5
considerations
16
1. Adequate Power

• Contribution of gene towards variance of trait
• or risk of disease is 1-8

• Adequately powered study require 1000s subjects
• frequency of polymorphism
• odds ratio 1.1-1.5 (10-50 relative ? in
  likelihood)

• Odds ratio 1.3
• similar allele frequency 2,000-10,000
  subjects
• rare disease allele 160,000
• 80 power
• Zondervan and Cardon (2004) Nat Rev Genet 5,
  89-100

17
Sample sizes in 579 genetic association studies
Adapted from Ioannidis (2003) Trends in Mol Med
9(4) 135-138
18
2. Replicability

• Validation by an independent dataset prior
• to publication
• in-house
• collaboration

• lt20 of all studies in AD genetics used an
• independent cohort

• 8 publications reporting intelligence genes
• 1 has replicated using two cohorts
• Quervain et al, 2003

19
3. Use of Haplotypes

• Haplotype analysis
• ? power
• ? number of tests
• ? false positives
• 5q cytokine locus/Crohn disease (Rioux et al,
  2001)
• Calpain 10/type 2 diabetes (Horikawa et al,
  2000)
• Tau/supranuclear palsy (Conrad et al, 1997)

• 2003 only 1/3 association studies examined
• gt1 polymorphism/gene
• half examined gt4 polymorphisms/gene

20

• In 2003 only 23 of AD publications fulfilled
• ?2 3 basic criterea
• sufficient power
• replication
• haplotype analysis

4. Biological plausibility majority of studies
meet this criterea animal models expression
studies linkage studies
21
5. Meta-analysis

• Combines results across studies

• Two results at p 0.06 are stronger evidence
  against
• the null (p 0.0014) than a single result at
  p 0.05
• Ten results at p 0.10 are stronger evidence
  against
• the null (p 0.000025) than five at p 0.05
  (p 0.00012)
• Rosenthal and DiMatteo (2001) Annu Rev Psychol
  52, 59-82

22
Meta-analysis publications in field of
psychiatric genetics
23
5HTT Neuroticism 44bp promoter
Munafò et al. (2003). Molecular Psychiatry, 8,
471-484
22 case-control studies 4 positive 18
negative p0.038
24
DRD4 Neuroticism 7 48bp rpt exon 3
Munafò et al. (2003). Molecular Psychiatry, 8,
471-484
17 case-control studies 4 positive 13
negative Not significant
25
DRD4 ADHD
Faraone et al, 2001,Am J Psy
8 case-control studies 4334 5 positive 3
negative p0.001, OR 1.9
14 TDT studies 3300 2 positive 12
negative p0.02, OR 1.4
26
Dopamine receptor D5 and ADHD

• Microsatelite marker upstream of DRD5 (Daly et
  al, 1999)

• 3 studies failed to replicate (non-significant
  trend)
• Barr et al, 2000
• Tahir et al, 2000
• Payton et al, 2001

• Meta-analysis of 14 studies (gt5000)
• p0.00005, OR 1.24, 95CI 1.12-1.38
• Lowe et al, 2004 (Human Genetics 2004 in press)

27

• Popularity of meta-analysis is that it can
  address
• issue of power

• Successful
• 20-30 studies show significance/combined
• Lohmueller et al, 2003

• Limitations
• investigate 2 genotype groups
• accurately adjust for heterogeneity

• Not as good as adequately powered primary
• association study

28
(No Transcript)
29
Manchester cognitive study

• Aim
• Establish a large cohort healthy older
  volunteers
• Investigate genetic association with cognitive
  ability and rate of cognitive decline

1. Power gt2200 volunteers involved
2. Replication 2 cohorts from Manchester and
Newcastle Each cohort gt1100 volunteers
30
3. Haplotypes
Cholinergic Gene Locus
31
4. Biological Plausibility
32
5. Study Design

• Well characterised cohort
• Caucasian
• Physiological tests
• Psychological tests
• General Health
• Demographic Data
• Array of cognitive tests
• Processing speed
• Vocabulary
• Memory

33

• All volunteers tested at five year intervals
• (minimum follow-up of fifteen years)
• Allows identification of genes that regulate
  both
• level of cognitive ability and rate of
  decline of
• cognitive domains

• Identical tests performed

34
Summary

• Current research
• minimum of 1000 subjects
• replication group

• Use of haplotypes ? power, ? tests, ? false
  positives

• Meta-analysis good for ? power

• Has research in the field of cognitive genetics
• learned anything from older association studies?

Not really!
35
THE END
"Only two things are infinite, the universe and
human stupidity, and I'm not sure about the
former" Albert Einstein
36
5. Analytical methods

• Population stratification

• Disease frequency varies between subpopulations
• chance affected individuals selected from
  subpopulation
• marker allele ? frequency in subpopulation

• TDT
• Spielman et al, 2003
• Genomic control to access ancestry
• Pritchard et al, 2000