Genetic association studies of complex diseases

1
Genetic association studies of complex diseases

• Mario A. Cleves, Ph.D. UAMS, College of
  Medicine,
• Department of PediatricsArkansas Center for
  Birth Defects Research and Prevention March 9,
  2005

2
OUTLINE

• Complex diseases
• Genetic epidemiology
• Genetic linkage linkage disequilibrium
• Association studies

3
Complex Diseases

• Characteristics
• Aggregate in families but do not segregate in
  Mendelian fashion

4
Complex Diseases

• Selected Characteristics
• Aggregate in families but do not segregate in
  mendelian fashion
• Multigenic
• Multiple environmental factors
• Phenotypic Genetic heterogeneity

5
Transposition of the great arteries
(TGA)Phenotypic Heterogeneity or genotypic
heterogeneity
6
Complex Diseases

• Selected Characteristics
• Aggregate in families but do not segregate in
  mendelian fashion
• Multigenic
• Multiple environmental factors
• Phenotypic Genetic heterogeneity
• Phenocopies
• Variable (incomplete) penetrance

7
Penetrance

• Probability of trait (disease) given the genotype
  
• penetrance p(D1genotype)

• Examples Assume 3 possible genotypes at a locus
  (AA,Aa,aa)
• p(DAA) 1, p(DAa) 1, p(Daa) 0
• p(DAA) 1, p(DAa) 0, p(Daa) 0
• p(DAA) lt 1, p(DAa) lt1, p(Daa) 0
• Incomplete penetrance penetrance lt 1
• Phenocopies p(Daa) gt 0

8
Peltonen et al. GENOMICS AND MEDICINE Dissecting
Human Disease in the Postgenomic Era. Science,
Vol 2915507, 1224-1229.
9
Complex Diseases

• Selected Characteristics
• Aggregate in families but do not segregate in
  mendelian fashion
• Multigenic
• Multiple environmental factors
• Phenotypic Genetic heterogeneity
• Phenocopies
• Variable penetrance
• Complex interplay between genetic and
  environmental factors

10
Complex Diseases

• Examples

11
Complex Diseases

• Mapping is easier for traits with high l
• Complex diseases tend to have low l
• Large l suggest a major disease genes
• Small l may indicate many genes, each
  contributing a small effect
• ls 500 Cystic Fibrosis
• ls 15 Type I Diabetes
• ls 1 4 Hypertension

12
(No Transcript)
13
GENETIC EPIDEMIOLOGY

• Broad goal
• To understand the complex interaction between
  genes and environmental factors in the etiology
  of diseases
• Ultimate goal
• Disease treatment, control and prevention.

14

• Goal Identify disease susceptibility genes

15
Research flow
16
Gene identification(General approaches)
Examples PKU, Sickle Cell Anemia
Examples Huntington Gene, Cystic Fibrosis
17
Positional cloning Statistical approaches to
gene mapping
Gene mapping locating genes on the genome

• Linkage analysis
• Association studies (Linkage disequilibrium)

18
Genetic linkage

• The tendency of genes to segregate together
• Two loci are linked if they are sufficiently
  close on a chromosome so that crossing over does
  not occur between them

19
(No Transcript)
20
Linkage analysis

• Determine the approximate location of a disease
  predisposing genes
• Uses a genetic marker map
• Looks for co-segregation with a marker
• Simple Idea
• Determine if marker allele at a known locations
  travels with the disease in a family

21
Linkage analysis

• Great success in identifying genes for simple
  Mendelian diseases
• Few success in identifying genes contributing to
  complex disease
• Unsuccessful in identifying genes contributing to
  common complex disease

22
Linkage disequilibrium (LD)

• The nonrandom association of alleles in the
  population
• Alleles at neighboring loci tend to cosegregate
• Linkage disequilibrium implies population allelic
  association

23
LD around an ancestral mutation
24
LD measures
Loci are in linkage equilibrium if p(A,C)
p(A)p(C)
D p(A,C) - p(A) p(C)
D p(A,C) p(G,T) - p(A,T) p(G,C)
25
LD measures

26
Decay of linkage disequilibrium
27
Linkage Disequilibrium Mapping

• Population based
• Look for variant allele in LD with disease
• If most affected individuals in a population
  share the same mutant allele, then LD can be used
  to locate the chromosomal region harboring the
  allele

28
Study designs

• Case-Control with unrelated controls
• Family-based studies
• Matched Case-Control studies
• Cohort Studies

29
Case-Control Studies with Unrelated Controls

• Common method in epidemiology
• Sample cases
• Sample controls from the same population
• This implies that cases and controls should have
  similar genetic backgrounds

30
Case-Control (unrelated controls)
31
Case-Control (unrelated controls)computing
Interaction effects
ORint OREG/ ORE ORG
ORint OREG/ (ORE ORG - 1)
32
Case-Control (unrelated controls)computing
Interaction effects
ORint 1.17/(0.77 x 2.46) 0.61824457
33
Case-Control (unrelated controls)

• Advantages
• Easy to implement
• Estimates main effects and interactions
• We know how to do these studies
• Disadvantages
• Admixture or population stratification
• Requires large sample sizes to detect
  interactions
• 4 times larger than for main effects if
  environmental and genetic factors are common
• Prohibited when factors are rare

34
Family-based Studies

• Use family members as controls
• Adv Eliminates bias due to admixture
• Two popular family designs
• Siblings or cousins as controls
• Case-parent triads

35
Non-affected siblings as controls

• Match case with unaffected sibling
• Analyze using conditional logistic regression
• Disadvantages
• Delayed age of onset
• Younger sibs may not be disease-free at the age
  that cases were diagnosed
• Exposure histories of older sibs may be different
• Older sibs may be more susceptible to recall bias

36
Non-affected Cousins as controls

• Match case with unaffected 1st cousin
• Analyze using conditional logistic regression
• Advantages over siblings as controls
• Closer matching on age
• Usually more cousins than sibs available
• Disadvantages
• Reduce participation rate (motivation or
  geography)
• No absolute protection from admixture effect

37
Case-parent triads

• Use affected child and both parents
• Disease onset must occur early enough for parents
  to be available

38
Transmission Disequilibrium Test (TDT)

• Looks at transmission of alleles from
  heterozygous parents to affected children
• Test if there is deviation from Mendelian
  segregation ratios
• If there is transmission distortion this suggest
  an etiologic association between the allele and
  the disease

39
a.
b.
M
M
M
M
M
M
M
M
1
2
1
1
2
1
2
2
M
M
M
M
1
1
1
2
a. 2 x (M1 Transmitted M2 nontransmitted) b.
(M1 Transmitted M2 nontransmitted) (M2
Transmitted M1 nontransmitted)
40

• For n families, 2n parents

Test with the McNemar c2 test for two related
samples
41
Case-parent triadsExtended-TDT for GxE
interaction

• Compare the estimated transmission probabilities
  of exposed and unexposed trios

42
Case-parent triads Conditional on parental
genotype (CPG)

• Alternative to the TDT that allows for a variety
  of genetic relative risk models.
• Unit of analysis is the triad, not the
  transmitted alleles
• Allows the inclusion of incomplete family data
  EM approach

43
Case-parent triads Conditional on parental
genotype (CPG)

• Assume 3 genotypes (AA,Aa,aa)
• Define A as the high risk allele, and the aa
  genotype as the baseline genotype.
• R1 genotype RR for heterozygous (Aa)
• R2 genotype RR for homozygous AA
• R1 R2 1 ? No disease-allele association

44
Case-parent triads Conditional on parental
genotype (CPG)

• Define 5 genetic relative risk models
• general arbitrary r1 and r2
• multiplicative allele effect r1 r, r2 r2
• additive allele effect r1 r, r2 2r -1
• dominant allele effect r1 r2 r
• recessive allele effect r1 1, r2 r
• The likelihood conditions the childs genotype on
  their parents genotypes

45
Probability of case genotype given mating type
L(r1, r2) ? r2 / (r1r2 ) n2r1 / (r1r2 )
n3r2 / (r22r11) n5 x 2r1 / (r22r11)
n61/ (r22r11) n7r1 / (r11) n8 x 1 /
(r11) n9
46
Conditional on parental genotype (CPG)

• Fit using a log-linear model
• Poisson regression software
• (Software must allow an offset and
  stratification)
• Can be parameterizes to allow for
• missing parents (incomplete triads)
• effects of maternal genotype
• the assessment of imprinting

47
Conditional on parental genotype (CPG)GxE
extension

• Test for GxE interaction

Distributed as a c2 with 2 df
Schaid, 1999
48
Conditional on parental genotype (CPG)

• Specific advantages of the CPG method
• LRT remains valid in the presence of incomplete
  population mixing
• Method has been extended for continuous
  environmental exposures and multiple alleles

49
(No Transcript)
50
The End